datafusion_python/
expr.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 datafusion::logical_expr::expr::{AggregateFunctionParams, FieldMetadata};
19use datafusion::logical_expr::utils::exprlist_to_fields;
20use datafusion::logical_expr::{
21    lit_with_metadata, ExprFuncBuilder, ExprFunctionExt, LogicalPlan, WindowFunctionDefinition,
22};
23use pyo3::IntoPyObjectExt;
24use pyo3::{basic::CompareOp, prelude::*};
25use std::collections::HashMap;
26use std::convert::{From, Into};
27use std::sync::Arc;
28use window::PyWindowFrame;
29
30use datafusion::arrow::datatypes::{DataType, Field};
31use datafusion::arrow::pyarrow::PyArrowType;
32use datafusion::functions::core::expr_ext::FieldAccessor;
33use datafusion::logical_expr::{
34    col,
35    expr::{AggregateFunction, InList, InSubquery, ScalarFunction, WindowFunction},
36    lit, Between, BinaryExpr, Case, Cast, Expr, Like, Operator, TryCast,
37};
38
39use crate::common::data_type::{DataTypeMap, NullTreatment, PyScalarValue, RexType};
40use crate::errors::{py_runtime_err, py_type_err, py_unsupported_variant_err, PyDataFusionResult};
41use crate::expr::aggregate_expr::PyAggregateFunction;
42use crate::expr::binary_expr::PyBinaryExpr;
43use crate::expr::column::PyColumn;
44use crate::expr::literal::PyLiteral;
45use crate::functions::add_builder_fns_to_window;
46use crate::pyarrow_util::scalar_to_pyarrow;
47use crate::sql::logical::PyLogicalPlan;
48
49use self::alias::PyAlias;
50use self::bool_expr::{
51    PyIsFalse, PyIsNotFalse, PyIsNotNull, PyIsNotTrue, PyIsNotUnknown, PyIsNull, PyIsTrue,
52    PyIsUnknown, PyNegative, PyNot,
53};
54use self::like::{PyILike, PyLike, PySimilarTo};
55use self::scalar_variable::PyScalarVariable;
56
57pub mod aggregate;
58pub mod aggregate_expr;
59pub mod alias;
60pub mod analyze;
61pub mod between;
62pub mod binary_expr;
63pub mod bool_expr;
64pub mod case;
65pub mod cast;
66pub mod column;
67pub mod conditional_expr;
68pub mod copy_to;
69pub mod create_catalog;
70pub mod create_catalog_schema;
71pub mod create_external_table;
72pub mod create_function;
73pub mod create_index;
74pub mod create_memory_table;
75pub mod create_view;
76pub mod describe_table;
77pub mod distinct;
78pub mod dml;
79pub mod drop_catalog_schema;
80pub mod drop_function;
81pub mod drop_table;
82pub mod drop_view;
83pub mod empty_relation;
84pub mod exists;
85pub mod explain;
86pub mod extension;
87pub mod filter;
88pub mod grouping_set;
89pub mod in_list;
90pub mod in_subquery;
91pub mod join;
92pub mod like;
93pub mod limit;
94pub mod literal;
95pub mod logical_node;
96pub mod placeholder;
97pub mod projection;
98pub mod recursive_query;
99pub mod repartition;
100pub mod scalar_subquery;
101pub mod scalar_variable;
102pub mod signature;
103pub mod sort;
104pub mod sort_expr;
105pub mod statement;
106pub mod subquery;
107pub mod subquery_alias;
108pub mod table_scan;
109pub mod union;
110pub mod unnest;
111pub mod unnest_expr;
112pub mod values;
113pub mod window;
114
115use sort_expr::{to_sort_expressions, PySortExpr};
116
117/// A PyExpr that can be used on a DataFrame
118#[pyclass(name = "RawExpr", module = "datafusion.expr", subclass)]
119#[derive(Debug, Clone)]
120pub struct PyExpr {
121    pub expr: Expr,
122}
123
124impl From<PyExpr> for Expr {
125    fn from(expr: PyExpr) -> Expr {
126        expr.expr
127    }
128}
129
130impl From<Expr> for PyExpr {
131    fn from(expr: Expr) -> PyExpr {
132        PyExpr { expr }
133    }
134}
135
136/// Convert a list of DataFusion Expr to PyExpr
137pub fn py_expr_list(expr: &[Expr]) -> PyResult<Vec<PyExpr>> {
138    Ok(expr.iter().map(|e| PyExpr::from(e.clone())).collect())
139}
140
141#[pymethods]
142impl PyExpr {
143    /// Return the specific expression
144    fn to_variant<'py>(&self, py: Python<'py>) -> PyResult<Bound<'py, PyAny>> {
145        Python::with_gil(|_| {
146            match &self.expr {
147            Expr::Alias(alias) => Ok(PyAlias::from(alias.clone()).into_bound_py_any(py)?),
148            Expr::Column(col) => Ok(PyColumn::from(col.clone()).into_bound_py_any(py)?),
149            Expr::ScalarVariable(data_type, variables) => {
150                Ok(PyScalarVariable::new(data_type, variables).into_bound_py_any(py)?)
151            }
152            Expr::Like(value) => Ok(PyLike::from(value.clone()).into_bound_py_any(py)?),
153            Expr::Literal(value, metadata) => Ok(PyLiteral::new_with_metadata(value.clone(), metadata.clone()).into_bound_py_any(py)?),
154            Expr::BinaryExpr(expr) => Ok(PyBinaryExpr::from(expr.clone()).into_bound_py_any(py)?),
155            Expr::Not(expr) => Ok(PyNot::new(*expr.clone()).into_bound_py_any(py)?),
156            Expr::IsNotNull(expr) => Ok(PyIsNotNull::new(*expr.clone()).into_bound_py_any(py)?),
157            Expr::IsNull(expr) => Ok(PyIsNull::new(*expr.clone()).into_bound_py_any(py)?),
158            Expr::IsTrue(expr) => Ok(PyIsTrue::new(*expr.clone()).into_bound_py_any(py)?),
159            Expr::IsFalse(expr) => Ok(PyIsFalse::new(*expr.clone()).into_bound_py_any(py)?),
160            Expr::IsUnknown(expr) => Ok(PyIsUnknown::new(*expr.clone()).into_bound_py_any(py)?),
161            Expr::IsNotTrue(expr) => Ok(PyIsNotTrue::new(*expr.clone()).into_bound_py_any(py)?),
162            Expr::IsNotFalse(expr) => Ok(PyIsNotFalse::new(*expr.clone()).into_bound_py_any(py)?),
163            Expr::IsNotUnknown(expr) => Ok(PyIsNotUnknown::new(*expr.clone()).into_bound_py_any(py)?),
164            Expr::Negative(expr) => Ok(PyNegative::new(*expr.clone()).into_bound_py_any(py)?),
165            Expr::AggregateFunction(expr) => {
166                Ok(PyAggregateFunction::from(expr.clone()).into_bound_py_any(py)?)
167            }
168            Expr::SimilarTo(value) => Ok(PySimilarTo::from(value.clone()).into_bound_py_any(py)?),
169            Expr::Between(value) => Ok(between::PyBetween::from(value.clone()).into_bound_py_any(py)?),
170            Expr::Case(value) => Ok(case::PyCase::from(value.clone()).into_bound_py_any(py)?),
171            Expr::Cast(value) => Ok(cast::PyCast::from(value.clone()).into_bound_py_any(py)?),
172            Expr::TryCast(value) => Ok(cast::PyTryCast::from(value.clone()).into_bound_py_any(py)?),
173            Expr::ScalarFunction(value) => Err(py_unsupported_variant_err(format!(
174                "Converting Expr::ScalarFunction to a Python object is not implemented: {value:?}"
175            ))),
176            Expr::WindowFunction(value) => Err(py_unsupported_variant_err(format!(
177                "Converting Expr::WindowFunction to a Python object is not implemented: {value:?}"
178            ))),
179            Expr::InList(value) => Ok(in_list::PyInList::from(value.clone()).into_bound_py_any(py)?),
180            Expr::Exists(value) => Ok(exists::PyExists::from(value.clone()).into_bound_py_any(py)?),
181            Expr::InSubquery(value) => {
182                Ok(in_subquery::PyInSubquery::from(value.clone()).into_bound_py_any(py)?)
183            }
184            Expr::ScalarSubquery(value) => {
185                Ok(scalar_subquery::PyScalarSubquery::from(value.clone()).into_bound_py_any(py)?)
186            }
187            #[allow(deprecated)]
188            Expr::Wildcard { qualifier, options } => Err(py_unsupported_variant_err(format!(
189                "Converting Expr::Wildcard to a Python object is not implemented : {qualifier:?} {options:?}"
190            ))),
191            Expr::GroupingSet(value) => {
192                Ok(grouping_set::PyGroupingSet::from(value.clone()).into_bound_py_any(py)?)
193            }
194            Expr::Placeholder(value) => {
195                Ok(placeholder::PyPlaceholder::from(value.clone()).into_bound_py_any(py)?)
196            }
197            Expr::OuterReferenceColumn(data_type, column) => Err(py_unsupported_variant_err(format!(
198                "Converting Expr::OuterReferenceColumn to a Python object is not implemented: {data_type:?} - {column:?}"
199            ))),
200            Expr::Unnest(value) => Ok(unnest_expr::PyUnnestExpr::from(value.clone()).into_bound_py_any(py)?),
201        }
202        })
203    }
204
205    /// Returns the name of this expression as it should appear in a schema. This name
206    /// will not include any CAST expressions.
207    fn schema_name(&self) -> PyResult<String> {
208        Ok(format!("{}", self.expr.schema_name()))
209    }
210
211    /// Returns a full and complete string representation of this expression.
212    fn canonical_name(&self) -> PyResult<String> {
213        Ok(format!("{}", self.expr))
214    }
215
216    /// Returns the name of the Expr variant.
217    /// Ex: 'IsNotNull', 'Literal', 'BinaryExpr', etc
218    fn variant_name(&self) -> PyResult<&str> {
219        Ok(self.expr.variant_name())
220    }
221
222    fn __richcmp__(&self, other: PyExpr, op: CompareOp) -> PyExpr {
223        let expr = match op {
224            CompareOp::Lt => self.expr.clone().lt(other.expr),
225            CompareOp::Le => self.expr.clone().lt_eq(other.expr),
226            CompareOp::Eq => self.expr.clone().eq(other.expr),
227            CompareOp::Ne => self.expr.clone().not_eq(other.expr),
228            CompareOp::Gt => self.expr.clone().gt(other.expr),
229            CompareOp::Ge => self.expr.clone().gt_eq(other.expr),
230        };
231        expr.into()
232    }
233
234    fn __repr__(&self) -> PyResult<String> {
235        Ok(format!("Expr({})", self.expr))
236    }
237
238    fn __add__(&self, rhs: PyExpr) -> PyResult<PyExpr> {
239        Ok((self.expr.clone() + rhs.expr).into())
240    }
241
242    fn __sub__(&self, rhs: PyExpr) -> PyResult<PyExpr> {
243        Ok((self.expr.clone() - rhs.expr).into())
244    }
245
246    fn __truediv__(&self, rhs: PyExpr) -> PyResult<PyExpr> {
247        Ok((self.expr.clone() / rhs.expr).into())
248    }
249
250    fn __mul__(&self, rhs: PyExpr) -> PyResult<PyExpr> {
251        Ok((self.expr.clone() * rhs.expr).into())
252    }
253
254    fn __mod__(&self, rhs: PyExpr) -> PyResult<PyExpr> {
255        let expr = self.expr.clone() % rhs.expr;
256        Ok(expr.into())
257    }
258
259    fn __and__(&self, rhs: PyExpr) -> PyResult<PyExpr> {
260        Ok(self.expr.clone().and(rhs.expr).into())
261    }
262
263    fn __or__(&self, rhs: PyExpr) -> PyResult<PyExpr> {
264        Ok(self.expr.clone().or(rhs.expr).into())
265    }
266
267    fn __invert__(&self) -> PyResult<PyExpr> {
268        let expr = !self.expr.clone();
269        Ok(expr.into())
270    }
271
272    fn __getitem__(&self, key: &str) -> PyResult<PyExpr> {
273        Ok(self.expr.clone().field(key).into())
274    }
275
276    #[staticmethod]
277    pub fn literal(value: PyScalarValue) -> PyExpr {
278        lit(value.0).into()
279    }
280
281    #[staticmethod]
282    pub fn literal_with_metadata(
283        value: PyScalarValue,
284        metadata: HashMap<String, String>,
285    ) -> PyExpr {
286        let metadata = FieldMetadata::new(metadata.into_iter().collect());
287        lit_with_metadata(value.0, Some(metadata)).into()
288    }
289
290    #[staticmethod]
291    pub fn column(value: &str) -> PyExpr {
292        col(value).into()
293    }
294
295    /// assign a name to the PyExpr
296    #[pyo3(signature = (name, metadata=None))]
297    pub fn alias(&self, name: &str, metadata: Option<HashMap<String, String>>) -> PyExpr {
298        let metadata = metadata.map(|m| FieldMetadata::new(m.into_iter().collect()));
299        self.expr.clone().alias_with_metadata(name, metadata).into()
300    }
301
302    /// Create a sort PyExpr from an existing PyExpr.
303    #[pyo3(signature = (ascending=true, nulls_first=true))]
304    pub fn sort(&self, ascending: bool, nulls_first: bool) -> PySortExpr {
305        self.expr.clone().sort(ascending, nulls_first).into()
306    }
307
308    pub fn is_null(&self) -> PyExpr {
309        self.expr.clone().is_null().into()
310    }
311
312    pub fn is_not_null(&self) -> PyExpr {
313        self.expr.clone().is_not_null().into()
314    }
315
316    pub fn cast(&self, to: PyArrowType<DataType>) -> PyExpr {
317        // self.expr.cast_to() requires DFSchema to validate that the cast
318        // is supported, omit that for now
319        let expr = Expr::Cast(Cast::new(Box::new(self.expr.clone()), to.0));
320        expr.into()
321    }
322
323    #[pyo3(signature = (low, high, negated=false))]
324    pub fn between(&self, low: PyExpr, high: PyExpr, negated: bool) -> PyExpr {
325        let expr = Expr::Between(Between::new(
326            Box::new(self.expr.clone()),
327            negated,
328            Box::new(low.into()),
329            Box::new(high.into()),
330        ));
331        expr.into()
332    }
333
334    /// A Rex (Row Expression) specifies a single row of data. That specification
335    /// could include user defined functions or types. RexType identifies the row
336    /// as one of the possible valid `RexTypes`.
337    pub fn rex_type(&self) -> PyResult<RexType> {
338        Ok(match self.expr {
339            Expr::Alias(..) => RexType::Alias,
340            Expr::Column(..) => RexType::Reference,
341            Expr::ScalarVariable(..) | Expr::Literal(..) => RexType::Literal,
342            Expr::BinaryExpr { .. }
343            | Expr::Not(..)
344            | Expr::IsNotNull(..)
345            | Expr::Negative(..)
346            | Expr::IsNull(..)
347            | Expr::Like { .. }
348            | Expr::SimilarTo { .. }
349            | Expr::Between { .. }
350            | Expr::Case { .. }
351            | Expr::Cast { .. }
352            | Expr::TryCast { .. }
353            | Expr::ScalarFunction { .. }
354            | Expr::AggregateFunction { .. }
355            | Expr::WindowFunction { .. }
356            | Expr::InList { .. }
357            | Expr::Exists { .. }
358            | Expr::InSubquery { .. }
359            | Expr::GroupingSet(..)
360            | Expr::IsTrue(..)
361            | Expr::IsFalse(..)
362            | Expr::IsUnknown(_)
363            | Expr::IsNotTrue(..)
364            | Expr::IsNotFalse(..)
365            | Expr::Placeholder { .. }
366            | Expr::OuterReferenceColumn(_, _)
367            | Expr::Unnest(_)
368            | Expr::IsNotUnknown(_) => RexType::Call,
369            Expr::ScalarSubquery(..) => RexType::ScalarSubquery,
370            #[allow(deprecated)]
371            Expr::Wildcard { .. } => {
372                return Err(py_unsupported_variant_err("Expr::Wildcard is unsupported"))
373            }
374        })
375    }
376
377    /// Given the current `Expr` return the DataTypeMap which represents the
378    /// PythonType, Arrow DataType, and SqlType Enum which represents
379    pub fn types(&self) -> PyResult<DataTypeMap> {
380        Self::_types(&self.expr)
381    }
382
383    /// Extracts the Expr value into a PyObject that can be shared with Python
384    pub fn python_value(&self, py: Python) -> PyResult<PyObject> {
385        match &self.expr {
386            Expr::Literal(scalar_value, _) => scalar_to_pyarrow(scalar_value, py),
387            _ => Err(py_type_err(format!(
388                "Non Expr::Literal encountered in types: {:?}",
389                &self.expr
390            ))),
391        }
392    }
393
394    /// Row expressions, Rex(s), operate on the concept of operands. Different variants of Expressions, Expr(s),
395    /// store those operands in different datastructures. This function examines the Expr variant and returns
396    /// the operands to the calling logic as a Vec of PyExpr instances.
397    pub fn rex_call_operands(&self) -> PyResult<Vec<PyExpr>> {
398        match &self.expr {
399            // Expr variants that are themselves the operand to return
400            Expr::Column(..) | Expr::ScalarVariable(..) | Expr::Literal(..) => {
401                Ok(vec![PyExpr::from(self.expr.clone())])
402            }
403
404            Expr::Alias(alias) => Ok(vec![PyExpr::from(*alias.expr.clone())]),
405
406            // Expr(s) that house the Expr instance to return in their bounded params
407            Expr::Not(expr)
408            | Expr::IsNull(expr)
409            | Expr::IsNotNull(expr)
410            | Expr::IsTrue(expr)
411            | Expr::IsFalse(expr)
412            | Expr::IsUnknown(expr)
413            | Expr::IsNotTrue(expr)
414            | Expr::IsNotFalse(expr)
415            | Expr::IsNotUnknown(expr)
416            | Expr::Negative(expr)
417            | Expr::Cast(Cast { expr, .. })
418            | Expr::TryCast(TryCast { expr, .. })
419            | Expr::InSubquery(InSubquery { expr, .. }) => Ok(vec![PyExpr::from(*expr.clone())]),
420
421            // Expr variants containing a collection of Expr(s) for operands
422            Expr::AggregateFunction(AggregateFunction {
423                params: AggregateFunctionParams { args, .. },
424                ..
425            })
426            | Expr::ScalarFunction(ScalarFunction { args, .. }) => {
427                Ok(args.iter().map(|arg| PyExpr::from(arg.clone())).collect())
428            }
429            Expr::WindowFunction(boxed_window_fn) => {
430                let args = &boxed_window_fn.params.args;
431                Ok(args.iter().map(|arg| PyExpr::from(arg.clone())).collect())
432            }
433
434            // Expr(s) that require more specific processing
435            Expr::Case(Case {
436                expr,
437                when_then_expr,
438                else_expr,
439            }) => {
440                let mut operands: Vec<PyExpr> = Vec::new();
441
442                if let Some(e) = expr {
443                    for (when, then) in when_then_expr {
444                        operands.push(PyExpr::from(Expr::BinaryExpr(BinaryExpr::new(
445                            Box::new(*e.clone()),
446                            Operator::Eq,
447                            Box::new(*when.clone()),
448                        ))));
449                        operands.push(PyExpr::from(*then.clone()));
450                    }
451                } else {
452                    for (when, then) in when_then_expr {
453                        operands.push(PyExpr::from(*when.clone()));
454                        operands.push(PyExpr::from(*then.clone()));
455                    }
456                };
457
458                if let Some(e) = else_expr {
459                    operands.push(PyExpr::from(*e.clone()));
460                };
461
462                Ok(operands)
463            }
464            Expr::InList(InList { expr, list, .. }) => {
465                let mut operands: Vec<PyExpr> = vec![PyExpr::from(*expr.clone())];
466                for list_elem in list {
467                    operands.push(PyExpr::from(list_elem.clone()));
468                }
469
470                Ok(operands)
471            }
472            Expr::BinaryExpr(BinaryExpr { left, right, .. }) => Ok(vec![
473                PyExpr::from(*left.clone()),
474                PyExpr::from(*right.clone()),
475            ]),
476            Expr::Like(Like { expr, pattern, .. }) => Ok(vec![
477                PyExpr::from(*expr.clone()),
478                PyExpr::from(*pattern.clone()),
479            ]),
480            Expr::SimilarTo(Like { expr, pattern, .. }) => Ok(vec![
481                PyExpr::from(*expr.clone()),
482                PyExpr::from(*pattern.clone()),
483            ]),
484            Expr::Between(Between {
485                expr,
486                negated: _,
487                low,
488                high,
489            }) => Ok(vec![
490                PyExpr::from(*expr.clone()),
491                PyExpr::from(*low.clone()),
492                PyExpr::from(*high.clone()),
493            ]),
494
495            // Currently un-support/implemented Expr types for Rex Call operations
496            Expr::GroupingSet(..)
497            | Expr::Unnest(_)
498            | Expr::OuterReferenceColumn(_, _)
499            | Expr::ScalarSubquery(..)
500            | Expr::Placeholder { .. }
501            | Expr::Exists { .. } => Err(py_runtime_err(format!(
502                "Unimplemented Expr type: {}",
503                self.expr
504            ))),
505
506            #[allow(deprecated)]
507            Expr::Wildcard { .. } => {
508                Err(py_unsupported_variant_err("Expr::Wildcard is unsupported"))
509            }
510        }
511    }
512
513    /// Extracts the operator associated with a RexType::Call
514    pub fn rex_call_operator(&self) -> PyResult<String> {
515        Ok(match &self.expr {
516            Expr::BinaryExpr(BinaryExpr {
517                left: _,
518                op,
519                right: _,
520            }) => format!("{op}"),
521            Expr::ScalarFunction(ScalarFunction { func, args: _ }) => func.name().to_string(),
522            Expr::Cast { .. } => "cast".to_string(),
523            Expr::Between { .. } => "between".to_string(),
524            Expr::Case { .. } => "case".to_string(),
525            Expr::IsNull(..) => "is null".to_string(),
526            Expr::IsNotNull(..) => "is not null".to_string(),
527            Expr::IsTrue(_) => "is true".to_string(),
528            Expr::IsFalse(_) => "is false".to_string(),
529            Expr::IsUnknown(_) => "is unknown".to_string(),
530            Expr::IsNotTrue(_) => "is not true".to_string(),
531            Expr::IsNotFalse(_) => "is not false".to_string(),
532            Expr::IsNotUnknown(_) => "is not unknown".to_string(),
533            Expr::InList { .. } => "in list".to_string(),
534            Expr::Negative(..) => "negative".to_string(),
535            Expr::Not(..) => "not".to_string(),
536            Expr::Like(Like {
537                negated,
538                case_insensitive,
539                ..
540            }) => {
541                let name = if *case_insensitive { "ilike" } else { "like" };
542                if *negated {
543                    format!("not {name}")
544                } else {
545                    name.to_string()
546                }
547            }
548            Expr::SimilarTo(Like { negated, .. }) => {
549                if *negated {
550                    "not similar to".to_string()
551                } else {
552                    "similar to".to_string()
553                }
554            }
555            _ => {
556                return Err(py_type_err(format!(
557                    "Catch all triggered in get_operator_name: {:?}",
558                    &self.expr
559                )))
560            }
561        })
562    }
563
564    pub fn column_name(&self, plan: PyLogicalPlan) -> PyResult<String> {
565        self._column_name(&plan.plan()).map_err(py_runtime_err)
566    }
567
568    // Expression Function Builder functions
569
570    pub fn order_by(&self, order_by: Vec<PySortExpr>) -> PyExprFuncBuilder {
571        self.expr
572            .clone()
573            .order_by(to_sort_expressions(order_by))
574            .into()
575    }
576
577    pub fn filter(&self, filter: PyExpr) -> PyExprFuncBuilder {
578        self.expr.clone().filter(filter.expr.clone()).into()
579    }
580
581    pub fn distinct(&self) -> PyExprFuncBuilder {
582        self.expr.clone().distinct().into()
583    }
584
585    pub fn null_treatment(&self, null_treatment: NullTreatment) -> PyExprFuncBuilder {
586        self.expr
587            .clone()
588            .null_treatment(Some(null_treatment.into()))
589            .into()
590    }
591
592    pub fn partition_by(&self, partition_by: Vec<PyExpr>) -> PyExprFuncBuilder {
593        let partition_by = partition_by.iter().map(|e| e.expr.clone()).collect();
594        self.expr.clone().partition_by(partition_by).into()
595    }
596
597    pub fn window_frame(&self, window_frame: PyWindowFrame) -> PyExprFuncBuilder {
598        self.expr.clone().window_frame(window_frame.into()).into()
599    }
600
601    #[pyo3(signature = (partition_by=None, window_frame=None, order_by=None, null_treatment=None))]
602    pub fn over(
603        &self,
604        partition_by: Option<Vec<PyExpr>>,
605        window_frame: Option<PyWindowFrame>,
606        order_by: Option<Vec<PySortExpr>>,
607        null_treatment: Option<NullTreatment>,
608    ) -> PyDataFusionResult<PyExpr> {
609        match &self.expr {
610            Expr::AggregateFunction(agg_fn) => {
611                let window_fn = Expr::WindowFunction(Box::new(WindowFunction::new(
612                    WindowFunctionDefinition::AggregateUDF(agg_fn.func.clone()),
613                    agg_fn.params.args.clone(),
614                )));
615
616                add_builder_fns_to_window(
617                    window_fn,
618                    partition_by,
619                    window_frame,
620                    order_by,
621                    null_treatment,
622                )
623            }
624            Expr::WindowFunction(_) => add_builder_fns_to_window(
625                self.expr.clone(),
626                partition_by,
627                window_frame,
628                order_by,
629                null_treatment,
630            ),
631            _ => Err(datafusion::error::DataFusionError::Plan(format!(
632                "Using {} with `over` is not allowed. Must use an aggregate or window function.",
633                self.expr.variant_name()
634            ))
635            .into()),
636        }
637    }
638}
639
640#[pyclass(name = "ExprFuncBuilder", module = "datafusion.expr", subclass)]
641#[derive(Debug, Clone)]
642pub struct PyExprFuncBuilder {
643    pub builder: ExprFuncBuilder,
644}
645
646impl From<ExprFuncBuilder> for PyExprFuncBuilder {
647    fn from(builder: ExprFuncBuilder) -> Self {
648        Self { builder }
649    }
650}
651
652#[pymethods]
653impl PyExprFuncBuilder {
654    pub fn order_by(&self, order_by: Vec<PySortExpr>) -> PyExprFuncBuilder {
655        self.builder
656            .clone()
657            .order_by(to_sort_expressions(order_by))
658            .into()
659    }
660
661    pub fn filter(&self, filter: PyExpr) -> PyExprFuncBuilder {
662        self.builder.clone().filter(filter.expr.clone()).into()
663    }
664
665    pub fn distinct(&self) -> PyExprFuncBuilder {
666        self.builder.clone().distinct().into()
667    }
668
669    pub fn null_treatment(&self, null_treatment: NullTreatment) -> PyExprFuncBuilder {
670        self.builder
671            .clone()
672            .null_treatment(Some(null_treatment.into()))
673            .into()
674    }
675
676    pub fn partition_by(&self, partition_by: Vec<PyExpr>) -> PyExprFuncBuilder {
677        let partition_by = partition_by.iter().map(|e| e.expr.clone()).collect();
678        self.builder.clone().partition_by(partition_by).into()
679    }
680
681    pub fn window_frame(&self, window_frame: PyWindowFrame) -> PyExprFuncBuilder {
682        self.builder
683            .clone()
684            .window_frame(window_frame.into())
685            .into()
686    }
687
688    pub fn build(&self) -> PyDataFusionResult<PyExpr> {
689        Ok(self.builder.clone().build().map(|expr| expr.into())?)
690    }
691}
692
693impl PyExpr {
694    pub fn _column_name(&self, plan: &LogicalPlan) -> PyDataFusionResult<String> {
695        let field = Self::expr_to_field(&self.expr, plan)?;
696        Ok(field.name().to_owned())
697    }
698
699    /// Create a [Field] representing an [Expr], given an input [LogicalPlan] to resolve against
700    pub fn expr_to_field(expr: &Expr, input_plan: &LogicalPlan) -> PyDataFusionResult<Arc<Field>> {
701        let fields = exprlist_to_fields(std::slice::from_ref(expr), input_plan)?;
702        Ok(fields[0].1.clone())
703    }
704    fn _types(expr: &Expr) -> PyResult<DataTypeMap> {
705        match expr {
706            Expr::BinaryExpr(BinaryExpr {
707                left: _,
708                op,
709                right: _,
710            }) => match op {
711                Operator::Eq
712                | Operator::NotEq
713                | Operator::Lt
714                | Operator::LtEq
715                | Operator::Gt
716                | Operator::GtEq
717                | Operator::And
718                | Operator::Or
719                | Operator::IsDistinctFrom
720                | Operator::IsNotDistinctFrom
721                | Operator::RegexMatch
722                | Operator::RegexIMatch
723                | Operator::RegexNotMatch
724                | Operator::RegexNotIMatch
725                | Operator::LikeMatch
726                | Operator::ILikeMatch
727                | Operator::NotLikeMatch
728                | Operator::NotILikeMatch => DataTypeMap::map_from_arrow_type(&DataType::Boolean),
729                Operator::Plus | Operator::Minus | Operator::Multiply | Operator::Modulo => {
730                    DataTypeMap::map_from_arrow_type(&DataType::Int64)
731                }
732                Operator::Divide => DataTypeMap::map_from_arrow_type(&DataType::Float64),
733                Operator::StringConcat => DataTypeMap::map_from_arrow_type(&DataType::Utf8),
734                Operator::BitwiseShiftLeft
735                | Operator::BitwiseShiftRight
736                | Operator::BitwiseXor
737                | Operator::BitwiseAnd
738                | Operator::BitwiseOr => DataTypeMap::map_from_arrow_type(&DataType::Binary),
739                Operator::AtArrow
740                | Operator::ArrowAt
741                | Operator::Arrow
742                | Operator::LongArrow
743                | Operator::HashArrow
744                | Operator::HashLongArrow
745                | Operator::AtAt
746                | Operator::IntegerDivide
747                | Operator::HashMinus
748                | Operator::AtQuestion
749                | Operator::Question
750                | Operator::QuestionAnd
751                | Operator::QuestionPipe => Err(py_type_err(format!("Unsupported expr: ${op}"))),
752            },
753            Expr::Cast(Cast { expr: _, data_type }) => DataTypeMap::map_from_arrow_type(data_type),
754            Expr::Literal(scalar_value, _) => DataTypeMap::map_from_scalar_value(scalar_value),
755            _ => Err(py_type_err(format!(
756                "Non Expr::Literal encountered in types: {expr:?}"
757            ))),
758        }
759    }
760}
761
762/// Initializes the `expr` module to match the pattern of `datafusion-expr` https://docs.rs/datafusion-expr/latest/datafusion_expr/
763pub(crate) fn init_module(m: &Bound<'_, PyModule>) -> PyResult<()> {
764    m.add_class::<PyExpr>()?;
765    m.add_class::<PyColumn>()?;
766    m.add_class::<PyLiteral>()?;
767    m.add_class::<PyBinaryExpr>()?;
768    m.add_class::<PyLiteral>()?;
769    m.add_class::<PyAggregateFunction>()?;
770    m.add_class::<PyNot>()?;
771    m.add_class::<PyIsNotNull>()?;
772    m.add_class::<PyIsNull>()?;
773    m.add_class::<PyIsTrue>()?;
774    m.add_class::<PyIsFalse>()?;
775    m.add_class::<PyIsUnknown>()?;
776    m.add_class::<PyIsNotTrue>()?;
777    m.add_class::<PyIsNotFalse>()?;
778    m.add_class::<PyIsNotUnknown>()?;
779    m.add_class::<PyNegative>()?;
780    m.add_class::<PyLike>()?;
781    m.add_class::<PyILike>()?;
782    m.add_class::<PySimilarTo>()?;
783    m.add_class::<PyScalarVariable>()?;
784    m.add_class::<alias::PyAlias>()?;
785    m.add_class::<in_list::PyInList>()?;
786    m.add_class::<exists::PyExists>()?;
787    m.add_class::<subquery::PySubquery>()?;
788    m.add_class::<in_subquery::PyInSubquery>()?;
789    m.add_class::<scalar_subquery::PyScalarSubquery>()?;
790    m.add_class::<placeholder::PyPlaceholder>()?;
791    m.add_class::<grouping_set::PyGroupingSet>()?;
792    m.add_class::<case::PyCase>()?;
793    m.add_class::<conditional_expr::PyCaseBuilder>()?;
794    m.add_class::<cast::PyCast>()?;
795    m.add_class::<cast::PyTryCast>()?;
796    m.add_class::<between::PyBetween>()?;
797    m.add_class::<explain::PyExplain>()?;
798    m.add_class::<limit::PyLimit>()?;
799    m.add_class::<aggregate::PyAggregate>()?;
800    m.add_class::<sort::PySort>()?;
801    m.add_class::<analyze::PyAnalyze>()?;
802    m.add_class::<empty_relation::PyEmptyRelation>()?;
803    m.add_class::<join::PyJoin>()?;
804    m.add_class::<join::PyJoinType>()?;
805    m.add_class::<join::PyJoinConstraint>()?;
806    m.add_class::<union::PyUnion>()?;
807    m.add_class::<unnest::PyUnnest>()?;
808    m.add_class::<unnest_expr::PyUnnestExpr>()?;
809    m.add_class::<extension::PyExtension>()?;
810    m.add_class::<filter::PyFilter>()?;
811    m.add_class::<projection::PyProjection>()?;
812    m.add_class::<table_scan::PyTableScan>()?;
813    m.add_class::<create_memory_table::PyCreateMemoryTable>()?;
814    m.add_class::<create_view::PyCreateView>()?;
815    m.add_class::<distinct::PyDistinct>()?;
816    m.add_class::<sort_expr::PySortExpr>()?;
817    m.add_class::<subquery_alias::PySubqueryAlias>()?;
818    m.add_class::<drop_table::PyDropTable>()?;
819    m.add_class::<repartition::PyPartitioning>()?;
820    m.add_class::<repartition::PyRepartition>()?;
821    m.add_class::<window::PyWindowExpr>()?;
822    m.add_class::<window::PyWindowFrame>()?;
823    m.add_class::<window::PyWindowFrameBound>()?;
824    m.add_class::<copy_to::PyCopyTo>()?;
825    m.add_class::<copy_to::PyFileType>()?;
826    m.add_class::<create_catalog::PyCreateCatalog>()?;
827    m.add_class::<create_catalog_schema::PyCreateCatalogSchema>()?;
828    m.add_class::<create_external_table::PyCreateExternalTable>()?;
829    m.add_class::<create_function::PyCreateFunction>()?;
830    m.add_class::<create_function::PyOperateFunctionArg>()?;
831    m.add_class::<create_function::PyCreateFunctionBody>()?;
832    m.add_class::<create_index::PyCreateIndex>()?;
833    m.add_class::<describe_table::PyDescribeTable>()?;
834    m.add_class::<dml::PyDmlStatement>()?;
835    m.add_class::<drop_catalog_schema::PyDropCatalogSchema>()?;
836    m.add_class::<drop_function::PyDropFunction>()?;
837    m.add_class::<drop_view::PyDropView>()?;
838    m.add_class::<recursive_query::PyRecursiveQuery>()?;
839
840    m.add_class::<statement::PyTransactionStart>()?;
841    m.add_class::<statement::PyTransactionEnd>()?;
842    m.add_class::<statement::PySetVariable>()?;
843    m.add_class::<statement::PyPrepare>()?;
844    m.add_class::<statement::PyExecute>()?;
845    m.add_class::<statement::PyDeallocate>()?;
846    m.add_class::<values::PyValues>()?;
847    m.add_class::<statement::PyTransactionAccessMode>()?;
848    m.add_class::<statement::PyTransactionConclusion>()?;
849    m.add_class::<statement::PyTransactionIsolationLevel>()?;
850
851    Ok(())
852}
datafusion_python/expr.rs

datafusion_python/
expr.rs
