()
| 301 | |
| 302 | |
| 303 | def test_hash_join(): |
| 304 | left = pa.table({'key': [1, 2, 3], 'a': [4, 5, 6]}) |
| 305 | left_source = Declaration("table_source", options=TableSourceNodeOptions(left)) |
| 306 | right = pa.table({'key': [2, 3, 4], 'b': [4, 5, 6]}) |
| 307 | right_source = Declaration("table_source", options=TableSourceNodeOptions(right)) |
| 308 | |
| 309 | # inner join |
| 310 | join_opts = HashJoinNodeOptions("inner", left_keys="key", right_keys="key") |
| 311 | joined = Declaration( |
| 312 | "hashjoin", options=join_opts, inputs=[left_source, right_source]) |
| 313 | result = joined.to_table() |
| 314 | expected = pa.table( |
| 315 | [[2, 3], [5, 6], [2, 3], [4, 5]], |
| 316 | names=["key", "a", "key", "b"]) |
| 317 | assert result.equals(expected) |
| 318 | |
| 319 | for keys in [field("key"), ["key"], [field("key")]]: |
| 320 | join_opts = HashJoinNodeOptions("inner", left_keys=keys, right_keys=keys) |
| 321 | joined = Declaration( |
| 322 | "hashjoin", options=join_opts, inputs=[left_source, right_source]) |
| 323 | result = joined.to_table() |
| 324 | assert result.equals(expected) |
| 325 | |
| 326 | # left join |
| 327 | join_opts = HashJoinNodeOptions( |
| 328 | "left outer", left_keys="key", right_keys="key") |
| 329 | joined = Declaration( |
| 330 | "hashjoin", options=join_opts, inputs=[left_source, right_source]) |
| 331 | result = joined.to_table() |
| 332 | expected = pa.table( |
| 333 | [[1, 2, 3], [4, 5, 6], [None, 2, 3], [None, 4, 5]], |
| 334 | names=["key", "a", "key", "b"] |
| 335 | ) |
| 336 | assert result.sort_by("a").equals(expected) |
| 337 | |
| 338 | # suffixes |
| 339 | join_opts = HashJoinNodeOptions( |
| 340 | "left outer", left_keys="key", right_keys="key", |
| 341 | output_suffix_for_left="_left", output_suffix_for_right="_right") |
| 342 | joined = Declaration( |
| 343 | "hashjoin", options=join_opts, inputs=[left_source, right_source]) |
| 344 | result = joined.to_table() |
| 345 | expected = pa.table( |
| 346 | [[1, 2, 3], [4, 5, 6], [None, 2, 3], [None, 4, 5]], |
| 347 | names=["key_left", "a", "key_right", "b"] |
| 348 | ) |
| 349 | assert result.sort_by("a").equals(expected) |
| 350 | |
| 351 | # manually specifying output columns |
| 352 | join_opts = HashJoinNodeOptions( |
| 353 | "left outer", left_keys="key", right_keys="key", |
| 354 | left_output=["key", "a"], right_output=[field("b")]) |
| 355 | joined = Declaration( |
| 356 | "hashjoin", options=join_opts, inputs=[left_source, right_source]) |
| 357 | result = joined.to_table() |
| 358 | expected = pa.table( |
| 359 | [[1, 2, 3], [4, 5, 6], [None, 4, 5]], |
| 360 | names=["key", "a", "b"] |
nothing calls this directly
no test coverage detected