arrow_schema/datatype.rs
1// Licensed to the Apache Software Foundation (ASF) under one
2// or more contributor license agreements. See the NOTICE file
3// distributed with this work for additional information
4// regarding copyright ownership. The ASF licenses this file
5// to you under the Apache License, Version 2.0 (the
6// "License"); you may not use this file except in compliance
7// with the License. You may obtain a copy of the License at
8//
9// http://www.apache.org/licenses/LICENSE-2.0
10//
11// Unless required by applicable law or agreed to in writing,
12// software distributed under the License is distributed on an
13// "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
14// KIND, either express or implied. See the License for the
15// specific language governing permissions and limitations
16// under the License.
17
18use std::str::FromStr;
19use std::sync::Arc;
20
21use crate::{ArrowError, Field, FieldRef, Fields, UnionFields};
22
23/// Datatypes supported by this implementation of Apache Arrow.
24///
25/// The variants of this enum include primitive fixed size types as well as
26/// parametric or nested types. See [`Schema.fbs`] for Arrow's specification.
27///
28/// # Examples
29///
30/// Primitive types
31/// ```
32/// # use arrow_schema::DataType;
33/// // create a new 32-bit signed integer
34/// let data_type = DataType::Int32;
35/// ```
36///
37/// Nested Types
38/// ```
39/// # use arrow_schema::{DataType, Field};
40/// # use std::sync::Arc;
41/// // create a new list of 32-bit signed integers directly
42/// let list_data_type = DataType::List(Arc::new(Field::new_list_field(DataType::Int32, true)));
43/// // Create the same list type with constructor
44/// let list_data_type2 = DataType::new_list(DataType::Int32, true);
45/// assert_eq!(list_data_type, list_data_type2);
46/// ```
47///
48/// Dictionary Types
49/// ```
50/// # use arrow_schema::{DataType};
51/// // String Dictionary (key type Int32 and value type Utf8)
52/// let data_type = DataType::Dictionary(Box::new(DataType::Int32), Box::new(DataType::Utf8));
53/// ```
54///
55/// Timestamp Types
56/// ```
57/// # use arrow_schema::{DataType, TimeUnit};
58/// // timestamp with millisecond precision without timezone specified
59/// let data_type = DataType::Timestamp(TimeUnit::Millisecond, None);
60/// // timestamp with nanosecond precision in UTC timezone
61/// let data_type = DataType::Timestamp(TimeUnit::Nanosecond, Some("UTC".into()));
62///```
63///
64/// # Display and FromStr
65///
66/// The `Display` and `FromStr` implementations for `DataType` are
67/// human-readable, parseable, and reversible.
68///
69/// ```
70/// # use arrow_schema::DataType;
71/// let data_type = DataType::Dictionary(Box::new(DataType::Int32), Box::new(DataType::Utf8));
72/// let data_type_string = data_type.to_string();
73/// assert_eq!(data_type_string, "Dictionary(Int32, Utf8)");
74/// // display can be parsed back into the original type
75/// let parsed_data_type: DataType = data_type.to_string().parse().unwrap();
76/// assert_eq!(data_type, parsed_data_type);
77/// ```
78///
79/// # Nested Support
80/// Currently, the Rust implementation supports the following nested types:
81/// - `List<T>`
82/// - `LargeList<T>`
83/// - `FixedSizeList<T>`
84/// - `Struct<T, U, V, ...>`
85/// - `Union<T, U, V, ...>`
86/// - `Map<K, V>`
87///
88/// Nested types can themselves be nested within other arrays.
89/// For more information on these types please see
90/// [the physical memory layout of Apache Arrow]
91///
92/// [`Schema.fbs`]: https://github.com/apache/arrow/blob/main/format/Schema.fbs
93/// [the physical memory layout of Apache Arrow]: https://arrow.apache.org/docs/format/Columnar.html#physical-memory-layout
94#[derive(Clone, Debug, PartialEq, Eq, Hash, PartialOrd, Ord)]
95#[cfg_attr(feature = "serde", derive(serde::Serialize, serde::Deserialize))]
96pub enum DataType {
97 /// Null type
98 Null,
99 /// A boolean datatype representing the values `true` and `false`.
100 Boolean,
101 /// A signed 8-bit integer.
102 Int8,
103 /// A signed 16-bit integer.
104 Int16,
105 /// A signed 32-bit integer.
106 Int32,
107 /// A signed 64-bit integer.
108 Int64,
109 /// An unsigned 8-bit integer.
110 UInt8,
111 /// An unsigned 16-bit integer.
112 UInt16,
113 /// An unsigned 32-bit integer.
114 UInt32,
115 /// An unsigned 64-bit integer.
116 UInt64,
117 /// A 16-bit floating point number.
118 Float16,
119 /// A 32-bit floating point number.
120 Float32,
121 /// A 64-bit floating point number.
122 Float64,
123 /// A timestamp with an optional timezone.
124 ///
125 /// Time is measured as a Unix epoch, counting the seconds from
126 /// 00:00:00.000 on 1 January 1970, excluding leap seconds,
127 /// as a signed 64-bit integer.
128 ///
129 /// The time zone is a string indicating the name of a time zone, one of:
130 ///
131 /// * As used in the Olson time zone database (the "tz database" or
132 /// "tzdata"), such as "America/New_York"
133 /// * An absolute time zone offset of the form +XX:XX or -XX:XX, such as +07:30
134 ///
135 /// Timestamps with a non-empty timezone
136 /// ------------------------------------
137 ///
138 /// If a Timestamp column has a non-empty timezone value, its epoch is
139 /// 1970-01-01 00:00:00 (January 1st 1970, midnight) in the *UTC* timezone
140 /// (the Unix epoch), regardless of the Timestamp's own timezone.
141 ///
142 /// Therefore, timestamp values with a non-empty timezone correspond to
143 /// physical points in time together with some additional information about
144 /// how the data was obtained and/or how to display it (the timezone).
145 ///
146 /// For example, the timestamp value 0 with the timezone string "Europe/Paris"
147 /// corresponds to "January 1st 1970, 00h00" in the UTC timezone, but the
148 /// application may prefer to display it as "January 1st 1970, 01h00" in
149 /// the Europe/Paris timezone (which is the same physical point in time).
150 ///
151 /// One consequence is that timestamp values with a non-empty timezone
152 /// can be compared and ordered directly, since they all share the same
153 /// well-known point of reference (the Unix epoch).
154 ///
155 /// Timestamps with an unset / empty timezone
156 /// -----------------------------------------
157 ///
158 /// If a Timestamp column has no timezone value, its epoch is
159 /// 1970-01-01 00:00:00 (January 1st 1970, midnight) in an *unknown* timezone.
160 ///
161 /// Therefore, timestamp values without a timezone cannot be meaningfully
162 /// interpreted as physical points in time, but only as calendar / clock
163 /// indications ("wall clock time") in an unspecified timezone.
164 ///
165 /// For example, the timestamp value 0 with an empty timezone string
166 /// corresponds to "January 1st 1970, 00h00" in an unknown timezone: there
167 /// is not enough information to interpret it as a well-defined physical
168 /// point in time.
169 ///
170 /// One consequence is that timestamp values without a timezone cannot
171 /// be reliably compared or ordered, since they may have different points of
172 /// reference. In particular, it is *not* possible to interpret an unset
173 /// or empty timezone as the same as "UTC".
174 ///
175 /// Conversion between timezones
176 /// ----------------------------
177 ///
178 /// If a Timestamp column has a non-empty timezone, changing the timezone
179 /// to a different non-empty value is a metadata-only operation:
180 /// the timestamp values need not change as their point of reference remains
181 /// the same (the Unix epoch).
182 ///
183 /// However, if a Timestamp column has no timezone value, changing it to a
184 /// non-empty value requires to think about the desired semantics.
185 /// One possibility is to assume that the original timestamp values are
186 /// relative to the epoch of the timezone being set; timestamp values should
187 /// then adjusted to the Unix epoch (for example, changing the timezone from
188 /// empty to "Europe/Paris" would require converting the timestamp values
189 /// from "Europe/Paris" to "UTC", which seems counter-intuitive but is
190 /// nevertheless correct).
191 ///
192 /// ```
193 /// # use arrow_schema::{DataType, TimeUnit};
194 /// DataType::Timestamp(TimeUnit::Second, None);
195 /// DataType::Timestamp(TimeUnit::Second, Some("literal".into()));
196 /// DataType::Timestamp(TimeUnit::Second, Some("string".to_string().into()));
197 /// ```
198 ///
199 /// # Timezone representation
200 /// ----------------------------
201 /// It is possible to use either the timezone string representation, such as "UTC", or the absolute time zone offset "+00:00".
202 /// For timezones with fixed offsets, such as "UTC" or "JST", the offset representation is recommended, as it is more explicit and less ambiguous.
203 ///
204 /// Most arrow-rs functionalities use the absolute offset representation,
205 /// such as [`PrimitiveArray::with_timezone_utc`] that applies a
206 /// UTC timezone to timestamp arrays.
207 ///
208 /// [`PrimitiveArray::with_timezone_utc`]: https://docs.rs/arrow/latest/arrow/array/struct.PrimitiveArray.html#method.with_timezone_utc
209 ///
210 /// Timezone string parsing
211 /// -----------------------
212 /// When feature `chrono-tz` is not enabled, allowed timezone strings are fixed offsets of the form "+09:00", "-09" or "+0930".
213 ///
214 /// When feature `chrono-tz` is enabled, additional strings supported by [chrono_tz](https://docs.rs/chrono-tz/latest/chrono_tz/)
215 /// are also allowed, which include [IANA database](https://en.wikipedia.org/wiki/List_of_tz_database_time_zones)
216 /// timezones.
217 Timestamp(TimeUnit, Option<Arc<str>>),
218 /// A signed 32-bit date representing the elapsed time since UNIX epoch (1970-01-01)
219 /// in days.
220 Date32,
221 /// A signed 64-bit date representing the elapsed time since UNIX epoch (1970-01-01)
222 /// in milliseconds.
223 ///
224 /// # Valid Ranges
225 ///
226 /// According to the Arrow specification ([Schema.fbs]), values of Date64
227 /// are treated as the number of *days*, in milliseconds, since the UNIX
228 /// epoch. Therefore, values of this type must be evenly divisible by
229 /// `86_400_000`, the number of milliseconds in a standard day.
230 ///
231 /// It is not valid to store milliseconds that do not represent an exact
232 /// day. The reason for this restriction is compatibility with other
233 /// language's native libraries (specifically Java), which historically
234 /// lacked a dedicated date type and only supported timestamps.
235 ///
236 /// # Validation
237 ///
238 /// This library does not validate or enforce that Date64 values are evenly
239 /// divisible by `86_400_000` for performance and usability reasons. Date64
240 /// values are treated similarly to `Timestamp(TimeUnit::Millisecond,
241 /// None)`: values will be displayed with a time of day if the value does
242 /// not represent an exact day, and arithmetic will be done at the
243 /// millisecond granularity.
244 ///
245 /// # Recommendation
246 ///
247 /// Users should prefer [`Date32`] to cleanly represent the number
248 /// of days, or one of the Timestamp variants to include time as part of the
249 /// representation, depending on their use case.
250 ///
251 /// # Further Reading
252 ///
253 /// For more details, see [#5288](https://github.com/apache/arrow-rs/issues/5288).
254 ///
255 /// [`Date32`]: Self::Date32
256 /// [Schema.fbs]: https://github.com/apache/arrow/blob/main/format/Schema.fbs
257 Date64,
258 /// A signed 32-bit time representing the elapsed time since midnight in the unit of `TimeUnit`.
259 /// Must be either seconds or milliseconds.
260 Time32(TimeUnit),
261 /// A signed 64-bit time representing the elapsed time since midnight in the unit of `TimeUnit`.
262 /// Must be either microseconds or nanoseconds.
263 Time64(TimeUnit),
264 /// Measure of elapsed time in either seconds, milliseconds, microseconds or nanoseconds.
265 Duration(TimeUnit),
266 /// A "calendar" interval which models types that don't necessarily
267 /// have a precise duration without the context of a base timestamp (e.g.
268 /// days can differ in length during day light savings time transitions).
269 Interval(IntervalUnit),
270 /// Opaque binary data of variable length.
271 ///
272 /// A single Binary array can store up to [`i32::MAX`] bytes
273 /// of binary data in total.
274 Binary,
275 /// Opaque binary data of fixed size.
276 /// Enum parameter specifies the number of bytes per value.
277 FixedSizeBinary(i32),
278 /// Opaque binary data of variable length and 64-bit offsets.
279 ///
280 /// A single LargeBinary array can store up to [`i64::MAX`] bytes
281 /// of binary data in total.
282 LargeBinary,
283 /// Opaque binary data of variable length.
284 ///
285 /// Logically the same as [`Binary`], but the internal representation uses a view
286 /// struct that contains the string length and either the string's entire data
287 /// inline (for small strings) or an inlined prefix, an index of another buffer,
288 /// and an offset pointing to a slice in that buffer (for non-small strings).
289 ///
290 /// [`Binary`]: Self::Binary
291 BinaryView,
292 /// A variable-length string in Unicode with UTF-8 encoding.
293 ///
294 /// A single Utf8 array can store up to [`i32::MAX`] bytes
295 /// of string data in total.
296 Utf8,
297 /// A variable-length string in Unicode with UFT-8 encoding and 64-bit offsets.
298 ///
299 /// A single LargeUtf8 array can store up to [`i64::MAX`] bytes
300 /// of string data in total.
301 LargeUtf8,
302 /// A variable-length string in Unicode with UTF-8 encoding
303 ///
304 /// Logically the same as [`Utf8`], but the internal representation uses a view
305 /// struct that contains the string length and either the string's entire data
306 /// inline (for small strings) or an inlined prefix, an index of another buffer,
307 /// and an offset pointing to a slice in that buffer (for non-small strings).
308 ///
309 /// [`Utf8`]: Self::Utf8
310 Utf8View,
311 /// A list of some logical data type with variable length.
312 ///
313 /// A single List array can store up to [`i32::MAX`] elements in total.
314 List(FieldRef),
315
316 /// (NOT YET FULLY SUPPORTED) A list of some logical data type with variable length.
317 ///
318 /// Logically the same as [`List`], but the internal representation differs in how child
319 /// data is referenced, allowing flexibility in how data is layed out.
320 ///
321 /// Note this data type is not yet fully supported. Using it with arrow APIs may result in `panic`s.
322 ///
323 /// [`List`]: Self::List
324 ListView(FieldRef),
325 /// A list of some logical data type with fixed length.
326 FixedSizeList(FieldRef, i32),
327 /// A list of some logical data type with variable length and 64-bit offsets.
328 ///
329 /// A single LargeList array can store up to [`i64::MAX`] elements in total.
330 LargeList(FieldRef),
331
332 /// (NOT YET FULLY SUPPORTED) A list of some logical data type with variable length and 64-bit offsets.
333 ///
334 /// Logically the same as [`LargeList`], but the internal representation differs in how child
335 /// data is referenced, allowing flexibility in how data is layed out.
336 ///
337 /// Note this data type is not yet fully supported. Using it with arrow APIs may result in `panic`s.
338 ///
339 /// [`LargeList`]: Self::LargeList
340 LargeListView(FieldRef),
341 /// A nested datatype that contains a number of sub-fields.
342 Struct(Fields),
343 /// A nested datatype that can represent slots of differing types. Components:
344 ///
345 /// 1. [`UnionFields`]
346 /// 2. The type of union (Sparse or Dense)
347 Union(UnionFields, UnionMode),
348 /// A dictionary encoded array (`key_type`, `value_type`), where
349 /// each array element is an index of `key_type` into an
350 /// associated dictionary of `value_type`.
351 ///
352 /// Dictionary arrays are used to store columns of `value_type`
353 /// that contain many repeated values using less memory, but with
354 /// a higher CPU overhead for some operations.
355 ///
356 /// This type mostly used to represent low cardinality string
357 /// arrays or a limited set of primitive types as integers.
358 Dictionary(Box<DataType>, Box<DataType>),
359 /// Exact 32-bit width decimal value with precision and scale
360 ///
361 /// * precision is the total number of digits
362 /// * scale is the number of digits past the decimal
363 ///
364 /// For example the number 123.45 has precision 5 and scale 2.
365 ///
366 /// In certain situations, scale could be negative number. For
367 /// negative scale, it is the number of padding 0 to the right
368 /// of the digits.
369 ///
370 /// For example the number 12300 could be treated as a decimal
371 /// has precision 3 and scale -2.
372 Decimal32(u8, i8),
373 /// Exact 64-bit width decimal value with precision and scale
374 ///
375 /// * precision is the total number of digits
376 /// * scale is the number of digits past the decimal
377 ///
378 /// For example the number 123.45 has precision 5 and scale 2.
379 ///
380 /// In certain situations, scale could be negative number. For
381 /// negative scale, it is the number of padding 0 to the right
382 /// of the digits.
383 ///
384 /// For example the number 12300 could be treated as a decimal
385 /// has precision 3 and scale -2.
386 Decimal64(u8, i8),
387 /// Exact 128-bit width decimal value with precision and scale
388 ///
389 /// * precision is the total number of digits
390 /// * scale is the number of digits past the decimal
391 ///
392 /// For example the number 123.45 has precision 5 and scale 2.
393 ///
394 /// In certain situations, scale could be negative number. For
395 /// negative scale, it is the number of padding 0 to the right
396 /// of the digits.
397 ///
398 /// For example the number 12300 could be treated as a decimal
399 /// has precision 3 and scale -2.
400 Decimal128(u8, i8),
401 /// Exact 256-bit width decimal value with precision and scale
402 ///
403 /// * precision is the total number of digits
404 /// * scale is the number of digits past the decimal
405 ///
406 /// For example the number 123.45 has precision 5 and scale 2.
407 ///
408 /// In certain situations, scale could be negative number. For
409 /// negative scale, it is the number of padding 0 to the right
410 /// of the digits.
411 ///
412 /// For example the number 12300 could be treated as a decimal
413 /// has precision 3 and scale -2.
414 Decimal256(u8, i8),
415 /// A Map is a logical nested type that is represented as
416 ///
417 /// `List<entries: Struct<key: K, value: V>>`
418 ///
419 /// The keys and values are each respectively contiguous.
420 /// The key and value types are not constrained, but keys should be
421 /// hashable and unique.
422 /// Whether the keys are sorted can be set in the `bool` after the `Field`.
423 ///
424 /// In a field with Map type, the field has a child Struct field, which then
425 /// has two children: key type and the second the value type. The names of the
426 /// child fields may be respectively "entries", "key", and "value", but this is
427 /// not enforced.
428 Map(FieldRef, bool),
429 /// A run-end encoding (REE) is a variation of run-length encoding (RLE). These
430 /// encodings are well-suited for representing data containing sequences of the
431 /// same value, called runs. Each run is represented as a value and an integer giving
432 /// the index in the array where the run ends.
433 ///
434 /// A run-end encoded array has no buffers by itself, but has two child arrays. The
435 /// first child array, called the run ends array, holds either 16, 32, or 64-bit
436 /// signed integers. The actual values of each run are held in the second child array.
437 ///
438 /// These child arrays are prescribed the standard names of "run_ends" and "values"
439 /// respectively.
440 RunEndEncoded(FieldRef, FieldRef),
441}
442
443/// An absolute length of time in seconds, milliseconds, microseconds or nanoseconds.
444#[derive(Debug, Clone, Copy, PartialEq, Eq, Hash, PartialOrd, Ord)]
445#[cfg_attr(feature = "serde", derive(serde::Serialize, serde::Deserialize))]
446pub enum TimeUnit {
447 /// Time in seconds.
448 Second,
449 /// Time in milliseconds.
450 Millisecond,
451 /// Time in microseconds.
452 Microsecond,
453 /// Time in nanoseconds.
454 Nanosecond,
455}
456
457impl std::fmt::Display for TimeUnit {
458 fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
459 match self {
460 TimeUnit::Second => write!(f, "s"),
461 TimeUnit::Millisecond => write!(f, "ms"),
462 TimeUnit::Microsecond => write!(f, "µs"),
463 TimeUnit::Nanosecond => write!(f, "ns"),
464 }
465 }
466}
467
468/// YEAR_MONTH, DAY_TIME, MONTH_DAY_NANO interval in SQL style.
469#[derive(Debug, Clone, Copy, PartialEq, Eq, Hash, PartialOrd, Ord)]
470#[cfg_attr(feature = "serde", derive(serde::Serialize, serde::Deserialize))]
471pub enum IntervalUnit {
472 /// Indicates the number of elapsed whole months, stored as 4-byte integers.
473 YearMonth,
474 /// Indicates the number of elapsed days and milliseconds,
475 /// stored as 2 contiguous 32-bit integers (days, milliseconds) (8-bytes in total).
476 DayTime,
477 /// A triple of the number of elapsed months, days, and nanoseconds.
478 /// The values are stored contiguously in 16 byte blocks. Months and
479 /// days are encoded as 32 bit integers and nanoseconds is encoded as a
480 /// 64 bit integer. All integers are signed. Each field is independent
481 /// (e.g. there is no constraint that nanoseconds have the same sign
482 /// as days or that the quantity of nanoseconds represents less
483 /// than a day's worth of time).
484 MonthDayNano,
485}
486
487/// Sparse or Dense union layouts
488#[derive(Debug, Clone, PartialEq, Eq, Hash, PartialOrd, Ord, Copy)]
489#[cfg_attr(feature = "serde", derive(serde::Serialize, serde::Deserialize))]
490pub enum UnionMode {
491 /// Sparse union layout
492 Sparse,
493 /// Dense union layout
494 Dense,
495}
496
497/// Parses `str` into a `DataType`.
498///
499/// This is the reverse of [`DataType`]'s `Display`
500/// impl, and maintains the invariant that
501/// `DataType::try_from(&data_type.to_string()).unwrap() == data_type`
502///
503/// # Example
504/// ```
505/// use arrow_schema::DataType;
506///
507/// let data_type: DataType = "Int32".parse().unwrap();
508/// assert_eq!(data_type, DataType::Int32);
509/// ```
510impl FromStr for DataType {
511 type Err = ArrowError;
512
513 fn from_str(s: &str) -> Result<Self, Self::Err> {
514 crate::datatype_parse::parse_data_type(s)
515 }
516}
517
518impl TryFrom<&str> for DataType {
519 type Error = ArrowError;
520
521 fn try_from(value: &str) -> Result<Self, Self::Error> {
522 value.parse()
523 }
524}
525
526impl DataType {
527 /// Returns true if the type is primitive: (numeric, temporal).
528 #[inline]
529 pub fn is_primitive(&self) -> bool {
530 self.is_numeric() || self.is_temporal()
531 }
532
533 /// Returns true if this type is numeric: (UInt*, Int*, Float*, Decimal*).
534 #[inline]
535 pub fn is_numeric(&self) -> bool {
536 use DataType::*;
537 matches!(
538 self,
539 UInt8
540 | UInt16
541 | UInt32
542 | UInt64
543 | Int8
544 | Int16
545 | Int32
546 | Int64
547 | Float16
548 | Float32
549 | Float64
550 | Decimal32(_, _)
551 | Decimal64(_, _)
552 | Decimal128(_, _)
553 | Decimal256(_, _)
554 )
555 }
556
557 /// Returns true if this type is temporal: (Date*, Time*, Duration, or Interval).
558 #[inline]
559 pub fn is_temporal(&self) -> bool {
560 use DataType::*;
561 matches!(
562 self,
563 Date32 | Date64 | Timestamp(_, _) | Time32(_) | Time64(_) | Duration(_) | Interval(_)
564 )
565 }
566
567 /// Returns true if this type is floating: (Float*).
568 #[inline]
569 pub fn is_floating(&self) -> bool {
570 use DataType::*;
571 matches!(self, Float16 | Float32 | Float64)
572 }
573
574 /// Returns true if this type is integer: (Int*, UInt*).
575 #[inline]
576 pub fn is_integer(&self) -> bool {
577 self.is_signed_integer() || self.is_unsigned_integer()
578 }
579
580 /// Returns true if this type is signed integer: (Int*).
581 #[inline]
582 pub fn is_signed_integer(&self) -> bool {
583 use DataType::*;
584 matches!(self, Int8 | Int16 | Int32 | Int64)
585 }
586
587 /// Returns true if this type is unsigned integer: (UInt*).
588 #[inline]
589 pub fn is_unsigned_integer(&self) -> bool {
590 use DataType::*;
591 matches!(self, UInt8 | UInt16 | UInt32 | UInt64)
592 }
593
594 /// Returns true if this type is decimal: (Decimal*).
595 #[inline]
596 pub fn is_decimal(&self) -> bool {
597 use DataType::*;
598 matches!(
599 self,
600 Decimal32(..) | Decimal64(..) | Decimal128(..) | Decimal256(..)
601 )
602 }
603
604 /// Returns true if this type is valid as a dictionary key
605 #[inline]
606 pub fn is_dictionary_key_type(&self) -> bool {
607 self.is_integer()
608 }
609
610 /// Returns true if this type is valid for run-ends array in RunArray
611 #[inline]
612 pub fn is_run_ends_type(&self) -> bool {
613 use DataType::*;
614 matches!(self, Int16 | Int32 | Int64)
615 }
616
617 /// Returns true if this type is nested (List, FixedSizeList, LargeList, ListView. LargeListView, Struct, Union,
618 /// or Map), or a dictionary of a nested type
619 #[inline]
620 pub fn is_nested(&self) -> bool {
621 use DataType::*;
622 match self {
623 Dictionary(_, v) => DataType::is_nested(v.as_ref()),
624 RunEndEncoded(_, v) => DataType::is_nested(v.data_type()),
625 List(_)
626 | FixedSizeList(_, _)
627 | LargeList(_)
628 | ListView(_)
629 | LargeListView(_)
630 | Struct(_)
631 | Union(_, _)
632 | Map(_, _) => true,
633 _ => false,
634 }
635 }
636
637 /// Returns true if this type is DataType::Null.
638 #[inline]
639 pub fn is_null(&self) -> bool {
640 use DataType::*;
641 matches!(self, Null)
642 }
643
644 /// Returns true if this type is a String type
645 #[inline]
646 pub fn is_string(&self) -> bool {
647 use DataType::*;
648 matches!(self, Utf8 | LargeUtf8 | Utf8View)
649 }
650
651 /// Compares the datatype with another, ignoring nested field names
652 /// and metadata.
653 pub fn equals_datatype(&self, other: &DataType) -> bool {
654 match (&self, other) {
655 (DataType::List(a), DataType::List(b))
656 | (DataType::LargeList(a), DataType::LargeList(b))
657 | (DataType::ListView(a), DataType::ListView(b))
658 | (DataType::LargeListView(a), DataType::LargeListView(b)) => {
659 a.is_nullable() == b.is_nullable() && a.data_type().equals_datatype(b.data_type())
660 }
661 (DataType::FixedSizeList(a, a_size), DataType::FixedSizeList(b, b_size)) => {
662 a_size == b_size
663 && a.is_nullable() == b.is_nullable()
664 && a.data_type().equals_datatype(b.data_type())
665 }
666 (DataType::Struct(a), DataType::Struct(b)) => {
667 a.len() == b.len()
668 && a.iter().zip(b).all(|(a, b)| {
669 a.is_nullable() == b.is_nullable()
670 && a.data_type().equals_datatype(b.data_type())
671 })
672 }
673 (DataType::Map(a_field, a_is_sorted), DataType::Map(b_field, b_is_sorted)) => {
674 a_field.is_nullable() == b_field.is_nullable()
675 && a_field.data_type().equals_datatype(b_field.data_type())
676 && a_is_sorted == b_is_sorted
677 }
678 (DataType::Dictionary(a_key, a_value), DataType::Dictionary(b_key, b_value)) => {
679 a_key.equals_datatype(b_key) && a_value.equals_datatype(b_value)
680 }
681 (
682 DataType::RunEndEncoded(a_run_ends, a_values),
683 DataType::RunEndEncoded(b_run_ends, b_values),
684 ) => {
685 a_run_ends.is_nullable() == b_run_ends.is_nullable()
686 && a_run_ends
687 .data_type()
688 .equals_datatype(b_run_ends.data_type())
689 && a_values.is_nullable() == b_values.is_nullable()
690 && a_values.data_type().equals_datatype(b_values.data_type())
691 }
692 (
693 DataType::Union(a_union_fields, a_union_mode),
694 DataType::Union(b_union_fields, b_union_mode),
695 ) => {
696 a_union_mode == b_union_mode
697 && a_union_fields.len() == b_union_fields.len()
698 && a_union_fields.iter().all(|a| {
699 b_union_fields.iter().any(|b| {
700 a.0 == b.0
701 && a.1.is_nullable() == b.1.is_nullable()
702 && a.1.data_type().equals_datatype(b.1.data_type())
703 })
704 })
705 }
706 _ => self == other,
707 }
708 }
709
710 /// Returns the byte width of this type if it is a primitive type
711 ///
712 /// Returns `None` if not a primitive type
713 #[inline]
714 pub fn primitive_width(&self) -> Option<usize> {
715 match self {
716 DataType::Null => None,
717 DataType::Boolean => None,
718 DataType::Int8 | DataType::UInt8 => Some(1),
719 DataType::Int16 | DataType::UInt16 | DataType::Float16 => Some(2),
720 DataType::Int32 | DataType::UInt32 | DataType::Float32 => Some(4),
721 DataType::Int64 | DataType::UInt64 | DataType::Float64 => Some(8),
722 DataType::Timestamp(_, _) => Some(8),
723 DataType::Date32 | DataType::Time32(_) => Some(4),
724 DataType::Date64 | DataType::Time64(_) => Some(8),
725 DataType::Duration(_) => Some(8),
726 DataType::Interval(IntervalUnit::YearMonth) => Some(4),
727 DataType::Interval(IntervalUnit::DayTime) => Some(8),
728 DataType::Interval(IntervalUnit::MonthDayNano) => Some(16),
729 DataType::Decimal32(_, _) => Some(4),
730 DataType::Decimal64(_, _) => Some(8),
731 DataType::Decimal128(_, _) => Some(16),
732 DataType::Decimal256(_, _) => Some(32),
733 DataType::Utf8 | DataType::LargeUtf8 | DataType::Utf8View => None,
734 DataType::Binary | DataType::LargeBinary | DataType::BinaryView => None,
735 DataType::FixedSizeBinary(_) => None,
736 DataType::List(_)
737 | DataType::ListView(_)
738 | DataType::LargeList(_)
739 | DataType::LargeListView(_)
740 | DataType::Map(_, _) => None,
741 DataType::FixedSizeList(_, _) => None,
742 DataType::Struct(_) => None,
743 DataType::Union(_, _) => None,
744 DataType::Dictionary(_, _) => None,
745 DataType::RunEndEncoded(_, _) => None,
746 }
747 }
748
749 /// Return size of this instance in bytes.
750 ///
751 /// Includes the size of `Self`.
752 pub fn size(&self) -> usize {
753 std::mem::size_of_val(self)
754 + match self {
755 DataType::Null
756 | DataType::Boolean
757 | DataType::Int8
758 | DataType::Int16
759 | DataType::Int32
760 | DataType::Int64
761 | DataType::UInt8
762 | DataType::UInt16
763 | DataType::UInt32
764 | DataType::UInt64
765 | DataType::Float16
766 | DataType::Float32
767 | DataType::Float64
768 | DataType::Date32
769 | DataType::Date64
770 | DataType::Time32(_)
771 | DataType::Time64(_)
772 | DataType::Duration(_)
773 | DataType::Interval(_)
774 | DataType::Binary
775 | DataType::FixedSizeBinary(_)
776 | DataType::LargeBinary
777 | DataType::BinaryView
778 | DataType::Utf8
779 | DataType::LargeUtf8
780 | DataType::Utf8View
781 | DataType::Decimal32(_, _)
782 | DataType::Decimal64(_, _)
783 | DataType::Decimal128(_, _)
784 | DataType::Decimal256(_, _) => 0,
785 DataType::Timestamp(_, s) => s.as_ref().map(|s| s.len()).unwrap_or_default(),
786 DataType::List(field)
787 | DataType::ListView(field)
788 | DataType::FixedSizeList(field, _)
789 | DataType::LargeList(field)
790 | DataType::LargeListView(field)
791 | DataType::Map(field, _) => field.size(),
792 DataType::Struct(fields) => fields.size(),
793 DataType::Union(fields, _) => fields.size(),
794 DataType::Dictionary(dt1, dt2) => dt1.size() + dt2.size(),
795 DataType::RunEndEncoded(run_ends, values) => {
796 run_ends.size() - std::mem::size_of_val(run_ends) + values.size()
797 - std::mem::size_of_val(values)
798 }
799 }
800 }
801
802 /// Check to see if `self` is a superset of `other`
803 ///
804 /// If DataType is a nested type, then it will check to see if the nested type is a superset of the other nested type
805 /// else it will check to see if the DataType is equal to the other DataType
806 pub fn contains(&self, other: &DataType) -> bool {
807 match (self, other) {
808 (DataType::List(f1), DataType::List(f2))
809 | (DataType::LargeList(f1), DataType::LargeList(f2))
810 | (DataType::ListView(f1), DataType::ListView(f2))
811 | (DataType::LargeListView(f1), DataType::LargeListView(f2)) => f1.contains(f2),
812 (DataType::FixedSizeList(f1, s1), DataType::FixedSizeList(f2, s2)) => {
813 s1 == s2 && f1.contains(f2)
814 }
815 (DataType::Map(f1, s1), DataType::Map(f2, s2)) => s1 == s2 && f1.contains(f2),
816 (DataType::Struct(f1), DataType::Struct(f2)) => f1.contains(f2),
817 (DataType::Union(f1, s1), DataType::Union(f2, s2)) => {
818 s1 == s2
819 && f1
820 .iter()
821 .all(|f1| f2.iter().any(|f2| f1.0 == f2.0 && f1.1.contains(f2.1)))
822 }
823 (DataType::Dictionary(k1, v1), DataType::Dictionary(k2, v2)) => {
824 k1.contains(k2) && v1.contains(v2)
825 }
826 _ => self == other,
827 }
828 }
829
830 /// Create a [`DataType::List`] with elements of the specified type
831 /// and nullability, and conventionally named inner [`Field`] (`"item"`).
832 ///
833 /// To specify field level metadata, construct the inner [`Field`]
834 /// directly via [`Field::new`] or [`Field::new_list_field`].
835 pub fn new_list(data_type: DataType, nullable: bool) -> Self {
836 DataType::List(Arc::new(Field::new_list_field(data_type, nullable)))
837 }
838
839 /// Create a [`DataType::LargeList`] with elements of the specified type
840 /// and nullability, and conventionally named inner [`Field`] (`"item"`).
841 ///
842 /// To specify field level metadata, construct the inner [`Field`]
843 /// directly via [`Field::new`] or [`Field::new_list_field`].
844 pub fn new_large_list(data_type: DataType, nullable: bool) -> Self {
845 DataType::LargeList(Arc::new(Field::new_list_field(data_type, nullable)))
846 }
847
848 /// Create a [`DataType::FixedSizeList`] with elements of the specified type, size
849 /// and nullability, and conventionally named inner [`Field`] (`"item"`).
850 ///
851 /// To specify field level metadata, construct the inner [`Field`]
852 /// directly via [`Field::new`] or [`Field::new_list_field`].
853 pub fn new_fixed_size_list(data_type: DataType, size: i32, nullable: bool) -> Self {
854 DataType::FixedSizeList(Arc::new(Field::new_list_field(data_type, nullable)), size)
855 }
856}
857
858/// The maximum precision for [DataType::Decimal32] values
859pub const DECIMAL32_MAX_PRECISION: u8 = 9;
860
861/// The maximum scale for [DataType::Decimal32] values
862pub const DECIMAL32_MAX_SCALE: i8 = 9;
863
864/// The maximum precision for [DataType::Decimal64] values
865pub const DECIMAL64_MAX_PRECISION: u8 = 18;
866
867/// The maximum scale for [DataType::Decimal64] values
868pub const DECIMAL64_MAX_SCALE: i8 = 18;
869
870/// The maximum precision for [DataType::Decimal128] values
871pub const DECIMAL128_MAX_PRECISION: u8 = 38;
872
873/// The maximum scale for [DataType::Decimal128] values
874pub const DECIMAL128_MAX_SCALE: i8 = 38;
875
876/// The maximum precision for [DataType::Decimal256] values
877pub const DECIMAL256_MAX_PRECISION: u8 = 76;
878
879/// The maximum scale for [DataType::Decimal256] values
880pub const DECIMAL256_MAX_SCALE: i8 = 76;
881
882/// The default scale for [DataType::Decimal32] values
883pub const DECIMAL32_DEFAULT_SCALE: i8 = 2;
884
885/// The default scale for [DataType::Decimal64] values
886pub const DECIMAL64_DEFAULT_SCALE: i8 = 6;
887
888/// The default scale for [DataType::Decimal128] and [DataType::Decimal256]
889/// values
890pub const DECIMAL_DEFAULT_SCALE: i8 = 10;
891
892#[cfg(test)]
893mod tests {
894 use super::*;
895
896 #[test]
897 #[cfg(feature = "serde")]
898 fn serde_struct_type() {
899 use std::collections::HashMap;
900
901 let kv_array = [("k".to_string(), "v".to_string())];
902 let field_metadata: HashMap<String, String> = kv_array.iter().cloned().collect();
903
904 // Non-empty map: should be converted as JSON obj { ... }
905 let first_name =
906 Field::new("first_name", DataType::Utf8, false).with_metadata(field_metadata);
907
908 // Empty map: should be omitted.
909 let last_name =
910 Field::new("last_name", DataType::Utf8, false).with_metadata(HashMap::default());
911
912 let person = DataType::Struct(Fields::from(vec![
913 first_name,
914 last_name,
915 Field::new(
916 "address",
917 DataType::Struct(Fields::from(vec![
918 Field::new("street", DataType::Utf8, false),
919 Field::new("zip", DataType::UInt16, false),
920 ])),
921 false,
922 ),
923 ]));
924
925 let serialized = serde_json::to_string(&person).unwrap();
926
927 // NOTE that this is testing the default (derived) serialization format, not the
928 // JSON format specified in metadata.md
929
930 assert_eq!(
931 "{\"Struct\":[\
932 {\"name\":\"first_name\",\"data_type\":\"Utf8\",\"nullable\":false,\"dict_id\":0,\"dict_is_ordered\":false,\"metadata\":{\"k\":\"v\"}},\
933 {\"name\":\"last_name\",\"data_type\":\"Utf8\",\"nullable\":false,\"dict_id\":0,\"dict_is_ordered\":false,\"metadata\":{}},\
934 {\"name\":\"address\",\"data_type\":{\"Struct\":\
935 [{\"name\":\"street\",\"data_type\":\"Utf8\",\"nullable\":false,\"dict_id\":0,\"dict_is_ordered\":false,\"metadata\":{}},\
936 {\"name\":\"zip\",\"data_type\":\"UInt16\",\"nullable\":false,\"dict_id\":0,\"dict_is_ordered\":false,\"metadata\":{}}\
937 ]},\"nullable\":false,\"dict_id\":0,\"dict_is_ordered\":false,\"metadata\":{}}]}",
938 serialized
939 );
940
941 let deserialized = serde_json::from_str(&serialized).unwrap();
942
943 assert_eq!(person, deserialized);
944 }
945
946 #[test]
947 fn test_list_datatype_equality() {
948 // tests that list type equality is checked while ignoring list names
949 let list_a = DataType::List(Arc::new(Field::new_list_field(DataType::Int32, true)));
950 let list_b = DataType::List(Arc::new(Field::new("array", DataType::Int32, true)));
951 let list_c = DataType::List(Arc::new(Field::new_list_field(DataType::Int32, false)));
952 let list_d = DataType::List(Arc::new(Field::new_list_field(DataType::UInt32, true)));
953 assert!(list_a.equals_datatype(&list_b));
954 assert!(!list_a.equals_datatype(&list_c));
955 assert!(!list_b.equals_datatype(&list_c));
956 assert!(!list_a.equals_datatype(&list_d));
957
958 let list_e =
959 DataType::FixedSizeList(Arc::new(Field::new_list_field(list_a.clone(), false)), 3);
960 let list_f =
961 DataType::FixedSizeList(Arc::new(Field::new("array", list_b.clone(), false)), 3);
962 let list_g = DataType::FixedSizeList(
963 Arc::new(Field::new_list_field(DataType::FixedSizeBinary(3), true)),
964 3,
965 );
966 assert!(list_e.equals_datatype(&list_f));
967 assert!(!list_e.equals_datatype(&list_g));
968 assert!(!list_f.equals_datatype(&list_g));
969
970 let list_h = DataType::Struct(Fields::from(vec![Field::new("f1", list_e, true)]));
971 let list_i = DataType::Struct(Fields::from(vec![Field::new("f1", list_f.clone(), true)]));
972 let list_j = DataType::Struct(Fields::from(vec![Field::new("f1", list_f.clone(), false)]));
973 let list_k = DataType::Struct(Fields::from(vec![
974 Field::new("f1", list_f.clone(), false),
975 Field::new("f2", list_g.clone(), false),
976 Field::new("f3", DataType::Utf8, true),
977 ]));
978 let list_l = DataType::Struct(Fields::from(vec![
979 Field::new("ff1", list_f.clone(), false),
980 Field::new("ff2", list_g.clone(), false),
981 Field::new("ff3", DataType::LargeUtf8, true),
982 ]));
983 let list_m = DataType::Struct(Fields::from(vec![
984 Field::new("ff1", list_f, false),
985 Field::new("ff2", list_g, false),
986 Field::new("ff3", DataType::Utf8, true),
987 ]));
988 assert!(list_h.equals_datatype(&list_i));
989 assert!(!list_h.equals_datatype(&list_j));
990 assert!(!list_k.equals_datatype(&list_l));
991 assert!(list_k.equals_datatype(&list_m));
992
993 let list_n = DataType::Map(Arc::new(Field::new("f1", list_a.clone(), true)), true);
994 let list_o = DataType::Map(Arc::new(Field::new("f2", list_b.clone(), true)), true);
995 let list_p = DataType::Map(Arc::new(Field::new("f2", list_b.clone(), true)), false);
996 let list_q = DataType::Map(Arc::new(Field::new("f2", list_c.clone(), true)), true);
997 let list_r = DataType::Map(Arc::new(Field::new("f1", list_a.clone(), false)), true);
998
999 assert!(list_n.equals_datatype(&list_o));
1000 assert!(!list_n.equals_datatype(&list_p));
1001 assert!(!list_n.equals_datatype(&list_q));
1002 assert!(!list_n.equals_datatype(&list_r));
1003
1004 let list_s = DataType::Dictionary(Box::new(DataType::UInt8), Box::new(list_a));
1005 let list_t = DataType::Dictionary(Box::new(DataType::UInt8), Box::new(list_b.clone()));
1006 let list_u = DataType::Dictionary(Box::new(DataType::Int8), Box::new(list_b));
1007 let list_v = DataType::Dictionary(Box::new(DataType::UInt8), Box::new(list_c));
1008
1009 assert!(list_s.equals_datatype(&list_t));
1010 assert!(!list_s.equals_datatype(&list_u));
1011 assert!(!list_s.equals_datatype(&list_v));
1012
1013 let union_a = DataType::Union(
1014 UnionFields::try_new(
1015 vec![1, 2],
1016 vec![
1017 Field::new("f1", DataType::Utf8, false),
1018 Field::new("f2", DataType::UInt8, false),
1019 ],
1020 )
1021 .unwrap(),
1022 UnionMode::Sparse,
1023 );
1024 let union_b = DataType::Union(
1025 UnionFields::try_new(
1026 vec![1, 2],
1027 vec![
1028 Field::new("ff1", DataType::Utf8, false),
1029 Field::new("ff2", DataType::UInt8, false),
1030 ],
1031 )
1032 .unwrap(),
1033 UnionMode::Sparse,
1034 );
1035 let union_c = DataType::Union(
1036 UnionFields::try_new(
1037 vec![2, 1],
1038 vec![
1039 Field::new("fff2", DataType::UInt8, false),
1040 Field::new("fff1", DataType::Utf8, false),
1041 ],
1042 )
1043 .unwrap(),
1044 UnionMode::Sparse,
1045 );
1046 let union_d = DataType::Union(
1047 UnionFields::try_new(
1048 vec![2, 1],
1049 vec![
1050 Field::new("fff1", DataType::Int8, false),
1051 Field::new("fff2", DataType::UInt8, false),
1052 ],
1053 )
1054 .unwrap(),
1055 UnionMode::Sparse,
1056 );
1057 let union_e = DataType::Union(
1058 UnionFields::try_new(
1059 vec![1, 2],
1060 vec![
1061 Field::new("f1", DataType::Utf8, true),
1062 Field::new("f2", DataType::UInt8, false),
1063 ],
1064 )
1065 .unwrap(),
1066 UnionMode::Sparse,
1067 );
1068
1069 assert!(union_a.equals_datatype(&union_b));
1070 assert!(union_a.equals_datatype(&union_c));
1071 assert!(!union_a.equals_datatype(&union_d));
1072 assert!(!union_a.equals_datatype(&union_e));
1073
1074 let list_w = DataType::RunEndEncoded(
1075 Arc::new(Field::new("f1", DataType::Int64, true)),
1076 Arc::new(Field::new("f2", DataType::Utf8, true)),
1077 );
1078 let list_x = DataType::RunEndEncoded(
1079 Arc::new(Field::new("ff1", DataType::Int64, true)),
1080 Arc::new(Field::new("ff2", DataType::Utf8, true)),
1081 );
1082 let list_y = DataType::RunEndEncoded(
1083 Arc::new(Field::new("ff1", DataType::UInt16, true)),
1084 Arc::new(Field::new("ff2", DataType::Utf8, true)),
1085 );
1086 let list_z = DataType::RunEndEncoded(
1087 Arc::new(Field::new("f1", DataType::Int64, false)),
1088 Arc::new(Field::new("f2", DataType::Utf8, true)),
1089 );
1090
1091 assert!(list_w.equals_datatype(&list_x));
1092 assert!(!list_w.equals_datatype(&list_y));
1093 assert!(!list_w.equals_datatype(&list_z));
1094 }
1095
1096 #[test]
1097 fn create_struct_type() {
1098 let _person = DataType::Struct(Fields::from(vec![
1099 Field::new("first_name", DataType::Utf8, false),
1100 Field::new("last_name", DataType::Utf8, false),
1101 Field::new(
1102 "address",
1103 DataType::Struct(Fields::from(vec![
1104 Field::new("street", DataType::Utf8, false),
1105 Field::new("zip", DataType::UInt16, false),
1106 ])),
1107 false,
1108 ),
1109 ]));
1110 }
1111
1112 #[test]
1113 fn test_nested() {
1114 let list = DataType::List(Arc::new(Field::new("foo", DataType::Utf8, true)));
1115 let list_view = DataType::ListView(Arc::new(Field::new("foo", DataType::Utf8, true)));
1116 let large_list_view =
1117 DataType::LargeListView(Arc::new(Field::new("foo", DataType::Utf8, true)));
1118
1119 assert!(!DataType::is_nested(&DataType::Boolean));
1120 assert!(!DataType::is_nested(&DataType::Int32));
1121 assert!(!DataType::is_nested(&DataType::Utf8));
1122 assert!(DataType::is_nested(&list));
1123 assert!(DataType::is_nested(&list_view));
1124 assert!(DataType::is_nested(&large_list_view));
1125
1126 assert!(!DataType::is_nested(&DataType::Dictionary(
1127 Box::new(DataType::Int32),
1128 Box::new(DataType::Boolean)
1129 )));
1130 assert!(!DataType::is_nested(&DataType::Dictionary(
1131 Box::new(DataType::Int32),
1132 Box::new(DataType::Int64)
1133 )));
1134 assert!(!DataType::is_nested(&DataType::Dictionary(
1135 Box::new(DataType::Int32),
1136 Box::new(DataType::LargeUtf8)
1137 )));
1138 assert!(DataType::is_nested(&DataType::Dictionary(
1139 Box::new(DataType::Int32),
1140 Box::new(list)
1141 )));
1142 }
1143
1144 #[test]
1145 fn test_integer() {
1146 // is_integer
1147 assert!(DataType::is_integer(&DataType::Int32));
1148 assert!(DataType::is_integer(&DataType::UInt64));
1149 assert!(!DataType::is_integer(&DataType::Float16));
1150
1151 // is_signed_integer
1152 assert!(DataType::is_signed_integer(&DataType::Int32));
1153 assert!(!DataType::is_signed_integer(&DataType::UInt64));
1154 assert!(!DataType::is_signed_integer(&DataType::Float16));
1155
1156 // is_unsigned_integer
1157 assert!(!DataType::is_unsigned_integer(&DataType::Int32));
1158 assert!(DataType::is_unsigned_integer(&DataType::UInt64));
1159 assert!(!DataType::is_unsigned_integer(&DataType::Float16));
1160
1161 // is_dictionary_key_type
1162 assert!(DataType::is_dictionary_key_type(&DataType::Int32));
1163 assert!(DataType::is_dictionary_key_type(&DataType::UInt64));
1164 assert!(!DataType::is_dictionary_key_type(&DataType::Float16));
1165 }
1166
1167 #[test]
1168 fn test_string() {
1169 assert!(DataType::is_string(&DataType::Utf8));
1170 assert!(DataType::is_string(&DataType::LargeUtf8));
1171 assert!(DataType::is_string(&DataType::Utf8View));
1172 assert!(!DataType::is_string(&DataType::Int32));
1173 }
1174
1175 #[test]
1176 fn test_floating() {
1177 assert!(DataType::is_floating(&DataType::Float16));
1178 assert!(!DataType::is_floating(&DataType::Int32));
1179 }
1180
1181 #[test]
1182 fn test_decimal() {
1183 assert!(DataType::is_decimal(&DataType::Decimal32(4, 2)));
1184 assert!(DataType::is_decimal(&DataType::Decimal64(4, 2)));
1185 assert!(DataType::is_decimal(&DataType::Decimal128(4, 2)));
1186 assert!(DataType::is_decimal(&DataType::Decimal256(4, 2)));
1187 assert!(!DataType::is_decimal(&DataType::Float16));
1188 }
1189
1190 #[test]
1191 fn test_datatype_is_null() {
1192 assert!(DataType::is_null(&DataType::Null));
1193 assert!(!DataType::is_null(&DataType::Int32));
1194 }
1195
1196 #[test]
1197 fn size_should_not_regress() {
1198 assert_eq!(std::mem::size_of::<DataType>(), 24);
1199 }
1200
1201 #[test]
1202 #[should_panic(expected = "duplicate type id: 1")]
1203 fn test_union_with_duplicated_type_id() {
1204 let type_ids = vec![1, 1];
1205 let _union = DataType::Union(
1206 UnionFields::try_new(
1207 type_ids,
1208 vec![
1209 Field::new("f1", DataType::Int32, false),
1210 Field::new("f2", DataType::Utf8, false),
1211 ],
1212 )
1213 .unwrap(),
1214 UnionMode::Dense,
1215 );
1216 }
1217
1218 #[test]
1219 fn test_try_from_str() {
1220 let data_type: DataType = "Int32".try_into().unwrap();
1221 assert_eq!(data_type, DataType::Int32);
1222 }
1223
1224 #[test]
1225 fn test_from_str() {
1226 let data_type: DataType = "UInt64".parse().unwrap();
1227 assert_eq!(data_type, DataType::UInt64);
1228 }
1229
1230 #[test]
1231 #[cfg_attr(miri, ignore)] // Can't handle the inlined strings of the assert_debug_snapshot macro
1232 fn test_debug_format_field() {
1233 // Make sure the `Debug` formatting of `DataType` is readable and not too long
1234 insta::assert_debug_snapshot!(DataType::new_list(DataType::Int8, false), @r"
1235 List(
1236 Field {
1237 data_type: Int8,
1238 },
1239 )
1240 ");
1241 }
1242}