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._ranges import Range
  13 from arvados.keep import KeepLocator
  14 from arvados.collection import Collection, CollectionReader
  15 from arvados.arvfile import ArvadosFile, ArvadosFileReader
  16
  17 import arvados_testutil as tutil
  18 from test_stream import StreamFileReaderTestCase, StreamRetryTestMixin
  19
  20 class ArvadosFileWriterTestCase(unittest.TestCase):
  21     class MockKeep(object):
  22         def __init__(self, blocks):
  23             self.blocks = blocks
  24             self.requests = []
  25         def get(self, locator, num_retries=0):
  26             self.requests.append(locator)
  27             return self.blocks.get(locator)
  28         def get_from_cache(self, locator):
  29             self.requests.append(locator)
  30             return self.blocks.get(locator)
  31         def put(self, data, num_retries=None):
  32             pdh = "%s+%i" % (hashlib.md5(data).hexdigest(), len(data))
  33             self.blocks[pdh] = str(data)
  34             return pdh
  35
  36     class MockApi(object):
  37         def __init__(self, b, r):
  38             self.body = b
  39             self.response = r
  40         class MockCollections(object):
  41             def __init__(self, b, r):
  42                 self.body = b
  43                 self.response = r
  44             class Execute(object):
  45                 def __init__(self, r):
  46                     self.response = r
  47                 def execute(self, num_retries=None):
  48                     return self.response
  49             def create(self, ensure_unique_name=False, body=None):
  50                 if body != self.body:
  51                     raise Exception("Body %s does not match expectation %s" % (body, self.body))
  52                 return ArvadosFileWriterTestCase.MockApi.MockCollections.Execute(self.response)
  53             def update(self, uuid=None, body=None):
  54                 return ArvadosFileWriterTestCase.MockApi.MockCollections.Execute(self.response)
  55         def collections(self):
  56             return ArvadosFileWriterTestCase.MockApi.MockCollections(self.body, self.response)
  57
  58
  59     def test_truncate(self):
  60         keep = ArvadosFileWriterTestCase.MockKeep({"781e5e245d69b566979b86e28d23f2c7+10": "0123456789"})
  61         api = ArvadosFileWriterTestCase.MockApi({"name":"test_truncate",
  62                                                  "manifest_text":". 781e5e245d69b566979b86e28d23f2c7+10 0:8:count.txt\n"},
  63                                                 {"uuid":"zzzzz-4zz18-mockcollection0",
  64                                                  "manifest_text":". 781e5e245d69b566979b86e28d23f2c7+10 0:8:count.txt\n"})
  65         with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n',
  66                              api_client=api, keep_client=keep) as c:
  67             writer = c.open("count.txt", "r+")
  68             self.assertEqual(writer.size(), 10)
  69             self.assertEqual("0123456789", writer.read(12))
  70
  71             writer.truncate(8)
  72
  73             # Make sure reading off the end doesn't break
  74             self.assertEqual("", writer.read(12))
  75
  76             self.assertEqual(writer.size(), 8)
  77             writer.seek(0, os.SEEK_SET)
  78             self.assertEqual("01234567", writer.read(12))
  79
  80             self.assertIsNone(c.manifest_locator())
  81             self.assertTrue(c.modified())
  82             c.save_new("test_truncate")
  83             self.assertEqual("zzzzz-4zz18-mockcollection0", c.manifest_locator())
  84             self.assertFalse(c.modified())
  85
  86     def test_write_to_end(self):
  87         keep = ArvadosFileWriterTestCase.MockKeep({"781e5e245d69b566979b86e28d23f2c7+10": "0123456789"})
  88         api = ArvadosFileWriterTestCase.MockApi({"name":"test_append",
  89                                                  "manifest_text": ". 781e5e245d69b566979b86e28d23f2c7+10 acbd18db4cc2f85cedef654fccc4a4d8+3 0:13:count.txt\n"},
  90                                                 {"uuid":"zzzzz-4zz18-mockcollection0",
  91                                                  "manifest_text": ". 781e5e245d69b566979b86e28d23f2c7+10 acbd18db4cc2f85cedef654fccc4a4d8+3 0:13:count.txt\n"})
  92         with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n',
  93                              api_client=api, keep_client=keep) as c:
  94             writer = c.open("count.txt", "r+")
  95             self.assertEqual(writer.size(), 10)
  96
  97             writer.seek(5, os.SEEK_SET)
  98             self.assertEqual("56789", writer.read(8))
  99
 100             writer.seek(10, os.SEEK_SET)
 101             writer.write("foo")
 102             self.assertEqual(writer.size(), 13)
 103
 104             writer.seek(5, os.SEEK_SET)
 105             self.assertEqual("56789foo", writer.read(8))
 106
 107             self.assertIsNone(c.manifest_locator())
 108             self.assertTrue(c.modified())
 109             self.assertIsNone(keep.get("acbd18db4cc2f85cedef654fccc4a4d8+3"))
 110
 111             c.save_new("test_append")
 112             self.assertEqual("zzzzz-4zz18-mockcollection0", c.manifest_locator())
 113             self.assertFalse(c.modified())
 114             self.assertEqual("foo", keep.get("acbd18db4cc2f85cedef654fccc4a4d8+3"))
 115
 116
 117     def test_append(self):
 118         keep = ArvadosFileWriterTestCase.MockKeep({"781e5e245d69b566979b86e28d23f2c7+10": "0123456789"})
 119         c = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n', keep_client=keep)
 120         writer = c.open("count.txt", "a+")
 121         self.assertEqual(writer.read(20), "0123456789")
 122         writer.seek(0, os.SEEK_SET)
 123
 124         writer.write("hello")
 125         self.assertEqual(writer.read(20), "0123456789hello")
 126         writer.seek(0, os.SEEK_SET)
 127
 128         writer.write("world")
 129         self.assertEqual(writer.read(20), "0123456789helloworld")
 130
 131         self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 fc5e038d38a57032085441e7fe7010b0+10 0:20:count.txt\n", c.portable_manifest_text())
 132
 133     def test_write_at_beginning(self):
 134         keep = ArvadosFileWriterTestCase.MockKeep({"781e5e245d69b566979b86e28d23f2c7+10": "0123456789"})
 135         with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n',
 136                              keep_client=keep) as c:
 137             writer = c.open("count.txt", "r+")
 138             self.assertEqual("0123456789", writer.readfrom(0, 13))
 139             writer.seek(0, os.SEEK_SET)
 140             writer.write("foo")
 141             self.assertEqual(writer.size(), 10)
 142             self.assertEqual("foo3456789", writer.readfrom(0, 13))
 143             self.assertEqual(". acbd18db4cc2f85cedef654fccc4a4d8+3 781e5e245d69b566979b86e28d23f2c7+10 0:3:count.txt 6:7:count.txt\n", c.portable_manifest_text())
 144
 145     def test_write_empty(self):
 146         keep = ArvadosFileWriterTestCase.MockKeep({})
 147         with Collection(keep_client=keep) as c:
 148             writer = c.open("count.txt", "w")
 149             self.assertEqual(writer.size(), 0)
 150             self.assertEqual(". d41d8cd98f00b204e9800998ecf8427e+0 0:0:count.txt\n", c.portable_manifest_text())
 151
 152     def test_save_manifest_text(self):
 153         keep = ArvadosFileWriterTestCase.MockKeep({})
 154         with Collection(keep_client=keep) as c:
 155             writer = c.open("count.txt", "w")
 156             writer.write("0123456789")
 157             self.assertEqual('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n', c.portable_manifest_text())
 158             self.assertNotIn('781e5e245d69b566979b86e28d23f2c7+10', keep.blocks)
 159
 160             self.assertEqual('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n', c.save_new(create_collection_record=False))
 161             self.assertIn('781e5e245d69b566979b86e28d23f2c7+10', keep.blocks)
 162
 163     def test_get_manifest_text_commits(self):
 164          keep = ArvadosFileWriterTestCase.MockKeep({})
 165          with Collection(keep_client=keep) as c:
 166              writer = c.open("count.txt", "w")
 167              writer.write("0123456789")
 168              self.assertEqual('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n', c.portable_manifest_text())
 169              self.assertNotIn('781e5e245d69b566979b86e28d23f2c7+10', keep.blocks)
 170              self.assertEqual('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n', c.manifest_text())
 171              self.assertIn('781e5e245d69b566979b86e28d23f2c7+10', keep.blocks)
 172
 173
 174     def test_write_in_middle(self):
 175         keep = ArvadosFileWriterTestCase.MockKeep({"781e5e245d69b566979b86e28d23f2c7+10": "0123456789"})
 176         with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n',
 177                              keep_client=keep) as c:
 178             writer = c.open("count.txt", "r+")
 179             self.assertEqual("0123456789", writer.readfrom(0, 13))
 180             writer.seek(3, os.SEEK_SET)
 181             writer.write("foo")
 182             self.assertEqual(writer.size(), 10)
 183             self.assertEqual("012foo6789", writer.readfrom(0, 13))
 184             self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 acbd18db4cc2f85cedef654fccc4a4d8+3 0:3:count.txt 10:3:count.txt 6:4:count.txt\n", c.portable_manifest_text())
 185
 186     def test_write_at_end(self):
 187         keep = ArvadosFileWriterTestCase.MockKeep({"781e5e245d69b566979b86e28d23f2c7+10": "0123456789"})
 188         with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n',
 189                              keep_client=keep) as c:
 190             writer = c.open("count.txt", "r+")
 191             self.assertEqual("0123456789", writer.readfrom(0, 13))
 192             writer.seek(7, os.SEEK_SET)
 193             writer.write("foo")
 194             self.assertEqual(writer.size(), 10)
 195             self.assertEqual("0123456foo", writer.readfrom(0, 13))
 196             self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 acbd18db4cc2f85cedef654fccc4a4d8+3 0:7:count.txt 10:3:count.txt\n", c.portable_manifest_text())
 197
 198     def test_write_across_segment_boundary(self):
 199         keep = ArvadosFileWriterTestCase.MockKeep({"781e5e245d69b566979b86e28d23f2c7+10": "0123456789"})
 200         with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt 0:10:count.txt\n',
 201                              keep_client=keep) as c:
 202             writer = c.open("count.txt", "r+")
 203             self.assertEqual("012345678901234", writer.readfrom(0, 15))
 204             writer.seek(7, os.SEEK_SET)
 205             writer.write("foobar")
 206             self.assertEqual(writer.size(), 20)
 207             self.assertEqual("0123456foobar34", writer.readfrom(0, 15))
 208             self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 3858f62230ac3c915f300c664312c63f+6 0:7:count.txt 10:6:count.txt 3:7:count.txt\n", c.portable_manifest_text())
 209
 210     def test_write_across_several_segments(self):
 211         keep = ArvadosFileWriterTestCase.MockKeep({"781e5e245d69b566979b86e28d23f2c7+10": "0123456789"})
 212         with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:4:count.txt 0:4:count.txt 0:4:count.txt',
 213                              keep_client=keep) as c:
 214             writer = c.open("count.txt", "r+")
 215             self.assertEqual("012301230123", writer.readfrom(0, 15))
 216             writer.seek(2, os.SEEK_SET)
 217             writer.write("abcdefg")
 218             self.assertEqual(writer.size(), 12)
 219             self.assertEqual("01abcdefg123", writer.readfrom(0, 15))
 220             self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 7ac66c0f148de9519b8bd264312c4d64+7 0:2:count.txt 10:7:count.txt 1:3:count.txt\n", c.portable_manifest_text())
 221
 222     def test_write_large(self):
 223         keep = ArvadosFileWriterTestCase.MockKeep({})
 224         api = ArvadosFileWriterTestCase.MockApi({"name":"test_write_large",
 225                                                  "manifest_text": ". a5de24f4417cfba9d5825eadc2f4ca49+67108000 598cc1a4ccaef8ab6e4724d87e675d78+32892000 0:100000000:count.txt\n"},
 226                                                 {"uuid":"zzzzz-4zz18-mockcollection0",
 227                                                  "manifest_text": ". a5de24f4417cfba9d5825eadc2f4ca49+67108000 598cc1a4ccaef8ab6e4724d87e675d78+32892000 0:100000000:count.txt\n"})
 228         with Collection('. ' + arvados.config.EMPTY_BLOCK_LOCATOR + ' 0:0:count.txt',
 229                              api_client=api, keep_client=keep) as c:
 230             writer = c.open("count.txt", "r+")
 231             text = "0123456789" * 100
 232             for b in xrange(0, 100000):
 233                 writer.write(text)
 234             self.assertEqual(writer.size(), 100000000)
 235
 236             self.assertIsNone(c.manifest_locator())
 237             self.assertTrue(c.modified())
 238             c.save_new("test_write_large")
 239             self.assertEqual("zzzzz-4zz18-mockcollection0", c.manifest_locator())
 240             self.assertFalse(c.modified())
 241
 242
 243     def test_large_write(self):
 244         keep = ArvadosFileWriterTestCase.MockKeep({})
 245         api = ArvadosFileWriterTestCase.MockApi({}, {})
 246         with Collection('. ' + arvados.config.EMPTY_BLOCK_LOCATOR + ' 0:0:count.txt',
 247                              api_client=api, keep_client=keep) as c:
 248             writer = c.open("count.txt", "r+")
 249             self.assertEqual(writer.size(), 0)
 250
 251             text = "0123456789"
 252             writer.write(text)
 253             text = "0123456789" * 9999999
 254             writer.write(text)
 255             self.assertEqual(writer.size(), 100000000)
 256
 257             self.assertEqual(c.manifest_text(), ". 781e5e245d69b566979b86e28d23f2c7+10 48dd23ea1645fd47d789804d71b5bb8e+67108864 77c57dc6ac5a10bb2205caaa73187994+32891126 0:100000000:count.txt\n")
 258
 259     def test_rewrite_on_empty_file(self):
 260         keep = ArvadosFileWriterTestCase.MockKeep({})
 261         with Collection('. ' + arvados.config.EMPTY_BLOCK_LOCATOR + ' 0:0:count.txt',
 262                              keep_client=keep) as c:
 263             writer = c.open("count.txt", "r+")
 264             for b in xrange(0, 10):
 265                 writer.seek(0, os.SEEK_SET)
 266                 writer.write("0123456789")
 267
 268             self.assertEqual(writer.size(), 10)
 269             self.assertEqual("0123456789", writer.readfrom(0, 20))
 270             self.assertEqual(". 7a08b07e84641703e5f2c836aa59a170+100 90:10:count.txt\n", c.portable_manifest_text())
 271             writer.flush()
 272             self.assertEqual(writer.size(), 10)
 273             self.assertEqual("0123456789", writer.readfrom(0, 20))
 274             self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n", c.portable_manifest_text())
 275
 276     def test_rewrite_append_existing_file(self):
 277         keep = ArvadosFileWriterTestCase.MockKeep({"781e5e245d69b566979b86e28d23f2c7+10": "0123456789"})
 278         with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt',
 279                              keep_client=keep) as c:
 280             writer = c.open("count.txt", "r+")
 281             for b in xrange(0, 10):
 282                 writer.seek(10, os.SEEK_SET)
 283                 writer.write("abcdefghij")
 284
 285             self.assertEqual(writer.size(), 20)
 286             self.assertEqual("0123456789abcdefghij", writer.readfrom(0, 20))
 287             self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 ae5f43bab79cf0be33f025fa97ae7398+100 0:10:count.txt 100:10:count.txt\n", c.portable_manifest_text())
 288
 289             writer.arvadosfile.flush()
 290             self.assertEqual(writer.size(), 20)
 291             self.assertEqual("0123456789abcdefghij", writer.readfrom(0, 20))
 292             self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 a925576942e94b2ef57a066101b48876+10 0:20:count.txt\n", c.portable_manifest_text())
 293
 294     def test_rewrite_over_existing_file(self):
 295         keep = ArvadosFileWriterTestCase.MockKeep({"781e5e245d69b566979b86e28d23f2c7+10": "0123456789"})
 296         with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt',
 297                              keep_client=keep) as c:
 298             writer = c.open("count.txt", "r+")
 299             for b in xrange(0, 10):
 300                 writer.seek(5, os.SEEK_SET)
 301                 writer.write("abcdefghij")
 302
 303             self.assertEqual(writer.size(), 15)
 304             self.assertEqual("01234abcdefghij", writer.readfrom(0, 20))
 305             self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 ae5f43bab79cf0be33f025fa97ae7398+100 0:5:count.txt 100:10:count.txt\n", c.portable_manifest_text())
 306
 307             writer.arvadosfile.flush()
 308
 309             self.assertEqual(writer.size(), 15)
 310             self.assertEqual("01234abcdefghij", writer.readfrom(0, 20))
 311             self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 a925576942e94b2ef57a066101b48876+10 0:5:count.txt 10:10:count.txt\n", c.portable_manifest_text())
 312
 313     def test_write_large_rewrite(self):
 314         keep = ArvadosFileWriterTestCase.MockKeep({})
 315         api = ArvadosFileWriterTestCase.MockApi({"name":"test_write_large",
 316                                                  "manifest_text": ". 37400a68af9abdd76ca5bf13e819e42a+32892003 a5de24f4417cfba9d5825eadc2f4ca49+67108000 32892000:3:count.txt 32892006:67107997:count.txt 0:32892000:count.txt\n"},
 317                                                 {"uuid":"zzzzz-4zz18-mockcollection0",
 318                                                  "manifest_text": ". 37400a68af9abdd76ca5bf13e819e42a+32892003 a5de24f4417cfba9d5825eadc2f4ca49+67108000 32892000:3:count.txt 32892006:67107997:count.txt 0:32892000:count.txt\n"})
 319         with Collection('. ' + arvados.config.EMPTY_BLOCK_LOCATOR + ' 0:0:count.txt',
 320                              api_client=api, keep_client=keep) as c:
 321             writer = c.open("count.txt", "r+")
 322             text = ''.join(["0123456789" for a in xrange(0, 100)])
 323             for b in xrange(0, 100000):
 324                 writer.write(text)
 325             writer.seek(0, os.SEEK_SET)
 326             writer.write("foo")
 327             self.assertEqual(writer.size(), 100000000)
 328
 329             self.assertIsNone(c.manifest_locator())
 330             self.assertTrue(c.modified())
 331             c.save_new("test_write_large")
 332             self.assertEqual("zzzzz-4zz18-mockcollection0", c.manifest_locator())
 333             self.assertFalse(c.modified())
 334
 335     def test_create(self):
 336         keep = ArvadosFileWriterTestCase.MockKeep({})
 337         api = ArvadosFileWriterTestCase.MockApi({"name":"test_create",
 338                                                  "manifest_text":". 2e9ec317e197819358fbc43afca7d837+8 0:8:count.txt\n"},
 339                                                 {"uuid":"zzzzz-4zz18-mockcollection0",
 340                                                  "manifest_text":". 2e9ec317e197819358fbc43afca7d837+8 0:8:count.txt\n"})
 341         with Collection(api_client=api, keep_client=keep) as c:
 342             writer = c.open("count.txt", "w+")
 343             self.assertEqual(writer.size(), 0)
 344             writer.write("01234567")
 345             self.assertEqual(writer.size(), 8)
 346
 347             self.assertIsNone(c.manifest_locator())
 348             self.assertTrue(c.modified())
 349             self.assertIsNone(keep.get("2e9ec317e197819358fbc43afca7d837+8"))
 350             c.save_new("test_create")
 351             self.assertEqual("zzzzz-4zz18-mockcollection0", c.manifest_locator())
 352             self.assertFalse(c.modified())
 353             self.assertEqual("01234567", keep.get("2e9ec317e197819358fbc43afca7d837+8"))
 354
 355
 356     def test_create_subdir(self):
 357         keep = ArvadosFileWriterTestCase.MockKeep({})
 358         api = ArvadosFileWriterTestCase.MockApi({"name":"test_create",
 359                                                  "manifest_text":"./foo/bar 2e9ec317e197819358fbc43afca7d837+8 0:8:count.txt\n"},
 360                                                 {"uuid":"zzzzz-4zz18-mockcollection0",
 361                                                  "manifest_text":"./foo/bar 2e9ec317e197819358fbc43afca7d837+8 0:8:count.txt\n"})
 362         with Collection(api_client=api, keep_client=keep) as c:
 363             self.assertIsNone(c.api_response())
 364             writer = c.open("foo/bar/count.txt", "w+")
 365             writer.write("01234567")
 366             self.assertFalse(c.committed())
 367             c.save_new("test_create")
 368             self.assertTrue(c.committed())
 369             self.assertEqual(c.api_response(), api.response)
 370
 371     def test_overwrite(self):
 372         keep = ArvadosFileWriterTestCase.MockKeep({"781e5e245d69b566979b86e28d23f2c7+10": "0123456789"})
 373         api = ArvadosFileWriterTestCase.MockApi({"name":"test_overwrite",
 374                                                  "manifest_text":". 2e9ec317e197819358fbc43afca7d837+8 0:8:count.txt\n"},
 375                                                 {"uuid":"zzzzz-4zz18-mockcollection0",
 376                                                  "manifest_text":". 2e9ec317e197819358fbc43afca7d837+8 0:8:count.txt\n"})
 377         with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n',
 378                              api_client=api, keep_client=keep) as c:
 379             writer = c.open("count.txt", "w+")
 380             self.assertEqual(writer.size(), 0)
 381             writer.write("01234567")
 382             self.assertEqual(writer.size(), 8)
 383
 384             self.assertIsNone(c.manifest_locator())
 385             self.assertTrue(c.modified())
 386             c.save_new("test_overwrite")
 387             self.assertEqual("zzzzz-4zz18-mockcollection0", c.manifest_locator())
 388             self.assertFalse(c.modified())
 389
 390     def test_file_not_found(self):
 391         with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n') as c:
 392             with self.assertRaises(IOError):
 393                 writer = c.open("nocount.txt", "r")
 394
 395     def test_cannot_open_directory(self):
 396         with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n') as c:
 397             with self.assertRaises(IOError):
 398                 writer = c.open(".", "r")
 399
 400     def test_create_multiple(self):
 401         keep = ArvadosFileWriterTestCase.MockKeep({})
 402         api = ArvadosFileWriterTestCase.MockApi({"name":"test_create_multiple",
 403                                                  "manifest_text":". 2e9ec317e197819358fbc43afca7d837+8 e8dc4081b13434b45189a720b77b6818+8 0:8:count1.txt 8:8:count2.txt\n"},
 404                                                 {"uuid":"zzzzz-4zz18-mockcollection0",
 405                                                  "manifest_text":". 2e9ec317e197819358fbc43afca7d837+8 e8dc4081b13434b45189a720b77b6818+8 0:8:count1.txt 8:8:count2.txt\n"})
 406         with Collection(api_client=api, keep_client=keep) as c:
 407             w1 = c.open("count1.txt", "w")
 408             w2 = c.open("count2.txt", "w")
 409             w1.write("01234567")
 410             w2.write("abcdefgh")
 411             self.assertEqual(w1.size(), 8)
 412             self.assertEqual(w2.size(), 8)
 413
 414             self.assertIsNone(c.manifest_locator())
 415             self.assertTrue(c.modified())
 416             self.assertIsNone(keep.get("2e9ec317e197819358fbc43afca7d837+8"))
 417             c.save_new("test_create_multiple")
 418             self.assertEqual("zzzzz-4zz18-mockcollection0", c.manifest_locator())
 419             self.assertFalse(c.modified())
 420             self.assertEqual("01234567", keep.get("2e9ec317e197819358fbc43afca7d837+8"))
 421
 422
 423 class ArvadosFileReaderTestCase(StreamFileReaderTestCase):
 424     class MockParent(object):
 425         class MockBlockMgr(object):
 426             def __init__(self, blocks, nocache):
 427                 self.blocks = blocks
 428                 self.nocache = nocache
 429
 430             def block_prefetch(self, loc):
 431                 pass
 432
 433             def get_block_contents(self, loc, num_retries=0, cache_only=False):
 434                 if self.nocache and cache_only:
 435                     return None
 436                 return self.blocks[loc]
 437
 438         def __init__(self, blocks, nocache):
 439             self.blocks = blocks
 440             self.nocache = nocache
 441             self.lock = arvados.arvfile.NoopLock()
 442
 443         def root_collection(self):
 444             return self
 445
 446         def _my_block_manager(self):
 447             return ArvadosFileReaderTestCase.MockParent.MockBlockMgr(self.blocks, self.nocache)
 448
 449
 450     def make_count_reader(self, nocache=False):
 451         stream = []
 452         n = 0
 453         blocks = {}
 454         for d in ['01234', '34567', '67890']:
 455             loc = '{}+{}'.format(hashlib.md5(d).hexdigest(), len(d))
 456             blocks[loc] = d
 457             stream.append(Range(loc, n, len(d)))
 458             n += len(d)
 459         af = ArvadosFile(ArvadosFileReaderTestCase.MockParent(blocks, nocache), "count.txt", stream=stream, segments=[Range(1, 0, 3), Range(6, 3, 3), Range(11, 6, 3)])
 460         return ArvadosFileReader(af)
 461
 462     def test_read_block_crossing_behavior(self):
 463         # read() needs to return all the data requested if possible, even if it
 464         # crosses uncached blocks: https://arvados.org/issues/5856
 465         sfile = self.make_count_reader(nocache=True)
 466         self.assertEqual('12345678', sfile.read(8))
 467
 468     def test_successive_reads(self):
 469         # Override StreamFileReaderTestCase.test_successive_reads
 470         sfile = self.make_count_reader(nocache=True)
 471         self.assertEqual('1234', sfile.read(4))
 472         self.assertEqual('5678', sfile.read(4))
 473         self.assertEqual('9', sfile.read(4))
 474         self.assertEqual('', sfile.read(4))
 475
 476     def test_tell_after_block_read(self):
 477         # Override StreamFileReaderTestCase.test_tell_after_block_read
 478         sfile = self.make_count_reader(nocache=True)
 479         self.assertEqual('12345678', sfile.read(8))
 480         self.assertEqual(8, sfile.tell())
 481
 482     def test_prefetch(self):
 483         keep = ArvadosFileWriterTestCase.MockKeep({"2e9ec317e197819358fbc43afca7d837+8": "01234567", "e8dc4081b13434b45189a720b77b6818+8": "abcdefgh"})
 484         with Collection(". 2e9ec317e197819358fbc43afca7d837+8 e8dc4081b13434b45189a720b77b6818+8 0:16:count.txt\n", keep_client=keep) as c:
 485             r = c.open("count.txt", "r")
 486             self.assertEqual("0123", r.read(4))
 487         self.assertIn("2e9ec317e197819358fbc43afca7d837+8", keep.requests)
 488         self.assertIn("e8dc4081b13434b45189a720b77b6818+8", keep.requests)
 489
 490     def test__eq__from_manifest(self):
 491         with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt') as c1:
 492             with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt') as c2:
 493                 self.assertTrue(c1["count1.txt"] == c2["count1.txt"])
 494                 self.assertFalse(c1["count1.txt"] != c2["count1.txt"])
 495
 496     def test__eq__from_writes(self):
 497         with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt') as c1:
 498             with Collection() as c2:
 499                 f = c2.open("count1.txt", "w")
 500                 f.write("0123456789")
 501
 502                 self.assertTrue(c1["count1.txt"] == c2["count1.txt"])
 503                 self.assertFalse(c1["count1.txt"] != c2["count1.txt"])
 504
 505     def test__ne__(self):
 506         with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt') as c1:
 507             with Collection() as c2:
 508                 f = c2.open("count1.txt", "w")
 509                 f.write("1234567890")
 510
 511                 self.assertTrue(c1["count1.txt"] != c2["count1.txt"])
 512                 self.assertFalse(c1["count1.txt"] == c2["count1.txt"])
 513
 514
 515 class ArvadosFileReadTestCase(unittest.TestCase, StreamRetryTestMixin):
 516     def reader_for(self, coll_name, **kwargs):
 517         stream = []
 518         segments = []
 519         n = 0
 520         for d in self.manifest_for(coll_name).split():
 521             try:
 522                 k = KeepLocator(d)
 523                 segments.append(Range(n, n, k.size))
 524                 stream.append(Range(d, n, k.size))
 525                 n += k.size
 526             except ValueError:
 527                 pass
 528
 529         blockmanager = arvados.arvfile._BlockManager(self.keep_client())
 530         blockmanager.prefetch_enabled = False
 531         col = Collection(keep_client=self.keep_client(), block_manager=blockmanager)
 532         af = ArvadosFile(col, "test",
 533                          stream=stream,
 534                          segments=segments)
 535         return ArvadosFileReader(af, **kwargs)
 536
 537     def read_for_test(self, reader, byte_count, **kwargs):
 538         return reader.read(byte_count, **kwargs)
 539
 540
 541 class ArvadosFileReadFromTestCase(ArvadosFileReadTestCase):
 542     def read_for_test(self, reader, byte_count, **kwargs):
 543         return reader.readfrom(0, byte_count, **kwargs)
 544
 545
 546 class ArvadosFileReadAllTestCase(ArvadosFileReadTestCase):
 547     def read_for_test(self, reader, byte_count, **kwargs):
 548         return ''.join(reader.readall(**kwargs))
 549
 550
 551 class ArvadosFileReadAllDecompressedTestCase(ArvadosFileReadTestCase):
 552     def read_for_test(self, reader, byte_count, **kwargs):
 553         return ''.join(reader.readall_decompressed(**kwargs))
 554
 555
 556 class ArvadosFileReadlinesTestCase(ArvadosFileReadTestCase):
 557     def read_for_test(self, reader, byte_count, **kwargs):
 558         return ''.join(reader.readlines(**kwargs))
 559
 560 class BlockManagerTest(unittest.TestCase):
 561     def test_bufferblock_append(self):
 562         keep = ArvadosFileWriterTestCase.MockKeep({})
 563         with arvados.arvfile._BlockManager(keep) as blockmanager:
 564             bufferblock = blockmanager.alloc_bufferblock()
 565             bufferblock.append("foo")
 566
 567             self.assertEqual(bufferblock.size(), 3)
 568             self.assertEqual(bufferblock.buffer_view[0:3], "foo")
 569             self.assertEqual(bufferblock.locator(), "acbd18db4cc2f85cedef654fccc4a4d8+3")
 570
 571             bufferblock.append("bar")
 572
 573             self.assertEqual(bufferblock.size(), 6)
 574             self.assertEqual(bufferblock.buffer_view[0:6], "foobar")
 575             self.assertEqual(bufferblock.locator(), "3858f62230ac3c915f300c664312c63f+6")
 576
 577             bufferblock.set_state(arvados.arvfile._BufferBlock.PENDING)
 578             with self.assertRaises(arvados.errors.AssertionError):
 579                 bufferblock.append("bar")
 580
 581     def test_bufferblock_dup(self):
 582         keep = ArvadosFileWriterTestCase.MockKeep({})
 583         with arvados.arvfile._BlockManager(keep) as blockmanager:
 584             bufferblock = blockmanager.alloc_bufferblock()
 585             bufferblock.append("foo")
 586
 587             self.assertEqual(bufferblock.size(), 3)
 588             self.assertEqual(bufferblock.buffer_view[0:3], "foo")
 589             self.assertEqual(bufferblock.locator(), "acbd18db4cc2f85cedef654fccc4a4d8+3")
 590             bufferblock.set_state(arvados.arvfile._BufferBlock.PENDING)
 591
 592             bufferblock2 = blockmanager.dup_block(bufferblock, None)
 593             self.assertNotEqual(bufferblock.blockid, bufferblock2.blockid)
 594
 595             bufferblock2.append("bar")
 596
 597             self.assertEqual(bufferblock2.size(), 6)
 598             self.assertEqual(bufferblock2.buffer_view[0:6], "foobar")
 599             self.assertEqual(bufferblock2.locator(), "3858f62230ac3c915f300c664312c63f+6")
 600
 601             self.assertEqual(bufferblock.size(), 3)
 602             self.assertEqual(bufferblock.buffer_view[0:3], "foo")
 603             self.assertEqual(bufferblock.locator(), "acbd18db4cc2f85cedef654fccc4a4d8+3")
 604
 605     def test_bufferblock_get(self):
 606         keep = ArvadosFileWriterTestCase.MockKeep({"781e5e245d69b566979b86e28d23f2c7+10": "0123456789"})
 607         with arvados.arvfile._BlockManager(keep) as blockmanager:
 608             bufferblock = blockmanager.alloc_bufferblock()
 609             bufferblock.append("foo")
 610
 611             self.assertEqual(blockmanager.get_block_contents("781e5e245d69b566979b86e28d23f2c7+10", 1), "0123456789")
 612             self.assertEqual(blockmanager.get_block_contents(bufferblock.blockid, 1), "foo")
 613
 614     def test_bufferblock_commit(self):
 615         mockkeep = mock.MagicMock()
 616         with arvados.arvfile._BlockManager(mockkeep) as blockmanager:
 617             bufferblock = blockmanager.alloc_bufferblock()
 618             bufferblock.owner = mock.MagicMock()
 619             def flush(sync=None):
 620                 blockmanager.commit_bufferblock(bufferblock, sync)
 621             bufferblock.owner.flush.side_effect = flush
 622             bufferblock.append("foo")
 623             blockmanager.commit_all()
 624             self.assertTrue(bufferblock.owner.flush.called)
 625             self.assertTrue(mockkeep.put.called)
 626             self.assertEqual(bufferblock.state(), arvados.arvfile._BufferBlock.COMMITTED)
 627             self.assertIsNone(bufferblock.buffer_view)
 628
 629
 630     def test_bufferblock_commit_with_error(self):
 631         mockkeep = mock.MagicMock()
 632         mockkeep.put.side_effect = arvados.errors.KeepWriteError("fail")
 633         with arvados.arvfile._BlockManager(mockkeep) as blockmanager:
 634             bufferblock = blockmanager.alloc_bufferblock()
 635             bufferblock.owner = mock.MagicMock()
 636             def flush(sync=None):
 637                 blockmanager.commit_bufferblock(bufferblock, sync)
 638             bufferblock.owner.flush.side_effect = flush
 639             bufferblock.append("foo")
 640             with self.assertRaises(arvados.errors.KeepWriteError) as err:
 641                 blockmanager.commit_all()
 642             self.assertTrue(bufferblock.owner.flush.called)
 643             self.assertEqual(str(err.exception), "Error writing some blocks: block acbd18db4cc2f85cedef654fccc4a4d8+3 raised KeepWriteError (fail)")
 644             self.assertEqual(bufferblock.state(), arvados.arvfile._BufferBlock.ERROR)