Repository navigation
Expand file tree
/
Copy pathudwf.rs
More file actions
337 lines (298 loc) · 11.1 KB
/
Copy pathudwf.rs
File metadata and controls
337 lines (298 loc) · 11.1 KB
1
2
3
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
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
// Licensed to the Apache Software Foundation (ASF) under one
// or more contributor license agreements. See the NOTICE file
// distributed with this work for additional information
// regarding copyright ownership. The ASF licenses this file
// to you under the Apache License, Version 2.0 (the
// "License"); you may not use this file except in compliance
// with the License. You may obtain a copy of the License at
//
// http://www.apache.org/licenses/LICENSE-2.0
//
// Unless required by applicable law or agreed to in writing,
// software distributed under the License is distributed on an
// "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
// KIND, either express or implied. See the License for the
// specific language governing permissions and limitations
// under the License.
use std::any::Any;
use std::ops::Range;
use std::sync::Arc;
use arrow::array::{make_array, Array, ArrayData, ArrayRef};
use datafusion::logical_expr::function::{PartitionEvaluatorArgs, WindowUDFFieldArgs};
use datafusion::logical_expr::window_state::WindowAggState;
use datafusion::scalar::ScalarValue;
use pyo3::exceptions::PyValueError;
use pyo3::prelude::*;
use crate::common::data_type::PyScalarValue;
use crate::errors::{py_datafusion_err, to_datafusion_err, PyDataFusionResult};
use crate::expr::PyExpr;
use crate::utils::{parse_volatility, validate_pycapsule};
use datafusion::arrow::datatypes::DataType;
use datafusion::arrow::pyarrow::{FromPyArrow, PyArrowType, ToPyArrow};
use datafusion::error::{DataFusionError, Result};
use datafusion::logical_expr::{
PartitionEvaluator, PartitionEvaluatorFactory, Signature, Volatility, WindowUDF, WindowUDFImpl,
};
use datafusion_ffi::udwf::{FFI_WindowUDF, ForeignWindowUDF};
use pyo3::types::{PyCapsule, PyList, PyTuple};
#[derive(Debug)]
struct RustPartitionEvaluator {
evaluator: PyObject,
}
impl RustPartitionEvaluator {
fn new(evaluator: PyObject) -> Self {
Self { evaluator }
}
}
impl PartitionEvaluator for RustPartitionEvaluator {
fn memoize(&mut self, _state: &mut WindowAggState) -> Result<()> {
Python::with_gil(|py| self.evaluator.bind(py).call_method0("memoize").map(|_| ()))
.map_err(|e| DataFusionError::Execution(format!("{e}")))
}
fn get_range(&self, idx: usize, n_rows: usize) -> Result<Range<usize>> {
Python::with_gil(|py| {
let py_args = vec![idx.into_pyobject(py)?, n_rows.into_pyobject(py)?];
let py_args = PyTuple::new(py, py_args)?;
self.evaluator
.bind(py)
.call_method1("get_range", py_args)
.and_then(|v| {
let tuple: Bound<'_, PyTuple> = v.extract()?;
if tuple.len() != 2 {
return Err(PyValueError::new_err(format!(
"Expected get_range to return tuple of length 2. Received length {}",
tuple.len()
)));
}
let start: usize = tuple.get_item(0).unwrap().extract()?;
let end: usize = tuple.get_item(1).unwrap().extract()?;
Ok(Range { start, end })
})
})
.map_err(|e| DataFusionError::Execution(format!("{e}")))
}
fn is_causal(&self) -> bool {
Python::with_gil(|py| {
self.evaluator
.bind(py)
.call_method0("is_causal")
.and_then(|v| v.extract())
.unwrap_or(false)
})
}
fn evaluate_all(&mut self, values: &[ArrayRef], num_rows: usize) -> Result<ArrayRef> {
println!("evaluate all called with number of values {}", values.len());
Python::with_gil(|py| {
let py_values = PyList::new(
py,
values
.iter()
.map(|arg| arg.into_data().to_pyarrow(py).unwrap()),
)?;
let py_num_rows = num_rows.into_pyobject(py)?;
let py_args = PyTuple::new(py, vec![py_values.as_any(), &py_num_rows])?;
self.evaluator
.bind(py)
.call_method1("evaluate_all", py_args)
.map(|v| {
let array_data = ArrayData::from_pyarrow_bound(&v).unwrap();
make_array(array_data)
})
})
.map_err(to_datafusion_err)
}
fn evaluate(&mut self, values: &[ArrayRef], range: &Range<usize>) -> Result<ScalarValue> {
Python::with_gil(|py| {
let py_values = PyList::new(
py,
values
.iter()
.map(|arg| arg.into_data().to_pyarrow(py).unwrap()),
)?;
let range_tuple = PyTuple::new(py, vec![range.start, range.end])?;
let py_args = PyTuple::new(py, vec![py_values.as_any(), range_tuple.as_any()])?;
self.evaluator
.bind(py)
.call_method1("evaluate", py_args)
.and_then(|v| v.extract::<PyScalarValue>())
.map(|v| v.0)
})
.map_err(to_datafusion_err)
}
fn evaluate_all_with_rank(
&self,
num_rows: usize,
ranks_in_partition: &[Range<usize>],
) -> Result<ArrayRef> {
Python::with_gil(|py| {
let ranks = ranks_in_partition
.iter()
.map(|r| PyTuple::new(py, vec![r.start, r.end]))
.collect::<PyResult<Vec<_>>>()?;
// 1. cast args to Pyarrow array
let py_args = vec![
num_rows.into_pyobject(py)?.into_any(),
PyList::new(py, ranks)?.into_any(),
];
let py_args = PyTuple::new(py, py_args)?;
// 2. call function
self.evaluator
.bind(py)
.call_method1("evaluate_all_with_rank", py_args)
.map(|v| {
let array_data = ArrayData::from_pyarrow_bound(&v).unwrap();
make_array(array_data)
})
})
.map_err(to_datafusion_err)
}
fn supports_bounded_execution(&self) -> bool {
Python::with_gil(|py| {
self.evaluator
.bind(py)
.call_method0("supports_bounded_execution")
.and_then(|v| v.extract())
.unwrap_or(false)
})
}
fn uses_window_frame(&self) -> bool {
Python::with_gil(|py| {
self.evaluator
.bind(py)
.call_method0("uses_window_frame")
.and_then(|v| v.extract())
.unwrap_or(false)
})
}
fn include_rank(&self) -> bool {
Python::with_gil(|py| {
self.evaluator
.bind(py)
.call_method0("include_rank")
.and_then(|v| v.extract())
.unwrap_or(false)
})
}
}
pub fn to_rust_partition_evaluator(evaluator: PyObject) -> PartitionEvaluatorFactory {
Arc::new(move || -> Result<Box<dyn PartitionEvaluator>> {
let evaluator = Python::with_gil(|py| {
evaluator
.call0(py)
.map_err(|e| DataFusionError::Execution(e.to_string()))
})?;
Ok(Box::new(RustPartitionEvaluator::new(evaluator)))
})
}
/// Represents an WindowUDF
#[pyclass(name = "WindowUDF", module = "datafusion", subclass)]
#[derive(Debug, Clone)]
pub struct PyWindowUDF {
pub(crate) function: WindowUDF,
}
#[pymethods]
impl PyWindowUDF {
#[new]
#[pyo3(signature=(name, evaluator, input_types, return_type, volatility))]
fn new(
name: &str,
evaluator: PyObject,
input_types: Vec<PyArrowType<DataType>>,
return_type: PyArrowType<DataType>,
volatility: &str,
) -> PyResult<Self> {
let return_type = return_type.0;
let input_types = input_types.into_iter().map(|t| t.0).collect();
let function = WindowUDF::from(MultiColumnWindowUDF::new(
name,
input_types,
return_type,
parse_volatility(volatility)?,
to_rust_partition_evaluator(evaluator),
));
Ok(Self { function })
}
/// creates a new PyExpr with the call of the udf
#[pyo3(signature = (*args))]
fn __call__(&self, args: Vec<PyExpr>) -> PyResult<PyExpr> {
let args = args.iter().map(|e| e.expr.clone()).collect();
Ok(self.function.call(args).into())
}
#[staticmethod]
pub fn from_pycapsule(func: Bound<'_, PyAny>) -> PyDataFusionResult<Self> {
if func.hasattr("__datafusion_window_udf__")? {
let capsule = func.getattr("__datafusion_window_udf__")?.call0()?;
let capsule = capsule.downcast::<PyCapsule>().map_err(py_datafusion_err)?;
validate_pycapsule(capsule, "datafusion_window_udf")?;
let udwf = unsafe { capsule.reference::<FFI_WindowUDF>() };
let udwf: ForeignWindowUDF = udwf.try_into()?;
Ok(Self {
function: udwf.into(),
})
} else {
Err(crate::errors::PyDataFusionError::Common(
"__datafusion_window_udf__ does not exist on WindowUDF object.".to_string(),
))
}
}
fn __repr__(&self) -> PyResult<String> {
Ok(format!("WindowUDF({})", self.function.name()))
}
}
pub struct MultiColumnWindowUDF {
name: String,
signature: Signature,
return_type: DataType,
partition_evaluator_factory: PartitionEvaluatorFactory,
}
impl std::fmt::Debug for MultiColumnWindowUDF {
fn fmt(&self, f: &mut std::fmt::Formatter) -> std::fmt::Result {
f.debug_struct("WindowUDF")
.field("name", &self.name)
.field("signature", &self.signature)
.field("return_type", &"<func>")
.field("partition_evaluator_factory", &"<FUNC>")
.finish()
}
}
impl MultiColumnWindowUDF {
pub fn new(
name: impl Into<String>,
input_types: Vec<DataType>,
return_type: DataType,
volatility: Volatility,
partition_evaluator_factory: PartitionEvaluatorFactory,
) -> Self {
let name = name.into();
let signature = Signature::exact(input_types, volatility);
Self {
name,
signature,
return_type,
partition_evaluator_factory,
}
}
}
impl WindowUDFImpl for MultiColumnWindowUDF {
fn as_any(&self) -> &dyn Any {
self
}
fn name(&self) -> &str {
&self.name
}
fn signature(&self) -> &Signature {
&self.signature
}
fn field(&self, field_args: WindowUDFFieldArgs) -> Result<arrow::datatypes::FieldRef> {
// TODO: Should nullable always be `true`?
Ok(arrow::datatypes::Field::new(field_args.name(), self.return_type.clone(), true).into())
}
// TODO: Enable passing partition_evaluator_args to python?
fn partition_evaluator(
&self,
_partition_evaluator_args: PartitionEvaluatorArgs,
) -> Result<Box<dyn PartitionEvaluator>> {
let _ = _partition_evaluator_args;
(self.partition_evaluator_factory)()
}
}