|
| 1 | +import io |
1 | 2 | import secrets |
2 | 3 |
|
3 | 4 | import pytest |
@@ -306,6 +307,41 @@ def test_seekable(fs, remote_dir): |
306 | 307 | file.seek(5) |
307 | 308 |
|
308 | 309 |
|
| 310 | +def test_readinto(fs, remote_dir): |
| 311 | + data = b"dvc.org" |
| 312 | + |
| 313 | + with fs.open(remote_dir + "/a.txt", "wb") as stream: |
| 314 | + stream.write(data) |
| 315 | + |
| 316 | + for seekable in [True, False]: |
| 317 | + with fs.open(remote_dir + "/a.txt", "rb", seekable=seekable) as file: |
| 318 | + buffer = bytearray(3) |
| 319 | + assert file.readinto(buffer) == 3 |
| 320 | + assert bytes(buffer) == data[:3] |
| 321 | + |
| 322 | + |
| 323 | +@pytest.mark.parametrize( |
| 324 | + "read", |
| 325 | + [ |
| 326 | + lambda buffered: buffered.read(3), |
| 327 | + lambda buffered: buffered.peek(3)[:3], |
| 328 | + lambda buffered: buffered.read1(3), |
| 329 | + ], |
| 330 | + ids=["read", "peek", "read1"], |
| 331 | +) |
| 332 | +def test_readinto_supports_a_buffered_reader(fs, remote_dir, read): |
| 333 | + # Each of these fills a buffer io.BufferedReader owns, so they go through |
| 334 | + # readinto rather than read(); an unsized read() does not. A gzip reader takes |
| 335 | + # the peek()/read1() path. |
| 336 | + data = b"dvc.org" |
| 337 | + |
| 338 | + with fs.open(remote_dir + "/a.txt", "wb") as stream: |
| 339 | + stream.write(data) |
| 340 | + |
| 341 | + with fs.open(remote_dir + "/a.txt", "rb") as file: |
| 342 | + assert read(io.BufferedReader(file)) == data[:3] |
| 343 | + |
| 344 | + |
309 | 345 | def test_get_kwargs_from_urls_hadoop_fs(): |
310 | 346 | kwargs = HadoopFileSystem._get_kwargs_from_urls( |
311 | 347 | "hdfs://user@localhost:8020/?replication=2" |
|
0 commit comments