|
1 | 1 | import collections.abc
|
2 | 2 | import os.path
|
| 3 | +from pathlib import Path |
| 4 | +from shutil import make_archive |
3 | 5 |
|
4 | 6 | import pytest
|
5 | 7 |
|
6 | 8 | import fsspec
|
7 | 9 | from fsspec.implementations.tests.test_archive import archive_data, tempzip
|
| 10 | +from fsspec.implementations.zip import ZipFileSystem |
8 | 11 |
|
9 | 12 |
|
10 | 13 | def test_info():
|
@@ -132,3 +135,340 @@ def test_append(m, tmpdir):
|
132 | 135 | fs.close()
|
133 | 136 |
|
134 | 137 | assert len(fsspec.open_files("zip://*::memory://out.zip")) == 2
|
| 138 | + |
| 139 | + |
| 140 | +@pytest.fixture(name="zip_file") |
| 141 | +def zip_file_fixture(tmp_path): |
| 142 | + data_dir = tmp_path / "data/" |
| 143 | + data_dir.mkdir() |
| 144 | + file1 = data_dir / "file1.txt" |
| 145 | + file1.write_text("Hello, World!") |
| 146 | + file2 = data_dir / "file2.txt" |
| 147 | + file2.write_text("Lorem ipsum dolor sit amet") |
| 148 | + |
| 149 | + empty_dir = data_dir / "dir1" |
| 150 | + empty_dir.mkdir() |
| 151 | + |
| 152 | + dir_with_files = data_dir / "dir2" |
| 153 | + dir_with_files.mkdir() |
| 154 | + file3 = dir_with_files / "file3.txt" |
| 155 | + file3.write_text("Hello!") |
| 156 | + |
| 157 | + potential_mix_up_path = data_dir / "dir2startwithsamename.txt" |
| 158 | + potential_mix_up_path.write_text("Hello again!") |
| 159 | + |
| 160 | + zip_file = tmp_path / "test" |
| 161 | + return Path(make_archive(zip_file, "zip", data_dir)) |
| 162 | + |
| 163 | + |
| 164 | +def _assert_all_except_context_dependent_variables(result, expected_result): |
| 165 | + for path in expected_result.keys(): |
| 166 | + assert result[path] |
| 167 | + result_without_date_time = result[path].copy() |
| 168 | + result_without_date_time.pop("date_time") |
| 169 | + result_without_date_time.pop("_raw_time") |
| 170 | + result_without_date_time.pop("external_attr") |
| 171 | + result_without_date_time.pop("create_system") |
| 172 | + |
| 173 | + expected_result_without_date_time = expected_result[path].copy() |
| 174 | + expected_result_without_date_time.pop("date_time") |
| 175 | + expected_result_without_date_time.pop("_raw_time") |
| 176 | + expected_result_without_date_time.pop("external_attr") |
| 177 | + expected_result_without_date_time.pop("create_system") |
| 178 | + assert result_without_date_time == expected_result_without_date_time |
| 179 | + |
| 180 | + |
| 181 | +def test_find_returns_expected_result_detail_true(zip_file): |
| 182 | + zip_file_system = ZipFileSystem(zip_file) |
| 183 | + |
| 184 | + result = zip_file_system.find("/", detail=True) |
| 185 | + |
| 186 | + expected_result = { |
| 187 | + "dir2/file3.txt": { |
| 188 | + "orig_filename": "dir2/file3.txt", |
| 189 | + "filename": "dir2/file3.txt", |
| 190 | + "date_time": (2024, 8, 16, 10, 46, 18), |
| 191 | + "compress_type": 8, |
| 192 | + "_compresslevel": None, |
| 193 | + "comment": b"", |
| 194 | + "extra": b"", |
| 195 | + "create_system": 3, |
| 196 | + "create_version": 20, |
| 197 | + "extract_version": 20, |
| 198 | + "reserved": 0, |
| 199 | + "flag_bits": 0, |
| 200 | + "volume": 0, |
| 201 | + "internal_attr": 0, |
| 202 | + "external_attr": 2175008768, |
| 203 | + "header_offset": 260, |
| 204 | + "CRC": 2636827734, |
| 205 | + "compress_size": 8, |
| 206 | + "file_size": 6, |
| 207 | + "_raw_time": 21961, |
| 208 | + "_end_offset": 312, |
| 209 | + "name": "dir2/file3.txt", |
| 210 | + "size": 6, |
| 211 | + "type": "file", |
| 212 | + }, |
| 213 | + "file1.txt": { |
| 214 | + "orig_filename": "file1.txt", |
| 215 | + "filename": "file1.txt", |
| 216 | + "date_time": (2024, 8, 16, 10, 46, 18), |
| 217 | + "compress_type": 8, |
| 218 | + "_compresslevel": None, |
| 219 | + "comment": b"", |
| 220 | + "extra": b"", |
| 221 | + "create_system": 3, |
| 222 | + "create_version": 20, |
| 223 | + "extract_version": 20, |
| 224 | + "reserved": 0, |
| 225 | + "flag_bits": 0, |
| 226 | + "volume": 0, |
| 227 | + "internal_attr": 0, |
| 228 | + "external_attr": 2175008768, |
| 229 | + "header_offset": 139, |
| 230 | + "CRC": 3964322768, |
| 231 | + "compress_size": 15, |
| 232 | + "file_size": 13, |
| 233 | + "_raw_time": 21961, |
| 234 | + "_end_offset": 193, |
| 235 | + "name": "file1.txt", |
| 236 | + "size": 13, |
| 237 | + "type": "file", |
| 238 | + }, |
| 239 | + "file2.txt": { |
| 240 | + "orig_filename": "file2.txt", |
| 241 | + "filename": "file2.txt", |
| 242 | + "date_time": (2024, 8, 16, 10, 46, 18), |
| 243 | + "compress_type": 8, |
| 244 | + "_compresslevel": None, |
| 245 | + "comment": b"", |
| 246 | + "extra": b"", |
| 247 | + "create_system": 3, |
| 248 | + "create_version": 20, |
| 249 | + "extract_version": 20, |
| 250 | + "reserved": 0, |
| 251 | + "flag_bits": 0, |
| 252 | + "volume": 0, |
| 253 | + "internal_attr": 0, |
| 254 | + "external_attr": 2175008768, |
| 255 | + "header_offset": 193, |
| 256 | + "CRC": 1596576865, |
| 257 | + "compress_size": 28, |
| 258 | + "file_size": 26, |
| 259 | + "_raw_time": 21961, |
| 260 | + "_end_offset": 260, |
| 261 | + "name": "file2.txt", |
| 262 | + "size": 26, |
| 263 | + "type": "file", |
| 264 | + }, |
| 265 | + } |
| 266 | + |
| 267 | + _assert_all_except_context_dependent_variables(result, expected_result) |
| 268 | + |
| 269 | + |
| 270 | +def test_find_returns_expected_result_detail_false(zip_file): |
| 271 | + zip_file_system = ZipFileSystem(zip_file) |
| 272 | + |
| 273 | + result = zip_file_system.find("/", detail=False) |
| 274 | + expected_result = [ |
| 275 | + "dir2/file3.txt", |
| 276 | + "dir2startwithsamename.txt", |
| 277 | + "file1.txt", |
| 278 | + "file2.txt", |
| 279 | + ] |
| 280 | + |
| 281 | + assert result == expected_result |
| 282 | + |
| 283 | + |
| 284 | +def test_find_returns_expected_result_detail_true_include_dirs(zip_file): |
| 285 | + zip_file_system = ZipFileSystem(zip_file) |
| 286 | + |
| 287 | + result = zip_file_system.find("/", detail=True, withdirs=True) |
| 288 | + expected_result = { |
| 289 | + "dir1": { |
| 290 | + "orig_filename": "dir1/", |
| 291 | + "filename": "dir1/", |
| 292 | + "date_time": (2024, 8, 16, 10, 54, 24), |
| 293 | + "compress_type": 0, |
| 294 | + "_compresslevel": None, |
| 295 | + "comment": b"", |
| 296 | + "extra": b"", |
| 297 | + "create_system": 3, |
| 298 | + "create_version": 20, |
| 299 | + "extract_version": 20, |
| 300 | + "reserved": 0, |
| 301 | + "flag_bits": 0, |
| 302 | + "volume": 0, |
| 303 | + "internal_attr": 0, |
| 304 | + "external_attr": 1106051088, |
| 305 | + "header_offset": 0, |
| 306 | + "CRC": 0, |
| 307 | + "compress_size": 0, |
| 308 | + "file_size": 0, |
| 309 | + "_raw_time": 22220, |
| 310 | + "_end_offset": 35, |
| 311 | + "name": "dir1", |
| 312 | + "size": 0, |
| 313 | + "type": "directory", |
| 314 | + }, |
| 315 | + "dir2": { |
| 316 | + "orig_filename": "dir2/", |
| 317 | + "filename": "dir2/", |
| 318 | + "date_time": (2024, 8, 16, 10, 54, 24), |
| 319 | + "compress_type": 0, |
| 320 | + "_compresslevel": None, |
| 321 | + "comment": b"", |
| 322 | + "extra": b"", |
| 323 | + "create_system": 3, |
| 324 | + "create_version": 20, |
| 325 | + "extract_version": 20, |
| 326 | + "reserved": 0, |
| 327 | + "flag_bits": 0, |
| 328 | + "volume": 0, |
| 329 | + "internal_attr": 0, |
| 330 | + "external_attr": 1106051088, |
| 331 | + "header_offset": 35, |
| 332 | + "CRC": 0, |
| 333 | + "compress_size": 0, |
| 334 | + "file_size": 0, |
| 335 | + "_raw_time": 22220, |
| 336 | + "_end_offset": 70, |
| 337 | + "name": "dir2", |
| 338 | + "size": 0, |
| 339 | + "type": "directory", |
| 340 | + }, |
| 341 | + "dir2/file3.txt": { |
| 342 | + "orig_filename": "dir2/file3.txt", |
| 343 | + "filename": "dir2/file3.txt", |
| 344 | + "date_time": (2024, 8, 16, 10, 54, 24), |
| 345 | + "compress_type": 8, |
| 346 | + "_compresslevel": None, |
| 347 | + "comment": b"", |
| 348 | + "extra": b"", |
| 349 | + "create_system": 3, |
| 350 | + "create_version": 20, |
| 351 | + "extract_version": 20, |
| 352 | + "reserved": 0, |
| 353 | + "flag_bits": 0, |
| 354 | + "volume": 0, |
| 355 | + "internal_attr": 0, |
| 356 | + "external_attr": 2175008768, |
| 357 | + "header_offset": 260, |
| 358 | + "CRC": 2636827734, |
| 359 | + "compress_size": 8, |
| 360 | + "file_size": 6, |
| 361 | + "_raw_time": 22220, |
| 362 | + "_end_offset": 312, |
| 363 | + "name": "dir2/file3.txt", |
| 364 | + "size": 6, |
| 365 | + "type": "file", |
| 366 | + }, |
| 367 | + "file1.txt": { |
| 368 | + "orig_filename": "file1.txt", |
| 369 | + "filename": "file1.txt", |
| 370 | + "date_time": (2024, 8, 16, 10, 54, 24), |
| 371 | + "compress_type": 8, |
| 372 | + "_compresslevel": None, |
| 373 | + "comment": b"", |
| 374 | + "extra": b"", |
| 375 | + "create_system": 3, |
| 376 | + "create_version": 20, |
| 377 | + "extract_version": 20, |
| 378 | + "reserved": 0, |
| 379 | + "flag_bits": 0, |
| 380 | + "volume": 0, |
| 381 | + "internal_attr": 0, |
| 382 | + "external_attr": 2175008768, |
| 383 | + "header_offset": 139, |
| 384 | + "CRC": 3964322768, |
| 385 | + "compress_size": 15, |
| 386 | + "file_size": 13, |
| 387 | + "_raw_time": 22220, |
| 388 | + "_end_offset": 193, |
| 389 | + "name": "file1.txt", |
| 390 | + "size": 13, |
| 391 | + "type": "file", |
| 392 | + }, |
| 393 | + "file2.txt": { |
| 394 | + "orig_filename": "file2.txt", |
| 395 | + "filename": "file2.txt", |
| 396 | + "date_time": (2024, 8, 16, 10, 54, 24), |
| 397 | + "compress_type": 8, |
| 398 | + "_compresslevel": None, |
| 399 | + "comment": b"", |
| 400 | + "extra": b"", |
| 401 | + "create_system": 3, |
| 402 | + "create_version": 20, |
| 403 | + "extract_version": 20, |
| 404 | + "reserved": 0, |
| 405 | + "flag_bits": 0, |
| 406 | + "volume": 0, |
| 407 | + "internal_attr": 0, |
| 408 | + "external_attr": 2175008768, |
| 409 | + "header_offset": 193, |
| 410 | + "CRC": 1596576865, |
| 411 | + "compress_size": 28, |
| 412 | + "file_size": 26, |
| 413 | + "_raw_time": 22220, |
| 414 | + "_end_offset": 260, |
| 415 | + "name": "file2.txt", |
| 416 | + "size": 26, |
| 417 | + "type": "file", |
| 418 | + }, |
| 419 | + } |
| 420 | + |
| 421 | + _assert_all_except_context_dependent_variables(result, expected_result) |
| 422 | + |
| 423 | + |
| 424 | +def test_find_returns_expected_result_detail_false_include_dirs(zip_file): |
| 425 | + zip_file_system = ZipFileSystem(zip_file) |
| 426 | + |
| 427 | + result = zip_file_system.find("/", detail=False, withdirs=True) |
| 428 | + expected_result = [ |
| 429 | + "dir1", |
| 430 | + "dir2", |
| 431 | + "dir2/file3.txt", |
| 432 | + "dir2startwithsamename.txt", |
| 433 | + "file1.txt", |
| 434 | + "file2.txt", |
| 435 | + ] |
| 436 | + |
| 437 | + assert result == expected_result |
| 438 | + |
| 439 | + |
| 440 | +def test_find_returns_expected_result_path_set(zip_file): |
| 441 | + zip_file_system = ZipFileSystem(zip_file) |
| 442 | + |
| 443 | + result = zip_file_system.find("/dir2") |
| 444 | + expected_result = ["dir2/file3.txt"] |
| 445 | + |
| 446 | + assert result == expected_result |
| 447 | + |
| 448 | + |
| 449 | +def test_find_with_and_without_slash_should_return_same_result(zip_file): |
| 450 | + zip_file_system = ZipFileSystem(zip_file) |
| 451 | + |
| 452 | + assert zip_file_system.find("/dir2/") == zip_file_system.find("/dir2") |
| 453 | + |
| 454 | + |
| 455 | +def test_find_should_return_file_if_exact_match(zip_file): |
| 456 | + zip_file_system = ZipFileSystem(zip_file) |
| 457 | + |
| 458 | + result = zip_file_system.find("/dir2startwithsamename.txt", detail=False) |
| 459 | + expected_result = ["dir2startwithsamename.txt"] |
| 460 | + |
| 461 | + assert result == expected_result |
| 462 | + |
| 463 | + |
| 464 | +def test_find_returns_expected_result_recursion_depth_set(zip_file): |
| 465 | + zip_file_system = ZipFileSystem(zip_file) |
| 466 | + result = zip_file_system.find("/", maxdepth=1) |
| 467 | + |
| 468 | + expected_result = [ |
| 469 | + "dir2startwithsamename.txt", |
| 470 | + "file1.txt", |
| 471 | + "file2.txt", |
| 472 | + ] |
| 473 | + |
| 474 | + assert result == expected_result |
0 commit comments