Repository navigation
Expand file tree
/
Copy pathtest_spark_functions.py
More file actions
545 lines (435 loc) · 18.3 KB
/
Copy pathtest_spark_functions.py
File metadata and controls
545 lines (435 loc) · 18.3 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
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
# 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.
"""Tests for the Spark-compatible function bindings."""
import math
import pyarrow as pa
import pytest
from datafusion import SessionContext, col, lit
from datafusion import functions as f
from datafusion.functions import spark
@pytest.fixture
def df():
ctx = SessionContext()
batch = pa.RecordBatch.from_arrays(
[
pa.array(["hello", "WORLD", "abc"]),
pa.array([1, 2, 3]),
pa.array([-5, 0, 7]),
pa.array([1.0, 2.5, 3.0]),
pa.array([[1, 2, 3], [4, 5, 6], [7, 8, 9]]),
],
names=["s", "i", "n", "f", "a"],
)
return ctx.create_dataframe([[batch]])
def _val(df, expr):
return df.select(expr.alias("v")).collect_column("v")[0].as_py()
def _ts():
import datetime as _d
naive = _d.datetime(2020, 1, 15, 14, 30, 45) # noqa: DTZ001
return lit(pa.scalar(naive, type=pa.timestamp("us")))
def _dt(*args):
import datetime as _d
return _d.datetime(*args) # noqa: DTZ001
# ---------------------------------------------------------------------------
# Math
# ---------------------------------------------------------------------------
@pytest.mark.parametrize(
("expr_factory", "expected"),
[
(lambda: spark.abs(lit(-5)), 5),
(lambda: spark.ceil(lit(1.2)), 2),
(lambda: spark.floor(lit(1.8)), 1),
(lambda: spark.bin(lit(7)), "111"),
(lambda: spark.hex(lit(255)), "FF"),
(lambda: spark.modulus(lit(10), lit(3)), 1),
(lambda: spark.pmod(lit(-1), lit(3)), 2),
(lambda: spark.rint(lit(2.5)), 2.0),
(lambda: spark.round(lit(2.5), lit(0)), 3.0),
(lambda: spark.negative(lit(3)), -3),
(lambda: spark.atan2(lit(1.0), lit(1.0)), 0.7853981633974483),
(lambda: spark.atan2(0.0, -1.0), 3.141592653589793),
(lambda: spark.hypot(lit(3.0), lit(4.0)), 5.0),
(lambda: spark.hypot(1e200, 1e200), math.hypot(1e200, 1e200)),
(lambda: spark.pow(lit(2), lit(10)), 1024.0),
(lambda: spark.pow(0.0, -1.0), float("inf")),
(lambda: spark.power(2, 3), 8.0),
],
)
def test_math(df, expr_factory, expected):
assert _val(df, expr_factory()) == expected
@pytest.mark.parametrize(
("expr_factory", "expected"),
[
(lambda: spark.monthname(_ts()), "Jan"),
(lambda: spark.weekday(_ts()), 2),
],
)
def test_monthname_weekday(df, expr_factory, expected):
assert _val(df, expr_factory()) == expected
def test_factorial(df):
# factorial wants Int32; lit(int) is Int64 by default.
expr = spark.factorial(lit(pa.scalar(5, type=pa.int32())))
assert _val(df, expr) == 120
# ---------------------------------------------------------------------------
# String
# ---------------------------------------------------------------------------
@pytest.mark.parametrize(
("expr_factory", "expected"),
[
(lambda: spark.ascii(lit("A")), 65),
(lambda: spark.char(lit(65)), "A"),
(lambda: spark.length(lit("hello")), 5),
(lambda: spark.like(lit("hello"), lit("h%")), True),
(lambda: spark.ilike(lit("HELLO"), lit("h%")), True),
(lambda: spark.substring(lit("hello"), lit(1), lit(3)), "hel"),
(lambda: spark.soundex(lit("Robert")), "R163"),
(lambda: spark.is_valid_utf8(lit("hi")), True),
(lambda: spark.concat(lit("a"), lit("b")), "ab"),
(lambda: spark.elt(lit(2), lit("a"), lit("b")), "b"),
(lambda: spark.quote(lit("it's")), "'it\\'s'"),
(lambda: spark.concat_ws(",", lit("a"), lit("b")), "a,b"),
(lambda: spark.concat_ws(lit("-"), lit("a"), lit(None), lit("b")), "a-b"),
(
lambda: spark.concat_ws(",", f.make_array(lit("a"), lit("b")), lit("c")),
"a,b,c",
),
],
)
def test_string(df, expr_factory, expected):
assert _val(df, expr_factory()) == expected
def test_space(df):
expr = spark.space(lit(pa.scalar(3, type=pa.int32())))
assert _val(df, expr) == " "
# ---------------------------------------------------------------------------
# Hash
# ---------------------------------------------------------------------------
@pytest.mark.parametrize(
("expr_factory", "expected"),
[
(
lambda: spark.sha1(lit("hello")),
"aaf4c61ddcc5e8a2dabede0f3b482cd9aea9434d",
),
(
lambda: spark.sha2(lit("hello"), lit(256)),
"2cf24dba5fb0a30e26e83b2ac5b9e29e1b161e5c1fa7425e73043362938b9824",
),
(lambda: spark.crc32(lit("ABC")), 2743272264),
],
)
def test_hash(df, expr_factory, expected):
assert _val(df, expr_factory()) == expected
# ---------------------------------------------------------------------------
# Bitwise
# ---------------------------------------------------------------------------
@pytest.mark.parametrize(
("expr_factory", "expected"),
[
(lambda: spark.bit_count(lit(7)), 3),
(lambda: spark.bit_get(lit(5), lit(0)), 1),
(lambda: spark.bitwise_not(lit(0)), -1),
(lambda: spark.shiftleft(lit(1), lit(3)), 8),
(lambda: spark.shiftright(lit(8), lit(2)), 2),
(lambda: spark.shiftrightunsigned(lit(8), lit(2)), 2),
],
)
def test_bitwise(df, expr_factory, expected):
assert _val(df, expr_factory()) == expected
# ---------------------------------------------------------------------------
# Array / collection / conditional
# ---------------------------------------------------------------------------
def test_array_and_size(df):
arr = spark.array(lit(1), lit(2), lit(3))
assert _val(df, arr) == [1, 2, 3]
assert _val(df, spark.size(arr)) == 3
assert _val(df, spark.array_contains(arr, lit(2))) is True
def test_slice(df):
assert _val(df, spark.slice(col("a"), lit(2), lit(2))) == [2, 3]
def test_slice_spark_array(df):
arr = spark.array(lit(1), lit(2), lit(3), lit(4))
assert _val(df, spark.slice(arr, lit(2), lit(2))) == [2, 3]
def test_array_repeat(df):
assert _val(df, spark.array_repeat(lit("a"), lit(3))) == ["a", "a", "a"]
def test_if(df):
assert _val(df, spark.if_(lit(True), lit("yes"), lit("no"))) == "yes"
assert _val(df, spark.if_(lit(False), lit("yes"), lit("no"))) == "no"
# ---------------------------------------------------------------------------
# Aggregate
# ---------------------------------------------------------------------------
def test_aggregate(df):
r = df.aggregate(
[],
[
spark.avg(col("f")).alias("avg"),
spark.try_sum(col("i")).alias("sum"),
],
).collect()
rec = pa.Table.from_batches(r)
assert rec.column("avg")[0].as_py() == pytest.approx(2.1666666, rel=1e-3)
assert rec.column("sum")[0].as_py() == 6
def test_collect_list_set(df):
r = df.aggregate(
[],
[
spark.collect_list(col("i")).alias("cl"),
spark.collect_set(col("i")).alias("cs"),
],
).collect()
rec = pa.Table.from_batches(r)
assert sorted(rec.column("cl")[0].as_py()) == [1, 2, 3]
assert sorted(rec.column("cs")[0].as_py()) == [1, 2, 3]
# ---------------------------------------------------------------------------
# Spark-semantics conflicts vs DataFusion defaults
# ---------------------------------------------------------------------------
def test_concat_null_propagates():
"""Spark concat returns NULL on any NULL input; default skips NULLs."""
ctx = SessionContext()
df = ctx.from_pydict({"x": [1]})
default_out = (
df.select(f.concat(lit("a"), lit(None), lit("b")).alias("v"))
.collect_column("v")[0]
.as_py()
)
spark_out = _val(df, spark.concat(lit("a"), lit(None), lit("b")))
assert default_out == "ab"
assert spark_out is None
def test_round_half_up():
"""Spark round uses HALF_UP rounding (2.5 → 3)."""
ctx = SessionContext()
df = ctx.from_pydict({"x": [1]})
assert _val(df, spark.round(lit(2.5), lit(0))) == 3.0
# ---------------------------------------------------------------------------
# Optional parameter defaults / NotImplementedError
# ---------------------------------------------------------------------------
def test_round_scale_default():
"""spark.round defaults scale to 0."""
ctx = SessionContext()
df = ctx.from_pydict({"x": [1]})
assert _val(df, spark.round(lit(2.5))) == 3.0
def test_make_dt_interval_defaults():
"""spark.make_dt_interval with no args returns a zero day-time interval."""
import datetime as dt
ctx = SessionContext()
df = ctx.from_pydict({"x": [1]})
assert _val(df, spark.make_dt_interval()) == dt.timedelta(0)
def test_make_interval_defaults():
"""spark.make_interval with no args returns a zero interval."""
ctx = SessionContext()
df = ctx.from_pydict({"x": [1]})
assert _val(df, spark.make_interval()).months == 0
def test_str_to_map_defaults():
"""spark.str_to_map defaults delimiters to ',' and ':'."""
ctx = SessionContext()
df = ctx.from_pydict({"x": [1]})
assert _val(df, spark.str_to_map(lit("a:1,b:2"))) == [("a", "1"), ("b", "2")]
def test_shuffle_seed_raises():
"""spark.shuffle(seed=...) raises NotImplementedError until Rust supports it."""
with pytest.raises(NotImplementedError, match="seed"):
spark.shuffle(spark.array(lit(1), lit(2)), seed=1)
def test_like_escape_raises():
"""spark.like/ilike escapeChar raises NotImplementedError until Rust supports."""
with pytest.raises(NotImplementedError, match="escapeChar"):
spark.like(lit("a"), lit("a"), escapeChar="\\")
with pytest.raises(NotImplementedError, match="escapeChar"):
spark.ilike(lit("a"), lit("a"), escapeChar="\\")
def test_parse_url_three_arg():
"""parse_url(url, partToExtract, key=...) extracts query parameters."""
ctx = SessionContext()
df = ctx.from_pydict({"x": [1]})
url = lit("http://example.com/p?q=hello&n=1")
assert _val(df, spark.parse_url(url, lit("QUERY"), key=lit("q"))) == "hello"
assert _val(df, spark.try_parse_url(url, lit("QUERY"), key=lit("n"))) == "1"
def test_format_string_plain_str_format():
"""format_string accepts a plain str format that is auto-promoted to lit."""
ctx = SessionContext()
df = ctx.from_pydict({"x": [1]})
assert _val(df, spark.format_string("%d-%s", lit(42), lit("hi"))) == "42-hi"
@pytest.mark.parametrize(
("native_expr", "wrapped_expr", "expected"),
[
# int literals coerced internally
(
lambda: spark.array_repeat(lit("a"), 3),
lambda: spark.array_repeat(lit("a"), lit(3)),
["a", "a", "a"],
),
(
lambda: spark.shiftleft(lit(1), 3),
lambda: spark.shiftleft(lit(1), lit(3)),
8,
),
(
lambda: spark.sha2(lit("hello"), 256),
lambda: spark.sha2(lit("hello"), lit(256)),
"2cf24dba5fb0a30e26e83b2ac5b9e29e1b161e5c1fa7425e73043362938b9824",
),
(
lambda: spark.round(lit(2.345), 2),
lambda: spark.round(lit(2.345), lit(2)),
2.35,
),
(
lambda: spark.substring(lit("hello"), 1, 3),
lambda: spark.substring(lit("hello"), lit(1), lit(3)),
"hel",
),
(
lambda: spark.width_bucket(lit(5.0), lit(0.0), lit(10.0), 5),
lambda: spark.width_bucket(lit(5.0), lit(0.0), lit(10.0), lit(5)),
3,
),
# str literals coerced internally
(
lambda: spark.date_trunc("month", _ts()),
lambda: spark.date_trunc(lit("month"), _ts()),
_dt(2020, 1, 1),
),
(
lambda: spark.spark_cast(lit(1579098645), "timestamp"),
lambda: spark.spark_cast(lit(1579098645), lit("timestamp")),
None,
), # compared separately
# int32-requiring args coerced to int32 internally
(
lambda: spark.space(3),
lambda: spark.space(lit(pa.scalar(3, type=pa.int32()))),
" ",
),
# Any-typed value coerced internally
(
lambda: spark.array_contains(spark.array(lit(1), lit(2)), 1),
lambda: spark.array_contains(spark.array(lit(1), lit(2)), lit(1)),
True,
),
],
)
def test_native_literal_coercion(df, native_expr, wrapped_expr, expected):
"""Native Python literals produce the same result as explicit lit() wrapping."""
if expected is None:
# spark_cast returns a timestamp; just assert native == wrapped.
assert _val(df, native_expr()) == _val(df, wrapped_expr())
return
assert _val(df, native_expr()) == expected
assert _val(df, wrapped_expr()) == expected
def test_native_literal_int32_datetime(df):
"""add_months/date_add/date_sub accept native ints (coerced to int32)."""
import datetime as dt
d = lit(pa.scalar(dt.date(2020, 1, 15), type=pa.date32()))
assert _val(df, spark.add_months(d, 2)) == dt.date(2020, 3, 15)
assert _val(df, spark.date_add(d, 5)) == dt.date(2020, 1, 20)
assert _val(df, spark.date_sub(d, 5)) == dt.date(2020, 1, 10)
assert _val(df, spark.next_day(d, "Mon")) == dt.date(2020, 1, 20)
def test_native_literal_make_interval(df):
"""make_dt_interval/make_interval accept native int/float parts."""
import datetime as dt
assert _val(
df, spark.make_dt_interval(days=1, hours=2, mins=3, secs=4.5)
) == dt.timedelta(days=1, seconds=7384, microseconds=500000)
assert _val(df, spark.make_interval(years=1)).months == 12
def test_str_to_map_pyspark_param_names(df):
"""str_to_map uses pyspark parameter names (pairDelim/keyValueDelim)."""
out = _val(
df, spark.str_to_map(lit("a=1;b=2"), pairDelim=lit(";"), keyValueDelim=lit("="))
)
assert out == [("a", "1"), ("b", "2")]
def test_column_or_name_str_is_column_ref():
"""ColumnOrName args treat a bare str as a column name (matching pyspark)."""
ctx = SessionContext()
df = ctx.from_pydict({"s": ["hello", "world"], "pat": ["h%", "z%"]})
# pattern given as a column name -> per-row pattern from column "pat"
out = (
df.select(spark.like(col("s"), "pat").alias("v"))
.collect_column("v")
.to_pylist()
)
assert out == [True, False]
def test_aggregate_positional_compat():
"""Pyspark-style positional calls still work after the rename to ``col``."""
ctx = SessionContext()
df = ctx.from_pydict({"a": [1.0, 2.0, 3.0]})
out = df.aggregate(
[],
[
spark.avg(col("a")).alias("av"),
spark.try_sum(col("a")).alias("ts"),
spark.collect_list(col("a")).alias("cl"),
spark.collect_set(col("a")).alias("cs"),
],
).collect()
rec = pa.Table.from_batches(out)
assert rec.column("av")[0].as_py() == 2.0
assert rec.column("ts")[0].as_py() == 6.0
# ---------------------------------------------------------------------------
# SQL path via enable_spark_functions
# ---------------------------------------------------------------------------
def test_sql_requires_enable():
"""Spark-only function (xxhash64) is not in SQL namespace by default."""
ctx = SessionContext()
with pytest.raises(Exception, match=r"(?i)xxhash64|Invalid function"):
ctx.sql("SELECT xxhash64('hello')").collect()
def test_sql_after_enable():
ctx = SessionContext()
ctx.enable_spark_functions()
out = ctx.sql("SELECT sha2('hello', 256) AS h").collect_column("h")[0].as_py()
assert out == ("2cf24dba5fb0a30e26e83b2ac5b9e29e1b161e5c1fa7425e73043362938b9824")
def test_sql_concat_semantics_override():
"""After enable, SQL concat propagates NULL like Spark."""
ctx = SessionContext()
default_out = (
ctx.sql("SELECT concat('a', NULL, 'b') AS c").collect_column("c")[0].as_py()
)
assert default_out == "ab"
ctx2 = SessionContext()
ctx2.enable_spark_functions()
spark_out = (
ctx2.sql("SELECT concat('a', NULL, 'b') AS c").collect_column("c")[0].as_py()
)
assert spark_out is None
@pytest.mark.parametrize(
("alias_fn", "primary_fn", "args"),
[
(spark.getbit, spark.bit_get, lambda: (lit(5), lit(0))),
(spark.dateadd, spark.date_add, lambda: (_ts().cast(pa.date32()), 3)),
(
spark.datediff,
spark.date_diff,
lambda: (_ts().cast(pa.date32()), lit("2020-01-01").cast(pa.date32())),
),
(spark.datepart, spark.date_part, lambda: ("YEAR", _ts())),
(spark.sha, spark.sha1, lambda: (lit("abc"),)),
(spark.ceiling, spark.ceil, lambda: (lit(1.2),)),
(spark.char_length, spark.length, lambda: (lit("hello"),)),
(spark.character_length, spark.length, lambda: (lit("hello"),)),
(spark.power, spark.pow, lambda: (lit(2), lit(3))),
(spark.substr, spark.substring, lambda: (lit("hello"), 2, 3)),
],
)
def test_aliases_match_primary(df, alias_fn, primary_fn, args):
assert _val(df, alias_fn(*args())) == _val(df, primary_fn(*args()))
def test_printf_str_is_column_name():
ctx = SessionContext()
df = ctx.from_pydict({"a": ["aa%d%s"], "b": [123], "c": ["cc"]})
assert _val(df, spark.printf("a", "b", "c")) == "aa123cc"
assert _val(df, spark.printf(col("a"), col("b"), col("c"))) == "aa123cc"
assert _val(df, spark.printf(lit("%d-%s"), lit(42), lit("hi"))) == "42-hi"
def test_substr_without_len(df):
assert _val(df, spark.substr(lit("hello"), 2)) == "ello"
assert _val(df, spark.substr(lit("hello"), -3)) == "llo"
def test_last_day_date_keyword(df):
import datetime as dt
d = lit(pa.scalar(dt.date(2024, 2, 10), type=pa.date32()))
assert _val(df, spark.last_day(date=d)) == dt.date(2024, 2, 29)