1 # Copyright (C) The Arvados Authors. All rights reserved.
3 # SPDX-License-Identifier: Apache-2.0
5 from __future__ import absolute_import
6 from builtins import hex
7 from builtins import str
8 from builtins import range
9 from builtins import object
15 from unittest import mock
18 from arvados._ranges import Range
19 from arvados.keep import KeepLocator
20 from arvados.collection import Collection
21 from arvados.arvfile import ArvadosFile, ArvadosFileReader
23 from . import arvados_testutil as tutil
24 from .test_stream import StreamFileReaderTestCase, StreamRetryTestMixin
26 class ArvadosFileWriterTestCase(unittest.TestCase):
27 class MockKeep(object):
28 def __init__(self, blocks):
31 self.num_prefetch_threads = 1
32 def get(self, locator, num_retries=0, prefetch=False):
33 self.requests.append(locator)
34 return self.blocks.get(locator)
35 def get_from_cache(self, locator):
36 self.requests.append(locator)
37 return self.blocks.get(locator)
38 def put(self, data, num_retries=None, copies=None, classes=[]):
39 pdh = tutil.str_keep_locator(data)
40 self.blocks[pdh] = bytes(data)
42 def block_prefetch(self, loc):
43 self.requests.append(loc)
45 class MockApi(object):
46 def __init__(self, b, r):
49 self._schema = ArvadosFileWriterTestCase.MockApi.MockSchema()
51 class MockSchema(object):
53 self.schemas = {'Collection': {'properties': {'replication_desired': {'type':'integer'}}}}
54 class MockCollections(object):
55 def __init__(self, b, r):
58 class Execute(object):
59 def __init__(self, r):
61 def execute(self, num_retries=None):
63 def create(self, ensure_unique_name=False, body=None):
65 raise Exception("Body %s does not match expectation %s" % (body, self.body))
66 return ArvadosFileWriterTestCase.MockApi.MockCollections.Execute(self.response)
67 def update(self, uuid=None, body=None):
68 return ArvadosFileWriterTestCase.MockApi.MockCollections.Execute(self.response)
69 def collections(self):
70 return ArvadosFileWriterTestCase.MockApi.MockCollections(self.body, self.response)
73 def test_truncate(self):
74 keep = ArvadosFileWriterTestCase.MockKeep({
75 "781e5e245d69b566979b86e28d23f2c7+10": b"0123456789",
77 api = ArvadosFileWriterTestCase.MockApi({
78 "name": "test_truncate",
79 "manifest_text": ". 781e5e245d69b566979b86e28d23f2c7+10 0:8:count.txt\n",
80 "replication_desired": None,
82 "uuid": "zzzzz-4zz18-mockcollection0",
83 "manifest_text": ". 781e5e245d69b566979b86e28d23f2c7+10 0:8:count.txt\n",
84 "portable_data_hash":"7fcd0eaac3aad4c31a6a0e756475da92+52",
86 with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n',
87 api_client=api, keep_client=keep) as c:
88 writer = c.open("count.txt", "rb+")
89 self.assertEqual(writer.size(), 10)
90 self.assertEqual(b"0123456789", writer.read(12))
94 # Make sure reading off the end doesn't break
95 self.assertEqual(b"", writer.read(12))
97 self.assertEqual(writer.size(), 8)
98 writer.seek(0, os.SEEK_SET)
99 self.assertEqual(b"01234567", writer.read(12))
101 self.assertIsNone(c.manifest_locator())
102 self.assertTrue(c.modified())
103 c.save_new("test_truncate")
104 self.assertEqual("zzzzz-4zz18-mockcollection0", c.manifest_locator())
105 self.assertFalse(c.modified())
108 def test_truncate2(self):
109 keep = ArvadosFileWriterTestCase.MockKeep({
110 "781e5e245d69b566979b86e28d23f2c7+10": b"0123456789",
112 api = ArvadosFileWriterTestCase.MockApi({
113 "name": "test_truncate2",
114 "manifest_text": ". 781e5e245d69b566979b86e28d23f2c7+10 7f614da9329cd3aebf59b91aadc30bf0+67108864 0:12:count.txt\n",
115 "replication_desired": None,
117 "uuid": "zzzzz-4zz18-mockcollection0",
118 "manifest_text": ". 781e5e245d69b566979b86e28d23f2c7+10 7f614da9329cd3aebf59b91aadc30bf0+67108864 0:12:count.txt\n",
119 "portable_data_hash": "272da898abdf86ddc71994835e3155f8+95",
121 with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n',
122 api_client=api, keep_client=keep) as c:
123 writer = c.open("count.txt", "rb+")
124 self.assertEqual(writer.size(), 10)
125 self.assertEqual(b"0123456789", writer.read(12))
130 self.assertEqual(writer.size(), 12)
131 writer.seek(0, os.SEEK_SET)
132 self.assertEqual(b"0123456789\x00\x00", writer.read(12))
134 self.assertIsNone(c.manifest_locator())
135 self.assertTrue(c.modified())
136 c.save_new("test_truncate2")
137 self.assertEqual("zzzzz-4zz18-mockcollection0", c.manifest_locator())
138 self.assertFalse(c.modified())
140 def test_truncate3(self):
141 keep = ArvadosFileWriterTestCase.MockKeep({
142 "781e5e245d69b566979b86e28d23f2c7+10": b"0123456789",
143 "a925576942e94b2ef57a066101b48876+10": b"abcdefghij",
145 api = ArvadosFileWriterTestCase.MockApi({
146 "name": "test_truncate",
147 "manifest_text": ". 781e5e245d69b566979b86e28d23f2c7+10 0:8:count.txt\n",
148 "replication_desired": None,
150 "uuid": "zzzzz-4zz18-mockcollection0",
151 "manifest_text": ". 781e5e245d69b566979b86e28d23f2c7+10 0:8:count.txt\n",
152 "portable_data_hash": "7fcd0eaac3aad4c31a6a0e756475da92+52",
154 with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 a925576942e94b2ef57a066101b48876+10 0:20:count.txt\n',
155 api_client=api, keep_client=keep) as c:
156 writer = c.open("count.txt", "rb+")
157 self.assertEqual(writer.size(), 20)
158 self.assertEqual(b"0123456789ab", writer.read(12))
159 self.assertEqual(12, writer.tell())
163 # Make sure reading off the end doesn't break
164 self.assertEqual(12, writer.tell())
165 self.assertEqual(b"", writer.read(12))
167 self.assertEqual(writer.size(), 8)
168 self.assertEqual(2, writer.seek(-10, os.SEEK_CUR))
169 self.assertEqual(b"234567", writer.read(12))
171 self.assertIsNone(c.manifest_locator())
172 self.assertTrue(c.modified())
173 c.save_new("test_truncate")
174 self.assertEqual("zzzzz-4zz18-mockcollection0", c.manifest_locator())
175 self.assertFalse(c.modified())
177 def test_write_to_end(self):
178 keep = ArvadosFileWriterTestCase.MockKeep({
179 "781e5e245d69b566979b86e28d23f2c7+10": b"0123456789",
181 api = ArvadosFileWriterTestCase.MockApi({
182 "name": "test_append",
183 "manifest_text": ". 781e5e245d69b566979b86e28d23f2c7+10 acbd18db4cc2f85cedef654fccc4a4d8+3 0:13:count.txt\n",
184 "replication_desired": None,
186 "uuid": "zzzzz-4zz18-mockcollection0",
187 "manifest_text": ". 781e5e245d69b566979b86e28d23f2c7+10 acbd18db4cc2f85cedef654fccc4a4d8+3 0:13:count.txt\n",
188 "portable_data_hash": "c5c3af76565c8efb6a806546bcf073f3+88",
190 with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n',
191 api_client=api, keep_client=keep) as c:
192 writer = c.open("count.txt", "rb+")
193 self.assertEqual(writer.size(), 10)
195 self.assertEqual(5, writer.seek(5, os.SEEK_SET))
196 self.assertEqual(b"56789", writer.read(8))
198 writer.seek(10, os.SEEK_SET)
200 self.assertEqual(writer.size(), 13)
202 writer.seek(5, os.SEEK_SET)
203 self.assertEqual(b"56789foo", writer.read(8))
205 self.assertIsNone(c.manifest_locator())
206 self.assertTrue(c.modified())
207 self.assertIsNone(keep.get("acbd18db4cc2f85cedef654fccc4a4d8+3"))
209 c.save_new("test_append")
210 self.assertEqual("zzzzz-4zz18-mockcollection0", c.manifest_locator())
211 self.assertFalse(c.modified())
212 self.assertEqual(b"foo", keep.get("acbd18db4cc2f85cedef654fccc4a4d8+3"))
215 def test_append(self):
216 keep = ArvadosFileWriterTestCase.MockKeep({
217 "781e5e245d69b566979b86e28d23f2c7+10": b"0123456789",
219 for (mode, convert) in (
220 ('a+', lambda data: data.decode(encoding='utf-8')),
221 ('at+', lambda data: data.decode(encoding='utf-8')),
222 ('ab+', lambda data: data)):
223 c = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n', keep_client=keep)
224 writer = c.open("count.txt", mode)
225 self.assertEqual(writer.read(20), convert(b"0123456789"))
227 writer.seek(0, os.SEEK_SET)
228 writer.write(convert(b"hello"))
229 self.assertEqual(writer.read(), convert(b""))
231 writer.seek(-5, os.SEEK_CUR)
232 self.assertEqual(writer.read(3), convert(b"hel"))
233 self.assertEqual(writer.read(), convert(b"lo"))
235 with self.assertRaises(IOError):
236 writer.seek(-5, os.SEEK_CUR)
237 with self.assertRaises(IOError):
238 writer.seek(-3, os.SEEK_END)
239 writer.seek(0, os.SEEK_SET)
241 self.assertEqual(7, writer.tell())
242 self.assertEqual(7, writer.seek(7, os.SEEK_SET))
244 writer.seek(0, os.SEEK_SET)
245 self.assertEqual(writer.read(), convert(b"0123456789hello"))
248 writer.write(convert(b"world"))
249 self.assertEqual(writer.read(), convert(b""))
251 self.assertEqual(writer.read(), convert(b"0123456789helloworld"))
253 self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 fc5e038d38a57032085441e7fe7010b0+10 0:20:count.txt\n", c.portable_manifest_text())
255 def test_write_at_beginning(self):
256 keep = ArvadosFileWriterTestCase.MockKeep({
257 "781e5e245d69b566979b86e28d23f2c7+10": b"0123456789",
259 with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n',
260 keep_client=keep) as c:
261 writer = c.open("count.txt", "rb+")
262 self.assertEqual(b"0123456789", writer.readfrom(0, 13))
263 writer.seek(0, os.SEEK_SET)
265 self.assertEqual(writer.size(), 10)
266 self.assertEqual(b"foo3456789", writer.readfrom(0, 13))
267 self.assertEqual(". acbd18db4cc2f85cedef654fccc4a4d8+3 781e5e245d69b566979b86e28d23f2c7+10 0:3:count.txt 6:7:count.txt\n", c.portable_manifest_text())
269 def test_write_empty(self):
270 keep = ArvadosFileWriterTestCase.MockKeep({})
271 with Collection(keep_client=keep) as c:
272 writer = c.open("count.txt", "wb")
273 self.assertEqual(writer.size(), 0)
274 self.assertEqual(". d41d8cd98f00b204e9800998ecf8427e+0 0:0:count.txt\n", c.portable_manifest_text())
276 def test_save_manifest_text(self):
277 keep = ArvadosFileWriterTestCase.MockKeep({})
278 with Collection(keep_client=keep) as c:
279 writer = c.open("count.txt", "wb")
280 writer.write(b"0123456789")
281 self.assertEqual('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n', c.portable_manifest_text())
282 self.assertNotIn('781e5e245d69b566979b86e28d23f2c7+10', keep.blocks)
284 self.assertEqual('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n', c.save_new(create_collection_record=False))
285 self.assertIn('781e5e245d69b566979b86e28d23f2c7+10', keep.blocks)
287 def test_get_manifest_text_commits(self):
288 keep = ArvadosFileWriterTestCase.MockKeep({})
289 with Collection(keep_client=keep) as c:
290 writer = c.open("count.txt", "wb")
291 writer.write("0123456789")
292 self.assertEqual('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n', c.portable_manifest_text())
293 self.assertNotIn('781e5e245d69b566979b86e28d23f2c7+10', keep.blocks)
294 self.assertEqual('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n', c.manifest_text())
295 self.assertIn('781e5e245d69b566979b86e28d23f2c7+10', keep.blocks)
298 def test_write_in_middle(self):
299 keep = ArvadosFileWriterTestCase.MockKeep({"781e5e245d69b566979b86e28d23f2c7+10": b"0123456789"})
300 with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n',
301 keep_client=keep) as c:
302 writer = c.open("count.txt", "rb+")
303 self.assertEqual(b"0123456789", writer.readfrom(0, 13))
304 writer.seek(3, os.SEEK_SET)
306 self.assertEqual(writer.size(), 10)
307 self.assertEqual(b"012foo6789", writer.readfrom(0, 13))
308 self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 acbd18db4cc2f85cedef654fccc4a4d8+3 0:3:count.txt 10:3:count.txt 6:4:count.txt\n", c.portable_manifest_text())
310 def test_write_at_end(self):
311 keep = ArvadosFileWriterTestCase.MockKeep({"781e5e245d69b566979b86e28d23f2c7+10": b"0123456789"})
312 with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n',
313 keep_client=keep) as c:
314 writer = c.open("count.txt", "rb+")
315 self.assertEqual(b"0123456789", writer.readfrom(0, 13))
316 writer.seek(7, os.SEEK_SET)
318 self.assertEqual(writer.size(), 10)
319 self.assertEqual(b"0123456foo", writer.readfrom(0, 13))
320 self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 acbd18db4cc2f85cedef654fccc4a4d8+3 0:7:count.txt 10:3:count.txt\n", c.portable_manifest_text())
322 def test_write_across_segment_boundary(self):
323 keep = ArvadosFileWriterTestCase.MockKeep({"781e5e245d69b566979b86e28d23f2c7+10": b"0123456789"})
324 with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt 0:10:count.txt\n',
325 keep_client=keep) as c:
326 writer = c.open("count.txt", "rb+")
327 self.assertEqual(b"012345678901234", writer.readfrom(0, 15))
328 writer.seek(7, os.SEEK_SET)
329 writer.write("foobar")
330 self.assertEqual(writer.size(), 20)
331 self.assertEqual(b"0123456foobar34", writer.readfrom(0, 15))
332 self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 3858f62230ac3c915f300c664312c63f+6 0:7:count.txt 10:6:count.txt 3:7:count.txt\n", c.portable_manifest_text())
334 def test_write_across_several_segments(self):
335 keep = ArvadosFileWriterTestCase.MockKeep({"781e5e245d69b566979b86e28d23f2c7+10": b"0123456789"})
336 with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:4:count.txt 0:4:count.txt 0:4:count.txt',
337 keep_client=keep) as c:
338 writer = c.open("count.txt", "rb+")
339 self.assertEqual(b"012301230123", writer.readfrom(0, 15))
340 writer.seek(2, os.SEEK_SET)
341 writer.write("abcdefg")
342 self.assertEqual(writer.size(), 12)
343 self.assertEqual(b"01abcdefg123", writer.readfrom(0, 15))
344 self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 7ac66c0f148de9519b8bd264312c4d64+7 0:2:count.txt 10:7:count.txt 1:3:count.txt\n", c.portable_manifest_text())
346 def test_write_large(self):
347 keep = ArvadosFileWriterTestCase.MockKeep({})
348 api = ArvadosFileWriterTestCase.MockApi({"name":"test_write_large",
349 "manifest_text": ". a5de24f4417cfba9d5825eadc2f4ca49+67108000 598cc1a4ccaef8ab6e4724d87e675d78+32892000 0:100000000:count.txt\n",
350 "replication_desired":None},
351 {"uuid":"zzzzz-4zz18-mockcollection0",
352 "manifest_text": ". a5de24f4417cfba9d5825eadc2f4ca49+67108000 598cc1a4ccaef8ab6e4724d87e675d78+32892000 0:100000000:count.txt\n",
353 "portable_data_hash":"9132ca8e3f671c76103a38f5bc24328c+108"})
354 with Collection('. ' + arvados.config.EMPTY_BLOCK_LOCATOR + ' 0:0:count.txt',
355 api_client=api, keep_client=keep) as c:
356 writer = c.open("count.txt", "rb+")
357 text = "0123456789" * 100
358 for b in range(0, 100000):
360 self.assertEqual(writer.size(), 100000000)
362 self.assertIsNone(c.manifest_locator())
363 self.assertTrue(c.modified())
364 c.save_new("test_write_large")
365 self.assertEqual("zzzzz-4zz18-mockcollection0", c.manifest_locator())
366 self.assertFalse(c.modified())
368 def test_large_write(self):
369 keep = ArvadosFileWriterTestCase.MockKeep({})
370 api = ArvadosFileWriterTestCase.MockApi({}, {})
371 with Collection('. ' + arvados.config.EMPTY_BLOCK_LOCATOR + ' 0:0:count.txt',
372 api_client=api, keep_client=keep) as c:
373 writer = c.open("count.txt", "rb+")
374 self.assertEqual(writer.size(), 0)
378 text = "0123456789" * 9999999
380 self.assertEqual(writer.size(), 100000000)
382 self.assertEqual(c.manifest_text(), ". 781e5e245d69b566979b86e28d23f2c7+10 48dd23ea1645fd47d789804d71b5bb8e+67108864 77c57dc6ac5a10bb2205caaa73187994+32891126 0:100000000:count.txt\n")
384 def test_sparse_write(self):
385 keep = ArvadosFileWriterTestCase.MockKeep({})
386 api = ArvadosFileWriterTestCase.MockApi({}, {})
387 with Collection('. ' + arvados.config.EMPTY_BLOCK_LOCATOR + ' 0:0:count.txt',
388 api_client=api, keep_client=keep) as c:
389 writer = c.open("count.txt", "rb+")
390 self.assertEqual(writer.size(), 0)
395 self.assertEqual(writer.size(), 12)
396 writer.seek(0, os.SEEK_SET)
397 self.assertEqual(writer.read(), b"\x00\x00"+text)
399 self.assertEqual(c.manifest_text(), ". 7f614da9329cd3aebf59b91aadc30bf0+67108864 781e5e245d69b566979b86e28d23f2c7+10 0:2:count.txt 67108864:10:count.txt\n")
401 def test_sparse_write2(self):
402 keep = ArvadosFileWriterTestCase.MockKeep({})
403 api = ArvadosFileWriterTestCase.MockApi({}, {})
404 with Collection('. ' + arvados.config.EMPTY_BLOCK_LOCATOR + ' 0:0:count.txt',
405 api_client=api, keep_client=keep) as c:
406 writer = c.open("count.txt", "rb+")
407 self.assertEqual(writer.size(), 0)
410 writer.seek((arvados.config.KEEP_BLOCK_SIZE*2) + 2)
412 self.assertEqual(writer.size(), (arvados.config.KEEP_BLOCK_SIZE*2) + 12)
413 writer.seek(0, os.SEEK_SET)
415 self.assertEqual(c.manifest_text(), ". 7f614da9329cd3aebf59b91aadc30bf0+67108864 781e5e245d69b566979b86e28d23f2c7+10 0:67108864:count.txt 0:67108864:count.txt 0:2:count.txt 67108864:10:count.txt\n")
417 def test_sparse_write3(self):
418 keep = ArvadosFileWriterTestCase.MockKeep({})
419 api = ArvadosFileWriterTestCase.MockApi({}, {})
420 for r in [[0, 1, 2, 3, 4], [4, 3, 2, 1, 0], [3, 2, 0, 4, 1]]:
421 with Collection(api_client=api, keep_client=keep) as c:
422 writer = c.open("count.txt", "rb+")
423 self.assertEqual(writer.size(), 0)
428 writer.write(w.encode())
430 self.assertEqual(writer.read(), b"00000000001111111111222222222233333333334444444444")
432 def test_sparse_write4(self):
433 keep = ArvadosFileWriterTestCase.MockKeep({})
434 api = ArvadosFileWriterTestCase.MockApi({}, {})
435 for r in [[0, 1, 2, 4], [4, 2, 1, 0], [2, 0, 4, 1]]:
436 with Collection(api_client=api, keep_client=keep) as c:
437 writer = c.open("count.txt", "rb+")
438 self.assertEqual(writer.size(), 0)
443 writer.write(w.encode())
445 self.assertEqual(writer.read(), b"000000000011111111112222222222\x00\x00\x00\x00\x00\x00\x00\x00\x00\x004444444444")
447 def test_rewrite_on_empty_file(self):
448 keep = ArvadosFileWriterTestCase.MockKeep({})
449 with Collection('. ' + arvados.config.EMPTY_BLOCK_LOCATOR + ' 0:0:count.txt',
450 keep_client=keep) as c:
451 writer = c.open("count.txt", "rb+")
452 for b in range(0, 10):
453 writer.seek(0, os.SEEK_SET)
454 writer.write("0123456789")
456 self.assertEqual(writer.size(), 10)
457 self.assertEqual(b"0123456789", writer.readfrom(0, 20))
458 self.assertEqual(". 7a08b07e84641703e5f2c836aa59a170+100 90:10:count.txt\n", c.portable_manifest_text())
460 self.assertEqual(writer.size(), 10)
461 self.assertEqual(b"0123456789", writer.readfrom(0, 20))
462 self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n", c.portable_manifest_text())
464 def test_rewrite_append_existing_file(self):
465 keep = ArvadosFileWriterTestCase.MockKeep({
466 "781e5e245d69b566979b86e28d23f2c7+10": b"0123456789",
468 with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt',
469 keep_client=keep) as c:
470 writer = c.open("count.txt", "rb+")
471 for b in range(0, 10):
472 writer.seek(10, os.SEEK_SET)
473 writer.write("abcdefghij")
475 self.assertEqual(writer.size(), 20)
476 self.assertEqual(b"0123456789abcdefghij", writer.readfrom(0, 20))
477 self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 ae5f43bab79cf0be33f025fa97ae7398+100 0:10:count.txt 100:10:count.txt\n", c.portable_manifest_text())
479 writer.arvadosfile.flush()
480 self.assertEqual(writer.size(), 20)
481 self.assertEqual(b"0123456789abcdefghij", writer.readfrom(0, 20))
482 self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 a925576942e94b2ef57a066101b48876+10 0:20:count.txt\n", c.portable_manifest_text())
484 def test_rewrite_over_existing_file(self):
485 keep = ArvadosFileWriterTestCase.MockKeep({
486 "781e5e245d69b566979b86e28d23f2c7+10": b"0123456789",
488 with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt',
489 keep_client=keep) as c:
490 writer = c.open("count.txt", "rb+")
491 for b in range(0, 10):
492 writer.seek(5, os.SEEK_SET)
493 writer.write("abcdefghij")
495 self.assertEqual(writer.size(), 15)
496 self.assertEqual(b"01234abcdefghij", writer.readfrom(0, 20))
497 self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 ae5f43bab79cf0be33f025fa97ae7398+100 0:5:count.txt 100:10:count.txt\n", c.portable_manifest_text())
499 writer.arvadosfile.flush()
501 self.assertEqual(writer.size(), 15)
502 self.assertEqual(b"01234abcdefghij", writer.readfrom(0, 20))
503 self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 a925576942e94b2ef57a066101b48876+10 0:5:count.txt 10:10:count.txt\n", c.portable_manifest_text())
505 def test_write_large_rewrite(self):
506 keep = ArvadosFileWriterTestCase.MockKeep({})
507 api = ArvadosFileWriterTestCase.MockApi({"name":"test_write_large",
508 "manifest_text": ". 3dc0d4bc21f48060bedcb2c91af4f906+32892003 a5de24f4417cfba9d5825eadc2f4ca49+67108000 0:3:count.txt 32892006:67107997:count.txt 3:32892000:count.txt\n",
509 "replication_desired":None},
510 {"uuid":"zzzzz-4zz18-mockcollection0",
511 "manifest_text": ". 3dc0d4bc21f48060bedcb2c91af4f906+32892003 a5de24f4417cfba9d5825eadc2f4ca49+67108000 0:3:count.txt 32892006:67107997:count.txt 3:32892000:count.txt\n",
512 "portable_data_hash":"217665c6b713e1b78dfba7ebd42344db+156"})
513 with Collection('. ' + arvados.config.EMPTY_BLOCK_LOCATOR + ' 0:0:count.txt',
514 api_client=api, keep_client=keep) as c:
515 writer = c.open("count.txt", "rb+")
516 text = b''.join([b"0123456789" for a in range(0, 100)])
517 for b in range(0, 100000):
519 writer.seek(0, os.SEEK_SET)
521 self.assertEqual(writer.size(), 100000000)
523 self.assertIsNone(c.manifest_locator())
524 self.assertTrue(c.modified())
525 c.save_new("test_write_large")
526 self.assertEqual("zzzzz-4zz18-mockcollection0", c.manifest_locator())
527 self.assertFalse(c.modified())
529 def test_create(self):
530 keep = ArvadosFileWriterTestCase.MockKeep({})
531 api = ArvadosFileWriterTestCase.MockApi({
532 "name":"test_create",
533 "manifest_text":". 2e9ec317e197819358fbc43afca7d837+8 0:8:count.txt\n",
534 "replication_desired":None,
536 "uuid":"zzzzz-4zz18-mockcollection0",
537 "manifest_text":". 2e9ec317e197819358fbc43afca7d837+8 0:8:count.txt\n",
538 "portable_data_hash":"7a461a8c58601798f690f8b368ac4423+51",
540 with Collection(api_client=api, keep_client=keep) as c:
541 writer = c.open("count.txt", "wb+")
542 self.assertEqual(writer.size(), 0)
543 writer.write("01234567")
544 self.assertEqual(writer.size(), 8)
546 self.assertIsNone(c.manifest_locator())
547 self.assertTrue(c.modified())
548 self.assertIsNone(keep.get("2e9ec317e197819358fbc43afca7d837+8"))
549 c.save_new("test_create")
550 self.assertEqual("zzzzz-4zz18-mockcollection0", c.manifest_locator())
551 self.assertFalse(c.modified())
552 self.assertEqual(b"01234567", keep.get("2e9ec317e197819358fbc43afca7d837+8"))
555 def test_create_subdir(self):
556 keep = ArvadosFileWriterTestCase.MockKeep({})
557 api = ArvadosFileWriterTestCase.MockApi({"name":"test_create",
558 "manifest_text":"./foo/bar 2e9ec317e197819358fbc43afca7d837+8 0:8:count.txt\n",
559 "replication_desired":None},
560 {"uuid":"zzzzz-4zz18-mockcollection0",
561 "manifest_text":"./foo/bar 2e9ec317e197819358fbc43afca7d837+8 0:8:count.txt\n",
562 "portable_data_hash":"1b02aaa62528d28a5be41651cbb9d7c7+59"})
563 with Collection(api_client=api, keep_client=keep) as c:
564 self.assertIsNone(c.api_response())
565 writer = c.open("foo/bar/count.txt", "wb+")
566 writer.write("01234567")
567 self.assertFalse(c.committed())
568 c.save_new("test_create")
569 self.assertTrue(c.committed())
570 self.assertEqual(c.api_response(), api.response)
572 def test_overwrite(self):
573 keep = ArvadosFileWriterTestCase.MockKeep({"781e5e245d69b566979b86e28d23f2c7+10": "0123456789"})
574 api = ArvadosFileWriterTestCase.MockApi({"name":"test_overwrite",
575 "manifest_text":". 2e9ec317e197819358fbc43afca7d837+8 0:8:count.txt\n",
576 "replication_desired":None},
577 {"uuid":"zzzzz-4zz18-mockcollection0",
578 "manifest_text":". 2e9ec317e197819358fbc43afca7d837+8 0:8:count.txt\n",
579 "portable_data_hash":"7a461a8c58601798f690f8b368ac4423+51"})
580 with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n',
581 api_client=api, keep_client=keep) as c:
582 writer = c.open("count.txt", "wb+")
583 self.assertEqual(writer.size(), 0)
584 writer.write("01234567")
585 self.assertEqual(writer.size(), 8)
587 self.assertIsNone(c.manifest_locator())
588 self.assertTrue(c.modified())
589 c.save_new("test_overwrite")
590 self.assertEqual("zzzzz-4zz18-mockcollection0", c.manifest_locator())
591 self.assertFalse(c.modified())
593 def test_file_not_found(self):
594 with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n') as c:
595 with self.assertRaises(IOError):
596 writer = c.open("nocount.txt", "rb")
598 def test_cannot_open_directory(self):
599 with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n') as c:
600 with self.assertRaises(IOError):
601 writer = c.open(".", "rb")
603 def test_create_multiple(self):
604 keep = ArvadosFileWriterTestCase.MockKeep({})
605 api = ArvadosFileWriterTestCase.MockApi({"name":"test_create_multiple",
606 "manifest_text":". 2e9ec317e197819358fbc43afca7d837+8 e8dc4081b13434b45189a720b77b6818+8 0:8:count1.txt 8:8:count2.txt\n",
607 "replication_desired":None},
608 {"uuid":"zzzzz-4zz18-mockcollection0",
609 "manifest_text":". 2e9ec317e197819358fbc43afca7d837+8 e8dc4081b13434b45189a720b77b6818+8 0:8:count1.txt 8:8:count2.txt\n",
610 "portable_data_hash":"71e7bb6c00d31fc2b4364199fd97be08+102"})
611 with Collection(api_client=api, keep_client=keep) as c:
612 w1 = c.open("count1.txt", "wb")
613 w2 = c.open("count2.txt", "wb")
616 self.assertEqual(w1.size(), 8)
617 self.assertEqual(w2.size(), 8)
619 self.assertIsNone(c.manifest_locator())
620 self.assertTrue(c.modified())
621 self.assertIsNone(keep.get("2e9ec317e197819358fbc43afca7d837+8"))
622 c.save_new("test_create_multiple")
623 self.assertEqual("zzzzz-4zz18-mockcollection0", c.manifest_locator())
624 self.assertFalse(c.modified())
625 self.assertEqual(b"01234567", keep.get("2e9ec317e197819358fbc43afca7d837+8"))
628 class ArvadosFileReaderTestCase(StreamFileReaderTestCase):
629 class MockParent(object):
630 class MockBlockMgr(object):
631 def __init__(self, blocks, nocache):
633 self.nocache = nocache
634 self._keep = ArvadosFileWriterTestCase.MockKeep({})
635 self.prefetch_lookahead = 0
637 def block_prefetch(self, loc):
640 def get_block_contents(self, loc, num_retries=0, cache_only=False):
641 if self.nocache and cache_only:
643 return self.blocks[loc]
645 def __init__(self, blocks, nocache):
647 self.nocache = nocache
648 self.lock = arvados.arvfile.NoopLock()
650 def root_collection(self):
653 def _my_block_manager(self):
654 return ArvadosFileReaderTestCase.MockParent.MockBlockMgr(self.blocks, self.nocache)
657 def make_count_reader(self, nocache=False):
661 for d in [b'01234', b'34567', b'67890']:
662 loc = tutil.str_keep_locator(d)
664 stream.append(Range(loc, n, len(d)))
666 af = ArvadosFile(ArvadosFileReaderTestCase.MockParent(blocks, nocache), "count.txt", stream=stream, segments=[Range(1, 0, 3), Range(6, 3, 3), Range(11, 6, 3)])
667 return ArvadosFileReader(af, mode="rb")
669 def test_read_block_crossing_behavior(self):
670 # read() needs to return all the data requested if possible, even if it
671 # crosses uncached blocks: https://arvados.org/issues/5856
672 sfile = self.make_count_reader(nocache=True)
673 self.assertEqual(b'12345678', sfile.read(8))
675 def test_successive_reads(self):
676 # Override StreamFileReaderTestCase.test_successive_reads
677 sfile = self.make_count_reader(nocache=True)
678 self.assertEqual(b'1234', sfile.read(4))
679 self.assertEqual(b'5678', sfile.read(4))
680 self.assertEqual(b'9', sfile.read(4))
681 self.assertEqual(b'', sfile.read(4))
683 def test_tell_after_block_read(self):
684 # Override StreamFileReaderTestCase.test_tell_after_block_read
685 sfile = self.make_count_reader(nocache=True)
686 self.assertEqual(b'12345678', sfile.read(8))
687 self.assertEqual(8, sfile.tell())
689 def test_prefetch(self):
690 keep = ArvadosFileWriterTestCase.MockKeep({
691 "2e9ec317e197819358fbc43afca7d837+8": b"01234567",
692 "e8dc4081b13434b45189a720b77b6818+8": b"abcdefgh",
694 with Collection(". 2e9ec317e197819358fbc43afca7d837+8 e8dc4081b13434b45189a720b77b6818+8 0:16:count.txt\n", keep_client=keep) as c:
695 r = c.open("count.txt", "rb")
696 self.assertEqual(b"0123", r.read(4))
697 self.assertEqual(["2e9ec317e197819358fbc43afca7d837+8",
698 "e8dc4081b13434b45189a720b77b6818+8"], keep.requests)
700 def test_prefetch_disabled(self):
701 keep = ArvadosFileWriterTestCase.MockKeep({
702 "2e9ec317e197819358fbc43afca7d837+8": b"01234567",
703 "e8dc4081b13434b45189a720b77b6818+8": b"abcdefgh",
705 keep.num_prefetch_threads = 0
706 with Collection(". 2e9ec317e197819358fbc43afca7d837+8 e8dc4081b13434b45189a720b77b6818+8 0:16:count.txt\n", keep_client=keep) as c:
707 r = c.open("count.txt", "rb")
708 self.assertEqual(b"0123", r.read(4))
710 self.assertEqual(["2e9ec317e197819358fbc43afca7d837+8"], keep.requests)
712 def test_prefetch_first_read_only(self):
713 # test behavior that prefetch only happens every 128 reads
714 # check that it doesn't make a prefetch request on the second read
715 keep = ArvadosFileWriterTestCase.MockKeep({
716 "2e9ec317e197819358fbc43afca7d837+8": b"01234567",
717 "e8dc4081b13434b45189a720b77b6818+8": b"abcdefgh",
719 with Collection(". 2e9ec317e197819358fbc43afca7d837+8 e8dc4081b13434b45189a720b77b6818+8 0:16:count.txt\n", keep_client=keep) as c:
720 r = c.open("count.txt", "rb")
721 self.assertEqual(b"0123", r.read(4))
722 self.assertEqual(b"45", r.read(2))
723 self.assertEqual(["2e9ec317e197819358fbc43afca7d837+8",
724 "e8dc4081b13434b45189a720b77b6818+8",
725 "2e9ec317e197819358fbc43afca7d837+8"], keep.requests)
726 self.assertEqual(3, len(keep.requests))
728 def test_prefetch_again(self):
729 # test behavior that prefetch only happens every 128 reads
730 # check that it does make another prefetch request after 128 reads
731 keep = ArvadosFileWriterTestCase.MockKeep({
732 "2e9ec317e197819358fbc43afca7d837+8": b"01234567",
733 "e8dc4081b13434b45189a720b77b6818+8": b"abcdefgh",
735 with Collection(". 2e9ec317e197819358fbc43afca7d837+8 e8dc4081b13434b45189a720b77b6818+8 0:16:count.txt\n", keep_client=keep) as c:
736 r = c.open("count.txt", "rb")
737 for i in range(0, 129):
739 self.assertEqual(b"0123", r.read(4))
740 self.assertEqual(["2e9ec317e197819358fbc43afca7d837+8",
741 "e8dc4081b13434b45189a720b77b6818+8",
742 "2e9ec317e197819358fbc43afca7d837+8",
743 "2e9ec317e197819358fbc43afca7d837+8"], keep.requests[0:4])
744 self.assertEqual(["2e9ec317e197819358fbc43afca7d837+8",
745 "2e9ec317e197819358fbc43afca7d837+8",
746 "2e9ec317e197819358fbc43afca7d837+8",
747 "e8dc4081b13434b45189a720b77b6818+8"], keep.requests[127:131])
748 # gets the 1st block 129 times from keep (cache),
749 # and the 2nd block twice to get 131 requests
750 self.assertEqual(131, len(keep.requests))
752 def test__eq__from_manifest(self):
753 with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt') as c1:
754 with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt') as c2:
755 self.assertTrue(c1["count1.txt"] == c2["count1.txt"])
756 self.assertFalse(c1["count1.txt"] != c2["count1.txt"])
758 def test__eq__from_writes(self):
759 with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt') as c1:
760 with Collection() as c2:
761 f = c2.open("count1.txt", "wb")
762 f.write("0123456789")
764 self.assertTrue(c1["count1.txt"] == c2["count1.txt"])
765 self.assertFalse(c1["count1.txt"] != c2["count1.txt"])
767 def test__ne__(self):
768 with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt') as c1:
769 with Collection() as c2:
770 f = c2.open("count1.txt", "wb")
771 f.write("1234567890")
773 self.assertTrue(c1["count1.txt"] != c2["count1.txt"])
774 self.assertFalse(c1["count1.txt"] == c2["count1.txt"])
777 class ArvadosFileReadTestCase(unittest.TestCase, StreamRetryTestMixin):
778 def reader_for(self, coll_name, **kwargs):
782 for d in self.manifest_for(coll_name).split():
785 segments.append(Range(n, n, k.size))
786 stream.append(Range(d, n, k.size))
791 blockmanager = arvados.arvfile._BlockManager(self.keep_client())
792 blockmanager.prefetch_enabled = False
793 col = Collection(keep_client=self.keep_client(), block_manager=blockmanager)
794 af = ArvadosFile(col, "test",
797 kwargs.setdefault('mode', 'rb')
798 return ArvadosFileReader(af, **kwargs)
800 def read_for_test(self, reader, byte_count, **kwargs):
801 return reader.read(byte_count, **kwargs)
804 class ArvadosFileReadFromTestCase(ArvadosFileReadTestCase):
805 def read_for_test(self, reader, byte_count, **kwargs):
806 return reader.readfrom(0, byte_count, **kwargs)
809 class ArvadosFileReadAllTestCase(ArvadosFileReadTestCase):
810 def read_for_test(self, reader, byte_count, **kwargs):
811 return b''.join(reader.readall(**kwargs))
814 class ArvadosFileReadAllDecompressedTestCase(ArvadosFileReadTestCase):
815 def read_for_test(self, reader, byte_count, **kwargs):
816 return b''.join(reader.readall_decompressed(**kwargs))
819 class ArvadosFileReadlinesTestCase(ArvadosFileReadTestCase):
820 def read_for_test(self, reader, byte_count, **kwargs):
821 return ''.join(reader.readlines(**kwargs)).encode()
824 class ArvadosFileTestCase(unittest.TestCase):
825 def datetime_to_hex(self, dt):
826 return hex(int(time.mktime(dt.timetuple())))[2:]
828 def test_permission_expired(self):
829 base_manifest = ". 781e5e245d69b566979b86e28d23f2c7+10+A715fd31f8111894f717eb1003c1b0216799dd9ec@{} 0:10:count.txt\n"
830 now = datetime.datetime.now()
831 a_week_ago = now - datetime.timedelta(days=7)
832 a_month_ago = now - datetime.timedelta(days=30)
833 a_week_from_now = now + datetime.timedelta(days=7)
834 with Collection(base_manifest.format(self.datetime_to_hex(a_week_from_now))) as c:
835 self.assertFalse(c.find('count.txt').permission_expired())
836 with Collection(base_manifest.format(self.datetime_to_hex(a_week_ago))) as c:
837 f = c.find('count.txt')
838 self.assertTrue(f.permission_expired())
839 self.assertTrue(f.permission_expired(a_week_from_now))
840 self.assertFalse(f.permission_expired(a_month_ago))
843 class BlockManagerTest(unittest.TestCase):
844 def test_bufferblock_append(self):
845 keep = ArvadosFileWriterTestCase.MockKeep({})
846 with arvados.arvfile._BlockManager(keep) as blockmanager:
847 bufferblock = blockmanager.alloc_bufferblock()
848 bufferblock.append("foo")
850 self.assertEqual(bufferblock.size(), 3)
851 self.assertEqual(bufferblock.buffer_view[0:3], b"foo")
852 self.assertEqual(bufferblock.locator(), "acbd18db4cc2f85cedef654fccc4a4d8+3")
854 bufferblock.append("bar")
856 self.assertEqual(bufferblock.size(), 6)
857 self.assertEqual(bufferblock.buffer_view[0:6], b"foobar")
858 self.assertEqual(bufferblock.locator(), "3858f62230ac3c915f300c664312c63f+6")
860 bufferblock.set_state(arvados.arvfile._BufferBlock.PENDING)
861 with self.assertRaises(arvados.errors.AssertionError):
862 bufferblock.append("bar")
864 def test_bufferblock_dup(self):
865 keep = ArvadosFileWriterTestCase.MockKeep({})
866 with arvados.arvfile._BlockManager(keep) as blockmanager:
867 bufferblock = blockmanager.alloc_bufferblock()
868 bufferblock.append("foo")
870 self.assertEqual(bufferblock.size(), 3)
871 self.assertEqual(bufferblock.buffer_view[0:3], b"foo")
872 self.assertEqual(bufferblock.locator(), "acbd18db4cc2f85cedef654fccc4a4d8+3")
873 bufferblock.set_state(arvados.arvfile._BufferBlock.PENDING)
875 bufferblock2 = blockmanager.dup_block(bufferblock, None)
876 self.assertNotEqual(bufferblock.blockid, bufferblock2.blockid)
878 bufferblock2.append("bar")
880 self.assertEqual(bufferblock2.size(), 6)
881 self.assertEqual(bufferblock2.buffer_view[0:6], b"foobar")
882 self.assertEqual(bufferblock2.locator(), "3858f62230ac3c915f300c664312c63f+6")
884 self.assertEqual(bufferblock.size(), 3)
885 self.assertEqual(bufferblock.buffer_view[0:3], b"foo")
886 self.assertEqual(bufferblock.locator(), "acbd18db4cc2f85cedef654fccc4a4d8+3")
888 def test_bufferblock_get(self):
889 keep = ArvadosFileWriterTestCase.MockKeep({
890 "781e5e245d69b566979b86e28d23f2c7+10": b"0123456789",
892 with arvados.arvfile._BlockManager(keep) as blockmanager:
893 bufferblock = blockmanager.alloc_bufferblock()
894 bufferblock.append("foo")
896 self.assertEqual(blockmanager.get_block_contents("781e5e245d69b566979b86e28d23f2c7+10", 1), b"0123456789")
897 self.assertEqual(blockmanager.get_block_contents(bufferblock.blockid, 1), b"foo")
899 def test_bufferblock_commit(self):
900 mockkeep = mock.MagicMock()
901 with arvados.arvfile._BlockManager(mockkeep) as blockmanager:
902 bufferblock = blockmanager.alloc_bufferblock()
903 bufferblock.owner = mock.MagicMock(spec=arvados.arvfile.ArvadosFile)
904 def flush(sync=None):
905 blockmanager.commit_bufferblock(bufferblock, sync)
906 bufferblock.owner.flush.side_effect = flush
907 bufferblock.append("foo")
908 blockmanager.commit_all()
909 self.assertTrue(bufferblock.owner.flush.called)
910 self.assertTrue(mockkeep.put.called)
911 self.assertEqual(bufferblock.state(), arvados.arvfile._BufferBlock.COMMITTED)
912 self.assertIsNone(bufferblock.buffer_view)
914 def test_bufferblock_commit_pending(self):
916 mockkeep = mock.MagicMock()
917 mockkeep.put.side_effect = lambda *args, **kwargs: time.sleep(1)
918 with arvados.arvfile._BlockManager(mockkeep) as blockmanager:
919 bufferblock = blockmanager.alloc_bufferblock()
920 bufferblock.append("foo")
922 blockmanager.commit_bufferblock(bufferblock, False)
923 self.assertEqual(bufferblock.state(), arvados.arvfile._BufferBlock.PENDING)
925 blockmanager.commit_bufferblock(bufferblock, True)
926 self.assertEqual(bufferblock.state(), arvados.arvfile._BufferBlock.COMMITTED)
928 def test_bufferblock_commit_with_error(self):
929 mockkeep = mock.MagicMock()
930 mockkeep.put.side_effect = arvados.errors.KeepWriteError("fail")
931 with arvados.arvfile._BlockManager(mockkeep) as blockmanager:
932 bufferblock = blockmanager.alloc_bufferblock()
933 bufferblock.owner = mock.MagicMock(spec=arvados.arvfile.ArvadosFile)
934 def flush(sync=None):
935 blockmanager.commit_bufferblock(bufferblock, sync)
936 bufferblock.owner.flush.side_effect = flush
937 bufferblock.append("foo")
938 with self.assertRaises(arvados.errors.KeepWriteError) as err:
939 blockmanager.commit_all()
940 self.assertTrue(bufferblock.owner.flush.called)
941 self.assertEqual(str(err.exception), "Error writing some blocks: block acbd18db4cc2f85cedef654fccc4a4d8+3 raised KeepWriteError (fail)")
942 self.assertEqual(bufferblock.state(), arvados.arvfile._BufferBlock.ERROR)