Struct polars_core::series::Series
source · pub struct Series(pub Arc<dyn SeriesTrait>);Expand description
Series
The columnar data type for a DataFrame.
Most of the available functions are defined in the SeriesTrait trait.
The Series struct consists
of typed ChunkedArray’s. To quickly cast
a Series to a ChunkedArray you can call the method with the name of the type:
let s: Series = [1, 2, 3].iter().collect();
// Quickly obtain the ChunkedArray wrapped by the Series.
let chunked_array = s.i32().unwrap();Arithmetic
You can do standard arithmetic on series.
let s = Series::new("a", [1 , 2, 3]);
let out_add = &s + &s;
let out_sub = &s - &s;
let out_div = &s / &s;
let out_mul = &s * &s;Or with series and numbers.
let s: Series = (1..3).collect();
let out_add_one = &s + 1;
let out_multiply = &s * 10;
// Could not overload left hand side operator.
let out_divide = 1.div(&s);
let out_add = 1.add(&s);
let out_subtract = 1.sub(&s);
let out_multiply = 1.mul(&s);Comparison
You can obtain boolean mask by comparing series.
let s = Series::new("dollars", &[1, 2, 3]);
let mask = s.equal(1).unwrap();
let valid = [true, false, false].iter();
assert!(mask
.into_iter()
.map(|opt_bool| opt_bool.unwrap()) // option, because series can be null
.zip(valid)
.all(|(a, b)| a == *b))See all the comparison operators in the CmpOps trait
Iterators
The Series variants contain differently typed ChunkedArray’s. These structs can be turned into iterators, making it possible to use any function/ closure you want on a Series.
These iterators return an Option<T> because the values of a series may be null.
use polars_core::prelude::*;
let pi = 3.14;
let s = Series::new("angle", [2f32 * pi, pi, 1.5 * pi].as_ref());
let s_cos: Series = s.f32()
.expect("series was not an f32 dtype")
.into_iter()
.map(|opt_angle| opt_angle.map(|angle| angle.cos()))
.collect();Creation
Series can be create from different data structures. Below we’ll show a few ways we can create a Series object.
// Series can be created from Vec's, slices and arrays
Series::new("boolean series", &[true, false, true]);
Series::new("int series", &[1, 2, 3]);
// And can be nullable
Series::new("got nulls", &[Some(1), None, Some(2)]);
// Series can also be collected from iterators
let from_iter: Series = (0..10)
.into_iter()
.collect();
Tuple Fields§
§0: Arc<dyn SeriesTrait>Implementations§
source§impl Series
impl Series
sourcepub fn sample_n(
&self,
n: usize,
with_replacement: bool,
shuffle: bool,
seed: Option<u64>
) -> PolarsResult<Self>
Available on crate feature random only.
pub fn sample_n(
&self,
n: usize,
with_replacement: bool,
shuffle: bool,
seed: Option<u64>
) -> PolarsResult<Self>
random only.sourcepub fn sample_frac(
&self,
frac: f64,
with_replacement: bool,
shuffle: bool,
seed: Option<u64>
) -> PolarsResult<Self>
Available on crate feature random only.
pub fn sample_frac(
&self,
frac: f64,
with_replacement: bool,
shuffle: bool,
seed: Option<u64>
) -> PolarsResult<Self>
random only.Sample a fraction between 0.0-1.0 of this ChunkedArray.
pub fn shuffle(&self, seed: Option<u64>) -> Self
Available on crate feature
random only.source§impl Series
impl Series
sourcepub fn from_any_values_and_dtype(
name: &str,
av: &[AnyValue<'_>],
dtype: &DataType
) -> PolarsResult<Series>
pub fn from_any_values_and_dtype(
name: &str,
av: &[AnyValue<'_>],
dtype: &DataType
) -> PolarsResult<Series>
Examples found in repository?
More examples
src/frame/row.rs (line 501)
482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503
pub fn into_series(self) -> Series {
use AnyValueBuffer::*;
match self {
Boolean(b) => b.finish().into_series(),
Int32(b) => b.finish().into_series(),
Int64(b) => b.finish().into_series(),
UInt32(b) => b.finish().into_series(),
UInt64(b) => b.finish().into_series(),
#[cfg(feature = "dtype-date")]
Date(b) => b.finish().into_date().into_series(),
#[cfg(feature = "dtype-datetime")]
Datetime(b, tu, tz) => b.finish().into_datetime(tu, tz).into_series(),
#[cfg(feature = "dtype-duration")]
Duration(b, tu) => b.finish().into_duration(tu).into_series(),
#[cfg(feature = "dtype-time")]
Time(b) => b.finish().into_time().into_series(),
Float32(b) => b.finish().into_series(),
Float64(b) => b.finish().into_series(),
Utf8(b) => b.finish().into_series(),
All(dtype, vals) => Series::from_any_values_and_dtype("", &vals, &dtype).unwrap(),
}
}sourcepub fn from_any_values(name: &str, av: &[AnyValue<'_>]) -> PolarsResult<Series>
pub fn from_any_values(name: &str, av: &[AnyValue<'_>]) -> PolarsResult<Series>
source§impl Series
impl Series
sourcepub fn array_ref(&self, chunk_idx: usize) -> &ArrayRef
pub fn array_ref(&self, chunk_idx: usize) -> &ArrayRef
Returns a reference to the Arrow ArrayRef
Examples found in repository?
More examples
src/series/ops/to_list.rs (line 34)
32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60
pub fn to_list(&self) -> PolarsResult<ListChunked> {
let s = self.rechunk();
let values = s.array_ref(0);
let offsets = vec![0i64, values.len() as i64];
let inner_type = self.dtype();
let data_type = ListArray::<i64>::default_datatype(inner_type.to_physical().to_arrow());
// Safety:
// offsets are correct;
let arr = unsafe {
ListArray::new(
data_type,
Offsets::new_unchecked(offsets).into(),
values.clone(),
None,
)
};
let name = self.name();
let mut ca = ListChunked::from_chunks(name, vec![Box::new(arr)]);
if self.dtype() != &self.dtype().to_physical() {
ca.to_logical(inner_type.clone())
}
ca.set_fast_explode();
Ok(ca)
}src/chunked_array/list/iterator.rs (line 110)
94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120
pub fn amortized_iter(&self) -> AmortizedListIter<impl Iterator<Item = Option<ArrayBox>> + '_> {
// we create the series container from the inner array
// so that the container has the proper dtype.
let arr = self.downcast_iter().next().unwrap();
let inner_values = arr.values();
// Safety:
// inner types logical type fits physical type
let series_container = unsafe {
Box::new(Series::from_chunks_and_dtype_unchecked(
"",
vec![inner_values.clone()],
&self.inner_dtype(),
))
};
let ptr = series_container.array_ref(0) as *const ArrayRef as *mut ArrayRef;
AmortizedListIter {
len: self.len(),
series_container,
inner: NonNull::new(ptr).unwrap(),
lifetime: PhantomData,
iter: self.downcast_iter().flat_map(|arr| arr.iter()),
inner_dtype: self.inner_dtype(),
}
}src/chunked_array/kernels/take.rs (line 56)
40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84
pub(crate) unsafe fn take_list_unchecked(
values: &ListArray<i64>,
indices: &IdxArr,
) -> ListArray<i64> {
// taking the whole list or a contiguous sublist
let (list_indices, offsets) = take_value_indices_from_list(values, indices);
// tmp series so that we can take primitives from it
let s = Series::try_from(("", values.values().clone() as ArrayRef)).unwrap();
let taken = s
.take_unchecked(&IdxCa::from_chunks(
"",
vec![Box::new(list_indices) as ArrayRef],
))
.unwrap();
let taken = taken.array_ref(0).clone();
let validity =
// if null count > 0
if values.has_validity() || indices.has_validity() {
// determine null buffer, which are a function of `values` and `indices`
let mut validity = MutableBitmap::with_capacity(indices.len());
let validity_ptr = validity.as_slice().as_ptr() as *mut u8;
validity.extend_constant(indices.len(), true);
{
offsets.as_slice().windows(2).enumerate().for_each(
|(i, window): (usize, &[i64])| {
if window[0] == window[1] {
// offsets are equal, slot is null
unset_bit_raw(validity_ptr, i);
}
},
);
}
Some(validity.into())
} else {
None
};
let dtype = ListArray::<i64>::default_datatype(taken.data_type().clone());
// Safety:
// offsets are monotonically increasing
ListArray::new(dtype, offsets.into(), taken, validity)
}src/chunked_array/cast.rs (line 197)
175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212
fn cast(&self, data_type: &DataType) -> PolarsResult<Series> {
match data_type {
DataType::List(child_type) => {
let phys_child = child_type.to_physical();
if phys_child.is_primitive() {
let mut ca = if child_type.to_physical() != self.inner_dtype().to_physical() {
let chunks = self
.downcast_iter()
.map(|list| cast_inner_list_type(list, &phys_child))
.collect::<PolarsResult<_>>()?;
ListChunked::from_chunks(self.name(), chunks)
} else {
self.clone()
};
ca.set_inner_dtype(*child_type.clone());
Ok(ca.into_series())
} else {
let ca = self.rechunk();
let arr = ca.downcast_iter().next().unwrap();
let s = Series::try_from(("", arr.values().clone())).unwrap();
let new_inner = s.cast(child_type)?;
let new_values = new_inner.array_ref(0).clone();
let data_type =
ListArray::<i64>::default_datatype(new_values.data_type().clone());
let new_arr = ListArray::<i64>::new(
data_type,
arr.offsets().clone(),
new_values,
arr.validity().cloned(),
);
Series::try_from((self.name(), Box::new(new_arr) as ArrayRef))
}
}
_ => Err(PolarsError::ComputeError("Cannot cast list type".into())),
}
}src/chunked_array/from.rs (line 37)
4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46
fn from_chunks_list_dtype(chunks: &mut Vec<ArrayRef>, dtype: DataType) -> DataType {
// ensure we don't get List<null>
let dtype = if let Some(arr) = chunks.get(0) {
arr.data_type().into()
} else {
dtype
};
match dtype {
#[cfg(feature = "dtype-categorical")]
// arrow dictionaries are not nested as dictionaries, but only by their keys, so we must
// change the list-value array to the keys and store the dicitonary values in the datatype.
// if a global string cache is set, we also must modify the keys.
DataType::List(inner) if *inner == DataType::Categorical(None) => {
use polars_arrow::kernels::concatenate::concatenate_owned_unchecked;
let array = concatenate_owned_unchecked(chunks).unwrap();
let list_arr = array.as_any().downcast_ref::<ListArray<i64>>().unwrap();
let values_arr = list_arr.values();
let cat = unsafe {
Series::try_from_arrow_unchecked(
"",
vec![values_arr.clone()],
values_arr.data_type(),
)
.unwrap()
};
// we nest only the physical representation
// the mapping is still in our rev-map
let arrow_dtype = ListArray::<i64>::default_datatype(ArrowDataType::UInt32);
let new_array = ListArray::new(
arrow_dtype,
list_arr.offsets().clone(),
cat.array_ref(0).clone(),
list_arr.validity().cloned(),
);
chunks.clear();
chunks.push(Box::new(new_array));
DataType::List(Box::new(cat.dtype().clone()))
}
_ => dtype,
}
}Additional examples can be found in:
sourcepub fn to_arrow(&self, chunk_idx: usize) -> ArrayRef
pub fn to_arrow(&self, chunk_idx: usize) -> ArrayRef
Convert a chunk in the Series to the correct Arrow type. This conversion is needed because polars doesn’t use a 1 on 1 mapping for logical/ categoricals, etc.
Examples found in repository?
More examples
src/series/into.rs (line 33)
21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76
pub fn to_arrow(&self, chunk_idx: usize) -> ArrayRef {
match self.dtype() {
// special list branch to
// make sure that we recursively apply all logical types.
DataType::List(inner) => {
let ca = self.list().unwrap();
let arr = ca.chunks[chunk_idx].clone();
let arr = arr.as_any().downcast_ref::<ListArray<i64>>().unwrap();
let s = unsafe {
Series::from_chunks_and_dtype_unchecked("", vec![arr.values().clone()], inner)
};
let new_values = s.to_arrow(0);
let data_type = ListArray::<i64>::default_datatype(inner.to_arrow());
let arr = ListArray::<i64>::new(
data_type,
arr.offsets().clone(),
new_values,
arr.validity().cloned(),
);
Box::new(arr)
}
#[cfg(feature = "dtype-categorical")]
DataType::Categorical(_) => {
let ca = self.categorical().unwrap();
let arr = ca.logical().chunks()[chunk_idx].clone();
let cats = UInt32Chunked::from_chunks("", vec![arr]);
// safety:
// we only take a single chunk and change nothing about the index/rev_map mapping
let new = unsafe {
CategoricalChunked::from_cats_and_rev_map_unchecked(
cats,
ca.get_rev_map().clone(),
)
};
let arr: DictionaryArray<u32> = (&new).into();
Box::new(arr) as ArrayRef
}
#[cfg(feature = "dtype-date")]
DataType::Date => cast(&*self.chunks()[chunk_idx], &DataType::Date.to_arrow()).unwrap(),
#[cfg(feature = "dtype-datetime")]
DataType::Datetime(_, _) => {
cast(&*self.chunks()[chunk_idx], &self.dtype().to_arrow()).unwrap()
}
#[cfg(feature = "dtype-duration")]
DataType::Duration(_) => {
cast(&*self.chunks()[chunk_idx], &self.dtype().to_arrow()).unwrap()
}
#[cfg(feature = "dtype-time")]
DataType::Time => cast(&*self.chunks()[chunk_idx], &DataType::Time.to_arrow()).unwrap(),
_ => self.array_ref(chunk_idx).clone(),
}
}source§impl Series
impl Series
sourcepub fn iter(&self) -> SeriesIter<'_> ⓘ
pub fn iter(&self) -> SeriesIter<'_> ⓘ
Examples found in repository?
src/frame/row.rs (line 161)
125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178
pub(crate) fn transpose_from_dtype(&self, dtype: &DataType) -> PolarsResult<DataFrame> {
let new_width = self.height();
let new_height = self.width();
match dtype {
#[cfg(feature = "dtype-i8")]
DataType::Int8 => numeric_transpose::<Int8Type>(&self.columns),
#[cfg(feature = "dtype-i16")]
DataType::Int16 => numeric_transpose::<Int16Type>(&self.columns),
DataType::Int32 => numeric_transpose::<Int32Type>(&self.columns),
DataType::Int64 => numeric_transpose::<Int64Type>(&self.columns),
#[cfg(feature = "dtype-u8")]
DataType::UInt8 => numeric_transpose::<UInt8Type>(&self.columns),
#[cfg(feature = "dtype-u16")]
DataType::UInt16 => numeric_transpose::<UInt16Type>(&self.columns),
DataType::UInt32 => numeric_transpose::<UInt32Type>(&self.columns),
DataType::UInt64 => numeric_transpose::<UInt64Type>(&self.columns),
DataType::Float32 => numeric_transpose::<Float32Type>(&self.columns),
DataType::Float64 => numeric_transpose::<Float64Type>(&self.columns),
_ => {
let mut buffers = (0..new_width)
.map(|_| {
let buf: AnyValueBuffer = (dtype, new_height).into();
buf
})
.collect::<Vec<_>>();
let columns = self
.columns
.iter()
.map(|s| s.cast(dtype).unwrap())
.collect::<Vec<_>>();
// this is very expensive. A lot of cache misses here.
// This is the part that is performance critical.
columns.iter().for_each(|s| {
s.iter().zip(buffers.iter_mut()).for_each(|(av, buf)| {
let _out = buf.add(av);
debug_assert!(_out.is_some());
});
});
let cols = buffers
.into_iter()
.enumerate()
.map(|(i, buf)| {
let mut s = buf.into_series();
s.rename(&format!("column_{i}"));
s
})
.collect::<Vec<_>>();
Ok(DataFrame::new_no_checks(cols))
}
}
}More examples
src/series/iterator.rs (line 139)
81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142
pub fn phys_iter(&self) -> SeriesPhysIter<'_> {
let dtype = self.dtype();
let phys_dtype = dtype.to_physical();
assert_eq!(dtype, &phys_dtype, "impl error");
assert_eq!(self.chunks().len(), 1, "impl error");
let arr = &*self.chunks()[0];
if phys_dtype.is_numeric() {
if arr.null_count() == 0 {
with_match_physical_numeric_type!(phys_dtype, |$T| {
let arr = arr.as_any().downcast_ref::<PrimitiveArray<$T>>().unwrap();
let values = arr.values().as_slice();
Box::new(values.iter().map(|&value| AnyValue::from(value))) as Box<dyn ExactSizeIterator<Item=AnyValue<'_>> + '_>
})
} else {
with_match_physical_numeric_type!(phys_dtype, |$T| {
let arr = arr.as_any().downcast_ref::<PrimitiveArray<$T>>().unwrap();
Box::new(arr.iter().map(|value| {
match value {
Some(value) => AnyValue::from(*value),
None => AnyValue::Null
}
})) as Box<dyn ExactSizeIterator<Item=AnyValue<'_>> + '_>
})
}
} else {
match dtype {
DataType::Utf8 => {
let arr = arr.as_any().downcast_ref::<Utf8Array<i64>>().unwrap();
if arr.null_count() == 0 {
Box::new(arr.values_iter().map(AnyValue::Utf8))
as Box<dyn ExactSizeIterator<Item = AnyValue<'_>> + '_>
} else {
let zipvalid = arr.iter();
Box::new(zipvalid.unwrap_optional().map(|v| match v {
Some(value) => AnyValue::Utf8(value),
None => AnyValue::Null,
}))
as Box<dyn ExactSizeIterator<Item = AnyValue<'_>> + '_>
}
}
DataType::Boolean => {
let arr = arr.as_any().downcast_ref::<BooleanArray>().unwrap();
if arr.null_count() == 0 {
Box::new(arr.values_iter().map(AnyValue::Boolean))
as Box<dyn ExactSizeIterator<Item = AnyValue<'_>> + '_>
} else {
let zipvalid = arr.iter();
Box::new(zipvalid.unwrap_optional().map(|v| match v {
Some(value) => AnyValue::Boolean(value),
None => AnyValue::Null,
}))
as Box<dyn ExactSizeIterator<Item = AnyValue<'_>> + '_>
}
}
_ => Box::new(self.iter()),
}
}
}pub fn phys_iter(&self) -> Box<dyn ExactSizeIterator<Item = AnyValue<'_>> + '_>
source§impl Series
impl Series
pub fn diff(&self, n: usize, null_behavior: NullBehavior) -> Series
Available on crate feature
diff only.source§impl Series
impl Series
sourcepub fn i8(&self) -> PolarsResult<&Int8Chunked>
pub fn i8(&self) -> PolarsResult<&Int8Chunked>
Unpack to ChunkedArray of dtype i8
Examples found in repository?
src/fmt.rs (line 227)
206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284
fn fmt(&self, f: &mut Formatter<'_>) -> fmt::Result {
match self.dtype() {
DataType::Boolean => {
format_array!(f, self.bool().unwrap(), "bool", self.name(), "Series")
}
DataType::Utf8 => {
format_array!(f, self.utf8().unwrap(), "str", self.name(), "Series")
}
DataType::UInt8 => {
format_array!(f, self.u8().unwrap(), "u8", self.name(), "Series")
}
DataType::UInt16 => {
format_array!(f, self.u16().unwrap(), "u16", self.name(), "Series")
}
DataType::UInt32 => {
format_array!(f, self.u32().unwrap(), "u32", self.name(), "Series")
}
DataType::UInt64 => {
format_array!(f, self.u64().unwrap(), "u64", self.name(), "Series")
}
DataType::Int8 => {
format_array!(f, self.i8().unwrap(), "i8", self.name(), "Series")
}
DataType::Int16 => {
format_array!(f, self.i16().unwrap(), "i16", self.name(), "Series")
}
DataType::Int32 => {
format_array!(f, self.i32().unwrap(), "i32", self.name(), "Series")
}
DataType::Int64 => {
format_array!(f, self.i64().unwrap(), "i64", self.name(), "Series")
}
DataType::Float32 => {
format_array!(f, self.f32().unwrap(), "f32", self.name(), "Series")
}
DataType::Float64 => {
format_array!(f, self.f64().unwrap(), "f64", self.name(), "Series")
}
#[cfg(feature = "dtype-date")]
DataType::Date => format_array!(f, self.date().unwrap(), "date", self.name(), "Series"),
#[cfg(feature = "dtype-datetime")]
DataType::Datetime(_, _) => {
let dt = format!("{}", self.dtype());
format_array!(f, self.datetime().unwrap(), &dt, self.name(), "Series")
}
#[cfg(feature = "dtype-time")]
DataType::Time => format_array!(f, self.time().unwrap(), "time", self.name(), "Series"),
#[cfg(feature = "dtype-duration")]
DataType::Duration(_) => {
let dt = format!("{}", self.dtype());
format_array!(f, self.duration().unwrap(), &dt, self.name(), "Series")
}
DataType::List(_) => {
format_array!(f, self.list().unwrap(), "list", self.name(), "Series")
}
#[cfg(feature = "object")]
DataType::Object(_) => format_object_array(f, self, self.name(), "Series"),
#[cfg(feature = "dtype-categorical")]
DataType::Categorical(_) => {
format_array!(f, self.categorical().unwrap(), "cat", self.name(), "Series")
}
#[cfg(feature = "dtype-struct")]
dt @ DataType::Struct(_) => format_array!(
f,
self.struct_().unwrap(),
format!("{dt}"),
self.name(),
"Series"
),
DataType::Null => {
writeln!(f, "nullarray")
}
#[cfg(feature = "dtype-binary")]
DataType::Binary => {
format_array!(f, self.binary().unwrap(), "binary", self.name(), "Series")
}
dt => panic!("{dt:?} not impl"),
}
}sourcepub fn i16(&self) -> PolarsResult<&Int16Chunked>
pub fn i16(&self) -> PolarsResult<&Int16Chunked>
Unpack to ChunkedArray i16
Examples found in repository?
src/fmt.rs (line 230)
206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284
fn fmt(&self, f: &mut Formatter<'_>) -> fmt::Result {
match self.dtype() {
DataType::Boolean => {
format_array!(f, self.bool().unwrap(), "bool", self.name(), "Series")
}
DataType::Utf8 => {
format_array!(f, self.utf8().unwrap(), "str", self.name(), "Series")
}
DataType::UInt8 => {
format_array!(f, self.u8().unwrap(), "u8", self.name(), "Series")
}
DataType::UInt16 => {
format_array!(f, self.u16().unwrap(), "u16", self.name(), "Series")
}
DataType::UInt32 => {
format_array!(f, self.u32().unwrap(), "u32", self.name(), "Series")
}
DataType::UInt64 => {
format_array!(f, self.u64().unwrap(), "u64", self.name(), "Series")
}
DataType::Int8 => {
format_array!(f, self.i8().unwrap(), "i8", self.name(), "Series")
}
DataType::Int16 => {
format_array!(f, self.i16().unwrap(), "i16", self.name(), "Series")
}
DataType::Int32 => {
format_array!(f, self.i32().unwrap(), "i32", self.name(), "Series")
}
DataType::Int64 => {
format_array!(f, self.i64().unwrap(), "i64", self.name(), "Series")
}
DataType::Float32 => {
format_array!(f, self.f32().unwrap(), "f32", self.name(), "Series")
}
DataType::Float64 => {
format_array!(f, self.f64().unwrap(), "f64", self.name(), "Series")
}
#[cfg(feature = "dtype-date")]
DataType::Date => format_array!(f, self.date().unwrap(), "date", self.name(), "Series"),
#[cfg(feature = "dtype-datetime")]
DataType::Datetime(_, _) => {
let dt = format!("{}", self.dtype());
format_array!(f, self.datetime().unwrap(), &dt, self.name(), "Series")
}
#[cfg(feature = "dtype-time")]
DataType::Time => format_array!(f, self.time().unwrap(), "time", self.name(), "Series"),
#[cfg(feature = "dtype-duration")]
DataType::Duration(_) => {
let dt = format!("{}", self.dtype());
format_array!(f, self.duration().unwrap(), &dt, self.name(), "Series")
}
DataType::List(_) => {
format_array!(f, self.list().unwrap(), "list", self.name(), "Series")
}
#[cfg(feature = "object")]
DataType::Object(_) => format_object_array(f, self, self.name(), "Series"),
#[cfg(feature = "dtype-categorical")]
DataType::Categorical(_) => {
format_array!(f, self.categorical().unwrap(), "cat", self.name(), "Series")
}
#[cfg(feature = "dtype-struct")]
dt @ DataType::Struct(_) => format_array!(
f,
self.struct_().unwrap(),
format!("{dt}"),
self.name(),
"Series"
),
DataType::Null => {
writeln!(f, "nullarray")
}
#[cfg(feature = "dtype-binary")]
DataType::Binary => {
format_array!(f, self.binary().unwrap(), "binary", self.name(), "Series")
}
dt => panic!("{dt:?} not impl"),
}
}sourcepub fn i32(&self) -> PolarsResult<&Int32Chunked>
pub fn i32(&self) -> PolarsResult<&Int32Chunked>
Unpack to ChunkedArray
let s = Series::new("foo", [1i32 ,2, 3]);
let s_squared: Series = s.i32()
.unwrap()
.into_iter()
.map(|opt_v| {
match opt_v {
Some(v) => Some(v * v),
None => None, // null value
}
}).collect();Examples found in repository?
src/series/mod.rs (line 576)
565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751 752 753 754 755 756 757 758 759 760 761 762 763 764 765 766 767 768 769 770 771 772 773 774 775 776 777 778 779 780 781 782 783 784 785 786 787 788 789 790 791 792 793 794 795 796 797 798 799 800 801 802 803 804 805 806 807 808 809 810 811 812 813 814 815 816 817 818 819 820 821 822
pub fn cumsum(&self, reverse: bool) -> Series {
#[cfg(feature = "cum_agg")]
{
use DataType::*;
match self.dtype() {
Boolean => self.cast(&DataType::UInt32).unwrap().cumsum(reverse),
Int8 | UInt8 | Int16 | UInt16 => {
let s = self.cast(&Int64).unwrap();
s.cumsum(reverse)
}
Int32 => {
let ca = self.i32().unwrap();
ca.cumsum(reverse).into_series()
}
UInt32 => {
let ca = self.u32().unwrap();
ca.cumsum(reverse).into_series()
}
UInt64 => {
let ca = self.u64().unwrap();
ca.cumsum(reverse).into_series()
}
Int64 => {
let ca = self.i64().unwrap();
ca.cumsum(reverse).into_series()
}
Float32 => {
let ca = self.f32().unwrap();
ca.cumsum(reverse).into_series()
}
Float64 => {
let ca = self.f64().unwrap();
ca.cumsum(reverse).into_series()
}
dt => panic!("cumsum not supported for dtype: {dt:?}"),
}
}
#[cfg(not(feature = "cum_agg"))]
{
panic!("activate 'cum_agg' feature")
}
}
/// Get an array with the cumulative product computed at every element
///
/// If the [`DataType`] is one of `{Int8, UInt8, Int16, UInt16, Int32, UInt32}` the `Series` is
/// first cast to `Int64` to prevent overflow issues.
#[cfg_attr(docsrs, doc(cfg(feature = "cum_agg")))]
#[allow(unused_variables)]
pub fn cumprod(&self, reverse: bool) -> Series {
#[cfg(feature = "cum_agg")]
{
use DataType::*;
match self.dtype() {
Boolean => self.cast(&DataType::Int64).unwrap().cumprod(reverse),
Int8 | UInt8 | Int16 | UInt16 | Int32 | UInt32 => {
let s = self.cast(&Int64).unwrap();
s.cumprod(reverse)
}
Int64 => {
let ca = self.i64().unwrap();
ca.cumprod(reverse).into_series()
}
UInt64 => {
let ca = self.u64().unwrap();
ca.cumprod(reverse).into_series()
}
Float32 => {
let ca = self.f32().unwrap();
ca.cumprod(reverse).into_series()
}
Float64 => {
let ca = self.f64().unwrap();
ca.cumprod(reverse).into_series()
}
dt => panic!("cumprod not supported for dtype: {dt:?}"),
}
}
#[cfg(not(feature = "cum_agg"))]
{
panic!("activate 'cum_agg' feature")
}
}
/// Get the product of an array.
///
/// If the [`DataType`] is one of `{Int8, UInt8, Int16, UInt16}` the `Series` is
/// first cast to `Int64` to prevent overflow issues.
#[cfg_attr(docsrs, doc(cfg(feature = "product")))]
pub fn product(&self) -> Series {
#[cfg(feature = "product")]
{
use DataType::*;
match self.dtype() {
Boolean => self.cast(&DataType::Int64).unwrap().product(),
Int8 | UInt8 | Int16 | UInt16 => {
let s = self.cast(&Int64).unwrap();
s.product()
}
Int64 => {
let ca = self.i64().unwrap();
ca.prod_as_series()
}
Float32 => {
let ca = self.f32().unwrap();
ca.prod_as_series()
}
Float64 => {
let ca = self.f64().unwrap();
ca.prod_as_series()
}
dt => panic!("cumprod not supported for dtype: {dt:?}"),
}
}
#[cfg(not(feature = "product"))]
{
panic!("activate 'product' feature")
}
}
#[cfg(feature = "rank")]
#[cfg_attr(docsrs, doc(cfg(feature = "rank")))]
pub fn rank(&self, options: RankOptions) -> Series {
rank(self, options.method, options.descending)
}
/// Cast throws an error if conversion had overflows
pub fn strict_cast(&self, data_type: &DataType) -> PolarsResult<Series> {
let s = self.cast(data_type)?;
if self.null_count() != s.null_count() {
let failure_mask = !self.is_null() & s.is_null();
let failures = self.filter_threaded(&failure_mask, false)?.unique()?;
Err(PolarsError::ComputeError(
format!(
"Strict conversion from {:?} to {:?} failed for values {}. \
If you were trying to cast Utf8 to Date, Time, or Datetime, \
consider using `strptime`.",
self.dtype(),
data_type,
failures.fmt_list(),
)
.into(),
))
} else {
Ok(s)
}
}
#[cfg(feature = "dtype-time")]
pub(crate) fn into_time(self) -> Series {
#[cfg(not(feature = "dtype-time"))]
{
panic!("activate feature dtype-time")
}
match self.dtype() {
DataType::Int64 => self.i64().unwrap().clone().into_time().into_series(),
DataType::Time => self
.time()
.unwrap()
.as_ref()
.clone()
.into_time()
.into_series(),
dt => panic!("date not implemented for {dt:?}"),
}
}
pub(crate) fn into_date(self) -> Series {
#[cfg(not(feature = "dtype-date"))]
{
panic!("activate feature dtype-date")
}
#[cfg(feature = "dtype-date")]
match self.dtype() {
DataType::Int32 => self.i32().unwrap().clone().into_date().into_series(),
DataType::Date => self
.date()
.unwrap()
.as_ref()
.clone()
.into_date()
.into_series(),
dt => panic!("date not implemented for {dt:?}"),
}
}
pub(crate) fn into_datetime(self, timeunit: TimeUnit, tz: Option<TimeZone>) -> Series {
#[cfg(not(feature = "dtype-datetime"))]
{
panic!("activate feature dtype-datetime")
}
#[cfg(feature = "dtype-datetime")]
match self.dtype() {
DataType::Int64 => self
.i64()
.unwrap()
.clone()
.into_datetime(timeunit, tz)
.into_series(),
DataType::Datetime(_, _) => self
.datetime()
.unwrap()
.as_ref()
.clone()
.into_datetime(timeunit, tz)
.into_series(),
dt => panic!("into_datetime not implemented for {dt:?}"),
}
}
pub(crate) fn into_duration(self, timeunit: TimeUnit) -> Series {
#[cfg(not(feature = "dtype-duration"))]
{
panic!("activate feature dtype-duration")
}
#[cfg(feature = "dtype-duration")]
match self.dtype() {
DataType::Int64 => self
.i64()
.unwrap()
.clone()
.into_duration(timeunit)
.into_series(),
DataType::Duration(_) => self
.duration()
.unwrap()
.as_ref()
.clone()
.into_duration(timeunit)
.into_series(),
dt => panic!("into_duration not implemented for {dt:?}"),
}
}
#[cfg(feature = "abs")]
#[cfg_attr(docsrs, doc(cfg(feature = "abs")))]
/// convert numerical values to their absolute value
pub fn abs(&self) -> PolarsResult<Series> {
let a = self.to_physical_repr();
use DataType::*;
let out = match a.dtype() {
#[cfg(feature = "dtype-i8")]
Int8 => a.i8().unwrap().abs().into_series(),
#[cfg(feature = "dtype-i16")]
Int16 => a.i16().unwrap().abs().into_series(),
Int32 => a.i32().unwrap().abs().into_series(),
Int64 => a.i64().unwrap().abs().into_series(),
UInt8 | UInt16 | UInt32 | UInt64 => self.clone(),
Float32 => a.f32().unwrap().abs().into_series(),
Float64 => a.f64().unwrap().abs().into_series(),
dt => {
return Err(PolarsError::InvalidOperation(
format!("abs not supported for series of type {dt:?}").into(),
));
}
};
Ok(out)
}More examples
src/frame/asof_join/mod.rs (line 135)
112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195
pub fn _join_asof(
&self,
other: &DataFrame,
left_on: &str,
right_on: &str,
strategy: AsofStrategy,
tolerance: Option<AnyValue<'static>>,
suffix: Option<String>,
slice: Option<(i64, usize)>,
) -> PolarsResult<DataFrame> {
let left_key = self.column(left_on)?;
let right_key = other.column(right_on)?;
check_asof_columns(left_key, right_key)?;
let left_key = left_key.to_physical_repr();
let right_key = right_key.to_physical_repr();
let take_idx = match left_key.dtype() {
DataType::Int64 => left_key
.i64()
.unwrap()
.join_asof(&right_key, strategy, tolerance),
DataType::Int32 => left_key
.i32()
.unwrap()
.join_asof(&right_key, strategy, tolerance),
DataType::UInt64 => left_key
.u64()
.unwrap()
.join_asof(&right_key, strategy, tolerance),
DataType::UInt32 => left_key
.u32()
.unwrap()
.join_asof(&right_key, strategy, tolerance),
DataType::Float32 => left_key
.f32()
.unwrap()
.join_asof(&right_key, strategy, tolerance),
DataType::Float64 => left_key
.f64()
.unwrap()
.join_asof(&right_key, strategy, tolerance),
_ => {
let left_key = left_key.cast(&DataType::Int32).unwrap();
let right_key = right_key.cast(&DataType::Int32).unwrap();
left_key
.i32()
.unwrap()
.join_asof(&right_key, strategy, tolerance)
}
}?;
// take_idx are sorted so this is a bound check for all
if let Some(Some(idx)) = take_idx.last() {
assert!((*idx as usize) < other.height())
}
// drop right join column
let other = if left_on == right_on {
Cow::Owned(other.drop(right_on)?)
} else {
Cow::Borrowed(other)
};
let mut left = self.clone();
let mut take_idx = &*take_idx;
if let Some((offset, len)) = slice {
left = left.slice(offset, len);
take_idx = slice_slice(take_idx, offset, len);
}
// Safety:
// join tuples are in bounds
let right_df = unsafe {
other.take_opt_iter_unchecked(
take_idx
.iter()
.map(|opt_idx| opt_idx.map(|idx| idx as usize)),
)
};
_finish_join(left, right_df, suffix.as_deref())
}src/series/arithmetic/borrowed.rs (line 255)
220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300
fn checked_div_num<T: ToPrimitive>(&self, rhs: T) -> PolarsResult<Series> {
use DataType::*;
let s = self.to_physical_repr();
let out = match s.dtype() {
#[cfg(feature = "dtype-u8")]
UInt8 => s
.u8()
.unwrap()
.apply_on_opt(|opt_v| opt_v.and_then(|v| v.checked_div(rhs.to_u8().unwrap())))
.into_series(),
#[cfg(feature = "dtype-i8")]
Int8 => s
.i8()
.unwrap()
.apply_on_opt(|opt_v| opt_v.and_then(|v| v.checked_div(rhs.to_i8().unwrap())))
.into_series(),
#[cfg(feature = "dtype-i16")]
Int16 => s
.i16()
.unwrap()
.apply_on_opt(|opt_v| opt_v.and_then(|v| v.checked_div(rhs.to_i16().unwrap())))
.into_series(),
#[cfg(feature = "dtype-u16")]
UInt16 => s
.u16()
.unwrap()
.apply_on_opt(|opt_v| opt_v.and_then(|v| v.checked_div(rhs.to_u16().unwrap())))
.into_series(),
UInt32 => s
.u32()
.unwrap()
.apply_on_opt(|opt_v| opt_v.and_then(|v| v.checked_div(rhs.to_u32().unwrap())))
.into_series(),
Int32 => s
.i32()
.unwrap()
.apply_on_opt(|opt_v| opt_v.and_then(|v| v.checked_div(rhs.to_i32().unwrap())))
.into_series(),
UInt64 => s
.u64()
.unwrap()
.apply_on_opt(|opt_v| opt_v.and_then(|v| v.checked_div(rhs.to_u64().unwrap())))
.into_series(),
Int64 => s
.i64()
.unwrap()
.apply_on_opt(|opt_v| opt_v.and_then(|v| v.checked_div(rhs.to_i64().unwrap())))
.into_series(),
Float32 => s
.f32()
.unwrap()
.apply_on_opt(|opt_v| {
opt_v.and_then(|v| {
let res = rhs.to_f32().unwrap();
if res.is_zero() {
None
} else {
Some(v / res)
}
})
})
.into_series(),
Float64 => s
.f64()
.unwrap()
.apply_on_opt(|opt_v| {
opt_v.and_then(|v| {
let res = rhs.to_f64().unwrap();
if res.is_zero() {
None
} else {
Some(v / res)
}
})
})
.into_series(),
_ => panic!("dtype not yet supported in checked div"),
};
out.cast(self.dtype())
}src/fmt.rs (line 233)
206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284
fn fmt(&self, f: &mut Formatter<'_>) -> fmt::Result {
match self.dtype() {
DataType::Boolean => {
format_array!(f, self.bool().unwrap(), "bool", self.name(), "Series")
}
DataType::Utf8 => {
format_array!(f, self.utf8().unwrap(), "str", self.name(), "Series")
}
DataType::UInt8 => {
format_array!(f, self.u8().unwrap(), "u8", self.name(), "Series")
}
DataType::UInt16 => {
format_array!(f, self.u16().unwrap(), "u16", self.name(), "Series")
}
DataType::UInt32 => {
format_array!(f, self.u32().unwrap(), "u32", self.name(), "Series")
}
DataType::UInt64 => {
format_array!(f, self.u64().unwrap(), "u64", self.name(), "Series")
}
DataType::Int8 => {
format_array!(f, self.i8().unwrap(), "i8", self.name(), "Series")
}
DataType::Int16 => {
format_array!(f, self.i16().unwrap(), "i16", self.name(), "Series")
}
DataType::Int32 => {
format_array!(f, self.i32().unwrap(), "i32", self.name(), "Series")
}
DataType::Int64 => {
format_array!(f, self.i64().unwrap(), "i64", self.name(), "Series")
}
DataType::Float32 => {
format_array!(f, self.f32().unwrap(), "f32", self.name(), "Series")
}
DataType::Float64 => {
format_array!(f, self.f64().unwrap(), "f64", self.name(), "Series")
}
#[cfg(feature = "dtype-date")]
DataType::Date => format_array!(f, self.date().unwrap(), "date", self.name(), "Series"),
#[cfg(feature = "dtype-datetime")]
DataType::Datetime(_, _) => {
let dt = format!("{}", self.dtype());
format_array!(f, self.datetime().unwrap(), &dt, self.name(), "Series")
}
#[cfg(feature = "dtype-time")]
DataType::Time => format_array!(f, self.time().unwrap(), "time", self.name(), "Series"),
#[cfg(feature = "dtype-duration")]
DataType::Duration(_) => {
let dt = format!("{}", self.dtype());
format_array!(f, self.duration().unwrap(), &dt, self.name(), "Series")
}
DataType::List(_) => {
format_array!(f, self.list().unwrap(), "list", self.name(), "Series")
}
#[cfg(feature = "object")]
DataType::Object(_) => format_object_array(f, self, self.name(), "Series"),
#[cfg(feature = "dtype-categorical")]
DataType::Categorical(_) => {
format_array!(f, self.categorical().unwrap(), "cat", self.name(), "Series")
}
#[cfg(feature = "dtype-struct")]
dt @ DataType::Struct(_) => format_array!(
f,
self.struct_().unwrap(),
format!("{dt}"),
self.name(),
"Series"
),
DataType::Null => {
writeln!(f, "nullarray")
}
#[cfg(feature = "dtype-binary")]
DataType::Binary => {
format_array!(f, self.binary().unwrap(), "binary", self.name(), "Series")
}
dt => panic!("{dt:?} not impl"),
}
}sourcepub fn i64(&self) -> PolarsResult<&Int64Chunked>
pub fn i64(&self) -> PolarsResult<&Int64Chunked>
Unpack to ChunkedArray of dtype i64
Examples found in repository?
src/series/mod.rs (line 588)
565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751 752 753 754 755 756 757 758 759 760 761 762 763 764 765 766 767 768 769 770 771 772 773 774 775 776 777 778 779 780 781 782 783 784 785 786 787 788 789 790 791 792 793 794 795 796 797 798 799 800 801 802 803 804 805 806 807 808 809 810 811 812 813 814 815 816 817 818 819 820 821 822
pub fn cumsum(&self, reverse: bool) -> Series {
#[cfg(feature = "cum_agg")]
{
use DataType::*;
match self.dtype() {
Boolean => self.cast(&DataType::UInt32).unwrap().cumsum(reverse),
Int8 | UInt8 | Int16 | UInt16 => {
let s = self.cast(&Int64).unwrap();
s.cumsum(reverse)
}
Int32 => {
let ca = self.i32().unwrap();
ca.cumsum(reverse).into_series()
}
UInt32 => {
let ca = self.u32().unwrap();
ca.cumsum(reverse).into_series()
}
UInt64 => {
let ca = self.u64().unwrap();
ca.cumsum(reverse).into_series()
}
Int64 => {
let ca = self.i64().unwrap();
ca.cumsum(reverse).into_series()
}
Float32 => {
let ca = self.f32().unwrap();
ca.cumsum(reverse).into_series()
}
Float64 => {
let ca = self.f64().unwrap();
ca.cumsum(reverse).into_series()
}
dt => panic!("cumsum not supported for dtype: {dt:?}"),
}
}
#[cfg(not(feature = "cum_agg"))]
{
panic!("activate 'cum_agg' feature")
}
}
/// Get an array with the cumulative product computed at every element
///
/// If the [`DataType`] is one of `{Int8, UInt8, Int16, UInt16, Int32, UInt32}` the `Series` is
/// first cast to `Int64` to prevent overflow issues.
#[cfg_attr(docsrs, doc(cfg(feature = "cum_agg")))]
#[allow(unused_variables)]
pub fn cumprod(&self, reverse: bool) -> Series {
#[cfg(feature = "cum_agg")]
{
use DataType::*;
match self.dtype() {
Boolean => self.cast(&DataType::Int64).unwrap().cumprod(reverse),
Int8 | UInt8 | Int16 | UInt16 | Int32 | UInt32 => {
let s = self.cast(&Int64).unwrap();
s.cumprod(reverse)
}
Int64 => {
let ca = self.i64().unwrap();
ca.cumprod(reverse).into_series()
}
UInt64 => {
let ca = self.u64().unwrap();
ca.cumprod(reverse).into_series()
}
Float32 => {
let ca = self.f32().unwrap();
ca.cumprod(reverse).into_series()
}
Float64 => {
let ca = self.f64().unwrap();
ca.cumprod(reverse).into_series()
}
dt => panic!("cumprod not supported for dtype: {dt:?}"),
}
}
#[cfg(not(feature = "cum_agg"))]
{
panic!("activate 'cum_agg' feature")
}
}
/// Get the product of an array.
///
/// If the [`DataType`] is one of `{Int8, UInt8, Int16, UInt16}` the `Series` is
/// first cast to `Int64` to prevent overflow issues.
#[cfg_attr(docsrs, doc(cfg(feature = "product")))]
pub fn product(&self) -> Series {
#[cfg(feature = "product")]
{
use DataType::*;
match self.dtype() {
Boolean => self.cast(&DataType::Int64).unwrap().product(),
Int8 | UInt8 | Int16 | UInt16 => {
let s = self.cast(&Int64).unwrap();
s.product()
}
Int64 => {
let ca = self.i64().unwrap();
ca.prod_as_series()
}
Float32 => {
let ca = self.f32().unwrap();
ca.prod_as_series()
}
Float64 => {
let ca = self.f64().unwrap();
ca.prod_as_series()
}
dt => panic!("cumprod not supported for dtype: {dt:?}"),
}
}
#[cfg(not(feature = "product"))]
{
panic!("activate 'product' feature")
}
}
#[cfg(feature = "rank")]
#[cfg_attr(docsrs, doc(cfg(feature = "rank")))]
pub fn rank(&self, options: RankOptions) -> Series {
rank(self, options.method, options.descending)
}
/// Cast throws an error if conversion had overflows
pub fn strict_cast(&self, data_type: &DataType) -> PolarsResult<Series> {
let s = self.cast(data_type)?;
if self.null_count() != s.null_count() {
let failure_mask = !self.is_null() & s.is_null();
let failures = self.filter_threaded(&failure_mask, false)?.unique()?;
Err(PolarsError::ComputeError(
format!(
"Strict conversion from {:?} to {:?} failed for values {}. \
If you were trying to cast Utf8 to Date, Time, or Datetime, \
consider using `strptime`.",
self.dtype(),
data_type,
failures.fmt_list(),
)
.into(),
))
} else {
Ok(s)
}
}
#[cfg(feature = "dtype-time")]
pub(crate) fn into_time(self) -> Series {
#[cfg(not(feature = "dtype-time"))]
{
panic!("activate feature dtype-time")
}
match self.dtype() {
DataType::Int64 => self.i64().unwrap().clone().into_time().into_series(),
DataType::Time => self
.time()
.unwrap()
.as_ref()
.clone()
.into_time()
.into_series(),
dt => panic!("date not implemented for {dt:?}"),
}
}
pub(crate) fn into_date(self) -> Series {
#[cfg(not(feature = "dtype-date"))]
{
panic!("activate feature dtype-date")
}
#[cfg(feature = "dtype-date")]
match self.dtype() {
DataType::Int32 => self.i32().unwrap().clone().into_date().into_series(),
DataType::Date => self
.date()
.unwrap()
.as_ref()
.clone()
.into_date()
.into_series(),
dt => panic!("date not implemented for {dt:?}"),
}
}
pub(crate) fn into_datetime(self, timeunit: TimeUnit, tz: Option<TimeZone>) -> Series {
#[cfg(not(feature = "dtype-datetime"))]
{
panic!("activate feature dtype-datetime")
}
#[cfg(feature = "dtype-datetime")]
match self.dtype() {
DataType::Int64 => self
.i64()
.unwrap()
.clone()
.into_datetime(timeunit, tz)
.into_series(),
DataType::Datetime(_, _) => self
.datetime()
.unwrap()
.as_ref()
.clone()
.into_datetime(timeunit, tz)
.into_series(),
dt => panic!("into_datetime not implemented for {dt:?}"),
}
}
pub(crate) fn into_duration(self, timeunit: TimeUnit) -> Series {
#[cfg(not(feature = "dtype-duration"))]
{
panic!("activate feature dtype-duration")
}
#[cfg(feature = "dtype-duration")]
match self.dtype() {
DataType::Int64 => self
.i64()
.unwrap()
.clone()
.into_duration(timeunit)
.into_series(),
DataType::Duration(_) => self
.duration()
.unwrap()
.as_ref()
.clone()
.into_duration(timeunit)
.into_series(),
dt => panic!("into_duration not implemented for {dt:?}"),
}
}
#[cfg(feature = "abs")]
#[cfg_attr(docsrs, doc(cfg(feature = "abs")))]
/// convert numerical values to their absolute value
pub fn abs(&self) -> PolarsResult<Series> {
let a = self.to_physical_repr();
use DataType::*;
let out = match a.dtype() {
#[cfg(feature = "dtype-i8")]
Int8 => a.i8().unwrap().abs().into_series(),
#[cfg(feature = "dtype-i16")]
Int16 => a.i16().unwrap().abs().into_series(),
Int32 => a.i32().unwrap().abs().into_series(),
Int64 => a.i64().unwrap().abs().into_series(),
UInt8 | UInt16 | UInt32 | UInt64 => self.clone(),
Float32 => a.f32().unwrap().abs().into_series(),
Float64 => a.f64().unwrap().abs().into_series(),
dt => {
return Err(PolarsError::InvalidOperation(
format!("abs not supported for series of type {dt:?}").into(),
));
}
};
Ok(out)
}More examples
src/frame/asof_join/mod.rs (line 131)
112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195
pub fn _join_asof(
&self,
other: &DataFrame,
left_on: &str,
right_on: &str,
strategy: AsofStrategy,
tolerance: Option<AnyValue<'static>>,
suffix: Option<String>,
slice: Option<(i64, usize)>,
) -> PolarsResult<DataFrame> {
let left_key = self.column(left_on)?;
let right_key = other.column(right_on)?;
check_asof_columns(left_key, right_key)?;
let left_key = left_key.to_physical_repr();
let right_key = right_key.to_physical_repr();
let take_idx = match left_key.dtype() {
DataType::Int64 => left_key
.i64()
.unwrap()
.join_asof(&right_key, strategy, tolerance),
DataType::Int32 => left_key
.i32()
.unwrap()
.join_asof(&right_key, strategy, tolerance),
DataType::UInt64 => left_key
.u64()
.unwrap()
.join_asof(&right_key, strategy, tolerance),
DataType::UInt32 => left_key
.u32()
.unwrap()
.join_asof(&right_key, strategy, tolerance),
DataType::Float32 => left_key
.f32()
.unwrap()
.join_asof(&right_key, strategy, tolerance),
DataType::Float64 => left_key
.f64()
.unwrap()
.join_asof(&right_key, strategy, tolerance),
_ => {
let left_key = left_key.cast(&DataType::Int32).unwrap();
let right_key = right_key.cast(&DataType::Int32).unwrap();
left_key
.i32()
.unwrap()
.join_asof(&right_key, strategy, tolerance)
}
}?;
// take_idx are sorted so this is a bound check for all
if let Some(Some(idx)) = take_idx.last() {
assert!((*idx as usize) < other.height())
}
// drop right join column
let other = if left_on == right_on {
Cow::Owned(other.drop(right_on)?)
} else {
Cow::Borrowed(other)
};
let mut left = self.clone();
let mut take_idx = &*take_idx;
if let Some((offset, len)) = slice {
left = left.slice(offset, len);
take_idx = slice_slice(take_idx, offset, len);
}
// Safety:
// join tuples are in bounds
let right_df = unsafe {
other.take_opt_iter_unchecked(
take_idx
.iter()
.map(|opt_idx| opt_idx.map(|idx| idx as usize)),
)
};
_finish_join(left, right_df, suffix.as_deref())
}src/series/arithmetic/borrowed.rs (line 265)
220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300
fn checked_div_num<T: ToPrimitive>(&self, rhs: T) -> PolarsResult<Series> {
use DataType::*;
let s = self.to_physical_repr();
let out = match s.dtype() {
#[cfg(feature = "dtype-u8")]
UInt8 => s
.u8()
.unwrap()
.apply_on_opt(|opt_v| opt_v.and_then(|v| v.checked_div(rhs.to_u8().unwrap())))
.into_series(),
#[cfg(feature = "dtype-i8")]
Int8 => s
.i8()
.unwrap()
.apply_on_opt(|opt_v| opt_v.and_then(|v| v.checked_div(rhs.to_i8().unwrap())))
.into_series(),
#[cfg(feature = "dtype-i16")]
Int16 => s
.i16()
.unwrap()
.apply_on_opt(|opt_v| opt_v.and_then(|v| v.checked_div(rhs.to_i16().unwrap())))
.into_series(),
#[cfg(feature = "dtype-u16")]
UInt16 => s
.u16()
.unwrap()
.apply_on_opt(|opt_v| opt_v.and_then(|v| v.checked_div(rhs.to_u16().unwrap())))
.into_series(),
UInt32 => s
.u32()
.unwrap()
.apply_on_opt(|opt_v| opt_v.and_then(|v| v.checked_div(rhs.to_u32().unwrap())))
.into_series(),
Int32 => s
.i32()
.unwrap()
.apply_on_opt(|opt_v| opt_v.and_then(|v| v.checked_div(rhs.to_i32().unwrap())))
.into_series(),
UInt64 => s
.u64()
.unwrap()
.apply_on_opt(|opt_v| opt_v.and_then(|v| v.checked_div(rhs.to_u64().unwrap())))
.into_series(),
Int64 => s
.i64()
.unwrap()
.apply_on_opt(|opt_v| opt_v.and_then(|v| v.checked_div(rhs.to_i64().unwrap())))
.into_series(),
Float32 => s
.f32()
.unwrap()
.apply_on_opt(|opt_v| {
opt_v.and_then(|v| {
let res = rhs.to_f32().unwrap();
if res.is_zero() {
None
} else {
Some(v / res)
}
})
})
.into_series(),
Float64 => s
.f64()
.unwrap()
.apply_on_opt(|opt_v| {
opt_v.and_then(|v| {
let res = rhs.to_f64().unwrap();
if res.is_zero() {
None
} else {
Some(v / res)
}
})
})
.into_series(),
_ => panic!("dtype not yet supported in checked div"),
};
out.cast(self.dtype())
}src/fmt.rs (line 236)
206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284
fn fmt(&self, f: &mut Formatter<'_>) -> fmt::Result {
match self.dtype() {
DataType::Boolean => {
format_array!(f, self.bool().unwrap(), "bool", self.name(), "Series")
}
DataType::Utf8 => {
format_array!(f, self.utf8().unwrap(), "str", self.name(), "Series")
}
DataType::UInt8 => {
format_array!(f, self.u8().unwrap(), "u8", self.name(), "Series")
}
DataType::UInt16 => {
format_array!(f, self.u16().unwrap(), "u16", self.name(), "Series")
}
DataType::UInt32 => {
format_array!(f, self.u32().unwrap(), "u32", self.name(), "Series")
}
DataType::UInt64 => {
format_array!(f, self.u64().unwrap(), "u64", self.name(), "Series")
}
DataType::Int8 => {
format_array!(f, self.i8().unwrap(), "i8", self.name(), "Series")
}
DataType::Int16 => {
format_array!(f, self.i16().unwrap(), "i16", self.name(), "Series")
}
DataType::Int32 => {
format_array!(f, self.i32().unwrap(), "i32", self.name(), "Series")
}
DataType::Int64 => {
format_array!(f, self.i64().unwrap(), "i64", self.name(), "Series")
}
DataType::Float32 => {
format_array!(f, self.f32().unwrap(), "f32", self.name(), "Series")
}
DataType::Float64 => {
format_array!(f, self.f64().unwrap(), "f64", self.name(), "Series")
}
#[cfg(feature = "dtype-date")]
DataType::Date => format_array!(f, self.date().unwrap(), "date", self.name(), "Series"),
#[cfg(feature = "dtype-datetime")]
DataType::Datetime(_, _) => {
let dt = format!("{}", self.dtype());
format_array!(f, self.datetime().unwrap(), &dt, self.name(), "Series")
}
#[cfg(feature = "dtype-time")]
DataType::Time => format_array!(f, self.time().unwrap(), "time", self.name(), "Series"),
#[cfg(feature = "dtype-duration")]
DataType::Duration(_) => {
let dt = format!("{}", self.dtype());
format_array!(f, self.duration().unwrap(), &dt, self.name(), "Series")
}
DataType::List(_) => {
format_array!(f, self.list().unwrap(), "list", self.name(), "Series")
}
#[cfg(feature = "object")]
DataType::Object(_) => format_object_array(f, self, self.name(), "Series"),
#[cfg(feature = "dtype-categorical")]
DataType::Categorical(_) => {
format_array!(f, self.categorical().unwrap(), "cat", self.name(), "Series")
}
#[cfg(feature = "dtype-struct")]
dt @ DataType::Struct(_) => format_array!(
f,
self.struct_().unwrap(),
format!("{dt}"),
self.name(),
"Series"
),
DataType::Null => {
writeln!(f, "nullarray")
}
#[cfg(feature = "dtype-binary")]
DataType::Binary => {
format_array!(f, self.binary().unwrap(), "binary", self.name(), "Series")
}
dt => panic!("{dt:?} not impl"),
}
}sourcepub fn f32(&self) -> PolarsResult<&Float32Chunked>
pub fn f32(&self) -> PolarsResult<&Float32Chunked>
Unpack to ChunkedArray of dtype f32
Examples found in repository?
src/series/mod.rs (line 308)
306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751 752 753 754 755 756 757 758 759 760 761 762 763 764 765 766 767 768 769 770 771 772 773 774 775 776 777 778 779 780 781 782 783 784 785 786 787 788 789 790 791 792 793 794 795 796 797 798 799 800 801 802 803 804 805 806 807 808 809 810 811 812 813 814 815 816 817 818 819 820 821 822
pub fn is_nan(&self) -> PolarsResult<BooleanChunked> {
match self.dtype() {
DataType::Float32 => Ok(self.f32().unwrap().is_nan()),
DataType::Float64 => Ok(self.f64().unwrap().is_nan()),
_ => Err(PolarsError::InvalidOperation(
format!(
"'is_nan' not supported for series with dtype {:?}",
self.dtype()
)
.into(),
)),
}
}
/// Check if float value is NaN (note this is different than missing/ null)
pub fn is_not_nan(&self) -> PolarsResult<BooleanChunked> {
match self.dtype() {
DataType::Float32 => Ok(self.f32().unwrap().is_not_nan()),
DataType::Float64 => Ok(self.f64().unwrap().is_not_nan()),
_ => Err(PolarsError::InvalidOperation(
format!(
"'is_not_nan' not supported for series with dtype {:?}",
self.dtype()
)
.into(),
)),
}
}
/// Check if float value is finite
pub fn is_finite(&self) -> PolarsResult<BooleanChunked> {
match self.dtype() {
DataType::Float32 => Ok(self.f32().unwrap().is_finite()),
DataType::Float64 => Ok(self.f64().unwrap().is_finite()),
_ => Err(PolarsError::InvalidOperation(
format!(
"'is_finite' not supported for series with dtype {:?}",
self.dtype()
)
.into(),
)),
}
}
/// Check if float value is infinite
pub fn is_infinite(&self) -> PolarsResult<BooleanChunked> {
match self.dtype() {
DataType::Float32 => Ok(self.f32().unwrap().is_infinite()),
DataType::Float64 => Ok(self.f64().unwrap().is_infinite()),
_ => Err(PolarsError::InvalidOperation(
format!(
"'is_infinite' not supported for series with dtype {:?}",
self.dtype()
)
.into(),
)),
}
}
/// Create a new ChunkedArray with values from self where the mask evaluates `true` and values
/// from `other` where the mask evaluates `false`
#[cfg(feature = "zip_with")]
#[cfg_attr(docsrs, doc(cfg(feature = "zip_with")))]
pub fn zip_with(&self, mask: &BooleanChunked, other: &Series) -> PolarsResult<Series> {
let (lhs, rhs) = coerce_lhs_rhs(self, other)?;
lhs.zip_with_same_type(mask, rhs.as_ref())
}
/// Cast a datelike Series to their physical representation.
/// Primitives remain unchanged
///
/// * Date -> Int32
/// * Datetime-> Int64
/// * Time -> Int64
/// * Categorical -> UInt32
///
pub fn to_physical_repr(&self) -> Cow<Series> {
use DataType::*;
match self.dtype() {
Date => Cow::Owned(self.cast(&DataType::Int32).unwrap()),
Datetime(_, _) | Duration(_) | Time => Cow::Owned(self.cast(&DataType::Int64).unwrap()),
#[cfg(feature = "dtype-categorical")]
Categorical(_) => Cow::Owned(self.cast(&DataType::UInt32).unwrap()),
_ => Cow::Borrowed(self),
}
}
fn finish_take_threaded(&self, s: Vec<Series>, rechunk: bool) -> Series {
let s = s
.into_iter()
.reduce(|mut s, s1| {
s.append(&s1).unwrap();
s
})
.unwrap();
if rechunk {
s.rechunk()
} else {
s
}
}
// take a function pointer to reduce bloat
fn threaded_op(
&self,
rechunk: bool,
len: usize,
func: &(dyn Fn(usize, usize) -> PolarsResult<Series> + Send + Sync),
) -> PolarsResult<Series> {
let n_threads = POOL.current_num_threads();
let offsets = _split_offsets(len, n_threads);
let series: PolarsResult<Vec<_>> = POOL.install(|| {
offsets
.into_par_iter()
.map(|(offset, len)| func(offset, len))
.collect()
});
Ok(self.finish_take_threaded(series?, rechunk))
}
/// Take by index if ChunkedArray contains a single chunk.
///
/// # Safety
/// This doesn't check any bounds. Null validity is checked.
pub unsafe fn take_unchecked_threaded(
&self,
idx: &IdxCa,
rechunk: bool,
) -> PolarsResult<Series> {
self.threaded_op(rechunk, idx.len(), &|offset, len| {
let idx = idx.slice(offset as i64, len);
self.take_unchecked(&idx)
})
}
/// # Safety
/// This doesn't check any bounds. Null validity is checked.
#[cfg(feature = "chunked_ids")]
pub(crate) unsafe fn _take_chunked_unchecked_threaded(
&self,
chunk_ids: &[ChunkId],
sorted: IsSorted,
rechunk: bool,
) -> Series {
self.threaded_op(rechunk, chunk_ids.len(), &|offset, len| {
let chunk_ids = &chunk_ids[offset..offset + len];
Ok(self._take_chunked_unchecked(chunk_ids, sorted))
})
.unwrap()
}
/// # Safety
/// This doesn't check any bounds. Null validity is checked.
#[cfg(feature = "chunked_ids")]
pub(crate) unsafe fn _take_opt_chunked_unchecked_threaded(
&self,
chunk_ids: &[Option<ChunkId>],
rechunk: bool,
) -> Series {
self.threaded_op(rechunk, chunk_ids.len(), &|offset, len| {
let chunk_ids = &chunk_ids[offset..offset + len];
Ok(self._take_opt_chunked_unchecked(chunk_ids))
})
.unwrap()
}
/// Take by index. This operation is clone.
///
/// # Notes
/// Out of bounds access doesn't Error but will return a Null value
pub fn take_threaded(&self, idx: &IdxCa, rechunk: bool) -> PolarsResult<Series> {
self.threaded_op(rechunk, idx.len(), &|offset, len| {
let idx = idx.slice(offset as i64, len);
self.take(&idx)
})
}
/// Filter by boolean mask. This operation clones data.
pub fn filter_threaded(&self, filter: &BooleanChunked, rechunk: bool) -> PolarsResult<Series> {
// this would fail if there is a broadcasting filter.
// because we cannot split that filter over threads
// besides they are a no-op, so we do the standard filter.
if filter.len() == 1 {
return self.filter(filter);
}
let n_threads = POOL.current_num_threads();
let filters = split_ca(filter, n_threads).unwrap();
let series = split_series(self, n_threads).unwrap();
let series: PolarsResult<Vec<_>> = POOL.install(|| {
filters
.par_iter()
.zip(series)
.map(|(filter, s)| s.filter(filter))
.collect()
});
Ok(self.finish_take_threaded(series?, rechunk))
}
#[cfg(feature = "dot_product")]
#[cfg_attr(docsrs, doc(cfg(feature = "dot_product")))]
pub fn dot(&self, other: &Series) -> Option<f64> {
(self * other).sum::<f64>()
}
/// Get the sum of the Series as a new Series of length 1.
/// Returns a Series with a single zeroed entry if self is an empty numeric series.
///
/// If the [`DataType`] is one of `{Int8, UInt8, Int16, UInt16}` the `Series` is
/// first cast to `Int64` to prevent overflow issues.
pub fn sum_as_series(&self) -> Series {
use DataType::*;
if self.is_empty() && self.dtype().is_numeric() {
return Series::new("", [0])
.cast(self.dtype())
.unwrap()
.sum_as_series();
}
match self.dtype() {
Int8 | UInt8 | Int16 | UInt16 => self.cast(&Int64).unwrap().sum_as_series(),
_ => self._sum_as_series(),
}
}
/// Get an array with the cumulative max computed at every element
#[cfg_attr(docsrs, doc(cfg(feature = "cum_agg")))]
pub fn cummax(&self, _reverse: bool) -> Series {
#[cfg(feature = "cum_agg")]
{
self._cummax(_reverse)
}
#[cfg(not(feature = "cum_agg"))]
{
panic!("activate 'cum_agg' feature")
}
}
/// Get an array with the cumulative min computed at every element
#[cfg_attr(docsrs, doc(cfg(feature = "cum_agg")))]
pub fn cummin(&self, _reverse: bool) -> Series {
#[cfg(feature = "cum_agg")]
{
self._cummin(_reverse)
}
#[cfg(not(feature = "cum_agg"))]
{
panic!("activate 'cum_agg' feature")
}
}
/// Get an array with the cumulative sum computed at every element
///
/// If the [`DataType`] is one of `{Int8, UInt8, Int16, UInt16}` the `Series` is
/// first cast to `Int64` to prevent overflow issues.
#[cfg_attr(docsrs, doc(cfg(feature = "cum_agg")))]
#[allow(unused_variables)]
pub fn cumsum(&self, reverse: bool) -> Series {
#[cfg(feature = "cum_agg")]
{
use DataType::*;
match self.dtype() {
Boolean => self.cast(&DataType::UInt32).unwrap().cumsum(reverse),
Int8 | UInt8 | Int16 | UInt16 => {
let s = self.cast(&Int64).unwrap();
s.cumsum(reverse)
}
Int32 => {
let ca = self.i32().unwrap();
ca.cumsum(reverse).into_series()
}
UInt32 => {
let ca = self.u32().unwrap();
ca.cumsum(reverse).into_series()
}
UInt64 => {
let ca = self.u64().unwrap();
ca.cumsum(reverse).into_series()
}
Int64 => {
let ca = self.i64().unwrap();
ca.cumsum(reverse).into_series()
}
Float32 => {
let ca = self.f32().unwrap();
ca.cumsum(reverse).into_series()
}
Float64 => {
let ca = self.f64().unwrap();
ca.cumsum(reverse).into_series()
}
dt => panic!("cumsum not supported for dtype: {dt:?}"),
}
}
#[cfg(not(feature = "cum_agg"))]
{
panic!("activate 'cum_agg' feature")
}
}
/// Get an array with the cumulative product computed at every element
///
/// If the [`DataType`] is one of `{Int8, UInt8, Int16, UInt16, Int32, UInt32}` the `Series` is
/// first cast to `Int64` to prevent overflow issues.
#[cfg_attr(docsrs, doc(cfg(feature = "cum_agg")))]
#[allow(unused_variables)]
pub fn cumprod(&self, reverse: bool) -> Series {
#[cfg(feature = "cum_agg")]
{
use DataType::*;
match self.dtype() {
Boolean => self.cast(&DataType::Int64).unwrap().cumprod(reverse),
Int8 | UInt8 | Int16 | UInt16 | Int32 | UInt32 => {
let s = self.cast(&Int64).unwrap();
s.cumprod(reverse)
}
Int64 => {
let ca = self.i64().unwrap();
ca.cumprod(reverse).into_series()
}
UInt64 => {
let ca = self.u64().unwrap();
ca.cumprod(reverse).into_series()
}
Float32 => {
let ca = self.f32().unwrap();
ca.cumprod(reverse).into_series()
}
Float64 => {
let ca = self.f64().unwrap();
ca.cumprod(reverse).into_series()
}
dt => panic!("cumprod not supported for dtype: {dt:?}"),
}
}
#[cfg(not(feature = "cum_agg"))]
{
panic!("activate 'cum_agg' feature")
}
}
/// Get the product of an array.
///
/// If the [`DataType`] is one of `{Int8, UInt8, Int16, UInt16}` the `Series` is
/// first cast to `Int64` to prevent overflow issues.
#[cfg_attr(docsrs, doc(cfg(feature = "product")))]
pub fn product(&self) -> Series {
#[cfg(feature = "product")]
{
use DataType::*;
match self.dtype() {
Boolean => self.cast(&DataType::Int64).unwrap().product(),
Int8 | UInt8 | Int16 | UInt16 => {
let s = self.cast(&Int64).unwrap();
s.product()
}
Int64 => {
let ca = self.i64().unwrap();
ca.prod_as_series()
}
Float32 => {
let ca = self.f32().unwrap();
ca.prod_as_series()
}
Float64 => {
let ca = self.f64().unwrap();
ca.prod_as_series()
}
dt => panic!("cumprod not supported for dtype: {dt:?}"),
}
}
#[cfg(not(feature = "product"))]
{
panic!("activate 'product' feature")
}
}
#[cfg(feature = "rank")]
#[cfg_attr(docsrs, doc(cfg(feature = "rank")))]
pub fn rank(&self, options: RankOptions) -> Series {
rank(self, options.method, options.descending)
}
/// Cast throws an error if conversion had overflows
pub fn strict_cast(&self, data_type: &DataType) -> PolarsResult<Series> {
let s = self.cast(data_type)?;
if self.null_count() != s.null_count() {
let failure_mask = !self.is_null() & s.is_null();
let failures = self.filter_threaded(&failure_mask, false)?.unique()?;
Err(PolarsError::ComputeError(
format!(
"Strict conversion from {:?} to {:?} failed for values {}. \
If you were trying to cast Utf8 to Date, Time, or Datetime, \
consider using `strptime`.",
self.dtype(),
data_type,
failures.fmt_list(),
)
.into(),
))
} else {
Ok(s)
}
}
#[cfg(feature = "dtype-time")]
pub(crate) fn into_time(self) -> Series {
#[cfg(not(feature = "dtype-time"))]
{
panic!("activate feature dtype-time")
}
match self.dtype() {
DataType::Int64 => self.i64().unwrap().clone().into_time().into_series(),
DataType::Time => self
.time()
.unwrap()
.as_ref()
.clone()
.into_time()
.into_series(),
dt => panic!("date not implemented for {dt:?}"),
}
}
pub(crate) fn into_date(self) -> Series {
#[cfg(not(feature = "dtype-date"))]
{
panic!("activate feature dtype-date")
}
#[cfg(feature = "dtype-date")]
match self.dtype() {
DataType::Int32 => self.i32().unwrap().clone().into_date().into_series(),
DataType::Date => self
.date()
.unwrap()
.as_ref()
.clone()
.into_date()
.into_series(),
dt => panic!("date not implemented for {dt:?}"),
}
}
pub(crate) fn into_datetime(self, timeunit: TimeUnit, tz: Option<TimeZone>) -> Series {
#[cfg(not(feature = "dtype-datetime"))]
{
panic!("activate feature dtype-datetime")
}
#[cfg(feature = "dtype-datetime")]
match self.dtype() {
DataType::Int64 => self
.i64()
.unwrap()
.clone()
.into_datetime(timeunit, tz)
.into_series(),
DataType::Datetime(_, _) => self
.datetime()
.unwrap()
.as_ref()
.clone()
.into_datetime(timeunit, tz)
.into_series(),
dt => panic!("into_datetime not implemented for {dt:?}"),
}
}
pub(crate) fn into_duration(self, timeunit: TimeUnit) -> Series {
#[cfg(not(feature = "dtype-duration"))]
{
panic!("activate feature dtype-duration")
}
#[cfg(feature = "dtype-duration")]
match self.dtype() {
DataType::Int64 => self
.i64()
.unwrap()
.clone()
.into_duration(timeunit)
.into_series(),
DataType::Duration(_) => self
.duration()
.unwrap()
.as_ref()
.clone()
.into_duration(timeunit)
.into_series(),
dt => panic!("into_duration not implemented for {dt:?}"),
}
}
#[cfg(feature = "abs")]
#[cfg_attr(docsrs, doc(cfg(feature = "abs")))]
/// convert numerical values to their absolute value
pub fn abs(&self) -> PolarsResult<Series> {
let a = self.to_physical_repr();
use DataType::*;
let out = match a.dtype() {
#[cfg(feature = "dtype-i8")]
Int8 => a.i8().unwrap().abs().into_series(),
#[cfg(feature = "dtype-i16")]
Int16 => a.i16().unwrap().abs().into_series(),
Int32 => a.i32().unwrap().abs().into_series(),
Int64 => a.i64().unwrap().abs().into_series(),
UInt8 | UInt16 | UInt32 | UInt64 => self.clone(),
Float32 => a.f32().unwrap().abs().into_series(),
Float64 => a.f64().unwrap().abs().into_series(),
dt => {
return Err(PolarsError::InvalidOperation(
format!("abs not supported for series of type {dt:?}").into(),
));
}
};
Ok(out)
}More examples
src/frame/groupby/aggregations/dispatch.rs (line 113)
109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184
pub unsafe fn agg_median(&self, groups: &GroupsProxy) -> Series {
use DataType::*;
match self.dtype() {
Float32 => SeriesWrap(self.f32().unwrap().clone()).agg_median(groups),
Float64 => SeriesWrap(self.f64().unwrap().clone()).agg_median(groups),
dt if dt.is_numeric() || dt.is_temporal() => {
let ca = self.to_physical_repr();
let physical_type = ca.dtype();
let s = apply_method_physical_integer!(ca, agg_median, groups);
if dt.is_logical() {
// back to physical and then
// back to logical type
s.cast(physical_type).unwrap().cast(dt).unwrap()
} else {
s
}
}
_ => Series::full_null("", groups.len(), self.dtype()),
}
}
#[doc(hidden)]
pub unsafe fn agg_quantile(
&self,
groups: &GroupsProxy,
quantile: f64,
interpol: QuantileInterpolOptions,
) -> Series {
use DataType::*;
match self.dtype() {
Float32 => {
SeriesWrap(self.f32().unwrap().clone()).agg_quantile(groups, quantile, interpol)
}
Float64 => {
SeriesWrap(self.f64().unwrap().clone()).agg_quantile(groups, quantile, interpol)
}
dt if dt.is_numeric() || dt.is_temporal() => {
let ca = self.to_physical_repr();
let physical_type = ca.dtype();
let s =
apply_method_physical_integer!(ca, agg_quantile, groups, quantile, interpol);
if dt.is_logical() {
// back to physical and then
// back to logical type
s.cast(physical_type).unwrap().cast(dt).unwrap()
} else {
s
}
}
_ => Series::full_null("", groups.len(), self.dtype()),
}
}
#[doc(hidden)]
pub unsafe fn agg_mean(&self, groups: &GroupsProxy) -> Series {
use DataType::*;
match self.dtype() {
Boolean => self.cast(&Float64).unwrap().agg_mean(groups),
Float32 => SeriesWrap(self.f32().unwrap().clone()).agg_mean(groups),
Float64 => SeriesWrap(self.f64().unwrap().clone()).agg_mean(groups),
dt if dt.is_numeric() => {
apply_method_physical_integer!(self, agg_mean, groups)
}
dt @ Duration(_) => {
let s = self.to_physical_repr();
// agg_mean returns Float64
let out = s.agg_mean(groups);
// cast back to Int64 and then to logical duration type
out.cast(&Int64).unwrap().cast(dt).unwrap()
}
_ => Series::full_null("", groups.len(), self.dtype()),
}
}src/series/ops/round.rs (line 8)
6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59
pub fn round(&self, decimals: u32) -> PolarsResult<Self> {
use num::traits::Pow;
if let Ok(ca) = self.f32() {
// Note we do the computation on f64 floats to not loose precision
// when the computation is done, we cast to f32
let multiplier = 10.0.pow(decimals as f64);
let s = ca
.apply(|val| ((val as f64 * multiplier).round() / multiplier) as f32)
.into_series();
return Ok(s);
}
if let Ok(ca) = self.f64() {
let multiplier = 10.0.pow(decimals as f64);
let s = ca
.apply(|val| (val * multiplier).round() / multiplier)
.into_series();
return Ok(s);
}
Err(PolarsError::SchemaMisMatch(
format!("{:?} is not a floating point datatype", self.dtype()).into(),
))
}
#[cfg_attr(docsrs, doc(cfg(feature = "round_series")))]
/// Floor underlying floating point array to the lowest integers smaller or equal to the float value.
pub fn floor(&self) -> PolarsResult<Self> {
if let Ok(ca) = self.f32() {
let s = ca.apply(|val| val.floor()).into_series();
return Ok(s);
}
if let Ok(ca) = self.f64() {
let s = ca.apply(|val| val.floor()).into_series();
return Ok(s);
}
Err(PolarsError::SchemaMisMatch(
format!("{:?} is not a floating point datatype", self.dtype()).into(),
))
}
#[cfg_attr(docsrs, doc(cfg(feature = "round_series")))]
/// Ceil underlying floating point array to the highest integers smaller or equal to the float value.
pub fn ceil(&self) -> PolarsResult<Self> {
if let Ok(ca) = self.f32() {
let s = ca.apply(|val| val.ceil()).into_series();
return Ok(s);
}
if let Ok(ca) = self.f64() {
let s = ca.apply(|val| val.ceil()).into_series();
return Ok(s);
}
Err(PolarsError::SchemaMisMatch(
format!("{:?} is not a floating point datatype", self.dtype()).into(),
))
}src/chunked_array/ndarray.rs (line 111)
100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163
pub fn to_ndarray<N>(&self) -> PolarsResult<Array2<N::Native>>
where
N: PolarsNumericType,
{
let columns = self
.get_columns()
.par_iter()
.map(|s| {
let s = s.cast(&N::get_dtype())?;
let s = match s.dtype() {
DataType::Float32 => {
let ca = s.f32().unwrap();
ca.none_to_nan().into_series()
}
DataType::Float64 => {
let ca = s.f64().unwrap();
ca.none_to_nan().into_series()
}
_ => s,
};
Ok(s.rechunk())
})
.collect::<PolarsResult<Vec<_>>>()?;
let shape = self.shape();
let height = self.height();
let mut membuf = Vec::with_capacity(shape.0 * shape.1);
let ptr = membuf.as_ptr() as usize;
columns.par_iter().enumerate().map(|(col_idx, s)| {
if s.null_count() != 0 {
return Err(PolarsError::ComputeError(
"Creation of ndarray with null values is not supported. Consider using floats and NaNs".into(),
));
}
// this is an Arc clone if already of type N
let s = s.cast(&N::get_dtype())?;
let ca = s.unpack::<N>()?;
let vals = ca.cont_slice().unwrap();
// Safety:
// we get parallel access to the vector
// but we make sure that we don't get aliased access by offsetting the column indices + length
unsafe {
let offset_ptr = (ptr as *mut N::Native).add(col_idx * height) ;
// Safety:
// this is uninitialized memory, so we must never read from this data
// copy_from_slice does not read
let buf = std::slice::from_raw_parts_mut(offset_ptr, height);
buf.copy_from_slice(vals)
}
Ok(())
}).collect::<PolarsResult<Vec<_>>>()?;
// Safety:
// we have written all data, so we can now safely set length
unsafe {
membuf.set_len(shape.0 * shape.1);
}
let ndarr = Array2::from_shape_vec((shape.1, shape.0), membuf).unwrap();
Ok(ndarr.reversed_axes())
}src/frame/asof_join/mod.rs (line 147)
112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195
pub fn _join_asof(
&self,
other: &DataFrame,
left_on: &str,
right_on: &str,
strategy: AsofStrategy,
tolerance: Option<AnyValue<'static>>,
suffix: Option<String>,
slice: Option<(i64, usize)>,
) -> PolarsResult<DataFrame> {
let left_key = self.column(left_on)?;
let right_key = other.column(right_on)?;
check_asof_columns(left_key, right_key)?;
let left_key = left_key.to_physical_repr();
let right_key = right_key.to_physical_repr();
let take_idx = match left_key.dtype() {
DataType::Int64 => left_key
.i64()
.unwrap()
.join_asof(&right_key, strategy, tolerance),
DataType::Int32 => left_key
.i32()
.unwrap()
.join_asof(&right_key, strategy, tolerance),
DataType::UInt64 => left_key
.u64()
.unwrap()
.join_asof(&right_key, strategy, tolerance),
DataType::UInt32 => left_key
.u32()
.unwrap()
.join_asof(&right_key, strategy, tolerance),
DataType::Float32 => left_key
.f32()
.unwrap()
.join_asof(&right_key, strategy, tolerance),
DataType::Float64 => left_key
.f64()
.unwrap()
.join_asof(&right_key, strategy, tolerance),
_ => {
let left_key = left_key.cast(&DataType::Int32).unwrap();
let right_key = right_key.cast(&DataType::Int32).unwrap();
left_key
.i32()
.unwrap()
.join_asof(&right_key, strategy, tolerance)
}
}?;
// take_idx are sorted so this is a bound check for all
if let Some(Some(idx)) = take_idx.last() {
assert!((*idx as usize) < other.height())
}
// drop right join column
let other = if left_on == right_on {
Cow::Owned(other.drop(right_on)?)
} else {
Cow::Borrowed(other)
};
let mut left = self.clone();
let mut take_idx = &*take_idx;
if let Some((offset, len)) = slice {
left = left.slice(offset, len);
take_idx = slice_slice(take_idx, offset, len);
}
// Safety:
// join tuples are in bounds
let right_df = unsafe {
other.take_opt_iter_unchecked(
take_idx
.iter()
.map(|opt_idx| opt_idx.map(|idx| idx as usize)),
)
};
_finish_join(left, right_df, suffix.as_deref())
}src/series/arithmetic/borrowed.rs (line 270)
220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300
fn checked_div_num<T: ToPrimitive>(&self, rhs: T) -> PolarsResult<Series> {
use DataType::*;
let s = self.to_physical_repr();
let out = match s.dtype() {
#[cfg(feature = "dtype-u8")]
UInt8 => s
.u8()
.unwrap()
.apply_on_opt(|opt_v| opt_v.and_then(|v| v.checked_div(rhs.to_u8().unwrap())))
.into_series(),
#[cfg(feature = "dtype-i8")]
Int8 => s
.i8()
.unwrap()
.apply_on_opt(|opt_v| opt_v.and_then(|v| v.checked_div(rhs.to_i8().unwrap())))
.into_series(),
#[cfg(feature = "dtype-i16")]
Int16 => s
.i16()
.unwrap()
.apply_on_opt(|opt_v| opt_v.and_then(|v| v.checked_div(rhs.to_i16().unwrap())))
.into_series(),
#[cfg(feature = "dtype-u16")]
UInt16 => s
.u16()
.unwrap()
.apply_on_opt(|opt_v| opt_v.and_then(|v| v.checked_div(rhs.to_u16().unwrap())))
.into_series(),
UInt32 => s
.u32()
.unwrap()
.apply_on_opt(|opt_v| opt_v.and_then(|v| v.checked_div(rhs.to_u32().unwrap())))
.into_series(),
Int32 => s
.i32()
.unwrap()
.apply_on_opt(|opt_v| opt_v.and_then(|v| v.checked_div(rhs.to_i32().unwrap())))
.into_series(),
UInt64 => s
.u64()
.unwrap()
.apply_on_opt(|opt_v| opt_v.and_then(|v| v.checked_div(rhs.to_u64().unwrap())))
.into_series(),
Int64 => s
.i64()
.unwrap()
.apply_on_opt(|opt_v| opt_v.and_then(|v| v.checked_div(rhs.to_i64().unwrap())))
.into_series(),
Float32 => s
.f32()
.unwrap()
.apply_on_opt(|opt_v| {
opt_v.and_then(|v| {
let res = rhs.to_f32().unwrap();
if res.is_zero() {
None
} else {
Some(v / res)
}
})
})
.into_series(),
Float64 => s
.f64()
.unwrap()
.apply_on_opt(|opt_v| {
opt_v.and_then(|v| {
let res = rhs.to_f64().unwrap();
if res.is_zero() {
None
} else {
Some(v / res)
}
})
})
.into_series(),
_ => panic!("dtype not yet supported in checked div"),
};
out.cast(self.dtype())
}Additional examples can be found in:
sourcepub fn f64(&self) -> PolarsResult<&Float64Chunked>
pub fn f64(&self) -> PolarsResult<&Float64Chunked>
Unpack to ChunkedArray of dtype f64
Examples found in repository?
src/series/mod.rs (line 253)
246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751 752 753 754 755 756 757 758 759 760 761 762 763 764 765 766 767 768 769 770 771 772 773 774 775 776 777 778 779 780 781 782 783 784 785 786 787 788 789 790 791 792 793 794 795 796 797 798 799 800 801 802 803 804 805 806 807 808 809 810 811 812 813 814 815 816 817 818 819 820 821 822
pub fn sum<T>(&self) -> Option<T>
where
T: NumCast,
{
self.sum_as_series()
.cast(&DataType::Float64)
.ok()
.and_then(|s| s.f64().unwrap().get(0).and_then(T::from))
}
/// Returns the minimum value in the array, according to the natural order.
/// Returns an option because the array is nullable.
/// ```
/// # use polars_core::prelude::*;
/// let s = Series::new("days", [1, 2, 3].as_ref());
/// assert_eq!(s.min(), Some(1));
/// ```
pub fn min<T>(&self) -> Option<T>
where
T: NumCast,
{
self.min_as_series()
.cast(&DataType::Float64)
.ok()
.and_then(|s| s.f64().unwrap().get(0).and_then(T::from))
}
/// Returns the maximum value in the array, according to the natural order.
/// Returns an option because the array is nullable.
/// ```
/// # use polars_core::prelude::*;
/// let s = Series::new("days", [1, 2, 3].as_ref());
/// assert_eq!(s.max(), Some(3));
/// ```
pub fn max<T>(&self) -> Option<T>
where
T: NumCast,
{
self.max_as_series()
.cast(&DataType::Float64)
.ok()
.and_then(|s| s.f64().unwrap().get(0).and_then(T::from))
}
/// Explode a list or utf8 Series. This expands every item to a new row..
pub fn explode(&self) -> PolarsResult<Series> {
match self.dtype() {
DataType::List(_) => self.list().unwrap().explode(),
DataType::Utf8 => self.utf8().unwrap().explode(),
_ => Err(PolarsError::InvalidOperation(
format!(
"explode not supported for Series with dtype {:?}",
self.dtype()
)
.into(),
)),
}
}
/// Check if float value is NaN (note this is different than missing/ null)
pub fn is_nan(&self) -> PolarsResult<BooleanChunked> {
match self.dtype() {
DataType::Float32 => Ok(self.f32().unwrap().is_nan()),
DataType::Float64 => Ok(self.f64().unwrap().is_nan()),
_ => Err(PolarsError::InvalidOperation(
format!(
"'is_nan' not supported for series with dtype {:?}",
self.dtype()
)
.into(),
)),
}
}
/// Check if float value is NaN (note this is different than missing/ null)
pub fn is_not_nan(&self) -> PolarsResult<BooleanChunked> {
match self.dtype() {
DataType::Float32 => Ok(self.f32().unwrap().is_not_nan()),
DataType::Float64 => Ok(self.f64().unwrap().is_not_nan()),
_ => Err(PolarsError::InvalidOperation(
format!(
"'is_not_nan' not supported for series with dtype {:?}",
self.dtype()
)
.into(),
)),
}
}
/// Check if float value is finite
pub fn is_finite(&self) -> PolarsResult<BooleanChunked> {
match self.dtype() {
DataType::Float32 => Ok(self.f32().unwrap().is_finite()),
DataType::Float64 => Ok(self.f64().unwrap().is_finite()),
_ => Err(PolarsError::InvalidOperation(
format!(
"'is_finite' not supported for series with dtype {:?}",
self.dtype()
)
.into(),
)),
}
}
/// Check if float value is infinite
pub fn is_infinite(&self) -> PolarsResult<BooleanChunked> {
match self.dtype() {
DataType::Float32 => Ok(self.f32().unwrap().is_infinite()),
DataType::Float64 => Ok(self.f64().unwrap().is_infinite()),
_ => Err(PolarsError::InvalidOperation(
format!(
"'is_infinite' not supported for series with dtype {:?}",
self.dtype()
)
.into(),
)),
}
}
/// Create a new ChunkedArray with values from self where the mask evaluates `true` and values
/// from `other` where the mask evaluates `false`
#[cfg(feature = "zip_with")]
#[cfg_attr(docsrs, doc(cfg(feature = "zip_with")))]
pub fn zip_with(&self, mask: &BooleanChunked, other: &Series) -> PolarsResult<Series> {
let (lhs, rhs) = coerce_lhs_rhs(self, other)?;
lhs.zip_with_same_type(mask, rhs.as_ref())
}
/// Cast a datelike Series to their physical representation.
/// Primitives remain unchanged
///
/// * Date -> Int32
/// * Datetime-> Int64
/// * Time -> Int64
/// * Categorical -> UInt32
///
pub fn to_physical_repr(&self) -> Cow<Series> {
use DataType::*;
match self.dtype() {
Date => Cow::Owned(self.cast(&DataType::Int32).unwrap()),
Datetime(_, _) | Duration(_) | Time => Cow::Owned(self.cast(&DataType::Int64).unwrap()),
#[cfg(feature = "dtype-categorical")]
Categorical(_) => Cow::Owned(self.cast(&DataType::UInt32).unwrap()),
_ => Cow::Borrowed(self),
}
}
fn finish_take_threaded(&self, s: Vec<Series>, rechunk: bool) -> Series {
let s = s
.into_iter()
.reduce(|mut s, s1| {
s.append(&s1).unwrap();
s
})
.unwrap();
if rechunk {
s.rechunk()
} else {
s
}
}
// take a function pointer to reduce bloat
fn threaded_op(
&self,
rechunk: bool,
len: usize,
func: &(dyn Fn(usize, usize) -> PolarsResult<Series> + Send + Sync),
) -> PolarsResult<Series> {
let n_threads = POOL.current_num_threads();
let offsets = _split_offsets(len, n_threads);
let series: PolarsResult<Vec<_>> = POOL.install(|| {
offsets
.into_par_iter()
.map(|(offset, len)| func(offset, len))
.collect()
});
Ok(self.finish_take_threaded(series?, rechunk))
}
/// Take by index if ChunkedArray contains a single chunk.
///
/// # Safety
/// This doesn't check any bounds. Null validity is checked.
pub unsafe fn take_unchecked_threaded(
&self,
idx: &IdxCa,
rechunk: bool,
) -> PolarsResult<Series> {
self.threaded_op(rechunk, idx.len(), &|offset, len| {
let idx = idx.slice(offset as i64, len);
self.take_unchecked(&idx)
})
}
/// # Safety
/// This doesn't check any bounds. Null validity is checked.
#[cfg(feature = "chunked_ids")]
pub(crate) unsafe fn _take_chunked_unchecked_threaded(
&self,
chunk_ids: &[ChunkId],
sorted: IsSorted,
rechunk: bool,
) -> Series {
self.threaded_op(rechunk, chunk_ids.len(), &|offset, len| {
let chunk_ids = &chunk_ids[offset..offset + len];
Ok(self._take_chunked_unchecked(chunk_ids, sorted))
})
.unwrap()
}
/// # Safety
/// This doesn't check any bounds. Null validity is checked.
#[cfg(feature = "chunked_ids")]
pub(crate) unsafe fn _take_opt_chunked_unchecked_threaded(
&self,
chunk_ids: &[Option<ChunkId>],
rechunk: bool,
) -> Series {
self.threaded_op(rechunk, chunk_ids.len(), &|offset, len| {
let chunk_ids = &chunk_ids[offset..offset + len];
Ok(self._take_opt_chunked_unchecked(chunk_ids))
})
.unwrap()
}
/// Take by index. This operation is clone.
///
/// # Notes
/// Out of bounds access doesn't Error but will return a Null value
pub fn take_threaded(&self, idx: &IdxCa, rechunk: bool) -> PolarsResult<Series> {
self.threaded_op(rechunk, idx.len(), &|offset, len| {
let idx = idx.slice(offset as i64, len);
self.take(&idx)
})
}
/// Filter by boolean mask. This operation clones data.
pub fn filter_threaded(&self, filter: &BooleanChunked, rechunk: bool) -> PolarsResult<Series> {
// this would fail if there is a broadcasting filter.
// because we cannot split that filter over threads
// besides they are a no-op, so we do the standard filter.
if filter.len() == 1 {
return self.filter(filter);
}
let n_threads = POOL.current_num_threads();
let filters = split_ca(filter, n_threads).unwrap();
let series = split_series(self, n_threads).unwrap();
let series: PolarsResult<Vec<_>> = POOL.install(|| {
filters
.par_iter()
.zip(series)
.map(|(filter, s)| s.filter(filter))
.collect()
});
Ok(self.finish_take_threaded(series?, rechunk))
}
#[cfg(feature = "dot_product")]
#[cfg_attr(docsrs, doc(cfg(feature = "dot_product")))]
pub fn dot(&self, other: &Series) -> Option<f64> {
(self * other).sum::<f64>()
}
/// Get the sum of the Series as a new Series of length 1.
/// Returns a Series with a single zeroed entry if self is an empty numeric series.
///
/// If the [`DataType`] is one of `{Int8, UInt8, Int16, UInt16}` the `Series` is
/// first cast to `Int64` to prevent overflow issues.
pub fn sum_as_series(&self) -> Series {
use DataType::*;
if self.is_empty() && self.dtype().is_numeric() {
return Series::new("", [0])
.cast(self.dtype())
.unwrap()
.sum_as_series();
}
match self.dtype() {
Int8 | UInt8 | Int16 | UInt16 => self.cast(&Int64).unwrap().sum_as_series(),
_ => self._sum_as_series(),
}
}
/// Get an array with the cumulative max computed at every element
#[cfg_attr(docsrs, doc(cfg(feature = "cum_agg")))]
pub fn cummax(&self, _reverse: bool) -> Series {
#[cfg(feature = "cum_agg")]
{
self._cummax(_reverse)
}
#[cfg(not(feature = "cum_agg"))]
{
panic!("activate 'cum_agg' feature")
}
}
/// Get an array with the cumulative min computed at every element
#[cfg_attr(docsrs, doc(cfg(feature = "cum_agg")))]
pub fn cummin(&self, _reverse: bool) -> Series {
#[cfg(feature = "cum_agg")]
{
self._cummin(_reverse)
}
#[cfg(not(feature = "cum_agg"))]
{
panic!("activate 'cum_agg' feature")
}
}
/// Get an array with the cumulative sum computed at every element
///
/// If the [`DataType`] is one of `{Int8, UInt8, Int16, UInt16}` the `Series` is
/// first cast to `Int64` to prevent overflow issues.
#[cfg_attr(docsrs, doc(cfg(feature = "cum_agg")))]
#[allow(unused_variables)]
pub fn cumsum(&self, reverse: bool) -> Series {
#[cfg(feature = "cum_agg")]
{
use DataType::*;
match self.dtype() {
Boolean => self.cast(&DataType::UInt32).unwrap().cumsum(reverse),
Int8 | UInt8 | Int16 | UInt16 => {
let s = self.cast(&Int64).unwrap();
s.cumsum(reverse)
}
Int32 => {
let ca = self.i32().unwrap();
ca.cumsum(reverse).into_series()
}
UInt32 => {
let ca = self.u32().unwrap();
ca.cumsum(reverse).into_series()
}
UInt64 => {
let ca = self.u64().unwrap();
ca.cumsum(reverse).into_series()
}
Int64 => {
let ca = self.i64().unwrap();
ca.cumsum(reverse).into_series()
}
Float32 => {
let ca = self.f32().unwrap();
ca.cumsum(reverse).into_series()
}
Float64 => {
let ca = self.f64().unwrap();
ca.cumsum(reverse).into_series()
}
dt => panic!("cumsum not supported for dtype: {dt:?}"),
}
}
#[cfg(not(feature = "cum_agg"))]
{
panic!("activate 'cum_agg' feature")
}
}
/// Get an array with the cumulative product computed at every element
///
/// If the [`DataType`] is one of `{Int8, UInt8, Int16, UInt16, Int32, UInt32}` the `Series` is
/// first cast to `Int64` to prevent overflow issues.
#[cfg_attr(docsrs, doc(cfg(feature = "cum_agg")))]
#[allow(unused_variables)]
pub fn cumprod(&self, reverse: bool) -> Series {
#[cfg(feature = "cum_agg")]
{
use DataType::*;
match self.dtype() {
Boolean => self.cast(&DataType::Int64).unwrap().cumprod(reverse),
Int8 | UInt8 | Int16 | UInt16 | Int32 | UInt32 => {
let s = self.cast(&Int64).unwrap();
s.cumprod(reverse)
}
Int64 => {
let ca = self.i64().unwrap();
ca.cumprod(reverse).into_series()
}
UInt64 => {
let ca = self.u64().unwrap();
ca.cumprod(reverse).into_series()
}
Float32 => {
let ca = self.f32().unwrap();
ca.cumprod(reverse).into_series()
}
Float64 => {
let ca = self.f64().unwrap();
ca.cumprod(reverse).into_series()
}
dt => panic!("cumprod not supported for dtype: {dt:?}"),
}
}
#[cfg(not(feature = "cum_agg"))]
{
panic!("activate 'cum_agg' feature")
}
}
/// Get the product of an array.
///
/// If the [`DataType`] is one of `{Int8, UInt8, Int16, UInt16}` the `Series` is
/// first cast to `Int64` to prevent overflow issues.
#[cfg_attr(docsrs, doc(cfg(feature = "product")))]
pub fn product(&self) -> Series {
#[cfg(feature = "product")]
{
use DataType::*;
match self.dtype() {
Boolean => self.cast(&DataType::Int64).unwrap().product(),
Int8 | UInt8 | Int16 | UInt16 => {
let s = self.cast(&Int64).unwrap();
s.product()
}
Int64 => {
let ca = self.i64().unwrap();
ca.prod_as_series()
}
Float32 => {
let ca = self.f32().unwrap();
ca.prod_as_series()
}
Float64 => {
let ca = self.f64().unwrap();
ca.prod_as_series()
}
dt => panic!("cumprod not supported for dtype: {dt:?}"),
}
}
#[cfg(not(feature = "product"))]
{
panic!("activate 'product' feature")
}
}
#[cfg(feature = "rank")]
#[cfg_attr(docsrs, doc(cfg(feature = "rank")))]
pub fn rank(&self, options: RankOptions) -> Series {
rank(self, options.method, options.descending)
}
/// Cast throws an error if conversion had overflows
pub fn strict_cast(&self, data_type: &DataType) -> PolarsResult<Series> {
let s = self.cast(data_type)?;
if self.null_count() != s.null_count() {
let failure_mask = !self.is_null() & s.is_null();
let failures = self.filter_threaded(&failure_mask, false)?.unique()?;
Err(PolarsError::ComputeError(
format!(
"Strict conversion from {:?} to {:?} failed for values {}. \
If you were trying to cast Utf8 to Date, Time, or Datetime, \
consider using `strptime`.",
self.dtype(),
data_type,
failures.fmt_list(),
)
.into(),
))
} else {
Ok(s)
}
}
#[cfg(feature = "dtype-time")]
pub(crate) fn into_time(self) -> Series {
#[cfg(not(feature = "dtype-time"))]
{
panic!("activate feature dtype-time")
}
match self.dtype() {
DataType::Int64 => self.i64().unwrap().clone().into_time().into_series(),
DataType::Time => self
.time()
.unwrap()
.as_ref()
.clone()
.into_time()
.into_series(),
dt => panic!("date not implemented for {dt:?}"),
}
}
pub(crate) fn into_date(self) -> Series {
#[cfg(not(feature = "dtype-date"))]
{
panic!("activate feature dtype-date")
}
#[cfg(feature = "dtype-date")]
match self.dtype() {
DataType::Int32 => self.i32().unwrap().clone().into_date().into_series(),
DataType::Date => self
.date()
.unwrap()
.as_ref()
.clone()
.into_date()
.into_series(),
dt => panic!("date not implemented for {dt:?}"),
}
}
pub(crate) fn into_datetime(self, timeunit: TimeUnit, tz: Option<TimeZone>) -> Series {
#[cfg(not(feature = "dtype-datetime"))]
{
panic!("activate feature dtype-datetime")
}
#[cfg(feature = "dtype-datetime")]
match self.dtype() {
DataType::Int64 => self
.i64()
.unwrap()
.clone()
.into_datetime(timeunit, tz)
.into_series(),
DataType::Datetime(_, _) => self
.datetime()
.unwrap()
.as_ref()
.clone()
.into_datetime(timeunit, tz)
.into_series(),
dt => panic!("into_datetime not implemented for {dt:?}"),
}
}
pub(crate) fn into_duration(self, timeunit: TimeUnit) -> Series {
#[cfg(not(feature = "dtype-duration"))]
{
panic!("activate feature dtype-duration")
}
#[cfg(feature = "dtype-duration")]
match self.dtype() {
DataType::Int64 => self
.i64()
.unwrap()
.clone()
.into_duration(timeunit)
.into_series(),
DataType::Duration(_) => self
.duration()
.unwrap()
.as_ref()
.clone()
.into_duration(timeunit)
.into_series(),
dt => panic!("into_duration not implemented for {dt:?}"),
}
}
#[cfg(feature = "abs")]
#[cfg_attr(docsrs, doc(cfg(feature = "abs")))]
/// convert numerical values to their absolute value
pub fn abs(&self) -> PolarsResult<Series> {
let a = self.to_physical_repr();
use DataType::*;
let out = match a.dtype() {
#[cfg(feature = "dtype-i8")]
Int8 => a.i8().unwrap().abs().into_series(),
#[cfg(feature = "dtype-i16")]
Int16 => a.i16().unwrap().abs().into_series(),
Int32 => a.i32().unwrap().abs().into_series(),
Int64 => a.i64().unwrap().abs().into_series(),
UInt8 | UInt16 | UInt32 | UInt64 => self.clone(),
Float32 => a.f32().unwrap().abs().into_series(),
Float64 => a.f64().unwrap().abs().into_series(),
dt => {
return Err(PolarsError::InvalidOperation(
format!("abs not supported for series of type {dt:?}").into(),
));
}
};
Ok(out)
}More examples
src/frame/groupby/aggregations/dispatch.rs (line 114)
109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184
pub unsafe fn agg_median(&self, groups: &GroupsProxy) -> Series {
use DataType::*;
match self.dtype() {
Float32 => SeriesWrap(self.f32().unwrap().clone()).agg_median(groups),
Float64 => SeriesWrap(self.f64().unwrap().clone()).agg_median(groups),
dt if dt.is_numeric() || dt.is_temporal() => {
let ca = self.to_physical_repr();
let physical_type = ca.dtype();
let s = apply_method_physical_integer!(ca, agg_median, groups);
if dt.is_logical() {
// back to physical and then
// back to logical type
s.cast(physical_type).unwrap().cast(dt).unwrap()
} else {
s
}
}
_ => Series::full_null("", groups.len(), self.dtype()),
}
}
#[doc(hidden)]
pub unsafe fn agg_quantile(
&self,
groups: &GroupsProxy,
quantile: f64,
interpol: QuantileInterpolOptions,
) -> Series {
use DataType::*;
match self.dtype() {
Float32 => {
SeriesWrap(self.f32().unwrap().clone()).agg_quantile(groups, quantile, interpol)
}
Float64 => {
SeriesWrap(self.f64().unwrap().clone()).agg_quantile(groups, quantile, interpol)
}
dt if dt.is_numeric() || dt.is_temporal() => {
let ca = self.to_physical_repr();
let physical_type = ca.dtype();
let s =
apply_method_physical_integer!(ca, agg_quantile, groups, quantile, interpol);
if dt.is_logical() {
// back to physical and then
// back to logical type
s.cast(physical_type).unwrap().cast(dt).unwrap()
} else {
s
}
}
_ => Series::full_null("", groups.len(), self.dtype()),
}
}
#[doc(hidden)]
pub unsafe fn agg_mean(&self, groups: &GroupsProxy) -> Series {
use DataType::*;
match self.dtype() {
Boolean => self.cast(&Float64).unwrap().agg_mean(groups),
Float32 => SeriesWrap(self.f32().unwrap().clone()).agg_mean(groups),
Float64 => SeriesWrap(self.f64().unwrap().clone()).agg_mean(groups),
dt if dt.is_numeric() => {
apply_method_physical_integer!(self, agg_mean, groups)
}
dt @ Duration(_) => {
let s = self.to_physical_repr();
// agg_mean returns Float64
let out = s.agg_mean(groups);
// cast back to Int64 and then to logical duration type
out.cast(&Int64).unwrap().cast(dt).unwrap()
}
_ => Series::full_null("", groups.len(), self.dtype()),
}
}src/series/ops/round.rs (line 17)
6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59
pub fn round(&self, decimals: u32) -> PolarsResult<Self> {
use num::traits::Pow;
if let Ok(ca) = self.f32() {
// Note we do the computation on f64 floats to not loose precision
// when the computation is done, we cast to f32
let multiplier = 10.0.pow(decimals as f64);
let s = ca
.apply(|val| ((val as f64 * multiplier).round() / multiplier) as f32)
.into_series();
return Ok(s);
}
if let Ok(ca) = self.f64() {
let multiplier = 10.0.pow(decimals as f64);
let s = ca
.apply(|val| (val * multiplier).round() / multiplier)
.into_series();
return Ok(s);
}
Err(PolarsError::SchemaMisMatch(
format!("{:?} is not a floating point datatype", self.dtype()).into(),
))
}
#[cfg_attr(docsrs, doc(cfg(feature = "round_series")))]
/// Floor underlying floating point array to the lowest integers smaller or equal to the float value.
pub fn floor(&self) -> PolarsResult<Self> {
if let Ok(ca) = self.f32() {
let s = ca.apply(|val| val.floor()).into_series();
return Ok(s);
}
if let Ok(ca) = self.f64() {
let s = ca.apply(|val| val.floor()).into_series();
return Ok(s);
}
Err(PolarsError::SchemaMisMatch(
format!("{:?} is not a floating point datatype", self.dtype()).into(),
))
}
#[cfg_attr(docsrs, doc(cfg(feature = "round_series")))]
/// Ceil underlying floating point array to the highest integers smaller or equal to the float value.
pub fn ceil(&self) -> PolarsResult<Self> {
if let Ok(ca) = self.f32() {
let s = ca.apply(|val| val.ceil()).into_series();
return Ok(s);
}
if let Ok(ca) = self.f64() {
let s = ca.apply(|val| val.ceil()).into_series();
return Ok(s);
}
Err(PolarsError::SchemaMisMatch(
format!("{:?} is not a floating point datatype", self.dtype()).into(),
))
}src/chunked_array/ndarray.rs (line 115)
100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163
pub fn to_ndarray<N>(&self) -> PolarsResult<Array2<N::Native>>
where
N: PolarsNumericType,
{
let columns = self
.get_columns()
.par_iter()
.map(|s| {
let s = s.cast(&N::get_dtype())?;
let s = match s.dtype() {
DataType::Float32 => {
let ca = s.f32().unwrap();
ca.none_to_nan().into_series()
}
DataType::Float64 => {
let ca = s.f64().unwrap();
ca.none_to_nan().into_series()
}
_ => s,
};
Ok(s.rechunk())
})
.collect::<PolarsResult<Vec<_>>>()?;
let shape = self.shape();
let height = self.height();
let mut membuf = Vec::with_capacity(shape.0 * shape.1);
let ptr = membuf.as_ptr() as usize;
columns.par_iter().enumerate().map(|(col_idx, s)| {
if s.null_count() != 0 {
return Err(PolarsError::ComputeError(
"Creation of ndarray with null values is not supported. Consider using floats and NaNs".into(),
));
}
// this is an Arc clone if already of type N
let s = s.cast(&N::get_dtype())?;
let ca = s.unpack::<N>()?;
let vals = ca.cont_slice().unwrap();
// Safety:
// we get parallel access to the vector
// but we make sure that we don't get aliased access by offsetting the column indices + length
unsafe {
let offset_ptr = (ptr as *mut N::Native).add(col_idx * height) ;
// Safety:
// this is uninitialized memory, so we must never read from this data
// copy_from_slice does not read
let buf = std::slice::from_raw_parts_mut(offset_ptr, height);
buf.copy_from_slice(vals)
}
Ok(())
}).collect::<PolarsResult<Vec<_>>>()?;
// Safety:
// we have written all data, so we can now safely set length
unsafe {
membuf.set_len(shape.0 * shape.1);
}
let ndarr = Array2::from_shape_vec((shape.1, shape.0), membuf).unwrap();
Ok(ndarr.reversed_axes())
}src/frame/asof_join/mod.rs (line 151)
112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195
pub fn _join_asof(
&self,
other: &DataFrame,
left_on: &str,
right_on: &str,
strategy: AsofStrategy,
tolerance: Option<AnyValue<'static>>,
suffix: Option<String>,
slice: Option<(i64, usize)>,
) -> PolarsResult<DataFrame> {
let left_key = self.column(left_on)?;
let right_key = other.column(right_on)?;
check_asof_columns(left_key, right_key)?;
let left_key = left_key.to_physical_repr();
let right_key = right_key.to_physical_repr();
let take_idx = match left_key.dtype() {
DataType::Int64 => left_key
.i64()
.unwrap()
.join_asof(&right_key, strategy, tolerance),
DataType::Int32 => left_key
.i32()
.unwrap()
.join_asof(&right_key, strategy, tolerance),
DataType::UInt64 => left_key
.u64()
.unwrap()
.join_asof(&right_key, strategy, tolerance),
DataType::UInt32 => left_key
.u32()
.unwrap()
.join_asof(&right_key, strategy, tolerance),
DataType::Float32 => left_key
.f32()
.unwrap()
.join_asof(&right_key, strategy, tolerance),
DataType::Float64 => left_key
.f64()
.unwrap()
.join_asof(&right_key, strategy, tolerance),
_ => {
let left_key = left_key.cast(&DataType::Int32).unwrap();
let right_key = right_key.cast(&DataType::Int32).unwrap();
left_key
.i32()
.unwrap()
.join_asof(&right_key, strategy, tolerance)
}
}?;
// take_idx are sorted so this is a bound check for all
if let Some(Some(idx)) = take_idx.last() {
assert!((*idx as usize) < other.height())
}
// drop right join column
let other = if left_on == right_on {
Cow::Owned(other.drop(right_on)?)
} else {
Cow::Borrowed(other)
};
let mut left = self.clone();
let mut take_idx = &*take_idx;
if let Some((offset, len)) = slice {
left = left.slice(offset, len);
take_idx = slice_slice(take_idx, offset, len);
}
// Safety:
// join tuples are in bounds
let right_df = unsafe {
other.take_opt_iter_unchecked(
take_idx
.iter()
.map(|opt_idx| opt_idx.map(|idx| idx as usize)),
)
};
_finish_join(left, right_df, suffix.as_deref())
}src/series/arithmetic/borrowed.rs (line 284)
220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300
fn checked_div_num<T: ToPrimitive>(&self, rhs: T) -> PolarsResult<Series> {
use DataType::*;
let s = self.to_physical_repr();
let out = match s.dtype() {
#[cfg(feature = "dtype-u8")]
UInt8 => s
.u8()
.unwrap()
.apply_on_opt(|opt_v| opt_v.and_then(|v| v.checked_div(rhs.to_u8().unwrap())))
.into_series(),
#[cfg(feature = "dtype-i8")]
Int8 => s
.i8()
.unwrap()
.apply_on_opt(|opt_v| opt_v.and_then(|v| v.checked_div(rhs.to_i8().unwrap())))
.into_series(),
#[cfg(feature = "dtype-i16")]
Int16 => s
.i16()
.unwrap()
.apply_on_opt(|opt_v| opt_v.and_then(|v| v.checked_div(rhs.to_i16().unwrap())))
.into_series(),
#[cfg(feature = "dtype-u16")]
UInt16 => s
.u16()
.unwrap()
.apply_on_opt(|opt_v| opt_v.and_then(|v| v.checked_div(rhs.to_u16().unwrap())))
.into_series(),
UInt32 => s
.u32()
.unwrap()
.apply_on_opt(|opt_v| opt_v.and_then(|v| v.checked_div(rhs.to_u32().unwrap())))
.into_series(),
Int32 => s
.i32()
.unwrap()
.apply_on_opt(|opt_v| opt_v.and_then(|v| v.checked_div(rhs.to_i32().unwrap())))
.into_series(),
UInt64 => s
.u64()
.unwrap()
.apply_on_opt(|opt_v| opt_v.and_then(|v| v.checked_div(rhs.to_u64().unwrap())))
.into_series(),
Int64 => s
.i64()
.unwrap()
.apply_on_opt(|opt_v| opt_v.and_then(|v| v.checked_div(rhs.to_i64().unwrap())))
.into_series(),
Float32 => s
.f32()
.unwrap()
.apply_on_opt(|opt_v| {
opt_v.and_then(|v| {
let res = rhs.to_f32().unwrap();
if res.is_zero() {
None
} else {
Some(v / res)
}
})
})
.into_series(),
Float64 => s
.f64()
.unwrap()
.apply_on_opt(|opt_v| {
opt_v.and_then(|v| {
let res = rhs.to_f64().unwrap();
if res.is_zero() {
None
} else {
Some(v / res)
}
})
})
.into_series(),
_ => panic!("dtype not yet supported in checked div"),
};
out.cast(self.dtype())
}Additional examples can be found in:
sourcepub fn u8(&self) -> PolarsResult<&UInt8Chunked>
pub fn u8(&self) -> PolarsResult<&UInt8Chunked>
Unpack to ChunkedArray of dtype u8
Examples found in repository?
src/fmt.rs (line 215)
206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284
fn fmt(&self, f: &mut Formatter<'_>) -> fmt::Result {
match self.dtype() {
DataType::Boolean => {
format_array!(f, self.bool().unwrap(), "bool", self.name(), "Series")
}
DataType::Utf8 => {
format_array!(f, self.utf8().unwrap(), "str", self.name(), "Series")
}
DataType::UInt8 => {
format_array!(f, self.u8().unwrap(), "u8", self.name(), "Series")
}
DataType::UInt16 => {
format_array!(f, self.u16().unwrap(), "u16", self.name(), "Series")
}
DataType::UInt32 => {
format_array!(f, self.u32().unwrap(), "u32", self.name(), "Series")
}
DataType::UInt64 => {
format_array!(f, self.u64().unwrap(), "u64", self.name(), "Series")
}
DataType::Int8 => {
format_array!(f, self.i8().unwrap(), "i8", self.name(), "Series")
}
DataType::Int16 => {
format_array!(f, self.i16().unwrap(), "i16", self.name(), "Series")
}
DataType::Int32 => {
format_array!(f, self.i32().unwrap(), "i32", self.name(), "Series")
}
DataType::Int64 => {
format_array!(f, self.i64().unwrap(), "i64", self.name(), "Series")
}
DataType::Float32 => {
format_array!(f, self.f32().unwrap(), "f32", self.name(), "Series")
}
DataType::Float64 => {
format_array!(f, self.f64().unwrap(), "f64", self.name(), "Series")
}
#[cfg(feature = "dtype-date")]
DataType::Date => format_array!(f, self.date().unwrap(), "date", self.name(), "Series"),
#[cfg(feature = "dtype-datetime")]
DataType::Datetime(_, _) => {
let dt = format!("{}", self.dtype());
format_array!(f, self.datetime().unwrap(), &dt, self.name(), "Series")
}
#[cfg(feature = "dtype-time")]
DataType::Time => format_array!(f, self.time().unwrap(), "time", self.name(), "Series"),
#[cfg(feature = "dtype-duration")]
DataType::Duration(_) => {
let dt = format!("{}", self.dtype());
format_array!(f, self.duration().unwrap(), &dt, self.name(), "Series")
}
DataType::List(_) => {
format_array!(f, self.list().unwrap(), "list", self.name(), "Series")
}
#[cfg(feature = "object")]
DataType::Object(_) => format_object_array(f, self, self.name(), "Series"),
#[cfg(feature = "dtype-categorical")]
DataType::Categorical(_) => {
format_array!(f, self.categorical().unwrap(), "cat", self.name(), "Series")
}
#[cfg(feature = "dtype-struct")]
dt @ DataType::Struct(_) => format_array!(
f,
self.struct_().unwrap(),
format!("{dt}"),
self.name(),
"Series"
),
DataType::Null => {
writeln!(f, "nullarray")
}
#[cfg(feature = "dtype-binary")]
DataType::Binary => {
format_array!(f, self.binary().unwrap(), "binary", self.name(), "Series")
}
dt => panic!("{dt:?} not impl"),
}
}sourcepub fn u16(&self) -> PolarsResult<&UInt16Chunked>
pub fn u16(&self) -> PolarsResult<&UInt16Chunked>
Unpack to ChunkedArray of dtype u16
Examples found in repository?
src/fmt.rs (line 218)
206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284
fn fmt(&self, f: &mut Formatter<'_>) -> fmt::Result {
match self.dtype() {
DataType::Boolean => {
format_array!(f, self.bool().unwrap(), "bool", self.name(), "Series")
}
DataType::Utf8 => {
format_array!(f, self.utf8().unwrap(), "str", self.name(), "Series")
}
DataType::UInt8 => {
format_array!(f, self.u8().unwrap(), "u8", self.name(), "Series")
}
DataType::UInt16 => {
format_array!(f, self.u16().unwrap(), "u16", self.name(), "Series")
}
DataType::UInt32 => {
format_array!(f, self.u32().unwrap(), "u32", self.name(), "Series")
}
DataType::UInt64 => {
format_array!(f, self.u64().unwrap(), "u64", self.name(), "Series")
}
DataType::Int8 => {
format_array!(f, self.i8().unwrap(), "i8", self.name(), "Series")
}
DataType::Int16 => {
format_array!(f, self.i16().unwrap(), "i16", self.name(), "Series")
}
DataType::Int32 => {
format_array!(f, self.i32().unwrap(), "i32", self.name(), "Series")
}
DataType::Int64 => {
format_array!(f, self.i64().unwrap(), "i64", self.name(), "Series")
}
DataType::Float32 => {
format_array!(f, self.f32().unwrap(), "f32", self.name(), "Series")
}
DataType::Float64 => {
format_array!(f, self.f64().unwrap(), "f64", self.name(), "Series")
}
#[cfg(feature = "dtype-date")]
DataType::Date => format_array!(f, self.date().unwrap(), "date", self.name(), "Series"),
#[cfg(feature = "dtype-datetime")]
DataType::Datetime(_, _) => {
let dt = format!("{}", self.dtype());
format_array!(f, self.datetime().unwrap(), &dt, self.name(), "Series")
}
#[cfg(feature = "dtype-time")]
DataType::Time => format_array!(f, self.time().unwrap(), "time", self.name(), "Series"),
#[cfg(feature = "dtype-duration")]
DataType::Duration(_) => {
let dt = format!("{}", self.dtype());
format_array!(f, self.duration().unwrap(), &dt, self.name(), "Series")
}
DataType::List(_) => {
format_array!(f, self.list().unwrap(), "list", self.name(), "Series")
}
#[cfg(feature = "object")]
DataType::Object(_) => format_object_array(f, self, self.name(), "Series"),
#[cfg(feature = "dtype-categorical")]
DataType::Categorical(_) => {
format_array!(f, self.categorical().unwrap(), "cat", self.name(), "Series")
}
#[cfg(feature = "dtype-struct")]
dt @ DataType::Struct(_) => format_array!(
f,
self.struct_().unwrap(),
format!("{dt}"),
self.name(),
"Series"
),
DataType::Null => {
writeln!(f, "nullarray")
}
#[cfg(feature = "dtype-binary")]
DataType::Binary => {
format_array!(f, self.binary().unwrap(), "binary", self.name(), "Series")
}
dt => panic!("{dt:?} not impl"),
}
}sourcepub fn u32(&self) -> PolarsResult<&UInt32Chunked>
pub fn u32(&self) -> PolarsResult<&UInt32Chunked>
Unpack to ChunkedArray of dtype u32
Examples found in repository?
More examples
src/series/implementations/categorical.rs (line 72)
69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137
fn explode_by_offsets(&self, offsets: &[i64]) -> Series {
// TODO! explode by offset should return concrete type
self.with_state(true, |cats| {
cats.explode_by_offsets(offsets).u32().unwrap().clone()
})
.into_series()
}
fn _set_sorted(&mut self, is_sorted: IsSorted) {
self.0.logical_mut().set_sorted2(is_sorted)
}
unsafe fn equal_element(&self, idx_self: usize, idx_other: usize, other: &Series) -> bool {
self.0.logical().equal_element(idx_self, idx_other, other)
}
#[cfg(feature = "zip_with")]
fn zip_with_same_type(&self, mask: &BooleanChunked, other: &Series) -> PolarsResult<Series> {
self.0
.zip_with(mask, other.categorical()?)
.map(|ca| ca.into_series())
}
fn into_partial_ord_inner<'a>(&'a self) -> Box<dyn PartialOrdInner + 'a> {
(&self.0).into_partial_ord_inner()
}
fn vec_hash(&self, random_state: RandomState, buf: &mut Vec<u64>) -> PolarsResult<()> {
self.0.logical().vec_hash(random_state, buf);
Ok(())
}
fn vec_hash_combine(&self, build_hasher: RandomState, hashes: &mut [u64]) -> PolarsResult<()> {
self.0.logical().vec_hash_combine(build_hasher, hashes);
Ok(())
}
unsafe fn agg_list(&self, groups: &GroupsProxy) -> Series {
// we cannot cast and dispatch as the inner type of the list would be incorrect
self.0
.logical()
.agg_list(groups)
.cast(&DataType::List(Box::new(self.dtype().clone())))
.unwrap()
}
fn zip_outer_join_column(
&self,
right_column: &Series,
opt_join_tuples: &[(Option<IdxSize>, Option<IdxSize>)],
) -> Series {
let new_rev_map = self
.0
.merge_categorical_map(right_column.categorical().unwrap())
.unwrap();
let left = self.0.logical();
let right = right_column
.categorical()
.unwrap()
.logical()
.clone()
.into_series();
let cats = left.zip_outer_join_column(&right, opt_join_tuples);
let cats = cats.u32().unwrap().clone();
unsafe {
CategoricalChunked::from_cats_and_rev_map_unchecked(cats, new_rev_map).into_series()
}
}src/chunked_array/logical/categorical/ops/unique.rs (line 52)
45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65
pub fn value_counts(&self) -> PolarsResult<DataFrame> {
let groups = self.logical().group_tuples(true, false).unwrap();
let logical_values = unsafe {
self.logical()
.clone()
.into_series()
.agg_first(&groups)
.u32()
.unwrap()
.clone()
};
let mut values = self.clone();
*values.logical_mut() = logical_values;
let mut counts = groups.group_count();
counts.rename("counts");
let cols = vec![values.into_series(), counts.into_series()];
let df = DataFrame::new_no_checks(cols);
df.sort(["counts"], true)
}src/series/mod.rs (line 580)
565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751 752 753 754 755 756 757 758 759 760 761 762 763 764 765 766 767 768 769 770 771 772 773 774 775 776 777 778 779 780 781 782 783 784 785 786 787 788 789 790 791 792 793 794 795 796 797 798 799 800 801 802 803 804 805 806 807 808 809 810 811 812 813 814 815 816 817 818 819 820 821 822 823 824 825 826 827 828 829 830 831 832 833 834 835 836 837 838 839 840 841 842 843 844 845 846 847 848 849 850 851 852 853 854 855 856 857 858 859 860 861 862 863 864 865 866 867 868 869 870 871 872 873 874 875 876 877 878 879 880 881 882 883 884 885 886 887 888 889 890
pub fn cumsum(&self, reverse: bool) -> Series {
#[cfg(feature = "cum_agg")]
{
use DataType::*;
match self.dtype() {
Boolean => self.cast(&DataType::UInt32).unwrap().cumsum(reverse),
Int8 | UInt8 | Int16 | UInt16 => {
let s = self.cast(&Int64).unwrap();
s.cumsum(reverse)
}
Int32 => {
let ca = self.i32().unwrap();
ca.cumsum(reverse).into_series()
}
UInt32 => {
let ca = self.u32().unwrap();
ca.cumsum(reverse).into_series()
}
UInt64 => {
let ca = self.u64().unwrap();
ca.cumsum(reverse).into_series()
}
Int64 => {
let ca = self.i64().unwrap();
ca.cumsum(reverse).into_series()
}
Float32 => {
let ca = self.f32().unwrap();
ca.cumsum(reverse).into_series()
}
Float64 => {
let ca = self.f64().unwrap();
ca.cumsum(reverse).into_series()
}
dt => panic!("cumsum not supported for dtype: {dt:?}"),
}
}
#[cfg(not(feature = "cum_agg"))]
{
panic!("activate 'cum_agg' feature")
}
}
/// Get an array with the cumulative product computed at every element
///
/// If the [`DataType`] is one of `{Int8, UInt8, Int16, UInt16, Int32, UInt32}` the `Series` is
/// first cast to `Int64` to prevent overflow issues.
#[cfg_attr(docsrs, doc(cfg(feature = "cum_agg")))]
#[allow(unused_variables)]
pub fn cumprod(&self, reverse: bool) -> Series {
#[cfg(feature = "cum_agg")]
{
use DataType::*;
match self.dtype() {
Boolean => self.cast(&DataType::Int64).unwrap().cumprod(reverse),
Int8 | UInt8 | Int16 | UInt16 | Int32 | UInt32 => {
let s = self.cast(&Int64).unwrap();
s.cumprod(reverse)
}
Int64 => {
let ca = self.i64().unwrap();
ca.cumprod(reverse).into_series()
}
UInt64 => {
let ca = self.u64().unwrap();
ca.cumprod(reverse).into_series()
}
Float32 => {
let ca = self.f32().unwrap();
ca.cumprod(reverse).into_series()
}
Float64 => {
let ca = self.f64().unwrap();
ca.cumprod(reverse).into_series()
}
dt => panic!("cumprod not supported for dtype: {dt:?}"),
}
}
#[cfg(not(feature = "cum_agg"))]
{
panic!("activate 'cum_agg' feature")
}
}
/// Get the product of an array.
///
/// If the [`DataType`] is one of `{Int8, UInt8, Int16, UInt16}` the `Series` is
/// first cast to `Int64` to prevent overflow issues.
#[cfg_attr(docsrs, doc(cfg(feature = "product")))]
pub fn product(&self) -> Series {
#[cfg(feature = "product")]
{
use DataType::*;
match self.dtype() {
Boolean => self.cast(&DataType::Int64).unwrap().product(),
Int8 | UInt8 | Int16 | UInt16 => {
let s = self.cast(&Int64).unwrap();
s.product()
}
Int64 => {
let ca = self.i64().unwrap();
ca.prod_as_series()
}
Float32 => {
let ca = self.f32().unwrap();
ca.prod_as_series()
}
Float64 => {
let ca = self.f64().unwrap();
ca.prod_as_series()
}
dt => panic!("cumprod not supported for dtype: {dt:?}"),
}
}
#[cfg(not(feature = "product"))]
{
panic!("activate 'product' feature")
}
}
#[cfg(feature = "rank")]
#[cfg_attr(docsrs, doc(cfg(feature = "rank")))]
pub fn rank(&self, options: RankOptions) -> Series {
rank(self, options.method, options.descending)
}
/// Cast throws an error if conversion had overflows
pub fn strict_cast(&self, data_type: &DataType) -> PolarsResult<Series> {
let s = self.cast(data_type)?;
if self.null_count() != s.null_count() {
let failure_mask = !self.is_null() & s.is_null();
let failures = self.filter_threaded(&failure_mask, false)?.unique()?;
Err(PolarsError::ComputeError(
format!(
"Strict conversion from {:?} to {:?} failed for values {}. \
If you were trying to cast Utf8 to Date, Time, or Datetime, \
consider using `strptime`.",
self.dtype(),
data_type,
failures.fmt_list(),
)
.into(),
))
} else {
Ok(s)
}
}
#[cfg(feature = "dtype-time")]
pub(crate) fn into_time(self) -> Series {
#[cfg(not(feature = "dtype-time"))]
{
panic!("activate feature dtype-time")
}
match self.dtype() {
DataType::Int64 => self.i64().unwrap().clone().into_time().into_series(),
DataType::Time => self
.time()
.unwrap()
.as_ref()
.clone()
.into_time()
.into_series(),
dt => panic!("date not implemented for {dt:?}"),
}
}
pub(crate) fn into_date(self) -> Series {
#[cfg(not(feature = "dtype-date"))]
{
panic!("activate feature dtype-date")
}
#[cfg(feature = "dtype-date")]
match self.dtype() {
DataType::Int32 => self.i32().unwrap().clone().into_date().into_series(),
DataType::Date => self
.date()
.unwrap()
.as_ref()
.clone()
.into_date()
.into_series(),
dt => panic!("date not implemented for {dt:?}"),
}
}
pub(crate) fn into_datetime(self, timeunit: TimeUnit, tz: Option<TimeZone>) -> Series {
#[cfg(not(feature = "dtype-datetime"))]
{
panic!("activate feature dtype-datetime")
}
#[cfg(feature = "dtype-datetime")]
match self.dtype() {
DataType::Int64 => self
.i64()
.unwrap()
.clone()
.into_datetime(timeunit, tz)
.into_series(),
DataType::Datetime(_, _) => self
.datetime()
.unwrap()
.as_ref()
.clone()
.into_datetime(timeunit, tz)
.into_series(),
dt => panic!("into_datetime not implemented for {dt:?}"),
}
}
pub(crate) fn into_duration(self, timeunit: TimeUnit) -> Series {
#[cfg(not(feature = "dtype-duration"))]
{
panic!("activate feature dtype-duration")
}
#[cfg(feature = "dtype-duration")]
match self.dtype() {
DataType::Int64 => self
.i64()
.unwrap()
.clone()
.into_duration(timeunit)
.into_series(),
DataType::Duration(_) => self
.duration()
.unwrap()
.as_ref()
.clone()
.into_duration(timeunit)
.into_series(),
dt => panic!("into_duration not implemented for {dt:?}"),
}
}
#[cfg(feature = "abs")]
#[cfg_attr(docsrs, doc(cfg(feature = "abs")))]
/// convert numerical values to their absolute value
pub fn abs(&self) -> PolarsResult<Series> {
let a = self.to_physical_repr();
use DataType::*;
let out = match a.dtype() {
#[cfg(feature = "dtype-i8")]
Int8 => a.i8().unwrap().abs().into_series(),
#[cfg(feature = "dtype-i16")]
Int16 => a.i16().unwrap().abs().into_series(),
Int32 => a.i32().unwrap().abs().into_series(),
Int64 => a.i64().unwrap().abs().into_series(),
UInt8 | UInt16 | UInt32 | UInt64 => self.clone(),
Float32 => a.f32().unwrap().abs().into_series(),
Float64 => a.f64().unwrap().abs().into_series(),
dt => {
return Err(PolarsError::InvalidOperation(
format!("abs not supported for series of type {dt:?}").into(),
));
}
};
Ok(out)
}
#[cfg(feature = "private")]
// used for formatting
pub fn str_value(&self, index: usize) -> PolarsResult<Cow<str>> {
let out = match self.0.get(index)? {
AnyValue::Utf8(s) => Cow::Borrowed(s),
AnyValue::Null => Cow::Borrowed("null"),
#[cfg(feature = "dtype-categorical")]
AnyValue::Categorical(idx, rev) => Cow::Borrowed(rev.get(idx)),
av => Cow::Owned(format!("{av}")),
};
Ok(out)
}
/// Get the head of the Series.
pub fn head(&self, length: Option<usize>) -> Series {
match length {
Some(len) => self.slice(0, std::cmp::min(len, self.len())),
None => self.slice(0, std::cmp::min(10, self.len())),
}
}
/// Get the tail of the Series.
pub fn tail(&self, length: Option<usize>) -> Series {
let len = match length {
Some(len) => std::cmp::min(len, self.len()),
None => std::cmp::min(10, self.len()),
};
self.slice(-(len as i64), len)
}
pub fn mean_as_series(&self) -> Series {
match self.dtype() {
DataType::Float32 => {
let val = &[self.mean().map(|m| m as f32)];
Series::new(self.name(), val)
}
dt if dt.is_numeric() || matches!(dt, DataType::Boolean) => {
let val = &[self.mean()];
Series::new(self.name(), val)
}
dt @ DataType::Duration(_) => {
Series::new(self.name(), &[self.mean().map(|v| v as i64)])
.cast(dt)
.unwrap()
}
_ => return Series::full_null(self.name(), 1, self.dtype()),
}
}
/// Compute the unique elements, but maintain order. This requires more work
/// than a naive [`Series::unique`](SeriesTrait::unique).
pub fn unique_stable(&self) -> PolarsResult<Series> {
let idx = self.arg_unique()?;
// Safety:
// Indices are in bounds.
unsafe { self.take_unchecked(&idx) }
}
pub fn idx(&self) -> PolarsResult<&IdxCa> {
#[cfg(feature = "bigidx")]
{
self.u64()
}
#[cfg(not(feature = "bigidx"))]
{
self.u32()
}
}src/chunked_array/ops/bit_repr.rs (line 143)
109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264
fn bit_repr_small(&self) -> UInt32Chunked {
if std::mem::size_of::<T::Native>() == 4 {
if matches!(self.dtype(), DataType::UInt32) {
let ca = self.clone();
// convince the compiler we are this type. This keeps flags
return unsafe { std::mem::transmute(ca) };
}
let chunks = self
.downcast_iter()
.map(|array| {
let buf = array.values().clone();
// Safety:
// we just check the size of T::Native to be 32 bits
// The fields can still be reordered between generic types
// so we do some extra assertions
let len = buf.len();
let offset = buf.offset();
let ptr = buf.as_slice().as_ptr() as usize;
#[allow(clippy::transmute_undefined_repr)]
let reinterpreted_buf = unsafe { std::mem::transmute::<_, Buffer<u32>>(buf) };
assert_eq!(reinterpreted_buf.len(), len);
assert_eq!(reinterpreted_buf.offset(), offset);
assert_eq!(reinterpreted_buf.as_slice().as_ptr() as usize, ptr);
Box::new(PrimitiveArray::new(
ArrowDataType::UInt32,
reinterpreted_buf,
array.validity().cloned(),
)) as ArrayRef
})
.collect::<Vec<_>>();
UInt32Chunked::from_chunks(self.name(), chunks)
} else {
self.cast_unchecked(&DataType::UInt32)
.unwrap()
.u32()
.unwrap()
.clone()
}
}
}
#[cfg(feature = "reinterpret")]
impl Reinterpret for UInt64Chunked {
fn reinterpret_signed(&self) -> Series {
let chunks = self
.downcast_iter()
.map(|array| {
let buf = array.values().clone();
// Safety
// same bit length u64 <-> i64
// The fields can still be reordered between generic types
// so we do some extra assertions
let len = buf.len();
let offset = buf.offset();
let ptr = buf.as_slice().as_ptr() as usize;
#[allow(clippy::transmute_undefined_repr)]
let reinterpreted_buf = unsafe { std::mem::transmute::<_, Buffer<i64>>(buf) };
assert_eq!(reinterpreted_buf.len(), len);
assert_eq!(reinterpreted_buf.offset(), offset);
assert_eq!(reinterpreted_buf.as_slice().as_ptr() as usize, ptr);
Box::new(PrimitiveArray::new(
ArrowDataType::Int64,
reinterpreted_buf,
array.validity().cloned(),
)) as ArrayRef
})
.collect::<Vec<_>>();
Int64Chunked::from_chunks(self.name(), chunks).into_series()
}
fn reinterpret_unsigned(&self) -> Series {
self.clone().into_series()
}
}
#[cfg(feature = "reinterpret")]
impl Reinterpret for Int64Chunked {
fn reinterpret_signed(&self) -> Series {
self.clone().into_series()
}
fn reinterpret_unsigned(&self) -> Series {
self.bit_repr_large().into_series()
}
}
impl UInt64Chunked {
#[doc(hidden)]
pub fn _reinterpret_float(&self) -> Float64Chunked {
let chunks = self
.downcast_iter()
.map(|array| {
let buf = array.values().clone();
// Safety
// same bit length u64 <-> f64
// The fields can still be reordered between generic types
// so we do some extra assertions
let len = buf.len();
let offset = buf.offset();
let ptr = buf.as_slice().as_ptr() as usize;
#[allow(clippy::transmute_undefined_repr)]
let reinterpreted_buf = unsafe { std::mem::transmute::<_, Buffer<f64>>(buf) };
assert_eq!(reinterpreted_buf.len(), len);
assert_eq!(reinterpreted_buf.offset(), offset);
assert_eq!(reinterpreted_buf.as_slice().as_ptr() as usize, ptr);
Box::new(PrimitiveArray::new(
ArrowDataType::Float64,
reinterpreted_buf,
array.validity().cloned(),
)) as ArrayRef
})
.collect::<Vec<_>>();
Float64Chunked::from_chunks(self.name(), chunks)
}
}
impl UInt32Chunked {
#[doc(hidden)]
pub fn _reinterpret_float(&self) -> Float32Chunked {
let chunks = self
.downcast_iter()
.map(|array| {
let buf = array.values().clone();
// Safety
// same bit length u32 <-> f32
// The fields can still be reordered between generic types
// so we do some extra assertions
let len = buf.len();
let offset = buf.offset();
let ptr = buf.as_slice().as_ptr() as usize;
#[allow(clippy::transmute_undefined_repr)]
let reinterpreted_buf = unsafe { std::mem::transmute::<_, Buffer<f32>>(buf) };
assert_eq!(reinterpreted_buf.len(), len);
assert_eq!(reinterpreted_buf.offset(), offset);
assert_eq!(reinterpreted_buf.as_slice().as_ptr() as usize, ptr);
Box::new(PrimitiveArray::new(
ArrowDataType::Float32,
reinterpreted_buf,
array.validity().cloned(),
)) as ArrayRef
})
.collect::<Vec<_>>();
Float32Chunked::from_chunks(self.name(), chunks)
}
}
/// Used to save compilation paths. Use carefully. Although this is safe,
/// if misused it can lead to incorrect results.
impl Float32Chunked {
pub(crate) fn apply_as_ints<F>(&self, f: F) -> Series
where
F: Fn(&Series) -> Series,
{
let s = self.bit_repr_small().into_series();
let out = f(&s);
let out = out.u32().unwrap();
out._reinterpret_float().into()
}src/frame/mod.rs (line 2935)
2901 2902 2903 2904 2905 2906 2907 2908 2909 2910 2911 2912 2913 2914 2915 2916 2917 2918 2919 2920 2921 2922 2923 2924 2925 2926 2927 2928 2929 2930 2931 2932 2933 2934 2935 2936 2937 2938 2939 2940 2941 2942 2943 2944 2945 2946 2947
pub fn hmean(&self, none_strategy: NullStrategy) -> PolarsResult<Option<Series>> {
match self.columns.len() {
0 => Ok(None),
1 => Ok(Some(self.columns[0].clone())),
_ => {
let columns = self
.columns
.iter()
.cloned()
.filter(|s| {
let dtype = s.dtype();
dtype.is_numeric() || matches!(dtype, DataType::Boolean)
})
.collect();
let numeric_df = DataFrame::new_no_checks(columns);
let sum = || numeric_df.hsum(none_strategy);
let null_count = || {
numeric_df
.columns
.par_iter()
.map(|s| s.is_null().cast(&DataType::UInt32).unwrap())
.reduce_with(|l, r| &l + &r)
// we can unwrap the option, because we are certain there is a column
// we started this operation on 2 columns
.unwrap()
};
let (sum, null_count) = POOL.install(|| rayon::join(sum, null_count));
let sum = sum?;
// value lengths: len - null_count
let value_length: UInt32Chunked =
(numeric_df.width().sub(&null_count)).u32().unwrap().clone();
// make sure that we do not divide by zero
// by replacing with None
let value_length = value_length
.set(&value_length.equal(0), None)?
.into_series()
.cast(&DataType::Float64)?;
Ok(sum.map(|sum| &sum / &value_length))
}
}
}sourcepub fn u64(&self) -> PolarsResult<&UInt64Chunked>
pub fn u64(&self) -> PolarsResult<&UInt64Chunked>
Unpack to ChunkedArray of dtype u64
Examples found in repository?
More examples
src/series/mod.rs (line 584)
565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647
pub fn cumsum(&self, reverse: bool) -> Series {
#[cfg(feature = "cum_agg")]
{
use DataType::*;
match self.dtype() {
Boolean => self.cast(&DataType::UInt32).unwrap().cumsum(reverse),
Int8 | UInt8 | Int16 | UInt16 => {
let s = self.cast(&Int64).unwrap();
s.cumsum(reverse)
}
Int32 => {
let ca = self.i32().unwrap();
ca.cumsum(reverse).into_series()
}
UInt32 => {
let ca = self.u32().unwrap();
ca.cumsum(reverse).into_series()
}
UInt64 => {
let ca = self.u64().unwrap();
ca.cumsum(reverse).into_series()
}
Int64 => {
let ca = self.i64().unwrap();
ca.cumsum(reverse).into_series()
}
Float32 => {
let ca = self.f32().unwrap();
ca.cumsum(reverse).into_series()
}
Float64 => {
let ca = self.f64().unwrap();
ca.cumsum(reverse).into_series()
}
dt => panic!("cumsum not supported for dtype: {dt:?}"),
}
}
#[cfg(not(feature = "cum_agg"))]
{
panic!("activate 'cum_agg' feature")
}
}
/// Get an array with the cumulative product computed at every element
///
/// If the [`DataType`] is one of `{Int8, UInt8, Int16, UInt16, Int32, UInt32}` the `Series` is
/// first cast to `Int64` to prevent overflow issues.
#[cfg_attr(docsrs, doc(cfg(feature = "cum_agg")))]
#[allow(unused_variables)]
pub fn cumprod(&self, reverse: bool) -> Series {
#[cfg(feature = "cum_agg")]
{
use DataType::*;
match self.dtype() {
Boolean => self.cast(&DataType::Int64).unwrap().cumprod(reverse),
Int8 | UInt8 | Int16 | UInt16 | Int32 | UInt32 => {
let s = self.cast(&Int64).unwrap();
s.cumprod(reverse)
}
Int64 => {
let ca = self.i64().unwrap();
ca.cumprod(reverse).into_series()
}
UInt64 => {
let ca = self.u64().unwrap();
ca.cumprod(reverse).into_series()
}
Float32 => {
let ca = self.f32().unwrap();
ca.cumprod(reverse).into_series()
}
Float64 => {
let ca = self.f64().unwrap();
ca.cumprod(reverse).into_series()
}
dt => panic!("cumprod not supported for dtype: {dt:?}"),
}
}
#[cfg(not(feature = "cum_agg"))]
{
panic!("activate 'cum_agg' feature")
}
}src/frame/asof_join/mod.rs (line 139)
112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195
pub fn _join_asof(
&self,
other: &DataFrame,
left_on: &str,
right_on: &str,
strategy: AsofStrategy,
tolerance: Option<AnyValue<'static>>,
suffix: Option<String>,
slice: Option<(i64, usize)>,
) -> PolarsResult<DataFrame> {
let left_key = self.column(left_on)?;
let right_key = other.column(right_on)?;
check_asof_columns(left_key, right_key)?;
let left_key = left_key.to_physical_repr();
let right_key = right_key.to_physical_repr();
let take_idx = match left_key.dtype() {
DataType::Int64 => left_key
.i64()
.unwrap()
.join_asof(&right_key, strategy, tolerance),
DataType::Int32 => left_key
.i32()
.unwrap()
.join_asof(&right_key, strategy, tolerance),
DataType::UInt64 => left_key
.u64()
.unwrap()
.join_asof(&right_key, strategy, tolerance),
DataType::UInt32 => left_key
.u32()
.unwrap()
.join_asof(&right_key, strategy, tolerance),
DataType::Float32 => left_key
.f32()
.unwrap()
.join_asof(&right_key, strategy, tolerance),
DataType::Float64 => left_key
.f64()
.unwrap()
.join_asof(&right_key, strategy, tolerance),
_ => {
let left_key = left_key.cast(&DataType::Int32).unwrap();
let right_key = right_key.cast(&DataType::Int32).unwrap();
left_key
.i32()
.unwrap()
.join_asof(&right_key, strategy, tolerance)
}
}?;
// take_idx are sorted so this is a bound check for all
if let Some(Some(idx)) = take_idx.last() {
assert!((*idx as usize) < other.height())
}
// drop right join column
let other = if left_on == right_on {
Cow::Owned(other.drop(right_on)?)
} else {
Cow::Borrowed(other)
};
let mut left = self.clone();
let mut take_idx = &*take_idx;
if let Some((offset, len)) = slice {
left = left.slice(offset, len);
take_idx = slice_slice(take_idx, offset, len);
}
// Safety:
// join tuples are in bounds
let right_df = unsafe {
other.take_opt_iter_unchecked(
take_idx
.iter()
.map(|opt_idx| opt_idx.map(|idx| idx as usize)),
)
};
_finish_join(left, right_df, suffix.as_deref())
}src/series/arithmetic/borrowed.rs (line 260)
220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300
fn checked_div_num<T: ToPrimitive>(&self, rhs: T) -> PolarsResult<Series> {
use DataType::*;
let s = self.to_physical_repr();
let out = match s.dtype() {
#[cfg(feature = "dtype-u8")]
UInt8 => s
.u8()
.unwrap()
.apply_on_opt(|opt_v| opt_v.and_then(|v| v.checked_div(rhs.to_u8().unwrap())))
.into_series(),
#[cfg(feature = "dtype-i8")]
Int8 => s
.i8()
.unwrap()
.apply_on_opt(|opt_v| opt_v.and_then(|v| v.checked_div(rhs.to_i8().unwrap())))
.into_series(),
#[cfg(feature = "dtype-i16")]
Int16 => s
.i16()
.unwrap()
.apply_on_opt(|opt_v| opt_v.and_then(|v| v.checked_div(rhs.to_i16().unwrap())))
.into_series(),
#[cfg(feature = "dtype-u16")]
UInt16 => s
.u16()
.unwrap()
.apply_on_opt(|opt_v| opt_v.and_then(|v| v.checked_div(rhs.to_u16().unwrap())))
.into_series(),
UInt32 => s
.u32()
.unwrap()
.apply_on_opt(|opt_v| opt_v.and_then(|v| v.checked_div(rhs.to_u32().unwrap())))
.into_series(),
Int32 => s
.i32()
.unwrap()
.apply_on_opt(|opt_v| opt_v.and_then(|v| v.checked_div(rhs.to_i32().unwrap())))
.into_series(),
UInt64 => s
.u64()
.unwrap()
.apply_on_opt(|opt_v| opt_v.and_then(|v| v.checked_div(rhs.to_u64().unwrap())))
.into_series(),
Int64 => s
.i64()
.unwrap()
.apply_on_opt(|opt_v| opt_v.and_then(|v| v.checked_div(rhs.to_i64().unwrap())))
.into_series(),
Float32 => s
.f32()
.unwrap()
.apply_on_opt(|opt_v| {
opt_v.and_then(|v| {
let res = rhs.to_f32().unwrap();
if res.is_zero() {
None
} else {
Some(v / res)
}
})
})
.into_series(),
Float64 => s
.f64()
.unwrap()
.apply_on_opt(|opt_v| {
opt_v.and_then(|v| {
let res = rhs.to_f64().unwrap();
if res.is_zero() {
None
} else {
Some(v / res)
}
})
})
.into_series(),
_ => panic!("dtype not yet supported in checked div"),
};
out.cast(self.dtype())
}src/fmt.rs (line 224)
206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284
fn fmt(&self, f: &mut Formatter<'_>) -> fmt::Result {
match self.dtype() {
DataType::Boolean => {
format_array!(f, self.bool().unwrap(), "bool", self.name(), "Series")
}
DataType::Utf8 => {
format_array!(f, self.utf8().unwrap(), "str", self.name(), "Series")
}
DataType::UInt8 => {
format_array!(f, self.u8().unwrap(), "u8", self.name(), "Series")
}
DataType::UInt16 => {
format_array!(f, self.u16().unwrap(), "u16", self.name(), "Series")
}
DataType::UInt32 => {
format_array!(f, self.u32().unwrap(), "u32", self.name(), "Series")
}
DataType::UInt64 => {
format_array!(f, self.u64().unwrap(), "u64", self.name(), "Series")
}
DataType::Int8 => {
format_array!(f, self.i8().unwrap(), "i8", self.name(), "Series")
}
DataType::Int16 => {
format_array!(f, self.i16().unwrap(), "i16", self.name(), "Series")
}
DataType::Int32 => {
format_array!(f, self.i32().unwrap(), "i32", self.name(), "Series")
}
DataType::Int64 => {
format_array!(f, self.i64().unwrap(), "i64", self.name(), "Series")
}
DataType::Float32 => {
format_array!(f, self.f32().unwrap(), "f32", self.name(), "Series")
}
DataType::Float64 => {
format_array!(f, self.f64().unwrap(), "f64", self.name(), "Series")
}
#[cfg(feature = "dtype-date")]
DataType::Date => format_array!(f, self.date().unwrap(), "date", self.name(), "Series"),
#[cfg(feature = "dtype-datetime")]
DataType::Datetime(_, _) => {
let dt = format!("{}", self.dtype());
format_array!(f, self.datetime().unwrap(), &dt, self.name(), "Series")
}
#[cfg(feature = "dtype-time")]
DataType::Time => format_array!(f, self.time().unwrap(), "time", self.name(), "Series"),
#[cfg(feature = "dtype-duration")]
DataType::Duration(_) => {
let dt = format!("{}", self.dtype());
format_array!(f, self.duration().unwrap(), &dt, self.name(), "Series")
}
DataType::List(_) => {
format_array!(f, self.list().unwrap(), "list", self.name(), "Series")
}
#[cfg(feature = "object")]
DataType::Object(_) => format_object_array(f, self, self.name(), "Series"),
#[cfg(feature = "dtype-categorical")]
DataType::Categorical(_) => {
format_array!(f, self.categorical().unwrap(), "cat", self.name(), "Series")
}
#[cfg(feature = "dtype-struct")]
dt @ DataType::Struct(_) => format_array!(
f,
self.struct_().unwrap(),
format!("{dt}"),
self.name(),
"Series"
),
DataType::Null => {
writeln!(f, "nullarray")
}
#[cfg(feature = "dtype-binary")]
DataType::Binary => {
format_array!(f, self.binary().unwrap(), "binary", self.name(), "Series")
}
dt => panic!("{dt:?} not impl"),
}
}sourcepub fn bool(&self) -> PolarsResult<&BooleanChunked>
pub fn bool(&self) -> PolarsResult<&BooleanChunked>
Unpack to ChunkedArray of dtype bool
Examples found in repository?
More examples
src/chunked_array/ops/is_in.rs (line 333)
298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351
fn is_in(&self, other: &Series) -> PolarsResult<BooleanChunked> {
match other.dtype() {
DataType::List(dt) if self.dtype() == &**dt => {
let mut ca: BooleanChunked = if self.len() == 1 && other.len() != 1 {
let value = self.get(0);
// safety: we know the iterators len
unsafe {
other
.list()?
.amortized_iter()
.map(|opt_s| {
opt_s.map(|s| {
let ca = s.as_ref().unpack::<BooleanType>().unwrap();
ca.into_iter().any(|a| a == value)
}) == Some(true)
})
.trust_my_length(other.len())
.collect_trusted()
}
} else {
self.into_iter()
.zip(other.list()?.amortized_iter())
.map(|(value, series)| match (value, series) {
(val, Some(series)) => {
let ca = series.as_ref().unpack::<BooleanType>().unwrap();
ca.into_iter().any(|a| a == val)
}
_ => false,
})
.collect_trusted()
};
ca.rename(self.name());
Ok(ca)
}
DataType::Boolean => {
let other = other.bool().unwrap();
let has_true = other.any();
let has_false = !other.all();
Ok(self.apply(|v| if v { has_true } else { has_false }))
}
_ => Err(PolarsError::SchemaMisMatch(
format!(
"cannot do is_in operation with left a dtype: {:?} and right a dtype {:?}",
self.dtype(),
other.dtype()
)
.into(),
)),
}
.map(|mut ca| {
ca.rename(self.name());
ca
})
}src/fmt.rs (line 209)
206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284
fn fmt(&self, f: &mut Formatter<'_>) -> fmt::Result {
match self.dtype() {
DataType::Boolean => {
format_array!(f, self.bool().unwrap(), "bool", self.name(), "Series")
}
DataType::Utf8 => {
format_array!(f, self.utf8().unwrap(), "str", self.name(), "Series")
}
DataType::UInt8 => {
format_array!(f, self.u8().unwrap(), "u8", self.name(), "Series")
}
DataType::UInt16 => {
format_array!(f, self.u16().unwrap(), "u16", self.name(), "Series")
}
DataType::UInt32 => {
format_array!(f, self.u32().unwrap(), "u32", self.name(), "Series")
}
DataType::UInt64 => {
format_array!(f, self.u64().unwrap(), "u64", self.name(), "Series")
}
DataType::Int8 => {
format_array!(f, self.i8().unwrap(), "i8", self.name(), "Series")
}
DataType::Int16 => {
format_array!(f, self.i16().unwrap(), "i16", self.name(), "Series")
}
DataType::Int32 => {
format_array!(f, self.i32().unwrap(), "i32", self.name(), "Series")
}
DataType::Int64 => {
format_array!(f, self.i64().unwrap(), "i64", self.name(), "Series")
}
DataType::Float32 => {
format_array!(f, self.f32().unwrap(), "f32", self.name(), "Series")
}
DataType::Float64 => {
format_array!(f, self.f64().unwrap(), "f64", self.name(), "Series")
}
#[cfg(feature = "dtype-date")]
DataType::Date => format_array!(f, self.date().unwrap(), "date", self.name(), "Series"),
#[cfg(feature = "dtype-datetime")]
DataType::Datetime(_, _) => {
let dt = format!("{}", self.dtype());
format_array!(f, self.datetime().unwrap(), &dt, self.name(), "Series")
}
#[cfg(feature = "dtype-time")]
DataType::Time => format_array!(f, self.time().unwrap(), "time", self.name(), "Series"),
#[cfg(feature = "dtype-duration")]
DataType::Duration(_) => {
let dt = format!("{}", self.dtype());
format_array!(f, self.duration().unwrap(), &dt, self.name(), "Series")
}
DataType::List(_) => {
format_array!(f, self.list().unwrap(), "list", self.name(), "Series")
}
#[cfg(feature = "object")]
DataType::Object(_) => format_object_array(f, self, self.name(), "Series"),
#[cfg(feature = "dtype-categorical")]
DataType::Categorical(_) => {
format_array!(f, self.categorical().unwrap(), "cat", self.name(), "Series")
}
#[cfg(feature = "dtype-struct")]
dt @ DataType::Struct(_) => format_array!(
f,
self.struct_().unwrap(),
format!("{dt}"),
self.name(),
"Series"
),
DataType::Null => {
writeln!(f, "nullarray")
}
#[cfg(feature = "dtype-binary")]
DataType::Binary => {
format_array!(f, self.binary().unwrap(), "binary", self.name(), "Series")
}
dt => panic!("{dt:?} not impl"),
}
}sourcepub fn utf8(&self) -> PolarsResult<&Utf8Chunked>
pub fn utf8(&self) -> PolarsResult<&Utf8Chunked>
Unpack to ChunkedArray of dtype utf8
Examples found in repository?
More examples
src/series/comparison.rs (line 96)
86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388
fn compare_cat_to_str_series<Compare>(
cat: &Series,
string: &Series,
name: &str,
compare: Compare,
fill_value: bool,
) -> PolarsResult<BooleanChunked>
where
Compare: Fn(&Series, u32) -> PolarsResult<BooleanChunked>,
{
match string.utf8()?.get(0) {
None => Ok(cat.is_null()),
Some(value) => compare_cat_to_str_value(cat, value, name, compare, fill_value),
}
}
fn validate_types(left: &DataType, right: &DataType) -> PolarsResult<()> {
use DataType::*;
#[cfg(feature = "dtype-categorical")]
{
if matches!(left, Utf8 | Categorical(_)) && right.is_numeric()
|| left.is_numeric() && matches!(right, Utf8 | Categorical(_))
{
Err(PolarsError::ComputeError(
"cannot compare Utf8 with numeric data".into(),
))
} else {
Ok(())
}
}
#[cfg(not(feature = "dtype-categorical"))]
{
if matches!(left, Utf8) && right.is_numeric() || left.is_numeric() && matches!(right, Utf8)
{
Err(PolarsError::ComputeError(
"cannot compare Utf8 with numeric data".into(),
))
} else {
Ok(())
}
}
}
impl ChunkCompare<&Series> for Series {
type Item = PolarsResult<BooleanChunked>;
/// Create a boolean mask by checking for equality.
fn equal(&self, rhs: &Series) -> PolarsResult<BooleanChunked> {
validate_types(self.dtype(), rhs.dtype())?;
#[cfg(feature = "dtype-categorical")]
use DataType::*;
let mut out = match (self.dtype(), rhs.dtype(), self.len(), rhs.len()) {
#[cfg(feature = "dtype-categorical")]
(Categorical(_), Utf8, _, 1) => {
return compare_cat_to_str_series(
self,
rhs,
self.name(),
|s, idx| s.equal(idx),
false,
);
}
#[cfg(feature = "dtype-categorical")]
(Utf8, Categorical(_), 1, _) => {
return compare_cat_to_str_series(
rhs,
self,
self.name(),
|s, idx| s.equal(idx),
false,
);
}
#[cfg(feature = "dtype-categorical")]
(Categorical(Some(rev_map_l)), Categorical(Some(rev_map_r)), _, _) => {
if rev_map_l.same_src(rev_map_r) {
self.categorical()
.unwrap()
.logical()
.equal(rhs.categorical().unwrap().logical())
} else {
return Err(PolarsError::ComputeError("Cannot compare categoricals originating from different sources. Consider setting a global string cache.".into()));
}
}
_ => {
impl_compare!(self, rhs, equal)
}
};
out.rename(self.name());
Ok(out)
}
/// Create a boolean mask by checking for inequality.
fn not_equal(&self, rhs: &Series) -> PolarsResult<BooleanChunked> {
validate_types(self.dtype(), rhs.dtype())?;
#[cfg(feature = "dtype-categorical")]
use DataType::*;
let mut out = match (self.dtype(), rhs.dtype(), self.len(), rhs.len()) {
#[cfg(feature = "dtype-categorical")]
(Categorical(_), Utf8, _, 1) => {
return compare_cat_to_str_series(
self,
rhs,
self.name(),
|s, idx| s.not_equal(idx),
true,
);
}
#[cfg(feature = "dtype-categorical")]
(Utf8, Categorical(_), 1, _) => {
return compare_cat_to_str_series(
rhs,
self,
self.name(),
|s, idx| s.not_equal(idx),
true,
);
}
#[cfg(feature = "dtype-categorical")]
(Categorical(Some(rev_map_l)), Categorical(Some(rev_map_r)), _, _) => {
if rev_map_l.same_src(rev_map_r) {
self.categorical()
.unwrap()
.logical()
.not_equal(rhs.categorical().unwrap().logical())
} else {
return Err(PolarsError::ComputeError("Cannot compare categoricals originating from different sources. Consider setting a global string cache.".into()));
}
}
_ => {
impl_compare!(self, rhs, not_equal)
}
};
out.rename(self.name());
Ok(out)
}
/// Create a boolean mask by checking if self > rhs.
fn gt(&self, rhs: &Series) -> PolarsResult<BooleanChunked> {
validate_types(self.dtype(), rhs.dtype())?;
let mut out = impl_compare!(self, rhs, gt);
out.rename(self.name());
Ok(out)
}
/// Create a boolean mask by checking if self >= rhs.
fn gt_eq(&self, rhs: &Series) -> PolarsResult<BooleanChunked> {
validate_types(self.dtype(), rhs.dtype())?;
let mut out = impl_compare!(self, rhs, gt_eq);
out.rename(self.name());
Ok(out)
}
/// Create a boolean mask by checking if self < rhs.
fn lt(&self, rhs: &Series) -> PolarsResult<BooleanChunked> {
validate_types(self.dtype(), rhs.dtype())?;
let mut out = impl_compare!(self, rhs, lt);
out.rename(self.name());
Ok(out)
}
/// Create a boolean mask by checking if self <= rhs.
fn lt_eq(&self, rhs: &Series) -> PolarsResult<BooleanChunked> {
validate_types(self.dtype(), rhs.dtype())?;
let mut out = impl_compare!(self, rhs, lt_eq);
out.rename(self.name());
Ok(out)
}
}
impl<Rhs> ChunkCompare<Rhs> for Series
where
Rhs: NumericNative,
{
type Item = PolarsResult<BooleanChunked>;
fn equal(&self, rhs: Rhs) -> PolarsResult<BooleanChunked> {
validate_types(self.dtype(), &DataType::Int8)?;
let s = self.to_physical_repr();
Ok(apply_method_physical_numeric!(&s, equal, rhs))
}
fn not_equal(&self, rhs: Rhs) -> PolarsResult<BooleanChunked> {
validate_types(self.dtype(), &DataType::Int8)?;
let s = self.to_physical_repr();
Ok(apply_method_physical_numeric!(&s, not_equal, rhs))
}
fn gt(&self, rhs: Rhs) -> PolarsResult<BooleanChunked> {
validate_types(self.dtype(), &DataType::Int8)?;
let s = self.to_physical_repr();
Ok(apply_method_physical_numeric!(&s, gt, rhs))
}
fn gt_eq(&self, rhs: Rhs) -> PolarsResult<BooleanChunked> {
validate_types(self.dtype(), &DataType::Int8)?;
let s = self.to_physical_repr();
Ok(apply_method_physical_numeric!(&s, gt_eq, rhs))
}
fn lt(&self, rhs: Rhs) -> PolarsResult<BooleanChunked> {
validate_types(self.dtype(), &DataType::Int8)?;
let s = self.to_physical_repr();
Ok(apply_method_physical_numeric!(&s, lt, rhs))
}
fn lt_eq(&self, rhs: Rhs) -> PolarsResult<BooleanChunked> {
validate_types(self.dtype(), &DataType::Int8)?;
let s = self.to_physical_repr();
Ok(apply_method_physical_numeric!(&s, lt_eq, rhs))
}
}
impl ChunkCompare<&str> for Series {
type Item = PolarsResult<BooleanChunked>;
fn equal(&self, rhs: &str) -> PolarsResult<BooleanChunked> {
validate_types(self.dtype(), &DataType::Utf8)?;
use DataType::*;
match self.dtype() {
Utf8 => Ok(self.utf8().unwrap().equal(rhs)),
#[cfg(feature = "dtype-categorical")]
Categorical(_) => {
compare_cat_to_str_value(self, rhs, self.name(), |lhs, idx| lhs.equal(idx), false)
}
_ => Ok(BooleanChunked::full(self.name(), false, self.len())),
}
}
fn not_equal(&self, rhs: &str) -> PolarsResult<BooleanChunked> {
validate_types(self.dtype(), &DataType::Utf8)?;
use DataType::*;
match self.dtype() {
Utf8 => Ok(self.utf8().unwrap().not_equal(rhs)),
#[cfg(feature = "dtype-categorical")]
Categorical(_) => compare_cat_to_str_value(
self,
rhs,
self.name(),
|lhs, idx| lhs.not_equal(idx),
true,
),
_ => Ok(BooleanChunked::full(self.name(), false, self.len())),
}
}
fn gt(&self, rhs: &str) -> PolarsResult<BooleanChunked> {
validate_types(self.dtype(), &DataType::Utf8)?;
if let Ok(a) = self.utf8() {
Ok(a.gt(rhs))
} else {
Err(PolarsError::ComputeError(
format!(
"cannot compare str value to series of type: {:?}",
self.dtype()
)
.into(),
))
}
}
fn gt_eq(&self, rhs: &str) -> PolarsResult<BooleanChunked> {
validate_types(self.dtype(), &DataType::Utf8)?;
if let Ok(a) = self.utf8() {
Ok(a.gt_eq(rhs))
} else {
Err(PolarsError::ComputeError(
format!(
"cannot compare str value to series of type: {:?}",
self.dtype()
)
.into(),
))
}
}
fn lt(&self, rhs: &str) -> PolarsResult<BooleanChunked> {
validate_types(self.dtype(), &DataType::Utf8)?;
if let Ok(a) = self.utf8() {
Ok(a.lt(rhs))
} else {
Err(PolarsError::ComputeError(
format!(
"cannot compare str value to series of type: {:?}",
self.dtype()
)
.into(),
))
}
}
fn lt_eq(&self, rhs: &str) -> PolarsResult<BooleanChunked> {
validate_types(self.dtype(), &DataType::Utf8)?;
if let Ok(a) = self.utf8() {
Ok(a.lt_eq(rhs))
} else {
Err(PolarsError::ComputeError(
format!(
"cannot compare str value to series of type: {:?}",
self.dtype()
)
.into(),
))
}
}src/series/mod.rs (line 294)
291 292 293 294 295 296 297 298 299 300 301 302 303
pub fn explode(&self) -> PolarsResult<Series> {
match self.dtype() {
DataType::List(_) => self.list().unwrap().explode(),
DataType::Utf8 => self.utf8().unwrap().explode(),
_ => Err(PolarsError::InvalidOperation(
format!(
"explode not supported for Series with dtype {:?}",
self.dtype()
)
.into(),
)),
}
}src/series/ops/unique.rs (line 44)
33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50
pub fn unique_counts(&self) -> IdxCa {
if self.dtype().to_physical().is_numeric() {
if self.bit_repr_is_large() {
let ca = self.bit_repr_large();
unique_counts(ca.into_iter())
} else {
let ca = self.bit_repr_small();
unique_counts(ca.into_iter())
}
} else {
match self.dtype() {
DataType::Utf8 => unique_counts(self.utf8().unwrap().into_iter()),
dt => {
panic!("'unique_counts' not implemented for {dt} data types")
}
}
}
}src/frame/hash_join/single_keys_dispatch.rs (line 14)
8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131
pub fn hash_join_left(&self, other: &Series) -> LeftJoinIds {
let (lhs, rhs) = (self.to_physical_repr(), other.to_physical_repr());
use DataType::*;
match lhs.dtype() {
Utf8 => {
let lhs = lhs.utf8().unwrap();
let rhs = rhs.utf8().unwrap();
lhs.hash_join_left(rhs)
}
#[cfg(feature = "dtype-binary")]
Binary => {
let lhs = lhs.binary().unwrap();
let rhs = rhs.binary().unwrap();
lhs.hash_join_left(rhs)
}
_ => {
if self.bit_repr_is_large() {
let lhs = lhs.bit_repr_large();
let rhs = rhs.bit_repr_large();
num_group_join_left(&lhs, &rhs)
} else {
let lhs = lhs.bit_repr_small();
let rhs = rhs.bit_repr_small();
num_group_join_left(&lhs, &rhs)
}
}
}
}
#[cfg(feature = "semi_anti_join")]
pub(super) fn hash_join_semi_anti(&self, other: &Series, anti: bool) -> Vec<IdxSize> {
let (lhs, rhs) = (self.to_physical_repr(), other.to_physical_repr());
use DataType::*;
match lhs.dtype() {
Utf8 => {
let lhs = lhs.utf8().unwrap();
let rhs = rhs.utf8().unwrap();
lhs.hash_join_semi_anti(rhs, anti)
}
#[cfg(feature = "dtype-binary")]
Binary => {
let lhs = lhs.binary().unwrap();
let rhs = rhs.binary().unwrap();
lhs.hash_join_semi_anti(rhs, anti)
}
_ => {
if self.bit_repr_is_large() {
let lhs = lhs.bit_repr_large();
let rhs = rhs.bit_repr_large();
num_group_join_anti_semi(&lhs, &rhs, anti)
} else {
let lhs = lhs.bit_repr_small();
let rhs = rhs.bit_repr_small();
num_group_join_anti_semi(&lhs, &rhs, anti)
}
}
}
}
// returns the join tuples and whether or not the lhs tuples are sorted
pub(super) fn hash_join_inner(&self, other: &Series) -> ((Vec<IdxSize>, Vec<IdxSize>), bool) {
let (lhs, rhs) = (self.to_physical_repr(), other.to_physical_repr());
use DataType::*;
match lhs.dtype() {
Utf8 => {
let lhs = lhs.utf8().unwrap();
let rhs = rhs.utf8().unwrap();
lhs.hash_join_inner(rhs)
}
#[cfg(feature = "dtype-binary")]
Binary => {
let lhs = lhs.binary().unwrap();
let rhs = rhs.binary().unwrap();
lhs.hash_join_inner(rhs)
}
_ => {
if self.bit_repr_is_large() {
let lhs = self.bit_repr_large();
let rhs = other.bit_repr_large();
num_group_join_inner(&lhs, &rhs)
} else {
let lhs = self.bit_repr_small();
let rhs = other.bit_repr_small();
num_group_join_inner(&lhs, &rhs)
}
}
}
}
pub(super) fn hash_join_outer(
&self,
other: &Series,
) -> Vec<(Option<IdxSize>, Option<IdxSize>)> {
let (lhs, rhs) = (self.to_physical_repr(), other.to_physical_repr());
use DataType::*;
match lhs.dtype() {
Utf8 => {
let lhs = lhs.utf8().unwrap();
let rhs = rhs.utf8().unwrap();
lhs.hash_join_outer(rhs)
}
#[cfg(feature = "dtype-binary")]
Binary => {
let lhs = lhs.binary().unwrap();
let rhs = rhs.binary().unwrap();
lhs.hash_join_outer(rhs)
}
_ => {
if self.bit_repr_is_large() {
let lhs = self.bit_repr_large();
let rhs = other.bit_repr_large();
lhs.hash_join_outer(&rhs)
} else {
let lhs = self.bit_repr_small();
let rhs = other.bit_repr_small();
lhs.hash_join_outer(&rhs)
}
}
}
}Additional examples can be found in:
sourcepub fn list(&self) -> PolarsResult<&ListChunked>
pub fn list(&self) -> PolarsResult<&ListChunked>
Unpack to ChunkedArray of dtype list
Examples found in repository?
More examples
src/series/mod.rs (line 293)
291 292 293 294 295 296 297 298 299 300 301 302 303
pub fn explode(&self) -> PolarsResult<Series> {
match self.dtype() {
DataType::List(_) => self.list().unwrap().explode(),
DataType::Utf8 => self.utf8().unwrap().explode(),
_ => Err(PolarsError::InvalidOperation(
format!(
"explode not supported for Series with dtype {:?}",
self.dtype()
)
.into(),
)),
}
}src/series/into.rs (line 26)
21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76
pub fn to_arrow(&self, chunk_idx: usize) -> ArrayRef {
match self.dtype() {
// special list branch to
// make sure that we recursively apply all logical types.
DataType::List(inner) => {
let ca = self.list().unwrap();
let arr = ca.chunks[chunk_idx].clone();
let arr = arr.as_any().downcast_ref::<ListArray<i64>>().unwrap();
let s = unsafe {
Series::from_chunks_and_dtype_unchecked("", vec![arr.values().clone()], inner)
};
let new_values = s.to_arrow(0);
let data_type = ListArray::<i64>::default_datatype(inner.to_arrow());
let arr = ListArray::<i64>::new(
data_type,
arr.offsets().clone(),
new_values,
arr.validity().cloned(),
);
Box::new(arr)
}
#[cfg(feature = "dtype-categorical")]
DataType::Categorical(_) => {
let ca = self.categorical().unwrap();
let arr = ca.logical().chunks()[chunk_idx].clone();
let cats = UInt32Chunked::from_chunks("", vec![arr]);
// safety:
// we only take a single chunk and change nothing about the index/rev_map mapping
let new = unsafe {
CategoricalChunked::from_cats_and_rev_map_unchecked(
cats,
ca.get_rev_map().clone(),
)
};
let arr: DictionaryArray<u32> = (&new).into();
Box::new(arr) as ArrayRef
}
#[cfg(feature = "dtype-date")]
DataType::Date => cast(&*self.chunks()[chunk_idx], &DataType::Date.to_arrow()).unwrap(),
#[cfg(feature = "dtype-datetime")]
DataType::Datetime(_, _) => {
cast(&*self.chunks()[chunk_idx], &self.dtype().to_arrow()).unwrap()
}
#[cfg(feature = "dtype-duration")]
DataType::Duration(_) => {
cast(&*self.chunks()[chunk_idx], &self.dtype().to_arrow()).unwrap()
}
#[cfg(feature = "dtype-time")]
DataType::Time => cast(&*self.chunks()[chunk_idx], &DataType::Time.to_arrow()).unwrap(),
_ => self.array_ref(chunk_idx).clone(),
}
}src/chunked_array/ops/is_in.rs (line 60)
45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351
fn is_in(&self, other: &Series) -> PolarsResult<BooleanChunked> {
// We check implicitly cast to supertype here
match other.dtype() {
DataType::List(dt) => {
let st = try_get_supertype(self.dtype(), dt)?;
if &st != self.dtype() {
let left = self.cast(&st)?;
let right = other.cast(&DataType::List(Box::new(st)))?;
return left.is_in(&right);
}
let mut ca: BooleanChunked = if self.len() == 1 && other.len() != 1 {
let value = self.get(0);
other
.list()?
.amortized_iter()
.map(|opt_s| {
opt_s.map(|s| {
let ca = s.as_ref().unpack::<T>().unwrap();
ca.into_iter().any(|a| a == value)
}) == Some(true)
})
.collect_trusted()
} else {
self.into_iter()
.zip(other.list()?.amortized_iter())
.map(|(value, series)| match (value, series) {
(val, Some(series)) => {
let ca = series.as_ref().unpack::<T>().unwrap();
ca.into_iter().any(|a| a == val)
}
_ => false,
})
.collect_trusted()
};
ca.rename(self.name());
Ok(ca)
}
_ => {
// first make sure that the types are equal
let st = try_get_supertype(self.dtype(), other.dtype())?;
if self.dtype() != other.dtype() {
let left = self.cast(&st)?;
let right = other.cast(&st)?;
return left.is_in(&right);
}
// now that the types are equal, we coerce every 32 bit array to u32
// and every 64 bit array to u64 (including floats)
// this allows hashing them and greatly reduces the number of code paths.
match self.dtype() {
DataType::UInt64 | DataType::Int64 | DataType::Float64 => unsafe {
is_in_helper::<T, u64>(self, other)
},
DataType::UInt32 | DataType::Int32 | DataType::Float32 => unsafe {
is_in_helper::<T, u32>(self, other)
},
DataType::UInt8 | DataType::Int8 => unsafe {
is_in_helper::<T, u8>(self, other)
},
DataType::UInt16 | DataType::Int16 => unsafe {
is_in_helper::<T, u16>(self, other)
},
_ => Err(PolarsError::ComputeError(
format!(
"Data type {:?} not supported in is_in operation",
self.dtype()
)
.into(),
)),
}
}
}
.map(|mut ca| {
ca.rename(self.name());
ca
})
}
}
impl IsIn for Utf8Chunked {
fn is_in(&self, other: &Series) -> PolarsResult<BooleanChunked> {
match other.dtype() {
#[cfg(feature = "dtype-categorical")]
DataType::List(dt) if matches!(&**dt, DataType::Categorical(_)) => {
if let DataType::Categorical(Some(rev_map)) = &**dt {
let opt_val = self.get(0);
let other = other.list()?;
match opt_val {
None => {
let mut ca: BooleanChunked = other
.amortized_iter()
.map(|opt_s| {
opt_s.map(|s| s.as_ref().null_count() > 0) == Some(true)
})
.collect_trusted();
ca.rename(self.name());
Ok(ca)
}
Some(value) => {
match rev_map.find(value) {
// all false
None => Ok(BooleanChunked::full(self.name(), false, other.len())),
Some(idx) => {
let mut ca: BooleanChunked = other
.amortized_iter()
.map(|opt_s| {
opt_s.map(|s| {
let s = s.as_ref().to_physical_repr();
let ca = s.as_ref().u32().unwrap();
if ca.null_count() == 0 {
ca.into_no_null_iter().any(|a| a == idx)
} else {
ca.into_iter().any(|a| a == Some(idx))
}
}) == Some(true)
})
.collect_trusted();
ca.rename(self.name());
Ok(ca)
}
}
}
}
} else {
unreachable!()
}
}
DataType::List(dt) if DataType::Utf8 == **dt => {
let mut ca: BooleanChunked = if self.len() == 1 && other.len() != 1 {
let value = self.get(0);
other
.list()?
.amortized_iter()
.map(|opt_s| {
opt_s.map(|s| {
let ca = s.as_ref().unpack::<Utf8Type>().unwrap();
ca.into_iter().any(|a| a == value)
}) == Some(true)
})
.collect_trusted()
} else {
self.into_iter()
.zip(other.list()?.amortized_iter())
.map(|(value, series)| match (value, series) {
(val, Some(series)) => {
let ca = series.as_ref().unpack::<Utf8Type>().unwrap();
ca.into_iter().any(|a| a == val)
}
_ => false,
})
.collect_trusted()
};
ca.rename(self.name());
Ok(ca)
}
DataType::Utf8 => {
let mut set = HashSet::with_capacity(other.len());
let other = other.utf8()?;
other.downcast_iter().for_each(|iter| {
iter.into_iter().for_each(|opt_val| {
set.insert(opt_val);
})
});
let mut ca: BooleanChunked = self
.into_iter()
.map(|opt_val| set.contains(&opt_val))
.collect_trusted();
ca.rename(self.name());
Ok(ca)
}
_ => Err(PolarsError::SchemaMisMatch(
format!(
"cannot do is_in operation with left a dtype: {:?} and right a dtype {:?}",
self.dtype(),
other.dtype()
)
.into(),
)),
}
.map(|mut ca| {
ca.rename(self.name());
ca
})
}
}
#[cfg(feature = "dtype-binary")]
impl IsIn for BinaryChunked {
fn is_in(&self, other: &Series) -> PolarsResult<BooleanChunked> {
match other.dtype() {
DataType::List(dt) if DataType::Binary == **dt => {
let mut ca: BooleanChunked = if self.len() == 1 && other.len() != 1 {
let value = self.get(0);
other
.list()?
.amortized_iter()
.map(|opt_b| {
opt_b.map(|s| {
let ca = s.as_ref().unpack::<BinaryType>().unwrap();
ca.into_iter().any(|a| a == value)
}) == Some(true)
})
.collect_trusted()
} else {
self.into_iter()
.zip(other.list()?.amortized_iter())
.map(|(value, series)| match (value, series) {
(val, Some(series)) => {
let ca = series.as_ref().unpack::<BinaryType>().unwrap();
ca.into_iter().any(|a| a == val)
}
_ => false,
})
.collect_trusted()
};
ca.rename(self.name());
Ok(ca)
}
DataType::Binary => {
let mut set = HashSet::with_capacity(other.len());
let other = other.binary()?;
other.downcast_iter().for_each(|iter| {
iter.into_iter().for_each(|opt_val| {
set.insert(opt_val);
})
});
let mut ca: BooleanChunked = self
.into_iter()
.map(|opt_val| set.contains(&opt_val))
.collect_trusted();
ca.rename(self.name());
Ok(ca)
}
_ => Err(PolarsError::SchemaMisMatch(
format!(
"cannot do is_in operation with left a dtype: {:?} and right a dtype {:?}",
self.dtype(),
other.dtype()
)
.into(),
)),
}
.map(|mut ca| {
ca.rename(self.name());
ca
})
}
}
impl IsIn for BooleanChunked {
fn is_in(&self, other: &Series) -> PolarsResult<BooleanChunked> {
match other.dtype() {
DataType::List(dt) if self.dtype() == &**dt => {
let mut ca: BooleanChunked = if self.len() == 1 && other.len() != 1 {
let value = self.get(0);
// safety: we know the iterators len
unsafe {
other
.list()?
.amortized_iter()
.map(|opt_s| {
opt_s.map(|s| {
let ca = s.as_ref().unpack::<BooleanType>().unwrap();
ca.into_iter().any(|a| a == value)
}) == Some(true)
})
.trust_my_length(other.len())
.collect_trusted()
}
} else {
self.into_iter()
.zip(other.list()?.amortized_iter())
.map(|(value, series)| match (value, series) {
(val, Some(series)) => {
let ca = series.as_ref().unpack::<BooleanType>().unwrap();
ca.into_iter().any(|a| a == val)
}
_ => false,
})
.collect_trusted()
};
ca.rename(self.name());
Ok(ca)
}
DataType::Boolean => {
let other = other.bool().unwrap();
let has_true = other.any();
let has_false = !other.all();
Ok(self.apply(|v| if v { has_true } else { has_false }))
}
_ => Err(PolarsError::SchemaMisMatch(
format!(
"cannot do is_in operation with left a dtype: {:?} and right a dtype {:?}",
self.dtype(),
other.dtype()
)
.into(),
)),
}
.map(|mut ca| {
ca.rename(self.name());
ca
})
}src/fmt.rs (line 259)
206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284
fn fmt(&self, f: &mut Formatter<'_>) -> fmt::Result {
match self.dtype() {
DataType::Boolean => {
format_array!(f, self.bool().unwrap(), "bool", self.name(), "Series")
}
DataType::Utf8 => {
format_array!(f, self.utf8().unwrap(), "str", self.name(), "Series")
}
DataType::UInt8 => {
format_array!(f, self.u8().unwrap(), "u8", self.name(), "Series")
}
DataType::UInt16 => {
format_array!(f, self.u16().unwrap(), "u16", self.name(), "Series")
}
DataType::UInt32 => {
format_array!(f, self.u32().unwrap(), "u32", self.name(), "Series")
}
DataType::UInt64 => {
format_array!(f, self.u64().unwrap(), "u64", self.name(), "Series")
}
DataType::Int8 => {
format_array!(f, self.i8().unwrap(), "i8", self.name(), "Series")
}
DataType::Int16 => {
format_array!(f, self.i16().unwrap(), "i16", self.name(), "Series")
}
DataType::Int32 => {
format_array!(f, self.i32().unwrap(), "i32", self.name(), "Series")
}
DataType::Int64 => {
format_array!(f, self.i64().unwrap(), "i64", self.name(), "Series")
}
DataType::Float32 => {
format_array!(f, self.f32().unwrap(), "f32", self.name(), "Series")
}
DataType::Float64 => {
format_array!(f, self.f64().unwrap(), "f64", self.name(), "Series")
}
#[cfg(feature = "dtype-date")]
DataType::Date => format_array!(f, self.date().unwrap(), "date", self.name(), "Series"),
#[cfg(feature = "dtype-datetime")]
DataType::Datetime(_, _) => {
let dt = format!("{}", self.dtype());
format_array!(f, self.datetime().unwrap(), &dt, self.name(), "Series")
}
#[cfg(feature = "dtype-time")]
DataType::Time => format_array!(f, self.time().unwrap(), "time", self.name(), "Series"),
#[cfg(feature = "dtype-duration")]
DataType::Duration(_) => {
let dt = format!("{}", self.dtype());
format_array!(f, self.duration().unwrap(), &dt, self.name(), "Series")
}
DataType::List(_) => {
format_array!(f, self.list().unwrap(), "list", self.name(), "Series")
}
#[cfg(feature = "object")]
DataType::Object(_) => format_object_array(f, self, self.name(), "Series"),
#[cfg(feature = "dtype-categorical")]
DataType::Categorical(_) => {
format_array!(f, self.categorical().unwrap(), "cat", self.name(), "Series")
}
#[cfg(feature = "dtype-struct")]
dt @ DataType::Struct(_) => format_array!(
f,
self.struct_().unwrap(),
format!("{dt}"),
self.name(),
"Series"
),
DataType::Null => {
writeln!(f, "nullarray")
}
#[cfg(feature = "dtype-binary")]
DataType::Binary => {
format_array!(f, self.binary().unwrap(), "binary", self.name(), "Series")
}
dt => panic!("{dt:?} not impl"),
}
}sourcepub fn categorical(&self) -> PolarsResult<&CategoricalChunked>
pub fn categorical(&self) -> PolarsResult<&CategoricalChunked>
Unpack to ChunkedArray of dtype categorical
Examples found in repository?
src/series/implementations/categorical.rs (line 88)
86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205
fn zip_with_same_type(&self, mask: &BooleanChunked, other: &Series) -> PolarsResult<Series> {
self.0
.zip_with(mask, other.categorical()?)
.map(|ca| ca.into_series())
}
fn into_partial_ord_inner<'a>(&'a self) -> Box<dyn PartialOrdInner + 'a> {
(&self.0).into_partial_ord_inner()
}
fn vec_hash(&self, random_state: RandomState, buf: &mut Vec<u64>) -> PolarsResult<()> {
self.0.logical().vec_hash(random_state, buf);
Ok(())
}
fn vec_hash_combine(&self, build_hasher: RandomState, hashes: &mut [u64]) -> PolarsResult<()> {
self.0.logical().vec_hash_combine(build_hasher, hashes);
Ok(())
}
unsafe fn agg_list(&self, groups: &GroupsProxy) -> Series {
// we cannot cast and dispatch as the inner type of the list would be incorrect
self.0
.logical()
.agg_list(groups)
.cast(&DataType::List(Box::new(self.dtype().clone())))
.unwrap()
}
fn zip_outer_join_column(
&self,
right_column: &Series,
opt_join_tuples: &[(Option<IdxSize>, Option<IdxSize>)],
) -> Series {
let new_rev_map = self
.0
.merge_categorical_map(right_column.categorical().unwrap())
.unwrap();
let left = self.0.logical();
let right = right_column
.categorical()
.unwrap()
.logical()
.clone()
.into_series();
let cats = left.zip_outer_join_column(&right, opt_join_tuples);
let cats = cats.u32().unwrap().clone();
unsafe {
CategoricalChunked::from_cats_and_rev_map_unchecked(cats, new_rev_map).into_series()
}
}
fn group_tuples(&self, multithreaded: bool, sorted: bool) -> PolarsResult<GroupsProxy> {
self.0.logical().group_tuples(multithreaded, sorted)
}
#[cfg(feature = "sort_multiple")]
fn argsort_multiple(&self, by: &[Series], reverse: &[bool]) -> PolarsResult<IdxCa> {
self.0.argsort_multiple(by, reverse)
}
}
impl SeriesTrait for SeriesWrap<CategoricalChunked> {
fn is_sorted(&self) -> IsSorted {
if self.0.logical().is_sorted() {
IsSorted::Ascending
} else if self.0.logical().is_sorted_reverse() {
IsSorted::Descending
} else {
IsSorted::Not
}
}
fn rename(&mut self, name: &str) {
self.0.logical_mut().rename(name);
}
fn chunk_lengths(&self) -> ChunkIdIter {
self.0.logical().chunk_id()
}
fn name(&self) -> &str {
self.0.logical().name()
}
fn chunks(&self) -> &Vec<ArrayRef> {
self.0.logical().chunks()
}
fn shrink_to_fit(&mut self) {
self.0.logical_mut().shrink_to_fit()
}
fn slice(&self, offset: i64, length: usize) -> Series {
self.with_state(false, |cats| cats.slice(offset, length))
.into_series()
}
fn append(&mut self, other: &Series) -> PolarsResult<()> {
if self.0.dtype() == other.dtype() {
self.0.append(other.categorical().unwrap())
} else {
Err(PolarsError::SchemaMisMatch(
"cannot append Series; data types don't match".into(),
))
}
}
fn extend(&mut self, other: &Series) -> PolarsResult<()> {
if self.0.dtype() == other.dtype() {
let other = other.categorical()?;
self.0.logical_mut().extend(other.logical());
let new_rev_map = self.0.merge_categorical_map(other)?;
// safety:
// rev_maps are merged
unsafe { self.0.set_rev_map(new_rev_map, false) };
Ok(())
} else {
Err(PolarsError::SchemaMisMatch(
"cannot extend Series; data types don't match".into(),
))
}
}More examples
src/series/comparison.rs (line 74)
64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220
fn compare_cat_to_str_value<Compare>(
cat: &Series,
value: &str,
name: &str,
compare: Compare,
fill_value: bool,
) -> PolarsResult<BooleanChunked>
where
Compare: Fn(&Series, u32) -> PolarsResult<BooleanChunked>,
{
let cat = cat.categorical().expect("should be categorical");
let cat_map = cat.get_rev_map();
match cat_map.find(value) {
None => Ok(BooleanChunked::full(name, fill_value, cat.len())),
Some(cat_idx) => {
let cat = cat.cast(&DataType::UInt32).unwrap();
compare(&cat, cat_idx)
}
}
}
#[cfg(feature = "dtype-categorical")]
fn compare_cat_to_str_series<Compare>(
cat: &Series,
string: &Series,
name: &str,
compare: Compare,
fill_value: bool,
) -> PolarsResult<BooleanChunked>
where
Compare: Fn(&Series, u32) -> PolarsResult<BooleanChunked>,
{
match string.utf8()?.get(0) {
None => Ok(cat.is_null()),
Some(value) => compare_cat_to_str_value(cat, value, name, compare, fill_value),
}
}
fn validate_types(left: &DataType, right: &DataType) -> PolarsResult<()> {
use DataType::*;
#[cfg(feature = "dtype-categorical")]
{
if matches!(left, Utf8 | Categorical(_)) && right.is_numeric()
|| left.is_numeric() && matches!(right, Utf8 | Categorical(_))
{
Err(PolarsError::ComputeError(
"cannot compare Utf8 with numeric data".into(),
))
} else {
Ok(())
}
}
#[cfg(not(feature = "dtype-categorical"))]
{
if matches!(left, Utf8) && right.is_numeric() || left.is_numeric() && matches!(right, Utf8)
{
Err(PolarsError::ComputeError(
"cannot compare Utf8 with numeric data".into(),
))
} else {
Ok(())
}
}
}
impl ChunkCompare<&Series> for Series {
type Item = PolarsResult<BooleanChunked>;
/// Create a boolean mask by checking for equality.
fn equal(&self, rhs: &Series) -> PolarsResult<BooleanChunked> {
validate_types(self.dtype(), rhs.dtype())?;
#[cfg(feature = "dtype-categorical")]
use DataType::*;
let mut out = match (self.dtype(), rhs.dtype(), self.len(), rhs.len()) {
#[cfg(feature = "dtype-categorical")]
(Categorical(_), Utf8, _, 1) => {
return compare_cat_to_str_series(
self,
rhs,
self.name(),
|s, idx| s.equal(idx),
false,
);
}
#[cfg(feature = "dtype-categorical")]
(Utf8, Categorical(_), 1, _) => {
return compare_cat_to_str_series(
rhs,
self,
self.name(),
|s, idx| s.equal(idx),
false,
);
}
#[cfg(feature = "dtype-categorical")]
(Categorical(Some(rev_map_l)), Categorical(Some(rev_map_r)), _, _) => {
if rev_map_l.same_src(rev_map_r) {
self.categorical()
.unwrap()
.logical()
.equal(rhs.categorical().unwrap().logical())
} else {
return Err(PolarsError::ComputeError("Cannot compare categoricals originating from different sources. Consider setting a global string cache.".into()));
}
}
_ => {
impl_compare!(self, rhs, equal)
}
};
out.rename(self.name());
Ok(out)
}
/// Create a boolean mask by checking for inequality.
fn not_equal(&self, rhs: &Series) -> PolarsResult<BooleanChunked> {
validate_types(self.dtype(), rhs.dtype())?;
#[cfg(feature = "dtype-categorical")]
use DataType::*;
let mut out = match (self.dtype(), rhs.dtype(), self.len(), rhs.len()) {
#[cfg(feature = "dtype-categorical")]
(Categorical(_), Utf8, _, 1) => {
return compare_cat_to_str_series(
self,
rhs,
self.name(),
|s, idx| s.not_equal(idx),
true,
);
}
#[cfg(feature = "dtype-categorical")]
(Utf8, Categorical(_), 1, _) => {
return compare_cat_to_str_series(
rhs,
self,
self.name(),
|s, idx| s.not_equal(idx),
true,
);
}
#[cfg(feature = "dtype-categorical")]
(Categorical(Some(rev_map_l)), Categorical(Some(rev_map_r)), _, _) => {
if rev_map_l.same_src(rev_map_r) {
self.categorical()
.unwrap()
.logical()
.not_equal(rhs.categorical().unwrap().logical())
} else {
return Err(PolarsError::ComputeError("Cannot compare categoricals originating from different sources. Consider setting a global string cache.".into()));
}
}
_ => {
impl_compare!(self, rhs, not_equal)
}
};
out.rename(self.name());
Ok(out)
}src/series/into.rs (line 46)
21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76
pub fn to_arrow(&self, chunk_idx: usize) -> ArrayRef {
match self.dtype() {
// special list branch to
// make sure that we recursively apply all logical types.
DataType::List(inner) => {
let ca = self.list().unwrap();
let arr = ca.chunks[chunk_idx].clone();
let arr = arr.as_any().downcast_ref::<ListArray<i64>>().unwrap();
let s = unsafe {
Series::from_chunks_and_dtype_unchecked("", vec![arr.values().clone()], inner)
};
let new_values = s.to_arrow(0);
let data_type = ListArray::<i64>::default_datatype(inner.to_arrow());
let arr = ListArray::<i64>::new(
data_type,
arr.offsets().clone(),
new_values,
arr.validity().cloned(),
);
Box::new(arr)
}
#[cfg(feature = "dtype-categorical")]
DataType::Categorical(_) => {
let ca = self.categorical().unwrap();
let arr = ca.logical().chunks()[chunk_idx].clone();
let cats = UInt32Chunked::from_chunks("", vec![arr]);
// safety:
// we only take a single chunk and change nothing about the index/rev_map mapping
let new = unsafe {
CategoricalChunked::from_cats_and_rev_map_unchecked(
cats,
ca.get_rev_map().clone(),
)
};
let arr: DictionaryArray<u32> = (&new).into();
Box::new(arr) as ArrayRef
}
#[cfg(feature = "dtype-date")]
DataType::Date => cast(&*self.chunks()[chunk_idx], &DataType::Date.to_arrow()).unwrap(),
#[cfg(feature = "dtype-datetime")]
DataType::Datetime(_, _) => {
cast(&*self.chunks()[chunk_idx], &self.dtype().to_arrow()).unwrap()
}
#[cfg(feature = "dtype-duration")]
DataType::Duration(_) => {
cast(&*self.chunks()[chunk_idx], &self.dtype().to_arrow()).unwrap()
}
#[cfg(feature = "dtype-time")]
DataType::Time => cast(&*self.chunks()[chunk_idx], &DataType::Time.to_arrow()).unwrap(),
_ => self.array_ref(chunk_idx).clone(),
}
}src/frame/hash_join/mod.rs (line 562)
515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581
pub fn _outer_join_from_series(
&self,
other: &DataFrame,
s_left: &Series,
s_right: &Series,
suffix: Option<String>,
slice: Option<(i64, usize)>,
) -> PolarsResult<DataFrame> {
#[cfg(feature = "dtype-categorical")]
_check_categorical_src(s_left.dtype(), s_right.dtype())?;
// store this so that we can keep original column order.
let join_column_index = self.iter().position(|s| s.name() == s_left.name()).unwrap();
// Get the indexes of the joined relations
let opt_join_tuples = s_left.hash_join_outer(s_right);
let mut opt_join_tuples = &*opt_join_tuples;
if let Some((offset, len)) = slice {
opt_join_tuples = slice_slice(opt_join_tuples, offset, len);
}
// Take the left and right dataframes by join tuples
let (mut df_left, df_right) = POOL.join(
|| unsafe {
self.drop(s_left.name()).unwrap().take_opt_iter_unchecked(
opt_join_tuples
.iter()
.map(|(left, _right)| left.map(|i| i as usize)),
)
},
|| unsafe {
other.drop(s_right.name()).unwrap().take_opt_iter_unchecked(
opt_join_tuples
.iter()
.map(|(_left, right)| right.map(|i| i as usize)),
)
},
);
let mut s = s_left
.to_physical_repr()
.zip_outer_join_column(&s_right.to_physical_repr(), opt_join_tuples);
s.rename(s_left.name());
let s = match s_left.dtype() {
#[cfg(feature = "dtype-categorical")]
DataType::Categorical(_) => {
let ca_left = s_left.categorical().unwrap();
let new_rev_map = ca_left.merge_categorical_map(s_right.categorical().unwrap())?;
let logical = s.u32().unwrap().clone();
// safety:
// categorical maps are merged
unsafe {
CategoricalChunked::from_cats_and_rev_map_unchecked(logical, new_rev_map)
.into_series()
}
}
dt @ DataType::Datetime(_, _)
| dt @ DataType::Time
| dt @ DataType::Date
| dt @ DataType::Duration(_) => s.cast(dt).unwrap(),
_ => s,
};
df_left.get_columns_mut().insert(join_column_index, s);
_finish_join(df_left, df_right, suffix.as_deref())
}src/fmt.rs (line 265)
206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284
fn fmt(&self, f: &mut Formatter<'_>) -> fmt::Result {
match self.dtype() {
DataType::Boolean => {
format_array!(f, self.bool().unwrap(), "bool", self.name(), "Series")
}
DataType::Utf8 => {
format_array!(f, self.utf8().unwrap(), "str", self.name(), "Series")
}
DataType::UInt8 => {
format_array!(f, self.u8().unwrap(), "u8", self.name(), "Series")
}
DataType::UInt16 => {
format_array!(f, self.u16().unwrap(), "u16", self.name(), "Series")
}
DataType::UInt32 => {
format_array!(f, self.u32().unwrap(), "u32", self.name(), "Series")
}
DataType::UInt64 => {
format_array!(f, self.u64().unwrap(), "u64", self.name(), "Series")
}
DataType::Int8 => {
format_array!(f, self.i8().unwrap(), "i8", self.name(), "Series")
}
DataType::Int16 => {
format_array!(f, self.i16().unwrap(), "i16", self.name(), "Series")
}
DataType::Int32 => {
format_array!(f, self.i32().unwrap(), "i32", self.name(), "Series")
}
DataType::Int64 => {
format_array!(f, self.i64().unwrap(), "i64", self.name(), "Series")
}
DataType::Float32 => {
format_array!(f, self.f32().unwrap(), "f32", self.name(), "Series")
}
DataType::Float64 => {
format_array!(f, self.f64().unwrap(), "f64", self.name(), "Series")
}
#[cfg(feature = "dtype-date")]
DataType::Date => format_array!(f, self.date().unwrap(), "date", self.name(), "Series"),
#[cfg(feature = "dtype-datetime")]
DataType::Datetime(_, _) => {
let dt = format!("{}", self.dtype());
format_array!(f, self.datetime().unwrap(), &dt, self.name(), "Series")
}
#[cfg(feature = "dtype-time")]
DataType::Time => format_array!(f, self.time().unwrap(), "time", self.name(), "Series"),
#[cfg(feature = "dtype-duration")]
DataType::Duration(_) => {
let dt = format!("{}", self.dtype());
format_array!(f, self.duration().unwrap(), &dt, self.name(), "Series")
}
DataType::List(_) => {
format_array!(f, self.list().unwrap(), "list", self.name(), "Series")
}
#[cfg(feature = "object")]
DataType::Object(_) => format_object_array(f, self, self.name(), "Series"),
#[cfg(feature = "dtype-categorical")]
DataType::Categorical(_) => {
format_array!(f, self.categorical().unwrap(), "cat", self.name(), "Series")
}
#[cfg(feature = "dtype-struct")]
dt @ DataType::Struct(_) => format_array!(
f,
self.struct_().unwrap(),
format!("{dt}"),
self.name(),
"Series"
),
DataType::Null => {
writeln!(f, "nullarray")
}
#[cfg(feature = "dtype-binary")]
DataType::Binary => {
format_array!(f, self.binary().unwrap(), "binary", self.name(), "Series")
}
dt => panic!("{dt:?} not impl"),
}
}source§impl Series
impl Series
sourcepub fn extend_constant(&self, value: AnyValue<'_>, n: usize) -> PolarsResult<Self>
pub fn extend_constant(&self, value: AnyValue<'_>, n: usize) -> PolarsResult<Self>
Extend with a constant value.
Examples found in repository?
src/functions.rs (line 236)
217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252
pub fn hor_concat_df(dfs: &[DataFrame]) -> PolarsResult<DataFrame> {
let max_len = dfs
.iter()
.map(|df| df.height())
.max()
.ok_or_else(|| PolarsError::ComputeError("cannot concat empty dataframes".into()))?;
let owned_df;
// if not all equal length, extend the DataFrame with nulls
let dfs = if !dfs.iter().all(|df| df.height() == max_len) {
owned_df = dfs
.iter()
.cloned()
.map(|mut df| {
if df.height() != max_len {
let diff = max_len - df.height();
df.columns
.iter_mut()
.for_each(|s| *s = s.extend_constant(AnyValue::Null, diff).unwrap());
}
df
})
.collect::<Vec<_>>();
owned_df.as_slice()
} else {
dfs
};
let mut first_df = dfs[0].clone();
for df in &dfs[1..] {
first_df.hstack_mut(df.get_columns())?;
}
Ok(first_df)
}More examples
src/frame/arithmetic.rs (line 131)
111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157
fn binary_aligned(
&self,
other: &DataFrame,
f: &(dyn Fn(&Series, &Series) -> PolarsResult<Series> + Sync + Send),
) -> PolarsResult<DataFrame> {
let max_len = std::cmp::max(self.height(), other.height());
let max_width = std::cmp::max(self.width(), other.width());
let mut cols = self
.get_columns()
.par_iter()
.zip(other.get_columns().par_iter())
.map(|(l, r)| {
let diff_l = max_len - l.len();
let diff_r = max_len - r.len();
let st = try_get_supertype(l.dtype(), r.dtype())?;
let mut l = l.cast(&st)?;
let mut r = r.cast(&st)?;
if diff_l > 0 {
l = l.extend_constant(AnyValue::Null, diff_l)?;
};
if diff_r > 0 {
r = r.extend_constant(AnyValue::Null, diff_r)?;
};
f(&l, &r)
})
.collect::<PolarsResult<Vec<_>>>()?;
let col_len = cols.len();
if col_len < max_width {
let df = if col_len < self.width() { self } else { other };
for i in col_len..max_len {
let s = &df.get_columns()[i];
let name = s.name();
let dtype = s.dtype();
// trick to fill a series with nulls
let vals: &[Option<i32>] = &[None];
let s = Series::new(name, vals).cast(dtype)?;
cols.push(s.new_from_index(0, max_len))
}
}
DataFrame::new(cols)
}source§impl Series
impl Series
sourcepub fn skew(&self, bias: bool) -> PolarsResult<Option<f64>>
Available on crate feature moment only.
pub fn skew(&self, bias: bool) -> PolarsResult<Option<f64>>
moment only.Compute the sample skewness of a data set.
For normally distributed data, the skewness should be about zero. For
uni-modal continuous distributions, a skewness value greater than zero means
that there is more weight in the right tail of the distribution. The
function skewtest can be used to determine if the skewness value
is close enough to zero, statistically speaking.
see: https://github.com/scipy/scipy/blob/47bb6febaa10658c72962b9615d5d5aa2513fa3a/scipy/stats/stats.py#L1024
sourcepub fn kurtosis(&self, fisher: bool, bias: bool) -> PolarsResult<Option<f64>>
Available on crate feature moment only.
pub fn kurtosis(&self, fisher: bool, bias: bool) -> PolarsResult<Option<f64>>
moment only.Compute the kurtosis (Fisher or Pearson) of a dataset.
Kurtosis is the fourth central moment divided by the square of the
variance. If Fisher’s definition is used, then 3.0 is subtracted from
the result to give 0.0 for a normal distribution.
If bias is false then the kurtosis is calculated using k statistics to
eliminate bias coming from biased moment estimators
see: https://github.com/scipy/scipy/blob/47bb6febaa10658c72962b9615d5d5aa2513fa3a/scipy/stats/stats.py#L1027
source§impl Series
impl Series
sourcepub fn full_null(name: &str, size: usize, dtype: &DataType) -> Self
pub fn full_null(name: &str, size: usize, dtype: &DataType) -> Self
Examples found in repository?
src/series/mod.rs (line 158)
157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751 752 753 754 755 756 757 758 759 760 761 762 763 764 765 766 767 768 769 770 771 772 773 774 775 776 777 778 779 780 781 782 783 784 785 786 787 788 789 790 791 792 793 794 795 796 797 798 799 800 801 802 803 804 805 806 807 808 809 810 811 812 813 814 815 816 817 818 819 820 821 822 823 824 825 826 827 828 829 830 831 832 833 834 835 836 837 838 839 840 841 842 843 844 845 846 847 848 849 850 851 852 853 854 855 856 857 858 859 860 861 862 863 864 865 866 867 868 869 870
pub fn new_empty(name: &str, dtype: &DataType) -> Series {
Series::full_null(name, 0, dtype)
}
#[doc(hidden)]
#[cfg(feature = "private")]
pub fn _get_inner_mut(&mut self) -> &mut dyn SeriesTrait {
if Arc::weak_count(&self.0) + Arc::strong_count(&self.0) != 1 {
self.0 = self.0.clone_inner();
}
Arc::get_mut(&mut self.0).expect("implementation error")
}
/// # Safety
/// The caller must ensure the length and the data types of `ArrayRef` does not change.
pub(crate) unsafe fn chunks_mut(&mut self) -> &mut Vec<ArrayRef> {
#[allow(unused_mut)]
let mut ca = self._get_inner_mut();
let chunks = ca.chunks() as *const Vec<ArrayRef> as *mut Vec<ArrayRef>;
// Safety
// ca is the owner of `chunks` and this we do not break aliasing rules
&mut *chunks
}
pub fn set_sorted(&mut self, sorted: IsSorted) {
let inner = self._get_inner_mut();
inner._set_sorted(sorted)
}
pub fn into_frame(self) -> DataFrame {
DataFrame::new_no_checks(vec![self])
}
/// Rename series.
pub fn rename(&mut self, name: &str) -> &mut Series {
self._get_inner_mut().rename(name);
self
}
/// Shrink the capacity of this array to fit its length.
pub fn shrink_to_fit(&mut self) {
self._get_inner_mut().shrink_to_fit()
}
/// Append in place. This is done by adding the chunks of `other` to this [`Series`].
///
/// See [`ChunkedArray::append`] and [`ChunkedArray::extend`].
pub fn append(&mut self, other: &Series) -> PolarsResult<&mut Self> {
self._get_inner_mut().append(other)?;
Ok(self)
}
/// Extend the memory backed by this array with the values from `other`.
///
/// See [`ChunkedArray::extend`] and [`ChunkedArray::append`].
pub fn extend(&mut self, other: &Series) -> PolarsResult<&mut Self> {
self._get_inner_mut().extend(other)?;
Ok(self)
}
pub fn sort(&self, reverse: bool) -> Self {
self.sort_with(SortOptions {
descending: reverse,
..Default::default()
})
}
/// Only implemented for numeric types
pub fn as_single_ptr(&mut self) -> PolarsResult<usize> {
self._get_inner_mut().as_single_ptr()
}
/// Cast `[Series]` to another `[DataType]`
pub fn cast(&self, dtype: &DataType) -> PolarsResult<Self> {
self.0.cast(dtype)
}
/// Compute the sum of all values in this Series.
/// Returns `Some(0)` if the array is empty, and `None` if the array only
/// contains null values.
///
/// If the [`DataType`] is one of `{Int8, UInt8, Int16, UInt16}` the `Series` is
/// first cast to `Int64` to prevent overflow issues.
///
/// ```
/// # use polars_core::prelude::*;
/// let s = Series::new("days", &[1, 2, 3]);
/// assert_eq!(s.sum(), Some(6));
/// ```
pub fn sum<T>(&self) -> Option<T>
where
T: NumCast,
{
self.sum_as_series()
.cast(&DataType::Float64)
.ok()
.and_then(|s| s.f64().unwrap().get(0).and_then(T::from))
}
/// Returns the minimum value in the array, according to the natural order.
/// Returns an option because the array is nullable.
/// ```
/// # use polars_core::prelude::*;
/// let s = Series::new("days", [1, 2, 3].as_ref());
/// assert_eq!(s.min(), Some(1));
/// ```
pub fn min<T>(&self) -> Option<T>
where
T: NumCast,
{
self.min_as_series()
.cast(&DataType::Float64)
.ok()
.and_then(|s| s.f64().unwrap().get(0).and_then(T::from))
}
/// Returns the maximum value in the array, according to the natural order.
/// Returns an option because the array is nullable.
/// ```
/// # use polars_core::prelude::*;
/// let s = Series::new("days", [1, 2, 3].as_ref());
/// assert_eq!(s.max(), Some(3));
/// ```
pub fn max<T>(&self) -> Option<T>
where
T: NumCast,
{
self.max_as_series()
.cast(&DataType::Float64)
.ok()
.and_then(|s| s.f64().unwrap().get(0).and_then(T::from))
}
/// Explode a list or utf8 Series. This expands every item to a new row..
pub fn explode(&self) -> PolarsResult<Series> {
match self.dtype() {
DataType::List(_) => self.list().unwrap().explode(),
DataType::Utf8 => self.utf8().unwrap().explode(),
_ => Err(PolarsError::InvalidOperation(
format!(
"explode not supported for Series with dtype {:?}",
self.dtype()
)
.into(),
)),
}
}
/// Check if float value is NaN (note this is different than missing/ null)
pub fn is_nan(&self) -> PolarsResult<BooleanChunked> {
match self.dtype() {
DataType::Float32 => Ok(self.f32().unwrap().is_nan()),
DataType::Float64 => Ok(self.f64().unwrap().is_nan()),
_ => Err(PolarsError::InvalidOperation(
format!(
"'is_nan' not supported for series with dtype {:?}",
self.dtype()
)
.into(),
)),
}
}
/// Check if float value is NaN (note this is different than missing/ null)
pub fn is_not_nan(&self) -> PolarsResult<BooleanChunked> {
match self.dtype() {
DataType::Float32 => Ok(self.f32().unwrap().is_not_nan()),
DataType::Float64 => Ok(self.f64().unwrap().is_not_nan()),
_ => Err(PolarsError::InvalidOperation(
format!(
"'is_not_nan' not supported for series with dtype {:?}",
self.dtype()
)
.into(),
)),
}
}
/// Check if float value is finite
pub fn is_finite(&self) -> PolarsResult<BooleanChunked> {
match self.dtype() {
DataType::Float32 => Ok(self.f32().unwrap().is_finite()),
DataType::Float64 => Ok(self.f64().unwrap().is_finite()),
_ => Err(PolarsError::InvalidOperation(
format!(
"'is_finite' not supported for series with dtype {:?}",
self.dtype()
)
.into(),
)),
}
}
/// Check if float value is infinite
pub fn is_infinite(&self) -> PolarsResult<BooleanChunked> {
match self.dtype() {
DataType::Float32 => Ok(self.f32().unwrap().is_infinite()),
DataType::Float64 => Ok(self.f64().unwrap().is_infinite()),
_ => Err(PolarsError::InvalidOperation(
format!(
"'is_infinite' not supported for series with dtype {:?}",
self.dtype()
)
.into(),
)),
}
}
/// Create a new ChunkedArray with values from self where the mask evaluates `true` and values
/// from `other` where the mask evaluates `false`
#[cfg(feature = "zip_with")]
#[cfg_attr(docsrs, doc(cfg(feature = "zip_with")))]
pub fn zip_with(&self, mask: &BooleanChunked, other: &Series) -> PolarsResult<Series> {
let (lhs, rhs) = coerce_lhs_rhs(self, other)?;
lhs.zip_with_same_type(mask, rhs.as_ref())
}
/// Cast a datelike Series to their physical representation.
/// Primitives remain unchanged
///
/// * Date -> Int32
/// * Datetime-> Int64
/// * Time -> Int64
/// * Categorical -> UInt32
///
pub fn to_physical_repr(&self) -> Cow<Series> {
use DataType::*;
match self.dtype() {
Date => Cow::Owned(self.cast(&DataType::Int32).unwrap()),
Datetime(_, _) | Duration(_) | Time => Cow::Owned(self.cast(&DataType::Int64).unwrap()),
#[cfg(feature = "dtype-categorical")]
Categorical(_) => Cow::Owned(self.cast(&DataType::UInt32).unwrap()),
_ => Cow::Borrowed(self),
}
}
fn finish_take_threaded(&self, s: Vec<Series>, rechunk: bool) -> Series {
let s = s
.into_iter()
.reduce(|mut s, s1| {
s.append(&s1).unwrap();
s
})
.unwrap();
if rechunk {
s.rechunk()
} else {
s
}
}
// take a function pointer to reduce bloat
fn threaded_op(
&self,
rechunk: bool,
len: usize,
func: &(dyn Fn(usize, usize) -> PolarsResult<Series> + Send + Sync),
) -> PolarsResult<Series> {
let n_threads = POOL.current_num_threads();
let offsets = _split_offsets(len, n_threads);
let series: PolarsResult<Vec<_>> = POOL.install(|| {
offsets
.into_par_iter()
.map(|(offset, len)| func(offset, len))
.collect()
});
Ok(self.finish_take_threaded(series?, rechunk))
}
/// Take by index if ChunkedArray contains a single chunk.
///
/// # Safety
/// This doesn't check any bounds. Null validity is checked.
pub unsafe fn take_unchecked_threaded(
&self,
idx: &IdxCa,
rechunk: bool,
) -> PolarsResult<Series> {
self.threaded_op(rechunk, idx.len(), &|offset, len| {
let idx = idx.slice(offset as i64, len);
self.take_unchecked(&idx)
})
}
/// # Safety
/// This doesn't check any bounds. Null validity is checked.
#[cfg(feature = "chunked_ids")]
pub(crate) unsafe fn _take_chunked_unchecked_threaded(
&self,
chunk_ids: &[ChunkId],
sorted: IsSorted,
rechunk: bool,
) -> Series {
self.threaded_op(rechunk, chunk_ids.len(), &|offset, len| {
let chunk_ids = &chunk_ids[offset..offset + len];
Ok(self._take_chunked_unchecked(chunk_ids, sorted))
})
.unwrap()
}
/// # Safety
/// This doesn't check any bounds. Null validity is checked.
#[cfg(feature = "chunked_ids")]
pub(crate) unsafe fn _take_opt_chunked_unchecked_threaded(
&self,
chunk_ids: &[Option<ChunkId>],
rechunk: bool,
) -> Series {
self.threaded_op(rechunk, chunk_ids.len(), &|offset, len| {
let chunk_ids = &chunk_ids[offset..offset + len];
Ok(self._take_opt_chunked_unchecked(chunk_ids))
})
.unwrap()
}
/// Take by index. This operation is clone.
///
/// # Notes
/// Out of bounds access doesn't Error but will return a Null value
pub fn take_threaded(&self, idx: &IdxCa, rechunk: bool) -> PolarsResult<Series> {
self.threaded_op(rechunk, idx.len(), &|offset, len| {
let idx = idx.slice(offset as i64, len);
self.take(&idx)
})
}
/// Filter by boolean mask. This operation clones data.
pub fn filter_threaded(&self, filter: &BooleanChunked, rechunk: bool) -> PolarsResult<Series> {
// this would fail if there is a broadcasting filter.
// because we cannot split that filter over threads
// besides they are a no-op, so we do the standard filter.
if filter.len() == 1 {
return self.filter(filter);
}
let n_threads = POOL.current_num_threads();
let filters = split_ca(filter, n_threads).unwrap();
let series = split_series(self, n_threads).unwrap();
let series: PolarsResult<Vec<_>> = POOL.install(|| {
filters
.par_iter()
.zip(series)
.map(|(filter, s)| s.filter(filter))
.collect()
});
Ok(self.finish_take_threaded(series?, rechunk))
}
#[cfg(feature = "dot_product")]
#[cfg_attr(docsrs, doc(cfg(feature = "dot_product")))]
pub fn dot(&self, other: &Series) -> Option<f64> {
(self * other).sum::<f64>()
}
/// Get the sum of the Series as a new Series of length 1.
/// Returns a Series with a single zeroed entry if self is an empty numeric series.
///
/// If the [`DataType`] is one of `{Int8, UInt8, Int16, UInt16}` the `Series` is
/// first cast to `Int64` to prevent overflow issues.
pub fn sum_as_series(&self) -> Series {
use DataType::*;
if self.is_empty() && self.dtype().is_numeric() {
return Series::new("", [0])
.cast(self.dtype())
.unwrap()
.sum_as_series();
}
match self.dtype() {
Int8 | UInt8 | Int16 | UInt16 => self.cast(&Int64).unwrap().sum_as_series(),
_ => self._sum_as_series(),
}
}
/// Get an array with the cumulative max computed at every element
#[cfg_attr(docsrs, doc(cfg(feature = "cum_agg")))]
pub fn cummax(&self, _reverse: bool) -> Series {
#[cfg(feature = "cum_agg")]
{
self._cummax(_reverse)
}
#[cfg(not(feature = "cum_agg"))]
{
panic!("activate 'cum_agg' feature")
}
}
/// Get an array with the cumulative min computed at every element
#[cfg_attr(docsrs, doc(cfg(feature = "cum_agg")))]
pub fn cummin(&self, _reverse: bool) -> Series {
#[cfg(feature = "cum_agg")]
{
self._cummin(_reverse)
}
#[cfg(not(feature = "cum_agg"))]
{
panic!("activate 'cum_agg' feature")
}
}
/// Get an array with the cumulative sum computed at every element
///
/// If the [`DataType`] is one of `{Int8, UInt8, Int16, UInt16}` the `Series` is
/// first cast to `Int64` to prevent overflow issues.
#[cfg_attr(docsrs, doc(cfg(feature = "cum_agg")))]
#[allow(unused_variables)]
pub fn cumsum(&self, reverse: bool) -> Series {
#[cfg(feature = "cum_agg")]
{
use DataType::*;
match self.dtype() {
Boolean => self.cast(&DataType::UInt32).unwrap().cumsum(reverse),
Int8 | UInt8 | Int16 | UInt16 => {
let s = self.cast(&Int64).unwrap();
s.cumsum(reverse)
}
Int32 => {
let ca = self.i32().unwrap();
ca.cumsum(reverse).into_series()
}
UInt32 => {
let ca = self.u32().unwrap();
ca.cumsum(reverse).into_series()
}
UInt64 => {
let ca = self.u64().unwrap();
ca.cumsum(reverse).into_series()
}
Int64 => {
let ca = self.i64().unwrap();
ca.cumsum(reverse).into_series()
}
Float32 => {
let ca = self.f32().unwrap();
ca.cumsum(reverse).into_series()
}
Float64 => {
let ca = self.f64().unwrap();
ca.cumsum(reverse).into_series()
}
dt => panic!("cumsum not supported for dtype: {dt:?}"),
}
}
#[cfg(not(feature = "cum_agg"))]
{
panic!("activate 'cum_agg' feature")
}
}
/// Get an array with the cumulative product computed at every element
///
/// If the [`DataType`] is one of `{Int8, UInt8, Int16, UInt16, Int32, UInt32}` the `Series` is
/// first cast to `Int64` to prevent overflow issues.
#[cfg_attr(docsrs, doc(cfg(feature = "cum_agg")))]
#[allow(unused_variables)]
pub fn cumprod(&self, reverse: bool) -> Series {
#[cfg(feature = "cum_agg")]
{
use DataType::*;
match self.dtype() {
Boolean => self.cast(&DataType::Int64).unwrap().cumprod(reverse),
Int8 | UInt8 | Int16 | UInt16 | Int32 | UInt32 => {
let s = self.cast(&Int64).unwrap();
s.cumprod(reverse)
}
Int64 => {
let ca = self.i64().unwrap();
ca.cumprod(reverse).into_series()
}
UInt64 => {
let ca = self.u64().unwrap();
ca.cumprod(reverse).into_series()
}
Float32 => {
let ca = self.f32().unwrap();
ca.cumprod(reverse).into_series()
}
Float64 => {
let ca = self.f64().unwrap();
ca.cumprod(reverse).into_series()
}
dt => panic!("cumprod not supported for dtype: {dt:?}"),
}
}
#[cfg(not(feature = "cum_agg"))]
{
panic!("activate 'cum_agg' feature")
}
}
/// Get the product of an array.
///
/// If the [`DataType`] is one of `{Int8, UInt8, Int16, UInt16}` the `Series` is
/// first cast to `Int64` to prevent overflow issues.
#[cfg_attr(docsrs, doc(cfg(feature = "product")))]
pub fn product(&self) -> Series {
#[cfg(feature = "product")]
{
use DataType::*;
match self.dtype() {
Boolean => self.cast(&DataType::Int64).unwrap().product(),
Int8 | UInt8 | Int16 | UInt16 => {
let s = self.cast(&Int64).unwrap();
s.product()
}
Int64 => {
let ca = self.i64().unwrap();
ca.prod_as_series()
}
Float32 => {
let ca = self.f32().unwrap();
ca.prod_as_series()
}
Float64 => {
let ca = self.f64().unwrap();
ca.prod_as_series()
}
dt => panic!("cumprod not supported for dtype: {dt:?}"),
}
}
#[cfg(not(feature = "product"))]
{
panic!("activate 'product' feature")
}
}
#[cfg(feature = "rank")]
#[cfg_attr(docsrs, doc(cfg(feature = "rank")))]
pub fn rank(&self, options: RankOptions) -> Series {
rank(self, options.method, options.descending)
}
/// Cast throws an error if conversion had overflows
pub fn strict_cast(&self, data_type: &DataType) -> PolarsResult<Series> {
let s = self.cast(data_type)?;
if self.null_count() != s.null_count() {
let failure_mask = !self.is_null() & s.is_null();
let failures = self.filter_threaded(&failure_mask, false)?.unique()?;
Err(PolarsError::ComputeError(
format!(
"Strict conversion from {:?} to {:?} failed for values {}. \
If you were trying to cast Utf8 to Date, Time, or Datetime, \
consider using `strptime`.",
self.dtype(),
data_type,
failures.fmt_list(),
)
.into(),
))
} else {
Ok(s)
}
}
#[cfg(feature = "dtype-time")]
pub(crate) fn into_time(self) -> Series {
#[cfg(not(feature = "dtype-time"))]
{
panic!("activate feature dtype-time")
}
match self.dtype() {
DataType::Int64 => self.i64().unwrap().clone().into_time().into_series(),
DataType::Time => self
.time()
.unwrap()
.as_ref()
.clone()
.into_time()
.into_series(),
dt => panic!("date not implemented for {dt:?}"),
}
}
pub(crate) fn into_date(self) -> Series {
#[cfg(not(feature = "dtype-date"))]
{
panic!("activate feature dtype-date")
}
#[cfg(feature = "dtype-date")]
match self.dtype() {
DataType::Int32 => self.i32().unwrap().clone().into_date().into_series(),
DataType::Date => self
.date()
.unwrap()
.as_ref()
.clone()
.into_date()
.into_series(),
dt => panic!("date not implemented for {dt:?}"),
}
}
pub(crate) fn into_datetime(self, timeunit: TimeUnit, tz: Option<TimeZone>) -> Series {
#[cfg(not(feature = "dtype-datetime"))]
{
panic!("activate feature dtype-datetime")
}
#[cfg(feature = "dtype-datetime")]
match self.dtype() {
DataType::Int64 => self
.i64()
.unwrap()
.clone()
.into_datetime(timeunit, tz)
.into_series(),
DataType::Datetime(_, _) => self
.datetime()
.unwrap()
.as_ref()
.clone()
.into_datetime(timeunit, tz)
.into_series(),
dt => panic!("into_datetime not implemented for {dt:?}"),
}
}
pub(crate) fn into_duration(self, timeunit: TimeUnit) -> Series {
#[cfg(not(feature = "dtype-duration"))]
{
panic!("activate feature dtype-duration")
}
#[cfg(feature = "dtype-duration")]
match self.dtype() {
DataType::Int64 => self
.i64()
.unwrap()
.clone()
.into_duration(timeunit)
.into_series(),
DataType::Duration(_) => self
.duration()
.unwrap()
.as_ref()
.clone()
.into_duration(timeunit)
.into_series(),
dt => panic!("into_duration not implemented for {dt:?}"),
}
}
#[cfg(feature = "abs")]
#[cfg_attr(docsrs, doc(cfg(feature = "abs")))]
/// convert numerical values to their absolute value
pub fn abs(&self) -> PolarsResult<Series> {
let a = self.to_physical_repr();
use DataType::*;
let out = match a.dtype() {
#[cfg(feature = "dtype-i8")]
Int8 => a.i8().unwrap().abs().into_series(),
#[cfg(feature = "dtype-i16")]
Int16 => a.i16().unwrap().abs().into_series(),
Int32 => a.i32().unwrap().abs().into_series(),
Int64 => a.i64().unwrap().abs().into_series(),
UInt8 | UInt16 | UInt32 | UInt64 => self.clone(),
Float32 => a.f32().unwrap().abs().into_series(),
Float64 => a.f64().unwrap().abs().into_series(),
dt => {
return Err(PolarsError::InvalidOperation(
format!("abs not supported for series of type {dt:?}").into(),
));
}
};
Ok(out)
}
#[cfg(feature = "private")]
// used for formatting
pub fn str_value(&self, index: usize) -> PolarsResult<Cow<str>> {
let out = match self.0.get(index)? {
AnyValue::Utf8(s) => Cow::Borrowed(s),
AnyValue::Null => Cow::Borrowed("null"),
#[cfg(feature = "dtype-categorical")]
AnyValue::Categorical(idx, rev) => Cow::Borrowed(rev.get(idx)),
av => Cow::Owned(format!("{av}")),
};
Ok(out)
}
/// Get the head of the Series.
pub fn head(&self, length: Option<usize>) -> Series {
match length {
Some(len) => self.slice(0, std::cmp::min(len, self.len())),
None => self.slice(0, std::cmp::min(10, self.len())),
}
}
/// Get the tail of the Series.
pub fn tail(&self, length: Option<usize>) -> Series {
let len = match length {
Some(len) => std::cmp::min(len, self.len()),
None => std::cmp::min(10, self.len()),
};
self.slice(-(len as i64), len)
}
pub fn mean_as_series(&self) -> Series {
match self.dtype() {
DataType::Float32 => {
let val = &[self.mean().map(|m| m as f32)];
Series::new(self.name(), val)
}
dt if dt.is_numeric() || matches!(dt, DataType::Boolean) => {
let val = &[self.mean()];
Series::new(self.name(), val)
}
dt @ DataType::Duration(_) => {
Series::new(self.name(), &[self.mean().map(|v| v as i64)])
.cast(dt)
.unwrap()
}
_ => return Series::full_null(self.name(), 1, self.dtype()),
}
}More examples
src/series/series_trait.rs (line 131)
130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592
unsafe fn agg_min(&self, groups: &GroupsProxy) -> Series {
Series::full_null(self._field().name(), groups.len(), self._dtype())
}
unsafe fn agg_max(&self, groups: &GroupsProxy) -> Series {
Series::full_null(self._field().name(), groups.len(), self._dtype())
}
/// If the [`DataType`] is one of `{Int8, UInt8, Int16, UInt16}` the `Series` is
/// first cast to `Int64` to prevent overflow issues.
unsafe fn agg_sum(&self, groups: &GroupsProxy) -> Series {
Series::full_null(self._field().name(), groups.len(), self._dtype())
}
unsafe fn agg_std(&self, groups: &GroupsProxy, _ddof: u8) -> Series {
Series::full_null(self._field().name(), groups.len(), self._dtype())
}
unsafe fn agg_var(&self, groups: &GroupsProxy, _ddof: u8) -> Series {
Series::full_null(self._field().name(), groups.len(), self._dtype())
}
unsafe fn agg_list(&self, groups: &GroupsProxy) -> Series {
Series::full_null(self._field().name(), groups.len(), self._dtype())
}
fn zip_outer_join_column(
&self,
_right_column: &Series,
_opt_join_tuples: &[(Option<IdxSize>, Option<IdxSize>)],
) -> Series {
invalid_operation_panic!(self)
}
fn subtract(&self, _rhs: &Series) -> PolarsResult<Series> {
invalid_operation_panic!(self)
}
fn add_to(&self, _rhs: &Series) -> PolarsResult<Series> {
invalid_operation_panic!(self)
}
fn multiply(&self, _rhs: &Series) -> PolarsResult<Series> {
invalid_operation_panic!(self)
}
fn divide(&self, _rhs: &Series) -> PolarsResult<Series> {
invalid_operation_panic!(self)
}
fn remainder(&self, _rhs: &Series) -> PolarsResult<Series> {
invalid_operation_panic!(self)
}
fn group_tuples(&self, _multithreaded: bool, _sorted: bool) -> PolarsResult<GroupsProxy> {
invalid_operation!(self)
}
fn zip_with_same_type(
&self,
_mask: &BooleanChunked,
_other: &Series,
) -> PolarsResult<Series> {
invalid_operation_panic!(self)
}
#[cfg(feature = "sort_multiple")]
fn argsort_multiple(&self, _by: &[Series], _reverse: &[bool]) -> PolarsResult<IdxCa> {
Err(PolarsError::InvalidOperation(
"argsort_multiple is not implemented for this Series".into(),
))
}
}
}
pub trait SeriesTrait:
Send + Sync + private::PrivateSeries + private::PrivateSeriesNumeric
{
/// Check if [`Series`] is sorted.
fn is_sorted(&self) -> IsSorted {
IsSorted::Not
}
/// Rename the Series.
fn rename(&mut self, name: &str);
fn bitand(&self, _other: &Series) -> PolarsResult<Series> {
Err(PolarsError::InvalidOperation(
format!(
"bitwise 'AND' operation not supported for dtype {:?}",
self.dtype()
)
.into(),
))
}
fn bitor(&self, _other: &Series) -> PolarsResult<Series> {
Err(PolarsError::InvalidOperation(
format!(
"bitwise 'OR' operation not supported for dtype {:?}",
self.dtype()
)
.into(),
))
}
fn bitxor(&self, _other: &Series) -> PolarsResult<Series> {
Err(PolarsError::InvalidOperation(
format!(
"bitwise 'XOR' operation not supported for dtype {:?}",
self.dtype()
)
.into(),
))
}
/// Get the lengths of the underlying chunks
fn chunk_lengths(&self) -> ChunkIdIter {
invalid_operation_panic!(self)
}
/// Name of series.
fn name(&self) -> &str {
invalid_operation_panic!(self)
}
/// Get field (used in schema)
fn field(&self) -> Cow<Field> {
self._field()
}
/// Get datatype of series.
fn dtype(&self) -> &DataType {
self._dtype()
}
/// Underlying chunks.
fn chunks(&self) -> &Vec<ArrayRef>;
/// Number of chunks in this Series
fn n_chunks(&self) -> usize {
self.chunks().len()
}
/// Shrink the capacity of this array to fit its length.
fn shrink_to_fit(&mut self) {
panic!("shrink to fit not supported for dtype {:?}", self.dtype())
}
/// Take `num_elements` from the top as a zero copy view.
fn limit(&self, num_elements: usize) -> Series {
self.slice(0, num_elements)
}
/// Get a zero copy view of the data.
///
/// When offset is negative the offset is counted from the
/// end of the array
fn slice(&self, _offset: i64, _length: usize) -> Series {
invalid_operation_panic!(self)
}
#[doc(hidden)]
fn append(&mut self, _other: &Series) -> PolarsResult<()> {
invalid_operation_panic!(self)
}
#[doc(hidden)]
fn extend(&mut self, _other: &Series) -> PolarsResult<()> {
invalid_operation_panic!(self)
}
/// Filter by boolean mask. This operation clones data.
fn filter(&self, _filter: &BooleanChunked) -> PolarsResult<Series> {
invalid_operation_panic!(self)
}
#[doc(hidden)]
#[cfg(feature = "chunked_ids")]
unsafe fn _take_chunked_unchecked(&self, by: &[ChunkId], sorted: IsSorted) -> Series;
#[doc(hidden)]
#[cfg(feature = "chunked_ids")]
unsafe fn _take_opt_chunked_unchecked(&self, by: &[Option<ChunkId>]) -> Series;
/// Take by index from an iterator. This operation clones the data.
fn take_iter(&self, _iter: &mut dyn TakeIterator) -> PolarsResult<Series>;
/// Take by index from an iterator. This operation clones the data.
///
/// # Safety
///
/// - This doesn't check any bounds.
/// - Iterator must be TrustedLen
unsafe fn take_iter_unchecked(&self, _iter: &mut dyn TakeIterator) -> Series;
/// Take by index if ChunkedArray contains a single chunk.
///
/// # Safety
/// This doesn't check any bounds.
unsafe fn take_unchecked(&self, _idx: &IdxCa) -> PolarsResult<Series>;
/// Take by index from an iterator. This operation clones the data.
///
/// # Safety
///
/// - This doesn't check any bounds.
/// - Iterator must be TrustedLen
unsafe fn take_opt_iter_unchecked(&self, _iter: &mut dyn TakeIteratorNulls) -> Series;
/// Take by index from an iterator. This operation clones the data.
/// todo! remove?
#[cfg(feature = "take_opt_iter")]
#[cfg_attr(docsrs, doc(cfg(feature = "take_opt_iter")))]
fn take_opt_iter(&self, _iter: &mut dyn TakeIteratorNulls) -> PolarsResult<Series> {
invalid_operation_panic!(self)
}
/// Take by index. This operation is clone.
fn take(&self, _indices: &IdxCa) -> PolarsResult<Series>;
/// Get length of series.
fn len(&self) -> usize;
/// Check if Series is empty.
fn is_empty(&self) -> bool {
self.len() == 0
}
/// Aggregate all chunks to a contiguous array of memory.
fn rechunk(&self) -> Series {
invalid_operation_panic!(self)
}
/// Take every nth value as a new Series
fn take_every(&self, n: usize) -> Series;
/// Drop all null values and return a new Series.
fn drop_nulls(&self) -> Series {
if self.null_count() == 0 {
Series(self.clone_inner())
} else {
self.filter(&self.is_not_null()).unwrap()
}
}
/// Returns the mean value in the array
/// Returns an option because the array is nullable.
fn mean(&self) -> Option<f64> {
None
}
/// Returns the median value in the array
/// Returns an option because the array is nullable.
fn median(&self) -> Option<f64> {
None
}
/// Create a new Series filled with values from the given index.
///
/// # Example
///
/// ```rust
/// use polars_core::prelude::*;
/// let s = Series::new("a", [0i32, 1, 8]);
/// let s2 = s.new_from_index(2, 4);
/// assert_eq!(Vec::from(s2.i32().unwrap()), &[Some(8), Some(8), Some(8), Some(8)])
/// ```
fn new_from_index(&self, _index: usize, _length: usize) -> Series {
invalid_operation_panic!(self)
}
fn cast(&self, _data_type: &DataType) -> PolarsResult<Series> {
invalid_operation_panic!(self)
}
/// Get a single value by index. Don't use this operation for loops as a runtime cast is
/// needed for every iteration.
fn get(&self, _index: usize) -> PolarsResult<AnyValue> {
invalid_operation_panic!(self)
}
/// Get a single value by index. Don't use this operation for loops as a runtime cast is
/// needed for every iteration.
///
/// This may refer to physical types
///
/// # Safety
/// Does not do any bounds checking
#[cfg(feature = "private")]
unsafe fn get_unchecked(&self, _index: usize) -> AnyValue {
invalid_operation_panic!(self)
}
fn sort_with(&self, _options: SortOptions) -> Series {
invalid_operation_panic!(self)
}
/// Retrieve the indexes needed for a sort.
#[allow(unused)]
fn argsort(&self, options: SortOptions) -> IdxCa {
invalid_operation_panic!(self)
}
/// Count the null values.
fn null_count(&self) -> usize {
invalid_operation_panic!(self)
}
/// Return if any the chunks in this `[ChunkedArray]` have a validity bitmap.
/// no bitmap means no null values.
fn has_validity(&self) -> bool;
/// Get unique values in the Series.
fn unique(&self) -> PolarsResult<Series> {
invalid_operation!(self)
}
/// Get unique values in the Series.
fn n_unique(&self) -> PolarsResult<usize> {
invalid_operation_panic!(self)
}
/// Get first indexes of unique values.
fn arg_unique(&self) -> PolarsResult<IdxCa> {
invalid_operation_panic!(self)
}
/// Get min index
fn arg_min(&self) -> Option<usize> {
None
}
/// Get max index
fn arg_max(&self) -> Option<usize> {
None
}
/// Get a mask of the null values.
fn is_null(&self) -> BooleanChunked {
invalid_operation_panic!(self)
}
/// Get a mask of the non-null values.
fn is_not_null(&self) -> BooleanChunked {
invalid_operation_panic!(self)
}
/// Get a mask of all the unique values.
fn is_unique(&self) -> PolarsResult<BooleanChunked> {
invalid_operation_panic!(self)
}
/// Get a mask of all the duplicated values.
fn is_duplicated(&self) -> PolarsResult<BooleanChunked> {
invalid_operation_panic!(self)
}
/// return a Series in reversed order
fn reverse(&self) -> Series {
invalid_operation_panic!(self)
}
/// Rechunk and return a pointer to the start of the Series.
/// Only implemented for numeric types
fn as_single_ptr(&mut self) -> PolarsResult<usize> {
Err(PolarsError::InvalidOperation(
"operation 'as_single_ptr' not supported".into(),
))
}
/// Shift the values by a given period and fill the parts that will be empty due to this operation
/// with `Nones`.
///
/// *NOTE: If you want to fill the Nones with a value use the
/// [`shift` operation on `ChunkedArray<T>`](../chunked_array/ops/trait.ChunkShift.html).*
///
/// # Example
///
/// ```rust
/// # use polars_core::prelude::*;
/// fn example() -> PolarsResult<()> {
/// let s = Series::new("series", &[1, 2, 3]);
///
/// let shifted = s.shift(1);
/// assert_eq!(Vec::from(shifted.i32()?), &[None, Some(1), Some(2)]);
///
/// let shifted = s.shift(-1);
/// assert_eq!(Vec::from(shifted.i32()?), &[Some(2), Some(3), None]);
///
/// let shifted = s.shift(2);
/// assert_eq!(Vec::from(shifted.i32()?), &[None, None, Some(1)]);
///
/// Ok(())
/// }
/// example();
/// ```
fn shift(&self, _periods: i64) -> Series {
invalid_operation_panic!(self)
}
/// Replace None values with one of the following strategies:
/// * Forward fill (replace None with the previous value)
/// * Backward fill (replace None with the next value)
/// * Mean fill (replace None with the mean of the whole array)
/// * Min fill (replace None with the minimum of the whole array)
/// * Max fill (replace None with the maximum of the whole array)
///
/// *NOTE: If you want to fill the Nones with a value use the
/// [`fill_null` operation on `ChunkedArray<T>`](../chunked_array/ops/trait.ChunkFillNull.html)*.
///
/// # Example
///
/// ```rust
/// # use polars_core::prelude::*;
/// fn example() -> PolarsResult<()> {
/// let s = Series::new("some_missing", &[Some(1), None, Some(2)]);
///
/// let filled = s.fill_null(FillNullStrategy::Forward(None))?;
/// assert_eq!(Vec::from(filled.i32()?), &[Some(1), Some(1), Some(2)]);
///
/// let filled = s.fill_null(FillNullStrategy::Backward(None))?;
/// assert_eq!(Vec::from(filled.i32()?), &[Some(1), Some(2), Some(2)]);
///
/// let filled = s.fill_null(FillNullStrategy::Min)?;
/// assert_eq!(Vec::from(filled.i32()?), &[Some(1), Some(1), Some(2)]);
///
/// let filled = s.fill_null(FillNullStrategy::Max)?;
/// assert_eq!(Vec::from(filled.i32()?), &[Some(1), Some(2), Some(2)]);
///
/// let filled = s.fill_null(FillNullStrategy::Mean)?;
/// assert_eq!(Vec::from(filled.i32()?), &[Some(1), Some(1), Some(2)]);
///
/// Ok(())
/// }
/// example();
/// ```
fn fill_null(&self, _strategy: FillNullStrategy) -> PolarsResult<Series> {
invalid_operation_panic!(self)
}
/// Get the sum of the Series as a new Series of length 1.
///
/// If the [`DataType`] is one of `{Int8, UInt8, Int16, UInt16}` the `Series` is
/// first cast to `Int64` to prevent overflow issues.
fn _sum_as_series(&self) -> Series {
invalid_operation_panic!(self)
}
/// Get the max of the Series as a new Series of length 1.
fn max_as_series(&self) -> Series {
invalid_operation_panic!(self)
}
/// Get the min of the Series as a new Series of length 1.
fn min_as_series(&self) -> Series {
invalid_operation_panic!(self)
}
/// Get the median of the Series as a new Series of length 1.
fn median_as_series(&self) -> Series {
Series::full_null(self.name(), 1, self.dtype())
}
/// Get the variance of the Series as a new Series of length 1.
fn var_as_series(&self, _ddof: u8) -> Series {
Series::full_null(self.name(), 1, self.dtype())
}
/// Get the standard deviation of the Series as a new Series of length 1.
fn std_as_series(&self, _ddof: u8) -> Series {
Series::full_null(self.name(), 1, self.dtype())
}
/// Get the quantile of the ChunkedArray as a new Series of length 1.
fn quantile_as_series(
&self,
_quantile: f64,
_interpol: QuantileInterpolOptions,
) -> PolarsResult<Series> {
Ok(Series::full_null(self.name(), 1, self.dtype()))
}src/frame/groupby/aggregations/dispatch.rs (line 127)
109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184
pub unsafe fn agg_median(&self, groups: &GroupsProxy) -> Series {
use DataType::*;
match self.dtype() {
Float32 => SeriesWrap(self.f32().unwrap().clone()).agg_median(groups),
Float64 => SeriesWrap(self.f64().unwrap().clone()).agg_median(groups),
dt if dt.is_numeric() || dt.is_temporal() => {
let ca = self.to_physical_repr();
let physical_type = ca.dtype();
let s = apply_method_physical_integer!(ca, agg_median, groups);
if dt.is_logical() {
// back to physical and then
// back to logical type
s.cast(physical_type).unwrap().cast(dt).unwrap()
} else {
s
}
}
_ => Series::full_null("", groups.len(), self.dtype()),
}
}
#[doc(hidden)]
pub unsafe fn agg_quantile(
&self,
groups: &GroupsProxy,
quantile: f64,
interpol: QuantileInterpolOptions,
) -> Series {
use DataType::*;
match self.dtype() {
Float32 => {
SeriesWrap(self.f32().unwrap().clone()).agg_quantile(groups, quantile, interpol)
}
Float64 => {
SeriesWrap(self.f64().unwrap().clone()).agg_quantile(groups, quantile, interpol)
}
dt if dt.is_numeric() || dt.is_temporal() => {
let ca = self.to_physical_repr();
let physical_type = ca.dtype();
let s =
apply_method_physical_integer!(ca, agg_quantile, groups, quantile, interpol);
if dt.is_logical() {
// back to physical and then
// back to logical type
s.cast(physical_type).unwrap().cast(dt).unwrap()
} else {
s
}
}
_ => Series::full_null("", groups.len(), self.dtype()),
}
}
#[doc(hidden)]
pub unsafe fn agg_mean(&self, groups: &GroupsProxy) -> Series {
use DataType::*;
match self.dtype() {
Boolean => self.cast(&Float64).unwrap().agg_mean(groups),
Float32 => SeriesWrap(self.f32().unwrap().clone()).agg_mean(groups),
Float64 => SeriesWrap(self.f64().unwrap().clone()).agg_mean(groups),
dt if dt.is_numeric() => {
apply_method_physical_integer!(self, agg_mean, groups)
}
dt @ Duration(_) => {
let s = self.to_physical_repr();
// agg_mean returns Float64
let out = s.agg_mean(groups);
// cast back to Int64 and then to logical duration type
out.cast(&Int64).unwrap().cast(dt).unwrap()
}
_ => Series::full_null("", groups.len(), self.dtype()),
}
}src/series/any_value.rs (line 59)
39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204
fn any_values_to_list(avs: &[AnyValue], inner_type: &DataType) -> ListChunked {
// this is handled downstream. The builder will choose the first non null type
if inner_type == &DataType::Null {
avs.iter()
.map(|av| match av {
AnyValue::List(b) => Some(b.clone()),
_ => None,
})
.collect_trusted()
}
// make sure that wrongly inferred anyvalues don't deviate from the datatype
else {
avs.iter()
.map(|av| match av {
AnyValue::List(b) => {
if b.dtype() == inner_type {
Some(b.clone())
} else {
match b.cast(inner_type) {
Ok(out) => Some(out),
Err(_) => Some(Series::full_null(b.name(), b.len(), inner_type)),
}
}
}
_ => None,
})
.collect_trusted()
}
}
impl<'a, T: AsRef<[AnyValue<'a>]>> NamedFrom<T, [AnyValue<'a>]> for Series {
fn new(name: &str, v: T) -> Self {
let av = v.as_ref();
Series::from_any_values(name, av).unwrap()
}
}
impl Series {
pub fn from_any_values_and_dtype(
name: &str,
av: &[AnyValue],
dtype: &DataType,
) -> PolarsResult<Series> {
let mut s = match dtype {
#[cfg(feature = "dtype-i8")]
DataType::Int8 => any_values_to_primitive::<Int8Type>(av).into_series(),
#[cfg(feature = "dtype-i16")]
DataType::Int16 => any_values_to_primitive::<Int16Type>(av).into_series(),
DataType::Int32 => any_values_to_primitive::<Int32Type>(av).into_series(),
DataType::Int64 => any_values_to_primitive::<Int64Type>(av).into_series(),
#[cfg(feature = "dtype-u8")]
DataType::UInt8 => any_values_to_primitive::<UInt8Type>(av).into_series(),
#[cfg(feature = "dtype-u16")]
DataType::UInt16 => any_values_to_primitive::<UInt16Type>(av).into_series(),
DataType::UInt32 => any_values_to_primitive::<UInt32Type>(av).into_series(),
DataType::UInt64 => any_values_to_primitive::<UInt64Type>(av).into_series(),
DataType::Float32 => any_values_to_primitive::<Float32Type>(av).into_series(),
DataType::Float64 => any_values_to_primitive::<Float64Type>(av).into_series(),
DataType::Utf8 => any_values_to_utf8(av).into_series(),
#[cfg(feature = "dtype-binary")]
DataType::Binary => any_values_to_binary(av).into_series(),
DataType::Boolean => any_values_to_bool(av).into_series(),
#[cfg(feature = "dtype-date")]
DataType::Date => any_values_to_primitive::<Int32Type>(av)
.into_date()
.into_series(),
#[cfg(feature = "dtype-datetime")]
DataType::Datetime(tu, tz) => any_values_to_primitive::<Int64Type>(av)
.into_datetime(*tu, (*tz).clone())
.into_series(),
#[cfg(feature = "dtype-time")]
DataType::Time => any_values_to_primitive::<Int64Type>(av)
.into_time()
.into_series(),
#[cfg(feature = "dtype-duration")]
DataType::Duration(tu) => any_values_to_primitive::<Int64Type>(av)
.into_duration(*tu)
.into_series(),
DataType::List(inner) => any_values_to_list(av, inner).into_series(),
#[cfg(feature = "dtype-struct")]
DataType::Struct(dtype_fields) => {
// the physical series fields of the struct
let mut series_fields = Vec::with_capacity(dtype_fields.len());
for (i, field) in dtype_fields.iter().enumerate() {
let mut field_avs = Vec::with_capacity(av.len());
for av in av.iter() {
match av {
AnyValue::StructOwned(payload) => {
let av_fields = &payload.1;
let av_values = &payload.0;
// all fields are available in this single value
// we can use the index to get value
if dtype_fields.len() == av_fields.len() {
for (l, r) in dtype_fields.iter().zip(av_fields.iter()) {
if l.name() != r.name() {
return Err(PolarsError::ComputeError(
"struct orders must remain the same".into(),
));
}
}
let av_val =
av_values.get(i).cloned().unwrap_or(AnyValue::Null);
field_avs.push(av_val)
}
// not all fields are available, we search the proper field
else {
// search for the name
let mut pushed = false;
for (av_fld, av_val) in av_fields.iter().zip(av_values) {
if av_fld.name == field.name {
field_avs.push(av_val.clone());
pushed = true;
break;
}
}
if !pushed {
field_avs.push(AnyValue::Null)
}
}
}
_ => field_avs.push(AnyValue::Null),
}
}
series_fields.push(Series::new(field.name(), &field_avs))
}
return Ok(StructChunked::new(name, &series_fields)
.unwrap()
.into_series());
}
#[cfg(feature = "object")]
DataType::Object(_) => {
use crate::chunked_array::object::registry;
let converter = registry::get_object_converter();
let mut builder = registry::get_object_builder(name, av.len());
for av in av {
if let AnyValue::Object(val) = av {
builder.append_value(val.as_any())
} else {
let any = converter(av.as_borrowed());
builder.append_value(&*any)
}
}
return Ok(builder.to_series());
}
DataType::Null => {
// TODO!
// use null dtype here and fix tests
Series::full_null(name, av.len(), &DataType::Int32)
}
dt => panic!("{dt:?} not supported"),
};
s.rename(name);
Ok(s)
}
pub fn from_any_values(name: &str, av: &[AnyValue]) -> PolarsResult<Series> {
match av.iter().find(|av| !matches!(av, AnyValue::Null)) {
None => Ok(Series::full_null(name, av.len(), &DataType::Int32)),
Some(av_) => {
let dtype: DataType = av_.into();
Series::from_any_values_and_dtype(name, av, &dtype)
}
}
}src/functions.rs (line 282)
258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290
pub fn diag_concat_df(dfs: &[DataFrame]) -> PolarsResult<DataFrame> {
// TODO! replace with lazy only?
let upper_bound_width = dfs.iter().map(|df| df.width()).sum();
let mut column_names = AHashSet::with_capacity(upper_bound_width);
let mut schema = Vec::with_capacity(upper_bound_width);
for df in dfs {
df.get_columns().iter().for_each(|s| {
let name = s.name();
if column_names.insert(name) {
schema.push((name, s.dtype()))
}
});
}
let dfs = dfs
.iter()
.map(|df| {
let height = df.height();
let mut columns = Vec::with_capacity(schema.len());
for (name, dtype) in &schema {
match df.column(name).ok() {
Some(s) => columns.push(s.clone()),
None => columns.push(Series::full_null(name, height, dtype)),
}
}
DataFrame::new_no_checks(columns)
})
.collect::<Vec<_>>();
concat_df(&dfs)
}Additional examples can be found in:
source§impl Series
impl Series
sourcepub fn round(&self, decimals: u32) -> PolarsResult<Self>
Available on crate feature round_series only.
pub fn round(&self, decimals: u32) -> PolarsResult<Self>
round_series only.Round underlying floating point array to given decimal.
sourcepub fn floor(&self) -> PolarsResult<Self>
Available on crate feature round_series only.
pub fn floor(&self) -> PolarsResult<Self>
round_series only.Floor underlying floating point array to the lowest integers smaller or equal to the float value.
sourcepub fn ceil(&self) -> PolarsResult<Self>
Available on crate feature round_series only.
pub fn ceil(&self) -> PolarsResult<Self>
round_series only.Ceil underlying floating point array to the highest integers smaller or equal to the float value.
sourcepub fn clip(self, min: AnyValue<'_>, max: AnyValue<'_>) -> PolarsResult<Self>
Available on crate feature round_series only.
pub fn clip(self, min: AnyValue<'_>, max: AnyValue<'_>) -> PolarsResult<Self>
round_series only.Clamp underlying values to the min and max values.
sourcepub fn clip_max(self, max: AnyValue<'_>) -> PolarsResult<Self>
Available on crate feature round_series only.
pub fn clip_max(self, max: AnyValue<'_>) -> PolarsResult<Self>
round_series only.Clamp underlying values to the max value.
sourcepub fn clip_min(self, min: AnyValue<'_>) -> PolarsResult<Self>
Available on crate feature round_series only.
pub fn clip_min(self, min: AnyValue<'_>) -> PolarsResult<Self>
round_series only.Clamp underlying values to the min value.
source§impl Series
impl Series
sourcepub fn to_list(&self) -> PolarsResult<ListChunked>
pub fn to_list(&self) -> PolarsResult<ListChunked>
Convert the values of this Series to a ListChunked with a length of 1,
So a Series of:
[1, 2, 3] becomes [[1, 2, 3]]
pub fn reshape(&self, dims: &[i64]) -> PolarsResult<Series>
source§impl Series
impl Series
sourcepub fn unique_counts(&self) -> IdxCa
Available on crate feature unique_counts only.
pub fn unique_counts(&self) -> IdxCa
unique_counts only.Returns a count of the unique values in the order of appearance.
source§impl Series
impl Series
sourcepub fn new_empty(name: &str, dtype: &DataType) -> Series
pub fn new_empty(name: &str, dtype: &DataType) -> Series
Create a new empty Series
Examples found in repository?
More examples
src/chunked_array/ops/explode.rs (line 388)
372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453
fn explode_and_offsets(&self) -> PolarsResult<(Series, OffsetsBuffer<i64>)> {
// A list array's memory layout is actually already 'exploded', so we can just take the values array
// of the list. And we also return a slice of the offsets. This slice can be used to find the old
// list layout or indexes to expand the DataFrame in the same manner as the 'explode' operation
let ca = self.rechunk();
let listarr: &LargeListArray = ca
.downcast_iter()
.next()
.ok_or_else(|| PolarsError::NoData("cannot explode empty list".into()))?;
let offsets_buf = listarr.offsets().clone();
let offsets = listarr.offsets().as_slice();
let mut values = listarr.values().clone();
// all empty
if offsets[offsets.len() - 1] == 0 {
return Ok((
Series::new_empty(self.name(), &self.inner_dtype()),
OffsetsBuffer::new(),
));
}
let mut s = if ca._can_fast_explode() {
// ensure that the value array is sliced
// as a list only slices its offsets on a slice operation
// we only do this in fast-explode as for the other
// branch the offsets must coincide with the values.
if !offsets.is_empty() {
let start = offsets[0] as usize;
let len = offsets[offsets.len() - 1] as usize - start;
// safety:
// we are in bounds
values = unsafe { values.slice_unchecked(start, len) };
}
Series::try_from((self.name(), values)).unwrap()
} else {
// during tests
// test that this code branch is not hit with list arrays that could be fast exploded
#[cfg(test)]
{
let mut last = offsets[0];
let mut has_empty = false;
for &o in &offsets[1..] {
if o == last {
has_empty = true;
}
last = o;
}
if !has_empty && offsets[0] == 0 {
panic!("could have fast exploded")
}
}
let values = Series::try_from((self.name(), values)).unwrap();
values.explode_by_offsets(offsets)
};
debug_assert_eq!(s.name(), self.name());
// make sure we restore the logical type
match self.inner_dtype() {
#[cfg(feature = "dtype-categorical")]
DataType::Categorical(rev_map) => {
let cats = s.u32().unwrap().clone();
// safety:
// rev_map is from same array, so we are still in bounds
s = unsafe {
CategoricalChunked::from_cats_and_rev_map_unchecked(cats, rev_map.unwrap())
.into_series()
};
}
#[cfg(feature = "dtype-date")]
DataType::Date => s = s.into_date(),
#[cfg(feature = "dtype-datetime")]
DataType::Datetime(tu, tz) => s = s.into_datetime(tu, tz),
#[cfg(feature = "dtype-duration")]
DataType::Duration(tu) => s = s.into_duration(tu),
#[cfg(feature = "dtype-time")]
DataType::Time => s = s.into_time(),
_ => {}
}
Ok((s, offsets_buf))
}sourcepub fn set_sorted(&mut self, sorted: IsSorted)
pub fn set_sorted(&mut self, sorted: IsSorted)
Examples found in repository?
src/utils/mod.rs (line 154)
143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165
fn flatten_df(df: &DataFrame) -> impl Iterator<Item = DataFrame> + '_ {
df.iter_chunks_physical().flat_map(|chunk| {
let df = DataFrame::new_no_checks(
df.iter()
.zip(chunk.into_arrays())
.map(|(s, arr)| {
// Safety:
// datatypes are correct
let mut out = unsafe {
Series::from_chunks_and_dtype_unchecked(s.name(), vec![arr], s.dtype())
};
out.set_sorted(s.is_sorted());
out
})
.collect(),
);
if df.height() == 0 {
None
} else {
Some(df)
}
})
}More examples
src/chunked_array/cast.rs (line 84)
61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89
fn cast_impl(&self, data_type: &DataType, checked: bool) -> PolarsResult<Series> {
match data_type {
#[cfg(feature = "dtype-categorical")]
DataType::Categorical(_) => {
Ok(CategoricalChunked::full_null(self.name(), self.len()).into_series())
}
#[cfg(feature = "dtype-struct")]
DataType::Struct(fields) => {
// cast to first field dtype
let fld = &fields[0];
let dtype = &fld.dtype;
let name = &fld.name;
let s = cast_impl_inner(name, &self.chunks, dtype, true)?;
Ok(StructChunked::new_unchecked(self.name(), &[s]).into_series())
}
_ => cast_impl_inner(self.name(), &self.chunks, data_type, checked).map(|mut s| {
// maintain sorted if data types remain signed
// this may still fail with overflow?
if ((self.dtype().is_signed() && data_type.is_signed())
|| (self.dtype().is_unsigned() && data_type.is_unsigned()))
&& (s.null_count() == self.null_count())
{
let is_sorted = self.is_sorted2();
s.set_sorted(is_sorted)
}
s
}),
}
}src/frame/groupby/mod.rs (line 348)
325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375
pub fn keys_sliced(&self, slice: Option<(i64, usize)>) -> Vec<Series> {
#[allow(unused_assignments)]
// needed to keep the lifetimes valid for this scope
let mut groups_owned = None;
let groups = if let Some((offset, len)) = slice {
groups_owned = Some(self.groups.slice(offset, len));
groups_owned.as_deref().unwrap()
} else {
&self.groups
};
POOL.install(|| {
self.selected_keys
.par_iter()
.map(|s| {
match groups {
GroupsProxy::Idx(groups) => {
let mut iter = groups.iter().map(|(first, _idx)| first as usize);
// Safety:
// groups are always in bounds
let mut out = unsafe { s.take_iter_unchecked(&mut iter) };
if groups.sorted {
out.set_sorted(s.is_sorted());
};
out
}
GroupsProxy::Slice { groups, rolling } => {
if *rolling && !groups.is_empty() {
// groups can be sliced
let offset = groups[0][0];
let [upper_offset, upper_len] = groups[groups.len() - 1];
return s.slice(
offset as i64,
((upper_offset + upper_len) - offset) as usize,
);
}
let mut iter = groups.iter().map(|&[first, _len]| first as usize);
// Safety:
// groups are always in bounds
let mut out = unsafe { s.take_iter_unchecked(&mut iter) };
// sliced groups are always in order of discovery
out.set_sorted(s.is_sorted());
out
}
}
})
.collect()
})
}src/frame/mod.rs (line 1891)
1827 1828 1829 1830 1831 1832 1833 1834 1835 1836 1837 1838 1839 1840 1841 1842 1843 1844 1845 1846 1847 1848 1849 1850 1851 1852 1853 1854 1855 1856 1857 1858 1859 1860 1861 1862 1863 1864 1865 1866 1867 1868 1869 1870 1871 1872 1873 1874 1875 1876 1877 1878 1879 1880 1881 1882 1883 1884 1885 1886 1887 1888 1889 1890 1891 1892 1893 1894 1895 1896 1897 1898
pub fn sort_impl(
&self,
by_column: Vec<Series>,
reverse: Vec<bool>,
nulls_last: bool,
slice: Option<(i64, usize)>,
) -> PolarsResult<Self> {
// note that the by_column argument also contains evaluated expression from polars-lazy
// that may not even be present in this dataframe.
// therefore when we try to set the first columns as sorted, we ignore the error
// as expressions are not present (they are renamed to _POLARS_SORT_COLUMN_i.
let first_reverse = reverse[0];
let first_by_column = by_column[0].name().to_string();
let mut take = match by_column.len() {
1 => {
let s = &by_column[0];
let options = SortOptions {
descending: reverse[0],
nulls_last,
};
// fast path for a frame with a single series
// no need to compute the sort indices and then take by these indices
// simply sort and return as frame
if self.width() == 1 && self.check_name_to_idx(s.name()).is_ok() {
let mut out = s.sort_with(options);
if let Some((offset, len)) = slice {
out = out.slice(offset, len);
}
return Ok(out.into_frame());
}
s.argsort(options)
}
_ => {
#[cfg(feature = "sort_multiple")]
{
let (first, by_column, reverse) = prepare_argsort(by_column, reverse)?;
first.argsort_multiple(&by_column, &reverse)?
}
#[cfg(not(feature = "sort_multiple"))]
{
panic!("activate `sort_multiple` feature gate to enable this functionality");
}
}
};
if let Some((offset, len)) = slice {
take = take.slice(offset, len);
}
// Safety:
// the created indices are in bounds
let mut df = if std::env::var("POLARS_VERT_PAR").is_ok() {
unsafe { self.take_unchecked_vectical(&take) }
} else {
unsafe { self.take_unchecked(&take) }
};
// Mark the first sort column as sorted
// if the column did not exists it is ok, because we sorted by an expression
// not present in the dataframe
let _ = df.apply(&first_by_column, |s| {
let mut s = s.clone();
if first_reverse {
s.set_sorted(IsSorted::Descending)
} else {
s.set_sorted(IsSorted::Ascending)
}
s
});
Ok(df)
}sourcepub fn into_frame(self) -> DataFrame
pub fn into_frame(self) -> DataFrame
Examples found in repository?
src/frame/mod.rs (line 1857)
1827 1828 1829 1830 1831 1832 1833 1834 1835 1836 1837 1838 1839 1840 1841 1842 1843 1844 1845 1846 1847 1848 1849 1850 1851 1852 1853 1854 1855 1856 1857 1858 1859 1860 1861 1862 1863 1864 1865 1866 1867 1868 1869 1870 1871 1872 1873 1874 1875 1876 1877 1878 1879 1880 1881 1882 1883 1884 1885 1886 1887 1888 1889 1890 1891 1892 1893 1894 1895 1896 1897 1898
pub fn sort_impl(
&self,
by_column: Vec<Series>,
reverse: Vec<bool>,
nulls_last: bool,
slice: Option<(i64, usize)>,
) -> PolarsResult<Self> {
// note that the by_column argument also contains evaluated expression from polars-lazy
// that may not even be present in this dataframe.
// therefore when we try to set the first columns as sorted, we ignore the error
// as expressions are not present (they are renamed to _POLARS_SORT_COLUMN_i.
let first_reverse = reverse[0];
let first_by_column = by_column[0].name().to_string();
let mut take = match by_column.len() {
1 => {
let s = &by_column[0];
let options = SortOptions {
descending: reverse[0],
nulls_last,
};
// fast path for a frame with a single series
// no need to compute the sort indices and then take by these indices
// simply sort and return as frame
if self.width() == 1 && self.check_name_to_idx(s.name()).is_ok() {
let mut out = s.sort_with(options);
if let Some((offset, len)) = slice {
out = out.slice(offset, len);
}
return Ok(out.into_frame());
}
s.argsort(options)
}
_ => {
#[cfg(feature = "sort_multiple")]
{
let (first, by_column, reverse) = prepare_argsort(by_column, reverse)?;
first.argsort_multiple(&by_column, &reverse)?
}
#[cfg(not(feature = "sort_multiple"))]
{
panic!("activate `sort_multiple` feature gate to enable this functionality");
}
}
};
if let Some((offset, len)) = slice {
take = take.slice(offset, len);
}
// Safety:
// the created indices are in bounds
let mut df = if std::env::var("POLARS_VERT_PAR").is_ok() {
unsafe { self.take_unchecked_vectical(&take) }
} else {
unsafe { self.take_unchecked(&take) }
};
// Mark the first sort column as sorted
// if the column did not exists it is ok, because we sorted by an expression
// not present in the dataframe
let _ = df.apply(&first_by_column, |s| {
let mut s = s.clone();
if first_reverse {
s.set_sorted(IsSorted::Descending)
} else {
s.set_sorted(IsSorted::Ascending)
}
s
});
Ok(df)
}sourcepub fn rename(&mut self, name: &str) -> &mut Series
pub fn rename(&mut self, name: &str) -> &mut Series
Rename series.
Examples found in repository?
src/named_from.rs (line 179)
177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450
fn new(name: &str, s: &Series) -> Self {
let mut s = s.clone();
s.rename(name);
s
}
}
impl<'a, T: AsRef<[&'a str]>> NamedFrom<T, [&'a str]> for Utf8Chunked {
fn new(name: &str, v: T) -> Self {
Utf8Chunked::from_slice(name, v.as_ref())
}
}
impl<'a, T: AsRef<[Option<&'a str>]>> NamedFrom<T, [Option<&'a str>]> for Series {
fn new(name: &str, v: T) -> Self {
Utf8Chunked::from_slice_options(name, v.as_ref()).into_series()
}
}
impl<'a, T: AsRef<[Option<&'a str>]>> NamedFrom<T, [Option<&'a str>]> for Utf8Chunked {
fn new(name: &str, v: T) -> Self {
Utf8Chunked::from_slice_options(name, v.as_ref())
}
}
impl<'a, T: AsRef<[Cow<'a, str>]>> NamedFrom<T, [Cow<'a, str>]> for Series {
fn new(name: &str, v: T) -> Self {
Utf8Chunked::from_iter_values(name, v.as_ref().iter().map(|value| value.as_ref()))
.into_series()
}
}
impl<'a, T: AsRef<[Cow<'a, str>]>> NamedFrom<T, [Cow<'a, str>]> for Utf8Chunked {
fn new(name: &str, v: T) -> Self {
Utf8Chunked::from_iter_values(name, v.as_ref().iter().map(|value| value.as_ref()))
}
}
impl<'a, T: AsRef<[Option<Cow<'a, str>>]>> NamedFrom<T, [Option<Cow<'a, str>>]> for Series {
fn new(name: &str, v: T) -> Self {
Utf8Chunked::new(name, v).into_series()
}
}
impl<'a, T: AsRef<[Option<Cow<'a, str>>]>> NamedFrom<T, [Option<Cow<'a, str>>]> for Utf8Chunked {
fn new(name: &str, v: T) -> Self {
Utf8Chunked::from_iter_options(
name,
v.as_ref()
.iter()
.map(|opt| opt.as_ref().map(|value| value.as_ref())),
)
}
}
#[cfg(feature = "dtype-binary")]
impl<'a, T: AsRef<[&'a [u8]]>> NamedFrom<T, [&'a [u8]]> for Series {
fn new(name: &str, v: T) -> Self {
BinaryChunked::from_slice(name, v.as_ref()).into_series()
}
}
#[cfg(feature = "dtype-binary")]
impl<'a, T: AsRef<[&'a [u8]]>> NamedFrom<T, [&'a [u8]]> for BinaryChunked {
fn new(name: &str, v: T) -> Self {
BinaryChunked::from_slice(name, v.as_ref())
}
}
#[cfg(feature = "dtype-binary")]
impl<'a, T: AsRef<[Option<&'a [u8]>]>> NamedFrom<T, [Option<&'a [u8]>]> for Series {
fn new(name: &str, v: T) -> Self {
BinaryChunked::from_slice_options(name, v.as_ref()).into_series()
}
}
#[cfg(feature = "dtype-binary")]
impl<'a, T: AsRef<[Option<&'a [u8]>]>> NamedFrom<T, [Option<&'a [u8]>]> for BinaryChunked {
fn new(name: &str, v: T) -> Self {
BinaryChunked::from_slice_options(name, v.as_ref())
}
}
#[cfg(feature = "dtype-binary")]
impl<'a, T: AsRef<[Cow<'a, [u8]>]>> NamedFrom<T, [Cow<'a, [u8]>]> for Series {
fn new(name: &str, v: T) -> Self {
BinaryChunked::from_iter_values(name, v.as_ref().iter().map(|value| value.as_ref()))
.into_series()
}
}
#[cfg(feature = "dtype-binary")]
impl<'a, T: AsRef<[Cow<'a, [u8]>]>> NamedFrom<T, [Cow<'a, [u8]>]> for BinaryChunked {
fn new(name: &str, v: T) -> Self {
BinaryChunked::from_iter_values(name, v.as_ref().iter().map(|value| value.as_ref()))
}
}
#[cfg(feature = "dtype-binary")]
impl<'a, T: AsRef<[Option<Cow<'a, [u8]>>]>> NamedFrom<T, [Option<Cow<'a, [u8]>>]> for Series {
fn new(name: &str, v: T) -> Self {
BinaryChunked::new(name, v).into_series()
}
}
#[cfg(feature = "dtype-binary")]
impl<'a, T: AsRef<[Option<Cow<'a, [u8]>>]>> NamedFrom<T, [Option<Cow<'a, [u8]>>]>
for BinaryChunked
{
fn new(name: &str, v: T) -> Self {
BinaryChunked::from_iter_options(
name,
v.as_ref()
.iter()
.map(|opt| opt.as_ref().map(|value| value.as_ref())),
)
}
}
#[cfg(feature = "dtype-date")]
impl<T: AsRef<[NaiveDate]>> NamedFrom<T, [NaiveDate]> for DateChunked {
fn new(name: &str, v: T) -> Self {
DateChunked::from_naive_date(name, v.as_ref().iter().copied())
}
}
#[cfg(feature = "dtype-date")]
impl<T: AsRef<[NaiveDate]>> NamedFrom<T, [NaiveDate]> for Series {
fn new(name: &str, v: T) -> Self {
DateChunked::new(name, v).into_series()
}
}
#[cfg(feature = "dtype-date")]
impl<T: AsRef<[Option<NaiveDate>]>> NamedFrom<T, [Option<NaiveDate>]> for DateChunked {
fn new(name: &str, v: T) -> Self {
DateChunked::from_naive_date_options(name, v.as_ref().iter().copied())
}
}
#[cfg(feature = "dtype-date")]
impl<T: AsRef<[Option<NaiveDate>]>> NamedFrom<T, [Option<NaiveDate>]> for Series {
fn new(name: &str, v: T) -> Self {
DateChunked::new(name, v).into_series()
}
}
#[cfg(feature = "dtype-datetime")]
impl<T: AsRef<[NaiveDateTime]>> NamedFrom<T, [NaiveDateTime]> for DatetimeChunked {
fn new(name: &str, v: T) -> Self {
DatetimeChunked::from_naive_datetime(
name,
v.as_ref().iter().copied(),
TimeUnit::Milliseconds,
)
}
}
#[cfg(feature = "dtype-datetime")]
impl<T: AsRef<[NaiveDateTime]>> NamedFrom<T, [NaiveDateTime]> for Series {
fn new(name: &str, v: T) -> Self {
DatetimeChunked::new(name, v).into_series()
}
}
#[cfg(feature = "dtype-datetime")]
impl<T: AsRef<[Option<NaiveDateTime>]>> NamedFrom<T, [Option<NaiveDateTime>]> for DatetimeChunked {
fn new(name: &str, v: T) -> Self {
DatetimeChunked::from_naive_datetime_options(
name,
v.as_ref().iter().copied(),
TimeUnit::Milliseconds,
)
}
}
#[cfg(feature = "dtype-datetime")]
impl<T: AsRef<[Option<NaiveDateTime>]>> NamedFrom<T, [Option<NaiveDateTime>]> for Series {
fn new(name: &str, v: T) -> Self {
DatetimeChunked::new(name, v).into_series()
}
}
#[cfg(feature = "dtype-duration")]
impl<T: AsRef<[ChronoDuration]>> NamedFrom<T, [ChronoDuration]> for DurationChunked {
fn new(name: &str, v: T) -> Self {
DurationChunked::from_duration(name, v.as_ref().iter().copied(), TimeUnit::Nanoseconds)
}
}
#[cfg(feature = "dtype-duration")]
impl<T: AsRef<[ChronoDuration]>> NamedFrom<T, [ChronoDuration]> for Series {
fn new(name: &str, v: T) -> Self {
DurationChunked::new(name, v).into_series()
}
}
#[cfg(feature = "dtype-duration")]
impl<T: AsRef<[Option<ChronoDuration>]>> NamedFrom<T, [Option<ChronoDuration>]>
for DurationChunked
{
fn new(name: &str, v: T) -> Self {
DurationChunked::from_duration_options(
name,
v.as_ref().iter().copied(),
TimeUnit::Nanoseconds,
)
}
}
#[cfg(feature = "dtype-duration")]
impl<T: AsRef<[Option<ChronoDuration>]>> NamedFrom<T, [Option<ChronoDuration>]> for Series {
fn new(name: &str, v: T) -> Self {
DurationChunked::new(name, v).into_series()
}
}
#[cfg(feature = "dtype-time")]
impl<T: AsRef<[NaiveTime]>> NamedFrom<T, [NaiveTime]> for TimeChunked {
fn new(name: &str, v: T) -> Self {
TimeChunked::from_naive_time(name, v.as_ref().iter().copied())
}
}
#[cfg(feature = "dtype-time")]
impl<T: AsRef<[NaiveTime]>> NamedFrom<T, [NaiveTime]> for Series {
fn new(name: &str, v: T) -> Self {
TimeChunked::new(name, v).into_series()
}
}
#[cfg(feature = "dtype-time")]
impl<T: AsRef<[Option<NaiveTime>]>> NamedFrom<T, [Option<NaiveTime>]> for TimeChunked {
fn new(name: &str, v: T) -> Self {
TimeChunked::from_naive_time_options(name, v.as_ref().iter().copied())
}
}
#[cfg(feature = "dtype-time")]
impl<T: AsRef<[Option<NaiveTime>]>> NamedFrom<T, [Option<NaiveTime>]> for Series {
fn new(name: &str, v: T) -> Self {
TimeChunked::new(name, v).into_series()
}
}
#[cfg(feature = "object")]
impl<T: PolarsObject> NamedFrom<&[T], &[T]> for ObjectChunked<T> {
fn new(name: &str, v: &[T]) -> Self {
ObjectChunked::from_slice(name, v)
}
}
#[cfg(feature = "object")]
impl<T: PolarsObject, S: AsRef<[Option<T>]>> NamedFrom<S, [Option<T>]> for ObjectChunked<T> {
fn new(name: &str, v: S) -> Self {
ObjectChunked::from_slice_options(name, v.as_ref())
}
}
impl<T: PolarsNumericType> ChunkedArray<T> {
/// Specialization that prevents an allocation
/// prefer this over ChunkedArray::new when you have a `Vec<T::Native>` and no null values.
pub fn new_vec(name: &str, v: Vec<T::Native>) -> Self {
ChunkedArray::from_vec(name, v)
}
}
/// For any [`ChunkedArray`] and [`Series`]
impl<T: IntoSeries> NamedFrom<T, T> for Series {
fn new(name: &str, t: T) -> Self {
let mut s = t.into_series();
s.rename(name);
s
}More examples
src/frame/groupby/mod.rs (line 432)
426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751 752 753 754 755 756 757 758 759 760 761 762 763 764 765 766 767 768 769 770 771 772 773 774 775 776 777 778 779 780 781 782 783 784 785 786 787 788 789 790 791 792 793 794 795 796 797 798 799 800 801 802 803 804 805 806 807 808 809 810 811 812 813 814 815 816 817 818 819 820 821 822 823 824 825 826 827 828 829 830 831 832 833 834 835 836 837 838 839 840 841 842 843 844 845 846 847 848 849 850
pub fn mean(&self) -> PolarsResult<DataFrame> {
let (mut cols, agg_cols) = self.prepare_agg()?;
for agg_col in agg_cols {
let new_name = fmt_groupby_column(agg_col.name(), GroupByMethod::Mean);
let mut agg = unsafe { agg_col.agg_mean(&self.groups) };
agg.rename(&new_name);
cols.push(agg);
}
DataFrame::new(cols)
}
/// Aggregate grouped series and compute the sum per group.
///
/// # Example
///
/// ```rust
/// # use polars_core::prelude::*;
/// fn example(df: DataFrame) -> PolarsResult<DataFrame> {
/// df.groupby(["date"])?.select(["temp"]).sum()
/// }
/// ```
/// Returns:
///
/// ```text
/// +------------+----------+
/// | date | temp_sum |
/// | --- | --- |
/// | Date | i32 |
/// +============+==========+
/// | 2020-08-23 | 9 |
/// +------------+----------+
/// | 2020-08-22 | 8 |
/// +------------+----------+
/// | 2020-08-21 | 30 |
/// +------------+----------+
/// ```
#[deprecated(since = "0.24.1", note = "use polars.lazy aggregations")]
pub fn sum(&self) -> PolarsResult<DataFrame> {
let (mut cols, agg_cols) = self.prepare_agg()?;
for agg_col in agg_cols {
let new_name = fmt_groupby_column(agg_col.name(), GroupByMethod::Sum);
let mut agg = unsafe { agg_col.agg_sum(&self.groups) };
agg.rename(&new_name);
cols.push(agg);
}
DataFrame::new(cols)
}
/// Aggregate grouped series and compute the minimal value per group.
///
/// # Example
///
/// ```rust
/// # use polars_core::prelude::*;
/// fn example(df: DataFrame) -> PolarsResult<DataFrame> {
/// df.groupby(["date"])?.select(["temp"]).min()
/// }
/// ```
/// Returns:
///
/// ```text
/// +------------+----------+
/// | date | temp_min |
/// | --- | --- |
/// | Date | i32 |
/// +============+==========+
/// | 2020-08-23 | 9 |
/// +------------+----------+
/// | 2020-08-22 | 1 |
/// +------------+----------+
/// | 2020-08-21 | 10 |
/// +------------+----------+
/// ```
#[deprecated(since = "0.24.1", note = "use polars.lazy aggregations")]
pub fn min(&self) -> PolarsResult<DataFrame> {
let (mut cols, agg_cols) = self.prepare_agg()?;
for agg_col in agg_cols {
let new_name = fmt_groupby_column(agg_col.name(), GroupByMethod::Min);
let mut agg = unsafe { agg_col.agg_min(&self.groups) };
agg.rename(&new_name);
cols.push(agg);
}
DataFrame::new(cols)
}
/// Aggregate grouped series and compute the maximum value per group.
///
/// # Example
///
/// ```rust
/// # use polars_core::prelude::*;
/// fn example(df: DataFrame) -> PolarsResult<DataFrame> {
/// df.groupby(["date"])?.select(["temp"]).max()
/// }
/// ```
/// Returns:
///
/// ```text
/// +------------+----------+
/// | date | temp_max |
/// | --- | --- |
/// | Date | i32 |
/// +============+==========+
/// | 2020-08-23 | 9 |
/// +------------+----------+
/// | 2020-08-22 | 7 |
/// +------------+----------+
/// | 2020-08-21 | 20 |
/// +------------+----------+
/// ```
#[deprecated(since = "0.24.1", note = "use polars.lazy aggregations")]
pub fn max(&self) -> PolarsResult<DataFrame> {
let (mut cols, agg_cols) = self.prepare_agg()?;
for agg_col in agg_cols {
let new_name = fmt_groupby_column(agg_col.name(), GroupByMethod::Max);
let mut agg = unsafe { agg_col.agg_max(&self.groups) };
agg.rename(&new_name);
cols.push(agg);
}
DataFrame::new(cols)
}
/// Aggregate grouped `Series` and find the first value per group.
///
/// # Example
///
/// ```rust
/// # use polars_core::prelude::*;
/// fn example(df: DataFrame) -> PolarsResult<DataFrame> {
/// df.groupby(["date"])?.select(["temp"]).first()
/// }
/// ```
/// Returns:
///
/// ```text
/// +------------+------------+
/// | date | temp_first |
/// | --- | --- |
/// | Date | i32 |
/// +============+============+
/// | 2020-08-23 | 9 |
/// +------------+------------+
/// | 2020-08-22 | 7 |
/// +------------+------------+
/// | 2020-08-21 | 20 |
/// +------------+------------+
/// ```
#[deprecated(since = "0.24.1", note = "use polars.lazy aggregations")]
pub fn first(&self) -> PolarsResult<DataFrame> {
let (mut cols, agg_cols) = self.prepare_agg()?;
for agg_col in agg_cols {
let new_name = fmt_groupby_column(agg_col.name(), GroupByMethod::First);
let mut agg = unsafe { agg_col.agg_first(&self.groups) };
agg.rename(&new_name);
cols.push(agg);
}
DataFrame::new(cols)
}
/// Aggregate grouped `Series` and return the last value per group.
///
/// # Example
///
/// ```rust
/// # use polars_core::prelude::*;
/// fn example(df: DataFrame) -> PolarsResult<DataFrame> {
/// df.groupby(["date"])?.select(["temp"]).last()
/// }
/// ```
/// Returns:
///
/// ```text
/// +------------+------------+
/// | date | temp_last |
/// | --- | --- |
/// | Date | i32 |
/// +============+============+
/// | 2020-08-23 | 9 |
/// +------------+------------+
/// | 2020-08-22 | 1 |
/// +------------+------------+
/// | 2020-08-21 | 10 |
/// +------------+------------+
/// ```
#[deprecated(since = "0.24.1", note = "use polars.lazy aggregations")]
pub fn last(&self) -> PolarsResult<DataFrame> {
let (mut cols, agg_cols) = self.prepare_agg()?;
for agg_col in agg_cols {
let new_name = fmt_groupby_column(agg_col.name(), GroupByMethod::Last);
let mut agg = unsafe { agg_col.agg_last(&self.groups) };
agg.rename(&new_name);
cols.push(agg);
}
DataFrame::new(cols)
}
/// Aggregate grouped `Series` by counting the number of unique values.
///
/// # Example
///
/// ```rust
/// # use polars_core::prelude::*;
/// fn example(df: DataFrame) -> PolarsResult<DataFrame> {
/// df.groupby(["date"])?.select(["temp"]).n_unique()
/// }
/// ```
/// Returns:
///
/// ```text
/// +------------+---------------+
/// | date | temp_n_unique |
/// | --- | --- |
/// | Date | u32 |
/// +============+===============+
/// | 2020-08-23 | 1 |
/// +------------+---------------+
/// | 2020-08-22 | 2 |
/// +------------+---------------+
/// | 2020-08-21 | 2 |
/// +------------+---------------+
/// ```
#[deprecated(since = "0.24.1", note = "use polars.lazy aggregations")]
pub fn n_unique(&self) -> PolarsResult<DataFrame> {
let (mut cols, agg_cols) = self.prepare_agg()?;
for agg_col in agg_cols {
let new_name = fmt_groupby_column(agg_col.name(), GroupByMethod::NUnique);
let mut agg = unsafe { agg_col.agg_n_unique(&self.groups) };
agg.rename(&new_name);
cols.push(agg.into_series());
}
DataFrame::new(cols)
}
/// Aggregate grouped `Series` and determine the quantile per group.
///
/// # Example
///
/// ```rust
/// # use polars_core::prelude::*;
/// # use polars_arrow::prelude::QuantileInterpolOptions;
///
/// fn example(df: DataFrame) -> PolarsResult<DataFrame> {
/// df.groupby(["date"])?.select(["temp"]).quantile(0.2, QuantileInterpolOptions::default())
/// }
/// ```
#[deprecated(since = "0.24.1", note = "use polars.lazy aggregations")]
pub fn quantile(
&self,
quantile: f64,
interpol: QuantileInterpolOptions,
) -> PolarsResult<DataFrame> {
if !(0.0..=1.0).contains(&quantile) {
return Err(PolarsError::ComputeError(
"quantile should be within 0.0 and 1.0".into(),
));
}
let (mut cols, agg_cols) = self.prepare_agg()?;
for agg_col in agg_cols {
let new_name =
fmt_groupby_column(agg_col.name(), GroupByMethod::Quantile(quantile, interpol));
let mut agg = unsafe { agg_col.agg_quantile(&self.groups, quantile, interpol) };
agg.rename(&new_name);
cols.push(agg.into_series());
}
DataFrame::new(cols)
}
/// Aggregate grouped `Series` and determine the median per group.
///
/// # Example
///
/// ```rust
/// # use polars_core::prelude::*;
/// fn example(df: DataFrame) -> PolarsResult<DataFrame> {
/// df.groupby(["date"])?.select(["temp"]).median()
/// }
/// ```
#[deprecated(since = "0.24.1", note = "use polars.lazy aggregations")]
pub fn median(&self) -> PolarsResult<DataFrame> {
let (mut cols, agg_cols) = self.prepare_agg()?;
for agg_col in agg_cols {
let new_name = fmt_groupby_column(agg_col.name(), GroupByMethod::Median);
let mut agg = unsafe { agg_col.agg_median(&self.groups) };
agg.rename(&new_name);
cols.push(agg.into_series());
}
DataFrame::new(cols)
}
/// Aggregate grouped `Series` and determine the variance per group.
#[deprecated(since = "0.24.1", note = "use polars.lazy aggregations")]
pub fn var(&self, ddof: u8) -> PolarsResult<DataFrame> {
let (mut cols, agg_cols) = self.prepare_agg()?;
for agg_col in agg_cols {
let new_name = fmt_groupby_column(agg_col.name(), GroupByMethod::Var(ddof));
let mut agg = unsafe { agg_col.agg_var(&self.groups, ddof) };
agg.rename(&new_name);
cols.push(agg.into_series());
}
DataFrame::new(cols)
}
/// Aggregate grouped `Series` and determine the standard deviation per group.
#[deprecated(since = "0.24.1", note = "use polars.lazy aggregations")]
pub fn std(&self, ddof: u8) -> PolarsResult<DataFrame> {
let (mut cols, agg_cols) = self.prepare_agg()?;
for agg_col in agg_cols {
let new_name = fmt_groupby_column(agg_col.name(), GroupByMethod::Std(ddof));
let mut agg = unsafe { agg_col.agg_std(&self.groups, ddof) };
agg.rename(&new_name);
cols.push(agg.into_series());
}
DataFrame::new(cols)
}
/// Aggregate grouped series and compute the number of values per group.
///
/// # Example
///
/// ```rust
/// # use polars_core::prelude::*;
/// fn example(df: DataFrame) -> PolarsResult<DataFrame> {
/// df.groupby(["date"])?.select(["temp"]).count()
/// }
/// ```
/// Returns:
///
/// ```text
/// +------------+------------+
/// | date | temp_count |
/// | --- | --- |
/// | Date | u32 |
/// +============+============+
/// | 2020-08-23 | 1 |
/// +------------+------------+
/// | 2020-08-22 | 2 |
/// +------------+------------+
/// | 2020-08-21 | 2 |
/// +------------+------------+
/// ```
pub fn count(&self) -> PolarsResult<DataFrame> {
let (mut cols, agg_cols) = self.prepare_agg()?;
for agg_col in agg_cols {
let new_name = fmt_groupby_column(agg_col.name(), GroupByMethod::Count);
let mut ca = self.groups.group_count();
ca.rename(&new_name);
cols.push(ca.into_series());
}
DataFrame::new(cols)
}
/// Get the groupby group indexes.
///
/// # Example
///
/// ```rust
/// # use polars_core::prelude::*;
/// fn example(df: DataFrame) -> PolarsResult<DataFrame> {
/// df.groupby(["date"])?.groups()
/// }
/// ```
/// Returns:
///
/// ```text
/// +--------------+------------+
/// | date | groups |
/// | --- | --- |
/// | Date(days) | list [u32] |
/// +==============+============+
/// | 2020-08-23 | "[3]" |
/// +--------------+------------+
/// | 2020-08-22 | "[2, 4]" |
/// +--------------+------------+
/// | 2020-08-21 | "[0, 1]" |
/// +--------------+------------+
/// ```
pub fn groups(&self) -> PolarsResult<DataFrame> {
let mut cols = self.keys();
let mut column = self.groups.as_list_chunked();
let new_name = fmt_groupby_column("", GroupByMethod::Groups);
column.rename(&new_name);
cols.push(column.into_series());
DataFrame::new(cols)
}
/// Aggregate the groups of the groupby operation into lists.
///
/// # Example
///
/// ```rust
/// # use polars_core::prelude::*;
/// fn example(df: DataFrame) -> PolarsResult<DataFrame> {
/// // GroupBy and aggregate to Lists
/// df.groupby(["date"])?.select(["temp"]).agg_list()
/// }
/// ```
/// Returns:
///
/// ```text
/// +------------+------------------------+
/// | date | temp_agg_list |
/// | --- | --- |
/// | Date | list [i32] |
/// +============+========================+
/// | 2020-08-23 | "[Some(9)]" |
/// +------------+------------------------+
/// | 2020-08-22 | "[Some(7), Some(1)]" |
/// +------------+------------------------+
/// | 2020-08-21 | "[Some(20), Some(10)]" |
/// +------------+------------------------+
/// ```
#[deprecated(since = "0.24.1", note = "use polars.lazy aggregations")]
pub fn agg_list(&self) -> PolarsResult<DataFrame> {
let (mut cols, agg_cols) = self.prepare_agg()?;
for agg_col in agg_cols {
let new_name = fmt_groupby_column(agg_col.name(), GroupByMethod::List);
let mut agg = unsafe { agg_col.agg_list(&self.groups) };
agg.rename(&new_name);
cols.push(agg);
}
DataFrame::new(cols)
}src/frame/cross_join.rs (line 106)
93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110
pub fn _cross_join_with_names(
&self,
other: &DataFrame,
names: &[String],
) -> PolarsResult<DataFrame> {
let (mut l_df, r_df) = self.cross_join_dfs(other, None, false)?;
l_df.get_columns_mut().extend_from_slice(&r_df.columns);
l_df.get_columns_mut()
.iter_mut()
.zip(names)
.for_each(|(s, name)| {
if s.name() != name {
s.rename(name);
}
});
Ok(l_df)
}src/frame/mod.rs (line 594)
576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751 752 753 754 755 756 757 758 759 760 761 762 763 764 765 766 767 768 769 770 771 772 773 774 775 776 777 778 779 780 781 782 783 784 785 786 787 788 789 790 791 792 793 794 795 796 797 798 799 800 801 802 803 804 805 806 807 808 809 810 811 812 813 814 815 816 817 818 819 820 821 822 823 824 825 826 827 828 829 830 831 832 833 834 835 836 837 838 839 840 841 842 843 844 845 846 847 848 849 850 851 852 853 854 855 856 857 858 859 860 861 862 863 864 865 866 867 868 869 870 871 872 873 874 875 876 877 878 879 880 881 882 883 884 885 886 887 888 889 890 891 892 893 894 895 896 897 898 899 900 901 902 903 904 905 906 907 908 909 910 911 912 913 914 915 916 917 918 919 920 921 922 923 924 925 926 927 928 929 930 931 932 933 934 935 936 937 938 939 940 941 942 943 944 945 946 947 948 949 950 951 952 953 954 955 956 957 958 959 960 961 962 963 964 965 966 967 968 969 970 971 972 973 974 975 976 977 978 979 980 981 982 983 984 985 986 987 988 989 990 991 992 993 994 995 996 997 998 999 1000 1001 1002 1003 1004 1005 1006 1007 1008 1009 1010 1011 1012 1013 1014 1015 1016 1017 1018 1019 1020 1021 1022 1023 1024 1025 1026 1027 1028 1029 1030 1031 1032 1033 1034 1035 1036 1037 1038 1039 1040 1041 1042 1043 1044 1045 1046 1047 1048 1049 1050 1051 1052 1053 1054 1055 1056 1057 1058 1059 1060 1061 1062 1063 1064 1065 1066 1067 1068 1069 1070 1071 1072 1073 1074 1075 1076 1077 1078 1079 1080 1081 1082 1083 1084 1085 1086 1087 1088 1089 1090 1091 1092 1093 1094 1095 1096 1097 1098 1099 1100 1101 1102 1103 1104 1105 1106 1107 1108 1109 1110 1111 1112 1113 1114 1115 1116 1117 1118 1119 1120 1121 1122 1123 1124 1125 1126 1127 1128 1129 1130 1131 1132 1133 1134 1135 1136 1137 1138 1139 1140 1141 1142 1143 1144 1145 1146 1147 1148 1149 1150 1151 1152 1153 1154 1155 1156 1157 1158 1159 1160 1161 1162 1163 1164 1165 1166 1167 1168 1169 1170 1171 1172 1173 1174 1175 1176 1177 1178 1179 1180 1181 1182 1183 1184 1185 1186 1187 1188 1189 1190 1191 1192 1193 1194 1195 1196 1197 1198 1199 1200 1201 1202 1203 1204 1205 1206 1207 1208 1209 1210 1211 1212 1213 1214 1215 1216 1217 1218 1219 1220 1221 1222 1223 1224 1225 1226 1227 1228 1229 1230 1231 1232 1233 1234 1235 1236 1237 1238 1239 1240 1241 1242 1243 1244 1245 1246 1247 1248 1249 1250 1251 1252 1253 1254 1255 1256 1257 1258 1259 1260 1261 1262 1263 1264 1265 1266 1267 1268 1269 1270 1271 1272 1273 1274 1275 1276 1277 1278 1279 1280 1281 1282 1283 1284 1285 1286 1287 1288 1289 1290 1291 1292 1293 1294 1295 1296 1297 1298 1299 1300 1301 1302 1303 1304 1305 1306 1307 1308 1309 1310 1311 1312 1313 1314 1315 1316 1317 1318 1319 1320 1321 1322 1323 1324 1325 1326 1327 1328 1329 1330 1331 1332 1333 1334 1335 1336 1337 1338 1339 1340 1341 1342 1343 1344 1345 1346 1347 1348 1349 1350 1351 1352 1353 1354 1355 1356 1357 1358 1359 1360 1361 1362 1363 1364 1365 1366 1367 1368 1369 1370 1371 1372 1373 1374 1375 1376 1377 1378 1379 1380 1381 1382 1383 1384 1385 1386 1387 1388 1389 1390 1391 1392 1393 1394 1395 1396 1397 1398 1399 1400 1401 1402 1403 1404 1405 1406 1407 1408 1409 1410 1411 1412 1413 1414 1415 1416 1417 1418 1419 1420 1421 1422 1423 1424 1425 1426 1427 1428 1429 1430 1431 1432 1433 1434 1435 1436 1437 1438 1439 1440 1441 1442 1443 1444 1445 1446 1447 1448 1449 1450 1451 1452 1453 1454 1455 1456 1457 1458 1459 1460 1461 1462 1463 1464 1465 1466 1467 1468 1469 1470 1471 1472 1473 1474 1475 1476 1477 1478 1479 1480 1481 1482 1483 1484 1485 1486 1487 1488 1489 1490 1491 1492 1493 1494 1495 1496 1497 1498 1499 1500 1501 1502 1503 1504 1505 1506 1507 1508 1509 1510 1511 1512 1513 1514 1515 1516 1517 1518 1519 1520 1521 1522 1523 1524 1525 1526 1527 1528 1529 1530 1531 1532 1533 1534 1535 1536 1537 1538 1539 1540 1541 1542 1543 1544 1545 1546 1547 1548 1549 1550 1551 1552 1553 1554 1555 1556 1557 1558 1559 1560 1561 1562 1563 1564 1565 1566 1567 1568 1569 1570 1571 1572 1573 1574 1575 1576 1577 1578 1579 1580 1581 1582 1583 1584 1585 1586 1587 1588 1589 1590 1591 1592 1593 1594 1595 1596 1597 1598 1599 1600 1601 1602 1603 1604 1605 1606 1607 1608 1609 1610 1611 1612 1613 1614 1615 1616 1617 1618 1619 1620 1621 1622 1623 1624 1625 1626 1627 1628 1629 1630 1631 1632 1633 1634 1635 1636 1637 1638 1639 1640 1641 1642 1643 1644 1645 1646 1647 1648 1649 1650 1651 1652 1653 1654 1655 1656 1657 1658 1659 1660 1661 1662 1663 1664 1665 1666 1667 1668 1669 1670 1671 1672 1673 1674 1675 1676 1677 1678 1679 1680 1681 1682 1683 1684 1685 1686 1687 1688 1689 1690 1691 1692 1693 1694 1695 1696 1697 1698 1699 1700 1701 1702 1703 1704 1705 1706 1707 1708 1709 1710 1711 1712 1713 1714 1715 1716 1717 1718 1719 1720 1721 1722 1723 1724 1725 1726 1727 1728 1729 1730 1731 1732 1733 1734 1735 1736 1737 1738 1739 1740 1741 1742 1743 1744 1745 1746 1747 1748 1749 1750 1751 1752 1753 1754 1755 1756 1757 1758 1759 1760 1761 1762 1763 1764 1765 1766 1767 1768 1769 1770 1771 1772 1773 1774 1775 1776 1777 1778 1779 1780 1781 1782 1783 1784 1785 1786 1787 1788 1789 1790 1791 1792 1793 1794 1795 1796 1797 1798 1799 1800 1801 1802 1803 1804 1805 1806 1807 1808 1809 1810 1811 1812 1813 1814 1815 1816 1817 1818 1819 1820 1821 1822 1823 1824 1825 1826 1827 1828 1829 1830 1831 1832 1833 1834 1835 1836 1837 1838 1839 1840 1841 1842 1843 1844 1845 1846 1847 1848 1849 1850 1851 1852 1853 1854 1855 1856 1857 1858 1859 1860 1861 1862 1863 1864 1865 1866 1867 1868 1869 1870 1871 1872 1873 1874 1875 1876 1877 1878 1879 1880 1881 1882 1883 1884 1885 1886 1887 1888 1889 1890 1891 1892 1893 1894 1895 1896 1897 1898 1899 1900 1901 1902 1903 1904 1905 1906 1907 1908 1909 1910 1911 1912 1913 1914 1915 1916 1917 1918 1919 1920 1921 1922 1923 1924 1925 1926 1927 1928 1929 1930 1931 1932 1933 1934 1935 1936 1937 1938 1939 1940 1941 1942 1943 1944 1945 1946 1947 1948 1949 1950 1951 1952 1953 1954 1955 1956 1957 1958 1959 1960 1961 1962 1963 1964 1965 1966 1967 1968 1969 1970 1971 1972 1973 1974 1975 1976 1977 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 2036 2037 2038 2039 2040 2041 2042 2043 2044 2045 2046 2047 2048 2049 2050 2051 2052 2053 2054 2055 2056 2057 2058 2059 2060 2061 2062 2063 2064 2065 2066 2067 2068 2069 2070 2071 2072 2073 2074 2075 2076 2077 2078 2079 2080 2081 2082 2083 2084 2085 2086 2087 2088 2089 2090 2091 2092 2093 2094 2095 2096 2097 2098 2099 2100 2101 2102 2103 2104 2105 2106 2107 2108 2109 2110 2111 2112 2113 2114 2115 2116 2117 2118 2119 2120 2121 2122 2123 2124 2125 2126 2127 2128 2129 2130 2131 2132 2133 2134 2135 2136 2137 2138 2139 2140 2141 2142 2143 2144 2145 2146 2147 2148 2149 2150 2151 2152 2153 2154 2155 2156 2157 2158 2159 2160 2161 2162 2163 2164 2165 2166 2167 2168 2169 2170 2171 2172 2173 2174 2175 2176 2177 2178 2179 2180 2181 2182 2183 2184 2185 2186 2187 2188 2189 2190 2191
pub fn set_column_names<S: AsRef<str>>(&mut self, names: &[S]) -> PolarsResult<()> {
if names.len() != self.columns.len() {
return Err(PolarsError::ShapeMisMatch("the provided slice with column names has not the same size as the DataFrame's width".into()));
}
let unique_names: AHashSet<&str, ahash::RandomState> =
AHashSet::from_iter(names.iter().map(|name| name.as_ref()));
if unique_names.len() != self.columns.len() {
return Err(PolarsError::SchemaMisMatch(
"duplicate column names found".into(),
));
}
let columns = mem::take(&mut self.columns);
self.columns = columns
.into_iter()
.zip(names)
.map(|(s, name)| {
let mut s = s;
s.rename(name.as_ref());
s
})
.collect();
Ok(())
}
/// Get the data types of the columns in the DataFrame.
///
/// # Example
///
/// ```rust
/// # use polars_core::prelude::*;
/// let venus_air: DataFrame = df!("Element" => &["Carbon dioxide", "Nitrogen"],
/// "Fraction" => &[0.965, 0.035])?;
///
/// assert_eq!(venus_air.dtypes(), &[DataType::Utf8, DataType::Float64]);
/// # Ok::<(), PolarsError>(())
/// ```
pub fn dtypes(&self) -> Vec<DataType> {
self.columns.iter().map(|s| s.dtype().clone()).collect()
}
/// The number of chunks per column
pub fn n_chunks(&self) -> usize {
match self.columns.get(0) {
None => 0,
Some(s) => s.n_chunks(),
}
}
/// Get a reference to the schema fields of the `DataFrame`.
///
/// # Example
///
/// ```rust
/// # use polars_core::prelude::*;
/// let earth: DataFrame = df!("Surface type" => &["Water", "Land"],
/// "Fraction" => &[0.708, 0.292])?;
///
/// let f1: Field = Field::new("Surface type", DataType::Utf8);
/// let f2: Field = Field::new("Fraction", DataType::Float64);
///
/// assert_eq!(earth.fields(), &[f1, f2]);
/// # Ok::<(), PolarsError>(())
/// ```
pub fn fields(&self) -> Vec<Field> {
self.columns
.iter()
.map(|s| s.field().into_owned())
.collect()
}
/// Get (height, width) of the `DataFrame`.
///
/// # Example
///
/// ```rust
/// # use polars_core::prelude::*;
/// let df0: DataFrame = DataFrame::default();
/// let df1: DataFrame = df!("1" => &[1, 2, 3, 4, 5])?;
/// let df2: DataFrame = df!("1" => &[1, 2, 3, 4, 5],
/// "2" => &[1, 2, 3, 4, 5])?;
///
/// assert_eq!(df0.shape(), (0 ,0));
/// assert_eq!(df1.shape(), (5, 1));
/// assert_eq!(df2.shape(), (5, 2));
/// # Ok::<(), PolarsError>(())
/// ```
pub fn shape(&self) -> (usize, usize) {
match self.columns.as_slice() {
&[] => (0, 0),
v => (v[0].len(), v.len()),
}
}
/// Get the width of the `DataFrame` which is the number of columns.
///
/// # Example
///
/// ```rust
/// # use polars_core::prelude::*;
/// let df0: DataFrame = DataFrame::default();
/// let df1: DataFrame = df!("Series 1" => &[0; 0])?;
/// let df2: DataFrame = df!("Series 1" => &[0; 0],
/// "Series 2" => &[0; 0])?;
///
/// assert_eq!(df0.width(), 0);
/// assert_eq!(df1.width(), 1);
/// assert_eq!(df2.width(), 2);
/// # Ok::<(), PolarsError>(())
/// ```
pub fn width(&self) -> usize {
self.columns.len()
}
/// Get the height of the `DataFrame` which is the number of rows.
///
/// # Example
///
/// ```rust
/// # use polars_core::prelude::*;
/// let df0: DataFrame = DataFrame::default();
/// let df1: DataFrame = df!("Currency" => &["€", "$"])?;
/// let df2: DataFrame = df!("Currency" => &["€", "$", "¥", "£", "₿"])?;
///
/// assert_eq!(df0.height(), 0);
/// assert_eq!(df1.height(), 2);
/// assert_eq!(df2.height(), 5);
/// # Ok::<(), PolarsError>(())
/// ```
pub fn height(&self) -> usize {
self.shape().0
}
/// Check if the `DataFrame` is empty.
///
/// # Example
///
/// ```rust
/// # use polars_core::prelude::*;
/// let df1: DataFrame = DataFrame::default();
/// assert!(df1.is_empty());
///
/// let df2: DataFrame = df!("First name" => &["Forever"],
/// "Last name" => &["Alone"])?;
/// assert!(!df2.is_empty());
/// # Ok::<(), PolarsError>(())
/// ```
pub fn is_empty(&self) -> bool {
self.columns.is_empty()
}
pub(crate) fn hstack_mut_no_checks(&mut self, columns: &[Series]) -> &mut Self {
for col in columns {
self.columns.push(col.clone());
}
self
}
/// Add multiple `Series` to a `DataFrame`.
/// The added `Series` are required to have the same length.
///
/// # Example
///
/// ```rust
/// # use polars_core::prelude::*;
/// fn stack(df: &mut DataFrame, columns: &[Series]) {
/// df.hstack_mut(columns);
/// }
/// ```
pub fn hstack_mut(&mut self, columns: &[Series]) -> PolarsResult<&mut Self> {
let mut names = PlHashSet::with_capacity(self.columns.len());
for s in &self.columns {
names.insert(s.name());
}
let height = self.height();
// first loop check validity. We don't do this in a single pass otherwise
// this DataFrame is already modified when an error occurs.
for col in columns {
if col.len() != height && height != 0 {
return Err(PolarsError::ShapeMisMatch(
format!("Could not horizontally stack Series. The Series length {} differs from the DataFrame height: {height}", col.len()).into()));
}
let name = col.name();
if names.contains(name) {
return Err(PolarsError::Duplicate(
format!("Cannot do hstack operation. Column with name: {name} already exists",)
.into(),
));
}
names.insert(name);
}
drop(names);
Ok(self.hstack_mut_no_checks(columns))
}
/// Add multiple `Series` to a `DataFrame`.
/// The added `Series` are required to have the same length.
///
/// # Example
///
/// ```rust
/// # use polars_core::prelude::*;
/// let df1: DataFrame = df!("Element" => &["Copper", "Silver", "Gold"])?;
/// let s1: Series = Series::new("Proton", &[29, 47, 79]);
/// let s2: Series = Series::new("Electron", &[29, 47, 79]);
///
/// let df2: DataFrame = df1.hstack(&[s1, s2])?;
/// assert_eq!(df2.shape(), (3, 3));
/// println!("{}", df2);
/// # Ok::<(), PolarsError>(())
/// ```
///
/// Output:
///
/// ```text
/// shape: (3, 3)
/// +---------+--------+----------+
/// | Element | Proton | Electron |
/// | --- | --- | --- |
/// | str | i32 | i32 |
/// +=========+========+==========+
/// | Copper | 29 | 29 |
/// +---------+--------+----------+
/// | Silver | 47 | 47 |
/// +---------+--------+----------+
/// | Gold | 79 | 79 |
/// +---------+--------+----------+
/// ```
pub fn hstack(&self, columns: &[Series]) -> PolarsResult<Self> {
let mut new_cols = self.columns.clone();
new_cols.extend_from_slice(columns);
DataFrame::new(new_cols)
}
/// Concatenate a `DataFrame` to this `DataFrame` and return as newly allocated `DataFrame`.
///
/// If many `vstack` operations are done, it is recommended to call [`DataFrame::rechunk`].
///
/// # Example
///
/// ```rust
/// # use polars_core::prelude::*;
/// let df1: DataFrame = df!("Element" => &["Copper", "Silver", "Gold"],
/// "Melting Point (K)" => &[1357.77, 1234.93, 1337.33])?;
/// let df2: DataFrame = df!("Element" => &["Platinum", "Palladium"],
/// "Melting Point (K)" => &[2041.4, 1828.05])?;
///
/// let df3: DataFrame = df1.vstack(&df2)?;
///
/// assert_eq!(df3.shape(), (5, 2));
/// println!("{}", df3);
/// # Ok::<(), PolarsError>(())
/// ```
///
/// Output:
///
/// ```text
/// shape: (5, 2)
/// +-----------+-------------------+
/// | Element | Melting Point (K) |
/// | --- | --- |
/// | str | f64 |
/// +===========+===================+
/// | Copper | 1357.77 |
/// +-----------+-------------------+
/// | Silver | 1234.93 |
/// +-----------+-------------------+
/// | Gold | 1337.33 |
/// +-----------+-------------------+
/// | Platinum | 2041.4 |
/// +-----------+-------------------+
/// | Palladium | 1828.05 |
/// +-----------+-------------------+
/// ```
pub fn vstack(&self, other: &DataFrame) -> PolarsResult<Self> {
let mut df = self.clone();
df.vstack_mut(other)?;
Ok(df)
}
/// Concatenate a DataFrame to this DataFrame
///
/// If many `vstack` operations are done, it is recommended to call [`DataFrame::rechunk`].
///
/// # Example
///
/// ```rust
/// # use polars_core::prelude::*;
/// let mut df1: DataFrame = df!("Element" => &["Copper", "Silver", "Gold"],
/// "Melting Point (K)" => &[1357.77, 1234.93, 1337.33])?;
/// let df2: DataFrame = df!("Element" => &["Platinum", "Palladium"],
/// "Melting Point (K)" => &[2041.4, 1828.05])?;
///
/// df1.vstack_mut(&df2)?;
///
/// assert_eq!(df1.shape(), (5, 2));
/// println!("{}", df1);
/// # Ok::<(), PolarsError>(())
/// ```
///
/// Output:
///
/// ```text
/// shape: (5, 2)
/// +-----------+-------------------+
/// | Element | Melting Point (K) |
/// | --- | --- |
/// | str | f64 |
/// +===========+===================+
/// | Copper | 1357.77 |
/// +-----------+-------------------+
/// | Silver | 1234.93 |
/// +-----------+-------------------+
/// | Gold | 1337.33 |
/// +-----------+-------------------+
/// | Platinum | 2041.4 |
/// +-----------+-------------------+
/// | Palladium | 1828.05 |
/// +-----------+-------------------+
/// ```
pub fn vstack_mut(&mut self, other: &DataFrame) -> PolarsResult<&mut Self> {
if self.width() != other.width() {
if self.width() == 0 {
self.columns = other.columns.clone();
return Ok(self);
}
return Err(PolarsError::ShapeMisMatch(
format!("Could not vertically stack DataFrame. The DataFrames appended width {} differs from the parent DataFrames width {}", self.width(), other.width()).into()
));
}
self.columns
.iter_mut()
.zip(other.columns.iter())
.try_for_each::<_, PolarsResult<_>>(|(left, right)| {
can_extend(left, right)?;
left.append(right).expect("should not fail");
Ok(())
})?;
Ok(self)
}
/// Does not check if schema is correct
pub(crate) fn vstack_mut_unchecked(&mut self, other: &DataFrame) {
self.columns
.iter_mut()
.zip(other.columns.iter())
.for_each(|(left, right)| {
left.append(right).expect("should not fail");
});
}
/// Extend the memory backed by this [`DataFrame`] with the values from `other`.
///
/// Different from [`vstack`](Self::vstack) which adds the chunks from `other` to the chunks of this [`DataFrame`]
/// `extend` appends the data from `other` to the underlying memory locations and thus may cause a reallocation.
///
/// If this does not cause a reallocation, the resulting data structure will not have any extra chunks
/// and thus will yield faster queries.
///
/// Prefer `extend` over `vstack` when you want to do a query after a single append. For instance during
/// online operations where you add `n` rows and rerun a query.
///
/// Prefer `vstack` over `extend` when you want to append many times before doing a query. For instance
/// when you read in multiple files and when to store them in a single `DataFrame`. In the latter case, finish the sequence
/// of `append` operations with a [`rechunk`](Self::rechunk).
pub fn extend(&mut self, other: &DataFrame) -> PolarsResult<()> {
if self.width() != other.width() {
return Err(PolarsError::ShapeMisMatch(
format!("Could not extend DataFrame. The DataFrames extended width {} differs from the parent DataFrames width {}", self.width(), other.width()).into()
));
}
self.columns
.iter_mut()
.zip(other.columns.iter())
.try_for_each::<_, PolarsResult<_>>(|(left, right)| {
can_extend(left, right)?;
left.extend(right).unwrap();
Ok(())
})?;
Ok(())
}
/// Remove a column by name and return the column removed.
///
/// # Example
///
/// ```rust
/// # use polars_core::prelude::*;
/// let mut df: DataFrame = df!("Animal" => &["Tiger", "Lion", "Great auk"],
/// "IUCN" => &["Endangered", "Vulnerable", "Extinct"])?;
///
/// let s1: PolarsResult<Series> = df.drop_in_place("Average weight");
/// assert!(s1.is_err());
///
/// let s2: Series = df.drop_in_place("Animal")?;
/// assert_eq!(s2, Series::new("Animal", &["Tiger", "Lion", "Great auk"]));
/// # Ok::<(), PolarsError>(())
/// ```
pub fn drop_in_place(&mut self, name: &str) -> PolarsResult<Series> {
let idx = self.check_name_to_idx(name)?;
Ok(self.columns.remove(idx))
}
/// Return a new `DataFrame` where all null values are dropped.
///
/// # Example
///
/// ```no_run
/// # use polars_core::prelude::*;
/// let df1: DataFrame = df!("Country" => ["Malta", "Liechtenstein", "North Korea"],
/// "Tax revenue (% GDP)" => [Some(32.7), None, None])?;
/// assert_eq!(df1.shape(), (3, 2));
///
/// let df2: DataFrame = df1.drop_nulls(None)?;
/// assert_eq!(df2.shape(), (1, 2));
/// println!("{}", df2);
/// # Ok::<(), PolarsError>(())
/// ```
///
/// Output:
///
/// ```text
/// shape: (1, 2)
/// +---------+---------------------+
/// | Country | Tax revenue (% GDP) |
/// | --- | --- |
/// | str | f64 |
/// +=========+=====================+
/// | Malta | 32.7 |
/// +---------+---------------------+
/// ```
pub fn drop_nulls(&self, subset: Option<&[String]>) -> PolarsResult<Self> {
let selected_series;
let mut iter = match subset {
Some(cols) => {
selected_series = self.select_series(cols)?;
selected_series.iter()
}
None => self.columns.iter(),
};
// fast path for no nulls in df
if iter.clone().all(|s| !s.has_validity()) {
return Ok(self.clone());
}
let mask = iter
.next()
.ok_or_else(|| PolarsError::NoData("No data to drop nulls from".into()))?;
let mut mask = mask.is_not_null();
for s in iter {
mask = mask & s.is_not_null();
}
self.filter(&mask)
}
/// Drop a column by name.
/// This is a pure method and will return a new `DataFrame` instead of modifying
/// the current one in place.
///
/// # Example
///
/// ```rust
/// # use polars_core::prelude::*;
/// let df1: DataFrame = df!("Ray type" => &["α", "β", "X", "γ"])?;
/// let df2: DataFrame = df1.drop("Ray type")?;
///
/// assert!(df2.is_empty());
/// # Ok::<(), PolarsError>(())
/// ```
pub fn drop(&self, name: &str) -> PolarsResult<Self> {
let idx = self.check_name_to_idx(name)?;
let mut new_cols = Vec::with_capacity(self.columns.len() - 1);
self.columns.iter().enumerate().for_each(|(i, s)| {
if i != idx {
new_cols.push(s.clone())
}
});
Ok(DataFrame::new_no_checks(new_cols))
}
pub fn drop_many<S: AsRef<str>>(&self, names: &[S]) -> Self {
let names = names.iter().map(|s| s.as_ref()).collect();
fn inner(df: &DataFrame, names: Vec<&str>) -> DataFrame {
let mut new_cols = Vec::with_capacity(df.columns.len() - names.len());
df.columns.iter().for_each(|s| {
if !names.contains(&s.name()) {
new_cols.push(s.clone())
}
});
DataFrame::new_no_checks(new_cols)
}
inner(self, names)
}
fn insert_at_idx_no_name_check(
&mut self,
index: usize,
series: Series,
) -> PolarsResult<&mut Self> {
if series.len() == self.height() {
self.columns.insert(index, series);
Ok(self)
} else {
Err(PolarsError::ShapeMisMatch(
format!(
"Could not add column. The Series length {} differs from the DataFrame height: {}",
series.len(),
self.height()
)
.into(),
))
}
}
/// Insert a new column at a given index.
pub fn insert_at_idx<S: IntoSeries>(
&mut self,
index: usize,
column: S,
) -> PolarsResult<&mut Self> {
let series = column.into_series();
self.check_already_present(series.name())?;
self.insert_at_idx_no_name_check(index, series)
}
fn add_column_by_search(&mut self, series: Series) -> PolarsResult<()> {
if let Some(idx) = self.find_idx_by_name(series.name()) {
self.replace_at_idx(idx, series)?;
} else {
self.columns.push(series);
}
Ok(())
}
/// Add a new column to this `DataFrame` or replace an existing one.
pub fn with_column<S: IntoSeries>(&mut self, column: S) -> PolarsResult<&mut Self> {
fn inner(df: &mut DataFrame, mut series: Series) -> PolarsResult<&mut DataFrame> {
let height = df.height();
if series.len() == 1 && height > 1 {
series = series.new_from_index(0, height);
}
if series.len() == height || df.is_empty() {
df.add_column_by_search(series)?;
Ok(df)
}
// special case for literals
else if height == 0 && series.len() == 1 {
let s = series.slice(0, 0);
df.add_column_by_search(s)?;
Ok(df)
} else {
Err(PolarsError::ShapeMisMatch(
format!(
"Could not add column. The Series length {} differs from the DataFrame height: {}",
series.len(),
df.height()
)
.into(),
))
}
}
let series = column.into_series();
inner(self, series)
}
fn add_column_by_schema(&mut self, s: Series, schema: &Schema) -> PolarsResult<()> {
let name = s.name();
if let Some((idx, _, _)) = schema.get_full(name) {
// schema is incorrect fallback to search
if self.columns.get(idx).map(|s| s.name()) != Some(name) {
self.add_column_by_search(s)?;
} else {
self.replace_at_idx(idx, s)?;
}
} else {
self.columns.push(s);
}
Ok(())
}
pub fn _add_columns(&mut self, columns: Vec<Series>, schema: &Schema) -> PolarsResult<()> {
for (i, s) in columns.into_iter().enumerate() {
// we need to branch here
// because users can add multiple columns with the same name
if i == 0 || schema.get(s.name()).is_some() {
self.with_column_and_schema(s, schema)?;
} else {
self.with_column(s.clone())?;
}
}
Ok(())
}
/// Add a new column to this `DataFrame` or replace an existing one.
/// Uses an existing schema to amortize lookups.
/// If the schema is incorrect, we will fallback to linear search.
pub fn with_column_and_schema<S: IntoSeries>(
&mut self,
column: S,
schema: &Schema,
) -> PolarsResult<&mut Self> {
let mut series = column.into_series();
let height = self.height();
if series.len() == 1 && height > 1 {
series = series.new_from_index(0, height);
}
if series.len() == height || self.is_empty() {
self.add_column_by_schema(series, schema)?;
Ok(self)
}
// special case for literals
else if height == 0 && series.len() == 1 {
let s = series.slice(0, 0);
self.add_column_by_schema(s, schema)?;
Ok(self)
} else {
Err(PolarsError::ShapeMisMatch(
format!(
"Could not add column. The Series length {} differs from the DataFrame height: {}",
series.len(),
self.height()
)
.into(),
))
}
}
/// Get a row in the `DataFrame`. Beware this is slow.
///
/// # Example
///
/// ```
/// # use polars_core::prelude::*;
/// fn example(df: &mut DataFrame, idx: usize) -> Option<Vec<AnyValue>> {
/// df.get(idx)
/// }
/// ```
pub fn get(&self, idx: usize) -> Option<Vec<AnyValue>> {
match self.columns.get(0) {
Some(s) => {
if s.len() <= idx {
return None;
}
}
None => return None,
}
// safety: we just checked bounds
unsafe { Some(self.columns.iter().map(|s| s.get_unchecked(idx)).collect()) }
}
/// Select a `Series` by index.
///
/// # Example
///
/// ```rust
/// # use polars_core::prelude::*;
/// let df: DataFrame = df!("Star" => &["Sun", "Betelgeuse", "Sirius A", "Sirius B"],
/// "Absolute magnitude" => &[4.83, -5.85, 1.42, 11.18])?;
///
/// let s1: Option<&Series> = df.select_at_idx(0);
/// let s2: Series = Series::new("Star", &["Sun", "Betelgeuse", "Sirius A", "Sirius B"]);
///
/// assert_eq!(s1, Some(&s2));
/// # Ok::<(), PolarsError>(())
/// ```
pub fn select_at_idx(&self, idx: usize) -> Option<&Series> {
self.columns.get(idx)
}
/// Select a mutable series by index.
///
/// *Note: the length of the Series should remain the same otherwise the DataFrame is invalid.*
/// For this reason the method is not public
fn select_at_idx_mut(&mut self, idx: usize) -> Option<&mut Series> {
self.columns.get_mut(idx)
}
/// Select column(s) from this `DataFrame` by range and return a new DataFrame
///
/// # Examples
///
/// ```rust
/// # use polars_core::prelude::*;
/// let df = df! {
/// "0" => &[0, 0, 0],
/// "1" => &[1, 1, 1],
/// "2" => &[2, 2, 2]
/// }?;
///
/// assert!(df.select(&["0", "1"])?.frame_equal(&df.select_by_range(0..=1)?));
/// assert!(df.frame_equal(&df.select_by_range(..)?));
/// # Ok::<(), PolarsError>(())
/// ```
pub fn select_by_range<R>(&self, range: R) -> PolarsResult<Self>
where
R: ops::RangeBounds<usize>,
{
// This function is copied from std::slice::range (https://doc.rust-lang.org/std/slice/fn.range.html)
// because it is the nightly feature. We should change here if this function were stable.
fn get_range<R>(range: R, bounds: ops::RangeTo<usize>) -> ops::Range<usize>
where
R: ops::RangeBounds<usize>,
{
let len = bounds.end;
let start: ops::Bound<&usize> = range.start_bound();
let start = match start {
ops::Bound::Included(&start) => start,
ops::Bound::Excluded(start) => start.checked_add(1).unwrap_or_else(|| {
panic!("attempted to index slice from after maximum usize");
}),
ops::Bound::Unbounded => 0,
};
let end: ops::Bound<&usize> = range.end_bound();
let end = match end {
ops::Bound::Included(end) => end.checked_add(1).unwrap_or_else(|| {
panic!("attempted to index slice up to maximum usize");
}),
ops::Bound::Excluded(&end) => end,
ops::Bound::Unbounded => len,
};
if start > end {
panic!("slice index starts at {start} but ends at {end}");
}
if end > len {
panic!("range end index {end} out of range for slice of length {len}",);
}
ops::Range { start, end }
}
let colnames = self.get_column_names_owned();
let range = get_range(range, ..colnames.len());
self.select_impl(&colnames[range])
}
/// Get column index of a `Series` by name.
/// # Example
///
/// ```rust
/// # use polars_core::prelude::*;
/// let df: DataFrame = df!("Name" => &["Player 1", "Player 2", "Player 3"],
/// "Health" => &[100, 200, 500],
/// "Mana" => &[250, 100, 0],
/// "Strength" => &[30, 150, 300])?;
///
/// assert_eq!(df.find_idx_by_name("Name"), Some(0));
/// assert_eq!(df.find_idx_by_name("Health"), Some(1));
/// assert_eq!(df.find_idx_by_name("Mana"), Some(2));
/// assert_eq!(df.find_idx_by_name("Strength"), Some(3));
/// assert_eq!(df.find_idx_by_name("Haste"), None);
/// # Ok::<(), PolarsError>(())
/// ```
pub fn find_idx_by_name(&self, name: &str) -> Option<usize> {
self.columns.iter().position(|s| s.name() == name)
}
/// Select a single column by name.
///
/// # Example
///
/// ```rust
/// # use polars_core::prelude::*;
/// let s1: Series = Series::new("Password", &["123456", "[]B$u$g$s$B#u#n#n#y[]{}"]);
/// let s2: Series = Series::new("Robustness", &["Weak", "Strong"]);
/// let df: DataFrame = DataFrame::new(vec![s1.clone(), s2])?;
///
/// assert_eq!(df.column("Password")?, &s1);
/// # Ok::<(), PolarsError>(())
/// ```
pub fn column(&self, name: &str) -> PolarsResult<&Series> {
let idx = self
.find_idx_by_name(name)
.ok_or_else(|| PolarsError::NotFound(name.to_string().into()))?;
Ok(self.select_at_idx(idx).unwrap())
}
/// Selected multiple columns by name.
///
/// # Example
///
/// ```rust
/// # use polars_core::prelude::*;
/// let df: DataFrame = df!("Latin name" => &["Oncorhynchus kisutch", "Salmo salar"],
/// "Max weight (kg)" => &[16.0, 35.89])?;
/// let sv: Vec<&Series> = df.columns(&["Latin name", "Max weight (kg)"])?;
///
/// assert_eq!(&df[0], sv[0]);
/// assert_eq!(&df[1], sv[1]);
/// # Ok::<(), PolarsError>(())
/// ```
pub fn columns<I, S>(&self, names: I) -> PolarsResult<Vec<&Series>>
where
I: IntoIterator<Item = S>,
S: AsRef<str>,
{
names
.into_iter()
.map(|name| self.column(name.as_ref()))
.collect()
}
/// Select column(s) from this `DataFrame` and return a new `DataFrame`.
///
/// # Examples
///
/// ```
/// # use polars_core::prelude::*;
/// fn example(df: &DataFrame) -> PolarsResult<DataFrame> {
/// df.select(["foo", "bar"])
/// }
/// ```
pub fn select<I, S>(&self, selection: I) -> PolarsResult<Self>
where
I: IntoIterator<Item = S>,
S: AsRef<str>,
{
let cols = selection
.into_iter()
.map(|s| s.as_ref().to_string())
.collect::<Vec<_>>();
self.select_impl(&cols)
}
fn select_impl(&self, cols: &[String]) -> PolarsResult<Self> {
self.select_check_duplicates(cols)?;
let selected = self.select_series_impl(cols)?;
Ok(DataFrame::new_no_checks(selected))
}
pub fn select_physical<I, S>(&self, selection: I) -> PolarsResult<Self>
where
I: IntoIterator<Item = S>,
S: AsRef<str>,
{
let cols = selection
.into_iter()
.map(|s| s.as_ref().to_string())
.collect::<Vec<_>>();
self.select_physical_impl(&cols)
}
fn select_physical_impl(&self, cols: &[String]) -> PolarsResult<Self> {
self.select_check_duplicates(cols)?;
let selected = self.select_series_physical_impl(cols)?;
Ok(DataFrame::new_no_checks(selected))
}
fn select_check_duplicates(&self, cols: &[String]) -> PolarsResult<()> {
let mut names = PlHashSet::with_capacity(cols.len());
for name in cols {
if !names.insert(name.as_str()) {
_duplicate_err(name)?
}
}
Ok(())
}
/// Select column(s) from this `DataFrame` and return them into a `Vec`.
///
/// # Example
///
/// ```rust
/// # use polars_core::prelude::*;
/// let df: DataFrame = df!("Name" => &["Methane", "Ethane", "Propane"],
/// "Carbon" => &[1, 2, 3],
/// "Hydrogen" => &[4, 6, 8])?;
/// let sv: Vec<Series> = df.select_series(&["Carbon", "Hydrogen"])?;
///
/// assert_eq!(df["Carbon"], sv[0]);
/// assert_eq!(df["Hydrogen"], sv[1]);
/// # Ok::<(), PolarsError>(())
/// ```
pub fn select_series(&self, selection: impl IntoVec<String>) -> PolarsResult<Vec<Series>> {
let cols = selection.into_vec();
self.select_series_impl(&cols)
}
fn _names_to_idx_map(&self) -> PlHashMap<&str, usize> {
self.columns
.iter()
.enumerate()
.map(|(i, s)| (s.name(), i))
.collect()
}
/// A non generic implementation to reduce compiler bloat.
fn select_series_physical_impl(&self, cols: &[String]) -> PolarsResult<Vec<Series>> {
let selected = if cols.len() > 1 && self.columns.len() > 10 {
let name_to_idx = self._names_to_idx_map();
cols.iter()
.map(|name| {
let idx = *name_to_idx
.get(name.as_str())
.ok_or_else(|| PolarsError::NotFound(name.to_string().into()))?;
Ok(self
.select_at_idx(idx)
.unwrap()
.to_physical_repr()
.into_owned())
})
.collect::<PolarsResult<Vec<_>>>()?
} else {
cols.iter()
.map(|c| self.column(c).map(|s| s.to_physical_repr().into_owned()))
.collect::<PolarsResult<Vec<_>>>()?
};
Ok(selected)
}
/// A non generic implementation to reduce compiler bloat.
fn select_series_impl(&self, cols: &[String]) -> PolarsResult<Vec<Series>> {
let selected = if cols.len() > 1 && self.columns.len() > 10 {
// we hash, because there are user that having millions of columns.
// # https://github.com/pola-rs/polars/issues/1023
let name_to_idx = self._names_to_idx_map();
cols.iter()
.map(|name| {
let idx = *name_to_idx
.get(name.as_str())
.ok_or_else(|| PolarsError::NotFound(name.to_string().into()))?;
Ok(self.select_at_idx(idx).unwrap().clone())
})
.collect::<PolarsResult<Vec<_>>>()?
} else {
cols.iter()
.map(|c| self.column(c).map(|s| s.clone()))
.collect::<PolarsResult<Vec<_>>>()?
};
Ok(selected)
}
/// Select a mutable series by name.
/// *Note: the length of the Series should remain the same otherwise the DataFrame is invalid.*
/// For this reason the method is not public
fn select_mut(&mut self, name: &str) -> Option<&mut Series> {
let opt_idx = self.find_idx_by_name(name);
match opt_idx {
Some(idx) => self.select_at_idx_mut(idx),
None => None,
}
}
/// Does a filter but splits thread chunks vertically instead of horizontally
/// This yields a DataFrame with `n_chunks == n_threads`.
fn filter_vertical(&mut self, mask: &BooleanChunked) -> PolarsResult<Self> {
let n_threads = POOL.current_num_threads();
let masks = split_ca(mask, n_threads).unwrap();
let dfs = split_df(self, n_threads).unwrap();
let dfs: PolarsResult<Vec<_>> = POOL.install(|| {
masks
.par_iter()
.zip(dfs)
.map(|(mask, df)| {
let cols = df
.columns
.iter()
.map(|s| s.filter(mask))
.collect::<PolarsResult<_>>()?;
Ok(DataFrame::new_no_checks(cols))
})
.collect()
});
let mut iter = dfs?.into_iter();
let first = iter.next().unwrap();
Ok(iter.fold(first, |mut acc, df| {
acc.vstack_mut(&df).unwrap();
acc
}))
}
/// Take the `DataFrame` rows by a boolean mask.
///
/// # Example
///
/// ```
/// # use polars_core::prelude::*;
/// fn example(df: &DataFrame) -> PolarsResult<DataFrame> {
/// let mask = df.column("sepal.width")?.is_not_null();
/// df.filter(&mask)
/// }
/// ```
pub fn filter(&self, mask: &BooleanChunked) -> PolarsResult<Self> {
if std::env::var("POLARS_VERT_PAR").is_ok() {
return self.clone().filter_vertical(mask);
}
let new_col = self.try_apply_columns_par(&|s| match s.dtype() {
DataType::Utf8 => s.filter_threaded(mask, true),
_ => s.filter(mask),
})?;
Ok(DataFrame::new_no_checks(new_col))
}
/// Same as `filter` but does not parallelize.
pub fn _filter_seq(&self, mask: &BooleanChunked) -> PolarsResult<Self> {
let new_col = self.try_apply_columns(&|s| s.filter(mask))?;
Ok(DataFrame::new_no_checks(new_col))
}
/// Take `DataFrame` value by indexes from an iterator.
///
/// # Example
///
/// ```
/// # use polars_core::prelude::*;
/// fn example(df: &DataFrame) -> PolarsResult<DataFrame> {
/// let iterator = (0..9).into_iter();
/// df.take_iter(iterator)
/// }
/// ```
pub fn take_iter<I>(&self, iter: I) -> PolarsResult<Self>
where
I: Iterator<Item = usize> + Clone + Sync + TrustedLen,
{
let new_col = self.try_apply_columns_par(&|s| {
let mut i = iter.clone();
s.take_iter(&mut i)
})?;
Ok(DataFrame::new_no_checks(new_col))
}
/// Take `DataFrame` values by indexes from an iterator.
///
/// # Safety
///
/// This doesn't do any bound checking but checks null validity.
#[must_use]
pub unsafe fn take_iter_unchecked<I>(&self, mut iter: I) -> Self
where
I: Iterator<Item = usize> + Clone + Sync + TrustedLen,
{
if std::env::var("POLARS_VERT_PAR").is_ok() {
let idx_ca: NoNull<IdxCa> = iter.into_iter().map(|idx| idx as IdxSize).collect();
return self.take_unchecked_vectical(&idx_ca.into_inner());
}
let n_chunks = self.n_chunks();
let has_utf8 = self
.columns
.iter()
.any(|s| matches!(s.dtype(), DataType::Utf8));
if (n_chunks == 1 && self.width() > 1) || has_utf8 {
let idx_ca: NoNull<IdxCa> = iter.into_iter().map(|idx| idx as IdxSize).collect();
let idx_ca = idx_ca.into_inner();
return self.take_unchecked(&idx_ca);
}
let new_col = if self.width() == 1 {
self.columns
.iter()
.map(|s| s.take_iter_unchecked(&mut iter))
.collect::<Vec<_>>()
} else {
self.apply_columns_par(&|s| {
let mut i = iter.clone();
s.take_iter_unchecked(&mut i)
})
};
DataFrame::new_no_checks(new_col)
}
/// Take `DataFrame` values by indexes from an iterator that may contain None values.
///
/// # Safety
///
/// This doesn't do any bound checking. Out of bounds may access uninitialized memory.
/// Null validity is checked
#[must_use]
pub unsafe fn take_opt_iter_unchecked<I>(&self, mut iter: I) -> Self
where
I: Iterator<Item = Option<usize>> + Clone + Sync + TrustedLen,
{
if std::env::var("POLARS_VERT_PAR").is_ok() {
let idx_ca: IdxCa = iter
.into_iter()
.map(|opt| opt.map(|v| v as IdxSize))
.collect();
return self.take_unchecked_vectical(&idx_ca);
}
let n_chunks = self.n_chunks();
let has_utf8 = self
.columns
.iter()
.any(|s| matches!(s.dtype(), DataType::Utf8));
if (n_chunks == 1 && self.width() > 1) || has_utf8 {
let idx_ca: IdxCa = iter
.into_iter()
.map(|opt| opt.map(|v| v as IdxSize))
.collect();
return self.take_unchecked(&idx_ca);
}
let new_col = if self.width() == 1 {
self.columns
.iter()
.map(|s| s.take_opt_iter_unchecked(&mut iter))
.collect::<Vec<_>>()
} else {
self.apply_columns_par(&|s| {
let mut i = iter.clone();
s.take_opt_iter_unchecked(&mut i)
})
};
DataFrame::new_no_checks(new_col)
}
/// Take `DataFrame` rows by index values.
///
/// # Example
///
/// ```
/// # use polars_core::prelude::*;
/// fn example(df: &DataFrame) -> PolarsResult<DataFrame> {
/// let idx = IdxCa::new("idx", &[0, 1, 9]);
/// df.take(&idx)
/// }
/// ```
pub fn take(&self, indices: &IdxCa) -> PolarsResult<Self> {
let indices = if indices.chunks.len() > 1 {
Cow::Owned(indices.rechunk())
} else {
Cow::Borrowed(indices)
};
let new_col = POOL.install(|| {
self.try_apply_columns_par(&|s| match s.dtype() {
DataType::Utf8 => s.take_threaded(&indices, true),
_ => s.take(&indices),
})
})?;
Ok(DataFrame::new_no_checks(new_col))
}
pub(crate) unsafe fn take_unchecked(&self, idx: &IdxCa) -> Self {
self.take_unchecked_impl(idx, true)
}
unsafe fn take_unchecked_impl(&self, idx: &IdxCa, allow_threads: bool) -> Self {
let cols = if allow_threads {
POOL.install(|| {
self.apply_columns_par(&|s| match s.dtype() {
DataType::Utf8 => s.take_unchecked_threaded(idx, true).unwrap(),
_ => s.take_unchecked(idx).unwrap(),
})
})
} else {
self.columns
.iter()
.map(|s| s.take_unchecked(idx).unwrap())
.collect()
};
DataFrame::new_no_checks(cols)
}
unsafe fn take_unchecked_vectical(&self, indices: &IdxCa) -> Self {
let n_threads = POOL.current_num_threads();
let idxs = split_ca(indices, n_threads).unwrap();
let dfs: Vec<_> = POOL.install(|| {
idxs.par_iter()
.map(|idx| {
let cols = self
.columns
.iter()
.map(|s| s.take_unchecked(idx).unwrap())
.collect();
DataFrame::new_no_checks(cols)
})
.collect()
});
let mut iter = dfs.into_iter();
let first = iter.next().unwrap();
iter.fold(first, |mut acc, df| {
acc.vstack_mut(&df).unwrap();
acc
})
}
/// Rename a column in the `DataFrame`.
///
/// # Example
///
/// ```
/// # use polars_core::prelude::*;
/// fn example(df: &mut DataFrame) -> PolarsResult<&mut DataFrame> {
/// let original_name = "foo";
/// let new_name = "bar";
/// df.rename(original_name, new_name)
/// }
/// ```
pub fn rename(&mut self, column: &str, name: &str) -> PolarsResult<&mut Self> {
self.select_mut(column)
.ok_or_else(|| PolarsError::NotFound(column.to_string().into()))
.map(|s| s.rename(name))?;
let unique_names: AHashSet<&str, ahash::RandomState> =
AHashSet::from_iter(self.columns.iter().map(|s| s.name()));
if unique_names.len() != self.columns.len() {
return Err(PolarsError::SchemaMisMatch(
"duplicate column names found".into(),
));
}
Ok(self)
}
/// Sort `DataFrame` in place by a column.
pub fn sort_in_place(
&mut self,
by_column: impl IntoVec<String>,
reverse: impl IntoVec<bool>,
) -> PolarsResult<&mut Self> {
// a lot of indirection in both sorting and take
self.as_single_chunk_par();
let by_column = self.select_series(by_column)?;
let reverse = reverse.into_vec();
self.columns = self.sort_impl(by_column, reverse, false, None)?.columns;
Ok(self)
}
/// This is the dispatch of Self::sort, and exists to reduce compile bloat by monomorphization.
#[cfg(feature = "private")]
pub fn sort_impl(
&self,
by_column: Vec<Series>,
reverse: Vec<bool>,
nulls_last: bool,
slice: Option<(i64, usize)>,
) -> PolarsResult<Self> {
// note that the by_column argument also contains evaluated expression from polars-lazy
// that may not even be present in this dataframe.
// therefore when we try to set the first columns as sorted, we ignore the error
// as expressions are not present (they are renamed to _POLARS_SORT_COLUMN_i.
let first_reverse = reverse[0];
let first_by_column = by_column[0].name().to_string();
let mut take = match by_column.len() {
1 => {
let s = &by_column[0];
let options = SortOptions {
descending: reverse[0],
nulls_last,
};
// fast path for a frame with a single series
// no need to compute the sort indices and then take by these indices
// simply sort and return as frame
if self.width() == 1 && self.check_name_to_idx(s.name()).is_ok() {
let mut out = s.sort_with(options);
if let Some((offset, len)) = slice {
out = out.slice(offset, len);
}
return Ok(out.into_frame());
}
s.argsort(options)
}
_ => {
#[cfg(feature = "sort_multiple")]
{
let (first, by_column, reverse) = prepare_argsort(by_column, reverse)?;
first.argsort_multiple(&by_column, &reverse)?
}
#[cfg(not(feature = "sort_multiple"))]
{
panic!("activate `sort_multiple` feature gate to enable this functionality");
}
}
};
if let Some((offset, len)) = slice {
take = take.slice(offset, len);
}
// Safety:
// the created indices are in bounds
let mut df = if std::env::var("POLARS_VERT_PAR").is_ok() {
unsafe { self.take_unchecked_vectical(&take) }
} else {
unsafe { self.take_unchecked(&take) }
};
// Mark the first sort column as sorted
// if the column did not exists it is ok, because we sorted by an expression
// not present in the dataframe
let _ = df.apply(&first_by_column, |s| {
let mut s = s.clone();
if first_reverse {
s.set_sorted(IsSorted::Descending)
} else {
s.set_sorted(IsSorted::Ascending)
}
s
});
Ok(df)
}
/// Return a sorted clone of this `DataFrame`.
///
/// # Example
///
/// ```
/// # use polars_core::prelude::*;
/// fn sort_example(df: &DataFrame, reverse: bool) -> PolarsResult<DataFrame> {
/// df.sort(["a"], reverse)
/// }
///
/// fn sort_by_multiple_columns_example(df: &DataFrame) -> PolarsResult<DataFrame> {
/// df.sort(&["a", "b"], vec![false, true])
/// }
/// ```
pub fn sort(
&self,
by_column: impl IntoVec<String>,
reverse: impl IntoVec<bool>,
) -> PolarsResult<Self> {
let mut df = self.clone();
df.sort_in_place(by_column, reverse)?;
Ok(df)
}
/// Sort the `DataFrame` by a single column with extra options.
pub fn sort_with_options(&self, by_column: &str, options: SortOptions) -> PolarsResult<Self> {
let mut df = self.clone();
// a lot of indirection in both sorting and take
df.as_single_chunk_par();
let by_column = vec![df.column(by_column)?.clone()];
let reverse = vec![options.descending];
df.columns = df
.sort_impl(by_column, reverse, options.nulls_last, None)?
.columns;
Ok(df)
}
/// Replace a column with a `Series`.
///
/// # Example
///
/// ```rust
/// # use polars_core::prelude::*;
/// let mut df: DataFrame = df!("Country" => &["United States", "China"],
/// "Area (km²)" => &[9_833_520, 9_596_961])?;
/// let s: Series = Series::new("Country", &["USA", "PRC"]);
///
/// assert!(df.replace("Nation", s.clone()).is_err());
/// assert!(df.replace("Country", s).is_ok());
/// # Ok::<(), PolarsError>(())
/// ```
pub fn replace<S: IntoSeries>(&mut self, column: &str, new_col: S) -> PolarsResult<&mut Self> {
self.apply(column, |_| new_col.into_series())
}
/// Replace or update a column. The difference between this method and [DataFrame::with_column]
/// is that now the value of `column: &str` determines the name of the column and not the name
/// of the `Series` passed to this method.
pub fn replace_or_add<S: IntoSeries>(
&mut self,
column: &str,
new_col: S,
) -> PolarsResult<&mut Self> {
let mut new_col = new_col.into_series();
new_col.rename(column);
self.with_column(new_col)
}
/// Replace column at index `idx` with a `Series`.
///
/// # Example
///
/// ```ignored
/// # use polars_core::prelude::*;
/// let s0 = Series::new("foo", &["ham", "spam", "egg"]);
/// let s1 = Series::new("ascii", &[70, 79, 79]);
/// let mut df = DataFrame::new(vec![s0, s1])?;
///
/// // Add 32 to get lowercase ascii values
/// df.replace_at_idx(1, df.select_at_idx(1).unwrap() + 32);
/// # Ok::<(), PolarsError>(())
/// ```
pub fn replace_at_idx<S: IntoSeries>(
&mut self,
idx: usize,
new_col: S,
) -> PolarsResult<&mut Self> {
let mut new_column = new_col.into_series();
if new_column.len() != self.height() {
return Err(PolarsError::ShapeMisMatch(
format!("Cannot replace Series at index {}. The shape of Series {} does not match that of the DataFrame {}",
idx, new_column.len(), self.height()
).into()));
};
if idx >= self.width() {
return Err(PolarsError::ComputeError(
format!(
"Column index: {} outside of DataFrame with {} columns",
idx,
self.width()
)
.into(),
));
}
let old_col = &mut self.columns[idx];
mem::swap(old_col, &mut new_column);
Ok(self)
}
/// Apply a closure to a column. This is the recommended way to do in place modification.
///
/// # Example
///
/// ```rust
/// # use polars_core::prelude::*;
/// let s0 = Series::new("foo", &["ham", "spam", "egg"]);
/// let s1 = Series::new("names", &["Jean", "Claude", "van"]);
/// let mut df = DataFrame::new(vec![s0, s1])?;
///
/// fn str_to_len(str_val: &Series) -> Series {
/// str_val.utf8()
/// .unwrap()
/// .into_iter()
/// .map(|opt_name: Option<&str>| {
/// opt_name.map(|name: &str| name.len() as u32)
/// })
/// .collect::<UInt32Chunked>()
/// .into_series()
/// }
///
/// // Replace the names column by the length of the names.
/// df.apply("names", str_to_len);
/// # Ok::<(), PolarsError>(())
/// ```
/// Results in:
///
/// ```text
/// +--------+-------+
/// | foo | |
/// | --- | names |
/// | str | u32 |
/// +========+=======+
/// | "ham" | 4 |
/// +--------+-------+
/// | "spam" | 6 |
/// +--------+-------+
/// | "egg" | 3 |
/// +--------+-------+
/// ```
pub fn apply<F, S>(&mut self, name: &str, f: F) -> PolarsResult<&mut Self>
where
F: FnOnce(&Series) -> S,
S: IntoSeries,
{
let idx = self.check_name_to_idx(name)?;
self.apply_at_idx(idx, f)
}
/// Apply a closure to a column at index `idx`. This is the recommended way to do in place
/// modification.
///
/// # Example
///
/// ```rust
/// # use polars_core::prelude::*;
/// let s0 = Series::new("foo", &["ham", "spam", "egg"]);
/// let s1 = Series::new("ascii", &[70, 79, 79]);
/// let mut df = DataFrame::new(vec![s0, s1])?;
///
/// // Add 32 to get lowercase ascii values
/// df.apply_at_idx(1, |s| s + 32);
/// # Ok::<(), PolarsError>(())
/// ```
/// Results in:
///
/// ```text
/// +--------+-------+
/// | foo | ascii |
/// | --- | --- |
/// | str | i32 |
/// +========+=======+
/// | "ham" | 102 |
/// +--------+-------+
/// | "spam" | 111 |
/// +--------+-------+
/// | "egg" | 111 |
/// +--------+-------+
/// ```
pub fn apply_at_idx<F, S>(&mut self, idx: usize, f: F) -> PolarsResult<&mut Self>
where
F: FnOnce(&Series) -> S,
S: IntoSeries,
{
let df_height = self.height();
let width = self.width();
let col = self.columns.get_mut(idx).ok_or_else(|| {
PolarsError::ComputeError(
format!("Column index: {idx} outside of DataFrame with {width} columns",).into(),
)
})?;
let name = col.name().to_string();
let new_col = f(col).into_series();
match new_col.len() {
1 => {
let new_col = new_col.new_from_index(0, df_height);
let _ = mem::replace(col, new_col);
}
len if (len == df_height) => {
let _ = mem::replace(col, new_col);
}
len => {
return Err(PolarsError::ShapeMisMatch(
format!(
"Result Series has shape {} where the DataFrame has height {}",
len,
self.height()
)
.into(),
));
}
}
// make sure the name remains the same after applying the closure
unsafe {
let col = self.columns.get_unchecked_mut(idx);
col.rename(&name);
}
Ok(self)
}
/// Apply a closure that may fail to a column at index `idx`. This is the recommended way to do in place
/// modification.
///
/// # Example
///
/// This is the idiomatic way to replace some values a column of a `DataFrame` given range of indexes.
///
/// ```rust
/// # use polars_core::prelude::*;
/// let s0 = Series::new("foo", &["ham", "spam", "egg", "bacon", "quack"]);
/// let s1 = Series::new("values", &[1, 2, 3, 4, 5]);
/// let mut df = DataFrame::new(vec![s0, s1])?;
///
/// let idx = vec![0, 1, 4];
///
/// df.try_apply("foo", |s| {
/// s.utf8()?
/// .set_at_idx_with(idx, |opt_val| opt_val.map(|string| format!("{}-is-modified", string)))
/// });
/// # Ok::<(), PolarsError>(())
/// ```
/// Results in:
///
/// ```text
/// +---------------------+--------+
/// | foo | values |
/// | --- | --- |
/// | str | i32 |
/// +=====================+========+
/// | "ham-is-modified" | 1 |
/// +---------------------+--------+
/// | "spam-is-modified" | 2 |
/// +---------------------+--------+
/// | "egg" | 3 |
/// +---------------------+--------+
/// | "bacon" | 4 |
/// +---------------------+--------+
/// | "quack-is-modified" | 5 |
/// +---------------------+--------+
/// ```
pub fn try_apply_at_idx<F, S>(&mut self, idx: usize, f: F) -> PolarsResult<&mut Self>
where
F: FnOnce(&Series) -> PolarsResult<S>,
S: IntoSeries,
{
let width = self.width();
let col = self.columns.get_mut(idx).ok_or_else(|| {
PolarsError::ComputeError(
format!("Column index: {idx} outside of DataFrame with {width} columns",).into(),
)
})?;
let name = col.name().to_string();
let _ = mem::replace(col, f(col).map(|s| s.into_series())?);
// make sure the name remains the same after applying the closure
unsafe {
let col = self.columns.get_unchecked_mut(idx);
col.rename(&name);
}
Ok(self)
}src/frame/row.rs (line 101)
69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178
pub fn from_rows_iter_and_schema<'a, I>(mut rows: I, schema: &Schema) -> PolarsResult<Self>
where
I: Iterator<Item = &'a Row<'a>>,
{
let capacity = rows.size_hint().0;
let mut buffers: Vec<_> = schema
.iter_dtypes()
.map(|dtype| {
let buf: AnyValueBuffer = (dtype, capacity).into();
buf
})
.collect();
let mut expected_len = 0;
rows.try_for_each::<_, PolarsResult<()>>(|row| {
expected_len += 1;
for (value, buf) in row.0.iter().zip(&mut buffers) {
buf.add_fallible(value)?
}
Ok(())
})?;
let v = buffers
.into_iter()
.zip(schema.iter_names())
.map(|(b, name)| {
let mut s = b.into_series();
// if the schema adds a column not in the rows, we
// fill it with nulls
if s.is_empty() {
Series::full_null(name, expected_len, s.dtype())
} else {
s.rename(name);
s
}
})
.collect();
DataFrame::new(v)
}
/// Create a new DataFrame from rows. This should only be used when you have row wise data,
/// as this is a lot slower than creating the `Series` in a columnar fashion
#[cfg_attr(docsrs, doc(cfg(feature = "rows")))]
pub fn from_rows(rows: &[Row]) -> PolarsResult<Self> {
let schema = rows_to_schema_first_non_null(rows, Some(50));
let has_nulls = schema
.iter_dtypes()
.any(|dtype| matches!(dtype, DataType::Null));
if has_nulls {
return Err(PolarsError::ComputeError(
"Could not infer row types, because of the null values".into(),
));
}
Self::from_rows_and_schema(rows, &schema)
}
pub(crate) fn transpose_from_dtype(&self, dtype: &DataType) -> PolarsResult<DataFrame> {
let new_width = self.height();
let new_height = self.width();
match dtype {
#[cfg(feature = "dtype-i8")]
DataType::Int8 => numeric_transpose::<Int8Type>(&self.columns),
#[cfg(feature = "dtype-i16")]
DataType::Int16 => numeric_transpose::<Int16Type>(&self.columns),
DataType::Int32 => numeric_transpose::<Int32Type>(&self.columns),
DataType::Int64 => numeric_transpose::<Int64Type>(&self.columns),
#[cfg(feature = "dtype-u8")]
DataType::UInt8 => numeric_transpose::<UInt8Type>(&self.columns),
#[cfg(feature = "dtype-u16")]
DataType::UInt16 => numeric_transpose::<UInt16Type>(&self.columns),
DataType::UInt32 => numeric_transpose::<UInt32Type>(&self.columns),
DataType::UInt64 => numeric_transpose::<UInt64Type>(&self.columns),
DataType::Float32 => numeric_transpose::<Float32Type>(&self.columns),
DataType::Float64 => numeric_transpose::<Float64Type>(&self.columns),
_ => {
let mut buffers = (0..new_width)
.map(|_| {
let buf: AnyValueBuffer = (dtype, new_height).into();
buf
})
.collect::<Vec<_>>();
let columns = self
.columns
.iter()
.map(|s| s.cast(dtype).unwrap())
.collect::<Vec<_>>();
// this is very expensive. A lot of cache misses here.
// This is the part that is performance critical.
columns.iter().for_each(|s| {
s.iter().zip(buffers.iter_mut()).for_each(|(av, buf)| {
let _out = buf.add(av);
debug_assert!(_out.is_some());
});
});
let cols = buffers
.into_iter()
.enumerate()
.map(|(i, buf)| {
let mut s = buf.into_series();
s.rename(&format!("column_{i}"));
s
})
.collect::<Vec<_>>();
Ok(DataFrame::new_no_checks(cols))
}
}
}src/frame/hash_join/mod.rs (line 558)
515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581
pub fn _outer_join_from_series(
&self,
other: &DataFrame,
s_left: &Series,
s_right: &Series,
suffix: Option<String>,
slice: Option<(i64, usize)>,
) -> PolarsResult<DataFrame> {
#[cfg(feature = "dtype-categorical")]
_check_categorical_src(s_left.dtype(), s_right.dtype())?;
// store this so that we can keep original column order.
let join_column_index = self.iter().position(|s| s.name() == s_left.name()).unwrap();
// Get the indexes of the joined relations
let opt_join_tuples = s_left.hash_join_outer(s_right);
let mut opt_join_tuples = &*opt_join_tuples;
if let Some((offset, len)) = slice {
opt_join_tuples = slice_slice(opt_join_tuples, offset, len);
}
// Take the left and right dataframes by join tuples
let (mut df_left, df_right) = POOL.join(
|| unsafe {
self.drop(s_left.name()).unwrap().take_opt_iter_unchecked(
opt_join_tuples
.iter()
.map(|(left, _right)| left.map(|i| i as usize)),
)
},
|| unsafe {
other.drop(s_right.name()).unwrap().take_opt_iter_unchecked(
opt_join_tuples
.iter()
.map(|(_left, right)| right.map(|i| i as usize)),
)
},
);
let mut s = s_left
.to_physical_repr()
.zip_outer_join_column(&s_right.to_physical_repr(), opt_join_tuples);
s.rename(s_left.name());
let s = match s_left.dtype() {
#[cfg(feature = "dtype-categorical")]
DataType::Categorical(_) => {
let ca_left = s_left.categorical().unwrap();
let new_rev_map = ca_left.merge_categorical_map(s_right.categorical().unwrap())?;
let logical = s.u32().unwrap().clone();
// safety:
// categorical maps are merged
unsafe {
CategoricalChunked::from_cats_and_rev_map_unchecked(logical, new_rev_map)
.into_series()
}
}
dt @ DataType::Datetime(_, _)
| dt @ DataType::Time
| dt @ DataType::Date
| dt @ DataType::Duration(_) => s.cast(dt).unwrap(),
_ => s,
};
df_left.get_columns_mut().insert(join_column_index, s);
_finish_join(df_left, df_right, suffix.as_deref())
}Additional examples can be found in:
sourcepub fn shrink_to_fit(&mut self)
pub fn shrink_to_fit(&mut self)
Shrink the capacity of this array to fit its length.
Examples found in repository?
src/frame/mod.rs (line 417)
414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751 752 753 754 755 756 757 758 759 760 761 762 763 764 765 766 767 768 769 770 771 772 773 774 775 776 777 778 779 780 781 782 783 784 785 786 787 788 789 790 791 792 793 794 795 796 797 798 799 800 801 802 803 804 805 806 807 808 809 810 811 812 813 814 815 816 817 818 819 820 821 822 823 824 825 826 827 828 829 830 831 832 833 834 835 836 837 838 839 840 841 842 843 844 845 846 847 848 849 850 851 852 853 854 855 856 857 858 859 860 861 862 863 864 865 866 867 868 869 870 871 872 873 874 875 876 877 878 879 880 881 882 883 884 885 886 887 888 889 890 891 892 893 894 895 896 897 898 899 900 901 902 903 904 905 906 907 908 909 910 911 912 913 914 915 916 917 918 919 920 921 922 923 924 925 926 927 928 929 930 931 932 933 934 935 936 937 938 939 940 941 942 943 944 945 946 947 948 949 950 951 952 953 954 955 956 957 958 959 960 961 962 963 964 965 966 967 968 969 970 971 972 973 974 975 976 977 978 979 980 981 982 983 984 985 986 987 988 989 990 991 992 993 994 995 996 997 998 999 1000 1001 1002 1003 1004 1005 1006 1007 1008 1009 1010 1011 1012 1013 1014 1015 1016 1017 1018 1019 1020 1021 1022 1023 1024 1025 1026 1027 1028 1029 1030 1031 1032 1033 1034 1035 1036 1037 1038 1039 1040 1041 1042 1043 1044 1045 1046 1047 1048 1049 1050 1051 1052 1053 1054 1055 1056 1057 1058 1059 1060 1061 1062 1063 1064 1065 1066 1067 1068 1069 1070 1071 1072 1073 1074 1075 1076 1077 1078 1079 1080 1081 1082 1083 1084 1085 1086 1087 1088 1089 1090 1091 1092 1093 1094 1095 1096 1097 1098 1099 1100 1101 1102 1103 1104 1105 1106 1107 1108 1109 1110 1111 1112 1113 1114 1115 1116 1117 1118 1119 1120 1121 1122 1123 1124 1125 1126 1127 1128 1129 1130 1131 1132 1133 1134 1135 1136 1137 1138 1139 1140 1141 1142 1143 1144 1145 1146 1147 1148 1149 1150 1151 1152 1153 1154 1155 1156 1157 1158 1159 1160 1161 1162 1163 1164 1165 1166 1167 1168 1169 1170 1171 1172 1173 1174 1175 1176 1177 1178 1179 1180 1181 1182 1183 1184 1185 1186 1187 1188 1189 1190 1191 1192 1193 1194 1195 1196 1197 1198 1199 1200 1201 1202 1203 1204 1205 1206 1207 1208 1209 1210 1211 1212 1213 1214 1215 1216 1217 1218 1219 1220 1221 1222 1223 1224 1225 1226 1227 1228 1229 1230 1231 1232 1233 1234 1235 1236 1237 1238 1239 1240 1241 1242 1243 1244 1245 1246 1247 1248 1249 1250 1251 1252 1253 1254 1255 1256 1257 1258 1259 1260 1261 1262 1263 1264 1265 1266 1267 1268 1269 1270 1271 1272 1273 1274 1275 1276 1277 1278 1279 1280 1281 1282 1283 1284 1285 1286 1287 1288 1289 1290 1291 1292 1293 1294 1295 1296 1297 1298 1299 1300 1301 1302 1303 1304 1305 1306 1307 1308 1309 1310 1311 1312 1313 1314 1315 1316 1317 1318 1319 1320 1321 1322 1323 1324 1325 1326 1327 1328 1329 1330 1331 1332 1333 1334 1335 1336 1337 1338 1339 1340 1341 1342 1343 1344 1345 1346 1347 1348 1349 1350 1351 1352 1353 1354 1355 1356 1357 1358 1359 1360 1361 1362 1363 1364 1365 1366 1367 1368 1369 1370 1371 1372 1373 1374 1375 1376 1377 1378 1379 1380 1381 1382 1383 1384 1385 1386 1387 1388 1389 1390 1391 1392 1393 1394 1395 1396 1397 1398 1399 1400 1401 1402 1403 1404 1405 1406 1407 1408 1409 1410 1411 1412 1413 1414 1415 1416 1417 1418 1419 1420 1421 1422 1423 1424 1425 1426 1427 1428 1429 1430 1431 1432 1433 1434 1435 1436 1437 1438 1439 1440 1441 1442 1443 1444 1445 1446 1447 1448 1449 1450 1451 1452 1453 1454 1455 1456 1457 1458 1459 1460 1461 1462 1463 1464 1465 1466 1467 1468 1469 1470 1471 1472 1473 1474 1475 1476 1477 1478 1479 1480 1481 1482 1483 1484 1485 1486 1487 1488 1489 1490 1491 1492 1493 1494 1495 1496 1497 1498 1499 1500 1501 1502 1503 1504 1505 1506 1507 1508 1509 1510 1511 1512 1513 1514 1515 1516 1517 1518 1519 1520 1521 1522 1523 1524 1525 1526 1527 1528 1529 1530 1531 1532 1533 1534 1535 1536 1537 1538 1539 1540 1541 1542 1543 1544 1545 1546 1547 1548 1549 1550 1551 1552 1553 1554 1555 1556 1557 1558 1559 1560 1561 1562 1563 1564 1565 1566 1567 1568 1569 1570 1571 1572 1573 1574 1575 1576 1577 1578 1579 1580 1581 1582 1583 1584 1585 1586 1587 1588 1589 1590 1591 1592 1593 1594 1595 1596 1597 1598 1599 1600 1601 1602 1603 1604 1605 1606 1607 1608 1609 1610 1611 1612 1613 1614 1615 1616 1617 1618 1619 1620 1621 1622 1623 1624 1625 1626 1627 1628 1629 1630 1631 1632 1633 1634 1635 1636 1637 1638 1639 1640 1641 1642 1643 1644 1645 1646 1647 1648 1649 1650 1651 1652 1653 1654 1655 1656 1657 1658 1659 1660 1661 1662 1663 1664 1665 1666 1667 1668 1669 1670 1671 1672 1673 1674 1675 1676 1677 1678 1679 1680 1681 1682 1683 1684 1685 1686 1687 1688 1689 1690 1691 1692 1693 1694 1695 1696 1697 1698 1699 1700 1701 1702 1703 1704 1705 1706 1707 1708 1709 1710 1711 1712 1713 1714 1715 1716 1717 1718 1719 1720 1721 1722 1723 1724 1725 1726 1727 1728 1729 1730 1731 1732 1733 1734 1735 1736 1737 1738 1739 1740 1741 1742 1743 1744 1745 1746 1747 1748 1749 1750 1751 1752 1753 1754 1755 1756 1757 1758 1759 1760 1761 1762 1763 1764 1765 1766 1767 1768 1769 1770 1771 1772 1773 1774 1775 1776 1777 1778 1779 1780 1781 1782 1783 1784 1785 1786 1787 1788 1789 1790 1791 1792 1793 1794 1795 1796 1797 1798 1799 1800 1801 1802 1803 1804 1805 1806 1807 1808 1809 1810 1811 1812 1813 1814 1815 1816 1817 1818 1819 1820 1821 1822 1823 1824 1825 1826 1827 1828 1829 1830 1831 1832 1833 1834 1835 1836 1837 1838 1839 1840 1841 1842 1843 1844 1845 1846 1847 1848 1849 1850 1851 1852 1853 1854 1855 1856 1857 1858 1859 1860 1861 1862 1863 1864 1865 1866 1867 1868 1869 1870 1871 1872 1873 1874 1875 1876 1877 1878 1879 1880 1881 1882 1883 1884 1885 1886 1887 1888 1889 1890 1891 1892 1893 1894 1895 1896 1897 1898 1899 1900 1901 1902 1903 1904 1905 1906 1907 1908 1909 1910 1911 1912 1913 1914 1915 1916 1917 1918 1919 1920 1921 1922 1923 1924 1925 1926 1927 1928 1929 1930 1931 1932 1933 1934 1935 1936 1937 1938 1939 1940 1941 1942 1943 1944 1945 1946 1947 1948 1949 1950 1951 1952 1953 1954 1955 1956 1957 1958 1959 1960 1961 1962 1963 1964 1965 1966 1967 1968 1969 1970 1971 1972 1973 1974 1975 1976 1977 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024 2025 2026 2027 2028 2029 2030 2031 2032 2033 2034 2035 2036 2037 2038 2039 2040 2041 2042 2043 2044 2045 2046 2047 2048 2049 2050 2051 2052 2053 2054 2055 2056 2057 2058 2059 2060 2061 2062 2063 2064 2065 2066 2067 2068 2069 2070 2071 2072 2073 2074 2075 2076 2077 2078 2079 2080 2081 2082 2083 2084 2085 2086 2087 2088 2089 2090 2091 2092 2093 2094 2095 2096 2097 2098 2099 2100 2101 2102 2103 2104 2105 2106 2107 2108 2109 2110 2111 2112 2113 2114 2115 2116 2117 2118 2119 2120 2121 2122 2123 2124 2125 2126 2127 2128 2129 2130 2131 2132 2133 2134 2135 2136 2137 2138 2139 2140 2141 2142 2143 2144 2145 2146 2147 2148 2149 2150 2151 2152 2153 2154 2155 2156 2157 2158 2159 2160 2161 2162 2163 2164 2165 2166 2167 2168 2169 2170 2171 2172 2173 2174 2175 2176 2177 2178 2179 2180 2181 2182 2183 2184 2185 2186 2187 2188 2189 2190 2191 2192 2193 2194 2195 2196 2197 2198 2199 2200 2201 2202 2203 2204 2205 2206 2207 2208 2209 2210 2211 2212 2213 2214 2215 2216 2217 2218 2219 2220 2221 2222 2223 2224 2225 2226 2227 2228 2229 2230 2231 2232 2233 2234 2235 2236 2237 2238 2239 2240 2241 2242 2243 2244 2245 2246 2247 2248 2249 2250 2251 2252 2253 2254 2255 2256 2257 2258 2259 2260 2261 2262 2263 2264 2265 2266 2267 2268 2269 2270 2271 2272 2273 2274 2275 2276 2277 2278 2279 2280 2281 2282 2283 2284 2285 2286 2287 2288 2289 2290 2291 2292 2293 2294 2295 2296 2297 2298 2299 2300 2301 2302 2303 2304 2305 2306
pub fn shrink_to_fit(&mut self) {
// Don't parallelize this. Memory overhead
for s in &mut self.columns {
s.shrink_to_fit();
}
}
/// Aggregate all the chunks in the DataFrame to a single chunk.
pub fn as_single_chunk(&mut self) -> &mut Self {
// Don't parallelize this. Memory overhead
for s in &mut self.columns {
*s = s.rechunk();
}
self
}
/// Aggregate all the chunks in the DataFrame to a single chunk in parallel.
/// This may lead to more peak memory consumption.
pub fn as_single_chunk_par(&mut self) -> &mut Self {
if self.columns.iter().any(|s| s.n_chunks() > 1) {
self.columns = self.apply_columns_par(&|s| s.rechunk());
}
self
}
/// Estimates of the DataFrames columns consist of the same chunk sizes
pub fn should_rechunk(&self) -> bool {
let hb = RandomState::default();
let hb2 = RandomState::with_seeds(392498, 98132457, 0, 412059);
!self
.columns
.iter()
// The idea is that we create a hash of the chunk lengths.
// Consisting of the combined hash + the sum (assuming collision probability is nihil)
// if not, we can add more hashes or at worst case we do an extra rechunk.
// the old solution to this was clone all lengths to a vec and compare the vecs
.map(|s| {
s.chunk_lengths().map(|i| i as u64).fold(
(0u64, 0u64, s.n_chunks()),
|(lhash, lh2, n), rval| {
let mut h = hb.build_hasher();
rval.hash(&mut h);
let rhash = h.finish();
let mut h = hb2.build_hasher();
rval.hash(&mut h);
let rh2 = h.finish();
(
_boost_hash_combine(lhash, rhash),
_boost_hash_combine(lh2, rh2),
n,
)
},
)
})
.all_equal()
}
/// Ensure all the chunks in the DataFrame are aligned.
pub fn rechunk(&mut self) -> &mut Self {
if self.should_rechunk() {
self.as_single_chunk_par()
} else {
self
}
}
/// Get the `DataFrame` schema.
///
/// # Example
///
/// ```rust
/// # use polars_core::prelude::*;
/// let df: DataFrame = df!("Thing" => &["Observable universe", "Human stupidity"],
/// "Diameter (m)" => &[8.8e26, f64::INFINITY])?;
///
/// let f1: Field = Field::new("Thing", DataType::Utf8);
/// let f2: Field = Field::new("Diameter (m)", DataType::Float64);
/// let sc: Schema = Schema::from(vec![f1, f2].into_iter());
///
/// assert_eq!(df.schema(), sc);
/// # Ok::<(), PolarsError>(())
/// ```
pub fn schema(&self) -> Schema {
Schema::from(self.iter().map(|s| s.field().into_owned()))
}
/// Get a reference to the `DataFrame` columns.
///
/// # Example
///
/// ```rust
/// # use polars_core::prelude::*;
/// let df: DataFrame = df!("Name" => &["Adenine", "Cytosine", "Guanine", "Thymine"],
/// "Symbol" => &["A", "C", "G", "T"])?;
/// let columns: &Vec<Series> = df.get_columns();
///
/// assert_eq!(columns[0].name(), "Name");
/// assert_eq!(columns[1].name(), "Symbol");
/// # Ok::<(), PolarsError>(())
/// ```
#[inline]
pub fn get_columns(&self) -> &Vec<Series> {
&self.columns
}
#[cfg(feature = "private")]
#[inline]
pub fn get_columns_mut(&mut self) -> &mut Vec<Series> {
&mut self.columns
}
/// Iterator over the columns as `Series`.
///
/// # Example
///
/// ```rust
/// # use polars_core::prelude::*;
/// let s1: Series = Series::new("Name", &["Pythagoras' theorem", "Shannon entropy"]);
/// let s2: Series = Series::new("Formula", &["a²+b²=c²", "H=-Σ[P(x)log|P(x)|]"]);
/// let df: DataFrame = DataFrame::new(vec![s1.clone(), s2.clone()])?;
///
/// let mut iterator = df.iter();
///
/// assert_eq!(iterator.next(), Some(&s1));
/// assert_eq!(iterator.next(), Some(&s2));
/// assert_eq!(iterator.next(), None);
/// # Ok::<(), PolarsError>(())
/// ```
pub fn iter(&self) -> std::slice::Iter<'_, Series> {
self.columns.iter()
}
/// # Example
///
/// ```rust
/// # use polars_core::prelude::*;
/// let df: DataFrame = df!("Language" => &["Rust", "Python"],
/// "Designer" => &["Graydon Hoare", "Guido van Rossum"])?;
///
/// assert_eq!(df.get_column_names(), &["Language", "Designer"]);
/// # Ok::<(), PolarsError>(())
/// ```
pub fn get_column_names(&self) -> Vec<&str> {
self.columns.iter().map(|s| s.name()).collect()
}
/// Get the `Vec<String>` representing the column names.
pub fn get_column_names_owned(&self) -> Vec<String> {
self.columns.iter().map(|s| s.name().to_string()).collect()
}
/// Set the column names.
/// # Example
///
/// ```rust
/// # use polars_core::prelude::*;
/// let mut df: DataFrame = df!("Mathematical set" => &["ℕ", "ℤ", "𝔻", "ℚ", "ℝ", "ℂ"])?;
/// df.set_column_names(&["Set"])?;
///
/// assert_eq!(df.get_column_names(), &["Set"]);
/// # Ok::<(), PolarsError>(())
/// ```
pub fn set_column_names<S: AsRef<str>>(&mut self, names: &[S]) -> PolarsResult<()> {
if names.len() != self.columns.len() {
return Err(PolarsError::ShapeMisMatch("the provided slice with column names has not the same size as the DataFrame's width".into()));
}
let unique_names: AHashSet<&str, ahash::RandomState> =
AHashSet::from_iter(names.iter().map(|name| name.as_ref()));
if unique_names.len() != self.columns.len() {
return Err(PolarsError::SchemaMisMatch(
"duplicate column names found".into(),
));
}
let columns = mem::take(&mut self.columns);
self.columns = columns
.into_iter()
.zip(names)
.map(|(s, name)| {
let mut s = s;
s.rename(name.as_ref());
s
})
.collect();
Ok(())
}
/// Get the data types of the columns in the DataFrame.
///
/// # Example
///
/// ```rust
/// # use polars_core::prelude::*;
/// let venus_air: DataFrame = df!("Element" => &["Carbon dioxide", "Nitrogen"],
/// "Fraction" => &[0.965, 0.035])?;
///
/// assert_eq!(venus_air.dtypes(), &[DataType::Utf8, DataType::Float64]);
/// # Ok::<(), PolarsError>(())
/// ```
pub fn dtypes(&self) -> Vec<DataType> {
self.columns.iter().map(|s| s.dtype().clone()).collect()
}
/// The number of chunks per column
pub fn n_chunks(&self) -> usize {
match self.columns.get(0) {
None => 0,
Some(s) => s.n_chunks(),
}
}
/// Get a reference to the schema fields of the `DataFrame`.
///
/// # Example
///
/// ```rust
/// # use polars_core::prelude::*;
/// let earth: DataFrame = df!("Surface type" => &["Water", "Land"],
/// "Fraction" => &[0.708, 0.292])?;
///
/// let f1: Field = Field::new("Surface type", DataType::Utf8);
/// let f2: Field = Field::new("Fraction", DataType::Float64);
///
/// assert_eq!(earth.fields(), &[f1, f2]);
/// # Ok::<(), PolarsError>(())
/// ```
pub fn fields(&self) -> Vec<Field> {
self.columns
.iter()
.map(|s| s.field().into_owned())
.collect()
}
/// Get (height, width) of the `DataFrame`.
///
/// # Example
///
/// ```rust
/// # use polars_core::prelude::*;
/// let df0: DataFrame = DataFrame::default();
/// let df1: DataFrame = df!("1" => &[1, 2, 3, 4, 5])?;
/// let df2: DataFrame = df!("1" => &[1, 2, 3, 4, 5],
/// "2" => &[1, 2, 3, 4, 5])?;
///
/// assert_eq!(df0.shape(), (0 ,0));
/// assert_eq!(df1.shape(), (5, 1));
/// assert_eq!(df2.shape(), (5, 2));
/// # Ok::<(), PolarsError>(())
/// ```
pub fn shape(&self) -> (usize, usize) {
match self.columns.as_slice() {
&[] => (0, 0),
v => (v[0].len(), v.len()),
}
}
/// Get the width of the `DataFrame` which is the number of columns.
///
/// # Example
///
/// ```rust
/// # use polars_core::prelude::*;
/// let df0: DataFrame = DataFrame::default();
/// let df1: DataFrame = df!("Series 1" => &[0; 0])?;
/// let df2: DataFrame = df!("Series 1" => &[0; 0],
/// "Series 2" => &[0; 0])?;
///
/// assert_eq!(df0.width(), 0);
/// assert_eq!(df1.width(), 1);
/// assert_eq!(df2.width(), 2);
/// # Ok::<(), PolarsError>(())
/// ```
pub fn width(&self) -> usize {
self.columns.len()
}
/// Get the height of the `DataFrame` which is the number of rows.
///
/// # Example
///
/// ```rust
/// # use polars_core::prelude::*;
/// let df0: DataFrame = DataFrame::default();
/// let df1: DataFrame = df!("Currency" => &["€", "$"])?;
/// let df2: DataFrame = df!("Currency" => &["€", "$", "¥", "£", "₿"])?;
///
/// assert_eq!(df0.height(), 0);
/// assert_eq!(df1.height(), 2);
/// assert_eq!(df2.height(), 5);
/// # Ok::<(), PolarsError>(())
/// ```
pub fn height(&self) -> usize {
self.shape().0
}
/// Check if the `DataFrame` is empty.
///
/// # Example
///
/// ```rust
/// # use polars_core::prelude::*;
/// let df1: DataFrame = DataFrame::default();
/// assert!(df1.is_empty());
///
/// let df2: DataFrame = df!("First name" => &["Forever"],
/// "Last name" => &["Alone"])?;
/// assert!(!df2.is_empty());
/// # Ok::<(), PolarsError>(())
/// ```
pub fn is_empty(&self) -> bool {
self.columns.is_empty()
}
pub(crate) fn hstack_mut_no_checks(&mut self, columns: &[Series]) -> &mut Self {
for col in columns {
self.columns.push(col.clone());
}
self
}
/// Add multiple `Series` to a `DataFrame`.
/// The added `Series` are required to have the same length.
///
/// # Example
///
/// ```rust
/// # use polars_core::prelude::*;
/// fn stack(df: &mut DataFrame, columns: &[Series]) {
/// df.hstack_mut(columns);
/// }
/// ```
pub fn hstack_mut(&mut self, columns: &[Series]) -> PolarsResult<&mut Self> {
let mut names = PlHashSet::with_capacity(self.columns.len());
for s in &self.columns {
names.insert(s.name());
}
let height = self.height();
// first loop check validity. We don't do this in a single pass otherwise
// this DataFrame is already modified when an error occurs.
for col in columns {
if col.len() != height && height != 0 {
return Err(PolarsError::ShapeMisMatch(
format!("Could not horizontally stack Series. The Series length {} differs from the DataFrame height: {height}", col.len()).into()));
}
let name = col.name();
if names.contains(name) {
return Err(PolarsError::Duplicate(
format!("Cannot do hstack operation. Column with name: {name} already exists",)
.into(),
));
}
names.insert(name);
}
drop(names);
Ok(self.hstack_mut_no_checks(columns))
}
/// Add multiple `Series` to a `DataFrame`.
/// The added `Series` are required to have the same length.
///
/// # Example
///
/// ```rust
/// # use polars_core::prelude::*;
/// let df1: DataFrame = df!("Element" => &["Copper", "Silver", "Gold"])?;
/// let s1: Series = Series::new("Proton", &[29, 47, 79]);
/// let s2: Series = Series::new("Electron", &[29, 47, 79]);
///
/// let df2: DataFrame = df1.hstack(&[s1, s2])?;
/// assert_eq!(df2.shape(), (3, 3));
/// println!("{}", df2);
/// # Ok::<(), PolarsError>(())
/// ```
///
/// Output:
///
/// ```text
/// shape: (3, 3)
/// +---------+--------+----------+
/// | Element | Proton | Electron |
/// | --- | --- | --- |
/// | str | i32 | i32 |
/// +=========+========+==========+
/// | Copper | 29 | 29 |
/// +---------+--------+----------+
/// | Silver | 47 | 47 |
/// +---------+--------+----------+
/// | Gold | 79 | 79 |
/// +---------+--------+----------+
/// ```
pub fn hstack(&self, columns: &[Series]) -> PolarsResult<Self> {
let mut new_cols = self.columns.clone();
new_cols.extend_from_slice(columns);
DataFrame::new(new_cols)
}
/// Concatenate a `DataFrame` to this `DataFrame` and return as newly allocated `DataFrame`.
///
/// If many `vstack` operations are done, it is recommended to call [`DataFrame::rechunk`].
///
/// # Example
///
/// ```rust
/// # use polars_core::prelude::*;
/// let df1: DataFrame = df!("Element" => &["Copper", "Silver", "Gold"],
/// "Melting Point (K)" => &[1357.77, 1234.93, 1337.33])?;
/// let df2: DataFrame = df!("Element" => &["Platinum", "Palladium"],
/// "Melting Point (K)" => &[2041.4, 1828.05])?;
///
/// let df3: DataFrame = df1.vstack(&df2)?;
///
/// assert_eq!(df3.shape(), (5, 2));
/// println!("{}", df3);
/// # Ok::<(), PolarsError>(())
/// ```
///
/// Output:
///
/// ```text
/// shape: (5, 2)
/// +-----------+-------------------+
/// | Element | Melting Point (K) |
/// | --- | --- |
/// | str | f64 |
/// +===========+===================+
/// | Copper | 1357.77 |
/// +-----------+-------------------+
/// | Silver | 1234.93 |
/// +-----------+-------------------+
/// | Gold | 1337.33 |
/// +-----------+-------------------+
/// | Platinum | 2041.4 |
/// +-----------+-------------------+
/// | Palladium | 1828.05 |
/// +-----------+-------------------+
/// ```
pub fn vstack(&self, other: &DataFrame) -> PolarsResult<Self> {
let mut df = self.clone();
df.vstack_mut(other)?;
Ok(df)
}
/// Concatenate a DataFrame to this DataFrame
///
/// If many `vstack` operations are done, it is recommended to call [`DataFrame::rechunk`].
///
/// # Example
///
/// ```rust
/// # use polars_core::prelude::*;
/// let mut df1: DataFrame = df!("Element" => &["Copper", "Silver", "Gold"],
/// "Melting Point (K)" => &[1357.77, 1234.93, 1337.33])?;
/// let df2: DataFrame = df!("Element" => &["Platinum", "Palladium"],
/// "Melting Point (K)" => &[2041.4, 1828.05])?;
///
/// df1.vstack_mut(&df2)?;
///
/// assert_eq!(df1.shape(), (5, 2));
/// println!("{}", df1);
/// # Ok::<(), PolarsError>(())
/// ```
///
/// Output:
///
/// ```text
/// shape: (5, 2)
/// +-----------+-------------------+
/// | Element | Melting Point (K) |
/// | --- | --- |
/// | str | f64 |
/// +===========+===================+
/// | Copper | 1357.77 |
/// +-----------+-------------------+
/// | Silver | 1234.93 |
/// +-----------+-------------------+
/// | Gold | 1337.33 |
/// +-----------+-------------------+
/// | Platinum | 2041.4 |
/// +-----------+-------------------+
/// | Palladium | 1828.05 |
/// +-----------+-------------------+
/// ```
pub fn vstack_mut(&mut self, other: &DataFrame) -> PolarsResult<&mut Self> {
if self.width() != other.width() {
if self.width() == 0 {
self.columns = other.columns.clone();
return Ok(self);
}
return Err(PolarsError::ShapeMisMatch(
format!("Could not vertically stack DataFrame. The DataFrames appended width {} differs from the parent DataFrames width {}", self.width(), other.width()).into()
));
}
self.columns
.iter_mut()
.zip(other.columns.iter())
.try_for_each::<_, PolarsResult<_>>(|(left, right)| {
can_extend(left, right)?;
left.append(right).expect("should not fail");
Ok(())
})?;
Ok(self)
}
/// Does not check if schema is correct
pub(crate) fn vstack_mut_unchecked(&mut self, other: &DataFrame) {
self.columns
.iter_mut()
.zip(other.columns.iter())
.for_each(|(left, right)| {
left.append(right).expect("should not fail");
});
}
/// Extend the memory backed by this [`DataFrame`] with the values from `other`.
///
/// Different from [`vstack`](Self::vstack) which adds the chunks from `other` to the chunks of this [`DataFrame`]
/// `extend` appends the data from `other` to the underlying memory locations and thus may cause a reallocation.
///
/// If this does not cause a reallocation, the resulting data structure will not have any extra chunks
/// and thus will yield faster queries.
///
/// Prefer `extend` over `vstack` when you want to do a query after a single append. For instance during
/// online operations where you add `n` rows and rerun a query.
///
/// Prefer `vstack` over `extend` when you want to append many times before doing a query. For instance
/// when you read in multiple files and when to store them in a single `DataFrame`. In the latter case, finish the sequence
/// of `append` operations with a [`rechunk`](Self::rechunk).
pub fn extend(&mut self, other: &DataFrame) -> PolarsResult<()> {
if self.width() != other.width() {
return Err(PolarsError::ShapeMisMatch(
format!("Could not extend DataFrame. The DataFrames extended width {} differs from the parent DataFrames width {}", self.width(), other.width()).into()
));
}
self.columns
.iter_mut()
.zip(other.columns.iter())
.try_for_each::<_, PolarsResult<_>>(|(left, right)| {
can_extend(left, right)?;
left.extend(right).unwrap();
Ok(())
})?;
Ok(())
}
/// Remove a column by name and return the column removed.
///
/// # Example
///
/// ```rust
/// # use polars_core::prelude::*;
/// let mut df: DataFrame = df!("Animal" => &["Tiger", "Lion", "Great auk"],
/// "IUCN" => &["Endangered", "Vulnerable", "Extinct"])?;
///
/// let s1: PolarsResult<Series> = df.drop_in_place("Average weight");
/// assert!(s1.is_err());
///
/// let s2: Series = df.drop_in_place("Animal")?;
/// assert_eq!(s2, Series::new("Animal", &["Tiger", "Lion", "Great auk"]));
/// # Ok::<(), PolarsError>(())
/// ```
pub fn drop_in_place(&mut self, name: &str) -> PolarsResult<Series> {
let idx = self.check_name_to_idx(name)?;
Ok(self.columns.remove(idx))
}
/// Return a new `DataFrame` where all null values are dropped.
///
/// # Example
///
/// ```no_run
/// # use polars_core::prelude::*;
/// let df1: DataFrame = df!("Country" => ["Malta", "Liechtenstein", "North Korea"],
/// "Tax revenue (% GDP)" => [Some(32.7), None, None])?;
/// assert_eq!(df1.shape(), (3, 2));
///
/// let df2: DataFrame = df1.drop_nulls(None)?;
/// assert_eq!(df2.shape(), (1, 2));
/// println!("{}", df2);
/// # Ok::<(), PolarsError>(())
/// ```
///
/// Output:
///
/// ```text
/// shape: (1, 2)
/// +---------+---------------------+
/// | Country | Tax revenue (% GDP) |
/// | --- | --- |
/// | str | f64 |
/// +=========+=====================+
/// | Malta | 32.7 |
/// +---------+---------------------+
/// ```
pub fn drop_nulls(&self, subset: Option<&[String]>) -> PolarsResult<Self> {
let selected_series;
let mut iter = match subset {
Some(cols) => {
selected_series = self.select_series(cols)?;
selected_series.iter()
}
None => self.columns.iter(),
};
// fast path for no nulls in df
if iter.clone().all(|s| !s.has_validity()) {
return Ok(self.clone());
}
let mask = iter
.next()
.ok_or_else(|| PolarsError::NoData("No data to drop nulls from".into()))?;
let mut mask = mask.is_not_null();
for s in iter {
mask = mask & s.is_not_null();
}
self.filter(&mask)
}
/// Drop a column by name.
/// This is a pure method and will return a new `DataFrame` instead of modifying
/// the current one in place.
///
/// # Example
///
/// ```rust
/// # use polars_core::prelude::*;
/// let df1: DataFrame = df!("Ray type" => &["α", "β", "X", "γ"])?;
/// let df2: DataFrame = df1.drop("Ray type")?;
///
/// assert!(df2.is_empty());
/// # Ok::<(), PolarsError>(())
/// ```
pub fn drop(&self, name: &str) -> PolarsResult<Self> {
let idx = self.check_name_to_idx(name)?;
let mut new_cols = Vec::with_capacity(self.columns.len() - 1);
self.columns.iter().enumerate().for_each(|(i, s)| {
if i != idx {
new_cols.push(s.clone())
}
});
Ok(DataFrame::new_no_checks(new_cols))
}
pub fn drop_many<S: AsRef<str>>(&self, names: &[S]) -> Self {
let names = names.iter().map(|s| s.as_ref()).collect();
fn inner(df: &DataFrame, names: Vec<&str>) -> DataFrame {
let mut new_cols = Vec::with_capacity(df.columns.len() - names.len());
df.columns.iter().for_each(|s| {
if !names.contains(&s.name()) {
new_cols.push(s.clone())
}
});
DataFrame::new_no_checks(new_cols)
}
inner(self, names)
}
fn insert_at_idx_no_name_check(
&mut self,
index: usize,
series: Series,
) -> PolarsResult<&mut Self> {
if series.len() == self.height() {
self.columns.insert(index, series);
Ok(self)
} else {
Err(PolarsError::ShapeMisMatch(
format!(
"Could not add column. The Series length {} differs from the DataFrame height: {}",
series.len(),
self.height()
)
.into(),
))
}
}
/// Insert a new column at a given index.
pub fn insert_at_idx<S: IntoSeries>(
&mut self,
index: usize,
column: S,
) -> PolarsResult<&mut Self> {
let series = column.into_series();
self.check_already_present(series.name())?;
self.insert_at_idx_no_name_check(index, series)
}
fn add_column_by_search(&mut self, series: Series) -> PolarsResult<()> {
if let Some(idx) = self.find_idx_by_name(series.name()) {
self.replace_at_idx(idx, series)?;
} else {
self.columns.push(series);
}
Ok(())
}
/// Add a new column to this `DataFrame` or replace an existing one.
pub fn with_column<S: IntoSeries>(&mut self, column: S) -> PolarsResult<&mut Self> {
fn inner(df: &mut DataFrame, mut series: Series) -> PolarsResult<&mut DataFrame> {
let height = df.height();
if series.len() == 1 && height > 1 {
series = series.new_from_index(0, height);
}
if series.len() == height || df.is_empty() {
df.add_column_by_search(series)?;
Ok(df)
}
// special case for literals
else if height == 0 && series.len() == 1 {
let s = series.slice(0, 0);
df.add_column_by_search(s)?;
Ok(df)
} else {
Err(PolarsError::ShapeMisMatch(
format!(
"Could not add column. The Series length {} differs from the DataFrame height: {}",
series.len(),
df.height()
)
.into(),
))
}
}
let series = column.into_series();
inner(self, series)
}
fn add_column_by_schema(&mut self, s: Series, schema: &Schema) -> PolarsResult<()> {
let name = s.name();
if let Some((idx, _, _)) = schema.get_full(name) {
// schema is incorrect fallback to search
if self.columns.get(idx).map(|s| s.name()) != Some(name) {
self.add_column_by_search(s)?;
} else {
self.replace_at_idx(idx, s)?;
}
} else {
self.columns.push(s);
}
Ok(())
}
pub fn _add_columns(&mut self, columns: Vec<Series>, schema: &Schema) -> PolarsResult<()> {
for (i, s) in columns.into_iter().enumerate() {
// we need to branch here
// because users can add multiple columns with the same name
if i == 0 || schema.get(s.name()).is_some() {
self.with_column_and_schema(s, schema)?;
} else {
self.with_column(s.clone())?;
}
}
Ok(())
}
/// Add a new column to this `DataFrame` or replace an existing one.
/// Uses an existing schema to amortize lookups.
/// If the schema is incorrect, we will fallback to linear search.
pub fn with_column_and_schema<S: IntoSeries>(
&mut self,
column: S,
schema: &Schema,
) -> PolarsResult<&mut Self> {
let mut series = column.into_series();
let height = self.height();
if series.len() == 1 && height > 1 {
series = series.new_from_index(0, height);
}
if series.len() == height || self.is_empty() {
self.add_column_by_schema(series, schema)?;
Ok(self)
}
// special case for literals
else if height == 0 && series.len() == 1 {
let s = series.slice(0, 0);
self.add_column_by_schema(s, schema)?;
Ok(self)
} else {
Err(PolarsError::ShapeMisMatch(
format!(
"Could not add column. The Series length {} differs from the DataFrame height: {}",
series.len(),
self.height()
)
.into(),
))
}
}
/// Get a row in the `DataFrame`. Beware this is slow.
///
/// # Example
///
/// ```
/// # use polars_core::prelude::*;
/// fn example(df: &mut DataFrame, idx: usize) -> Option<Vec<AnyValue>> {
/// df.get(idx)
/// }
/// ```
pub fn get(&self, idx: usize) -> Option<Vec<AnyValue>> {
match self.columns.get(0) {
Some(s) => {
if s.len() <= idx {
return None;
}
}
None => return None,
}
// safety: we just checked bounds
unsafe { Some(self.columns.iter().map(|s| s.get_unchecked(idx)).collect()) }
}
/// Select a `Series` by index.
///
/// # Example
///
/// ```rust
/// # use polars_core::prelude::*;
/// let df: DataFrame = df!("Star" => &["Sun", "Betelgeuse", "Sirius A", "Sirius B"],
/// "Absolute magnitude" => &[4.83, -5.85, 1.42, 11.18])?;
///
/// let s1: Option<&Series> = df.select_at_idx(0);
/// let s2: Series = Series::new("Star", &["Sun", "Betelgeuse", "Sirius A", "Sirius B"]);
///
/// assert_eq!(s1, Some(&s2));
/// # Ok::<(), PolarsError>(())
/// ```
pub fn select_at_idx(&self, idx: usize) -> Option<&Series> {
self.columns.get(idx)
}
/// Select a mutable series by index.
///
/// *Note: the length of the Series should remain the same otherwise the DataFrame is invalid.*
/// For this reason the method is not public
fn select_at_idx_mut(&mut self, idx: usize) -> Option<&mut Series> {
self.columns.get_mut(idx)
}
/// Select column(s) from this `DataFrame` by range and return a new DataFrame
///
/// # Examples
///
/// ```rust
/// # use polars_core::prelude::*;
/// let df = df! {
/// "0" => &[0, 0, 0],
/// "1" => &[1, 1, 1],
/// "2" => &[2, 2, 2]
/// }?;
///
/// assert!(df.select(&["0", "1"])?.frame_equal(&df.select_by_range(0..=1)?));
/// assert!(df.frame_equal(&df.select_by_range(..)?));
/// # Ok::<(), PolarsError>(())
/// ```
pub fn select_by_range<R>(&self, range: R) -> PolarsResult<Self>
where
R: ops::RangeBounds<usize>,
{
// This function is copied from std::slice::range (https://doc.rust-lang.org/std/slice/fn.range.html)
// because it is the nightly feature. We should change here if this function were stable.
fn get_range<R>(range: R, bounds: ops::RangeTo<usize>) -> ops::Range<usize>
where
R: ops::RangeBounds<usize>,
{
let len = bounds.end;
let start: ops::Bound<&usize> = range.start_bound();
let start = match start {
ops::Bound::Included(&start) => start,
ops::Bound::Excluded(start) => start.checked_add(1).unwrap_or_else(|| {
panic!("attempted to index slice from after maximum usize");
}),
ops::Bound::Unbounded => 0,
};
let end: ops::Bound<&usize> = range.end_bound();
let end = match end {
ops::Bound::Included(end) => end.checked_add(1).unwrap_or_else(|| {
panic!("attempted to index slice up to maximum usize");
}),
ops::Bound::Excluded(&end) => end,
ops::Bound::Unbounded => len,
};
if start > end {
panic!("slice index starts at {start} but ends at {end}");
}
if end > len {
panic!("range end index {end} out of range for slice of length {len}",);
}
ops::Range { start, end }
}
let colnames = self.get_column_names_owned();
let range = get_range(range, ..colnames.len());
self.select_impl(&colnames[range])
}
/// Get column index of a `Series` by name.
/// # Example
///
/// ```rust
/// # use polars_core::prelude::*;
/// let df: DataFrame = df!("Name" => &["Player 1", "Player 2", "Player 3"],
/// "Health" => &[100, 200, 500],
/// "Mana" => &[250, 100, 0],
/// "Strength" => &[30, 150, 300])?;
///
/// assert_eq!(df.find_idx_by_name("Name"), Some(0));
/// assert_eq!(df.find_idx_by_name("Health"), Some(1));
/// assert_eq!(df.find_idx_by_name("Mana"), Some(2));
/// assert_eq!(df.find_idx_by_name("Strength"), Some(3));
/// assert_eq!(df.find_idx_by_name("Haste"), None);
/// # Ok::<(), PolarsError>(())
/// ```
pub fn find_idx_by_name(&self, name: &str) -> Option<usize> {
self.columns.iter().position(|s| s.name() == name)
}
/// Select a single column by name.
///
/// # Example
///
/// ```rust
/// # use polars_core::prelude::*;
/// let s1: Series = Series::new("Password", &["123456", "[]B$u$g$s$B#u#n#n#y[]{}"]);
/// let s2: Series = Series::new("Robustness", &["Weak", "Strong"]);
/// let df: DataFrame = DataFrame::new(vec![s1.clone(), s2])?;
///
/// assert_eq!(df.column("Password")?, &s1);
/// # Ok::<(), PolarsError>(())
/// ```
pub fn column(&self, name: &str) -> PolarsResult<&Series> {
let idx = self
.find_idx_by_name(name)
.ok_or_else(|| PolarsError::NotFound(name.to_string().into()))?;
Ok(self.select_at_idx(idx).unwrap())
}
/// Selected multiple columns by name.
///
/// # Example
///
/// ```rust
/// # use polars_core::prelude::*;
/// let df: DataFrame = df!("Latin name" => &["Oncorhynchus kisutch", "Salmo salar"],
/// "Max weight (kg)" => &[16.0, 35.89])?;
/// let sv: Vec<&Series> = df.columns(&["Latin name", "Max weight (kg)"])?;
///
/// assert_eq!(&df[0], sv[0]);
/// assert_eq!(&df[1], sv[1]);
/// # Ok::<(), PolarsError>(())
/// ```
pub fn columns<I, S>(&self, names: I) -> PolarsResult<Vec<&Series>>
where
I: IntoIterator<Item = S>,
S: AsRef<str>,
{
names
.into_iter()
.map(|name| self.column(name.as_ref()))
.collect()
}
/// Select column(s) from this `DataFrame` and return a new `DataFrame`.
///
/// # Examples
///
/// ```
/// # use polars_core::prelude::*;
/// fn example(df: &DataFrame) -> PolarsResult<DataFrame> {
/// df.select(["foo", "bar"])
/// }
/// ```
pub fn select<I, S>(&self, selection: I) -> PolarsResult<Self>
where
I: IntoIterator<Item = S>,
S: AsRef<str>,
{
let cols = selection
.into_iter()
.map(|s| s.as_ref().to_string())
.collect::<Vec<_>>();
self.select_impl(&cols)
}
fn select_impl(&self, cols: &[String]) -> PolarsResult<Self> {
self.select_check_duplicates(cols)?;
let selected = self.select_series_impl(cols)?;
Ok(DataFrame::new_no_checks(selected))
}
pub fn select_physical<I, S>(&self, selection: I) -> PolarsResult<Self>
where
I: IntoIterator<Item = S>,
S: AsRef<str>,
{
let cols = selection
.into_iter()
.map(|s| s.as_ref().to_string())
.collect::<Vec<_>>();
self.select_physical_impl(&cols)
}
fn select_physical_impl(&self, cols: &[String]) -> PolarsResult<Self> {
self.select_check_duplicates(cols)?;
let selected = self.select_series_physical_impl(cols)?;
Ok(DataFrame::new_no_checks(selected))
}
fn select_check_duplicates(&self, cols: &[String]) -> PolarsResult<()> {
let mut names = PlHashSet::with_capacity(cols.len());
for name in cols {
if !names.insert(name.as_str()) {
_duplicate_err(name)?
}
}
Ok(())
}
/// Select column(s) from this `DataFrame` and return them into a `Vec`.
///
/// # Example
///
/// ```rust
/// # use polars_core::prelude::*;
/// let df: DataFrame = df!("Name" => &["Methane", "Ethane", "Propane"],
/// "Carbon" => &[1, 2, 3],
/// "Hydrogen" => &[4, 6, 8])?;
/// let sv: Vec<Series> = df.select_series(&["Carbon", "Hydrogen"])?;
///
/// assert_eq!(df["Carbon"], sv[0]);
/// assert_eq!(df["Hydrogen"], sv[1]);
/// # Ok::<(), PolarsError>(())
/// ```
pub fn select_series(&self, selection: impl IntoVec<String>) -> PolarsResult<Vec<Series>> {
let cols = selection.into_vec();
self.select_series_impl(&cols)
}
fn _names_to_idx_map(&self) -> PlHashMap<&str, usize> {
self.columns
.iter()
.enumerate()
.map(|(i, s)| (s.name(), i))
.collect()
}
/// A non generic implementation to reduce compiler bloat.
fn select_series_physical_impl(&self, cols: &[String]) -> PolarsResult<Vec<Series>> {
let selected = if cols.len() > 1 && self.columns.len() > 10 {
let name_to_idx = self._names_to_idx_map();
cols.iter()
.map(|name| {
let idx = *name_to_idx
.get(name.as_str())
.ok_or_else(|| PolarsError::NotFound(name.to_string().into()))?;
Ok(self
.select_at_idx(idx)
.unwrap()
.to_physical_repr()
.into_owned())
})
.collect::<PolarsResult<Vec<_>>>()?
} else {
cols.iter()
.map(|c| self.column(c).map(|s| s.to_physical_repr().into_owned()))
.collect::<PolarsResult<Vec<_>>>()?
};
Ok(selected)
}
/// A non generic implementation to reduce compiler bloat.
fn select_series_impl(&self, cols: &[String]) -> PolarsResult<Vec<Series>> {
let selected = if cols.len() > 1 && self.columns.len() > 10 {
// we hash, because there are user that having millions of columns.
// # https://github.com/pola-rs/polars/issues/1023
let name_to_idx = self._names_to_idx_map();
cols.iter()
.map(|name| {
let idx = *name_to_idx
.get(name.as_str())
.ok_or_else(|| PolarsError::NotFound(name.to_string().into()))?;
Ok(self.select_at_idx(idx).unwrap().clone())
})
.collect::<PolarsResult<Vec<_>>>()?
} else {
cols.iter()
.map(|c| self.column(c).map(|s| s.clone()))
.collect::<PolarsResult<Vec<_>>>()?
};
Ok(selected)
}
/// Select a mutable series by name.
/// *Note: the length of the Series should remain the same otherwise the DataFrame is invalid.*
/// For this reason the method is not public
fn select_mut(&mut self, name: &str) -> Option<&mut Series> {
let opt_idx = self.find_idx_by_name(name);
match opt_idx {
Some(idx) => self.select_at_idx_mut(idx),
None => None,
}
}
/// Does a filter but splits thread chunks vertically instead of horizontally
/// This yields a DataFrame with `n_chunks == n_threads`.
fn filter_vertical(&mut self, mask: &BooleanChunked) -> PolarsResult<Self> {
let n_threads = POOL.current_num_threads();
let masks = split_ca(mask, n_threads).unwrap();
let dfs = split_df(self, n_threads).unwrap();
let dfs: PolarsResult<Vec<_>> = POOL.install(|| {
masks
.par_iter()
.zip(dfs)
.map(|(mask, df)| {
let cols = df
.columns
.iter()
.map(|s| s.filter(mask))
.collect::<PolarsResult<_>>()?;
Ok(DataFrame::new_no_checks(cols))
})
.collect()
});
let mut iter = dfs?.into_iter();
let first = iter.next().unwrap();
Ok(iter.fold(first, |mut acc, df| {
acc.vstack_mut(&df).unwrap();
acc
}))
}
/// Take the `DataFrame` rows by a boolean mask.
///
/// # Example
///
/// ```
/// # use polars_core::prelude::*;
/// fn example(df: &DataFrame) -> PolarsResult<DataFrame> {
/// let mask = df.column("sepal.width")?.is_not_null();
/// df.filter(&mask)
/// }
/// ```
pub fn filter(&self, mask: &BooleanChunked) -> PolarsResult<Self> {
if std::env::var("POLARS_VERT_PAR").is_ok() {
return self.clone().filter_vertical(mask);
}
let new_col = self.try_apply_columns_par(&|s| match s.dtype() {
DataType::Utf8 => s.filter_threaded(mask, true),
_ => s.filter(mask),
})?;
Ok(DataFrame::new_no_checks(new_col))
}
/// Same as `filter` but does not parallelize.
pub fn _filter_seq(&self, mask: &BooleanChunked) -> PolarsResult<Self> {
let new_col = self.try_apply_columns(&|s| s.filter(mask))?;
Ok(DataFrame::new_no_checks(new_col))
}
/// Take `DataFrame` value by indexes from an iterator.
///
/// # Example
///
/// ```
/// # use polars_core::prelude::*;
/// fn example(df: &DataFrame) -> PolarsResult<DataFrame> {
/// let iterator = (0..9).into_iter();
/// df.take_iter(iterator)
/// }
/// ```
pub fn take_iter<I>(&self, iter: I) -> PolarsResult<Self>
where
I: Iterator<Item = usize> + Clone + Sync + TrustedLen,
{
let new_col = self.try_apply_columns_par(&|s| {
let mut i = iter.clone();
s.take_iter(&mut i)
})?;
Ok(DataFrame::new_no_checks(new_col))
}
/// Take `DataFrame` values by indexes from an iterator.
///
/// # Safety
///
/// This doesn't do any bound checking but checks null validity.
#[must_use]
pub unsafe fn take_iter_unchecked<I>(&self, mut iter: I) -> Self
where
I: Iterator<Item = usize> + Clone + Sync + TrustedLen,
{
if std::env::var("POLARS_VERT_PAR").is_ok() {
let idx_ca: NoNull<IdxCa> = iter.into_iter().map(|idx| idx as IdxSize).collect();
return self.take_unchecked_vectical(&idx_ca.into_inner());
}
let n_chunks = self.n_chunks();
let has_utf8 = self
.columns
.iter()
.any(|s| matches!(s.dtype(), DataType::Utf8));
if (n_chunks == 1 && self.width() > 1) || has_utf8 {
let idx_ca: NoNull<IdxCa> = iter.into_iter().map(|idx| idx as IdxSize).collect();
let idx_ca = idx_ca.into_inner();
return self.take_unchecked(&idx_ca);
}
let new_col = if self.width() == 1 {
self.columns
.iter()
.map(|s| s.take_iter_unchecked(&mut iter))
.collect::<Vec<_>>()
} else {
self.apply_columns_par(&|s| {
let mut i = iter.clone();
s.take_iter_unchecked(&mut i)
})
};
DataFrame::new_no_checks(new_col)
}
/// Take `DataFrame` values by indexes from an iterator that may contain None values.
///
/// # Safety
///
/// This doesn't do any bound checking. Out of bounds may access uninitialized memory.
/// Null validity is checked
#[must_use]
pub unsafe fn take_opt_iter_unchecked<I>(&self, mut iter: I) -> Self
where
I: Iterator<Item = Option<usize>> + Clone + Sync + TrustedLen,
{
if std::env::var("POLARS_VERT_PAR").is_ok() {
let idx_ca: IdxCa = iter
.into_iter()
.map(|opt| opt.map(|v| v as IdxSize))
.collect();
return self.take_unchecked_vectical(&idx_ca);
}
let n_chunks = self.n_chunks();
let has_utf8 = self
.columns
.iter()
.any(|s| matches!(s.dtype(), DataType::Utf8));
if (n_chunks == 1 && self.width() > 1) || has_utf8 {
let idx_ca: IdxCa = iter
.into_iter()
.map(|opt| opt.map(|v| v as IdxSize))
.collect();
return self.take_unchecked(&idx_ca);
}
let new_col = if self.width() == 1 {
self.columns
.iter()
.map(|s| s.take_opt_iter_unchecked(&mut iter))
.collect::<Vec<_>>()
} else {
self.apply_columns_par(&|s| {
let mut i = iter.clone();
s.take_opt_iter_unchecked(&mut i)
})
};
DataFrame::new_no_checks(new_col)
}
/// Take `DataFrame` rows by index values.
///
/// # Example
///
/// ```
/// # use polars_core::prelude::*;
/// fn example(df: &DataFrame) -> PolarsResult<DataFrame> {
/// let idx = IdxCa::new("idx", &[0, 1, 9]);
/// df.take(&idx)
/// }
/// ```
pub fn take(&self, indices: &IdxCa) -> PolarsResult<Self> {
let indices = if indices.chunks.len() > 1 {
Cow::Owned(indices.rechunk())
} else {
Cow::Borrowed(indices)
};
let new_col = POOL.install(|| {
self.try_apply_columns_par(&|s| match s.dtype() {
DataType::Utf8 => s.take_threaded(&indices, true),
_ => s.take(&indices),
})
})?;
Ok(DataFrame::new_no_checks(new_col))
}
pub(crate) unsafe fn take_unchecked(&self, idx: &IdxCa) -> Self {
self.take_unchecked_impl(idx, true)
}
unsafe fn take_unchecked_impl(&self, idx: &IdxCa, allow_threads: bool) -> Self {
let cols = if allow_threads {
POOL.install(|| {
self.apply_columns_par(&|s| match s.dtype() {
DataType::Utf8 => s.take_unchecked_threaded(idx, true).unwrap(),
_ => s.take_unchecked(idx).unwrap(),
})
})
} else {
self.columns
.iter()
.map(|s| s.take_unchecked(idx).unwrap())
.collect()
};
DataFrame::new_no_checks(cols)
}
unsafe fn take_unchecked_vectical(&self, indices: &IdxCa) -> Self {
let n_threads = POOL.current_num_threads();
let idxs = split_ca(indices, n_threads).unwrap();
let dfs: Vec<_> = POOL.install(|| {
idxs.par_iter()
.map(|idx| {
let cols = self
.columns
.iter()
.map(|s| s.take_unchecked(idx).unwrap())
.collect();
DataFrame::new_no_checks(cols)
})
.collect()
});
let mut iter = dfs.into_iter();
let first = iter.next().unwrap();
iter.fold(first, |mut acc, df| {
acc.vstack_mut(&df).unwrap();
acc
})
}
/// Rename a column in the `DataFrame`.
///
/// # Example
///
/// ```
/// # use polars_core::prelude::*;
/// fn example(df: &mut DataFrame) -> PolarsResult<&mut DataFrame> {
/// let original_name = "foo";
/// let new_name = "bar";
/// df.rename(original_name, new_name)
/// }
/// ```
pub fn rename(&mut self, column: &str, name: &str) -> PolarsResult<&mut Self> {
self.select_mut(column)
.ok_or_else(|| PolarsError::NotFound(column.to_string().into()))
.map(|s| s.rename(name))?;
let unique_names: AHashSet<&str, ahash::RandomState> =
AHashSet::from_iter(self.columns.iter().map(|s| s.name()));
if unique_names.len() != self.columns.len() {
return Err(PolarsError::SchemaMisMatch(
"duplicate column names found".into(),
));
}
Ok(self)
}
/// Sort `DataFrame` in place by a column.
pub fn sort_in_place(
&mut self,
by_column: impl IntoVec<String>,
reverse: impl IntoVec<bool>,
) -> PolarsResult<&mut Self> {
// a lot of indirection in both sorting and take
self.as_single_chunk_par();
let by_column = self.select_series(by_column)?;
let reverse = reverse.into_vec();
self.columns = self.sort_impl(by_column, reverse, false, None)?.columns;
Ok(self)
}
/// This is the dispatch of Self::sort, and exists to reduce compile bloat by monomorphization.
#[cfg(feature = "private")]
pub fn sort_impl(
&self,
by_column: Vec<Series>,
reverse: Vec<bool>,
nulls_last: bool,
slice: Option<(i64, usize)>,
) -> PolarsResult<Self> {
// note that the by_column argument also contains evaluated expression from polars-lazy
// that may not even be present in this dataframe.
// therefore when we try to set the first columns as sorted, we ignore the error
// as expressions are not present (they are renamed to _POLARS_SORT_COLUMN_i.
let first_reverse = reverse[0];
let first_by_column = by_column[0].name().to_string();
let mut take = match by_column.len() {
1 => {
let s = &by_column[0];
let options = SortOptions {
descending: reverse[0],
nulls_last,
};
// fast path for a frame with a single series
// no need to compute the sort indices and then take by these indices
// simply sort and return as frame
if self.width() == 1 && self.check_name_to_idx(s.name()).is_ok() {
let mut out = s.sort_with(options);
if let Some((offset, len)) = slice {
out = out.slice(offset, len);
}
return Ok(out.into_frame());
}
s.argsort(options)
}
_ => {
#[cfg(feature = "sort_multiple")]
{
let (first, by_column, reverse) = prepare_argsort(by_column, reverse)?;
first.argsort_multiple(&by_column, &reverse)?
}
#[cfg(not(feature = "sort_multiple"))]
{
panic!("activate `sort_multiple` feature gate to enable this functionality");
}
}
};
if let Some((offset, len)) = slice {
take = take.slice(offset, len);
}
// Safety:
// the created indices are in bounds
let mut df = if std::env::var("POLARS_VERT_PAR").is_ok() {
unsafe { self.take_unchecked_vectical(&take) }
} else {
unsafe { self.take_unchecked(&take) }
};
// Mark the first sort column as sorted
// if the column did not exists it is ok, because we sorted by an expression
// not present in the dataframe
let _ = df.apply(&first_by_column, |s| {
let mut s = s.clone();
if first_reverse {
s.set_sorted(IsSorted::Descending)
} else {
s.set_sorted(IsSorted::Ascending)
}
s
});
Ok(df)
}
/// Return a sorted clone of this `DataFrame`.
///
/// # Example
///
/// ```
/// # use polars_core::prelude::*;
/// fn sort_example(df: &DataFrame, reverse: bool) -> PolarsResult<DataFrame> {
/// df.sort(["a"], reverse)
/// }
///
/// fn sort_by_multiple_columns_example(df: &DataFrame) -> PolarsResult<DataFrame> {
/// df.sort(&["a", "b"], vec![false, true])
/// }
/// ```
pub fn sort(
&self,
by_column: impl IntoVec<String>,
reverse: impl IntoVec<bool>,
) -> PolarsResult<Self> {
let mut df = self.clone();
df.sort_in_place(by_column, reverse)?;
Ok(df)
}
/// Sort the `DataFrame` by a single column with extra options.
pub fn sort_with_options(&self, by_column: &str, options: SortOptions) -> PolarsResult<Self> {
let mut df = self.clone();
// a lot of indirection in both sorting and take
df.as_single_chunk_par();
let by_column = vec![df.column(by_column)?.clone()];
let reverse = vec![options.descending];
df.columns = df
.sort_impl(by_column, reverse, options.nulls_last, None)?
.columns;
Ok(df)
}
/// Replace a column with a `Series`.
///
/// # Example
///
/// ```rust
/// # use polars_core::prelude::*;
/// let mut df: DataFrame = df!("Country" => &["United States", "China"],
/// "Area (km²)" => &[9_833_520, 9_596_961])?;
/// let s: Series = Series::new("Country", &["USA", "PRC"]);
///
/// assert!(df.replace("Nation", s.clone()).is_err());
/// assert!(df.replace("Country", s).is_ok());
/// # Ok::<(), PolarsError>(())
/// ```
pub fn replace<S: IntoSeries>(&mut self, column: &str, new_col: S) -> PolarsResult<&mut Self> {
self.apply(column, |_| new_col.into_series())
}
/// Replace or update a column. The difference between this method and [DataFrame::with_column]
/// is that now the value of `column: &str` determines the name of the column and not the name
/// of the `Series` passed to this method.
pub fn replace_or_add<S: IntoSeries>(
&mut self,
column: &str,
new_col: S,
) -> PolarsResult<&mut Self> {
let mut new_col = new_col.into_series();
new_col.rename(column);
self.with_column(new_col)
}
/// Replace column at index `idx` with a `Series`.
///
/// # Example
///
/// ```ignored
/// # use polars_core::prelude::*;
/// let s0 = Series::new("foo", &["ham", "spam", "egg"]);
/// let s1 = Series::new("ascii", &[70, 79, 79]);
/// let mut df = DataFrame::new(vec![s0, s1])?;
///
/// // Add 32 to get lowercase ascii values
/// df.replace_at_idx(1, df.select_at_idx(1).unwrap() + 32);
/// # Ok::<(), PolarsError>(())
/// ```
pub fn replace_at_idx<S: IntoSeries>(
&mut self,
idx: usize,
new_col: S,
) -> PolarsResult<&mut Self> {
let mut new_column = new_col.into_series();
if new_column.len() != self.height() {
return Err(PolarsError::ShapeMisMatch(
format!("Cannot replace Series at index {}. The shape of Series {} does not match that of the DataFrame {}",
idx, new_column.len(), self.height()
).into()));
};
if idx >= self.width() {
return Err(PolarsError::ComputeError(
format!(
"Column index: {} outside of DataFrame with {} columns",
idx,
self.width()
)
.into(),
));
}
let old_col = &mut self.columns[idx];
mem::swap(old_col, &mut new_column);
Ok(self)
}
/// Apply a closure to a column. This is the recommended way to do in place modification.
///
/// # Example
///
/// ```rust
/// # use polars_core::prelude::*;
/// let s0 = Series::new("foo", &["ham", "spam", "egg"]);
/// let s1 = Series::new("names", &["Jean", "Claude", "van"]);
/// let mut df = DataFrame::new(vec![s0, s1])?;
///
/// fn str_to_len(str_val: &Series) -> Series {
/// str_val.utf8()
/// .unwrap()
/// .into_iter()
/// .map(|opt_name: Option<&str>| {
/// opt_name.map(|name: &str| name.len() as u32)
/// })
/// .collect::<UInt32Chunked>()
/// .into_series()
/// }
///
/// // Replace the names column by the length of the names.
/// df.apply("names", str_to_len);
/// # Ok::<(), PolarsError>(())
/// ```
/// Results in:
///
/// ```text
/// +--------+-------+
/// | foo | |
/// | --- | names |
/// | str | u32 |
/// +========+=======+
/// | "ham" | 4 |
/// +--------+-------+
/// | "spam" | 6 |
/// +--------+-------+
/// | "egg" | 3 |
/// +--------+-------+
/// ```
pub fn apply<F, S>(&mut self, name: &str, f: F) -> PolarsResult<&mut Self>
where
F: FnOnce(&Series) -> S,
S: IntoSeries,
{
let idx = self.check_name_to_idx(name)?;
self.apply_at_idx(idx, f)
}
/// Apply a closure to a column at index `idx`. This is the recommended way to do in place
/// modification.
///
/// # Example
///
/// ```rust
/// # use polars_core::prelude::*;
/// let s0 = Series::new("foo", &["ham", "spam", "egg"]);
/// let s1 = Series::new("ascii", &[70, 79, 79]);
/// let mut df = DataFrame::new(vec![s0, s1])?;
///
/// // Add 32 to get lowercase ascii values
/// df.apply_at_idx(1, |s| s + 32);
/// # Ok::<(), PolarsError>(())
/// ```
/// Results in:
///
/// ```text
/// +--------+-------+
/// | foo | ascii |
/// | --- | --- |
/// | str | i32 |
/// +========+=======+
/// | "ham" | 102 |
/// +--------+-------+
/// | "spam" | 111 |
/// +--------+-------+
/// | "egg" | 111 |
/// +--------+-------+
/// ```
pub fn apply_at_idx<F, S>(&mut self, idx: usize, f: F) -> PolarsResult<&mut Self>
where
F: FnOnce(&Series) -> S,
S: IntoSeries,
{
let df_height = self.height();
let width = self.width();
let col = self.columns.get_mut(idx).ok_or_else(|| {
PolarsError::ComputeError(
format!("Column index: {idx} outside of DataFrame with {width} columns",).into(),
)
})?;
let name = col.name().to_string();
let new_col = f(col).into_series();
match new_col.len() {
1 => {
let new_col = new_col.new_from_index(0, df_height);
let _ = mem::replace(col, new_col);
}
len if (len == df_height) => {
let _ = mem::replace(col, new_col);
}
len => {
return Err(PolarsError::ShapeMisMatch(
format!(
"Result Series has shape {} where the DataFrame has height {}",
len,
self.height()
)
.into(),
));
}
}
// make sure the name remains the same after applying the closure
unsafe {
let col = self.columns.get_unchecked_mut(idx);
col.rename(&name);
}
Ok(self)
}
/// Apply a closure that may fail to a column at index `idx`. This is the recommended way to do in place
/// modification.
///
/// # Example
///
/// This is the idiomatic way to replace some values a column of a `DataFrame` given range of indexes.
///
/// ```rust
/// # use polars_core::prelude::*;
/// let s0 = Series::new("foo", &["ham", "spam", "egg", "bacon", "quack"]);
/// let s1 = Series::new("values", &[1, 2, 3, 4, 5]);
/// let mut df = DataFrame::new(vec![s0, s1])?;
///
/// let idx = vec![0, 1, 4];
///
/// df.try_apply("foo", |s| {
/// s.utf8()?
/// .set_at_idx_with(idx, |opt_val| opt_val.map(|string| format!("{}-is-modified", string)))
/// });
/// # Ok::<(), PolarsError>(())
/// ```
/// Results in:
///
/// ```text
/// +---------------------+--------+
/// | foo | values |
/// | --- | --- |
/// | str | i32 |
/// +=====================+========+
/// | "ham-is-modified" | 1 |
/// +---------------------+--------+
/// | "spam-is-modified" | 2 |
/// +---------------------+--------+
/// | "egg" | 3 |
/// +---------------------+--------+
/// | "bacon" | 4 |
/// +---------------------+--------+
/// | "quack-is-modified" | 5 |
/// +---------------------+--------+
/// ```
pub fn try_apply_at_idx<F, S>(&mut self, idx: usize, f: F) -> PolarsResult<&mut Self>
where
F: FnOnce(&Series) -> PolarsResult<S>,
S: IntoSeries,
{
let width = self.width();
let col = self.columns.get_mut(idx).ok_or_else(|| {
PolarsError::ComputeError(
format!("Column index: {idx} outside of DataFrame with {width} columns",).into(),
)
})?;
let name = col.name().to_string();
let _ = mem::replace(col, f(col).map(|s| s.into_series())?);
// make sure the name remains the same after applying the closure
unsafe {
let col = self.columns.get_unchecked_mut(idx);
col.rename(&name);
}
Ok(self)
}
/// Apply a closure that may fail to a column. This is the recommended way to do in place
/// modification.
///
/// # Example
///
/// This is the idiomatic way to replace some values a column of a `DataFrame` given a boolean mask.
///
/// ```rust
/// # use polars_core::prelude::*;
/// let s0 = Series::new("foo", &["ham", "spam", "egg", "bacon", "quack"]);
/// let s1 = Series::new("values", &[1, 2, 3, 4, 5]);
/// let mut df = DataFrame::new(vec![s0, s1])?;
///
/// // create a mask
/// let values = df.column("values")?;
/// let mask = values.lt_eq(1)? | values.gt_eq(5_i32)?;
///
/// df.try_apply("foo", |s| {
/// s.utf8()?
/// .set(&mask, Some("not_within_bounds"))
/// });
/// # Ok::<(), PolarsError>(())
/// ```
/// Results in:
///
/// ```text
/// +---------------------+--------+
/// | foo | values |
/// | --- | --- |
/// | str | i32 |
/// +=====================+========+
/// | "not_within_bounds" | 1 |
/// +---------------------+--------+
/// | "spam" | 2 |
/// +---------------------+--------+
/// | "egg" | 3 |
/// +---------------------+--------+
/// | "bacon" | 4 |
/// +---------------------+--------+
/// | "not_within_bounds" | 5 |
/// +---------------------+--------+
/// ```
pub fn try_apply<F, S>(&mut self, column: &str, f: F) -> PolarsResult<&mut Self>
where
F: FnOnce(&Series) -> PolarsResult<S>,
S: IntoSeries,
{
let idx = self
.find_idx_by_name(column)
.ok_or_else(|| PolarsError::NotFound(column.to_string().into()))?;
self.try_apply_at_idx(idx, f)
}
/// Slice the `DataFrame` along the rows.
///
/// # Example
///
/// ```rust
/// # use polars_core::prelude::*;
/// let df: DataFrame = df!("Fruit" => &["Apple", "Grape", "Grape", "Fig", "Fig"],
/// "Color" => &["Green", "Red", "White", "White", "Red"])?;
/// let sl: DataFrame = df.slice(2, 3);
///
/// assert_eq!(sl.shape(), (3, 2));
/// println!("{}", sl);
/// # Ok::<(), PolarsError>(())
/// ```
/// Output:
/// ```text
/// shape: (3, 2)
/// +-------+-------+
/// | Fruit | Color |
/// | --- | --- |
/// | str | str |
/// +=======+=======+
/// | Grape | White |
/// +-------+-------+
/// | Fig | White |
/// +-------+-------+
/// | Fig | Red |
/// +-------+-------+
/// ```
#[must_use]
pub fn slice(&self, offset: i64, length: usize) -> Self {
if offset == 0 && length == self.height() {
return self.clone();
}
let col = self
.columns
.iter()
.map(|s| s.slice(offset, length))
.collect::<Vec<_>>();
DataFrame::new_no_checks(col)
}
#[must_use]
pub fn slice_par(&self, offset: i64, length: usize) -> Self {
if offset == 0 && length == self.height() {
return self.clone();
}
DataFrame::new_no_checks(self.apply_columns_par(&|s| s.slice(offset, length)))
}
#[must_use]
pub fn _slice_and_realloc(&self, offset: i64, length: usize) -> Self {
if offset == 0 && length == self.height() {
return self.clone();
}
DataFrame::new_no_checks(self.apply_columns(&|s| {
let mut out = s.slice(offset, length);
out.shrink_to_fit();
out
}))
}sourcepub fn append(&mut self, other: &Series) -> PolarsResult<&mut Self>
pub fn append(&mut self, other: &Series) -> PolarsResult<&mut Self>
Append in place. This is done by adding the chunks of other to this Series.
See ChunkedArray::append and ChunkedArray::extend.
Examples found in repository?
src/series/mod.rs (line 397)
393 394 395 396 397 398 399 400 401 402 403 404 405 406
fn finish_take_threaded(&self, s: Vec<Series>, rechunk: bool) -> Series {
let s = s
.into_iter()
.reduce(|mut s, s1| {
s.append(&s1).unwrap();
s
})
.unwrap();
if rechunk {
s.rechunk()
} else {
s
}
}More examples
src/utils/mod.rs (line 656)
636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661
pub fn parallel_op_series<F>(f: F, s: Series, n_threads: Option<usize>) -> PolarsResult<Series>
where
F: Fn(Series) -> PolarsResult<Series> + Send + Sync,
{
let n_threads = n_threads.unwrap_or_else(|| POOL.current_num_threads());
let splits = _split_offsets(s.len(), n_threads);
let chunks = POOL.install(|| {
splits
.into_par_iter()
.map(|(offset, len)| {
let s = s.slice(offset as i64, len);
f(s)
})
.collect::<PolarsResult<Vec<_>>>()
})?;
let mut iter = chunks.into_iter();
let first = iter.next().unwrap();
let out = iter.fold(first, |mut acc, s| {
acc.append(&s).unwrap();
acc
});
f(out)
}src/frame/mod.rs (line 915)
898 899 900 901 902 903 904 905 906 907 908 909 910 911 912 913 914 915 916 917 918 919 920 921 922 923 924 925 926 927 928 929
pub fn vstack_mut(&mut self, other: &DataFrame) -> PolarsResult<&mut Self> {
if self.width() != other.width() {
if self.width() == 0 {
self.columns = other.columns.clone();
return Ok(self);
}
return Err(PolarsError::ShapeMisMatch(
format!("Could not vertically stack DataFrame. The DataFrames appended width {} differs from the parent DataFrames width {}", self.width(), other.width()).into()
));
}
self.columns
.iter_mut()
.zip(other.columns.iter())
.try_for_each::<_, PolarsResult<_>>(|(left, right)| {
can_extend(left, right)?;
left.append(right).expect("should not fail");
Ok(())
})?;
Ok(self)
}
/// Does not check if schema is correct
pub(crate) fn vstack_mut_unchecked(&mut self, other: &DataFrame) {
self.columns
.iter_mut()
.zip(other.columns.iter())
.for_each(|(left, right)| {
left.append(right).expect("should not fail");
});
}src/series/ops/extend.rs (line 23)
5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25
pub fn extend_constant(&self, value: AnyValue, n: usize) -> PolarsResult<Self> {
use AnyValue::*;
let s = match value {
Float32(v) => Series::new("", vec![v]),
Float64(v) => Series::new("", vec![v]),
UInt32(v) => Series::new("", vec![v]),
UInt64(v) => Series::new("", vec![v]),
Int32(v) => Series::new("", vec![v]),
Int64(v) => Series::new("", vec![v]),
Utf8(v) => Series::new("", vec![v]),
Boolean(v) => Series::new("", vec![v]),
Null => BooleanChunked::full_null("", 1).into_series(),
dt => panic!("{dt:?} not supported"),
};
let s = s.cast(self.dtype())?;
let to_append = s.new_from_index(0, n);
let mut out = self.clone();
out.append(&to_append)?;
Ok(out)
}sourcepub fn extend(&mut self, other: &Series) -> PolarsResult<&mut Self>
pub fn extend(&mut self, other: &Series) -> PolarsResult<&mut Self>
Extend the memory backed by this array with the values from other.
See ChunkedArray::extend and ChunkedArray::append.
Examples found in repository?
src/frame/mod.rs (line 957)
945 946 947 948 949 950 951 952 953 954 955 956 957 958 959 960 961
pub fn extend(&mut self, other: &DataFrame) -> PolarsResult<()> {
if self.width() != other.width() {
return Err(PolarsError::ShapeMisMatch(
format!("Could not extend DataFrame. The DataFrames extended width {} differs from the parent DataFrames width {}", self.width(), other.width()).into()
));
}
self.columns
.iter_mut()
.zip(other.columns.iter())
.try_for_each::<_, PolarsResult<_>>(|(left, right)| {
can_extend(left, right)?;
left.extend(right).unwrap();
Ok(())
})?;
Ok(())
}pub fn sort(&self, reverse: bool) -> Self
sourcepub fn as_single_ptr(&mut self) -> PolarsResult<usize>
pub fn as_single_ptr(&mut self) -> PolarsResult<usize>
Only implemented for numeric types
sourcepub fn cast(&self, dtype: &DataType) -> PolarsResult<Self>
pub fn cast(&self, dtype: &DataType) -> PolarsResult<Self>
Cast [Series] to another [DataType]
Examples found in repository?
src/frame/groupby/aggregations/dispatch.rs (line 11)
9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184
fn restore_logical(&self, out: Series) -> Series {
if self.dtype().is_logical() {
out.cast(self.dtype()).unwrap()
} else {
out
}
}
#[doc(hidden)]
pub fn agg_valid_count(&self, groups: &GroupsProxy) -> Series {
match groups {
GroupsProxy::Idx(groups) => agg_helper_idx_on_all::<IdxType, _>(groups, |idx| {
debug_assert!(idx.len() <= self.len());
if idx.is_empty() {
None
} else if !self.has_validity() {
Some(idx.len() as IdxSize)
} else {
let take =
unsafe { self.take_iter_unchecked(&mut idx.iter().map(|i| *i as usize)) };
Some((take.len() - take.null_count()) as IdxSize)
}
}),
GroupsProxy::Slice { groups, .. } => {
_agg_helper_slice::<IdxType, _>(groups, |[first, len]| {
debug_assert!(len <= self.len() as IdxSize);
if len == 0 {
None
} else if !self.has_validity() {
Some(len)
} else {
let take = self.slice_from_offsets(first, len);
Some((take.len() - take.null_count()) as IdxSize)
}
})
}
}
}
#[doc(hidden)]
pub unsafe fn agg_first(&self, groups: &GroupsProxy) -> Series {
let out = match groups {
GroupsProxy::Idx(groups) => {
let mut iter = groups.iter().map(|(first, idx)| {
if idx.is_empty() {
None
} else {
Some(first as usize)
}
});
// Safety:
// groups are always in bounds
self.take_opt_iter_unchecked(&mut iter)
}
GroupsProxy::Slice { groups, .. } => {
let mut iter =
groups.iter().map(
|&[first, len]| {
if len == 0 {
None
} else {
Some(first as usize)
}
},
);
// Safety:
// groups are always in bounds
self.take_opt_iter_unchecked(&mut iter)
}
};
self.restore_logical(out)
}
#[doc(hidden)]
pub unsafe fn agg_n_unique(&self, groups: &GroupsProxy) -> Series {
match groups {
GroupsProxy::Idx(groups) => agg_helper_idx_on_all::<IdxType, _>(groups, |idx| {
debug_assert!(idx.len() <= self.len());
if idx.is_empty() {
None
} else {
let take = self.take_iter_unchecked(&mut idx.iter().map(|i| *i as usize));
take.n_unique().ok().map(|v| v as IdxSize)
}
}),
GroupsProxy::Slice { groups, .. } => {
_agg_helper_slice::<IdxType, _>(groups, |[first, len]| {
debug_assert!(len <= self.len() as IdxSize);
if len == 0 {
None
} else {
let take = self.slice_from_offsets(first, len);
take.n_unique().ok().map(|v| v as IdxSize)
}
})
}
}
}
#[doc(hidden)]
pub unsafe fn agg_median(&self, groups: &GroupsProxy) -> Series {
use DataType::*;
match self.dtype() {
Float32 => SeriesWrap(self.f32().unwrap().clone()).agg_median(groups),
Float64 => SeriesWrap(self.f64().unwrap().clone()).agg_median(groups),
dt if dt.is_numeric() || dt.is_temporal() => {
let ca = self.to_physical_repr();
let physical_type = ca.dtype();
let s = apply_method_physical_integer!(ca, agg_median, groups);
if dt.is_logical() {
// back to physical and then
// back to logical type
s.cast(physical_type).unwrap().cast(dt).unwrap()
} else {
s
}
}
_ => Series::full_null("", groups.len(), self.dtype()),
}
}
#[doc(hidden)]
pub unsafe fn agg_quantile(
&self,
groups: &GroupsProxy,
quantile: f64,
interpol: QuantileInterpolOptions,
) -> Series {
use DataType::*;
match self.dtype() {
Float32 => {
SeriesWrap(self.f32().unwrap().clone()).agg_quantile(groups, quantile, interpol)
}
Float64 => {
SeriesWrap(self.f64().unwrap().clone()).agg_quantile(groups, quantile, interpol)
}
dt if dt.is_numeric() || dt.is_temporal() => {
let ca = self.to_physical_repr();
let physical_type = ca.dtype();
let s =
apply_method_physical_integer!(ca, agg_quantile, groups, quantile, interpol);
if dt.is_logical() {
// back to physical and then
// back to logical type
s.cast(physical_type).unwrap().cast(dt).unwrap()
} else {
s
}
}
_ => Series::full_null("", groups.len(), self.dtype()),
}
}
#[doc(hidden)]
pub unsafe fn agg_mean(&self, groups: &GroupsProxy) -> Series {
use DataType::*;
match self.dtype() {
Boolean => self.cast(&Float64).unwrap().agg_mean(groups),
Float32 => SeriesWrap(self.f32().unwrap().clone()).agg_mean(groups),
Float64 => SeriesWrap(self.f64().unwrap().clone()).agg_mean(groups),
dt if dt.is_numeric() => {
apply_method_physical_integer!(self, agg_mean, groups)
}
dt @ Duration(_) => {
let s = self.to_physical_repr();
// agg_mean returns Float64
let out = s.agg_mean(groups);
// cast back to Int64 and then to logical duration type
out.cast(&Int64).unwrap().cast(dt).unwrap()
}
_ => Series::full_null("", groups.len(), self.dtype()),
}
}More examples
src/series/mod.rs (line 251)
246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751 752 753 754 755 756 757 758 759 760 761 762 763 764 765 766 767 768 769 770 771 772 773 774 775 776 777 778 779 780 781 782 783 784 785 786 787 788 789 790 791 792 793 794 795 796 797 798 799 800 801 802 803 804 805 806 807 808 809 810 811 812 813 814 815 816 817 818 819 820 821 822 823 824 825 826 827 828 829 830 831 832 833 834 835 836 837 838 839 840 841 842 843 844 845 846 847 848 849 850 851 852 853 854 855 856 857 858 859 860 861 862 863 864 865 866 867 868 869 870
pub fn sum<T>(&self) -> Option<T>
where
T: NumCast,
{
self.sum_as_series()
.cast(&DataType::Float64)
.ok()
.and_then(|s| s.f64().unwrap().get(0).and_then(T::from))
}
/// Returns the minimum value in the array, according to the natural order.
/// Returns an option because the array is nullable.
/// ```
/// # use polars_core::prelude::*;
/// let s = Series::new("days", [1, 2, 3].as_ref());
/// assert_eq!(s.min(), Some(1));
/// ```
pub fn min<T>(&self) -> Option<T>
where
T: NumCast,
{
self.min_as_series()
.cast(&DataType::Float64)
.ok()
.and_then(|s| s.f64().unwrap().get(0).and_then(T::from))
}
/// Returns the maximum value in the array, according to the natural order.
/// Returns an option because the array is nullable.
/// ```
/// # use polars_core::prelude::*;
/// let s = Series::new("days", [1, 2, 3].as_ref());
/// assert_eq!(s.max(), Some(3));
/// ```
pub fn max<T>(&self) -> Option<T>
where
T: NumCast,
{
self.max_as_series()
.cast(&DataType::Float64)
.ok()
.and_then(|s| s.f64().unwrap().get(0).and_then(T::from))
}
/// Explode a list or utf8 Series. This expands every item to a new row..
pub fn explode(&self) -> PolarsResult<Series> {
match self.dtype() {
DataType::List(_) => self.list().unwrap().explode(),
DataType::Utf8 => self.utf8().unwrap().explode(),
_ => Err(PolarsError::InvalidOperation(
format!(
"explode not supported for Series with dtype {:?}",
self.dtype()
)
.into(),
)),
}
}
/// Check if float value is NaN (note this is different than missing/ null)
pub fn is_nan(&self) -> PolarsResult<BooleanChunked> {
match self.dtype() {
DataType::Float32 => Ok(self.f32().unwrap().is_nan()),
DataType::Float64 => Ok(self.f64().unwrap().is_nan()),
_ => Err(PolarsError::InvalidOperation(
format!(
"'is_nan' not supported for series with dtype {:?}",
self.dtype()
)
.into(),
)),
}
}
/// Check if float value is NaN (note this is different than missing/ null)
pub fn is_not_nan(&self) -> PolarsResult<BooleanChunked> {
match self.dtype() {
DataType::Float32 => Ok(self.f32().unwrap().is_not_nan()),
DataType::Float64 => Ok(self.f64().unwrap().is_not_nan()),
_ => Err(PolarsError::InvalidOperation(
format!(
"'is_not_nan' not supported for series with dtype {:?}",
self.dtype()
)
.into(),
)),
}
}
/// Check if float value is finite
pub fn is_finite(&self) -> PolarsResult<BooleanChunked> {
match self.dtype() {
DataType::Float32 => Ok(self.f32().unwrap().is_finite()),
DataType::Float64 => Ok(self.f64().unwrap().is_finite()),
_ => Err(PolarsError::InvalidOperation(
format!(
"'is_finite' not supported for series with dtype {:?}",
self.dtype()
)
.into(),
)),
}
}
/// Check if float value is infinite
pub fn is_infinite(&self) -> PolarsResult<BooleanChunked> {
match self.dtype() {
DataType::Float32 => Ok(self.f32().unwrap().is_infinite()),
DataType::Float64 => Ok(self.f64().unwrap().is_infinite()),
_ => Err(PolarsError::InvalidOperation(
format!(
"'is_infinite' not supported for series with dtype {:?}",
self.dtype()
)
.into(),
)),
}
}
/// Create a new ChunkedArray with values from self where the mask evaluates `true` and values
/// from `other` where the mask evaluates `false`
#[cfg(feature = "zip_with")]
#[cfg_attr(docsrs, doc(cfg(feature = "zip_with")))]
pub fn zip_with(&self, mask: &BooleanChunked, other: &Series) -> PolarsResult<Series> {
let (lhs, rhs) = coerce_lhs_rhs(self, other)?;
lhs.zip_with_same_type(mask, rhs.as_ref())
}
/// Cast a datelike Series to their physical representation.
/// Primitives remain unchanged
///
/// * Date -> Int32
/// * Datetime-> Int64
/// * Time -> Int64
/// * Categorical -> UInt32
///
pub fn to_physical_repr(&self) -> Cow<Series> {
use DataType::*;
match self.dtype() {
Date => Cow::Owned(self.cast(&DataType::Int32).unwrap()),
Datetime(_, _) | Duration(_) | Time => Cow::Owned(self.cast(&DataType::Int64).unwrap()),
#[cfg(feature = "dtype-categorical")]
Categorical(_) => Cow::Owned(self.cast(&DataType::UInt32).unwrap()),
_ => Cow::Borrowed(self),
}
}
fn finish_take_threaded(&self, s: Vec<Series>, rechunk: bool) -> Series {
let s = s
.into_iter()
.reduce(|mut s, s1| {
s.append(&s1).unwrap();
s
})
.unwrap();
if rechunk {
s.rechunk()
} else {
s
}
}
// take a function pointer to reduce bloat
fn threaded_op(
&self,
rechunk: bool,
len: usize,
func: &(dyn Fn(usize, usize) -> PolarsResult<Series> + Send + Sync),
) -> PolarsResult<Series> {
let n_threads = POOL.current_num_threads();
let offsets = _split_offsets(len, n_threads);
let series: PolarsResult<Vec<_>> = POOL.install(|| {
offsets
.into_par_iter()
.map(|(offset, len)| func(offset, len))
.collect()
});
Ok(self.finish_take_threaded(series?, rechunk))
}
/// Take by index if ChunkedArray contains a single chunk.
///
/// # Safety
/// This doesn't check any bounds. Null validity is checked.
pub unsafe fn take_unchecked_threaded(
&self,
idx: &IdxCa,
rechunk: bool,
) -> PolarsResult<Series> {
self.threaded_op(rechunk, idx.len(), &|offset, len| {
let idx = idx.slice(offset as i64, len);
self.take_unchecked(&idx)
})
}
/// # Safety
/// This doesn't check any bounds. Null validity is checked.
#[cfg(feature = "chunked_ids")]
pub(crate) unsafe fn _take_chunked_unchecked_threaded(
&self,
chunk_ids: &[ChunkId],
sorted: IsSorted,
rechunk: bool,
) -> Series {
self.threaded_op(rechunk, chunk_ids.len(), &|offset, len| {
let chunk_ids = &chunk_ids[offset..offset + len];
Ok(self._take_chunked_unchecked(chunk_ids, sorted))
})
.unwrap()
}
/// # Safety
/// This doesn't check any bounds. Null validity is checked.
#[cfg(feature = "chunked_ids")]
pub(crate) unsafe fn _take_opt_chunked_unchecked_threaded(
&self,
chunk_ids: &[Option<ChunkId>],
rechunk: bool,
) -> Series {
self.threaded_op(rechunk, chunk_ids.len(), &|offset, len| {
let chunk_ids = &chunk_ids[offset..offset + len];
Ok(self._take_opt_chunked_unchecked(chunk_ids))
})
.unwrap()
}
/// Take by index. This operation is clone.
///
/// # Notes
/// Out of bounds access doesn't Error but will return a Null value
pub fn take_threaded(&self, idx: &IdxCa, rechunk: bool) -> PolarsResult<Series> {
self.threaded_op(rechunk, idx.len(), &|offset, len| {
let idx = idx.slice(offset as i64, len);
self.take(&idx)
})
}
/// Filter by boolean mask. This operation clones data.
pub fn filter_threaded(&self, filter: &BooleanChunked, rechunk: bool) -> PolarsResult<Series> {
// this would fail if there is a broadcasting filter.
// because we cannot split that filter over threads
// besides they are a no-op, so we do the standard filter.
if filter.len() == 1 {
return self.filter(filter);
}
let n_threads = POOL.current_num_threads();
let filters = split_ca(filter, n_threads).unwrap();
let series = split_series(self, n_threads).unwrap();
let series: PolarsResult<Vec<_>> = POOL.install(|| {
filters
.par_iter()
.zip(series)
.map(|(filter, s)| s.filter(filter))
.collect()
});
Ok(self.finish_take_threaded(series?, rechunk))
}
#[cfg(feature = "dot_product")]
#[cfg_attr(docsrs, doc(cfg(feature = "dot_product")))]
pub fn dot(&self, other: &Series) -> Option<f64> {
(self * other).sum::<f64>()
}
/// Get the sum of the Series as a new Series of length 1.
/// Returns a Series with a single zeroed entry if self is an empty numeric series.
///
/// If the [`DataType`] is one of `{Int8, UInt8, Int16, UInt16}` the `Series` is
/// first cast to `Int64` to prevent overflow issues.
pub fn sum_as_series(&self) -> Series {
use DataType::*;
if self.is_empty() && self.dtype().is_numeric() {
return Series::new("", [0])
.cast(self.dtype())
.unwrap()
.sum_as_series();
}
match self.dtype() {
Int8 | UInt8 | Int16 | UInt16 => self.cast(&Int64).unwrap().sum_as_series(),
_ => self._sum_as_series(),
}
}
/// Get an array with the cumulative max computed at every element
#[cfg_attr(docsrs, doc(cfg(feature = "cum_agg")))]
pub fn cummax(&self, _reverse: bool) -> Series {
#[cfg(feature = "cum_agg")]
{
self._cummax(_reverse)
}
#[cfg(not(feature = "cum_agg"))]
{
panic!("activate 'cum_agg' feature")
}
}
/// Get an array with the cumulative min computed at every element
#[cfg_attr(docsrs, doc(cfg(feature = "cum_agg")))]
pub fn cummin(&self, _reverse: bool) -> Series {
#[cfg(feature = "cum_agg")]
{
self._cummin(_reverse)
}
#[cfg(not(feature = "cum_agg"))]
{
panic!("activate 'cum_agg' feature")
}
}
/// Get an array with the cumulative sum computed at every element
///
/// If the [`DataType`] is one of `{Int8, UInt8, Int16, UInt16}` the `Series` is
/// first cast to `Int64` to prevent overflow issues.
#[cfg_attr(docsrs, doc(cfg(feature = "cum_agg")))]
#[allow(unused_variables)]
pub fn cumsum(&self, reverse: bool) -> Series {
#[cfg(feature = "cum_agg")]
{
use DataType::*;
match self.dtype() {
Boolean => self.cast(&DataType::UInt32).unwrap().cumsum(reverse),
Int8 | UInt8 | Int16 | UInt16 => {
let s = self.cast(&Int64).unwrap();
s.cumsum(reverse)
}
Int32 => {
let ca = self.i32().unwrap();
ca.cumsum(reverse).into_series()
}
UInt32 => {
let ca = self.u32().unwrap();
ca.cumsum(reverse).into_series()
}
UInt64 => {
let ca = self.u64().unwrap();
ca.cumsum(reverse).into_series()
}
Int64 => {
let ca = self.i64().unwrap();
ca.cumsum(reverse).into_series()
}
Float32 => {
let ca = self.f32().unwrap();
ca.cumsum(reverse).into_series()
}
Float64 => {
let ca = self.f64().unwrap();
ca.cumsum(reverse).into_series()
}
dt => panic!("cumsum not supported for dtype: {dt:?}"),
}
}
#[cfg(not(feature = "cum_agg"))]
{
panic!("activate 'cum_agg' feature")
}
}
/// Get an array with the cumulative product computed at every element
///
/// If the [`DataType`] is one of `{Int8, UInt8, Int16, UInt16, Int32, UInt32}` the `Series` is
/// first cast to `Int64` to prevent overflow issues.
#[cfg_attr(docsrs, doc(cfg(feature = "cum_agg")))]
#[allow(unused_variables)]
pub fn cumprod(&self, reverse: bool) -> Series {
#[cfg(feature = "cum_agg")]
{
use DataType::*;
match self.dtype() {
Boolean => self.cast(&DataType::Int64).unwrap().cumprod(reverse),
Int8 | UInt8 | Int16 | UInt16 | Int32 | UInt32 => {
let s = self.cast(&Int64).unwrap();
s.cumprod(reverse)
}
Int64 => {
let ca = self.i64().unwrap();
ca.cumprod(reverse).into_series()
}
UInt64 => {
let ca = self.u64().unwrap();
ca.cumprod(reverse).into_series()
}
Float32 => {
let ca = self.f32().unwrap();
ca.cumprod(reverse).into_series()
}
Float64 => {
let ca = self.f64().unwrap();
ca.cumprod(reverse).into_series()
}
dt => panic!("cumprod not supported for dtype: {dt:?}"),
}
}
#[cfg(not(feature = "cum_agg"))]
{
panic!("activate 'cum_agg' feature")
}
}
/// Get the product of an array.
///
/// If the [`DataType`] is one of `{Int8, UInt8, Int16, UInt16}` the `Series` is
/// first cast to `Int64` to prevent overflow issues.
#[cfg_attr(docsrs, doc(cfg(feature = "product")))]
pub fn product(&self) -> Series {
#[cfg(feature = "product")]
{
use DataType::*;
match self.dtype() {
Boolean => self.cast(&DataType::Int64).unwrap().product(),
Int8 | UInt8 | Int16 | UInt16 => {
let s = self.cast(&Int64).unwrap();
s.product()
}
Int64 => {
let ca = self.i64().unwrap();
ca.prod_as_series()
}
Float32 => {
let ca = self.f32().unwrap();
ca.prod_as_series()
}
Float64 => {
let ca = self.f64().unwrap();
ca.prod_as_series()
}
dt => panic!("cumprod not supported for dtype: {dt:?}"),
}
}
#[cfg(not(feature = "product"))]
{
panic!("activate 'product' feature")
}
}
#[cfg(feature = "rank")]
#[cfg_attr(docsrs, doc(cfg(feature = "rank")))]
pub fn rank(&self, options: RankOptions) -> Series {
rank(self, options.method, options.descending)
}
/// Cast throws an error if conversion had overflows
pub fn strict_cast(&self, data_type: &DataType) -> PolarsResult<Series> {
let s = self.cast(data_type)?;
if self.null_count() != s.null_count() {
let failure_mask = !self.is_null() & s.is_null();
let failures = self.filter_threaded(&failure_mask, false)?.unique()?;
Err(PolarsError::ComputeError(
format!(
"Strict conversion from {:?} to {:?} failed for values {}. \
If you were trying to cast Utf8 to Date, Time, or Datetime, \
consider using `strptime`.",
self.dtype(),
data_type,
failures.fmt_list(),
)
.into(),
))
} else {
Ok(s)
}
}
#[cfg(feature = "dtype-time")]
pub(crate) fn into_time(self) -> Series {
#[cfg(not(feature = "dtype-time"))]
{
panic!("activate feature dtype-time")
}
match self.dtype() {
DataType::Int64 => self.i64().unwrap().clone().into_time().into_series(),
DataType::Time => self
.time()
.unwrap()
.as_ref()
.clone()
.into_time()
.into_series(),
dt => panic!("date not implemented for {dt:?}"),
}
}
pub(crate) fn into_date(self) -> Series {
#[cfg(not(feature = "dtype-date"))]
{
panic!("activate feature dtype-date")
}
#[cfg(feature = "dtype-date")]
match self.dtype() {
DataType::Int32 => self.i32().unwrap().clone().into_date().into_series(),
DataType::Date => self
.date()
.unwrap()
.as_ref()
.clone()
.into_date()
.into_series(),
dt => panic!("date not implemented for {dt:?}"),
}
}
pub(crate) fn into_datetime(self, timeunit: TimeUnit, tz: Option<TimeZone>) -> Series {
#[cfg(not(feature = "dtype-datetime"))]
{
panic!("activate feature dtype-datetime")
}
#[cfg(feature = "dtype-datetime")]
match self.dtype() {
DataType::Int64 => self
.i64()
.unwrap()
.clone()
.into_datetime(timeunit, tz)
.into_series(),
DataType::Datetime(_, _) => self
.datetime()
.unwrap()
.as_ref()
.clone()
.into_datetime(timeunit, tz)
.into_series(),
dt => panic!("into_datetime not implemented for {dt:?}"),
}
}
pub(crate) fn into_duration(self, timeunit: TimeUnit) -> Series {
#[cfg(not(feature = "dtype-duration"))]
{
panic!("activate feature dtype-duration")
}
#[cfg(feature = "dtype-duration")]
match self.dtype() {
DataType::Int64 => self
.i64()
.unwrap()
.clone()
.into_duration(timeunit)
.into_series(),
DataType::Duration(_) => self
.duration()
.unwrap()
.as_ref()
.clone()
.into_duration(timeunit)
.into_series(),
dt => panic!("into_duration not implemented for {dt:?}"),
}
}
#[cfg(feature = "abs")]
#[cfg_attr(docsrs, doc(cfg(feature = "abs")))]
/// convert numerical values to their absolute value
pub fn abs(&self) -> PolarsResult<Series> {
let a = self.to_physical_repr();
use DataType::*;
let out = match a.dtype() {
#[cfg(feature = "dtype-i8")]
Int8 => a.i8().unwrap().abs().into_series(),
#[cfg(feature = "dtype-i16")]
Int16 => a.i16().unwrap().abs().into_series(),
Int32 => a.i32().unwrap().abs().into_series(),
Int64 => a.i64().unwrap().abs().into_series(),
UInt8 | UInt16 | UInt32 | UInt64 => self.clone(),
Float32 => a.f32().unwrap().abs().into_series(),
Float64 => a.f64().unwrap().abs().into_series(),
dt => {
return Err(PolarsError::InvalidOperation(
format!("abs not supported for series of type {dt:?}").into(),
));
}
};
Ok(out)
}
#[cfg(feature = "private")]
// used for formatting
pub fn str_value(&self, index: usize) -> PolarsResult<Cow<str>> {
let out = match self.0.get(index)? {
AnyValue::Utf8(s) => Cow::Borrowed(s),
AnyValue::Null => Cow::Borrowed("null"),
#[cfg(feature = "dtype-categorical")]
AnyValue::Categorical(idx, rev) => Cow::Borrowed(rev.get(idx)),
av => Cow::Owned(format!("{av}")),
};
Ok(out)
}
/// Get the head of the Series.
pub fn head(&self, length: Option<usize>) -> Series {
match length {
Some(len) => self.slice(0, std::cmp::min(len, self.len())),
None => self.slice(0, std::cmp::min(10, self.len())),
}
}
/// Get the tail of the Series.
pub fn tail(&self, length: Option<usize>) -> Series {
let len = match length {
Some(len) => std::cmp::min(len, self.len()),
None => std::cmp::min(10, self.len()),
};
self.slice(-(len as i64), len)
}
pub fn mean_as_series(&self) -> Series {
match self.dtype() {
DataType::Float32 => {
let val = &[self.mean().map(|m| m as f32)];
Series::new(self.name(), val)
}
dt if dt.is_numeric() || matches!(dt, DataType::Boolean) => {
let val = &[self.mean()];
Series::new(self.name(), val)
}
dt @ DataType::Duration(_) => {
Series::new(self.name(), &[self.mean().map(|v| v as i64)])
.cast(dt)
.unwrap()
}
_ => return Series::full_null(self.name(), 1, self.dtype()),
}
}src/frame/mod.rs (line 2506)
2502 2503 2504 2505 2506 2507 2508 2509 2510 2511 2512 2513 2514 2515 2516 2517 2518 2519 2520 2521 2522 2523 2524 2525 2526 2527 2528 2529 2530 2531 2532 2533 2534 2535 2536 2537 2538 2539 2540 2541 2542 2543 2544 2545 2546 2547 2548 2549 2550 2551 2552 2553 2554 2555 2556 2557 2558 2559 2560 2561 2562 2563 2564 2565 2566 2567 2568 2569 2570 2571 2572 2573 2574 2575 2576 2577 2578 2579 2580 2581 2582 2583 2584 2585 2586 2587 2588 2589 2590 2591 2592 2593 2594 2595 2596 2597 2598 2599 2600 2601 2602 2603 2604 2605 2606 2607 2608 2609 2610 2611 2612 2613 2614 2615 2616 2617 2618 2619 2620 2621 2622 2623 2624 2625 2626 2627 2628 2629 2630 2631 2632 2633 2634 2635 2636 2637 2638 2639 2640 2641 2642 2643 2644 2645 2646 2647 2648 2649 2650 2651 2652 2653 2654 2655 2656 2657 2658 2659 2660 2661 2662 2663 2664 2665 2666 2667 2668 2669 2670 2671 2672 2673 2674 2675 2676 2677 2678 2679 2680 2681 2682 2683 2684 2685 2686 2687 2688 2689 2690 2691 2692 2693 2694 2695 2696 2697 2698 2699 2700 2701 2702 2703 2704 2705 2706 2707 2708 2709 2710 2711 2712 2713 2714 2715 2716 2717 2718 2719 2720 2721 2722 2723 2724 2725 2726 2727 2728 2729 2730 2731 2732 2733 2734 2735 2736 2737 2738 2739 2740 2741 2742 2743 2744 2745 2746 2747 2748 2749 2750 2751 2752 2753 2754 2755 2756 2757 2758 2759 2760 2761 2762 2763 2764 2765 2766 2767 2768 2769 2770 2771 2772 2773 2774 2775 2776 2777 2778 2779 2780 2781 2782 2783 2784 2785 2786 2787 2788 2789 2790 2791 2792 2793 2794 2795 2796 2797 2798 2799 2800 2801 2802 2803 2804 2805 2806 2807 2808 2809 2810 2811 2812 2813 2814 2815 2816 2817 2818 2819 2820 2821 2822 2823 2824 2825 2826 2827 2828 2829 2830 2831 2832 2833 2834 2835 2836 2837 2838 2839 2840 2841 2842 2843 2844 2845 2846 2847 2848 2849 2850 2851 2852 2853 2854 2855 2856 2857 2858 2859 2860 2861 2862 2863 2864 2865 2866 2867 2868 2869 2870 2871 2872 2873 2874 2875 2876 2877 2878 2879 2880 2881 2882 2883 2884 2885 2886 2887 2888 2889 2890 2891 2892 2893 2894 2895 2896 2897 2898 2899 2900 2901 2902 2903 2904 2905 2906 2907 2908 2909 2910 2911 2912 2913 2914 2915 2916 2917 2918 2919 2920 2921 2922 2923 2924 2925 2926 2927 2928 2929 2930 2931 2932 2933 2934 2935 2936 2937 2938 2939 2940 2941 2942 2943 2944 2945 2946 2947
fn describe_cast(df: &DataFrame) -> DataFrame {
let mut columns: Vec<Series> = vec![];
for s in df.columns.iter() {
columns.push(s.cast(&DataType::Float64).expect("cast to float failed"));
}
DataFrame::new(columns).unwrap()
}
fn count(df: &DataFrame) -> DataFrame {
let columns = df.apply_columns_par(&|s| Series::new(s.name(), [s.len() as IdxSize]));
DataFrame::new_no_checks(columns)
}
let percentiles = percentiles.unwrap_or(&[0.25, 0.5, 0.75]);
let mut headers: Vec<String> = vec![
"count".to_string(),
"mean".to_string(),
"std".to_string(),
"min".to_string(),
];
let mut tmp: Vec<DataFrame> = vec![
describe_cast(&count(self)),
describe_cast(&self.mean()),
describe_cast(&self.std(1)),
describe_cast(&self.min()),
];
for p in percentiles {
tmp.push(describe_cast(
&self
.quantile(*p, QuantileInterpolOptions::Linear)
.expect("quantile failed"),
));
headers.push(format!("{}%", *p * 100.0));
}
// Keep order same as pandas
tmp.push(describe_cast(&self.max()));
headers.push("max".to_string());
let mut summary = concat_df_unchecked(&tmp);
summary
.insert_at_idx(0, Series::new("describe", headers))
.expect("insert of header failed");
summary
}
/// Aggregate the columns to their maximum values.
///
/// # Example
///
/// ```no_run
/// # use polars_core::prelude::*;
/// let df1: DataFrame = df!("Die n°1" => &[1, 3, 1, 5, 6],
/// "Die n°2" => &[3, 2, 3, 5, 3])?;
/// assert_eq!(df1.shape(), (5, 2));
///
/// let df2: DataFrame = df1.max();
/// assert_eq!(df2.shape(), (1, 2));
/// println!("{}", df2);
/// # Ok::<(), PolarsError>(())
/// ```
///
/// Output:
///
/// ```text
/// shape: (1, 2)
/// +---------+---------+
/// | Die n°1 | Die n°2 |
/// | --- | --- |
/// | i32 | i32 |
/// +=========+=========+
/// | 6 | 5 |
/// +---------+---------+
/// ```
#[must_use]
pub fn max(&self) -> Self {
let columns = self.apply_columns_par(&|s| s.max_as_series());
DataFrame::new_no_checks(columns)
}
/// Aggregate the columns to their standard deviation values.
///
/// # Example
///
/// ```no_run
/// # use polars_core::prelude::*;
/// let df1: DataFrame = df!("Die n°1" => &[1, 3, 1, 5, 6],
/// "Die n°2" => &[3, 2, 3, 5, 3])?;
/// assert_eq!(df1.shape(), (5, 2));
///
/// let df2: DataFrame = df1.std(1);
/// assert_eq!(df2.shape(), (1, 2));
/// println!("{}", df2);
/// # Ok::<(), PolarsError>(())
/// ```
///
/// Output:
///
/// ```text
/// shape: (1, 2)
/// +-------------------+--------------------+
/// | Die n°1 | Die n°2 |
/// | --- | --- |
/// | f64 | f64 |
/// +===================+====================+
/// | 2.280350850198276 | 1.0954451150103321 |
/// +-------------------+--------------------+
/// ```
#[must_use]
pub fn std(&self, ddof: u8) -> Self {
let columns = self.apply_columns_par(&|s| s.std_as_series(ddof));
DataFrame::new_no_checks(columns)
}
/// Aggregate the columns to their variation values.
///
/// # Example
///
/// ```no_run
/// # use polars_core::prelude::*;
/// let df1: DataFrame = df!("Die n°1" => &[1, 3, 1, 5, 6],
/// "Die n°2" => &[3, 2, 3, 5, 3])?;
/// assert_eq!(df1.shape(), (5, 2));
///
/// let df2: DataFrame = df1.var(1);
/// assert_eq!(df2.shape(), (1, 2));
/// println!("{}", df2);
/// # Ok::<(), PolarsError>(())
/// ```
///
/// Output:
///
/// ```text
/// shape: (1, 2)
/// +---------+---------+
/// | Die n°1 | Die n°2 |
/// | --- | --- |
/// | f64 | f64 |
/// +=========+=========+
/// | 5.2 | 1.2 |
/// +---------+---------+
/// ```
#[must_use]
pub fn var(&self, ddof: u8) -> Self {
let columns = self.apply_columns_par(&|s| s.var_as_series(ddof));
DataFrame::new_no_checks(columns)
}
/// Aggregate the columns to their minimum values.
///
/// # Example
///
/// ```no_run
/// # use polars_core::prelude::*;
/// let df1: DataFrame = df!("Die n°1" => &[1, 3, 1, 5, 6],
/// "Die n°2" => &[3, 2, 3, 5, 3])?;
/// assert_eq!(df1.shape(), (5, 2));
///
/// let df2: DataFrame = df1.min();
/// assert_eq!(df2.shape(), (1, 2));
/// println!("{}", df2);
/// # Ok::<(), PolarsError>(())
/// ```
///
/// Output:
///
/// ```text
/// shape: (1, 2)
/// +---------+---------+
/// | Die n°1 | Die n°2 |
/// | --- | --- |
/// | i32 | i32 |
/// +=========+=========+
/// | 1 | 2 |
/// +---------+---------+
/// ```
#[must_use]
pub fn min(&self) -> Self {
let columns = self.apply_columns_par(&|s| s.min_as_series());
DataFrame::new_no_checks(columns)
}
/// Aggregate the columns to their sum values.
///
/// # Example
///
/// ```no_run
/// # use polars_core::prelude::*;
/// let df1: DataFrame = df!("Die n°1" => &[1, 3, 1, 5, 6],
/// "Die n°2" => &[3, 2, 3, 5, 3])?;
/// assert_eq!(df1.shape(), (5, 2));
///
/// let df2: DataFrame = df1.sum();
/// assert_eq!(df2.shape(), (1, 2));
/// println!("{}", df2);
/// # Ok::<(), PolarsError>(())
/// ```
///
/// Output:
///
/// ```text
/// shape: (1, 2)
/// +---------+---------+
/// | Die n°1 | Die n°2 |
/// | --- | --- |
/// | i32 | i32 |
/// +=========+=========+
/// | 16 | 16 |
/// +---------+---------+
/// ```
#[must_use]
pub fn sum(&self) -> Self {
let columns = self.apply_columns_par(&|s| s.sum_as_series());
DataFrame::new_no_checks(columns)
}
/// Aggregate the columns to their mean values.
///
/// # Example
///
/// ```no_run
/// # use polars_core::prelude::*;
/// let df1: DataFrame = df!("Die n°1" => &[1, 3, 1, 5, 6],
/// "Die n°2" => &[3, 2, 3, 5, 3])?;
/// assert_eq!(df1.shape(), (5, 2));
///
/// let df2: DataFrame = df1.mean();
/// assert_eq!(df2.shape(), (1, 2));
/// println!("{}", df2);
/// # Ok::<(), PolarsError>(())
/// ```
///
/// Output:
///
/// ```text
/// shape: (1, 2)
/// +---------+---------+
/// | Die n°1 | Die n°2 |
/// | --- | --- |
/// | f64 | f64 |
/// +=========+=========+
/// | 3.2 | 3.2 |
/// +---------+---------+
/// ```
#[must_use]
pub fn mean(&self) -> Self {
let columns = self.apply_columns_par(&|s| s.mean_as_series());
DataFrame::new_no_checks(columns)
}
/// Aggregate the columns to their median values.
///
/// # Example
///
/// ```no_run
/// # use polars_core::prelude::*;
/// let df1: DataFrame = df!("Die n°1" => &[1, 3, 1, 5, 6],
/// "Die n°2" => &[3, 2, 3, 5, 3])?;
/// assert_eq!(df1.shape(), (5, 2));
///
/// let df2: DataFrame = df1.median();
/// assert_eq!(df2.shape(), (1, 2));
/// println!("{}", df2);
/// # Ok::<(), PolarsError>(())
/// ```
///
/// Output:
///
/// ```text
/// shape: (1, 2)
/// +---------+---------+
/// | Die n°1 | Die n°2 |
/// | --- | --- |
/// | i32 | i32 |
/// +=========+=========+
/// | 3 | 3 |
/// +---------+---------+
/// ```
#[must_use]
pub fn median(&self) -> Self {
let columns = self.apply_columns_par(&|s| s.median_as_series());
DataFrame::new_no_checks(columns)
}
/// Aggregate the columns to their quantile values.
pub fn quantile(&self, quantile: f64, interpol: QuantileInterpolOptions) -> PolarsResult<Self> {
let columns = self.try_apply_columns_par(&|s| s.quantile_as_series(quantile, interpol))?;
Ok(DataFrame::new_no_checks(columns))
}
/// Aggregate the column horizontally to their min values.
#[cfg(feature = "zip_with")]
#[cfg_attr(docsrs, doc(cfg(feature = "zip_with")))]
pub fn hmin(&self) -> PolarsResult<Option<Series>> {
let min_fn = |acc: &Series, s: &Series| {
let mask = acc.lt(s)? & acc.is_not_null() | s.is_null();
acc.zip_with(&mask, s)
};
match self.columns.len() {
0 => Ok(None),
1 => Ok(Some(self.columns[0].clone())),
2 => min_fn(&self.columns[0], &self.columns[1]).map(Some),
_ => {
// the try_reduce_with is a bit slower in parallelism,
// but I don't think it matters here as we parallelize over columns, not over elements
POOL.install(|| {
self.columns
.par_iter()
.map(|s| Ok(Cow::Borrowed(s)))
.try_reduce_with(|l, r| min_fn(&l, &r).map(Cow::Owned))
// we can unwrap the option, because we are certain there is a column
// we started this operation on 3 columns
.unwrap()
.map(|cow| Some(cow.into_owned()))
})
}
}
}
/// Aggregate the column horizontally to their max values.
#[cfg(feature = "zip_with")]
#[cfg_attr(docsrs, doc(cfg(feature = "zip_with")))]
pub fn hmax(&self) -> PolarsResult<Option<Series>> {
let max_fn = |acc: &Series, s: &Series| {
let mask = acc.gt(s)? & acc.is_not_null() | s.is_null();
acc.zip_with(&mask, s)
};
match self.columns.len() {
0 => Ok(None),
1 => Ok(Some(self.columns[0].clone())),
2 => max_fn(&self.columns[0], &self.columns[1]).map(Some),
_ => {
// the try_reduce_with is a bit slower in parallelism,
// but I don't think it matters here as we parallelize over columns, not over elements
POOL.install(|| {
self.columns
.par_iter()
.map(|s| Ok(Cow::Borrowed(s)))
.try_reduce_with(|l, r| max_fn(&l, &r).map(Cow::Owned))
// we can unwrap the option, because we are certain there is a column
// we started this operation on 3 columns
.unwrap()
.map(|cow| Some(cow.into_owned()))
})
}
}
}
/// Aggregate the column horizontally to their sum values.
pub fn hsum(&self, none_strategy: NullStrategy) -> PolarsResult<Option<Series>> {
let sum_fn =
|acc: &Series, s: &Series, none_strategy: NullStrategy| -> PolarsResult<Series> {
let mut acc = acc.clone();
let mut s = s.clone();
if let NullStrategy::Ignore = none_strategy {
// if has nulls
if acc.has_validity() {
acc = acc.fill_null(FillNullStrategy::Zero)?;
}
if s.has_validity() {
s = s.fill_null(FillNullStrategy::Zero)?;
}
}
Ok(&acc + &s)
};
match self.columns.len() {
0 => Ok(None),
1 => Ok(Some(self.columns[0].clone())),
2 => sum_fn(&self.columns[0], &self.columns[1], none_strategy).map(Some),
_ => {
// the try_reduce_with is a bit slower in parallelism,
// but I don't think it matters here as we parallelize over columns, not over elements
POOL.install(|| {
self.columns
.par_iter()
.map(|s| Ok(Cow::Borrowed(s)))
.try_reduce_with(|l, r| sum_fn(&l, &r, none_strategy).map(Cow::Owned))
// we can unwrap the option, because we are certain there is a column
// we started this operation on 3 columns
.unwrap()
.map(|cow| Some(cow.into_owned()))
})
}
}
}
/// Aggregate the column horizontally to their mean values.
pub fn hmean(&self, none_strategy: NullStrategy) -> PolarsResult<Option<Series>> {
match self.columns.len() {
0 => Ok(None),
1 => Ok(Some(self.columns[0].clone())),
_ => {
let columns = self
.columns
.iter()
.cloned()
.filter(|s| {
let dtype = s.dtype();
dtype.is_numeric() || matches!(dtype, DataType::Boolean)
})
.collect();
let numeric_df = DataFrame::new_no_checks(columns);
let sum = || numeric_df.hsum(none_strategy);
let null_count = || {
numeric_df
.columns
.par_iter()
.map(|s| s.is_null().cast(&DataType::UInt32).unwrap())
.reduce_with(|l, r| &l + &r)
// we can unwrap the option, because we are certain there is a column
// we started this operation on 2 columns
.unwrap()
};
let (sum, null_count) = POOL.install(|| rayon::join(sum, null_count));
let sum = sum?;
// value lengths: len - null_count
let value_length: UInt32Chunked =
(numeric_df.width().sub(&null_count)).u32().unwrap().clone();
// make sure that we do not divide by zero
// by replacing with None
let value_length = value_length
.set(&value_length.equal(0), None)?
.into_series()
.cast(&DataType::Float64)?;
Ok(sum.map(|sum| &sum / &value_length))
}
}
}src/series/implementations/boolean.rs (line 345)
338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369
fn median_as_series(&self) -> Series {
// first convert array to f32 as that's cheaper
// finally the single value to f64
self.0
.cast(&DataType::Float32)
.unwrap()
.median_as_series()
.cast(&DataType::Float64)
.unwrap()
}
/// Get the variance of the Series as a new Series of length 1.
fn var_as_series(&self, _ddof: u8) -> Series {
// first convert array to f32 as that's cheaper
// finally the single value to f64
self.0
.cast(&DataType::Float32)
.unwrap()
.var_as_series(_ddof)
.cast(&DataType::Float64)
.unwrap()
}
/// Get the standard deviation of the Series as a new Series of length 1.
fn std_as_series(&self, _ddof: u8) -> Series {
// first convert array to f32 as that's cheaper
// finally the single value to f64
self.0
.cast(&DataType::Float32)
.unwrap()
.std_as_series(_ddof)
.cast(&DataType::Float64)
.unwrap()
}src/frame/groupby/mod.rs (line 31)
26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47
fn prepare_dataframe_unsorted(by: &[Series]) -> DataFrame {
DataFrame::new_no_checks(
by.iter()
.map(|s| match s.dtype() {
#[cfg(feature = "dtype-categorical")]
DataType::Categorical(_) => s.cast(&DataType::UInt32).unwrap(),
_ => {
if s.dtype().to_physical().is_numeric() {
let s = s.to_physical_repr();
if s.bit_repr_is_large() {
s.bit_repr_large().into_series()
} else {
s.bit_repr_small().into_series()
}
} else {
s.clone()
}
}
})
.collect(),
)
}Additional examples can be found in:
- src/series/ops/extend.rs
- src/series/any_value.rs
- src/series/ops/moment.rs
- src/chunked_array/ops/sort/mod.rs
- src/frame/arithmetic.rs
- src/chunked_array/cast.rs
- src/chunked_array/ndarray.rs
- src/functions.rs
- src/frame/row.rs
- src/frame/hash_join/mod.rs
- src/frame/asof_join/mod.rs
- src/frame/explode.rs
- src/series/arithmetic/borrowed.rs
- src/chunked_array/ops/is_in.rs
- src/chunked_array/ops/any_value.rs
- src/chunked_array/ops/rolling_window.rs
sourcepub fn sum<T>(&self) -> Option<T>where
T: NumCast,
pub fn sum<T>(&self) -> Option<T>where
T: NumCast,
Compute the sum of all values in this Series.
Returns Some(0) if the array is empty, and None if the array only
contains null values.
If the DataType is one of {Int8, UInt8, Int16, UInt16} the Series is
first cast to Int64 to prevent overflow issues.
let s = Series::new("days", &[1, 2, 3]);
assert_eq!(s.sum(), Some(6));sourcepub fn min<T>(&self) -> Option<T>where
T: NumCast,
pub fn min<T>(&self) -> Option<T>where
T: NumCast,
Returns the minimum value in the array, according to the natural order. Returns an option because the array is nullable.
let s = Series::new("days", [1, 2, 3].as_ref());
assert_eq!(s.min(), Some(1));sourcepub fn max<T>(&self) -> Option<T>where
T: NumCast,
pub fn max<T>(&self) -> Option<T>where
T: NumCast,
Returns the maximum value in the array, according to the natural order. Returns an option because the array is nullable.
let s = Series::new("days", [1, 2, 3].as_ref());
assert_eq!(s.max(), Some(3));sourcepub fn explode(&self) -> PolarsResult<Series>
pub fn explode(&self) -> PolarsResult<Series>
Explode a list or utf8 Series. This expands every item to a new row..
Examples found in repository?
src/frame/explode.rs (line 35)
31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81
pub fn explode_impl(&self, mut columns: Vec<Series>) -> PolarsResult<DataFrame> {
let mut df = self.clone();
if self.height() == 0 {
for s in &columns {
df.with_column(s.explode()?)?;
}
return Ok(df);
}
columns.sort_by(|sa, sb| {
self.check_name_to_idx(sa.name())
.expect("checked above")
.partial_cmp(&self.check_name_to_idx(sb.name()).expect("checked above"))
.expect("cmp usize -> Ordering")
});
// first remove all the exploded columns
for s in &columns {
df = df.drop(s.name())?;
}
for (i, s) in columns.iter().enumerate() {
// Safety:
// offsets don't have indices exceeding Series length.
if let Ok((exploded, offsets)) = get_exploded(s) {
let col_idx = self.check_name_to_idx(s.name())?;
// expand all the other columns based the exploded first column
if i == 0 {
let row_idx = offsets_to_indexes(offsets.as_slice(), exploded.len());
let mut row_idx = IdxCa::from_vec("", row_idx);
row_idx.set_sorted(false);
// Safety
// We just created indices that are in bounds.
df = unsafe { df.take_unchecked(&row_idx) };
}
if exploded.len() == df.height() || df.width() == 0 {
df.columns.insert(col_idx, exploded);
} else {
return Err(PolarsError::ShapeMisMatch(
format!("The exploded column(s) don't have the same length. Length DataFrame: {}. Length exploded column {}: {}", df.height(), exploded.name(), exploded.len()).into(),
));
}
} else {
return Err(PolarsError::InvalidOperation(
format!("cannot explode dtype: {:?}", s.dtype()).into(),
));
}
}
Ok(df)
}More examples
src/series/ops/to_list.rs (line 67)
62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134
pub fn reshape(&self, dims: &[i64]) -> PolarsResult<Series> {
if dims.is_empty() {
panic!("dimensions cannot be empty")
}
let s = if let DataType::List(_) = self.dtype() {
Cow::Owned(self.explode()?)
} else {
Cow::Borrowed(self)
};
// no rows
if dims[0] == 0 {
let s = reshape_fast_path(self.name(), &s);
return Ok(s);
}
let s_ref = s.as_ref();
let mut dims = dims.to_vec();
if let Some(idx) = dims.iter().position(|i| *i == -1) {
let mut product = 1;
for (cnt, dim) in dims.iter().enumerate() {
if cnt != idx {
product *= *dim
}
}
dims[idx] = s_ref.len() as i64 / product;
}
let prod = dims.iter().product::<i64>() as usize;
if prod != s_ref.len() {
return Err(PolarsError::ComputeError(
format!("cannot reshape len {} into shape {:?}", s_ref.len(), dims).into(),
));
}
match dims.len() {
1 => Ok(s_ref.slice(0, dims[0] as usize)),
2 => {
let mut rows = dims[0];
let mut cols = dims[1];
// infer dimension
if rows == -1 {
rows = cols / s_ref.len() as i64
}
if cols == -1 {
cols = rows / s_ref.len() as i64
}
// fast path, we can create a unit list so we only allocate offsets
if rows as usize == s_ref.len() && cols == 1 {
let s = reshape_fast_path(self.name(), s_ref);
return Ok(s);
}
let mut builder =
get_list_builder(s_ref.dtype(), s_ref.len(), rows as usize, self.name())?;
let mut offset = 0i64;
for _ in 0..rows {
let row = s_ref.slice(offset, cols as usize);
builder.append_series(&row);
offset += cols;
}
Ok(builder.finish().into_series())
}
_ => {
panic!("more than two dimensions not yet supported");
}
}
}sourcepub fn is_nan(&self) -> PolarsResult<BooleanChunked>
pub fn is_nan(&self) -> PolarsResult<BooleanChunked>
Check if float value is NaN (note this is different than missing/ null)
sourcepub fn is_not_nan(&self) -> PolarsResult<BooleanChunked>
pub fn is_not_nan(&self) -> PolarsResult<BooleanChunked>
Check if float value is NaN (note this is different than missing/ null)
sourcepub fn is_finite(&self) -> PolarsResult<BooleanChunked>
pub fn is_finite(&self) -> PolarsResult<BooleanChunked>
Check if float value is finite
sourcepub fn is_infinite(&self) -> PolarsResult<BooleanChunked>
pub fn is_infinite(&self) -> PolarsResult<BooleanChunked>
Check if float value is infinite
sourcepub fn zip_with(
&self,
mask: &BooleanChunked,
other: &Series
) -> PolarsResult<Series>
Available on crate feature zip_with only.
pub fn zip_with(
&self,
mask: &BooleanChunked,
other: &Series
) -> PolarsResult<Series>
zip_with only.Create a new ChunkedArray with values from self where the mask evaluates true and values
from other where the mask evaluates false
Examples found in repository?
src/frame/mod.rs (line 2807)
2804 2805 2806 2807 2808 2809 2810 2811 2812 2813 2814 2815 2816 2817 2818 2819 2820 2821 2822 2823 2824 2825 2826 2827 2828 2829 2830 2831 2832 2833 2834 2835 2836 2837 2838 2839 2840 2841 2842 2843 2844 2845 2846 2847 2848 2849 2850 2851 2852 2853 2854 2855 2856 2857 2858 2859
pub fn hmin(&self) -> PolarsResult<Option<Series>> {
let min_fn = |acc: &Series, s: &Series| {
let mask = acc.lt(s)? & acc.is_not_null() | s.is_null();
acc.zip_with(&mask, s)
};
match self.columns.len() {
0 => Ok(None),
1 => Ok(Some(self.columns[0].clone())),
2 => min_fn(&self.columns[0], &self.columns[1]).map(Some),
_ => {
// the try_reduce_with is a bit slower in parallelism,
// but I don't think it matters here as we parallelize over columns, not over elements
POOL.install(|| {
self.columns
.par_iter()
.map(|s| Ok(Cow::Borrowed(s)))
.try_reduce_with(|l, r| min_fn(&l, &r).map(Cow::Owned))
// we can unwrap the option, because we are certain there is a column
// we started this operation on 3 columns
.unwrap()
.map(|cow| Some(cow.into_owned()))
})
}
}
}
/// Aggregate the column horizontally to their max values.
#[cfg(feature = "zip_with")]
#[cfg_attr(docsrs, doc(cfg(feature = "zip_with")))]
pub fn hmax(&self) -> PolarsResult<Option<Series>> {
let max_fn = |acc: &Series, s: &Series| {
let mask = acc.gt(s)? & acc.is_not_null() | s.is_null();
acc.zip_with(&mask, s)
};
match self.columns.len() {
0 => Ok(None),
1 => Ok(Some(self.columns[0].clone())),
2 => max_fn(&self.columns[0], &self.columns[1]).map(Some),
_ => {
// the try_reduce_with is a bit slower in parallelism,
// but I don't think it matters here as we parallelize over columns, not over elements
POOL.install(|| {
self.columns
.par_iter()
.map(|s| Ok(Cow::Borrowed(s)))
.try_reduce_with(|l, r| max_fn(&l, &r).map(Cow::Owned))
// we can unwrap the option, because we are certain there is a column
// we started this operation on 3 columns
.unwrap()
.map(|cow| Some(cow.into_owned()))
})
}
}
}sourcepub fn to_physical_repr(&self) -> Cow<'_, Series>
pub fn to_physical_repr(&self) -> Cow<'_, Series>
Cast a datelike Series to their physical representation. Primitives remain unchanged
- Date -> Int32
- Datetime-> Int64
- Time -> Int64
- Categorical -> UInt32
Examples found in repository?
More examples
src/series/comparison.rs (line 263)
261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295
fn equal(&self, rhs: Rhs) -> PolarsResult<BooleanChunked> {
validate_types(self.dtype(), &DataType::Int8)?;
let s = self.to_physical_repr();
Ok(apply_method_physical_numeric!(&s, equal, rhs))
}
fn not_equal(&self, rhs: Rhs) -> PolarsResult<BooleanChunked> {
validate_types(self.dtype(), &DataType::Int8)?;
let s = self.to_physical_repr();
Ok(apply_method_physical_numeric!(&s, not_equal, rhs))
}
fn gt(&self, rhs: Rhs) -> PolarsResult<BooleanChunked> {
validate_types(self.dtype(), &DataType::Int8)?;
let s = self.to_physical_repr();
Ok(apply_method_physical_numeric!(&s, gt, rhs))
}
fn gt_eq(&self, rhs: Rhs) -> PolarsResult<BooleanChunked> {
validate_types(self.dtype(), &DataType::Int8)?;
let s = self.to_physical_repr();
Ok(apply_method_physical_numeric!(&s, gt_eq, rhs))
}
fn lt(&self, rhs: Rhs) -> PolarsResult<BooleanChunked> {
validate_types(self.dtype(), &DataType::Int8)?;
let s = self.to_physical_repr();
Ok(apply_method_physical_numeric!(&s, lt, rhs))
}
fn lt_eq(&self, rhs: Rhs) -> PolarsResult<BooleanChunked> {
validate_types(self.dtype(), &DataType::Int8)?;
let s = self.to_physical_repr();
Ok(apply_method_physical_numeric!(&s, lt_eq, rhs))
}src/utils/series.rs (line 9)
6 7 8 9 10 11 12 13 14 15 16 17 18 19
pub fn _to_physical_and_bit_repr(s: &[Series]) -> Vec<Series> {
s.iter()
.map(|s| {
let physical = s.to_physical_repr();
match physical.dtype() {
DataType::Int64 => physical.bit_repr_large().into_series(),
DataType::Int32 => physical.bit_repr_small().into_series(),
DataType::Float32 => physical.bit_repr_small().into_series(),
DataType::Float64 => physical.bit_repr_large().into_series(),
_ => physical.into_owned(),
}
})
.collect()
}src/chunked_array/builder/list.rs (line 159)
155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174
fn append_series(&mut self, s: &Series) {
if s.is_empty() {
self.fast_explode = false;
}
let physical = s.to_physical_repr();
let ca = physical.unpack::<T>().unwrap();
let values = self.builder.mut_values();
ca.downcast_iter().for_each(|arr| {
if !arr.has_validity() {
values.extend_from_slice(arr.values().as_slice())
} else {
// Safety:
// Arrow arrays are trusted length iterators.
unsafe { values.extend_trusted_len_unchecked(arr.into_iter()) }
}
});
// overflow of i64 is far beyond polars capable lengths.
unsafe { self.builder.try_push_valid().unwrap_unchecked() };
}src/frame/groupby/mod.rs (line 34)
26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47
fn prepare_dataframe_unsorted(by: &[Series]) -> DataFrame {
DataFrame::new_no_checks(
by.iter()
.map(|s| match s.dtype() {
#[cfg(feature = "dtype-categorical")]
DataType::Categorical(_) => s.cast(&DataType::UInt32).unwrap(),
_ => {
if s.dtype().to_physical().is_numeric() {
let s = s.to_physical_repr();
if s.bit_repr_is_large() {
s.bit_repr_large().into_series()
} else {
s.bit_repr_small().into_series()
}
} else {
s.clone()
}
}
})
.collect(),
)
}src/series/mod.rs (line 803)
802 803 804 805 806 807 808 809 810 811 812 813 814 815 816 817 818 819 820 821 822
pub fn abs(&self) -> PolarsResult<Series> {
let a = self.to_physical_repr();
use DataType::*;
let out = match a.dtype() {
#[cfg(feature = "dtype-i8")]
Int8 => a.i8().unwrap().abs().into_series(),
#[cfg(feature = "dtype-i16")]
Int16 => a.i16().unwrap().abs().into_series(),
Int32 => a.i32().unwrap().abs().into_series(),
Int64 => a.i64().unwrap().abs().into_series(),
UInt8 | UInt16 | UInt32 | UInt64 => self.clone(),
Float32 => a.f32().unwrap().abs().into_series(),
Float64 => a.f64().unwrap().abs().into_series(),
dt => {
return Err(PolarsError::InvalidOperation(
format!("abs not supported for series of type {dt:?}").into(),
));
}
};
Ok(out)
}sourcepub unsafe fn take_unchecked_threaded(
&self,
idx: &IdxCa,
rechunk: bool
) -> PolarsResult<Series>
pub unsafe fn take_unchecked_threaded(
&self,
idx: &IdxCa,
rechunk: bool
) -> PolarsResult<Series>
Take by index if ChunkedArray contains a single chunk.
Safety
This doesn’t check any bounds. Null validity is checked.
Examples found in repository?
src/frame/mod.rs (line 1746)
1742 1743 1744 1745 1746 1747 1748 1749 1750 1751 1752 1753 1754 1755 1756 1757
unsafe fn take_unchecked_impl(&self, idx: &IdxCa, allow_threads: bool) -> Self {
let cols = if allow_threads {
POOL.install(|| {
self.apply_columns_par(&|s| match s.dtype() {
DataType::Utf8 => s.take_unchecked_threaded(idx, true).unwrap(),
_ => s.take_unchecked(idx).unwrap(),
})
})
} else {
self.columns
.iter()
.map(|s| s.take_unchecked(idx).unwrap())
.collect()
};
DataFrame::new_no_checks(cols)
}sourcepub fn take_threaded(&self, idx: &IdxCa, rechunk: bool) -> PolarsResult<Series>
pub fn take_threaded(&self, idx: &IdxCa, rechunk: bool) -> PolarsResult<Series>
Take by index. This operation is clone.
Notes
Out of bounds access doesn’t Error but will return a Null value
Examples found in repository?
src/frame/mod.rs (line 1730)
1722 1723 1724 1725 1726 1727 1728 1729 1730 1731 1732 1733 1734 1735 1736
pub fn take(&self, indices: &IdxCa) -> PolarsResult<Self> {
let indices = if indices.chunks.len() > 1 {
Cow::Owned(indices.rechunk())
} else {
Cow::Borrowed(indices)
};
let new_col = POOL.install(|| {
self.try_apply_columns_par(&|s| match s.dtype() {
DataType::Utf8 => s.take_threaded(&indices, true),
_ => s.take(&indices),
})
})?;
Ok(DataFrame::new_no_checks(new_col))
}sourcepub fn filter_threaded(
&self,
filter: &BooleanChunked,
rechunk: bool
) -> PolarsResult<Series>
pub fn filter_threaded(
&self,
filter: &BooleanChunked,
rechunk: bool
) -> PolarsResult<Series>
Filter by boolean mask. This operation clones data.
Examples found in repository?
More examples
src/series/mod.rs (line 696)
692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711
pub fn strict_cast(&self, data_type: &DataType) -> PolarsResult<Series> {
let s = self.cast(data_type)?;
if self.null_count() != s.null_count() {
let failure_mask = !self.is_null() & s.is_null();
let failures = self.filter_threaded(&failure_mask, false)?.unique()?;
Err(PolarsError::ComputeError(
format!(
"Strict conversion from {:?} to {:?} failed for values {}. \
If you were trying to cast Utf8 to Date, Time, or Datetime, \
consider using `strptime`.",
self.dtype(),
data_type,
failures.fmt_list(),
)
.into(),
))
} else {
Ok(s)
}
}pub fn dot(&self, other: &Series) -> Option<f64>
Available on crate feature
dot_product only.sourcepub fn sum_as_series(&self) -> Series
pub fn sum_as_series(&self) -> Series
Get the sum of the Series as a new Series of length 1. Returns a Series with a single zeroed entry if self is an empty numeric series.
If the DataType is one of {Int8, UInt8, Int16, UInt16} the Series is
first cast to Int64 to prevent overflow issues.
Examples found in repository?
More examples
src/series/mod.rs (line 250)
246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531
pub fn sum<T>(&self) -> Option<T>
where
T: NumCast,
{
self.sum_as_series()
.cast(&DataType::Float64)
.ok()
.and_then(|s| s.f64().unwrap().get(0).and_then(T::from))
}
/// Returns the minimum value in the array, according to the natural order.
/// Returns an option because the array is nullable.
/// ```
/// # use polars_core::prelude::*;
/// let s = Series::new("days", [1, 2, 3].as_ref());
/// assert_eq!(s.min(), Some(1));
/// ```
pub fn min<T>(&self) -> Option<T>
where
T: NumCast,
{
self.min_as_series()
.cast(&DataType::Float64)
.ok()
.and_then(|s| s.f64().unwrap().get(0).and_then(T::from))
}
/// Returns the maximum value in the array, according to the natural order.
/// Returns an option because the array is nullable.
/// ```
/// # use polars_core::prelude::*;
/// let s = Series::new("days", [1, 2, 3].as_ref());
/// assert_eq!(s.max(), Some(3));
/// ```
pub fn max<T>(&self) -> Option<T>
where
T: NumCast,
{
self.max_as_series()
.cast(&DataType::Float64)
.ok()
.and_then(|s| s.f64().unwrap().get(0).and_then(T::from))
}
/// Explode a list or utf8 Series. This expands every item to a new row..
pub fn explode(&self) -> PolarsResult<Series> {
match self.dtype() {
DataType::List(_) => self.list().unwrap().explode(),
DataType::Utf8 => self.utf8().unwrap().explode(),
_ => Err(PolarsError::InvalidOperation(
format!(
"explode not supported for Series with dtype {:?}",
self.dtype()
)
.into(),
)),
}
}
/// Check if float value is NaN (note this is different than missing/ null)
pub fn is_nan(&self) -> PolarsResult<BooleanChunked> {
match self.dtype() {
DataType::Float32 => Ok(self.f32().unwrap().is_nan()),
DataType::Float64 => Ok(self.f64().unwrap().is_nan()),
_ => Err(PolarsError::InvalidOperation(
format!(
"'is_nan' not supported for series with dtype {:?}",
self.dtype()
)
.into(),
)),
}
}
/// Check if float value is NaN (note this is different than missing/ null)
pub fn is_not_nan(&self) -> PolarsResult<BooleanChunked> {
match self.dtype() {
DataType::Float32 => Ok(self.f32().unwrap().is_not_nan()),
DataType::Float64 => Ok(self.f64().unwrap().is_not_nan()),
_ => Err(PolarsError::InvalidOperation(
format!(
"'is_not_nan' not supported for series with dtype {:?}",
self.dtype()
)
.into(),
)),
}
}
/// Check if float value is finite
pub fn is_finite(&self) -> PolarsResult<BooleanChunked> {
match self.dtype() {
DataType::Float32 => Ok(self.f32().unwrap().is_finite()),
DataType::Float64 => Ok(self.f64().unwrap().is_finite()),
_ => Err(PolarsError::InvalidOperation(
format!(
"'is_finite' not supported for series with dtype {:?}",
self.dtype()
)
.into(),
)),
}
}
/// Check if float value is infinite
pub fn is_infinite(&self) -> PolarsResult<BooleanChunked> {
match self.dtype() {
DataType::Float32 => Ok(self.f32().unwrap().is_infinite()),
DataType::Float64 => Ok(self.f64().unwrap().is_infinite()),
_ => Err(PolarsError::InvalidOperation(
format!(
"'is_infinite' not supported for series with dtype {:?}",
self.dtype()
)
.into(),
)),
}
}
/// Create a new ChunkedArray with values from self where the mask evaluates `true` and values
/// from `other` where the mask evaluates `false`
#[cfg(feature = "zip_with")]
#[cfg_attr(docsrs, doc(cfg(feature = "zip_with")))]
pub fn zip_with(&self, mask: &BooleanChunked, other: &Series) -> PolarsResult<Series> {
let (lhs, rhs) = coerce_lhs_rhs(self, other)?;
lhs.zip_with_same_type(mask, rhs.as_ref())
}
/// Cast a datelike Series to their physical representation.
/// Primitives remain unchanged
///
/// * Date -> Int32
/// * Datetime-> Int64
/// * Time -> Int64
/// * Categorical -> UInt32
///
pub fn to_physical_repr(&self) -> Cow<Series> {
use DataType::*;
match self.dtype() {
Date => Cow::Owned(self.cast(&DataType::Int32).unwrap()),
Datetime(_, _) | Duration(_) | Time => Cow::Owned(self.cast(&DataType::Int64).unwrap()),
#[cfg(feature = "dtype-categorical")]
Categorical(_) => Cow::Owned(self.cast(&DataType::UInt32).unwrap()),
_ => Cow::Borrowed(self),
}
}
fn finish_take_threaded(&self, s: Vec<Series>, rechunk: bool) -> Series {
let s = s
.into_iter()
.reduce(|mut s, s1| {
s.append(&s1).unwrap();
s
})
.unwrap();
if rechunk {
s.rechunk()
} else {
s
}
}
// take a function pointer to reduce bloat
fn threaded_op(
&self,
rechunk: bool,
len: usize,
func: &(dyn Fn(usize, usize) -> PolarsResult<Series> + Send + Sync),
) -> PolarsResult<Series> {
let n_threads = POOL.current_num_threads();
let offsets = _split_offsets(len, n_threads);
let series: PolarsResult<Vec<_>> = POOL.install(|| {
offsets
.into_par_iter()
.map(|(offset, len)| func(offset, len))
.collect()
});
Ok(self.finish_take_threaded(series?, rechunk))
}
/// Take by index if ChunkedArray contains a single chunk.
///
/// # Safety
/// This doesn't check any bounds. Null validity is checked.
pub unsafe fn take_unchecked_threaded(
&self,
idx: &IdxCa,
rechunk: bool,
) -> PolarsResult<Series> {
self.threaded_op(rechunk, idx.len(), &|offset, len| {
let idx = idx.slice(offset as i64, len);
self.take_unchecked(&idx)
})
}
/// # Safety
/// This doesn't check any bounds. Null validity is checked.
#[cfg(feature = "chunked_ids")]
pub(crate) unsafe fn _take_chunked_unchecked_threaded(
&self,
chunk_ids: &[ChunkId],
sorted: IsSorted,
rechunk: bool,
) -> Series {
self.threaded_op(rechunk, chunk_ids.len(), &|offset, len| {
let chunk_ids = &chunk_ids[offset..offset + len];
Ok(self._take_chunked_unchecked(chunk_ids, sorted))
})
.unwrap()
}
/// # Safety
/// This doesn't check any bounds. Null validity is checked.
#[cfg(feature = "chunked_ids")]
pub(crate) unsafe fn _take_opt_chunked_unchecked_threaded(
&self,
chunk_ids: &[Option<ChunkId>],
rechunk: bool,
) -> Series {
self.threaded_op(rechunk, chunk_ids.len(), &|offset, len| {
let chunk_ids = &chunk_ids[offset..offset + len];
Ok(self._take_opt_chunked_unchecked(chunk_ids))
})
.unwrap()
}
/// Take by index. This operation is clone.
///
/// # Notes
/// Out of bounds access doesn't Error but will return a Null value
pub fn take_threaded(&self, idx: &IdxCa, rechunk: bool) -> PolarsResult<Series> {
self.threaded_op(rechunk, idx.len(), &|offset, len| {
let idx = idx.slice(offset as i64, len);
self.take(&idx)
})
}
/// Filter by boolean mask. This operation clones data.
pub fn filter_threaded(&self, filter: &BooleanChunked, rechunk: bool) -> PolarsResult<Series> {
// this would fail if there is a broadcasting filter.
// because we cannot split that filter over threads
// besides they are a no-op, so we do the standard filter.
if filter.len() == 1 {
return self.filter(filter);
}
let n_threads = POOL.current_num_threads();
let filters = split_ca(filter, n_threads).unwrap();
let series = split_series(self, n_threads).unwrap();
let series: PolarsResult<Vec<_>> = POOL.install(|| {
filters
.par_iter()
.zip(series)
.map(|(filter, s)| s.filter(filter))
.collect()
});
Ok(self.finish_take_threaded(series?, rechunk))
}
#[cfg(feature = "dot_product")]
#[cfg_attr(docsrs, doc(cfg(feature = "dot_product")))]
pub fn dot(&self, other: &Series) -> Option<f64> {
(self * other).sum::<f64>()
}
/// Get the sum of the Series as a new Series of length 1.
/// Returns a Series with a single zeroed entry if self is an empty numeric series.
///
/// If the [`DataType`] is one of `{Int8, UInt8, Int16, UInt16}` the `Series` is
/// first cast to `Int64` to prevent overflow issues.
pub fn sum_as_series(&self) -> Series {
use DataType::*;
if self.is_empty() && self.dtype().is_numeric() {
return Series::new("", [0])
.cast(self.dtype())
.unwrap()
.sum_as_series();
}
match self.dtype() {
Int8 | UInt8 | Int16 | UInt16 => self.cast(&Int64).unwrap().sum_as_series(),
_ => self._sum_as_series(),
}
}sourcepub fn cummax(&self, _reverse: bool) -> Series
Available on crate feature cum_agg only.
pub fn cummax(&self, _reverse: bool) -> Series
cum_agg only.Get an array with the cumulative max computed at every element
sourcepub fn cummin(&self, _reverse: bool) -> Series
Available on crate feature cum_agg only.
pub fn cummin(&self, _reverse: bool) -> Series
cum_agg only.Get an array with the cumulative min computed at every element
sourcepub fn cumsum(&self, reverse: bool) -> Series
Available on crate feature cum_agg only.
pub fn cumsum(&self, reverse: bool) -> Series
cum_agg only.Get an array with the cumulative sum computed at every element
If the DataType is one of {Int8, UInt8, Int16, UInt16} the Series is
first cast to Int64 to prevent overflow issues.
Examples found in repository?
src/series/mod.rs (line 570)
565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606
pub fn cumsum(&self, reverse: bool) -> Series {
#[cfg(feature = "cum_agg")]
{
use DataType::*;
match self.dtype() {
Boolean => self.cast(&DataType::UInt32).unwrap().cumsum(reverse),
Int8 | UInt8 | Int16 | UInt16 => {
let s = self.cast(&Int64).unwrap();
s.cumsum(reverse)
}
Int32 => {
let ca = self.i32().unwrap();
ca.cumsum(reverse).into_series()
}
UInt32 => {
let ca = self.u32().unwrap();
ca.cumsum(reverse).into_series()
}
UInt64 => {
let ca = self.u64().unwrap();
ca.cumsum(reverse).into_series()
}
Int64 => {
let ca = self.i64().unwrap();
ca.cumsum(reverse).into_series()
}
Float32 => {
let ca = self.f32().unwrap();
ca.cumsum(reverse).into_series()
}
Float64 => {
let ca = self.f64().unwrap();
ca.cumsum(reverse).into_series()
}
dt => panic!("cumsum not supported for dtype: {dt:?}"),
}
}
#[cfg(not(feature = "cum_agg"))]
{
panic!("activate 'cum_agg' feature")
}
}sourcepub fn cumprod(&self, reverse: bool) -> Series
Available on crate feature cum_agg only.
pub fn cumprod(&self, reverse: bool) -> Series
cum_agg only.Get an array with the cumulative product computed at every element
If the DataType is one of {Int8, UInt8, Int16, UInt16, Int32, UInt32} the Series is
first cast to Int64 to prevent overflow issues.
Examples found in repository?
src/series/mod.rs (line 619)
614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647
pub fn cumprod(&self, reverse: bool) -> Series {
#[cfg(feature = "cum_agg")]
{
use DataType::*;
match self.dtype() {
Boolean => self.cast(&DataType::Int64).unwrap().cumprod(reverse),
Int8 | UInt8 | Int16 | UInt16 | Int32 | UInt32 => {
let s = self.cast(&Int64).unwrap();
s.cumprod(reverse)
}
Int64 => {
let ca = self.i64().unwrap();
ca.cumprod(reverse).into_series()
}
UInt64 => {
let ca = self.u64().unwrap();
ca.cumprod(reverse).into_series()
}
Float32 => {
let ca = self.f32().unwrap();
ca.cumprod(reverse).into_series()
}
Float64 => {
let ca = self.f64().unwrap();
ca.cumprod(reverse).into_series()
}
dt => panic!("cumprod not supported for dtype: {dt:?}"),
}
}
#[cfg(not(feature = "cum_agg"))]
{
panic!("activate 'cum_agg' feature")
}
}sourcepub fn product(&self) -> Series
Available on crate feature product only.
pub fn product(&self) -> Series
product only.Get the product of an array.
If the DataType is one of {Int8, UInt8, Int16, UInt16} the Series is
first cast to Int64 to prevent overflow issues.
Examples found in repository?
src/series/mod.rs (line 659)
654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683
pub fn product(&self) -> Series {
#[cfg(feature = "product")]
{
use DataType::*;
match self.dtype() {
Boolean => self.cast(&DataType::Int64).unwrap().product(),
Int8 | UInt8 | Int16 | UInt16 => {
let s = self.cast(&Int64).unwrap();
s.product()
}
Int64 => {
let ca = self.i64().unwrap();
ca.prod_as_series()
}
Float32 => {
let ca = self.f32().unwrap();
ca.prod_as_series()
}
Float64 => {
let ca = self.f64().unwrap();
ca.prod_as_series()
}
dt => panic!("cumprod not supported for dtype: {dt:?}"),
}
}
#[cfg(not(feature = "product"))]
{
panic!("activate 'product' feature")
}
}pub fn rank(&self, options: RankOptions) -> Series
Available on crate feature
rank only.sourcepub fn strict_cast(&self, data_type: &DataType) -> PolarsResult<Series>
pub fn strict_cast(&self, data_type: &DataType) -> PolarsResult<Series>
Cast throws an error if conversion had overflows
sourcepub fn abs(&self) -> PolarsResult<Series>
Available on crate feature abs only.
pub fn abs(&self) -> PolarsResult<Series>
abs only.convert numerical values to their absolute value
sourcepub fn str_value(&self, index: usize) -> PolarsResult<Cow<'_, str>>
pub fn str_value(&self, index: usize) -> PolarsResult<Cow<'_, str>>
Examples found in repository?
src/fmt.rs (line 118)
98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126
fn format_object_array(
f: &mut Formatter<'_>,
object: &Series,
name: &str,
array_type: &str,
) -> fmt::Result {
match object.dtype() {
DataType::Object(inner_type) => {
let limit = std::cmp::min(LIMIT, object.len());
write!(
f,
"shape: ({},)\n{}: '{}' [o][{}]\n[\n",
object.len(),
array_type,
name,
inner_type
)?;
for i in 0..limit {
let v = object.str_value(i);
writeln!(f, "\t{}", v.unwrap())?;
}
write!(f, "]")
}
_ => unreachable!(),
}
}sourcepub fn mean_as_series(&self) -> Series
pub fn mean_as_series(&self) -> Series
sourcepub fn unique_stable(&self) -> PolarsResult<Series>
pub fn unique_stable(&self) -> PolarsResult<Series>
Compute the unique elements, but maintain order. This requires more work
than a naive Series::unique.
pub fn idx(&self) -> PolarsResult<&IdxCa>
sourcepub fn estimated_size(&self) -> usize
pub fn estimated_size(&self) -> usize
Returns an estimation of the total (heap) allocated size of the Series in bytes.
Implementation
This estimation is the sum of the size of its buffers, validity, including nested arrays.
Multiple arrays may share buffers and bitmaps. Therefore, the size of 2 arrays is not the
sum of the sizes computed from this function. In particular, StructArray’s size is an upper bound.
When an array is sliced, its allocated size remains constant because the buffer unchanged. However, this function will yield a smaller number. This is because this function returns the visible size of the buffer, not its total capacity.
FFI buffers are included in this estimation.
source§impl Series
impl Series
sourcepub fn series_equal(&self, other: &Series) -> bool
pub fn series_equal(&self, other: &Series) -> bool
Check if series are equal. Note that None == None evaluates to false
sourcepub fn series_equal_missing(&self, other: &Series) -> bool
pub fn series_equal_missing(&self, other: &Series) -> bool
Check if all values in series are equal where None == None evaluates to true.
Two Datetime series are not equal if their timezones are different, regardless
if they represent the same UTC time or not.
Examples found in repository?
More examples
src/testing.rs (line 12)
8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131
pub fn series_equal(&self, other: &Series) -> bool {
if self.null_count() > 0 || other.null_count() > 0 || self.dtype() != other.dtype() {
false
} else {
self.series_equal_missing(other)
}
}
/// Check if all values in series are equal where `None == None` evaluates to `true`.
/// Two `Datetime` series are *not* equal if their timezones are different, regardless
/// if they represent the same UTC time or not.
pub fn series_equal_missing(&self, other: &Series) -> bool {
// TODO! remove this? Default behavior already includes equal missing
#[cfg(feature = "timezones")]
{
use crate::datatypes::DataType::Datetime;
if let Datetime(_, tz_lhs) = self.dtype() {
if let Datetime(_, tz_rhs) = other.dtype() {
if tz_lhs != tz_rhs {
return false;
}
} else {
return false;
}
}
}
// differences from Partial::eq in that numerical dtype may be different
self.len() == other.len()
&& self.name() == other.name()
&& self.null_count() == other.null_count()
&& {
let eq = self.equal(other);
match eq {
Ok(b) => b.sum().map(|s| s as usize).unwrap_or(0) == self.len(),
Err(_) => false,
}
}
}
/// Get a pointer to the underlying data of this Series.
/// Can be useful for fast comparisons.
pub fn get_data_ptr(&self) -> usize {
let object = self.0.deref();
// Safety:
// A fat pointer consists of a data ptr and a ptr to the vtable.
// we specifically check that we only transmute &dyn SeriesTrait e.g.
// a trait object, therefore this is sound.
#[allow(clippy::transmute_undefined_repr)]
let (data_ptr, _vtable_ptr) =
unsafe { std::mem::transmute::<&dyn SeriesTrait, (usize, usize)>(object) };
data_ptr
}
}
impl PartialEq for Series {
fn eq(&self, other: &Self) -> bool {
self.len() == other.len()
&& self.field() == other.field()
&& self.null_count() == other.null_count()
&& self
.equal(other)
.unwrap()
.sum()
.map(|s| s as usize)
.unwrap_or(0)
== self.len()
}
}
impl DataFrame {
/// Check if `DataFrames` are equal. Note that `None == None` evaluates to `false`
///
/// # Example
///
/// ```rust
/// # use polars_core::prelude::*;
/// let df1: DataFrame = df!("Atomic number" => &[1, 51, 300],
/// "Element" => &[Some("Hydrogen"), Some("Antimony"), None])?;
/// let df2: DataFrame = df!("Atomic number" => &[1, 51, 300],
/// "Element" => &[Some("Hydrogen"), Some("Antimony"), None])?;
///
/// assert!(!df1.frame_equal(&df2));
/// # Ok::<(), PolarsError>(())
/// ```
pub fn frame_equal(&self, other: &DataFrame) -> bool {
if self.shape() != other.shape() {
return false;
}
for (left, right) in self.get_columns().iter().zip(other.get_columns()) {
if !left.series_equal(right) {
return false;
}
}
true
}
/// Check if all values in `DataFrames` are equal where `None == None` evaluates to `true`.
///
/// # Example
///
/// ```rust
/// # use polars_core::prelude::*;
/// let df1: DataFrame = df!("Atomic number" => &[1, 51, 300],
/// "Element" => &[Some("Hydrogen"), Some("Antimony"), None])?;
/// let df2: DataFrame = df!("Atomic number" => &[1, 51, 300],
/// "Element" => &[Some("Hydrogen"), Some("Antimony"), None])?;
///
/// assert!(df1.frame_equal_missing(&df2));
/// # Ok::<(), PolarsError>(())
/// ```
pub fn frame_equal_missing(&self, other: &DataFrame) -> bool {
if self.shape() != other.shape() {
return false;
}
for (left, right) in self.get_columns().iter().zip(other.get_columns()) {
if !left.series_equal_missing(right) {
return false;
}
}
true
}src/chunked_array/comparison.rs (line 898)
893 894 895 896 897 898 899 900 901 902 903 904 905 906 907 908 909 910 911 912 913 914 915 916
fn equal(&self, rhs: &ListChunked) -> BooleanChunked {
self.amortized_iter()
.zip(rhs.amortized_iter())
.map(|(left, right)| match (left, right) {
(None, None) => true,
(Some(l), Some(r)) => l.as_ref().series_equal_missing(r.as_ref()),
_ => false,
})
.collect_trusted()
}
fn not_equal(&self, rhs: &ListChunked) -> BooleanChunked {
self.amortized_iter()
.zip(rhs.amortized_iter())
.map(|(left, right)| {
let out = match (left, right) {
(None, None) => true,
(Some(l), Some(r)) => l.as_ref().series_equal_missing(r.as_ref()),
_ => false,
};
!out
})
.collect_trusted()
}sourcepub fn get_data_ptr(&self) -> usize
pub fn get_data_ptr(&self) -> usize
Get a pointer to the underlying data of this Series. Can be useful for fast comparisons.
Methods from Deref<Target = dyn SeriesTrait>§
sourcepub fn unpack<N>(&self) -> PolarsResult<&ChunkedArray<N>>where
N: PolarsDataType + 'static,
pub fn unpack<N>(&self) -> PolarsResult<&ChunkedArray<N>>where
N: PolarsDataType + 'static,
Examples found in repository?
src/chunked_array/builder/list.rs (line 160)
155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174
fn append_series(&mut self, s: &Series) {
if s.is_empty() {
self.fast_explode = false;
}
let physical = s.to_physical_repr();
let ca = physical.unpack::<T>().unwrap();
let values = self.builder.mut_values();
ca.downcast_iter().for_each(|arr| {
if !arr.has_validity() {
values.extend_from_slice(arr.values().as_slice())
} else {
// Safety:
// Arrow arrays are trusted length iterators.
unsafe { values.extend_trusted_len_unchecked(arr.into_iter()) }
}
});
// overflow of i64 is far beyond polars capable lengths.
unsafe { self.builder.try_push_valid().unwrap_unchecked() };
}More examples
src/chunked_array/ndarray.rs (line 44)
22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163
pub fn to_ndarray<N>(&self) -> PolarsResult<Array2<N::Native>>
where
N: PolarsNumericType,
{
if self.null_count() != 0 {
Err(PolarsError::ComputeError(
"Creation of ndarray with null values is not supported.".into(),
))
} else {
let mut iter = self.into_no_null_iter();
let mut ndarray;
let width;
// first iteration determine the size
if let Some(series) = iter.next() {
width = series.len();
let mut row_idx = 0;
ndarray = ndarray::Array::uninit((self.len(), width));
let series = series.cast(&N::get_dtype())?;
let ca = series.unpack::<N>()?;
let a = ca.to_ndarray()?;
let mut row = ndarray.slice_mut(s![row_idx, ..]);
a.assign_to(&mut row);
row_idx += 1;
for series in iter {
if series.len() != width {
return Err(PolarsError::ShapeMisMatch(
"Could not create a 2D array. Series have different lengths".into(),
));
}
let series = series.cast(&N::get_dtype())?;
let ca = series.unpack::<N>()?;
let a = ca.to_ndarray()?;
let mut row = ndarray.slice_mut(s![row_idx, ..]);
a.assign_to(&mut row);
row_idx += 1;
}
debug_assert_eq!(row_idx, self.len());
// Safety:
// We have assigned to every row and element of the array
unsafe { Ok(ndarray.assume_init()) }
} else {
Err(PolarsError::NoData(
"cannot create ndarray of empty ListChunked".into(),
))
}
}
}
}
impl DataFrame {
/// Create a 2D `ndarray::Array` from this `DataFrame`. This requires all columns in the
/// `DataFrame` to be non-null and numeric. They will be casted to the same data type
/// (if they aren't already).
///
/// For floating point data we implicitly convert `None` to `NaN` without failure.
///
/// ```rust
/// use polars_core::prelude::*;
/// let a = UInt32Chunked::new("a", &[1, 2, 3]).into_series();
/// let b = Float64Chunked::new("b", &[10., 8., 6.]).into_series();
///
/// let df = DataFrame::new(vec![a, b]).unwrap();
/// let ndarray = df.to_ndarray::<Float64Type>().unwrap();
/// println!("{:?}", ndarray);
/// ```
/// Outputs:
/// ```text
/// [[1.0, 10.0],
/// [2.0, 8.0],
/// [3.0, 6.0]], shape=[3, 2], strides=[2, 1], layout=C (0x1), const ndim=2/
/// ```
#[cfg_attr(docsrs, doc(cfg(feature = "ndarray")))]
pub fn to_ndarray<N>(&self) -> PolarsResult<Array2<N::Native>>
where
N: PolarsNumericType,
{
let columns = self
.get_columns()
.par_iter()
.map(|s| {
let s = s.cast(&N::get_dtype())?;
let s = match s.dtype() {
DataType::Float32 => {
let ca = s.f32().unwrap();
ca.none_to_nan().into_series()
}
DataType::Float64 => {
let ca = s.f64().unwrap();
ca.none_to_nan().into_series()
}
_ => s,
};
Ok(s.rechunk())
})
.collect::<PolarsResult<Vec<_>>>()?;
let shape = self.shape();
let height = self.height();
let mut membuf = Vec::with_capacity(shape.0 * shape.1);
let ptr = membuf.as_ptr() as usize;
columns.par_iter().enumerate().map(|(col_idx, s)| {
if s.null_count() != 0 {
return Err(PolarsError::ComputeError(
"Creation of ndarray with null values is not supported. Consider using floats and NaNs".into(),
));
}
// this is an Arc clone if already of type N
let s = s.cast(&N::get_dtype())?;
let ca = s.unpack::<N>()?;
let vals = ca.cont_slice().unwrap();
// Safety:
// we get parallel access to the vector
// but we make sure that we don't get aliased access by offsetting the column indices + length
unsafe {
let offset_ptr = (ptr as *mut N::Native).add(col_idx * height) ;
// Safety:
// this is uninitialized memory, so we must never read from this data
// copy_from_slice does not read
let buf = std::slice::from_raw_parts_mut(offset_ptr, height);
buf.copy_from_slice(vals)
}
Ok(())
}).collect::<PolarsResult<Vec<_>>>()?;
// Safety:
// we have written all data, so we can now safely set length
unsafe {
membuf.set_len(shape.0 * shape.1);
}
let ndarr = Array2::from_shape_vec((shape.1, shape.0), membuf).unwrap();
Ok(ndarr.reversed_axes())
}src/frame/groupby/aggregations/mod.rs (line 781)
772 773 774 775 776 777 778 779 780 781 782 783 784 785 786 787 788 789 790 791 792 793 794 795 796 797 798 799 800 801 802 803 804 805 806 807 808 809 810 811 812 813 814 815 816 817 818 819 820 821 822 823 824 825 826 827 828 829 830 831 832 833 834 835 836 837 838 839 840 841 842 843 844 845 846 847 848 849 850 851 852 853 854 855 856 857 858 859 860 861 862 863 864 865 866 867 868 869 870 871 872 873 874 875 876 877 878 879 880 881 882 883 884 885 886 887 888 889 890 891 892 893 894 895 896 897 898 899 900 901 902 903 904 905 906 907 908 909 910 911 912 913 914 915 916 917 918 919 920 921 922 923 924 925 926 927 928 929 930 931 932 933 934 935 936 937 938 939 940 941 942 943 944 945 946 947 948 949 950 951 952 953 954 955 956 957 958 959 960 961 962 963 964 965 966 967 968 969 970 971 972 973 974 975 976 977 978 979 980 981 982 983 984 985 986 987 988 989 990 991 992 993 994 995 996 997 998 999 1000 1001 1002 1003 1004 1005 1006 1007 1008 1009 1010 1011 1012 1013 1014 1015 1016 1017 1018 1019 1020 1021 1022 1023 1024 1025 1026 1027 1028 1029 1030 1031 1032 1033 1034 1035 1036 1037 1038 1039 1040 1041 1042 1043 1044 1045 1046 1047 1048 1049 1050 1051 1052 1053 1054 1055 1056 1057 1058 1059 1060 1061 1062 1063 1064 1065 1066 1067 1068 1069 1070 1071 1072 1073 1074 1075 1076 1077 1078 1079 1080 1081 1082 1083 1084 1085 1086 1087 1088 1089 1090 1091 1092 1093 1094 1095 1096 1097 1098 1099 1100 1101 1102 1103 1104 1105 1106 1107 1108 1109 1110 1111 1112 1113 1114 1115 1116 1117 1118 1119 1120 1121 1122 1123 1124 1125 1126 1127 1128 1129 1130 1131 1132 1133 1134 1135 1136 1137 1138 1139 1140 1141 1142 1143 1144 1145 1146 1147 1148 1149 1150 1151 1152 1153 1154 1155 1156 1157 1158 1159 1160 1161 1162 1163 1164 1165 1166 1167 1168 1169 1170 1171 1172 1173 1174 1175 1176 1177
pub(crate) unsafe fn agg_var(&self, groups: &GroupsProxy, ddof: u8) -> Series {
let ca = &self.0;
match groups {
GroupsProxy::Idx(groups) => agg_helper_idx_on_all::<T, _>(groups, |idx| {
debug_assert!(idx.len() <= ca.len());
if idx.is_empty() {
return None;
}
let take = { ca.take_unchecked(idx.into()) };
take.var_as_series(ddof).unpack::<T>().unwrap().get(0)
}),
GroupsProxy::Slice { groups, .. } => {
if _use_rolling_kernels(groups, self.chunks()) {
let arr = self.downcast_iter().next().unwrap();
let values = arr.values().as_slice();
let offset_iter = groups.iter().map(|[first, len]| (*first, *len));
let arr = match arr.validity() {
None => _rolling_apply_agg_window_no_nulls::<VarWindow<_>, _, _>(
values,
offset_iter,
),
Some(validity) => _rolling_apply_agg_window_nulls::<
rolling::nulls::VarWindow<_>,
_,
_,
>(values, validity, offset_iter),
};
ChunkedArray::<T>::from_chunks("", vec![arr]).into_series()
} else {
_agg_helper_slice::<T, _>(groups, |[first, len]| {
debug_assert!(len <= self.len() as IdxSize);
match len {
0 => None,
1 => NumCast::from(0),
_ => {
let arr_group = _slice_from_offsets(self, first, len);
arr_group.var(ddof).map(|flt| NumCast::from(flt).unwrap())
}
}
})
}
}
}
}
pub(crate) unsafe fn agg_std(&self, groups: &GroupsProxy, ddof: u8) -> Series {
let ca = &self.0;
match groups {
GroupsProxy::Idx(groups) => agg_helper_idx_on_all::<T, _>(groups, |idx| {
debug_assert!(idx.len() <= ca.len());
if idx.is_empty() {
return None;
}
let take = { ca.take_unchecked(idx.into()) };
take.std_as_series(ddof).unpack::<T>().unwrap().get(0)
}),
GroupsProxy::Slice { groups, .. } => {
if _use_rolling_kernels(groups, self.chunks()) {
let arr = self.downcast_iter().next().unwrap();
let values = arr.values().as_slice();
let offset_iter = groups.iter().map(|[first, len]| (*first, *len));
let arr = match arr.validity() {
None => _rolling_apply_agg_window_no_nulls::<StdWindow<_>, _, _>(
values,
offset_iter,
),
Some(validity) => _rolling_apply_agg_window_nulls::<
rolling::nulls::StdWindow<_>,
_,
_,
>(values, validity, offset_iter),
};
ChunkedArray::<T>::from_chunks("", vec![arr]).into_series()
} else {
_agg_helper_slice::<T, _>(groups, |[first, len]| {
debug_assert!(len <= self.len() as IdxSize);
match len {
0 => None,
1 => NumCast::from(0),
_ => {
let arr_group = _slice_from_offsets(self, first, len);
arr_group.std(ddof).map(|flt| NumCast::from(flt).unwrap())
}
}
})
}
}
}
}
pub(crate) unsafe fn agg_quantile(
&self,
groups: &GroupsProxy,
quantile: f64,
interpol: QuantileInterpolOptions,
) -> Series {
let ca = &self.0;
let invalid_quantile = !(0.0..=1.0).contains(&quantile);
match groups {
GroupsProxy::Idx(groups) => agg_helper_idx_on_all::<T, _>(groups, |idx| {
debug_assert!(idx.len() <= ca.len());
if idx.is_empty() | invalid_quantile {
return None;
}
let take = { ca.take_unchecked(idx.into()) };
take.quantile_as_series(quantile, interpol)
.unwrap() // checked with invalid quantile check
.unpack::<T>()
.unwrap()
.get(0)
}),
GroupsProxy::Slice { groups, .. } => {
if _use_rolling_kernels(groups, self.chunks()) {
let arr = self.downcast_iter().next().unwrap();
let values = arr.values().as_slice();
let offset_iter = groups.iter().map(|[first, len]| (*first, *len));
let arr = match arr.validity() {
None => rolling::no_nulls::rolling_quantile_by_iter(
values,
quantile,
interpol,
offset_iter,
),
Some(validity) => rolling::nulls::rolling_quantile_by_iter(
values,
validity,
quantile,
interpol,
offset_iter,
),
};
ChunkedArray::<T>::from_chunks("", vec![arr]).into_series()
} else {
_agg_helper_slice::<T, _>(groups, |[first, len]| {
debug_assert!(first + len <= self.len() as IdxSize);
match len {
0 => None,
1 => self.get(first as usize),
_ => {
let arr_group = _slice_from_offsets(self, first, len);
// unwrap checked with invalid quantile check
arr_group
.quantile(quantile, interpol)
.unwrap()
.map(|flt| NumCast::from(flt).unwrap())
}
}
})
}
}
}
}
pub(crate) unsafe fn agg_median(&self, groups: &GroupsProxy) -> Series {
let ca = &self.0;
match groups {
GroupsProxy::Idx(groups) => agg_helper_idx_on_all::<T, _>(groups, |idx| {
debug_assert!(idx.len() <= ca.len());
if idx.is_empty() {
return None;
}
let take = { ca.take_unchecked(idx.into()) };
take.median_as_series().unpack::<T>().unwrap().get(0)
}),
GroupsProxy::Slice { .. } => {
self.agg_quantile(groups, 0.5, QuantileInterpolOptions::Linear)
}
}
}
}
impl<T> ChunkedArray<T>
where
T: PolarsIntegerType,
ChunkedArray<T>: IntoSeries,
T::Native: NumericNative + Ord,
<T::Native as Simd>::Simd: std::ops::Add<Output = <T::Native as Simd>::Simd>
+ arrow::compute::aggregate::Sum<T::Native>
+ arrow::compute::aggregate::SimdOrd<T::Native>,
{
pub(crate) unsafe fn agg_mean(&self, groups: &GroupsProxy) -> Series {
match groups {
GroupsProxy::Idx(groups) => {
_agg_helper_idx::<Float64Type, _>(groups, |(first, idx)| {
// this can fail due to a bug in lazy code.
// here users can create filters in aggregations
// and thereby creating shorter columns than the original group tuples.
// the group tuples are modified, but if that's done incorrect there can be out of bounds
// access
debug_assert!(idx.len() <= self.len());
if idx.is_empty() {
None
} else if idx.len() == 1 {
self.get(first as usize).map(|sum| sum.to_f64().unwrap())
} else {
match (self.has_validity(), self.chunks.len()) {
(false, 1) => {
take_agg_no_null_primitive_iter_unchecked(
self.downcast_iter().next().unwrap(),
idx.iter().map(|i| *i as usize),
|a, b| a + b,
0.0f64,
)
}
.to_f64()
.map(|sum| sum / idx.len() as f64),
(_, 1) => {
{
take_agg_primitive_iter_unchecked_count_nulls::<
T::Native,
f64,
_,
_,
>(
self.downcast_iter().next().unwrap(),
idx.iter().map(|i| *i as usize),
|a, b| a + b,
0.0,
idx.len() as IdxSize,
)
}
.map(|(sum, null_count)| {
sum / (idx.len() as f64 - null_count as f64)
})
}
_ => {
let take = { self.take_unchecked(idx.into()) };
take.mean()
}
}
}
})
}
GroupsProxy::Slice {
groups: groups_slice,
..
} => {
if _use_rolling_kernels(groups_slice, self.chunks()) {
let ca = self.cast(&DataType::Float64).unwrap();
ca.agg_mean(groups)
} else {
_agg_helper_slice::<Float64Type, _>(groups_slice, |[first, len]| {
debug_assert!(first + len <= self.len() as IdxSize);
match len {
0 => None,
1 => self.get(first as usize).map(|v| NumCast::from(v).unwrap()),
_ => {
let arr_group = _slice_from_offsets(self, first, len);
arr_group.mean()
}
}
})
}
}
}
}
pub(crate) unsafe fn agg_var(&self, groups: &GroupsProxy, ddof: u8) -> Series {
match groups {
GroupsProxy::Idx(groups) => agg_helper_idx_on_all::<Float64Type, _>(groups, |idx| {
debug_assert!(idx.len() <= self.len());
if idx.is_empty() {
return None;
}
let take = { self.take_unchecked(idx.into()) };
take.var_as_series(ddof)
.unpack::<Float64Type>()
.unwrap()
.get(0)
}),
GroupsProxy::Slice {
groups: groups_slice,
..
} => {
if _use_rolling_kernels(groups_slice, self.chunks()) {
let ca = self.cast(&DataType::Float64).unwrap();
ca.agg_var(groups, ddof)
} else {
_agg_helper_slice::<Float64Type, _>(groups_slice, |[first, len]| {
debug_assert!(first + len <= self.len() as IdxSize);
match len {
0 => None,
1 => NumCast::from(0),
_ => {
let arr_group = _slice_from_offsets(self, first, len);
arr_group.var(ddof)
}
}
})
}
}
}
}
pub(crate) unsafe fn agg_std(&self, groups: &GroupsProxy, ddof: u8) -> Series {
match groups {
GroupsProxy::Idx(groups) => agg_helper_idx_on_all::<Float64Type, _>(groups, |idx| {
debug_assert!(idx.len() <= self.len());
if idx.is_empty() {
return None;
}
let take = { self.take_unchecked(idx.into()) };
take.std_as_series(ddof)
.unpack::<Float64Type>()
.unwrap()
.get(0)
}),
GroupsProxy::Slice {
groups: groups_slice,
..
} => {
if _use_rolling_kernels(groups_slice, self.chunks()) {
let ca = self.cast(&DataType::Float64).unwrap();
ca.agg_std(groups, ddof)
} else {
_agg_helper_slice::<Float64Type, _>(groups_slice, |[first, len]| {
debug_assert!(first + len <= self.len() as IdxSize);
match len {
0 => None,
1 => NumCast::from(0),
_ => {
let arr_group = _slice_from_offsets(self, first, len);
arr_group.std(ddof)
}
}
})
}
}
}
}
pub(crate) unsafe fn agg_quantile(
&self,
groups: &GroupsProxy,
quantile: f64,
interpol: QuantileInterpolOptions,
) -> Series {
match groups {
GroupsProxy::Idx(groups) => agg_helper_idx_on_all::<Float64Type, _>(groups, |idx| {
debug_assert!(idx.len() <= self.len());
if idx.is_empty() {
return None;
}
let take = self.take_unchecked(idx.into());
take.quantile_as_series(quantile, interpol)
.unwrap()
.unpack::<Float64Type>()
.unwrap()
.get(0)
}),
GroupsProxy::Slice {
groups: groups_slice,
..
} => {
if _use_rolling_kernels(groups_slice, self.chunks()) {
let ca = self.cast(&DataType::Float64).unwrap();
ca.agg_quantile(groups, quantile, interpol)
} else {
_agg_helper_slice::<Float64Type, _>(groups_slice, |[first, len]| {
debug_assert!(len <= self.len() as IdxSize);
match len {
0 => None,
1 => self.get(first as usize).map(|v| NumCast::from(v).unwrap()),
_ => {
let arr_group = _slice_from_offsets(self, first, len);
arr_group.quantile(quantile, interpol).unwrap()
}
}
})
}
}
}
}
pub(crate) unsafe fn agg_median(&self, groups: &GroupsProxy) -> Series {
match groups {
GroupsProxy::Idx(groups) => agg_helper_idx_on_all::<Float64Type, _>(groups, |idx| {
debug_assert!(idx.len() <= self.len());
if idx.is_empty() {
return None;
}
let take = self.take_unchecked(idx.into());
take.median_as_series()
.unpack::<Float64Type>()
.unwrap()
.get(0)
}),
GroupsProxy::Slice {
groups: groups_slice,
..
} => {
if _use_rolling_kernels(groups_slice, self.chunks()) {
let ca = self.cast(&DataType::Float64).unwrap();
ca.agg_median(groups)
} else {
_agg_helper_slice::<Float64Type, _>(groups_slice, |[first, len]| {
debug_assert!(len <= self.len() as IdxSize);
match len {
0 => None,
1 => self.get(first as usize).map(|v| NumCast::from(v).unwrap()),
_ => {
let arr_group = _slice_from_offsets(self, first, len);
arr_group.median()
}
}
})
}
}
}
}src/chunked_array/ops/is_in.rs (line 64)
45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351
fn is_in(&self, other: &Series) -> PolarsResult<BooleanChunked> {
// We check implicitly cast to supertype here
match other.dtype() {
DataType::List(dt) => {
let st = try_get_supertype(self.dtype(), dt)?;
if &st != self.dtype() {
let left = self.cast(&st)?;
let right = other.cast(&DataType::List(Box::new(st)))?;
return left.is_in(&right);
}
let mut ca: BooleanChunked = if self.len() == 1 && other.len() != 1 {
let value = self.get(0);
other
.list()?
.amortized_iter()
.map(|opt_s| {
opt_s.map(|s| {
let ca = s.as_ref().unpack::<T>().unwrap();
ca.into_iter().any(|a| a == value)
}) == Some(true)
})
.collect_trusted()
} else {
self.into_iter()
.zip(other.list()?.amortized_iter())
.map(|(value, series)| match (value, series) {
(val, Some(series)) => {
let ca = series.as_ref().unpack::<T>().unwrap();
ca.into_iter().any(|a| a == val)
}
_ => false,
})
.collect_trusted()
};
ca.rename(self.name());
Ok(ca)
}
_ => {
// first make sure that the types are equal
let st = try_get_supertype(self.dtype(), other.dtype())?;
if self.dtype() != other.dtype() {
let left = self.cast(&st)?;
let right = other.cast(&st)?;
return left.is_in(&right);
}
// now that the types are equal, we coerce every 32 bit array to u32
// and every 64 bit array to u64 (including floats)
// this allows hashing them and greatly reduces the number of code paths.
match self.dtype() {
DataType::UInt64 | DataType::Int64 | DataType::Float64 => unsafe {
is_in_helper::<T, u64>(self, other)
},
DataType::UInt32 | DataType::Int32 | DataType::Float32 => unsafe {
is_in_helper::<T, u32>(self, other)
},
DataType::UInt8 | DataType::Int8 => unsafe {
is_in_helper::<T, u8>(self, other)
},
DataType::UInt16 | DataType::Int16 => unsafe {
is_in_helper::<T, u16>(self, other)
},
_ => Err(PolarsError::ComputeError(
format!(
"Data type {:?} not supported in is_in operation",
self.dtype()
)
.into(),
)),
}
}
}
.map(|mut ca| {
ca.rename(self.name());
ca
})
}
}
impl IsIn for Utf8Chunked {
fn is_in(&self, other: &Series) -> PolarsResult<BooleanChunked> {
match other.dtype() {
#[cfg(feature = "dtype-categorical")]
DataType::List(dt) if matches!(&**dt, DataType::Categorical(_)) => {
if let DataType::Categorical(Some(rev_map)) = &**dt {
let opt_val = self.get(0);
let other = other.list()?;
match opt_val {
None => {
let mut ca: BooleanChunked = other
.amortized_iter()
.map(|opt_s| {
opt_s.map(|s| s.as_ref().null_count() > 0) == Some(true)
})
.collect_trusted();
ca.rename(self.name());
Ok(ca)
}
Some(value) => {
match rev_map.find(value) {
// all false
None => Ok(BooleanChunked::full(self.name(), false, other.len())),
Some(idx) => {
let mut ca: BooleanChunked = other
.amortized_iter()
.map(|opt_s| {
opt_s.map(|s| {
let s = s.as_ref().to_physical_repr();
let ca = s.as_ref().u32().unwrap();
if ca.null_count() == 0 {
ca.into_no_null_iter().any(|a| a == idx)
} else {
ca.into_iter().any(|a| a == Some(idx))
}
}) == Some(true)
})
.collect_trusted();
ca.rename(self.name());
Ok(ca)
}
}
}
}
} else {
unreachable!()
}
}
DataType::List(dt) if DataType::Utf8 == **dt => {
let mut ca: BooleanChunked = if self.len() == 1 && other.len() != 1 {
let value = self.get(0);
other
.list()?
.amortized_iter()
.map(|opt_s| {
opt_s.map(|s| {
let ca = s.as_ref().unpack::<Utf8Type>().unwrap();
ca.into_iter().any(|a| a == value)
}) == Some(true)
})
.collect_trusted()
} else {
self.into_iter()
.zip(other.list()?.amortized_iter())
.map(|(value, series)| match (value, series) {
(val, Some(series)) => {
let ca = series.as_ref().unpack::<Utf8Type>().unwrap();
ca.into_iter().any(|a| a == val)
}
_ => false,
})
.collect_trusted()
};
ca.rename(self.name());
Ok(ca)
}
DataType::Utf8 => {
let mut set = HashSet::with_capacity(other.len());
let other = other.utf8()?;
other.downcast_iter().for_each(|iter| {
iter.into_iter().for_each(|opt_val| {
set.insert(opt_val);
})
});
let mut ca: BooleanChunked = self
.into_iter()
.map(|opt_val| set.contains(&opt_val))
.collect_trusted();
ca.rename(self.name());
Ok(ca)
}
_ => Err(PolarsError::SchemaMisMatch(
format!(
"cannot do is_in operation with left a dtype: {:?} and right a dtype {:?}",
self.dtype(),
other.dtype()
)
.into(),
)),
}
.map(|mut ca| {
ca.rename(self.name());
ca
})
}
}
#[cfg(feature = "dtype-binary")]
impl IsIn for BinaryChunked {
fn is_in(&self, other: &Series) -> PolarsResult<BooleanChunked> {
match other.dtype() {
DataType::List(dt) if DataType::Binary == **dt => {
let mut ca: BooleanChunked = if self.len() == 1 && other.len() != 1 {
let value = self.get(0);
other
.list()?
.amortized_iter()
.map(|opt_b| {
opt_b.map(|s| {
let ca = s.as_ref().unpack::<BinaryType>().unwrap();
ca.into_iter().any(|a| a == value)
}) == Some(true)
})
.collect_trusted()
} else {
self.into_iter()
.zip(other.list()?.amortized_iter())
.map(|(value, series)| match (value, series) {
(val, Some(series)) => {
let ca = series.as_ref().unpack::<BinaryType>().unwrap();
ca.into_iter().any(|a| a == val)
}
_ => false,
})
.collect_trusted()
};
ca.rename(self.name());
Ok(ca)
}
DataType::Binary => {
let mut set = HashSet::with_capacity(other.len());
let other = other.binary()?;
other.downcast_iter().for_each(|iter| {
iter.into_iter().for_each(|opt_val| {
set.insert(opt_val);
})
});
let mut ca: BooleanChunked = self
.into_iter()
.map(|opt_val| set.contains(&opt_val))
.collect_trusted();
ca.rename(self.name());
Ok(ca)
}
_ => Err(PolarsError::SchemaMisMatch(
format!(
"cannot do is_in operation with left a dtype: {:?} and right a dtype {:?}",
self.dtype(),
other.dtype()
)
.into(),
)),
}
.map(|mut ca| {
ca.rename(self.name());
ca
})
}
}
impl IsIn for BooleanChunked {
fn is_in(&self, other: &Series) -> PolarsResult<BooleanChunked> {
match other.dtype() {
DataType::List(dt) if self.dtype() == &**dt => {
let mut ca: BooleanChunked = if self.len() == 1 && other.len() != 1 {
let value = self.get(0);
// safety: we know the iterators len
unsafe {
other
.list()?
.amortized_iter()
.map(|opt_s| {
opt_s.map(|s| {
let ca = s.as_ref().unpack::<BooleanType>().unwrap();
ca.into_iter().any(|a| a == value)
}) == Some(true)
})
.trust_my_length(other.len())
.collect_trusted()
}
} else {
self.into_iter()
.zip(other.list()?.amortized_iter())
.map(|(value, series)| match (value, series) {
(val, Some(series)) => {
let ca = series.as_ref().unpack::<BooleanType>().unwrap();
ca.into_iter().any(|a| a == val)
}
_ => false,
})
.collect_trusted()
};
ca.rename(self.name());
Ok(ca)
}
DataType::Boolean => {
let other = other.bool().unwrap();
let has_true = other.any();
let has_false = !other.all();
Ok(self.apply(|v| if v { has_true } else { has_false }))
}
_ => Err(PolarsError::SchemaMisMatch(
format!(
"cannot do is_in operation with left a dtype: {:?} and right a dtype {:?}",
self.dtype(),
other.dtype()
)
.into(),
)),
}
.map(|mut ca| {
ca.rename(self.name());
ca
})
}src/frame/row.rs (line 579)
552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651
fn numeric_transpose<T>(cols: &[Series]) -> PolarsResult<DataFrame>
where
T: PolarsNumericType,
ChunkedArray<T>: IntoSeries,
{
let new_width = cols[0].len();
let new_height = cols.len();
let has_nulls = cols.iter().any(|s| s.null_count() > 0);
let mut values_buf: Vec<Vec<T::Native>> = (0..new_width)
.map(|_| Vec::with_capacity(new_height))
.collect();
let mut validity_buf: Vec<_> = if has_nulls {
// we first use bools instead of bits, because we can access these in parallel without aliasing
(0..new_width).map(|_| vec![true; new_height]).collect()
} else {
(0..new_width).map(|_| vec![]).collect()
};
// work with *mut pointers because we it is UB write to &refs.
let values_buf_ptr = &mut values_buf as *mut Vec<Vec<T::Native>> as usize;
let validity_buf_ptr = &mut validity_buf as *mut Vec<Vec<bool>> as usize;
POOL.install(|| {
cols.iter().enumerate().for_each(|(row_idx, s)| {
let s = s.cast(&T::get_dtype()).unwrap();
let ca = s.unpack::<T>().unwrap();
// Safety
// we access in parallel, but every access is unique, so we don't break aliasing rules
// we also ensured we allocated enough memory, so we never reallocate and thus
// the pointers remain valid.
if has_nulls {
for (col_idx, opt_v) in ca.into_iter().enumerate() {
match opt_v {
None => unsafe {
let column = (*(validity_buf_ptr as *mut Vec<Vec<bool>>))
.get_unchecked_mut(col_idx);
let el_ptr = column.as_mut_ptr();
*el_ptr.add(row_idx) = false;
// we must initialize this memory otherwise downstream code
// might access uninitialized memory when the masked out values
// are changed.
add_value(values_buf_ptr, col_idx, row_idx, T::Native::default());
},
Some(v) => unsafe {
add_value(values_buf_ptr, col_idx, row_idx, v);
},
}
}
} else {
for (col_idx, v) in ca.into_no_null_iter().enumerate() {
unsafe {
let column = (*(values_buf_ptr as *mut Vec<Vec<T::Native>>))
.get_unchecked_mut(col_idx);
let el_ptr = column.as_mut_ptr();
*el_ptr.add(row_idx) = v;
}
}
}
})
});
let series = POOL.install(|| {
values_buf
.into_par_iter()
.zip(validity_buf)
.enumerate()
.map(|(i, (mut values, validity))| {
// Safety:
// all values are written we can now set len
unsafe {
values.set_len(new_height);
}
let validity = if has_nulls {
let validity = Bitmap::from_trusted_len_iter(validity.iter().copied());
if validity.unset_bits() > 0 {
Some(validity)
} else {
None
}
} else {
None
};
let arr = PrimitiveArray::<T::Native>::new(
T::get_dtype().to_arrow(),
values.into(),
validity,
);
let name = format!("column_{i}");
ChunkedArray::<T>::from_chunks(&name, vec![Box::new(arr) as ArrayRef]).into_series()
})
.collect()
});
Ok(DataFrame::new_no_checks(series))
}Trait Implementations§
source§impl AsMut<Series> for UnstableSeries<'_>
impl AsMut<Series> for UnstableSeries<'_>
source§impl AsRef<Series> for UnstableSeries<'_>
impl AsRef<Series> for UnstableSeries<'_>
We don’t implement Deref so that the caller is aware of converting to Series
source§impl<'a> AsRef<dyn SeriesTrait + 'a> for Series
impl<'a> AsRef<dyn SeriesTrait + 'a> for Series
source§fn as_ref(&self) -> &(dyn SeriesTrait + 'a)
fn as_ref(&self) -> &(dyn SeriesTrait + 'a)
Converts this type into a shared reference of the (usually inferred) input type.
source§impl<'a> ChunkApply<'a, Series, Series> for ListChunked
impl<'a> ChunkApply<'a, Series, Series> for ListChunked
source§fn apply<F>(&'a self, f: F) -> Selfwhere
F: Fn(Series) -> Series + Copy,
fn apply<F>(&'a self, f: F) -> Selfwhere
F: Fn(Series) -> Series + Copy,
Apply a closure F elementwise.
source§fn apply_with_idx<F>(&'a self, f: F) -> Selfwhere
F: Fn((usize, Series)) -> Series + Copy,
fn apply_with_idx<F>(&'a self, f: F) -> Selfwhere
F: Fn((usize, Series)) -> Series + Copy,
Apply a closure elementwise. The closure gets the index of the element as first argument.
source§fn apply_with_idx_on_opt<F>(&'a self, f: F) -> Selfwhere
F: Fn((usize, Option<Series>)) -> Option<Series> + Copy,
fn apply_with_idx_on_opt<F>(&'a self, f: F) -> Selfwhere
F: Fn((usize, Option<Series>)) -> Option<Series> + Copy,
Apply a closure elementwise. The closure gets the index of the element as first argument.
source§fn apply_cast_numeric<F, S>(&self, f: F) -> ChunkedArray<S>where
F: Fn(Series) -> S::Native + Copy,
S: PolarsNumericType,
fn apply_cast_numeric<F, S>(&self, f: F) -> ChunkedArray<S>where
F: Fn(Series) -> S::Native + Copy,
S: PolarsNumericType,
Apply a closure elementwise and cast to a Numeric ChunkedArray. This is fastest when the null check branching is more expensive
than the closure application. Read more
source§fn branch_apply_cast_numeric_no_null<F, S>(&self, f: F) -> ChunkedArray<S>where
F: Fn(Option<Series>) -> S::Native + Copy,
S: PolarsNumericType,
fn branch_apply_cast_numeric_no_null<F, S>(&self, f: F) -> ChunkedArray<S>where
F: Fn(Option<Series>) -> S::Native + Copy,
S: PolarsNumericType,
Apply a closure on optional values and cast to Numeric ChunkedArray without null values.
fn try_apply<F>(&'a self, f: F) -> PolarsResult<Self>where
F: Fn(Series) -> PolarsResult<Series> + Copy,
source§impl ChunkCompare<&Series> for Series
impl ChunkCompare<&Series> for Series
source§fn equal(&self, rhs: &Series) -> PolarsResult<BooleanChunked>
fn equal(&self, rhs: &Series) -> PolarsResult<BooleanChunked>
Create a boolean mask by checking for equality.
source§fn not_equal(&self, rhs: &Series) -> PolarsResult<BooleanChunked>
fn not_equal(&self, rhs: &Series) -> PolarsResult<BooleanChunked>
Create a boolean mask by checking for inequality.
source§fn gt(&self, rhs: &Series) -> PolarsResult<BooleanChunked>
fn gt(&self, rhs: &Series) -> PolarsResult<BooleanChunked>
Create a boolean mask by checking if self > rhs.
source§fn gt_eq(&self, rhs: &Series) -> PolarsResult<BooleanChunked>
fn gt_eq(&self, rhs: &Series) -> PolarsResult<BooleanChunked>
Create a boolean mask by checking if self >= rhs.
source§fn lt(&self, rhs: &Series) -> PolarsResult<BooleanChunked>
fn lt(&self, rhs: &Series) -> PolarsResult<BooleanChunked>
Create a boolean mask by checking if self < rhs.
source§fn lt_eq(&self, rhs: &Series) -> PolarsResult<BooleanChunked>
fn lt_eq(&self, rhs: &Series) -> PolarsResult<BooleanChunked>
Create a boolean mask by checking if self <= rhs.
type Item = Result<ChunkedArray<BooleanType>, PolarsError>
source§impl ChunkCompare<&str> for Series
impl ChunkCompare<&str> for Series
type Item = Result<ChunkedArray<BooleanType>, PolarsError>
source§fn equal(&self, rhs: &str) -> PolarsResult<BooleanChunked>
fn equal(&self, rhs: &str) -> PolarsResult<BooleanChunked>
Check for equality.
source§fn not_equal(&self, rhs: &str) -> PolarsResult<BooleanChunked>
fn not_equal(&self, rhs: &str) -> PolarsResult<BooleanChunked>
Check for inequality.
source§fn gt(&self, rhs: &str) -> PolarsResult<BooleanChunked>
fn gt(&self, rhs: &str) -> PolarsResult<BooleanChunked>
Greater than comparison.
source§fn gt_eq(&self, rhs: &str) -> PolarsResult<BooleanChunked>
fn gt_eq(&self, rhs: &str) -> PolarsResult<BooleanChunked>
Greater than or equal comparison.
source§fn lt(&self, rhs: &str) -> PolarsResult<BooleanChunked>
fn lt(&self, rhs: &str) -> PolarsResult<BooleanChunked>
Less than comparison.
source§fn lt_eq(&self, rhs: &str) -> PolarsResult<BooleanChunked>
fn lt_eq(&self, rhs: &str) -> PolarsResult<BooleanChunked>
Less than or equal comparison
source§impl<Rhs> ChunkCompare<Rhs> for Serieswhere
Rhs: NumericNative,
impl<Rhs> ChunkCompare<Rhs> for Serieswhere
Rhs: NumericNative,
type Item = Result<ChunkedArray<BooleanType>, PolarsError>
source§fn equal(&self, rhs: Rhs) -> PolarsResult<BooleanChunked>
fn equal(&self, rhs: Rhs) -> PolarsResult<BooleanChunked>
Check for equality.
source§fn not_equal(&self, rhs: Rhs) -> PolarsResult<BooleanChunked>
fn not_equal(&self, rhs: Rhs) -> PolarsResult<BooleanChunked>
Check for inequality.
source§fn gt(&self, rhs: Rhs) -> PolarsResult<BooleanChunked>
fn gt(&self, rhs: Rhs) -> PolarsResult<BooleanChunked>
Greater than comparison.
source§fn gt_eq(&self, rhs: Rhs) -> PolarsResult<BooleanChunked>
fn gt_eq(&self, rhs: Rhs) -> PolarsResult<BooleanChunked>
Greater than or equal comparison.
source§fn lt(&self, rhs: Rhs) -> PolarsResult<BooleanChunked>
fn lt(&self, rhs: Rhs) -> PolarsResult<BooleanChunked>
Less than comparison.
source§fn lt_eq(&self, rhs: Rhs) -> PolarsResult<BooleanChunked>
fn lt_eq(&self, rhs: Rhs) -> PolarsResult<BooleanChunked>
Less than or equal comparison
source§impl ChunkFillNullValue<&Series> for ListChunked
impl ChunkFillNullValue<&Series> for ListChunked
source§fn fill_null_with_values(&self, _value: &Series) -> PolarsResult<Self>
fn fill_null_with_values(&self, _value: &Series) -> PolarsResult<Self>
Replace None values with a give value
T.source§impl ChunkFull<&Series> for ListChunked
impl ChunkFull<&Series> for ListChunked
source§impl ChunkQuantile<Series> for ListChunked
impl ChunkQuantile<Series> for ListChunked
source§fn median(&self) -> Option<T>
fn median(&self) -> Option<T>
Returns the mean value in the array.
Returns
None if the array is empty or only contains null values.source§fn quantile(
&self,
_quantile: f64,
_interpol: QuantileInterpolOptions
) -> PolarsResult<Option<T>>
fn quantile(
&self,
_quantile: f64,
_interpol: QuantileInterpolOptions
) -> PolarsResult<Option<T>>
Aggregate a given quantile of the ChunkedArray.
Returns
None if the array is empty or only contains null values.source§impl<T: PolarsObject> ChunkQuantile<Series> for ObjectChunked<T>
impl<T: PolarsObject> ChunkQuantile<Series> for ObjectChunked<T>
source§fn median(&self) -> Option<T>
fn median(&self) -> Option<T>
Returns the mean value in the array.
Returns
None if the array is empty or only contains null values.source§fn quantile(
&self,
_quantile: f64,
_interpol: QuantileInterpolOptions
) -> PolarsResult<Option<T>>
fn quantile(
&self,
_quantile: f64,
_interpol: QuantileInterpolOptions
) -> PolarsResult<Option<T>>
Aggregate a given quantile of the ChunkedArray.
Returns
None if the array is empty or only contains null values.source§impl ChunkVar<Series> for ListChunked
impl ChunkVar<Series> for ListChunked
source§impl<T: PolarsObject> ChunkVar<Series> for ObjectChunked<T>
impl<T: PolarsObject> ChunkVar<Series> for ObjectChunked<T>
source§impl<T> From<ChunkedArray<T>> for Serieswhere
T: PolarsDataType,
ChunkedArray<T>: IntoSeries,
impl<T> From<ChunkedArray<T>> for Serieswhere
T: PolarsDataType,
ChunkedArray<T>: IntoSeries,
source§fn from(ca: ChunkedArray<T>) -> Self
fn from(ca: ChunkedArray<T>) -> Self
Converts to this type from the input type.
source§impl<'a> FromIterator<&'a bool> for Series
impl<'a> FromIterator<&'a bool> for Series
source§impl<'a> FromIterator<&'a f32> for Series
impl<'a> FromIterator<&'a f32> for Series
source§impl<'a> FromIterator<&'a f64> for Series
impl<'a> FromIterator<&'a f64> for Series
source§impl<'a> FromIterator<&'a i32> for Series
impl<'a> FromIterator<&'a i32> for Series
source§impl<'a> FromIterator<&'a i64> for Series
impl<'a> FromIterator<&'a i64> for Series
source§impl<'a> FromIterator<&'a str> for Series
impl<'a> FromIterator<&'a str> for Series
source§impl<'a> FromIterator<&'a u32> for Series
impl<'a> FromIterator<&'a u32> for Series
source§impl<'a> FromIterator<&'a u64> for Series
impl<'a> FromIterator<&'a u64> for Series
source§impl FromIterator<Series> for DataFrame
impl FromIterator<Series> for DataFrame
source§impl FromIterator<String> for Series
impl FromIterator<String> for Series
source§impl FromIterator<bool> for Series
impl FromIterator<bool> for Series
source§impl FromIterator<f32> for Series
impl FromIterator<f32> for Series
source§impl FromIterator<f64> for Series
impl FromIterator<f64> for Series
source§impl FromIterator<i32> for Series
impl FromIterator<i32> for Series
source§impl FromIterator<i64> for Series
impl FromIterator<i64> for Series
source§impl FromIterator<u32> for Series
impl FromIterator<u32> for Series
source§impl FromIterator<u64> for Series
impl FromIterator<u64> for Series
source§impl<T: IntoSeries> NamedFrom<T, T> for Series
impl<T: IntoSeries> NamedFrom<T, T> for Series
For any ChunkedArray and Series
source§impl NumOpsDispatchChecked for Series
impl NumOpsDispatchChecked for Series
source§fn checked_div(&self, rhs: &Series) -> PolarsResult<Series>
fn checked_div(&self, rhs: &Series) -> PolarsResult<Series>
Checked integer division. Computes self / rhs, returning None if rhs == 0 or the division results in overflow.