sdk/python/tests/test_arvfile.py

   1 # Copyright (C) The Arvados Authors. All rights reserved.
   2 #
   3 # SPDX-License-Identifier: Apache-2.0
   4
   5 import datetime
   6 import os
   7 import time
   8 import unittest
   9
  10 from unittest import mock
  11
  12 import arvados
  13 from arvados._ranges import Range
  14 from arvados.keep import KeepLocator
  15 from arvados.collection import Collection
  16 from arvados.arvfile import ArvadosFile, ArvadosFileReader
  17
  18 from . import arvados_testutil as tutil
  19 from .test_stream import StreamFileReaderTestMixin, StreamRetryTestMixin
  20
  21 class ArvadosFileWriterTestCase(unittest.TestCase):
  22     class MockKeep(object):
  23         def __init__(self, blocks):
  24             self.blocks = blocks
  25             self.requests = []
  26             self.num_prefetch_threads = 1
  27         def get(self, locator, num_retries=0, prefetch=False):
  28             self.requests.append(locator)
  29             return self.blocks.get(locator)
  30         def get_from_cache(self, locator):
  31             self.requests.append(locator)
  32             return self.blocks.get(locator)
  33         def put(self, data, num_retries=None, copies=None, classes=[]):
  34             pdh = tutil.str_keep_locator(data)
  35             self.blocks[pdh] = bytes(data)
  36             return pdh
  37         def block_prefetch(self, loc):
  38             self.requests.append(loc)
  39
  40     class MockApi(object):
  41         def __init__(self, b, r):
  42             self.body = b
  43             self.response = r
  44             self._schema = ArvadosFileWriterTestCase.MockApi.MockSchema()
  45             self._rootDesc = {}
  46         class MockSchema(object):
  47             def __init__(self):
  48                 self.schemas = {'Collection': {'properties': {'replication_desired': {'type':'integer'}}}}
  49         class MockCollections(object):
  50             def __init__(self, b, r):
  51                 self.body = b
  52                 self.response = r
  53             class Execute(object):
  54                 def __init__(self, r):
  55                     self.response = r
  56                 def execute(self, num_retries=None):
  57                     return self.response
  58             def create(self, ensure_unique_name=False, body=None):
  59                 if body != self.body:
  60                     raise Exception("Body %s does not match expectation %s" % (body, self.body))
  61                 return ArvadosFileWriterTestCase.MockApi.MockCollections.Execute(self.response)
  62             def update(self, uuid=None, body=None):
  63                 return ArvadosFileWriterTestCase.MockApi.MockCollections.Execute(self.response)
  64         def collections(self):
  65             return ArvadosFileWriterTestCase.MockApi.MockCollections(self.body, self.response)
  66
  67
  68     def test_truncate(self):
  69         keep = ArvadosFileWriterTestCase.MockKeep({
  70             "781e5e245d69b566979b86e28d23f2c7+10": b"0123456789",
  71         })
  72         api = ArvadosFileWriterTestCase.MockApi({
  73             "name": "test_truncate",
  74             "manifest_text": ". 781e5e245d69b566979b86e28d23f2c7+10 0:8:count.txt\n",
  75             "replication_desired": None,
  76         }, {
  77             "uuid": "zzzzz-4zz18-mockcollection0",
  78             "manifest_text": ". 781e5e245d69b566979b86e28d23f2c7+10 0:8:count.txt\n",
  79             "portable_data_hash":"7fcd0eaac3aad4c31a6a0e756475da92+52",
  80         })
  81         with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n',
  82                         api_client=api, keep_client=keep) as c:
  83             writer = c.open("count.txt", "rb+")
  84             self.assertEqual(writer.size(), 10)
  85             self.assertEqual(b"0123456789", writer.read(12))
  86
  87             writer.truncate(8)
  88
  89             # Make sure reading off the end doesn't break
  90             self.assertEqual(b"", writer.read(12))
  91
  92             self.assertEqual(writer.size(), 8)
  93             writer.seek(0, os.SEEK_SET)
  94             self.assertEqual(b"01234567", writer.read(12))
  95
  96             self.assertIsNone(c.manifest_locator())
  97             self.assertTrue(c.modified())
  98             c.save_new("test_truncate")
  99             self.assertEqual("zzzzz-4zz18-mockcollection0", c.manifest_locator())
 100             self.assertFalse(c.modified())
 101
 102
 103     def test_truncate2(self):
 104         keep = ArvadosFileWriterTestCase.MockKeep({
 105             "781e5e245d69b566979b86e28d23f2c7+10": b"0123456789",
 106         })
 107         api = ArvadosFileWriterTestCase.MockApi({
 108             "name": "test_truncate2",
 109             "manifest_text": ". 781e5e245d69b566979b86e28d23f2c7+10 7f614da9329cd3aebf59b91aadc30bf0+67108864 0:12:count.txt\n",
 110             "replication_desired": None,
 111         }, {
 112             "uuid": "zzzzz-4zz18-mockcollection0",
 113             "manifest_text": ". 781e5e245d69b566979b86e28d23f2c7+10 7f614da9329cd3aebf59b91aadc30bf0+67108864 0:12:count.txt\n",
 114             "portable_data_hash": "272da898abdf86ddc71994835e3155f8+95",
 115         })
 116         with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n',
 117                         api_client=api, keep_client=keep) as c:
 118             writer = c.open("count.txt", "rb+")
 119             self.assertEqual(writer.size(), 10)
 120             self.assertEqual(b"0123456789", writer.read(12))
 121
 122             # extend file size
 123             writer.truncate(12)
 124
 125             self.assertEqual(writer.size(), 12)
 126             writer.seek(0, os.SEEK_SET)
 127             self.assertEqual(b"0123456789\x00\x00", writer.read(12))
 128
 129             self.assertIsNone(c.manifest_locator())
 130             self.assertTrue(c.modified())
 131             c.save_new("test_truncate2")
 132             self.assertEqual("zzzzz-4zz18-mockcollection0", c.manifest_locator())
 133             self.assertFalse(c.modified())
 134
 135     def test_truncate3(self):
 136         keep = ArvadosFileWriterTestCase.MockKeep({
 137             "781e5e245d69b566979b86e28d23f2c7+10": b"0123456789",
 138             "a925576942e94b2ef57a066101b48876+10": b"abcdefghij",
 139         })
 140         api = ArvadosFileWriterTestCase.MockApi({
 141             "name": "test_truncate",
 142             "manifest_text": ". 781e5e245d69b566979b86e28d23f2c7+10 0:8:count.txt\n",
 143             "replication_desired": None,
 144         }, {
 145             "uuid": "zzzzz-4zz18-mockcollection0",
 146             "manifest_text": ". 781e5e245d69b566979b86e28d23f2c7+10 0:8:count.txt\n",
 147             "portable_data_hash": "7fcd0eaac3aad4c31a6a0e756475da92+52",
 148         })
 149         with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 a925576942e94b2ef57a066101b48876+10 0:20:count.txt\n',
 150                         api_client=api, keep_client=keep) as c:
 151             writer = c.open("count.txt", "rb+")
 152             self.assertEqual(writer.size(), 20)
 153             self.assertEqual(b"0123456789ab", writer.read(12))
 154             self.assertEqual(12, writer.tell())
 155
 156             writer.truncate(8)
 157
 158             # Make sure reading off the end doesn't break
 159             self.assertEqual(12, writer.tell())
 160             self.assertEqual(b"", writer.read(12))
 161
 162             self.assertEqual(writer.size(), 8)
 163             self.assertEqual(2, writer.seek(-10, os.SEEK_CUR))
 164             self.assertEqual(b"234567", writer.read(12))
 165
 166             self.assertIsNone(c.manifest_locator())
 167             self.assertTrue(c.modified())
 168             c.save_new("test_truncate")
 169             self.assertEqual("zzzzz-4zz18-mockcollection0", c.manifest_locator())
 170             self.assertFalse(c.modified())
 171
 172     def test_write_to_end(self):
 173         keep = ArvadosFileWriterTestCase.MockKeep({
 174             "781e5e245d69b566979b86e28d23f2c7+10": b"0123456789",
 175         })
 176         api = ArvadosFileWriterTestCase.MockApi({
 177             "name": "test_append",
 178             "manifest_text": ". 781e5e245d69b566979b86e28d23f2c7+10 acbd18db4cc2f85cedef654fccc4a4d8+3 0:13:count.txt\n",
 179             "replication_desired": None,
 180         }, {
 181             "uuid": "zzzzz-4zz18-mockcollection0",
 182             "manifest_text": ". 781e5e245d69b566979b86e28d23f2c7+10 acbd18db4cc2f85cedef654fccc4a4d8+3 0:13:count.txt\n",
 183             "portable_data_hash": "c5c3af76565c8efb6a806546bcf073f3+88",
 184         })
 185         with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n',
 186                              api_client=api, keep_client=keep) as c:
 187             writer = c.open("count.txt", "rb+")
 188             self.assertEqual(writer.size(), 10)
 189
 190             self.assertEqual(5, writer.seek(5, os.SEEK_SET))
 191             self.assertEqual(b"56789", writer.read(8))
 192
 193             writer.seek(10, os.SEEK_SET)
 194             writer.write("foo")
 195             self.assertEqual(writer.size(), 13)
 196
 197             writer.seek(5, os.SEEK_SET)
 198             self.assertEqual(b"56789foo", writer.read(8))
 199
 200             self.assertIsNone(c.manifest_locator())
 201             self.assertTrue(c.modified())
 202             self.assertIsNone(keep.get("acbd18db4cc2f85cedef654fccc4a4d8+3"))
 203
 204             c.save_new("test_append")
 205             self.assertEqual("zzzzz-4zz18-mockcollection0", c.manifest_locator())
 206             self.assertFalse(c.modified())
 207             self.assertEqual(b"foo", keep.get("acbd18db4cc2f85cedef654fccc4a4d8+3"))
 208
 209
 210     def test_append(self):
 211         keep = ArvadosFileWriterTestCase.MockKeep({
 212             "781e5e245d69b566979b86e28d23f2c7+10": b"0123456789",
 213         })
 214         for (mode, convert) in (
 215                 ('a+', lambda data: data.decode(encoding='utf-8')),
 216                 ('at+', lambda data: data.decode(encoding='utf-8')),
 217                 ('ab+', lambda data: data)):
 218             c = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n', keep_client=keep)
 219             writer = c.open("count.txt", mode)
 220             self.assertEqual(writer.read(20), convert(b"0123456789"))
 221
 222             writer.seek(0, os.SEEK_SET)
 223             writer.write(convert(b"hello"))
 224             self.assertEqual(writer.read(), convert(b""))
 225             if 'b' in mode:
 226                 writer.seek(-5, os.SEEK_CUR)
 227                 self.assertEqual(writer.read(3), convert(b"hel"))
 228                 self.assertEqual(writer.read(), convert(b"lo"))
 229             else:
 230                 with self.assertRaises(IOError):
 231                     writer.seek(-5, os.SEEK_CUR)
 232                 with self.assertRaises(IOError):
 233                     writer.seek(-3, os.SEEK_END)
 234             writer.seek(0, os.SEEK_SET)
 235             writer.read(7)
 236             self.assertEqual(7, writer.tell())
 237             self.assertEqual(7, writer.seek(7, os.SEEK_SET))
 238
 239             writer.seek(0, os.SEEK_SET)
 240             self.assertEqual(writer.read(), convert(b"0123456789hello"))
 241
 242             writer.seek(0)
 243             writer.write(convert(b"world"))
 244             self.assertEqual(writer.read(), convert(b""))
 245             writer.seek(0)
 246             self.assertEqual(writer.read(), convert(b"0123456789helloworld"))
 247
 248             self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 fc5e038d38a57032085441e7fe7010b0+10 0:20:count.txt\n", c.portable_manifest_text())
 249
 250     def test_write_at_beginning(self):
 251         keep = ArvadosFileWriterTestCase.MockKeep({
 252             "781e5e245d69b566979b86e28d23f2c7+10": b"0123456789",
 253         })
 254         with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n',
 255                              keep_client=keep) as c:
 256             writer = c.open("count.txt", "rb+")
 257             self.assertEqual(b"0123456789", writer.readfrom(0, 13))
 258             writer.seek(0, os.SEEK_SET)
 259             writer.write("foo")
 260             self.assertEqual(writer.size(), 10)
 261             self.assertEqual(b"foo3456789", writer.readfrom(0, 13))
 262             self.assertEqual(". acbd18db4cc2f85cedef654fccc4a4d8+3 781e5e245d69b566979b86e28d23f2c7+10 0:3:count.txt 6:7:count.txt\n", c.portable_manifest_text())
 263
 264     def test_write_empty(self):
 265         keep = ArvadosFileWriterTestCase.MockKeep({})
 266         with Collection(keep_client=keep) as c:
 267             writer = c.open("count.txt", "wb")
 268             self.assertEqual(writer.size(), 0)
 269             self.assertEqual(". d41d8cd98f00b204e9800998ecf8427e+0 0:0:count.txt\n", c.portable_manifest_text())
 270
 271     def test_save_manifest_text(self):
 272         keep = ArvadosFileWriterTestCase.MockKeep({})
 273         with Collection(keep_client=keep) as c:
 274             writer = c.open("count.txt", "wb")
 275             writer.write(b"0123456789")
 276             self.assertEqual('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n', c.portable_manifest_text())
 277             self.assertNotIn('781e5e245d69b566979b86e28d23f2c7+10', keep.blocks)
 278
 279             self.assertEqual('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n', c.save_new(create_collection_record=False))
 280             self.assertIn('781e5e245d69b566979b86e28d23f2c7+10', keep.blocks)
 281
 282     def test_get_manifest_text_commits(self):
 283          keep = ArvadosFileWriterTestCase.MockKeep({})
 284          with Collection(keep_client=keep) as c:
 285              writer = c.open("count.txt", "wb")
 286              writer.write("0123456789")
 287              self.assertEqual('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n', c.portable_manifest_text())
 288              self.assertNotIn('781e5e245d69b566979b86e28d23f2c7+10', keep.blocks)
 289              self.assertEqual('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n', c.manifest_text())
 290              self.assertIn('781e5e245d69b566979b86e28d23f2c7+10', keep.blocks)
 291
 292
 293     def test_write_in_middle(self):
 294         keep = ArvadosFileWriterTestCase.MockKeep({"781e5e245d69b566979b86e28d23f2c7+10": b"0123456789"})
 295         with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n',
 296                              keep_client=keep) as c:
 297             writer = c.open("count.txt", "rb+")
 298             self.assertEqual(b"0123456789", writer.readfrom(0, 13))
 299             writer.seek(3, os.SEEK_SET)
 300             writer.write("foo")
 301             self.assertEqual(writer.size(), 10)
 302             self.assertEqual(b"012foo6789", writer.readfrom(0, 13))
 303             self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 acbd18db4cc2f85cedef654fccc4a4d8+3 0:3:count.txt 10:3:count.txt 6:4:count.txt\n", c.portable_manifest_text())
 304
 305     def test_write_at_end(self):
 306         keep = ArvadosFileWriterTestCase.MockKeep({"781e5e245d69b566979b86e28d23f2c7+10": b"0123456789"})
 307         with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n',
 308                              keep_client=keep) as c:
 309             writer = c.open("count.txt", "rb+")
 310             self.assertEqual(b"0123456789", writer.readfrom(0, 13))
 311             writer.seek(7, os.SEEK_SET)
 312             writer.write("foo")
 313             self.assertEqual(writer.size(), 10)
 314             self.assertEqual(b"0123456foo", writer.readfrom(0, 13))
 315             self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 acbd18db4cc2f85cedef654fccc4a4d8+3 0:7:count.txt 10:3:count.txt\n", c.portable_manifest_text())
 316
 317     def test_write_across_segment_boundary(self):
 318         keep = ArvadosFileWriterTestCase.MockKeep({"781e5e245d69b566979b86e28d23f2c7+10": b"0123456789"})
 319         with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt 0:10:count.txt\n',
 320                              keep_client=keep) as c:
 321             writer = c.open("count.txt", "rb+")
 322             self.assertEqual(b"012345678901234", writer.readfrom(0, 15))
 323             writer.seek(7, os.SEEK_SET)
 324             writer.write("foobar")
 325             self.assertEqual(writer.size(), 20)
 326             self.assertEqual(b"0123456foobar34", writer.readfrom(0, 15))
 327             self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 3858f62230ac3c915f300c664312c63f+6 0:7:count.txt 10:6:count.txt 3:7:count.txt\n", c.portable_manifest_text())
 328
 329     def test_write_across_several_segments(self):
 330         keep = ArvadosFileWriterTestCase.MockKeep({"781e5e245d69b566979b86e28d23f2c7+10": b"0123456789"})
 331         with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:4:count.txt 0:4:count.txt 0:4:count.txt',
 332                              keep_client=keep) as c:
 333             writer = c.open("count.txt", "rb+")
 334             self.assertEqual(b"012301230123", writer.readfrom(0, 15))
 335             writer.seek(2, os.SEEK_SET)
 336             writer.write("abcdefg")
 337             self.assertEqual(writer.size(), 12)
 338             self.assertEqual(b"01abcdefg123", writer.readfrom(0, 15))
 339             self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 7ac66c0f148de9519b8bd264312c4d64+7 0:2:count.txt 10:7:count.txt 1:3:count.txt\n", c.portable_manifest_text())
 340
 341     def test_write_large(self):
 342         keep = ArvadosFileWriterTestCase.MockKeep({})
 343         api = ArvadosFileWriterTestCase.MockApi({"name":"test_write_large",
 344                                                  "manifest_text": ". a5de24f4417cfba9d5825eadc2f4ca49+67108000 598cc1a4ccaef8ab6e4724d87e675d78+32892000 0:100000000:count.txt\n",
 345                                                  "replication_desired":None},
 346                                                 {"uuid":"zzzzz-4zz18-mockcollection0",
 347                                                  "manifest_text": ". a5de24f4417cfba9d5825eadc2f4ca49+67108000 598cc1a4ccaef8ab6e4724d87e675d78+32892000 0:100000000:count.txt\n",
 348                                                  "portable_data_hash":"9132ca8e3f671c76103a38f5bc24328c+108"})
 349         with Collection('. ' + arvados.config.EMPTY_BLOCK_LOCATOR + ' 0:0:count.txt',
 350                              api_client=api, keep_client=keep) as c:
 351             writer = c.open("count.txt", "rb+")
 352             text = "0123456789" * 100
 353             for b in range(0, 100000):
 354                 writer.write(text)
 355             self.assertEqual(writer.size(), 100000000)
 356
 357             self.assertIsNone(c.manifest_locator())
 358             self.assertTrue(c.modified())
 359             c.save_new("test_write_large")
 360             self.assertEqual("zzzzz-4zz18-mockcollection0", c.manifest_locator())
 361             self.assertFalse(c.modified())
 362
 363     def test_large_write(self):
 364         keep = ArvadosFileWriterTestCase.MockKeep({})
 365         api = ArvadosFileWriterTestCase.MockApi({}, {})
 366         with Collection('. ' + arvados.config.EMPTY_BLOCK_LOCATOR + ' 0:0:count.txt',
 367                              api_client=api, keep_client=keep) as c:
 368             writer = c.open("count.txt", "rb+")
 369             self.assertEqual(writer.size(), 0)
 370
 371             text = "0123456789"
 372             writer.write(text)
 373             text = "0123456789" * 9999999
 374             writer.write(text)
 375             self.assertEqual(writer.size(), 100000000)
 376
 377             self.assertEqual(c.manifest_text(), ". 781e5e245d69b566979b86e28d23f2c7+10 48dd23ea1645fd47d789804d71b5bb8e+67108864 77c57dc6ac5a10bb2205caaa73187994+32891126 0:100000000:count.txt\n")
 378
 379     def test_sparse_write(self):
 380         keep = ArvadosFileWriterTestCase.MockKeep({})
 381         api = ArvadosFileWriterTestCase.MockApi({}, {})
 382         with Collection('. ' + arvados.config.EMPTY_BLOCK_LOCATOR + ' 0:0:count.txt',
 383                              api_client=api, keep_client=keep) as c:
 384             writer = c.open("count.txt", "rb+")
 385             self.assertEqual(writer.size(), 0)
 386
 387             text = b"0123456789"
 388             writer.seek(2)
 389             writer.write(text)
 390             self.assertEqual(writer.size(), 12)
 391             writer.seek(0, os.SEEK_SET)
 392             self.assertEqual(writer.read(), b"\x00\x00"+text)
 393
 394             self.assertEqual(c.manifest_text(), ". 7f614da9329cd3aebf59b91aadc30bf0+67108864 781e5e245d69b566979b86e28d23f2c7+10 0:2:count.txt 67108864:10:count.txt\n")
 395
 396     def test_sparse_write2(self):
 397         keep = ArvadosFileWriterTestCase.MockKeep({})
 398         api = ArvadosFileWriterTestCase.MockApi({}, {})
 399         with Collection('. ' + arvados.config.EMPTY_BLOCK_LOCATOR + ' 0:0:count.txt',
 400                              api_client=api, keep_client=keep) as c:
 401             writer = c.open("count.txt", "rb+")
 402             self.assertEqual(writer.size(), 0)
 403
 404             text = "0123456789"
 405             writer.seek((arvados.config.KEEP_BLOCK_SIZE*2) + 2)
 406             writer.write(text)
 407             self.assertEqual(writer.size(), (arvados.config.KEEP_BLOCK_SIZE*2) + 12)
 408             writer.seek(0, os.SEEK_SET)
 409
 410             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")
 411
 412     def test_sparse_write3(self):
 413         keep = ArvadosFileWriterTestCase.MockKeep({})
 414         api = ArvadosFileWriterTestCase.MockApi({}, {})
 415         for r in [[0, 1, 2, 3, 4], [4, 3, 2, 1, 0], [3, 2, 0, 4, 1]]:
 416             with Collection(api_client=api, keep_client=keep) as c:
 417                 writer = c.open("count.txt", "rb+")
 418                 self.assertEqual(writer.size(), 0)
 419
 420                 for i in r:
 421                     w = ("%s" % i) * 10
 422                     writer.seek(i*10)
 423                     writer.write(w.encode())
 424                 writer.seek(0)
 425                 self.assertEqual(writer.read(), b"00000000001111111111222222222233333333334444444444")
 426
 427     def test_sparse_write4(self):
 428         keep = ArvadosFileWriterTestCase.MockKeep({})
 429         api = ArvadosFileWriterTestCase.MockApi({}, {})
 430         for r in [[0, 1, 2, 4], [4, 2, 1, 0], [2, 0, 4, 1]]:
 431             with Collection(api_client=api, keep_client=keep) as c:
 432                 writer = c.open("count.txt", "rb+")
 433                 self.assertEqual(writer.size(), 0)
 434
 435                 for i in r:
 436                     w = ("%s" % i) * 10
 437                     writer.seek(i*10)
 438                     writer.write(w.encode())
 439                 writer.seek(0)
 440                 self.assertEqual(writer.read(), b"000000000011111111112222222222\x00\x00\x00\x00\x00\x00\x00\x00\x00\x004444444444")
 441
 442     def test_rewrite_on_empty_file(self):
 443         keep = ArvadosFileWriterTestCase.MockKeep({})
 444         with Collection('. ' + arvados.config.EMPTY_BLOCK_LOCATOR + ' 0:0:count.txt',
 445                              keep_client=keep) as c:
 446             writer = c.open("count.txt", "rb+")
 447             for b in range(0, 10):
 448                 writer.seek(0, os.SEEK_SET)
 449                 writer.write("0123456789")
 450
 451             self.assertEqual(writer.size(), 10)
 452             self.assertEqual(b"0123456789", writer.readfrom(0, 20))
 453             self.assertEqual(". 7a08b07e84641703e5f2c836aa59a170+100 90:10:count.txt\n", c.portable_manifest_text())
 454             writer.flush()
 455             self.assertEqual(writer.size(), 10)
 456             self.assertEqual(b"0123456789", writer.readfrom(0, 20))
 457             self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n", c.portable_manifest_text())
 458
 459     def test_rewrite_append_existing_file(self):
 460         keep = ArvadosFileWriterTestCase.MockKeep({
 461             "781e5e245d69b566979b86e28d23f2c7+10": b"0123456789",
 462         })
 463         with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt',
 464                              keep_client=keep) as c:
 465             writer = c.open("count.txt", "rb+")
 466             for b in range(0, 10):
 467                 writer.seek(10, os.SEEK_SET)
 468                 writer.write("abcdefghij")
 469
 470             self.assertEqual(writer.size(), 20)
 471             self.assertEqual(b"0123456789abcdefghij", writer.readfrom(0, 20))
 472             self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 ae5f43bab79cf0be33f025fa97ae7398+100 0:10:count.txt 100:10:count.txt\n", c.portable_manifest_text())
 473
 474             writer.arvadosfile.flush()
 475             self.assertEqual(writer.size(), 20)
 476             self.assertEqual(b"0123456789abcdefghij", writer.readfrom(0, 20))
 477             self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 a925576942e94b2ef57a066101b48876+10 0:20:count.txt\n", c.portable_manifest_text())
 478
 479     def test_rewrite_over_existing_file(self):
 480         keep = ArvadosFileWriterTestCase.MockKeep({
 481             "781e5e245d69b566979b86e28d23f2c7+10": b"0123456789",
 482         })
 483         with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt',
 484                              keep_client=keep) as c:
 485             writer = c.open("count.txt", "rb+")
 486             for b in range(0, 10):
 487                 writer.seek(5, os.SEEK_SET)
 488                 writer.write("abcdefghij")
 489
 490             self.assertEqual(writer.size(), 15)
 491             self.assertEqual(b"01234abcdefghij", writer.readfrom(0, 20))
 492             self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 ae5f43bab79cf0be33f025fa97ae7398+100 0:5:count.txt 100:10:count.txt\n", c.portable_manifest_text())
 493
 494             writer.arvadosfile.flush()
 495
 496             self.assertEqual(writer.size(), 15)
 497             self.assertEqual(b"01234abcdefghij", writer.readfrom(0, 20))
 498             self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 a925576942e94b2ef57a066101b48876+10 0:5:count.txt 10:10:count.txt\n", c.portable_manifest_text())
 499
 500     def test_write_large_rewrite(self):
 501         keep = ArvadosFileWriterTestCase.MockKeep({})
 502         api = ArvadosFileWriterTestCase.MockApi({"name":"test_write_large",
 503                                                  "manifest_text": ". 3dc0d4bc21f48060bedcb2c91af4f906+32892003 a5de24f4417cfba9d5825eadc2f4ca49+67108000 0:3:count.txt 32892006:67107997:count.txt 3:32892000:count.txt\n",
 504                                                  "replication_desired":None},
 505                                                 {"uuid":"zzzzz-4zz18-mockcollection0",
 506                                                  "manifest_text": ". 3dc0d4bc21f48060bedcb2c91af4f906+32892003 a5de24f4417cfba9d5825eadc2f4ca49+67108000 0:3:count.txt 32892006:67107997:count.txt 3:32892000:count.txt\n",
 507                                                  "portable_data_hash":"217665c6b713e1b78dfba7ebd42344db+156"})
 508         with Collection('. ' + arvados.config.EMPTY_BLOCK_LOCATOR + ' 0:0:count.txt',
 509                              api_client=api, keep_client=keep) as c:
 510             writer = c.open("count.txt", "rb+")
 511             text = b''.join([b"0123456789" for a in range(0, 100)])
 512             for b in range(0, 100000):
 513                 writer.write(text)
 514             writer.seek(0, os.SEEK_SET)
 515             writer.write("foo")
 516             self.assertEqual(writer.size(), 100000000)
 517
 518             self.assertIsNone(c.manifest_locator())
 519             self.assertTrue(c.modified())
 520             c.save_new("test_write_large")
 521             self.assertEqual("zzzzz-4zz18-mockcollection0", c.manifest_locator())
 522             self.assertFalse(c.modified())
 523
 524     def test_create(self):
 525         keep = ArvadosFileWriterTestCase.MockKeep({})
 526         api = ArvadosFileWriterTestCase.MockApi({
 527             "name":"test_create",
 528             "manifest_text":". 2e9ec317e197819358fbc43afca7d837+8 0:8:count.txt\n",
 529             "replication_desired":None,
 530         }, {
 531             "uuid":"zzzzz-4zz18-mockcollection0",
 532             "manifest_text":". 2e9ec317e197819358fbc43afca7d837+8 0:8:count.txt\n",
 533             "portable_data_hash":"7a461a8c58601798f690f8b368ac4423+51",
 534         })
 535         with Collection(api_client=api, keep_client=keep) as c:
 536             writer = c.open("count.txt", "wb+")
 537             self.assertEqual(writer.size(), 0)
 538             writer.write("01234567")
 539             self.assertEqual(writer.size(), 8)
 540
 541             self.assertIsNone(c.manifest_locator())
 542             self.assertTrue(c.modified())
 543             self.assertIsNone(keep.get("2e9ec317e197819358fbc43afca7d837+8"))
 544             c.save_new("test_create")
 545             self.assertEqual("zzzzz-4zz18-mockcollection0", c.manifest_locator())
 546             self.assertFalse(c.modified())
 547             self.assertEqual(b"01234567", keep.get("2e9ec317e197819358fbc43afca7d837+8"))
 548
 549
 550     def test_create_subdir(self):
 551         keep = ArvadosFileWriterTestCase.MockKeep({})
 552         api = ArvadosFileWriterTestCase.MockApi({"name":"test_create",
 553                                                  "manifest_text":"./foo/bar 2e9ec317e197819358fbc43afca7d837+8 0:8:count.txt\n",
 554                                                  "replication_desired":None},
 555                                                 {"uuid":"zzzzz-4zz18-mockcollection0",
 556                                                  "manifest_text":"./foo/bar 2e9ec317e197819358fbc43afca7d837+8 0:8:count.txt\n",
 557                                                  "portable_data_hash":"1b02aaa62528d28a5be41651cbb9d7c7+59"})
 558         with Collection(api_client=api, keep_client=keep) as c:
 559             self.assertIsNone(c.api_response())
 560             writer = c.open("foo/bar/count.txt", "wb+")
 561             writer.write("01234567")
 562             self.assertFalse(c.committed())
 563             c.save_new("test_create")
 564             self.assertTrue(c.committed())
 565             self.assertEqual(c.api_response(), api.response)
 566
 567     def test_overwrite(self):
 568         keep = ArvadosFileWriterTestCase.MockKeep({"781e5e245d69b566979b86e28d23f2c7+10": "0123456789"})
 569         api = ArvadosFileWriterTestCase.MockApi({"name":"test_overwrite",
 570                                                  "manifest_text":". 2e9ec317e197819358fbc43afca7d837+8 0:8:count.txt\n",
 571                                                  "replication_desired":None},
 572                                                 {"uuid":"zzzzz-4zz18-mockcollection0",
 573                                                  "manifest_text":". 2e9ec317e197819358fbc43afca7d837+8 0:8:count.txt\n",
 574                                                  "portable_data_hash":"7a461a8c58601798f690f8b368ac4423+51"})
 575         with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n',
 576                              api_client=api, keep_client=keep) as c:
 577             writer = c.open("count.txt", "wb+")
 578             self.assertEqual(writer.size(), 0)
 579             writer.write("01234567")
 580             self.assertEqual(writer.size(), 8)
 581
 582             self.assertIsNone(c.manifest_locator())
 583             self.assertTrue(c.modified())
 584             c.save_new("test_overwrite")
 585             self.assertEqual("zzzzz-4zz18-mockcollection0", c.manifest_locator())
 586             self.assertFalse(c.modified())
 587
 588     def test_file_not_found(self):
 589         with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n') as c:
 590             with self.assertRaises(IOError):
 591                 writer = c.open("nocount.txt", "rb")
 592
 593     def test_cannot_open_directory(self):
 594         with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n') as c:
 595             with self.assertRaises(IOError):
 596                 writer = c.open(".", "rb")
 597
 598     def test_create_multiple(self):
 599         keep = ArvadosFileWriterTestCase.MockKeep({})
 600         api = ArvadosFileWriterTestCase.MockApi({"name":"test_create_multiple",
 601                                                  "manifest_text":". 2e9ec317e197819358fbc43afca7d837+8 e8dc4081b13434b45189a720b77b6818+8 0:8:count1.txt 8:8:count2.txt\n",
 602                                                  "replication_desired":None},
 603                                                 {"uuid":"zzzzz-4zz18-mockcollection0",
 604                                                  "manifest_text":". 2e9ec317e197819358fbc43afca7d837+8 e8dc4081b13434b45189a720b77b6818+8 0:8:count1.txt 8:8:count2.txt\n",
 605                                                  "portable_data_hash":"71e7bb6c00d31fc2b4364199fd97be08+102"})
 606         with Collection(api_client=api, keep_client=keep) as c:
 607             w1 = c.open("count1.txt", "wb")
 608             w2 = c.open("count2.txt", "wb")
 609             w1.write("01234567")
 610             w2.write("abcdefgh")
 611             self.assertEqual(w1.size(), 8)
 612             self.assertEqual(w2.size(), 8)
 613
 614             self.assertIsNone(c.manifest_locator())
 615             self.assertTrue(c.modified())
 616             self.assertIsNone(keep.get("2e9ec317e197819358fbc43afca7d837+8"))
 617             c.save_new("test_create_multiple")
 618             self.assertEqual("zzzzz-4zz18-mockcollection0", c.manifest_locator())
 619             self.assertFalse(c.modified())
 620             self.assertEqual(b"01234567", keep.get("2e9ec317e197819358fbc43afca7d837+8"))
 621
 622
 623 class ArvadosFileReaderTestCase(unittest.TestCase, StreamFileReaderTestMixin):
 624     class MockParent(object):
 625         class MockBlockMgr(object):
 626             def __init__(self, blocks, nocache):
 627                 self.blocks = blocks
 628                 self.nocache = nocache
 629                 self._keep = ArvadosFileWriterTestCase.MockKeep({})
 630                 self.prefetch_lookahead = 0
 631
 632             def block_prefetch(self, loc):
 633                 pass
 634
 635             def get_block_contents(self, loc, num_retries=0, cache_only=False):
 636                 if self.nocache and cache_only:
 637                     return None
 638                 return self.blocks[loc]
 639
 640         def __init__(self, blocks, nocache):
 641             self.blocks = blocks
 642             self.nocache = nocache
 643             self.lock = arvados.arvfile.NoopLock()
 644
 645         def root_collection(self):
 646             return self
 647
 648         def _my_block_manager(self):
 649             return ArvadosFileReaderTestCase.MockParent.MockBlockMgr(self.blocks, self.nocache)
 650
 651
 652     def make_file_reader(self, name='emptyfile', data='', nocache=False):
 653         loc = tutil.str_keep_locator(data)
 654         af = ArvadosFile(ArvadosFileReaderTestCase.MockParent({loc: data}, nocache=nocache), name, stream=[Range(loc, 0, len(data))], segments=[Range(0, len(data), len(data))])
 655         return ArvadosFileReader(af, mode='rb')
 656
 657     def make_count_reader(self, nocache=False):
 658         stream = []
 659         n = 0
 660         blocks = {}
 661         for d in [b'01234', b'34567', b'67890']:
 662             loc = tutil.str_keep_locator(d)
 663             blocks[loc] = d
 664             stream.append(Range(loc, n, len(d)))
 665             n += len(d)
 666         af = ArvadosFile(ArvadosFileReaderTestCase.MockParent(blocks, nocache=nocache), "count.txt", stream=stream, segments=[Range(1, 0, 3), Range(6, 3, 3), Range(11, 6, 3)])
 667         return ArvadosFileReader(af, mode="rb")
 668
 669     def make_newlines_reader(self, nocache=False):
 670         stream = []
 671         segments = []
 672         n = 0
 673         blocks = {}
 674         for d in [b'one\ntwo\n\nth', b'ree\nfour\n\n']:
 675             loc = tutil.str_keep_locator(d)
 676             blocks[loc] = d
 677             stream.append(Range(loc, n, len(d)))
 678             segments.append(Range(n, len(d), n+len(d)))
 679             n += len(d)
 680         af = ArvadosFile(ArvadosFileReaderTestCase.MockParent(blocks, nocache=nocache), "count.txt", stream=stream, segments=segments)
 681         return ArvadosFileReader(af, mode="rb")
 682
 683     def test_read_block_crossing_behavior(self):
 684         # read() needs to return all the data requested if possible, even if it
 685         # crosses uncached blocks: https://arvados.org/issues/5856
 686         sfile = self.make_count_reader(nocache=True)
 687         self.assertEqual(b'12345678', sfile.read(8))
 688
 689     def test_tell_after_block_read(self):
 690         sfile = self.make_count_reader(nocache=True)
 691         self.assertEqual(b'12345678', sfile.read(8))
 692         self.assertEqual(8, sfile.tell())
 693
 694     def test_prefetch(self):
 695         keep = ArvadosFileWriterTestCase.MockKeep({
 696             "2e9ec317e197819358fbc43afca7d837+8": b"01234567",
 697             "e8dc4081b13434b45189a720b77b6818+8": b"abcdefgh",
 698         })
 699         with Collection(". 2e9ec317e197819358fbc43afca7d837+8 e8dc4081b13434b45189a720b77b6818+8 0:16:count.txt\n", keep_client=keep) as c:
 700             r = c.open("count.txt", "rb")
 701             self.assertEqual(b"0123", r.read(4))
 702         self.assertEqual(["2e9ec317e197819358fbc43afca7d837+8",
 703                           "e8dc4081b13434b45189a720b77b6818+8"], keep.requests)
 704
 705     def test_prefetch_disabled(self):
 706         keep = ArvadosFileWriterTestCase.MockKeep({
 707             "2e9ec317e197819358fbc43afca7d837+8": b"01234567",
 708             "e8dc4081b13434b45189a720b77b6818+8": b"abcdefgh",
 709         })
 710         keep.num_prefetch_threads = 0
 711         with Collection(". 2e9ec317e197819358fbc43afca7d837+8 e8dc4081b13434b45189a720b77b6818+8 0:16:count.txt\n", keep_client=keep) as c:
 712             r = c.open("count.txt", "rb")
 713             self.assertEqual(b"0123", r.read(4))
 714
 715         self.assertEqual(["2e9ec317e197819358fbc43afca7d837+8"], keep.requests)
 716
 717     def test_prefetch_first_read_only(self):
 718         # test behavior that prefetch only happens every 128 reads
 719         # check that it doesn't make a prefetch request on the second read
 720         keep = ArvadosFileWriterTestCase.MockKeep({
 721             "2e9ec317e197819358fbc43afca7d837+8": b"01234567",
 722             "e8dc4081b13434b45189a720b77b6818+8": b"abcdefgh",
 723         })
 724         with Collection(". 2e9ec317e197819358fbc43afca7d837+8 e8dc4081b13434b45189a720b77b6818+8 0:16:count.txt\n", keep_client=keep) as c:
 725             r = c.open("count.txt", "rb")
 726             self.assertEqual(b"0123", r.read(4))
 727             self.assertEqual(b"45", r.read(2))
 728         self.assertEqual(["2e9ec317e197819358fbc43afca7d837+8",
 729                           "e8dc4081b13434b45189a720b77b6818+8",
 730                           "2e9ec317e197819358fbc43afca7d837+8"], keep.requests)
 731         self.assertEqual(3, len(keep.requests))
 732
 733     def test_prefetch_again(self):
 734         # test behavior that prefetch only happens every 128 reads
 735         # check that it does make another prefetch request after 128 reads
 736         keep = ArvadosFileWriterTestCase.MockKeep({
 737             "2e9ec317e197819358fbc43afca7d837+8": b"01234567",
 738             "e8dc4081b13434b45189a720b77b6818+8": b"abcdefgh",
 739         })
 740         with Collection(". 2e9ec317e197819358fbc43afca7d837+8 e8dc4081b13434b45189a720b77b6818+8 0:16:count.txt\n", keep_client=keep) as c:
 741             r = c.open("count.txt", "rb")
 742             for i in range(0, 129):
 743                 r.seek(0)
 744                 self.assertEqual(b"0123", r.read(4))
 745         self.assertEqual(["2e9ec317e197819358fbc43afca7d837+8",
 746                           "e8dc4081b13434b45189a720b77b6818+8",
 747                           "2e9ec317e197819358fbc43afca7d837+8",
 748                           "2e9ec317e197819358fbc43afca7d837+8"], keep.requests[0:4])
 749         self.assertEqual(["2e9ec317e197819358fbc43afca7d837+8",
 750                           "2e9ec317e197819358fbc43afca7d837+8",
 751                           "2e9ec317e197819358fbc43afca7d837+8",
 752                           "e8dc4081b13434b45189a720b77b6818+8"], keep.requests[127:131])
 753         # gets the 1st block 129 times from keep (cache),
 754         # and the 2nd block twice to get 131 requests
 755         self.assertEqual(131, len(keep.requests))
 756
 757     def test__eq__from_manifest(self):
 758         with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt') as c1:
 759             with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt') as c2:
 760                 self.assertTrue(c1["count1.txt"] == c2["count1.txt"])
 761                 self.assertFalse(c1["count1.txt"] != c2["count1.txt"])
 762
 763     def test__eq__from_writes(self):
 764         with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt') as c1:
 765             with Collection() as c2:
 766                 f = c2.open("count1.txt", "wb")
 767                 f.write("0123456789")
 768
 769                 self.assertTrue(c1["count1.txt"] == c2["count1.txt"])
 770                 self.assertFalse(c1["count1.txt"] != c2["count1.txt"])
 771
 772     def test__ne__(self):
 773         with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt') as c1:
 774             with Collection() as c2:
 775                 f = c2.open("count1.txt", "wb")
 776                 f.write("1234567890")
 777
 778                 self.assertTrue(c1["count1.txt"] != c2["count1.txt"])
 779                 self.assertFalse(c1["count1.txt"] == c2["count1.txt"])
 780
 781
 782 class ArvadosFileReadTestCase(unittest.TestCase, StreamRetryTestMixin):
 783     def reader_for(self, coll_name, **kwargs):
 784         stream = []
 785         segments = []
 786         n = 0
 787         for d in self.manifest_for(coll_name).split():
 788             try:
 789                 k = KeepLocator(d)
 790                 segments.append(Range(n, n, k.size))
 791                 stream.append(Range(d, n, k.size))
 792                 n += k.size
 793             except ValueError:
 794                 pass
 795
 796         blockmanager = arvados.arvfile._BlockManager(self.keep_client())
 797         blockmanager.prefetch_enabled = False
 798         col = Collection(keep_client=self.keep_client(), block_manager=blockmanager)
 799         af = ArvadosFile(col, "test",
 800                          stream=stream,
 801                          segments=segments)
 802         kwargs.setdefault('mode', 'rb')
 803         return ArvadosFileReader(af, **kwargs)
 804
 805     def read_for_test(self, reader, byte_count, **kwargs):
 806         return reader.read(byte_count, **kwargs)
 807
 808
 809 class ArvadosFileReadFromTestCase(ArvadosFileReadTestCase):
 810     def read_for_test(self, reader, byte_count, **kwargs):
 811         return reader.readfrom(0, byte_count, **kwargs)
 812
 813
 814 class ArvadosFileReadAllTestCase(ArvadosFileReadTestCase):
 815     def read_for_test(self, reader, byte_count, **kwargs):
 816         return b''.join(reader.readall(**kwargs))
 817
 818
 819 class ArvadosFileReadAllDecompressedTestCase(ArvadosFileReadTestCase):
 820     def read_for_test(self, reader, byte_count, **kwargs):
 821         return b''.join(reader.readall_decompressed(**kwargs))
 822
 823
 824 class ArvadosFileReadlinesTestCase(ArvadosFileReadTestCase):
 825     def read_for_test(self, reader, byte_count, **kwargs):
 826         return ''.join(reader.readlines(**kwargs)).encode()
 827
 828
 829 class ArvadosFileTestCase(unittest.TestCase):
 830     def datetime_to_hex(self, dt):
 831         return hex(int(time.mktime(dt.timetuple())))[2:]
 832
 833     def test_permission_expired(self):
 834         base_manifest = ". 781e5e245d69b566979b86e28d23f2c7+10+A715fd31f8111894f717eb1003c1b0216799dd9ec@{} 0:10:count.txt\n"
 835         now = datetime.datetime.now()
 836         a_week_ago = now - datetime.timedelta(days=7)
 837         a_month_ago = now - datetime.timedelta(days=30)
 838         a_week_from_now = now + datetime.timedelta(days=7)
 839         with Collection(base_manifest.format(self.datetime_to_hex(a_week_from_now))) as c:
 840             self.assertFalse(c.find('count.txt').permission_expired())
 841         with Collection(base_manifest.format(self.datetime_to_hex(a_week_ago))) as c:
 842             f = c.find('count.txt')
 843             self.assertTrue(f.permission_expired())
 844             self.assertTrue(f.permission_expired(a_week_from_now))
 845             self.assertFalse(f.permission_expired(a_month_ago))
 846
 847
 848 class BlockManagerTest(unittest.TestCase):
 849     def test_bufferblock_append(self):
 850         keep = ArvadosFileWriterTestCase.MockKeep({})
 851         with arvados.arvfile._BlockManager(keep) as blockmanager:
 852             bufferblock = blockmanager.alloc_bufferblock()
 853             bufferblock.append("foo")
 854
 855             self.assertEqual(bufferblock.size(), 3)
 856             self.assertEqual(bufferblock.buffer_view[0:3], b"foo")
 857             self.assertEqual(bufferblock.locator(), "acbd18db4cc2f85cedef654fccc4a4d8+3")
 858
 859             bufferblock.append("bar")
 860
 861             self.assertEqual(bufferblock.size(), 6)
 862             self.assertEqual(bufferblock.buffer_view[0:6], b"foobar")
 863             self.assertEqual(bufferblock.locator(), "3858f62230ac3c915f300c664312c63f+6")
 864
 865             bufferblock.set_state(arvados.arvfile._BufferBlock.PENDING)
 866             with self.assertRaises(arvados.errors.AssertionError):
 867                 bufferblock.append("bar")
 868
 869     def test_bufferblock_dup(self):
 870         keep = ArvadosFileWriterTestCase.MockKeep({})
 871         with arvados.arvfile._BlockManager(keep) as blockmanager:
 872             bufferblock = blockmanager.alloc_bufferblock()
 873             bufferblock.append("foo")
 874
 875             self.assertEqual(bufferblock.size(), 3)
 876             self.assertEqual(bufferblock.buffer_view[0:3], b"foo")
 877             self.assertEqual(bufferblock.locator(), "acbd18db4cc2f85cedef654fccc4a4d8+3")
 878             bufferblock.set_state(arvados.arvfile._BufferBlock.PENDING)
 879
 880             bufferblock2 = blockmanager.dup_block(bufferblock, None)
 881             self.assertNotEqual(bufferblock.blockid, bufferblock2.blockid)
 882
 883             bufferblock2.append("bar")
 884
 885             self.assertEqual(bufferblock2.size(), 6)
 886             self.assertEqual(bufferblock2.buffer_view[0:6], b"foobar")
 887             self.assertEqual(bufferblock2.locator(), "3858f62230ac3c915f300c664312c63f+6")
 888
 889             self.assertEqual(bufferblock.size(), 3)
 890             self.assertEqual(bufferblock.buffer_view[0:3], b"foo")
 891             self.assertEqual(bufferblock.locator(), "acbd18db4cc2f85cedef654fccc4a4d8+3")
 892
 893     def test_bufferblock_get(self):
 894         keep = ArvadosFileWriterTestCase.MockKeep({
 895             "781e5e245d69b566979b86e28d23f2c7+10": b"0123456789",
 896         })
 897         with arvados.arvfile._BlockManager(keep) as blockmanager:
 898             bufferblock = blockmanager.alloc_bufferblock()
 899             bufferblock.append("foo")
 900
 901             self.assertEqual(blockmanager.get_block_contents("781e5e245d69b566979b86e28d23f2c7+10", 1), b"0123456789")
 902             self.assertEqual(blockmanager.get_block_contents(bufferblock.blockid, 1), b"foo")
 903
 904     def test_bufferblock_commit(self):
 905         mockkeep = mock.MagicMock()
 906         with arvados.arvfile._BlockManager(mockkeep) as blockmanager:
 907             bufferblock = blockmanager.alloc_bufferblock()
 908             bufferblock.owner = mock.MagicMock(spec=arvados.arvfile.ArvadosFile)
 909             def flush(sync=None):
 910                 blockmanager.commit_bufferblock(bufferblock, sync)
 911             bufferblock.owner.flush.side_effect = flush
 912             bufferblock.append("foo")
 913             blockmanager.commit_all()
 914             self.assertTrue(bufferblock.owner.flush.called)
 915             self.assertTrue(mockkeep.put.called)
 916             self.assertEqual(bufferblock.state(), arvados.arvfile._BufferBlock.COMMITTED)
 917             self.assertIsNone(bufferblock.buffer_view)
 918
 919     def test_bufferblock_commit_pending(self):
 920         # Test for bug #7225
 921         mockkeep = mock.MagicMock()
 922         mockkeep.put.side_effect = lambda *args, **kwargs: time.sleep(1)
 923         with arvados.arvfile._BlockManager(mockkeep) as blockmanager:
 924             bufferblock = blockmanager.alloc_bufferblock()
 925             bufferblock.append("foo")
 926
 927             blockmanager.commit_bufferblock(bufferblock, False)
 928             self.assertEqual(bufferblock.state(), arvados.arvfile._BufferBlock.PENDING)
 929
 930             blockmanager.commit_bufferblock(bufferblock, True)
 931             self.assertEqual(bufferblock.state(), arvados.arvfile._BufferBlock.COMMITTED)
 932
 933     def test_bufferblock_commit_with_error(self):
 934         mockkeep = mock.MagicMock()
 935         mockkeep.put.side_effect = arvados.errors.KeepWriteError("fail")
 936         with arvados.arvfile._BlockManager(mockkeep) as blockmanager:
 937             bufferblock = blockmanager.alloc_bufferblock()
 938             bufferblock.owner = mock.MagicMock(spec=arvados.arvfile.ArvadosFile)
 939             def flush(sync=None):
 940                 blockmanager.commit_bufferblock(bufferblock, sync)
 941             bufferblock.owner.flush.side_effect = flush
 942             bufferblock.append("foo")
 943             with self.assertRaises(arvados.errors.KeepWriteError) as err:
 944                 blockmanager.commit_all()
 945             self.assertTrue(bufferblock.owner.flush.called)
 946             self.assertEqual(str(err.exception), "Error writing some blocks: block acbd18db4cc2f85cedef654fccc4a4d8+3 raised KeepWriteError (fail)")
 947             self.assertEqual(bufferblock.state(), arvados.arvfile._BufferBlock.ERROR)