sdk/python/tests/test_arvfile.py

   1 #!/usr/bin/env python
   2
   3 import bz2
   4 import gzip
   5 import io
   6 import mock
   7 import os
   8 import unittest
   9 import hashlib
  10
  11 import arvados
  12 from arvados import Range, KeepLocator
  13 from arvados.collection import import_manifest, export_manifest, ReadOnlyCollection, WritableCollection
  14 from arvados.arvfile import ArvadosFile, ArvadosFileReader, SYNC_READONLY, SYNC_EXPLICIT
  15
  16 import arvados_testutil as tutil
  17 from test_stream import StreamFileReaderTestCase, StreamRetryTestMixin
  18
  19 class ArvadosFileWriterTestCase(unittest.TestCase):
  20     class MockKeep(object):
  21         def __init__(self, blocks):
  22             self.blocks = blocks
  23             self.requests = []
  24         def get(self, locator, num_retries=0):
  25             self.requests.append(locator)
  26             return self.blocks.get(locator)
  27         def get_from_cache(self, locator):
  28             self.requests.append(locator)
  29             return self.blocks.get(locator)
  30         def put(self, data, num_retries=None):
  31             pdh = "%s+%i" % (hashlib.md5(data).hexdigest(), len(data))
  32             self.blocks[pdh] = str(data)
  33             return pdh
  34
  35     class MockApi(object):
  36         def __init__(self, b, r):
  37             self.b = b
  38             self.r = r
  39         class MockCollections(object):
  40             def __init__(self, b, r):
  41                 self.b = b
  42                 self.r = r
  43             class Execute(object):
  44                 def __init__(self, r):
  45                     self.r = r
  46                 def execute(self, num_retries=None):
  47                     return self.r
  48             def create(self, ensure_unique_name=False, body=None):
  49                 if body != self.b:
  50                     raise Exception("Body %s does not match expectation %s" % (body, self.b))
  51                 return ArvadosFileWriterTestCase.MockApi.MockCollections.Execute(self.r)
  52             def update(self, uuid=None, body=None):
  53                 return ArvadosFileWriterTestCase.MockApi.MockCollections.Execute(self.r)
  54         def collections(self):
  55             return ArvadosFileWriterTestCase.MockApi.MockCollections(self.b, self.r)
  56
  57
  58     def test_truncate(self):
  59         keep = ArvadosFileWriterTestCase.MockKeep({"781e5e245d69b566979b86e28d23f2c7+10": "0123456789"})
  60         api = ArvadosFileWriterTestCase.MockApi({"name":"test_truncate",
  61                                                  "manifest_text":". 781e5e245d69b566979b86e28d23f2c7+10 0:8:count.txt\n"},
  62                                                 {"uuid":"zzzzz-4zz18-mockcollection0"})
  63         with WritableCollection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n',
  64                              api_client=api, keep_client=keep) as c:
  65             writer = c.open("count.txt", "r+")
  66             self.assertEqual(writer.size(), 10)
  67             writer.seek(5)
  68             self.assertEqual("56789", writer.read(8))
  69             writer.truncate(8)
  70             writer.seek(5, os.SEEK_SET)
  71             self.assertEqual("567", writer.read(8))
  72
  73             self.assertEqual(None, c._manifest_locator)
  74             self.assertEqual(True, c.modified())
  75             c.save_new("test_truncate")
  76             self.assertEqual("zzzzz-4zz18-mockcollection0", c._manifest_locator)
  77             self.assertEqual(False, c.modified())
  78
  79     def test_append0(self):
  80         keep = ArvadosFileWriterTestCase.MockKeep({"781e5e245d69b566979b86e28d23f2c7+10": "0123456789"})
  81         api = ArvadosFileWriterTestCase.MockApi({"name":"test_append",
  82                                                  "manifest_text": ". 781e5e245d69b566979b86e28d23f2c7+10 acbd18db4cc2f85cedef654fccc4a4d8+3 0:13:count.txt\n"},
  83                                                 {"uuid":"zzzzz-4zz18-mockcollection0"})
  84         with WritableCollection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n',
  85                              api_client=api, keep_client=keep) as c:
  86             writer = c.open("count.txt", "r+")
  87             self.assertEqual(writer.size(), 10)
  88
  89             writer.seek(5, os.SEEK_SET)
  90             self.assertEqual("56789", writer.read(8))
  91
  92             writer.seek(10, os.SEEK_SET)
  93             writer.write("foo")
  94             self.assertEqual(writer.size(), 13)
  95
  96             writer.seek(5, os.SEEK_SET)
  97             self.assertEqual("56789foo", writer.read(8))
  98
  99             self.assertEqual(None, c._manifest_locator)
 100             self.assertEqual(True, c.modified())
 101             self.assertEqual(None, keep.get("acbd18db4cc2f85cedef654fccc4a4d8+3"))
 102
 103             c.save_new("test_append")
 104             self.assertEqual("zzzzz-4zz18-mockcollection0", c._manifest_locator)
 105             self.assertEqual(False, c.modified())
 106             self.assertEqual("foo", keep.get("acbd18db4cc2f85cedef654fccc4a4d8+3"))
 107
 108
 109     def test_append1(self):
 110         keep = ArvadosFileWriterTestCase.MockKeep({"781e5e245d69b566979b86e28d23f2c7+10": "0123456789"})
 111         c = WritableCollection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n', keep_client=keep)
 112         writer = c.open("count.txt", "a+")
 113         self.assertEqual(writer.read(20), "0123456789")
 114         writer.seek(0, os.SEEK_SET)
 115
 116         writer.write("hello")
 117         self.assertEqual(writer.read(20), "0123456789hello")
 118         writer.seek(0, os.SEEK_SET)
 119
 120         writer.write("world")
 121         self.assertEqual(writer.read(20), "0123456789helloworld")
 122
 123         self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 fc5e038d38a57032085441e7fe7010b0+10 0:20:count.txt\n", export_manifest(c))
 124
 125     def test_write0(self):
 126         keep = ArvadosFileWriterTestCase.MockKeep({"781e5e245d69b566979b86e28d23f2c7+10": "0123456789"})
 127         with WritableCollection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n',
 128                              keep_client=keep) as c:
 129             writer = c.open("count.txt", "r+")
 130             self.assertEqual("0123456789", writer.readfrom(0, 13))
 131             writer.seek(0, os.SEEK_SET)
 132             writer.write("foo")
 133             self.assertEqual(writer.size(), 10)
 134             self.assertEqual("foo3456789", writer.readfrom(0, 13))
 135             self.assertEqual(". acbd18db4cc2f85cedef654fccc4a4d8+3 781e5e245d69b566979b86e28d23f2c7+10 0:3:count.txt 6:7:count.txt\n", export_manifest(c))
 136
 137     def test_write1(self):
 138         keep = ArvadosFileWriterTestCase.MockKeep({"781e5e245d69b566979b86e28d23f2c7+10": "0123456789"})
 139         with WritableCollection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n',
 140                              keep_client=keep) as c:
 141             writer = c.open("count.txt", "r+")
 142             self.assertEqual("0123456789", writer.readfrom(0, 13))
 143             writer.seek(3, os.SEEK_SET)
 144             writer.write("foo")
 145             self.assertEqual(writer.size(), 10)
 146             self.assertEqual("012foo6789", writer.readfrom(0, 13))
 147             self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 acbd18db4cc2f85cedef654fccc4a4d8+3 0:3:count.txt 10:3:count.txt 6:4:count.txt\n", export_manifest(c))
 148
 149     def test_write2(self):
 150         keep = ArvadosFileWriterTestCase.MockKeep({"781e5e245d69b566979b86e28d23f2c7+10": "0123456789"})
 151         with WritableCollection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n',
 152                              keep_client=keep) as c:
 153             writer = c.open("count.txt", "r+")
 154             self.assertEqual("0123456789", writer.readfrom(0, 13))
 155             writer.seek(7, os.SEEK_SET)
 156             writer.write("foo")
 157             self.assertEqual(writer.size(), 10)
 158             self.assertEqual("0123456foo", writer.readfrom(0, 13))
 159             self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 acbd18db4cc2f85cedef654fccc4a4d8+3 0:7:count.txt 10:3:count.txt\n", export_manifest(c))
 160
 161     def test_write3(self):
 162         keep = ArvadosFileWriterTestCase.MockKeep({"781e5e245d69b566979b86e28d23f2c7+10": "0123456789"})
 163         with WritableCollection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt 0:10:count.txt\n',
 164                              keep_client=keep) as c:
 165             writer = c.open("count.txt", "r+")
 166             self.assertEqual("012345678901234", writer.readfrom(0, 15))
 167             writer.seek(7, os.SEEK_SET)
 168             writer.write("foobar")
 169             self.assertEqual(writer.size(), 20)
 170             self.assertEqual("0123456foobar34", writer.readfrom(0, 15))
 171             self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 3858f62230ac3c915f300c664312c63f+6 0:7:count.txt 10:6:count.txt 3:7:count.txt\n", export_manifest(c))
 172
 173     def test_write4(self):
 174         keep = ArvadosFileWriterTestCase.MockKeep({"781e5e245d69b566979b86e28d23f2c7+10": "0123456789"})
 175         with WritableCollection('. 781e5e245d69b566979b86e28d23f2c7+10 0:4:count.txt 0:4:count.txt 0:4:count.txt',
 176                              keep_client=keep) as c:
 177             writer = c.open("count.txt", "r+")
 178             self.assertEqual("012301230123", writer.readfrom(0, 15))
 179             writer.seek(2, os.SEEK_SET)
 180             writer.write("abcdefg")
 181             self.assertEqual(writer.size(), 12)
 182             self.assertEqual("01abcdefg123", writer.readfrom(0, 15))
 183             self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 7ac66c0f148de9519b8bd264312c4d64+7 0:2:count.txt 10:7:count.txt 1:3:count.txt\n", export_manifest(c))
 184
 185     def test_write_large(self):
 186         keep = ArvadosFileWriterTestCase.MockKeep({})
 187         api = ArvadosFileWriterTestCase.MockApi({"name":"test_write_large",
 188                                                  "manifest_text": ". a5de24f4417cfba9d5825eadc2f4ca49+67108000 598cc1a4ccaef8ab6e4724d87e675d78+32892000 0:100000000:count.txt\n"},
 189                                                 {"uuid":"zzzzz-4zz18-mockcollection0"})
 190         with WritableCollection('. ' + arvados.config.EMPTY_BLOCK_LOCATOR + ' 0:0:count.txt',
 191                              api_client=api, keep_client=keep) as c:
 192             writer = c.open("count.txt", "r+")
 193             text = ''.join(["0123456789" for a in xrange(0, 100)])
 194             for b in xrange(0, 100000):
 195                 writer.write(text)
 196             self.assertEqual(writer.size(), 100000000)
 197
 198             self.assertEqual(None, c._manifest_locator)
 199             self.assertEqual(True, c.modified())
 200             c.save_new("test_write_large")
 201             self.assertEqual("zzzzz-4zz18-mockcollection0", c._manifest_locator)
 202             self.assertEqual(False, c.modified())
 203
 204     def test_write_rewrite0(self):
 205         keep = ArvadosFileWriterTestCase.MockKeep({})
 206         with WritableCollection('. ' + arvados.config.EMPTY_BLOCK_LOCATOR + ' 0:0:count.txt',
 207                              keep_client=keep) as c:
 208             writer = c.open("count.txt", "r+")
 209             for b in xrange(0, 10):
 210                 writer.seek(0, os.SEEK_SET)
 211                 writer.write("0123456789")
 212             writer.arvadosfile._repack_writes()
 213             self.assertEqual(writer.size(), 10)
 214             self.assertEqual("0123456789", writer.readfrom(0, 20))
 215             self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n", export_manifest(c))
 216
 217     def test_write_rewrite1(self):
 218         keep = ArvadosFileWriterTestCase.MockKeep({"781e5e245d69b566979b86e28d23f2c7+10": "0123456789"})
 219         with WritableCollection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt',
 220                              keep_client=keep) as c:
 221             writer = c.open("count.txt", "r+")
 222             for b in xrange(0, 10):
 223                 writer.seek(10, os.SEEK_SET)
 224                 writer.write("abcdefghij")
 225             writer.arvadosfile._repack_writes()
 226             self.assertEqual(writer.size(), 20)
 227             self.assertEqual("0123456789abcdefghij", writer.readfrom(0, 20))
 228             self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 a925576942e94b2ef57a066101b48876+10 0:20:count.txt\n", export_manifest(c))
 229
 230     def test_write_rewrite2(self):
 231         keep = ArvadosFileWriterTestCase.MockKeep({"781e5e245d69b566979b86e28d23f2c7+10": "0123456789"})
 232         with WritableCollection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt',
 233                              keep_client=keep) as c:
 234             writer = c.open("count.txt", "r+")
 235             for b in xrange(0, 10):
 236                 writer.seek(5, os.SEEK_SET)
 237                 writer.write("abcdefghij")
 238             writer.arvadosfile._repack_writes()
 239             self.assertEqual(writer.size(), 15)
 240             self.assertEqual("01234abcdefghij", writer.readfrom(0, 20))
 241             self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 a925576942e94b2ef57a066101b48876+10 0:5:count.txt 10:10:count.txt\n", export_manifest(c))
 242
 243     def test_write_large_rewrite0(self):
 244         keep = ArvadosFileWriterTestCase.MockKeep({})
 245         api = ArvadosFileWriterTestCase.MockApi({"name":"test_write_large",
 246                                                  "manifest_text": ". 37400a68af9abdd76ca5bf13e819e42a+32892003 a5de24f4417cfba9d5825eadc2f4ca49+67108000 32892000:3:count.txt 32892006:67107997:count.txt 0:32892000:count.txt\n"},
 247                                                 {"uuid":"zzzzz-4zz18-mockcollection0"})
 248         with WritableCollection('. ' + arvados.config.EMPTY_BLOCK_LOCATOR + ' 0:0:count.txt',
 249                              api_client=api, keep_client=keep) as c:
 250             writer = c.open("count.txt", "r+")
 251             text = ''.join(["0123456789" for a in xrange(0, 100)])
 252             for b in xrange(0, 100000):
 253                 writer.write(text)
 254             writer.seek(0, os.SEEK_SET)
 255             writer.write("foo")
 256             self.assertEqual(writer.size(), 100000000)
 257
 258             self.assertEqual(None, c._manifest_locator)
 259             self.assertEqual(True, c.modified())
 260             c.save_new("test_write_large")
 261             self.assertEqual("zzzzz-4zz18-mockcollection0", c._manifest_locator)
 262             self.assertEqual(False, c.modified())
 263
 264     def test_create(self):
 265         keep = ArvadosFileWriterTestCase.MockKeep({})
 266         api = ArvadosFileWriterTestCase.MockApi({"name":"test_create",
 267                                                  "manifest_text":". 2e9ec317e197819358fbc43afca7d837+8 0:8:count.txt\n"},
 268                                                 {"uuid":"zzzzz-4zz18-mockcollection0"})
 269         with WritableCollection(api_client=api, keep_client=keep) as c:
 270             writer = c.open("count.txt", "w+")
 271             self.assertEqual(writer.size(), 0)
 272             writer.write("01234567")
 273             self.assertEqual(writer.size(), 8)
 274
 275             self.assertEqual(None, c._manifest_locator)
 276             self.assertEqual(True, c.modified())
 277             self.assertEqual(None, keep.get("2e9ec317e197819358fbc43afca7d837+8"))
 278             c.save_new("test_create")
 279             self.assertEqual("zzzzz-4zz18-mockcollection0", c._manifest_locator)
 280             self.assertEqual(False, c.modified())
 281             self.assertEqual("01234567", keep.get("2e9ec317e197819358fbc43afca7d837+8"))
 282
 283
 284     def test_create_subdir(self):
 285         keep = ArvadosFileWriterTestCase.MockKeep({})
 286         api = ArvadosFileWriterTestCase.MockApi({"name":"test_create",
 287                                                  "manifest_text":"./foo/bar 2e9ec317e197819358fbc43afca7d837+8 0:8:count.txt\n"},
 288                                                 {"uuid":"zzzzz-4zz18-mockcollection0"})
 289         with WritableCollection(api_client=api, keep_client=keep) as c:
 290             writer = c.open("foo/bar/count.txt", "w+")
 291             writer.write("01234567")
 292             c.save_new("test_create")
 293
 294     def test_overwrite(self):
 295         keep = ArvadosFileWriterTestCase.MockKeep({"781e5e245d69b566979b86e28d23f2c7+10": "0123456789"})
 296         api = ArvadosFileWriterTestCase.MockApi({"name":"test_overwrite",
 297                                                  "manifest_text":". 2e9ec317e197819358fbc43afca7d837+8 0:8:count.txt\n"},
 298                                                 {"uuid":"zzzzz-4zz18-mockcollection0"})
 299         with WritableCollection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n',
 300                              api_client=api, keep_client=keep) as c:
 301             writer = c.open("count.txt", "w+")
 302             self.assertEqual(writer.size(), 0)
 303             writer.write("01234567")
 304             self.assertEqual(writer.size(), 8)
 305
 306             self.assertEqual(None, c._manifest_locator)
 307             self.assertEqual(True, c.modified())
 308             c.save_new("test_overwrite")
 309             self.assertEqual("zzzzz-4zz18-mockcollection0", c._manifest_locator)
 310             self.assertEqual(False, c.modified())
 311
 312     def test_file_not_found(self):
 313         with WritableCollection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n') as c:
 314             with self.assertRaises(IOError):
 315                 writer = c.open("nocount.txt", "r")
 316
 317     def test_cannot_open_directory(self):
 318         with WritableCollection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n') as c:
 319             with self.assertRaises(IOError):
 320                 writer = c.open(".", "r")
 321
 322     def test_create_multiple(self):
 323         keep = ArvadosFileWriterTestCase.MockKeep({})
 324         api = ArvadosFileWriterTestCase.MockApi({"name":"test_create_multiple",
 325                                                  "manifest_text":". 2e9ec317e197819358fbc43afca7d837+8 e8dc4081b13434b45189a720b77b6818+8 0:8:count1.txt 8:8:count2.txt\n"},
 326                                                 {"uuid":"zzzzz-4zz18-mockcollection0"})
 327         with WritableCollection(api_client=api, keep_client=keep) as c:
 328             w1 = c.open("count1.txt", "w")
 329             w2 = c.open("count2.txt", "w")
 330             w1.write("01234567")
 331             w2.write("abcdefgh")
 332             self.assertEqual(w1.size(), 8)
 333             self.assertEqual(w2.size(), 8)
 334
 335             self.assertEqual(None, c._manifest_locator)
 336             self.assertEqual(True, c.modified())
 337             self.assertEqual(None, keep.get("2e9ec317e197819358fbc43afca7d837+8"))
 338             c.save_new("test_create_multiple")
 339             self.assertEqual("zzzzz-4zz18-mockcollection0", c._manifest_locator)
 340             self.assertEqual(False, c.modified())
 341             self.assertEqual("01234567", keep.get("2e9ec317e197819358fbc43afca7d837+8"))
 342
 343
 344 class ArvadosFileReaderTestCase(StreamFileReaderTestCase):
 345     class MockParent(object):
 346         class MockBlockMgr(object):
 347             def __init__(self, blocks, nocache):
 348                 self.blocks = blocks
 349                 self.nocache = nocache
 350
 351             def block_prefetch(self, loc):
 352                 pass
 353
 354             def get_block_contents(self, loc, num_retries=0, cache_only=False):
 355                 if self.nocache and cache_only:
 356                     return None
 357                 return self.blocks[loc]
 358
 359         def __init__(self, blocks, nocache):
 360             self.blocks = blocks
 361             self.nocache = nocache
 362             self.lock = arvados.arvfile.NoopLock()
 363
 364         def root_collection(self):
 365             return self
 366
 367         def _my_block_manager(self):
 368             return ArvadosFileReaderTestCase.MockParent.MockBlockMgr(self.blocks, self.nocache)
 369
 370         def sync_mode(self):
 371             return SYNC_READONLY
 372
 373
 374     def make_count_reader(self, nocache=False):
 375         stream = []
 376         n = 0
 377         blocks = {}
 378         for d in ['01234', '34567', '67890']:
 379             loc = '{}+{}'.format(hashlib.md5(d).hexdigest(), len(d))
 380             blocks[loc] = d
 381             stream.append(Range(loc, n, len(d)))
 382             n += len(d)
 383         af = ArvadosFile(ArvadosFileReaderTestCase.MockParent(blocks, nocache), stream=stream, segments=[Range(1, 0, 3), Range(6, 3, 3), Range(11, 6, 3)])
 384         return ArvadosFileReader(af, "count.txt")
 385
 386     def test_read_returns_first_block(self):
 387         # read() calls will be aligned on block boundaries - see #3663.
 388         sfile = self.make_count_reader(nocache=True)
 389         self.assertEqual('123', sfile.read(10))
 390
 391     def test_successive_reads(self):
 392         sfile = self.make_count_reader(nocache=True)
 393         for expect in ['123', '456', '789', '']:
 394             self.assertEqual(expect, sfile.read(10))
 395
 396     def test_tell_after_block_read(self):
 397         sfile = self.make_count_reader(nocache=True)
 398         sfile.read(5)
 399         self.assertEqual(3, sfile.tell())
 400
 401     def test_prefetch(self):
 402         keep = ArvadosFileWriterTestCase.MockKeep({"2e9ec317e197819358fbc43afca7d837+8": "01234567", "e8dc4081b13434b45189a720b77b6818+8": "abcdefgh"})
 403         with WritableCollection(". 2e9ec317e197819358fbc43afca7d837+8 e8dc4081b13434b45189a720b77b6818+8 0:16:count.txt\n", keep_client=keep) as c:
 404             r = c.open("count.txt", "r")
 405             self.assertEqual("0123", r.read(4))
 406         self.assertIn("2e9ec317e197819358fbc43afca7d837+8", keep.requests)
 407         self.assertIn("e8dc4081b13434b45189a720b77b6818+8", keep.requests)
 408
 409     def test__eq__1(self):
 410         with WritableCollection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt') as c1:
 411             with WritableCollection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt') as c2:
 412                 self.assertTrue(c1["count1.txt"] == c2["count1.txt"])
 413                 self.assertFalse(c1["count1.txt"] != c2["count1.txt"])
 414
 415     def test__eq__2(self):
 416         with WritableCollection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt') as c1:
 417             with WritableCollection() as c2:
 418                 with c2.open("count1.txt", "w") as f:
 419                     f.write("0123456789")
 420
 421                 self.assertTrue(c1["count1.txt"] == c2["count1.txt"])
 422                 self.assertFalse(c1["count1.txt"] != c2["count1.txt"])
 423
 424     def test__ne__(self):
 425         with WritableCollection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt') as c1:
 426             with WritableCollection() as c2:
 427                 with c2.open("count1.txt", "w") as f:
 428                     f.write("1234567890")
 429
 430                 self.assertTrue(c1["count1.txt"] != c2["count1.txt"])
 431                 self.assertFalse(c1["count1.txt"] == c2["count1.txt"])
 432
 433
 434 class ArvadosFileReadTestCase(unittest.TestCase, StreamRetryTestMixin):
 435     def reader_for(self, coll_name, **kwargs):
 436         stream = []
 437         segments = []
 438         n = 0
 439         for d in self.manifest_for(coll_name).split():
 440             try:
 441                 k = KeepLocator(d)
 442                 segments.append(Range(n, n, k.size))
 443                 stream.append(Range(d, n, k.size))
 444                 n += k.size
 445             except ValueError:
 446                 pass
 447         col = ReadOnlyCollection(keep_client=self.keep_client())
 448         col._my_block_manager().prefetch_enabled = False
 449         af = ArvadosFile(col,
 450                          stream=stream,
 451                          segments=segments)
 452         return ArvadosFileReader(af, "test", **kwargs)
 453
 454     def read_for_test(self, reader, byte_count, **kwargs):
 455         return reader.read(byte_count, **kwargs)
 456
 457
 458 class ArvadosFileReadFromTestCase(ArvadosFileReadTestCase):
 459     def read_for_test(self, reader, byte_count, **kwargs):
 460         return reader.readfrom(0, byte_count, **kwargs)
 461
 462
 463 class ArvadosFileReadAllTestCase(ArvadosFileReadTestCase):
 464     def read_for_test(self, reader, byte_count, **kwargs):
 465         return ''.join(reader.readall(**kwargs))
 466
 467
 468 class ArvadosFileReadAllDecompressedTestCase(ArvadosFileReadTestCase):
 469     def read_for_test(self, reader, byte_count, **kwargs):
 470         return ''.join(reader.readall_decompressed(**kwargs))
 471
 472
 473 class ArvadosFileReadlinesTestCase(ArvadosFileReadTestCase):
 474     def read_for_test(self, reader, byte_count, **kwargs):
 475         return ''.join(reader.readlines(**kwargs))
 476
 477 class BlockManagerTest(unittest.TestCase):
 478     def test_bufferblock_append(self):
 479         keep = ArvadosFileWriterTestCase.MockKeep({})
 480         blockmanager = arvados.arvfile.BlockManager(keep)
 481         bufferblock = blockmanager.alloc_bufferblock()
 482         bufferblock.append("foo")
 483
 484         self.assertEqual(bufferblock.size(), 3)
 485         self.assertEqual(bufferblock.buffer_view[0:3], "foo")
 486         self.assertEqual(bufferblock.locator(), "acbd18db4cc2f85cedef654fccc4a4d8+3")
 487
 488         bufferblock.append("bar")
 489
 490         self.assertEqual(bufferblock.size(), 6)
 491         self.assertEqual(bufferblock.buffer_view[0:6], "foobar")
 492         self.assertEqual(bufferblock.locator(), "3858f62230ac3c915f300c664312c63f+6")
 493
 494         bufferblock.set_state(arvados.arvfile.BufferBlock.PENDING)
 495         with self.assertRaises(arvados.errors.AssertionError):
 496             bufferblock.append("bar")
 497
 498     def test_bufferblock_dup(self):
 499         keep = ArvadosFileWriterTestCase.MockKeep({})
 500         blockmanager = arvados.arvfile.BlockManager(keep)
 501         bufferblock = blockmanager.alloc_bufferblock()
 502         bufferblock.append("foo")
 503
 504         self.assertEqual(bufferblock.size(), 3)
 505         self.assertEqual(bufferblock.buffer_view[0:3], "foo")
 506         self.assertEqual(bufferblock.locator(), "acbd18db4cc2f85cedef654fccc4a4d8+3")
 507         bufferblock.set_state(arvados.arvfile.BufferBlock.PENDING)
 508
 509         bufferblock2 = blockmanager.dup_block(bufferblock, None)
 510         self.assertNotEqual(bufferblock.blockid, bufferblock2.blockid)
 511
 512         bufferblock2.append("bar")
 513
 514         self.assertEqual(bufferblock2.size(), 6)
 515         self.assertEqual(bufferblock2.buffer_view[0:6], "foobar")
 516         self.assertEqual(bufferblock2.locator(), "3858f62230ac3c915f300c664312c63f+6")
 517
 518         self.assertEqual(bufferblock.size(), 3)
 519         self.assertEqual(bufferblock.buffer_view[0:3], "foo")
 520         self.assertEqual(bufferblock.locator(), "acbd18db4cc2f85cedef654fccc4a4d8+3")
 521
 522     def test_bufferblock_get(self):
 523         keep = ArvadosFileWriterTestCase.MockKeep({"781e5e245d69b566979b86e28d23f2c7+10": "0123456789"})
 524         blockmanager = arvados.arvfile.BlockManager(keep)
 525         bufferblock = blockmanager.alloc_bufferblock()
 526         bufferblock.append("foo")
 527
 528         self.assertEqual(blockmanager.get_block_contents("781e5e245d69b566979b86e28d23f2c7+10", 1), "0123456789")
 529         self.assertEqual(blockmanager.get_block_contents(bufferblock.blockid, 1), "foo")
 530
 531     def test_bufferblock_commit(self):
 532         mockkeep = mock.MagicMock()
 533         blockmanager = arvados.arvfile.BlockManager(mockkeep)
 534         bufferblock = blockmanager.alloc_bufferblock()
 535         bufferblock.append("foo")
 536         blockmanager.commit_all()
 537         self.assertTrue(mockkeep.put.called)
 538         self.assertEqual(bufferblock.state(), arvados.arvfile.BufferBlock.COMMITTED)
 539         self.assertIsNone(bufferblock.buffer_view)
 540
 541
 542     def test_bufferblock_commit_with_error(self):
 543         mockkeep = mock.MagicMock()
 544         mockkeep.put.side_effect = arvados.errors.KeepWriteError("fail")
 545         blockmanager = arvados.arvfile.BlockManager(mockkeep)
 546         bufferblock = blockmanager.alloc_bufferblock()
 547         bufferblock.append("foo")
 548         with self.assertRaises(arvados.errors.KeepWriteError) as err:
 549             blockmanager.commit_all()
 550         self.assertEquals(str(err.exception), "Error writing some blocks: acbd18db4cc2f85cedef654fccc4a4d8+3 raised KeepWriteError (fail)")
 551         self.assertEqual(bufferblock.state(), arvados.arvfile.BufferBlock.PENDING)