sdk/python/tests/test_arvfile.py

   1 #!/usr/bin/env python
   2
   3 import bz2
   4 import datetime
   5 import gzip
   6 import io
   7 import mock
   8 import os
   9 import unittest
  10 import time
  11
  12 import arvados
  13 from arvados._ranges import Range
  14 from arvados.keep import KeepLocator
  15 from arvados.collection import Collection, CollectionReader
  16 from arvados.arvfile import ArvadosFile, ArvadosFileReader
  17
  18 import arvados_testutil as tutil
  19 from test_stream import StreamFileReaderTestCase, StreamRetryTestMixin
  20
  21 class ArvadosFileWriterTestCase(unittest.TestCase):
  22     class MockKeep(object):
  23         def __init__(self, blocks):
  24             self.blocks = blocks
  25             self.requests = []
  26         def get(self, locator, num_retries=0):
  27             self.requests.append(locator)
  28             return self.blocks.get(locator)
  29         def get_from_cache(self, locator):
  30             self.requests.append(locator)
  31             return self.blocks.get(locator)
  32         def put(self, data, num_retries=None, copies=None):
  33             pdh = tutil.str_keep_locator(data)
  34             self.blocks[pdh] = str(data)
  35             return pdh
  36
  37     class MockApi(object):
  38         def __init__(self, b, r):
  39             self.body = b
  40             self.response = r
  41             self._schema = ArvadosFileWriterTestCase.MockApi.MockSchema()
  42             self._rootDesc = {}
  43         class MockSchema(object):
  44             def __init__(self):
  45                 self.schemas = {'Collection': {'properties': {'replication_desired': {'type':'integer'}}}}
  46         class MockCollections(object):
  47             def __init__(self, b, r):
  48                 self.body = b
  49                 self.response = r
  50             class Execute(object):
  51                 def __init__(self, r):
  52                     self.response = r
  53                 def execute(self, num_retries=None):
  54                     return self.response
  55             def create(self, ensure_unique_name=False, body=None):
  56                 if body != self.body:
  57                     raise Exception("Body %s does not match expectation %s" % (body, self.body))
  58                 return ArvadosFileWriterTestCase.MockApi.MockCollections.Execute(self.response)
  59             def update(self, uuid=None, body=None):
  60                 return ArvadosFileWriterTestCase.MockApi.MockCollections.Execute(self.response)
  61         def collections(self):
  62             return ArvadosFileWriterTestCase.MockApi.MockCollections(self.body, self.response)
  63
  64
  65     def test_truncate(self):
  66         keep = ArvadosFileWriterTestCase.MockKeep({"781e5e245d69b566979b86e28d23f2c7+10": "0123456789"})
  67         api = ArvadosFileWriterTestCase.MockApi({"name":"test_truncate",
  68                                                  "manifest_text":". 781e5e245d69b566979b86e28d23f2c7+10 0:8:count.txt\n",
  69                                                  "replication_desired":None},
  70                                                 {"uuid":"zzzzz-4zz18-mockcollection0",
  71                                                  "manifest_text":". 781e5e245d69b566979b86e28d23f2c7+10 0:8:count.txt\n",
  72                                                  "portable_data_hash":"7fcd0eaac3aad4c31a6a0e756475da92+52"})
  73         with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n',
  74                              api_client=api, keep_client=keep) as c:
  75             writer = c.open("count.txt", "r+")
  76             self.assertEqual(writer.size(), 10)
  77             self.assertEqual("0123456789", writer.read(12))
  78
  79             writer.truncate(8)
  80
  81             # Make sure reading off the end doesn't break
  82             self.assertEqual("", writer.read(12))
  83
  84             self.assertEqual(writer.size(), 8)
  85             writer.seek(0, os.SEEK_SET)
  86             self.assertEqual("01234567", writer.read(12))
  87
  88             self.assertIsNone(c.manifest_locator())
  89             self.assertTrue(c.modified())
  90             c.save_new("test_truncate")
  91             self.assertEqual("zzzzz-4zz18-mockcollection0", c.manifest_locator())
  92             self.assertFalse(c.modified())
  93
  94
  95     def test_truncate2(self):
  96         keep = ArvadosFileWriterTestCase.MockKeep({"781e5e245d69b566979b86e28d23f2c7+10": "0123456789"})
  97         api = ArvadosFileWriterTestCase.MockApi({"name":"test_truncate2",
  98                                                  "manifest_text":". 781e5e245d69b566979b86e28d23f2c7+10 7f614da9329cd3aebf59b91aadc30bf0+67108864 0:12:count.txt\n",
  99                                                  "replication_desired":None},
 100                                                 {"uuid":"zzzzz-4zz18-mockcollection0",
 101                                                  "manifest_text":". 781e5e245d69b566979b86e28d23f2c7+10 7f614da9329cd3aebf59b91aadc30bf0+67108864 0:12:count.txt\n",
 102                                                  "portable_data_hash":"272da898abdf86ddc71994835e3155f8+95"})
 103         with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n',
 104                              api_client=api, keep_client=keep) as c:
 105             writer = c.open("count.txt", "r+")
 106             self.assertEqual(writer.size(), 10)
 107             self.assertEqual("0123456789", writer.read(12))
 108
 109             # extend file size
 110             writer.truncate(12)
 111
 112             self.assertEqual(writer.size(), 12)
 113             writer.seek(0, os.SEEK_SET)
 114             self.assertEqual(b"0123456789\x00\x00", writer.read(12))
 115
 116             self.assertIsNone(c.manifest_locator())
 117             self.assertTrue(c.modified())
 118             c.save_new("test_truncate2")
 119             self.assertEqual("zzzzz-4zz18-mockcollection0", c.manifest_locator())
 120             self.assertFalse(c.modified())
 121
 122     def test_truncate3(self):
 123         keep = ArvadosFileWriterTestCase.MockKeep({"781e5e245d69b566979b86e28d23f2c7+10": "0123456789",
 124                                                    "a925576942e94b2ef57a066101b48876+10": "abcdefghij"})
 125         api = ArvadosFileWriterTestCase.MockApi({"name":"test_truncate",
 126                                                  "manifest_text":". 781e5e245d69b566979b86e28d23f2c7+10 0:8:count.txt\n",
 127                                                  "replication_desired":None},
 128                                                 {"uuid":"zzzzz-4zz18-mockcollection0",
 129                                                  "manifest_text":". 781e5e245d69b566979b86e28d23f2c7+10 0:8:count.txt\n",
 130                                                  "portable_data_hash":"7fcd0eaac3aad4c31a6a0e756475da92+52"})
 131         with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 a925576942e94b2ef57a066101b48876+10 0:20:count.txt\n',
 132                              api_client=api, keep_client=keep) as c:
 133             writer = c.open("count.txt", "r+")
 134             self.assertEqual(writer.size(), 20)
 135             self.assertEqual("0123456789ab", writer.read(12))
 136             self.assertEqual(12, writer.tell())
 137
 138             writer.truncate(8)
 139
 140             # Make sure reading off the end doesn't break
 141             self.assertEqual(12, writer.tell())
 142             self.assertEqual("", writer.read(12))
 143
 144             self.assertEqual(writer.size(), 8)
 145             self.assertEqual(2, writer.seek(-10, os.SEEK_CUR))
 146             self.assertEqual("234567", writer.read(12))
 147
 148             self.assertIsNone(c.manifest_locator())
 149             self.assertTrue(c.modified())
 150             c.save_new("test_truncate")
 151             self.assertEqual("zzzzz-4zz18-mockcollection0", c.manifest_locator())
 152             self.assertFalse(c.modified())
 153
 154
 155
 156     def test_write_to_end(self):
 157         keep = ArvadosFileWriterTestCase.MockKeep({"781e5e245d69b566979b86e28d23f2c7+10": "0123456789"})
 158         api = ArvadosFileWriterTestCase.MockApi({"name":"test_append",
 159                                                  "manifest_text": ". 781e5e245d69b566979b86e28d23f2c7+10 acbd18db4cc2f85cedef654fccc4a4d8+3 0:13:count.txt\n",
 160                                                  "replication_desired":None},
 161                                                 {"uuid":"zzzzz-4zz18-mockcollection0",
 162                                                  "manifest_text": ". 781e5e245d69b566979b86e28d23f2c7+10 acbd18db4cc2f85cedef654fccc4a4d8+3 0:13:count.txt\n",
 163                                                  "portable_data_hash":"c5c3af76565c8efb6a806546bcf073f3+88"})
 164         with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n',
 165                              api_client=api, keep_client=keep) as c:
 166             writer = c.open("count.txt", "r+")
 167             self.assertEqual(writer.size(), 10)
 168
 169             self.assertEqual(5, writer.seek(5, os.SEEK_SET))
 170             self.assertEqual("56789", writer.read(8))
 171
 172             writer.seek(10, os.SEEK_SET)
 173             writer.write("foo")
 174             self.assertEqual(writer.size(), 13)
 175
 176             writer.seek(5, os.SEEK_SET)
 177             self.assertEqual("56789foo", writer.read(8))
 178
 179             self.assertIsNone(c.manifest_locator())
 180             self.assertTrue(c.modified())
 181             self.assertIsNone(keep.get("acbd18db4cc2f85cedef654fccc4a4d8+3"))
 182
 183             c.save_new("test_append")
 184             self.assertEqual("zzzzz-4zz18-mockcollection0", c.manifest_locator())
 185             self.assertFalse(c.modified())
 186             self.assertEqual("foo", keep.get("acbd18db4cc2f85cedef654fccc4a4d8+3"))
 187
 188
 189     def test_append(self):
 190         keep = ArvadosFileWriterTestCase.MockKeep({"781e5e245d69b566979b86e28d23f2c7+10": "0123456789"})
 191         c = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n', keep_client=keep)
 192         writer = c.open("count.txt", "a+")
 193         self.assertEqual(writer.read(20), "0123456789")
 194         writer.seek(0, os.SEEK_SET)
 195
 196         writer.write("hello")
 197         self.assertEqual(writer.read(20), "0123456789hello")
 198         writer.seek(0, os.SEEK_SET)
 199
 200         writer.write("world")
 201         self.assertEqual(writer.read(20), "0123456789helloworld")
 202
 203         self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 fc5e038d38a57032085441e7fe7010b0+10 0:20:count.txt\n", c.portable_manifest_text())
 204
 205     def test_write_at_beginning(self):
 206         keep = ArvadosFileWriterTestCase.MockKeep({"781e5e245d69b566979b86e28d23f2c7+10": "0123456789"})
 207         with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n',
 208                              keep_client=keep) as c:
 209             writer = c.open("count.txt", "r+")
 210             self.assertEqual("0123456789", writer.readfrom(0, 13))
 211             writer.seek(0, os.SEEK_SET)
 212             writer.write("foo")
 213             self.assertEqual(writer.size(), 10)
 214             self.assertEqual("foo3456789", writer.readfrom(0, 13))
 215             self.assertEqual(". acbd18db4cc2f85cedef654fccc4a4d8+3 781e5e245d69b566979b86e28d23f2c7+10 0:3:count.txt 6:7:count.txt\n", c.portable_manifest_text())
 216
 217     def test_write_empty(self):
 218         keep = ArvadosFileWriterTestCase.MockKeep({})
 219         with Collection(keep_client=keep) as c:
 220             writer = c.open("count.txt", "w")
 221             self.assertEqual(writer.size(), 0)
 222             self.assertEqual(". d41d8cd98f00b204e9800998ecf8427e+0 0:0:count.txt\n", c.portable_manifest_text())
 223
 224     def test_save_manifest_text(self):
 225         keep = ArvadosFileWriterTestCase.MockKeep({})
 226         with Collection(keep_client=keep) as c:
 227             writer = c.open("count.txt", "w")
 228             writer.write("0123456789")
 229             self.assertEqual('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n', c.portable_manifest_text())
 230             self.assertNotIn('781e5e245d69b566979b86e28d23f2c7+10', keep.blocks)
 231
 232             self.assertEqual('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n', c.save_new(create_collection_record=False))
 233             self.assertIn('781e5e245d69b566979b86e28d23f2c7+10', keep.blocks)
 234
 235     def test_get_manifest_text_commits(self):
 236          keep = ArvadosFileWriterTestCase.MockKeep({})
 237          with Collection(keep_client=keep) as c:
 238              writer = c.open("count.txt", "w")
 239              writer.write("0123456789")
 240              self.assertEqual('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n', c.portable_manifest_text())
 241              self.assertNotIn('781e5e245d69b566979b86e28d23f2c7+10', keep.blocks)
 242              self.assertEqual('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n', c.manifest_text())
 243              self.assertIn('781e5e245d69b566979b86e28d23f2c7+10', keep.blocks)
 244
 245
 246     def test_write_in_middle(self):
 247         keep = ArvadosFileWriterTestCase.MockKeep({"781e5e245d69b566979b86e28d23f2c7+10": "0123456789"})
 248         with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n',
 249                              keep_client=keep) as c:
 250             writer = c.open("count.txt", "r+")
 251             self.assertEqual("0123456789", writer.readfrom(0, 13))
 252             writer.seek(3, os.SEEK_SET)
 253             writer.write("foo")
 254             self.assertEqual(writer.size(), 10)
 255             self.assertEqual("012foo6789", writer.readfrom(0, 13))
 256             self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 acbd18db4cc2f85cedef654fccc4a4d8+3 0:3:count.txt 10:3:count.txt 6:4:count.txt\n", c.portable_manifest_text())
 257
 258     def test_write_at_end(self):
 259         keep = ArvadosFileWriterTestCase.MockKeep({"781e5e245d69b566979b86e28d23f2c7+10": "0123456789"})
 260         with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n',
 261                              keep_client=keep) as c:
 262             writer = c.open("count.txt", "r+")
 263             self.assertEqual("0123456789", writer.readfrom(0, 13))
 264             writer.seek(7, os.SEEK_SET)
 265             writer.write("foo")
 266             self.assertEqual(writer.size(), 10)
 267             self.assertEqual("0123456foo", writer.readfrom(0, 13))
 268             self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 acbd18db4cc2f85cedef654fccc4a4d8+3 0:7:count.txt 10:3:count.txt\n", c.portable_manifest_text())
 269
 270     def test_write_across_segment_boundary(self):
 271         keep = ArvadosFileWriterTestCase.MockKeep({"781e5e245d69b566979b86e28d23f2c7+10": "0123456789"})
 272         with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt 0:10:count.txt\n',
 273                              keep_client=keep) as c:
 274             writer = c.open("count.txt", "r+")
 275             self.assertEqual("012345678901234", writer.readfrom(0, 15))
 276             writer.seek(7, os.SEEK_SET)
 277             writer.write("foobar")
 278             self.assertEqual(writer.size(), 20)
 279             self.assertEqual("0123456foobar34", writer.readfrom(0, 15))
 280             self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 3858f62230ac3c915f300c664312c63f+6 0:7:count.txt 10:6:count.txt 3:7:count.txt\n", c.portable_manifest_text())
 281
 282     def test_write_across_several_segments(self):
 283         keep = ArvadosFileWriterTestCase.MockKeep({"781e5e245d69b566979b86e28d23f2c7+10": "0123456789"})
 284         with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:4:count.txt 0:4:count.txt 0:4:count.txt',
 285                              keep_client=keep) as c:
 286             writer = c.open("count.txt", "r+")
 287             self.assertEqual("012301230123", writer.readfrom(0, 15))
 288             writer.seek(2, os.SEEK_SET)
 289             writer.write("abcdefg")
 290             self.assertEqual(writer.size(), 12)
 291             self.assertEqual("01abcdefg123", writer.readfrom(0, 15))
 292             self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 7ac66c0f148de9519b8bd264312c4d64+7 0:2:count.txt 10:7:count.txt 1:3:count.txt\n", c.portable_manifest_text())
 293
 294     def test_write_large(self):
 295         keep = ArvadosFileWriterTestCase.MockKeep({})
 296         api = ArvadosFileWriterTestCase.MockApi({"name":"test_write_large",
 297                                                  "manifest_text": ". a5de24f4417cfba9d5825eadc2f4ca49+67108000 598cc1a4ccaef8ab6e4724d87e675d78+32892000 0:100000000:count.txt\n",
 298                                                  "replication_desired":None},
 299                                                 {"uuid":"zzzzz-4zz18-mockcollection0",
 300                                                  "manifest_text": ". a5de24f4417cfba9d5825eadc2f4ca49+67108000 598cc1a4ccaef8ab6e4724d87e675d78+32892000 0:100000000:count.txt\n",
 301                                                  "portable_data_hash":"9132ca8e3f671c76103a38f5bc24328c+108"})
 302         with Collection('. ' + arvados.config.EMPTY_BLOCK_LOCATOR + ' 0:0:count.txt',
 303                              api_client=api, keep_client=keep) as c:
 304             writer = c.open("count.txt", "r+")
 305             text = "0123456789" * 100
 306             for b in xrange(0, 100000):
 307                 writer.write(text)
 308             self.assertEqual(writer.size(), 100000000)
 309
 310             self.assertIsNone(c.manifest_locator())
 311             self.assertTrue(c.modified())
 312             c.save_new("test_write_large")
 313             self.assertEqual("zzzzz-4zz18-mockcollection0", c.manifest_locator())
 314             self.assertFalse(c.modified())
 315
 316
 317     def test_large_write(self):
 318         keep = ArvadosFileWriterTestCase.MockKeep({})
 319         api = ArvadosFileWriterTestCase.MockApi({}, {})
 320         with Collection('. ' + arvados.config.EMPTY_BLOCK_LOCATOR + ' 0:0:count.txt',
 321                              api_client=api, keep_client=keep) as c:
 322             writer = c.open("count.txt", "r+")
 323             self.assertEqual(writer.size(), 0)
 324
 325             text = "0123456789"
 326             writer.write(text)
 327             text = "0123456789" * 9999999
 328             writer.write(text)
 329             self.assertEqual(writer.size(), 100000000)
 330
 331             self.assertEqual(c.manifest_text(), ". 781e5e245d69b566979b86e28d23f2c7+10 48dd23ea1645fd47d789804d71b5bb8e+67108864 77c57dc6ac5a10bb2205caaa73187994+32891126 0:100000000:count.txt\n")
 332
 333     def test_sparse_write(self):
 334         keep = ArvadosFileWriterTestCase.MockKeep({})
 335         api = ArvadosFileWriterTestCase.MockApi({}, {})
 336         with Collection('. ' + arvados.config.EMPTY_BLOCK_LOCATOR + ' 0:0:count.txt',
 337                              api_client=api, keep_client=keep) as c:
 338             writer = c.open("count.txt", "r+")
 339             self.assertEqual(writer.size(), 0)
 340
 341             text = "0123456789"
 342             writer.seek(2)
 343             writer.write(text)
 344             self.assertEqual(writer.size(), 12)
 345             writer.seek(0, os.SEEK_SET)
 346             self.assertEqual(writer.read(), b"\x00\x00"+text)
 347
 348             self.assertEqual(c.manifest_text(), ". 7f614da9329cd3aebf59b91aadc30bf0+67108864 781e5e245d69b566979b86e28d23f2c7+10 0:2:count.txt 67108864:10:count.txt\n")
 349
 350
 351     def test_sparse_write2(self):
 352         keep = ArvadosFileWriterTestCase.MockKeep({})
 353         api = ArvadosFileWriterTestCase.MockApi({}, {})
 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", "r+")
 357             self.assertEqual(writer.size(), 0)
 358
 359             text = "0123456789"
 360             writer.seek((arvados.config.KEEP_BLOCK_SIZE*2) + 2)
 361             writer.write(text)
 362             self.assertEqual(writer.size(), (arvados.config.KEEP_BLOCK_SIZE*2) + 12)
 363             writer.seek(0, os.SEEK_SET)
 364
 365             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")
 366
 367
 368     def test_sparse_write3(self):
 369         keep = ArvadosFileWriterTestCase.MockKeep({})
 370         api = ArvadosFileWriterTestCase.MockApi({}, {})
 371         for r in [[0, 1, 2, 3, 4], [4, 3, 2, 1, 0], [3, 2, 0, 4, 1]]:
 372             with Collection() as c:
 373                 writer = c.open("count.txt", "r+")
 374                 self.assertEqual(writer.size(), 0)
 375
 376                 for i in r:
 377                     w = ("%s" % i) * 10
 378                     writer.seek(i*10)
 379                     writer.write(w)
 380                 writer.seek(0)
 381                 self.assertEqual(writer.read(), "00000000001111111111222222222233333333334444444444")
 382
 383     def test_sparse_write4(self):
 384         keep = ArvadosFileWriterTestCase.MockKeep({})
 385         api = ArvadosFileWriterTestCase.MockApi({}, {})
 386         for r in [[0, 1, 2, 4], [4, 2, 1, 0], [2, 0, 4, 1]]:
 387             with Collection() as c:
 388                 writer = c.open("count.txt", "r+")
 389                 self.assertEqual(writer.size(), 0)
 390
 391                 for i in r:
 392                     w = ("%s" % i) * 10
 393                     writer.seek(i*10)
 394                     writer.write(w)
 395                 writer.seek(0)
 396                 self.assertEqual(writer.read(), b"000000000011111111112222222222\x00\x00\x00\x00\x00\x00\x00\x00\x00\x004444444444")
 397
 398
 399     def test_rewrite_on_empty_file(self):
 400         keep = ArvadosFileWriterTestCase.MockKeep({})
 401         with Collection('. ' + arvados.config.EMPTY_BLOCK_LOCATOR + ' 0:0:count.txt',
 402                              keep_client=keep) as c:
 403             writer = c.open("count.txt", "r+")
 404             for b in xrange(0, 10):
 405                 writer.seek(0, os.SEEK_SET)
 406                 writer.write("0123456789")
 407
 408             self.assertEqual(writer.size(), 10)
 409             self.assertEqual("0123456789", writer.readfrom(0, 20))
 410             self.assertEqual(". 7a08b07e84641703e5f2c836aa59a170+100 90:10:count.txt\n", c.portable_manifest_text())
 411             writer.flush()
 412             self.assertEqual(writer.size(), 10)
 413             self.assertEqual("0123456789", writer.readfrom(0, 20))
 414             self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n", c.portable_manifest_text())
 415
 416     def test_rewrite_append_existing_file(self):
 417         keep = ArvadosFileWriterTestCase.MockKeep({"781e5e245d69b566979b86e28d23f2c7+10": "0123456789"})
 418         with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt',
 419                              keep_client=keep) as c:
 420             writer = c.open("count.txt", "r+")
 421             for b in xrange(0, 10):
 422                 writer.seek(10, os.SEEK_SET)
 423                 writer.write("abcdefghij")
 424
 425             self.assertEqual(writer.size(), 20)
 426             self.assertEqual("0123456789abcdefghij", writer.readfrom(0, 20))
 427             self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 ae5f43bab79cf0be33f025fa97ae7398+100 0:10:count.txt 100:10:count.txt\n", c.portable_manifest_text())
 428
 429             writer.arvadosfile.flush()
 430             self.assertEqual(writer.size(), 20)
 431             self.assertEqual("0123456789abcdefghij", writer.readfrom(0, 20))
 432             self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 a925576942e94b2ef57a066101b48876+10 0:20:count.txt\n", c.portable_manifest_text())
 433
 434     def test_rewrite_over_existing_file(self):
 435         keep = ArvadosFileWriterTestCase.MockKeep({"781e5e245d69b566979b86e28d23f2c7+10": "0123456789"})
 436         with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt',
 437                              keep_client=keep) as c:
 438             writer = c.open("count.txt", "r+")
 439             for b in xrange(0, 10):
 440                 writer.seek(5, os.SEEK_SET)
 441                 writer.write("abcdefghij")
 442
 443             self.assertEqual(writer.size(), 15)
 444             self.assertEqual("01234abcdefghij", writer.readfrom(0, 20))
 445             self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 ae5f43bab79cf0be33f025fa97ae7398+100 0:5:count.txt 100:10:count.txt\n", c.portable_manifest_text())
 446
 447             writer.arvadosfile.flush()
 448
 449             self.assertEqual(writer.size(), 15)
 450             self.assertEqual("01234abcdefghij", writer.readfrom(0, 20))
 451             self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 a925576942e94b2ef57a066101b48876+10 0:5:count.txt 10:10:count.txt\n", c.portable_manifest_text())
 452
 453     def test_write_large_rewrite(self):
 454         keep = ArvadosFileWriterTestCase.MockKeep({})
 455         api = ArvadosFileWriterTestCase.MockApi({"name":"test_write_large",
 456                                                  "manifest_text": ". 3dc0d4bc21f48060bedcb2c91af4f906+32892003 a5de24f4417cfba9d5825eadc2f4ca49+67108000 0:3:count.txt 32892006:67107997:count.txt 3:32892000:count.txt\n",
 457                                                  "replication_desired":None},
 458                                                 {"uuid":"zzzzz-4zz18-mockcollection0",
 459                                                  "manifest_text": ". 3dc0d4bc21f48060bedcb2c91af4f906+32892003 a5de24f4417cfba9d5825eadc2f4ca49+67108000 0:3:count.txt 32892006:67107997:count.txt 3:32892000:count.txt\n",
 460                                                  "portable_data_hash":"217665c6b713e1b78dfba7ebd42344db+156"})
 461         with Collection('. ' + arvados.config.EMPTY_BLOCK_LOCATOR + ' 0:0:count.txt',
 462                              api_client=api, keep_client=keep) as c:
 463             writer = c.open("count.txt", "r+")
 464             text = ''.join(["0123456789" for a in xrange(0, 100)])
 465             for b in xrange(0, 100000):
 466                 writer.write(text)
 467             writer.seek(0, os.SEEK_SET)
 468             writer.write("foo")
 469             self.assertEqual(writer.size(), 100000000)
 470
 471             self.assertIsNone(c.manifest_locator())
 472             self.assertTrue(c.modified())
 473             c.save_new("test_write_large")
 474             self.assertEqual("zzzzz-4zz18-mockcollection0", c.manifest_locator())
 475             self.assertFalse(c.modified())
 476
 477     def test_create(self):
 478         keep = ArvadosFileWriterTestCase.MockKeep({})
 479         api = ArvadosFileWriterTestCase.MockApi({"name":"test_create",
 480                                                  "manifest_text":". 2e9ec317e197819358fbc43afca7d837+8 0:8:count.txt\n",
 481                                                  "replication_desired":None},
 482                                                 {"uuid":"zzzzz-4zz18-mockcollection0",
 483                                                  "manifest_text":". 2e9ec317e197819358fbc43afca7d837+8 0:8:count.txt\n",
 484                                                  "portable_data_hash":"7a461a8c58601798f690f8b368ac4423+51"})
 485         with Collection(api_client=api, keep_client=keep) as c:
 486             writer = c.open("count.txt", "w+")
 487             self.assertEqual(writer.size(), 0)
 488             writer.write("01234567")
 489             self.assertEqual(writer.size(), 8)
 490
 491             self.assertIsNone(c.manifest_locator())
 492             self.assertTrue(c.modified())
 493             self.assertIsNone(keep.get("2e9ec317e197819358fbc43afca7d837+8"))
 494             c.save_new("test_create")
 495             self.assertEqual("zzzzz-4zz18-mockcollection0", c.manifest_locator())
 496             self.assertFalse(c.modified())
 497             self.assertEqual("01234567", keep.get("2e9ec317e197819358fbc43afca7d837+8"))
 498
 499
 500     def test_create_subdir(self):
 501         keep = ArvadosFileWriterTestCase.MockKeep({})
 502         api = ArvadosFileWriterTestCase.MockApi({"name":"test_create",
 503                                                  "manifest_text":"./foo/bar 2e9ec317e197819358fbc43afca7d837+8 0:8:count.txt\n",
 504                                                  "replication_desired":None},
 505                                                 {"uuid":"zzzzz-4zz18-mockcollection0",
 506                                                  "manifest_text":"./foo/bar 2e9ec317e197819358fbc43afca7d837+8 0:8:count.txt\n",
 507                                                  "portable_data_hash":"1b02aaa62528d28a5be41651cbb9d7c7+59"})
 508         with Collection(api_client=api, keep_client=keep) as c:
 509             self.assertIsNone(c.api_response())
 510             writer = c.open("foo/bar/count.txt", "w+")
 511             writer.write("01234567")
 512             self.assertFalse(c.committed())
 513             c.save_new("test_create")
 514             self.assertTrue(c.committed())
 515             self.assertEqual(c.api_response(), api.response)
 516
 517     def test_overwrite(self):
 518         keep = ArvadosFileWriterTestCase.MockKeep({"781e5e245d69b566979b86e28d23f2c7+10": "0123456789"})
 519         api = ArvadosFileWriterTestCase.MockApi({"name":"test_overwrite",
 520                                                  "manifest_text":". 2e9ec317e197819358fbc43afca7d837+8 0:8:count.txt\n",
 521                                                  "replication_desired":None},
 522                                                 {"uuid":"zzzzz-4zz18-mockcollection0",
 523                                                  "manifest_text":". 2e9ec317e197819358fbc43afca7d837+8 0:8:count.txt\n",
 524                                                  "portable_data_hash":"7a461a8c58601798f690f8b368ac4423+51"})
 525         with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n',
 526                              api_client=api, keep_client=keep) as c:
 527             writer = c.open("count.txt", "w+")
 528             self.assertEqual(writer.size(), 0)
 529             writer.write("01234567")
 530             self.assertEqual(writer.size(), 8)
 531
 532             self.assertIsNone(c.manifest_locator())
 533             self.assertTrue(c.modified())
 534             c.save_new("test_overwrite")
 535             self.assertEqual("zzzzz-4zz18-mockcollection0", c.manifest_locator())
 536             self.assertFalse(c.modified())
 537
 538     def test_file_not_found(self):
 539         with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n') as c:
 540             with self.assertRaises(IOError):
 541                 writer = c.open("nocount.txt", "r")
 542
 543     def test_cannot_open_directory(self):
 544         with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n') as c:
 545             with self.assertRaises(IOError):
 546                 writer = c.open(".", "r")
 547
 548     def test_create_multiple(self):
 549         keep = ArvadosFileWriterTestCase.MockKeep({})
 550         api = ArvadosFileWriterTestCase.MockApi({"name":"test_create_multiple",
 551                                                  "manifest_text":". 2e9ec317e197819358fbc43afca7d837+8 e8dc4081b13434b45189a720b77b6818+8 0:8:count1.txt 8:8:count2.txt\n",
 552                                                  "replication_desired":None},
 553                                                 {"uuid":"zzzzz-4zz18-mockcollection0",
 554                                                  "manifest_text":". 2e9ec317e197819358fbc43afca7d837+8 e8dc4081b13434b45189a720b77b6818+8 0:8:count1.txt 8:8:count2.txt\n",
 555                                                  "portable_data_hash":"71e7bb6c00d31fc2b4364199fd97be08+102"})
 556         with Collection(api_client=api, keep_client=keep) as c:
 557             w1 = c.open("count1.txt", "w")
 558             w2 = c.open("count2.txt", "w")
 559             w1.write("01234567")
 560             w2.write("abcdefgh")
 561             self.assertEqual(w1.size(), 8)
 562             self.assertEqual(w2.size(), 8)
 563
 564             self.assertIsNone(c.manifest_locator())
 565             self.assertTrue(c.modified())
 566             self.assertIsNone(keep.get("2e9ec317e197819358fbc43afca7d837+8"))
 567             c.save_new("test_create_multiple")
 568             self.assertEqual("zzzzz-4zz18-mockcollection0", c.manifest_locator())
 569             self.assertFalse(c.modified())
 570             self.assertEqual("01234567", keep.get("2e9ec317e197819358fbc43afca7d837+8"))
 571
 572
 573 class ArvadosFileReaderTestCase(StreamFileReaderTestCase):
 574     class MockParent(object):
 575         class MockBlockMgr(object):
 576             def __init__(self, blocks, nocache):
 577                 self.blocks = blocks
 578                 self.nocache = nocache
 579
 580             def block_prefetch(self, loc):
 581                 pass
 582
 583             def get_block_contents(self, loc, num_retries=0, cache_only=False):
 584                 if self.nocache and cache_only:
 585                     return None
 586                 return self.blocks[loc]
 587
 588         def __init__(self, blocks, nocache):
 589             self.blocks = blocks
 590             self.nocache = nocache
 591             self.lock = arvados.arvfile.NoopLock()
 592
 593         def root_collection(self):
 594             return self
 595
 596         def _my_block_manager(self):
 597             return ArvadosFileReaderTestCase.MockParent.MockBlockMgr(self.blocks, self.nocache)
 598
 599
 600     def make_count_reader(self, nocache=False):
 601         stream = []
 602         n = 0
 603         blocks = {}
 604         for d in ['01234', '34567', '67890']:
 605             loc = tutil.str_keep_locator(d)
 606             blocks[loc] = d
 607             stream.append(Range(loc, n, len(d)))
 608             n += len(d)
 609         af = ArvadosFile(ArvadosFileReaderTestCase.MockParent(blocks, nocache), "count.txt", stream=stream, segments=[Range(1, 0, 3), Range(6, 3, 3), Range(11, 6, 3)])
 610         return ArvadosFileReader(af)
 611
 612     def test_read_block_crossing_behavior(self):
 613         # read() needs to return all the data requested if possible, even if it
 614         # crosses uncached blocks: https://arvados.org/issues/5856
 615         sfile = self.make_count_reader(nocache=True)
 616         self.assertEqual('12345678', sfile.read(8))
 617
 618     def test_successive_reads(self):
 619         # Override StreamFileReaderTestCase.test_successive_reads
 620         sfile = self.make_count_reader(nocache=True)
 621         self.assertEqual('1234', sfile.read(4))
 622         self.assertEqual('5678', sfile.read(4))
 623         self.assertEqual('9', sfile.read(4))
 624         self.assertEqual('', sfile.read(4))
 625
 626     def test_tell_after_block_read(self):
 627         # Override StreamFileReaderTestCase.test_tell_after_block_read
 628         sfile = self.make_count_reader(nocache=True)
 629         self.assertEqual('12345678', sfile.read(8))
 630         self.assertEqual(8, sfile.tell())
 631
 632     def test_prefetch(self):
 633         keep = ArvadosFileWriterTestCase.MockKeep({"2e9ec317e197819358fbc43afca7d837+8": "01234567", "e8dc4081b13434b45189a720b77b6818+8": "abcdefgh"})
 634         with Collection(". 2e9ec317e197819358fbc43afca7d837+8 e8dc4081b13434b45189a720b77b6818+8 0:16:count.txt\n", keep_client=keep) as c:
 635             r = c.open("count.txt", "r")
 636             self.assertEqual("0123", r.read(4))
 637         self.assertIn("2e9ec317e197819358fbc43afca7d837+8", keep.requests)
 638         self.assertIn("e8dc4081b13434b45189a720b77b6818+8", keep.requests)
 639
 640     def test__eq__from_manifest(self):
 641         with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt') as c1:
 642             with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt') as c2:
 643                 self.assertTrue(c1["count1.txt"] == c2["count1.txt"])
 644                 self.assertFalse(c1["count1.txt"] != c2["count1.txt"])
 645
 646     def test__eq__from_writes(self):
 647         with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt') as c1:
 648             with Collection() as c2:
 649                 f = c2.open("count1.txt", "w")
 650                 f.write("0123456789")
 651
 652                 self.assertTrue(c1["count1.txt"] == c2["count1.txt"])
 653                 self.assertFalse(c1["count1.txt"] != c2["count1.txt"])
 654
 655     def test__ne__(self):
 656         with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt') as c1:
 657             with Collection() as c2:
 658                 f = c2.open("count1.txt", "w")
 659                 f.write("1234567890")
 660
 661                 self.assertTrue(c1["count1.txt"] != c2["count1.txt"])
 662                 self.assertFalse(c1["count1.txt"] == c2["count1.txt"])
 663
 664
 665 class ArvadosFileReadTestCase(unittest.TestCase, StreamRetryTestMixin):
 666     def reader_for(self, coll_name, **kwargs):
 667         stream = []
 668         segments = []
 669         n = 0
 670         for d in self.manifest_for(coll_name).split():
 671             try:
 672                 k = KeepLocator(d)
 673                 segments.append(Range(n, n, k.size))
 674                 stream.append(Range(d, n, k.size))
 675                 n += k.size
 676             except ValueError:
 677                 pass
 678
 679         blockmanager = arvados.arvfile._BlockManager(self.keep_client())
 680         blockmanager.prefetch_enabled = False
 681         col = Collection(keep_client=self.keep_client(), block_manager=blockmanager)
 682         af = ArvadosFile(col, "test",
 683                          stream=stream,
 684                          segments=segments)
 685         return ArvadosFileReader(af, **kwargs)
 686
 687     def read_for_test(self, reader, byte_count, **kwargs):
 688         return reader.read(byte_count, **kwargs)
 689
 690
 691 class ArvadosFileReadFromTestCase(ArvadosFileReadTestCase):
 692     def read_for_test(self, reader, byte_count, **kwargs):
 693         return reader.readfrom(0, byte_count, **kwargs)
 694
 695
 696 class ArvadosFileReadAllTestCase(ArvadosFileReadTestCase):
 697     def read_for_test(self, reader, byte_count, **kwargs):
 698         return ''.join(reader.readall(**kwargs))
 699
 700
 701 class ArvadosFileReadAllDecompressedTestCase(ArvadosFileReadTestCase):
 702     def read_for_test(self, reader, byte_count, **kwargs):
 703         return ''.join(reader.readall_decompressed(**kwargs))
 704
 705
 706 class ArvadosFileReadlinesTestCase(ArvadosFileReadTestCase):
 707     def read_for_test(self, reader, byte_count, **kwargs):
 708         return ''.join(reader.readlines(**kwargs))
 709
 710
 711 class ArvadosFileTestCase(unittest.TestCase):
 712     def datetime_to_hex(self, dt):
 713         return hex(int(time.mktime(dt.timetuple())))[2:]
 714
 715     def test_permission_expired(self):
 716         base_manifest = ". 781e5e245d69b566979b86e28d23f2c7+10+A715fd31f8111894f717eb1003c1b0216799dd9ec@{} 0:10:count.txt\n"
 717         now = datetime.datetime.now()
 718         a_week_ago = now - datetime.timedelta(days=7)
 719         a_month_ago = now - datetime.timedelta(days=30)
 720         a_week_from_now = now + datetime.timedelta(days=7)
 721         with Collection(base_manifest.format(self.datetime_to_hex(a_week_from_now))) as c:
 722             self.assertFalse(c.find('count.txt').permission_expired())
 723         with Collection(base_manifest.format(self.datetime_to_hex(a_week_ago))) as c:
 724             f = c.find('count.txt')
 725             self.assertTrue(f.permission_expired())
 726             self.assertTrue(f.permission_expired(a_week_from_now))
 727             self.assertFalse(f.permission_expired(a_month_ago))
 728
 729
 730 class BlockManagerTest(unittest.TestCase):
 731     def test_bufferblock_append(self):
 732         keep = ArvadosFileWriterTestCase.MockKeep({})
 733         with arvados.arvfile._BlockManager(keep) as blockmanager:
 734             bufferblock = blockmanager.alloc_bufferblock()
 735             bufferblock.append("foo")
 736
 737             self.assertEqual(bufferblock.size(), 3)
 738             self.assertEqual(bufferblock.buffer_view[0:3], "foo")
 739             self.assertEqual(bufferblock.locator(), "acbd18db4cc2f85cedef654fccc4a4d8+3")
 740
 741             bufferblock.append("bar")
 742
 743             self.assertEqual(bufferblock.size(), 6)
 744             self.assertEqual(bufferblock.buffer_view[0:6], "foobar")
 745             self.assertEqual(bufferblock.locator(), "3858f62230ac3c915f300c664312c63f+6")
 746
 747             bufferblock.set_state(arvados.arvfile._BufferBlock.PENDING)
 748             with self.assertRaises(arvados.errors.AssertionError):
 749                 bufferblock.append("bar")
 750
 751     def test_bufferblock_dup(self):
 752         keep = ArvadosFileWriterTestCase.MockKeep({})
 753         with arvados.arvfile._BlockManager(keep) as blockmanager:
 754             bufferblock = blockmanager.alloc_bufferblock()
 755             bufferblock.append("foo")
 756
 757             self.assertEqual(bufferblock.size(), 3)
 758             self.assertEqual(bufferblock.buffer_view[0:3], "foo")
 759             self.assertEqual(bufferblock.locator(), "acbd18db4cc2f85cedef654fccc4a4d8+3")
 760             bufferblock.set_state(arvados.arvfile._BufferBlock.PENDING)
 761
 762             bufferblock2 = blockmanager.dup_block(bufferblock, None)
 763             self.assertNotEqual(bufferblock.blockid, bufferblock2.blockid)
 764
 765             bufferblock2.append("bar")
 766
 767             self.assertEqual(bufferblock2.size(), 6)
 768             self.assertEqual(bufferblock2.buffer_view[0:6], "foobar")
 769             self.assertEqual(bufferblock2.locator(), "3858f62230ac3c915f300c664312c63f+6")
 770
 771             self.assertEqual(bufferblock.size(), 3)
 772             self.assertEqual(bufferblock.buffer_view[0:3], "foo")
 773             self.assertEqual(bufferblock.locator(), "acbd18db4cc2f85cedef654fccc4a4d8+3")
 774
 775     def test_bufferblock_get(self):
 776         keep = ArvadosFileWriterTestCase.MockKeep({"781e5e245d69b566979b86e28d23f2c7+10": "0123456789"})
 777         with arvados.arvfile._BlockManager(keep) as blockmanager:
 778             bufferblock = blockmanager.alloc_bufferblock()
 779             bufferblock.append("foo")
 780
 781             self.assertEqual(blockmanager.get_block_contents("781e5e245d69b566979b86e28d23f2c7+10", 1), "0123456789")
 782             self.assertEqual(blockmanager.get_block_contents(bufferblock.blockid, 1), "foo")
 783
 784     def test_bufferblock_commit(self):
 785         mockkeep = mock.MagicMock()
 786         with arvados.arvfile._BlockManager(mockkeep) as blockmanager:
 787             bufferblock = blockmanager.alloc_bufferblock()
 788             bufferblock.owner = mock.MagicMock()
 789             def flush(sync=None):
 790                 blockmanager.commit_bufferblock(bufferblock, sync)
 791             bufferblock.owner.flush.side_effect = flush
 792             bufferblock.append("foo")
 793             blockmanager.commit_all()
 794             self.assertTrue(bufferblock.owner.flush.called)
 795             self.assertTrue(mockkeep.put.called)
 796             self.assertEqual(bufferblock.state(), arvados.arvfile._BufferBlock.COMMITTED)
 797             self.assertIsNone(bufferblock.buffer_view)
 798
 799     def test_bufferblock_commit_pending(self):
 800         # Test for bug #7225
 801         mockkeep = mock.MagicMock()
 802         mockkeep.put.side_effect = lambda x: time.sleep(1)
 803         with arvados.arvfile._BlockManager(mockkeep) as blockmanager:
 804             bufferblock = blockmanager.alloc_bufferblock()
 805             bufferblock.append("foo")
 806
 807             blockmanager.commit_bufferblock(bufferblock, False)
 808             self.assertEqual(bufferblock.state(), arvados.arvfile._BufferBlock.PENDING)
 809
 810             blockmanager.commit_bufferblock(bufferblock, True)
 811             self.assertEqual(bufferblock.state(), arvados.arvfile._BufferBlock.COMMITTED)
 812
 813
 814     def test_bufferblock_commit_with_error(self):
 815         mockkeep = mock.MagicMock()
 816         mockkeep.put.side_effect = arvados.errors.KeepWriteError("fail")
 817         with arvados.arvfile._BlockManager(mockkeep) as blockmanager:
 818             bufferblock = blockmanager.alloc_bufferblock()
 819             bufferblock.owner = mock.MagicMock()
 820             def flush(sync=None):
 821                 blockmanager.commit_bufferblock(bufferblock, sync)
 822             bufferblock.owner.flush.side_effect = flush
 823             bufferblock.append("foo")
 824             with self.assertRaises(arvados.errors.KeepWriteError) as err:
 825                 blockmanager.commit_all()
 826             self.assertTrue(bufferblock.owner.flush.called)
 827             self.assertEqual(str(err.exception), "Error writing some blocks: block acbd18db4cc2f85cedef654fccc4a4d8+3 raised KeepWriteError (fail)")
 828             self.assertEqual(bufferblock.state(), arvados.arvfile._BufferBlock.ERROR)