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.assertEqual(None, c.manifest_locator())
  81             self.assertEqual(True, c.modified())
  82             c.save_new("test_truncate")
  83             self.assertEqual("zzzzz-4zz18-mockcollection0", c.manifest_locator())
  84             self.assertEqual(False, 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.assertEqual(None, c.manifest_locator())
 108             self.assertEqual(True, c.modified())
 109             self.assertEqual(None, keep.get("acbd18db4cc2f85cedef654fccc4a4d8+3"))
 110
 111             c.save_new("test_append")
 112             self.assertEqual("zzzzz-4zz18-mockcollection0", c.manifest_locator())
 113             self.assertEqual(False, 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.assertEqual(None, c.manifest_locator())
 237             self.assertEqual(True, c.modified())
 238             c.save_new("test_write_large")
 239             self.assertEqual("zzzzz-4zz18-mockcollection0", c.manifest_locator())
 240             self.assertEqual(False, c.modified())
 241
 242
 243     def test_large_write(self):
 244         keep = ArvadosFileWriterTestCase.MockKeep({})
 245         api = ArvadosFileWriterTestCase.MockApi({"name":"test_write_large",
 246                                                  "manifest_text": ". 48dd23ea1645fd47d789804d71b5bb8e+67108864 b373ad0ffebb84efb524947b247ea89a+32891136 0:100000000:count.txt\n"},
 247                                                 {"uuid":"zzzzz-4zz18-mockcollection0",
 248                                                  "manifest_text": ". 48dd23ea1645fd47d789804d71b5bb8e+67108864 598cc1a4ccaef8ab6e4724d87e675d78+32891136 0:100000000:count.txt\n"})
 249         with Collection('. ' + arvados.config.EMPTY_BLOCK_LOCATOR + ' 0:0:count.txt',
 250                              api_client=api, keep_client=keep) as c:
 251             writer = c.open("count.txt", "r+")
 252             text = "0123456789" * 10000000
 253             writer.write(text)
 254             self.assertEqual(writer.size(), 100000000)
 255
 256             self.assertEqual(None, c.manifest_locator())
 257             self.assertEqual(True, c.modified())
 258             c.save_new("test_write_large")
 259             self.assertEqual("zzzzz-4zz18-mockcollection0", c.manifest_locator())
 260             self.assertEqual(False, c.modified())
 261
 262     def test_rewrite_on_empty_file(self):
 263         keep = ArvadosFileWriterTestCase.MockKeep({})
 264         with Collection('. ' + arvados.config.EMPTY_BLOCK_LOCATOR + ' 0:0:count.txt',
 265                              keep_client=keep) as c:
 266             writer = c.open("count.txt", "r+")
 267             for b in xrange(0, 10):
 268                 writer.seek(0, os.SEEK_SET)
 269                 writer.write("0123456789")
 270
 271             self.assertEqual(writer.size(), 10)
 272             self.assertEqual("0123456789", writer.readfrom(0, 20))
 273             self.assertEqual(". 7a08b07e84641703e5f2c836aa59a170+100 90:10:count.txt\n", c.portable_manifest_text())
 274             writer.flush()
 275             self.assertEqual(writer.size(), 10)
 276             self.assertEqual("0123456789", writer.readfrom(0, 20))
 277             self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n", c.portable_manifest_text())
 278
 279     def test_rewrite_append_existing_file(self):
 280         keep = ArvadosFileWriterTestCase.MockKeep({"781e5e245d69b566979b86e28d23f2c7+10": "0123456789"})
 281         with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt',
 282                              keep_client=keep) as c:
 283             writer = c.open("count.txt", "r+")
 284             for b in xrange(0, 10):
 285                 writer.seek(10, os.SEEK_SET)
 286                 writer.write("abcdefghij")
 287
 288             self.assertEqual(writer.size(), 20)
 289             self.assertEqual("0123456789abcdefghij", writer.readfrom(0, 20))
 290             self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 ae5f43bab79cf0be33f025fa97ae7398+100 0:10:count.txt 100:10:count.txt\n", c.portable_manifest_text())
 291
 292             writer.arvadosfile.flush()
 293             self.assertEqual(writer.size(), 20)
 294             self.assertEqual("0123456789abcdefghij", writer.readfrom(0, 20))
 295             self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 a925576942e94b2ef57a066101b48876+10 0:20:count.txt\n", c.portable_manifest_text())
 296
 297     def test_rewrite_over_existing_file(self):
 298         keep = ArvadosFileWriterTestCase.MockKeep({"781e5e245d69b566979b86e28d23f2c7+10": "0123456789"})
 299         with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt',
 300                              keep_client=keep) as c:
 301             writer = c.open("count.txt", "r+")
 302             for b in xrange(0, 10):
 303                 writer.seek(5, os.SEEK_SET)
 304                 writer.write("abcdefghij")
 305
 306             self.assertEqual(writer.size(), 15)
 307             self.assertEqual("01234abcdefghij", writer.readfrom(0, 20))
 308             self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 ae5f43bab79cf0be33f025fa97ae7398+100 0:5:count.txt 100:10:count.txt\n", c.portable_manifest_text())
 309
 310             writer.arvadosfile.flush()
 311
 312             self.assertEqual(writer.size(), 15)
 313             self.assertEqual("01234abcdefghij", writer.readfrom(0, 20))
 314             self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 a925576942e94b2ef57a066101b48876+10 0:5:count.txt 10:10:count.txt\n", c.portable_manifest_text())
 315
 316     def test_write_large_rewrite(self):
 317         keep = ArvadosFileWriterTestCase.MockKeep({})
 318         api = ArvadosFileWriterTestCase.MockApi({"name":"test_write_large",
 319                                                  "manifest_text": ". 37400a68af9abdd76ca5bf13e819e42a+32892003 a5de24f4417cfba9d5825eadc2f4ca49+67108000 32892000:3:count.txt 32892006:67107997:count.txt 0:32892000:count.txt\n"},
 320                                                 {"uuid":"zzzzz-4zz18-mockcollection0",
 321                                                  "manifest_text": ". 37400a68af9abdd76ca5bf13e819e42a+32892003 a5de24f4417cfba9d5825eadc2f4ca49+67108000 32892000:3:count.txt 32892006:67107997:count.txt 0:32892000:count.txt\n"})
 322         with Collection('. ' + arvados.config.EMPTY_BLOCK_LOCATOR + ' 0:0:count.txt',
 323                              api_client=api, keep_client=keep) as c:
 324             writer = c.open("count.txt", "r+")
 325             text = ''.join(["0123456789" for a in xrange(0, 100)])
 326             for b in xrange(0, 100000):
 327                 writer.write(text)
 328             writer.seek(0, os.SEEK_SET)
 329             writer.write("foo")
 330             self.assertEqual(writer.size(), 100000000)
 331
 332             self.assertEqual(None, c.manifest_locator())
 333             self.assertEqual(True, c.modified())
 334             c.save_new("test_write_large")
 335             self.assertEqual("zzzzz-4zz18-mockcollection0", c.manifest_locator())
 336             self.assertEqual(False, c.modified())
 337
 338     def test_create(self):
 339         keep = ArvadosFileWriterTestCase.MockKeep({})
 340         api = ArvadosFileWriterTestCase.MockApi({"name":"test_create",
 341                                                  "manifest_text":". 2e9ec317e197819358fbc43afca7d837+8 0:8:count.txt\n"},
 342                                                 {"uuid":"zzzzz-4zz18-mockcollection0",
 343                                                  "manifest_text":". 2e9ec317e197819358fbc43afca7d837+8 0:8:count.txt\n"})
 344         with Collection(api_client=api, keep_client=keep) as c:
 345             writer = c.open("count.txt", "w+")
 346             self.assertEqual(writer.size(), 0)
 347             writer.write("01234567")
 348             self.assertEqual(writer.size(), 8)
 349
 350             self.assertEqual(None, c.manifest_locator())
 351             self.assertEqual(True, c.modified())
 352             self.assertEqual(None, keep.get("2e9ec317e197819358fbc43afca7d837+8"))
 353             c.save_new("test_create")
 354             self.assertEqual("zzzzz-4zz18-mockcollection0", c.manifest_locator())
 355             self.assertEqual(False, c.modified())
 356             self.assertEqual("01234567", keep.get("2e9ec317e197819358fbc43afca7d837+8"))
 357
 358
 359     def test_create_subdir(self):
 360         keep = ArvadosFileWriterTestCase.MockKeep({})
 361         api = ArvadosFileWriterTestCase.MockApi({"name":"test_create",
 362                                                  "manifest_text":"./foo/bar 2e9ec317e197819358fbc43afca7d837+8 0:8:count.txt\n"},
 363                                                 {"uuid":"zzzzz-4zz18-mockcollection0",
 364                                                  "manifest_text":"./foo/bar 2e9ec317e197819358fbc43afca7d837+8 0:8:count.txt\n"})
 365         with Collection(api_client=api, keep_client=keep) as c:
 366             self.assertIsNone(c.api_response())
 367             writer = c.open("foo/bar/count.txt", "w+")
 368             writer.write("01234567")
 369             c.save_new("test_create")
 370             self.assertEqual(c.api_response(), api.response)
 371
 372     def test_overwrite(self):
 373         keep = ArvadosFileWriterTestCase.MockKeep({"781e5e245d69b566979b86e28d23f2c7+10": "0123456789"})
 374         api = ArvadosFileWriterTestCase.MockApi({"name":"test_overwrite",
 375                                                  "manifest_text":". 2e9ec317e197819358fbc43afca7d837+8 0:8:count.txt\n"},
 376                                                 {"uuid":"zzzzz-4zz18-mockcollection0",
 377                                                  "manifest_text":". 2e9ec317e197819358fbc43afca7d837+8 0:8:count.txt\n"})
 378         with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n',
 379                              api_client=api, keep_client=keep) as c:
 380             writer = c.open("count.txt", "w+")
 381             self.assertEqual(writer.size(), 0)
 382             writer.write("01234567")
 383             self.assertEqual(writer.size(), 8)
 384
 385             self.assertEqual(None, c.manifest_locator())
 386             self.assertEqual(True, c.modified())
 387             c.save_new("test_overwrite")
 388             self.assertEqual("zzzzz-4zz18-mockcollection0", c.manifest_locator())
 389             self.assertEqual(False, c.modified())
 390
 391     def test_file_not_found(self):
 392         with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n') as c:
 393             with self.assertRaises(IOError):
 394                 writer = c.open("nocount.txt", "r")
 395
 396     def test_cannot_open_directory(self):
 397         with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n') as c:
 398             with self.assertRaises(IOError):
 399                 writer = c.open(".", "r")
 400
 401     def test_create_multiple(self):
 402         keep = ArvadosFileWriterTestCase.MockKeep({})
 403         api = ArvadosFileWriterTestCase.MockApi({"name":"test_create_multiple",
 404                                                  "manifest_text":". 2e9ec317e197819358fbc43afca7d837+8 e8dc4081b13434b45189a720b77b6818+8 0:8:count1.txt 8:8:count2.txt\n"},
 405                                                 {"uuid":"zzzzz-4zz18-mockcollection0",
 406                                                  "manifest_text":". 2e9ec317e197819358fbc43afca7d837+8 e8dc4081b13434b45189a720b77b6818+8 0:8:count1.txt 8:8:count2.txt\n"})
 407         with Collection(api_client=api, keep_client=keep) as c:
 408             w1 = c.open("count1.txt", "w")
 409             w2 = c.open("count2.txt", "w")
 410             w1.write("01234567")
 411             w2.write("abcdefgh")
 412             self.assertEqual(w1.size(), 8)
 413             self.assertEqual(w2.size(), 8)
 414
 415             self.assertEqual(None, c.manifest_locator())
 416             self.assertEqual(True, c.modified())
 417             self.assertEqual(None, keep.get("2e9ec317e197819358fbc43afca7d837+8"))
 418             c.save_new("test_create_multiple")
 419             self.assertEqual("zzzzz-4zz18-mockcollection0", c.manifest_locator())
 420             self.assertEqual(False, c.modified())
 421             self.assertEqual("01234567", keep.get("2e9ec317e197819358fbc43afca7d837+8"))
 422
 423
 424 class ArvadosFileReaderTestCase(StreamFileReaderTestCase):
 425     class MockParent(object):
 426         class MockBlockMgr(object):
 427             def __init__(self, blocks, nocache):
 428                 self.blocks = blocks
 429                 self.nocache = nocache
 430
 431             def block_prefetch(self, loc):
 432                 pass
 433
 434             def get_block_contents(self, loc, num_retries=0, cache_only=False):
 435                 if self.nocache and cache_only:
 436                     return None
 437                 return self.blocks[loc]
 438
 439         def __init__(self, blocks, nocache):
 440             self.blocks = blocks
 441             self.nocache = nocache
 442             self.lock = arvados.arvfile.NoopLock()
 443
 444         def root_collection(self):
 445             return self
 446
 447         def _my_block_manager(self):
 448             return ArvadosFileReaderTestCase.MockParent.MockBlockMgr(self.blocks, self.nocache)
 449
 450
 451     def make_count_reader(self, nocache=False):
 452         stream = []
 453         n = 0
 454         blocks = {}
 455         for d in ['01234', '34567', '67890']:
 456             loc = '{}+{}'.format(hashlib.md5(d).hexdigest(), len(d))
 457             blocks[loc] = d
 458             stream.append(Range(loc, n, len(d)))
 459             n += len(d)
 460         af = ArvadosFile(ArvadosFileReaderTestCase.MockParent(blocks, nocache), stream=stream, segments=[Range(1, 0, 3), Range(6, 3, 3), Range(11, 6, 3)])
 461         return ArvadosFileReader(af, "count.txt")
 462
 463     def test_read_block_crossing_behavior(self):
 464         # read() needs to return all the data requested if possible, even if it
 465         # crosses uncached blocks: https://arvados.org/issues/5856
 466         sfile = self.make_count_reader(nocache=True)
 467         self.assertEqual('12345678', sfile.read(8))
 468
 469     def test_successive_reads(self):
 470         # Override StreamFileReaderTestCase.test_successive_reads
 471         sfile = self.make_count_reader(nocache=True)
 472         self.assertEqual('1234', sfile.read(4))
 473         self.assertEqual('5678', sfile.read(4))
 474         self.assertEqual('9', sfile.read(4))
 475         self.assertEqual('', sfile.read(4))
 476
 477     def test_tell_after_block_read(self):
 478         # Override StreamFileReaderTestCase.test_tell_after_block_read
 479         sfile = self.make_count_reader(nocache=True)
 480         self.assertEqual('12345678', sfile.read(8))
 481         self.assertEqual(8, sfile.tell())
 482
 483     def test_prefetch(self):
 484         keep = ArvadosFileWriterTestCase.MockKeep({"2e9ec317e197819358fbc43afca7d837+8": "01234567", "e8dc4081b13434b45189a720b77b6818+8": "abcdefgh"})
 485         with Collection(". 2e9ec317e197819358fbc43afca7d837+8 e8dc4081b13434b45189a720b77b6818+8 0:16:count.txt\n", keep_client=keep) as c:
 486             r = c.open("count.txt", "r")
 487             self.assertEqual("0123", r.read(4))
 488         self.assertIn("2e9ec317e197819358fbc43afca7d837+8", keep.requests)
 489         self.assertIn("e8dc4081b13434b45189a720b77b6818+8", keep.requests)
 490
 491     def test__eq__from_manifest(self):
 492         with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt') as c1:
 493             with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt') as c2:
 494                 self.assertTrue(c1["count1.txt"] == c2["count1.txt"])
 495                 self.assertFalse(c1["count1.txt"] != c2["count1.txt"])
 496
 497     def test__eq__from_writes(self):
 498         with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt') as c1:
 499             with Collection() as c2:
 500                 with c2.open("count1.txt", "w") as f:
 501                     f.write("0123456789")
 502
 503                 self.assertTrue(c1["count1.txt"] == c2["count1.txt"])
 504                 self.assertFalse(c1["count1.txt"] != c2["count1.txt"])
 505
 506     def test__ne__(self):
 507         with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt') as c1:
 508             with Collection() as c2:
 509                 with c2.open("count1.txt", "w") as f:
 510                     f.write("1234567890")
 511
 512                 self.assertTrue(c1["count1.txt"] != c2["count1.txt"])
 513                 self.assertFalse(c1["count1.txt"] == c2["count1.txt"])
 514
 515
 516 class ArvadosFileReadTestCase(unittest.TestCase, StreamRetryTestMixin):
 517     def reader_for(self, coll_name, **kwargs):
 518         stream = []
 519         segments = []
 520         n = 0
 521         for d in self.manifest_for(coll_name).split():
 522             try:
 523                 k = KeepLocator(d)
 524                 segments.append(Range(n, n, k.size))
 525                 stream.append(Range(d, n, k.size))
 526                 n += k.size
 527             except ValueError:
 528                 pass
 529
 530         blockmanager = arvados.arvfile._BlockManager(self.keep_client())
 531         blockmanager.prefetch_enabled = False
 532         col = Collection(keep_client=self.keep_client(), block_manager=blockmanager)
 533         af = ArvadosFile(col,
 534                          stream=stream,
 535                          segments=segments)
 536         return ArvadosFileReader(af, "test", **kwargs)
 537
 538     def read_for_test(self, reader, byte_count, **kwargs):
 539         return reader.read(byte_count, **kwargs)
 540
 541
 542 class ArvadosFileReadFromTestCase(ArvadosFileReadTestCase):
 543     def read_for_test(self, reader, byte_count, **kwargs):
 544         return reader.readfrom(0, byte_count, **kwargs)
 545
 546
 547 class ArvadosFileReadAllTestCase(ArvadosFileReadTestCase):
 548     def read_for_test(self, reader, byte_count, **kwargs):
 549         return ''.join(reader.readall(**kwargs))
 550
 551
 552 class ArvadosFileReadAllDecompressedTestCase(ArvadosFileReadTestCase):
 553     def read_for_test(self, reader, byte_count, **kwargs):
 554         return ''.join(reader.readall_decompressed(**kwargs))
 555
 556
 557 class ArvadosFileReadlinesTestCase(ArvadosFileReadTestCase):
 558     def read_for_test(self, reader, byte_count, **kwargs):
 559         return ''.join(reader.readlines(**kwargs))
 560
 561 class BlockManagerTest(unittest.TestCase):
 562     def test_bufferblock_append(self):
 563         keep = ArvadosFileWriterTestCase.MockKeep({})
 564         blockmanager = arvados.arvfile._BlockManager(keep)
 565         bufferblock = blockmanager.alloc_bufferblock()
 566         bufferblock.append("foo")
 567
 568         self.assertEqual(bufferblock.size(), 3)
 569         self.assertEqual(bufferblock.buffer_view[0:3], "foo")
 570         self.assertEqual(bufferblock.locator(), "acbd18db4cc2f85cedef654fccc4a4d8+3")
 571
 572         bufferblock.append("bar")
 573
 574         self.assertEqual(bufferblock.size(), 6)
 575         self.assertEqual(bufferblock.buffer_view[0:6], "foobar")
 576         self.assertEqual(bufferblock.locator(), "3858f62230ac3c915f300c664312c63f+6")
 577
 578         bufferblock.set_state(arvados.arvfile._BufferBlock.PENDING)
 579         with self.assertRaises(arvados.errors.AssertionError):
 580             bufferblock.append("bar")
 581
 582     def test_bufferblock_dup(self):
 583         keep = ArvadosFileWriterTestCase.MockKeep({})
 584         blockmanager = arvados.arvfile._BlockManager(keep)
 585         bufferblock = blockmanager.alloc_bufferblock()
 586         bufferblock.append("foo")
 587
 588         self.assertEqual(bufferblock.size(), 3)
 589         self.assertEqual(bufferblock.buffer_view[0:3], "foo")
 590         self.assertEqual(bufferblock.locator(), "acbd18db4cc2f85cedef654fccc4a4d8+3")
 591         bufferblock.set_state(arvados.arvfile._BufferBlock.PENDING)
 592
 593         bufferblock2 = blockmanager.dup_block(bufferblock, None)
 594         self.assertNotEqual(bufferblock.blockid, bufferblock2.blockid)
 595
 596         bufferblock2.append("bar")
 597
 598         self.assertEqual(bufferblock2.size(), 6)
 599         self.assertEqual(bufferblock2.buffer_view[0:6], "foobar")
 600         self.assertEqual(bufferblock2.locator(), "3858f62230ac3c915f300c664312c63f+6")
 601
 602         self.assertEqual(bufferblock.size(), 3)
 603         self.assertEqual(bufferblock.buffer_view[0:3], "foo")
 604         self.assertEqual(bufferblock.locator(), "acbd18db4cc2f85cedef654fccc4a4d8+3")
 605
 606     def test_bufferblock_get(self):
 607         keep = ArvadosFileWriterTestCase.MockKeep({"781e5e245d69b566979b86e28d23f2c7+10": "0123456789"})
 608         blockmanager = arvados.arvfile._BlockManager(keep)
 609         bufferblock = blockmanager.alloc_bufferblock()
 610         bufferblock.append("foo")
 611
 612         self.assertEqual(blockmanager.get_block_contents("781e5e245d69b566979b86e28d23f2c7+10", 1), "0123456789")
 613         self.assertEqual(blockmanager.get_block_contents(bufferblock.blockid, 1), "foo")
 614
 615     def test_bufferblock_commit(self):
 616         mockkeep = mock.MagicMock()
 617         blockmanager = arvados.arvfile._BlockManager(mockkeep)
 618         bufferblock = blockmanager.alloc_bufferblock()
 619         bufferblock.owner = mock.MagicMock()
 620         bufferblock.owner.flush.side_effect = lambda: blockmanager.commit_bufferblock(bufferblock)
 621         bufferblock.append("foo")
 622         blockmanager.commit_all()
 623         self.assertTrue(bufferblock.owner.flush.called)
 624         self.assertTrue(mockkeep.put.called)
 625         self.assertEqual(bufferblock.state(), arvados.arvfile._BufferBlock.COMMITTED)
 626         self.assertIsNone(bufferblock.buffer_view)
 627
 628
 629     def test_bufferblock_commit_with_error(self):
 630         mockkeep = mock.MagicMock()
 631         mockkeep.put.side_effect = arvados.errors.KeepWriteError("fail")
 632         blockmanager = arvados.arvfile._BlockManager(mockkeep)
 633         bufferblock = blockmanager.alloc_bufferblock()
 634         bufferblock.owner = mock.MagicMock()
 635         bufferblock.owner.flush.side_effect = lambda: blockmanager.commit_bufferblock(bufferblock)
 636         bufferblock.append("foo")
 637         with self.assertRaises(arvados.errors.KeepWriteError) as err:
 638             blockmanager.commit_all()
 639         self.assertTrue(bufferblock.owner.flush.called)
 640         self.assertEqual(str(err.exception), "Error writing some blocks: block acbd18db4cc2f85cedef654fccc4a4d8+3 raised KeepWriteError (fail)")
 641         self.assertEqual(bufferblock.state(), arvados.arvfile._BufferBlock.PENDING)