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.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.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.manifest_text())
 151
 152     def test_write_in_middle(self):
 153         keep = ArvadosFileWriterTestCase.MockKeep({"781e5e245d69b566979b86e28d23f2c7+10": "0123456789"})
 154         with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n',
 155                              keep_client=keep) as c:
 156             writer = c.open("count.txt", "r+")
 157             self.assertEqual("0123456789", writer.readfrom(0, 13))
 158             writer.seek(3, os.SEEK_SET)
 159             writer.write("foo")
 160             self.assertEqual(writer.size(), 10)
 161             self.assertEqual("012foo6789", writer.readfrom(0, 13))
 162             self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 acbd18db4cc2f85cedef654fccc4a4d8+3 0:3:count.txt 10:3:count.txt 6:4:count.txt\n", c.manifest_text())
 163
 164     def test_write_at_end(self):
 165         keep = ArvadosFileWriterTestCase.MockKeep({"781e5e245d69b566979b86e28d23f2c7+10": "0123456789"})
 166         with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n',
 167                              keep_client=keep) as c:
 168             writer = c.open("count.txt", "r+")
 169             self.assertEqual("0123456789", writer.readfrom(0, 13))
 170             writer.seek(7, os.SEEK_SET)
 171             writer.write("foo")
 172             self.assertEqual(writer.size(), 10)
 173             self.assertEqual("0123456foo", writer.readfrom(0, 13))
 174             self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 acbd18db4cc2f85cedef654fccc4a4d8+3 0:7:count.txt 10:3:count.txt\n", c.manifest_text())
 175
 176     def test_write_across_segment_boundary(self):
 177         keep = ArvadosFileWriterTestCase.MockKeep({"781e5e245d69b566979b86e28d23f2c7+10": "0123456789"})
 178         with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt 0:10:count.txt\n',
 179                              keep_client=keep) as c:
 180             writer = c.open("count.txt", "r+")
 181             self.assertEqual("012345678901234", writer.readfrom(0, 15))
 182             writer.seek(7, os.SEEK_SET)
 183             writer.write("foobar")
 184             self.assertEqual(writer.size(), 20)
 185             self.assertEqual("0123456foobar34", writer.readfrom(0, 15))
 186             self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 3858f62230ac3c915f300c664312c63f+6 0:7:count.txt 10:6:count.txt 3:7:count.txt\n", c.manifest_text())
 187
 188     def test_write_across_several_segments(self):
 189         keep = ArvadosFileWriterTestCase.MockKeep({"781e5e245d69b566979b86e28d23f2c7+10": "0123456789"})
 190         with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:4:count.txt 0:4:count.txt 0:4:count.txt',
 191                              keep_client=keep) as c:
 192             writer = c.open("count.txt", "r+")
 193             self.assertEqual("012301230123", writer.readfrom(0, 15))
 194             writer.seek(2, os.SEEK_SET)
 195             writer.write("abcdefg")
 196             self.assertEqual(writer.size(), 12)
 197             self.assertEqual("01abcdefg123", writer.readfrom(0, 15))
 198             self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 7ac66c0f148de9519b8bd264312c4d64+7 0:2:count.txt 10:7:count.txt 1:3:count.txt\n", c.manifest_text())
 199
 200     def test_write_large(self):
 201         keep = ArvadosFileWriterTestCase.MockKeep({})
 202         api = ArvadosFileWriterTestCase.MockApi({"name":"test_write_large",
 203                                                  "manifest_text": ". a5de24f4417cfba9d5825eadc2f4ca49+67108000 598cc1a4ccaef8ab6e4724d87e675d78+32892000 0:100000000:count.txt\n"},
 204                                                 {"uuid":"zzzzz-4zz18-mockcollection0",
 205                                                  "manifest_text": ". a5de24f4417cfba9d5825eadc2f4ca49+67108000 598cc1a4ccaef8ab6e4724d87e675d78+32892000 0:100000000:count.txt\n"})
 206         with Collection('. ' + arvados.config.EMPTY_BLOCK_LOCATOR + ' 0:0:count.txt',
 207                              api_client=api, keep_client=keep) as c:
 208             writer = c.open("count.txt", "r+")
 209             text = ''.join(["0123456789" for a in xrange(0, 100)])
 210             for b in xrange(0, 100000):
 211                 writer.write(text)
 212             self.assertEqual(writer.size(), 100000000)
 213
 214             self.assertEqual(None, c.manifest_locator())
 215             self.assertEqual(True, c.modified())
 216             c.save_new("test_write_large")
 217             self.assertEqual("zzzzz-4zz18-mockcollection0", c.manifest_locator())
 218             self.assertEqual(False, c.modified())
 219
 220     def test_rewrite_on_empty_file(self):
 221         keep = ArvadosFileWriterTestCase.MockKeep({})
 222         with Collection('. ' + arvados.config.EMPTY_BLOCK_LOCATOR + ' 0:0:count.txt',
 223                              keep_client=keep) as c:
 224             writer = c.open("count.txt", "r+")
 225             for b in xrange(0, 10):
 226                 writer.seek(0, os.SEEK_SET)
 227                 writer.write("0123456789")
 228
 229             self.assertEqual(writer.size(), 10)
 230             self.assertEqual("0123456789", writer.readfrom(0, 20))
 231             self.assertEqual(". 7a08b07e84641703e5f2c836aa59a170+100 90:10:count.txt\n", c.manifest_text())
 232             writer.flush()
 233             self.assertEqual(writer.size(), 10)
 234             self.assertEqual("0123456789", writer.readfrom(0, 20))
 235             self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n", c.manifest_text())
 236
 237     def test_rewrite_append_existing_file(self):
 238         keep = ArvadosFileWriterTestCase.MockKeep({"781e5e245d69b566979b86e28d23f2c7+10": "0123456789"})
 239         with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt',
 240                              keep_client=keep) as c:
 241             writer = c.open("count.txt", "r+")
 242             for b in xrange(0, 10):
 243                 writer.seek(10, os.SEEK_SET)
 244                 writer.write("abcdefghij")
 245
 246             self.assertEqual(writer.size(), 20)
 247             self.assertEqual("0123456789abcdefghij", writer.readfrom(0, 20))
 248             self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 ae5f43bab79cf0be33f025fa97ae7398+100 0:10:count.txt 100:10:count.txt\n", c.manifest_text())
 249
 250             writer.arvadosfile.flush()
 251             self.assertEqual(writer.size(), 20)
 252             self.assertEqual("0123456789abcdefghij", writer.readfrom(0, 20))
 253             self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 a925576942e94b2ef57a066101b48876+10 0:20:count.txt\n", c.manifest_text())
 254
 255     def test_rewrite_over_existing_file(self):
 256         keep = ArvadosFileWriterTestCase.MockKeep({"781e5e245d69b566979b86e28d23f2c7+10": "0123456789"})
 257         with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt',
 258                              keep_client=keep) as c:
 259             writer = c.open("count.txt", "r+")
 260             for b in xrange(0, 10):
 261                 writer.seek(5, os.SEEK_SET)
 262                 writer.write("abcdefghij")
 263
 264             self.assertEqual(writer.size(), 15)
 265             self.assertEqual("01234abcdefghij", writer.readfrom(0, 20))
 266             self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 ae5f43bab79cf0be33f025fa97ae7398+100 0:5:count.txt 100:10:count.txt\n", c.manifest_text())
 267
 268             writer.arvadosfile.flush()
 269
 270             self.assertEqual(writer.size(), 15)
 271             self.assertEqual("01234abcdefghij", writer.readfrom(0, 20))
 272             self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 a925576942e94b2ef57a066101b48876+10 0:5:count.txt 10:10:count.txt\n", c.manifest_text())
 273
 274     def test_write_large_rewrite(self):
 275         keep = ArvadosFileWriterTestCase.MockKeep({})
 276         api = ArvadosFileWriterTestCase.MockApi({"name":"test_write_large",
 277                                                  "manifest_text": ". 37400a68af9abdd76ca5bf13e819e42a+32892003 a5de24f4417cfba9d5825eadc2f4ca49+67108000 32892000:3:count.txt 32892006:67107997:count.txt 0:32892000:count.txt\n"},
 278                                                 {"uuid":"zzzzz-4zz18-mockcollection0",
 279                                                  "manifest_text": ". 37400a68af9abdd76ca5bf13e819e42a+32892003 a5de24f4417cfba9d5825eadc2f4ca49+67108000 32892000:3:count.txt 32892006:67107997:count.txt 0:32892000:count.txt\n"})
 280         with Collection('. ' + arvados.config.EMPTY_BLOCK_LOCATOR + ' 0:0:count.txt',
 281                              api_client=api, keep_client=keep) as c:
 282             writer = c.open("count.txt", "r+")
 283             text = ''.join(["0123456789" for a in xrange(0, 100)])
 284             for b in xrange(0, 100000):
 285                 writer.write(text)
 286             writer.seek(0, os.SEEK_SET)
 287             writer.write("foo")
 288             self.assertEqual(writer.size(), 100000000)
 289
 290             self.assertEqual(None, c.manifest_locator())
 291             self.assertEqual(True, c.modified())
 292             c.save_new("test_write_large")
 293             self.assertEqual("zzzzz-4zz18-mockcollection0", c.manifest_locator())
 294             self.assertEqual(False, c.modified())
 295
 296     def test_create(self):
 297         keep = ArvadosFileWriterTestCase.MockKeep({})
 298         api = ArvadosFileWriterTestCase.MockApi({"name":"test_create",
 299                                                  "manifest_text":". 2e9ec317e197819358fbc43afca7d837+8 0:8:count.txt\n"},
 300                                                 {"uuid":"zzzzz-4zz18-mockcollection0",
 301                                                  "manifest_text":". 2e9ec317e197819358fbc43afca7d837+8 0:8:count.txt\n"})
 302         with Collection(api_client=api, keep_client=keep) as c:
 303             writer = c.open("count.txt", "w+")
 304             self.assertEqual(writer.size(), 0)
 305             writer.write("01234567")
 306             self.assertEqual(writer.size(), 8)
 307
 308             self.assertEqual(None, c.manifest_locator())
 309             self.assertEqual(True, c.modified())
 310             self.assertEqual(None, keep.get("2e9ec317e197819358fbc43afca7d837+8"))
 311             c.save_new("test_create")
 312             self.assertEqual("zzzzz-4zz18-mockcollection0", c.manifest_locator())
 313             self.assertEqual(False, c.modified())
 314             self.assertEqual("01234567", keep.get("2e9ec317e197819358fbc43afca7d837+8"))
 315
 316
 317     def test_create_subdir(self):
 318         keep = ArvadosFileWriterTestCase.MockKeep({})
 319         api = ArvadosFileWriterTestCase.MockApi({"name":"test_create",
 320                                                  "manifest_text":"./foo/bar 2e9ec317e197819358fbc43afca7d837+8 0:8:count.txt\n"},
 321                                                 {"uuid":"zzzzz-4zz18-mockcollection0",
 322                                                  "manifest_text":"./foo/bar 2e9ec317e197819358fbc43afca7d837+8 0:8:count.txt\n"})
 323         with Collection(api_client=api, keep_client=keep) as c:
 324             self.assertIsNone(c.api_response())
 325             writer = c.open("foo/bar/count.txt", "w+")
 326             writer.write("01234567")
 327             c.save_new("test_create")
 328             self.assertEqual(c.api_response(), api.response)
 329
 330     def test_overwrite(self):
 331         keep = ArvadosFileWriterTestCase.MockKeep({"781e5e245d69b566979b86e28d23f2c7+10": "0123456789"})
 332         api = ArvadosFileWriterTestCase.MockApi({"name":"test_overwrite",
 333                                                  "manifest_text":". 2e9ec317e197819358fbc43afca7d837+8 0:8:count.txt\n"},
 334                                                 {"uuid":"zzzzz-4zz18-mockcollection0",
 335                                                  "manifest_text":". 2e9ec317e197819358fbc43afca7d837+8 0:8:count.txt\n"})
 336         with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n',
 337                              api_client=api, keep_client=keep) as c:
 338             writer = c.open("count.txt", "w+")
 339             self.assertEqual(writer.size(), 0)
 340             writer.write("01234567")
 341             self.assertEqual(writer.size(), 8)
 342
 343             self.assertEqual(None, c.manifest_locator())
 344             self.assertEqual(True, c.modified())
 345             c.save_new("test_overwrite")
 346             self.assertEqual("zzzzz-4zz18-mockcollection0", c.manifest_locator())
 347             self.assertEqual(False, c.modified())
 348
 349     def test_file_not_found(self):
 350         with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n') as c:
 351             with self.assertRaises(IOError):
 352                 writer = c.open("nocount.txt", "r")
 353
 354     def test_cannot_open_directory(self):
 355         with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n') as c:
 356             with self.assertRaises(IOError):
 357                 writer = c.open(".", "r")
 358
 359     def test_create_multiple(self):
 360         keep = ArvadosFileWriterTestCase.MockKeep({})
 361         api = ArvadosFileWriterTestCase.MockApi({"name":"test_create_multiple",
 362                                                  "manifest_text":". 2e9ec317e197819358fbc43afca7d837+8 e8dc4081b13434b45189a720b77b6818+8 0:8:count1.txt 8:8:count2.txt\n"},
 363                                                 {"uuid":"zzzzz-4zz18-mockcollection0",
 364                                                  "manifest_text":". 2e9ec317e197819358fbc43afca7d837+8 e8dc4081b13434b45189a720b77b6818+8 0:8:count1.txt 8:8:count2.txt\n"})
 365         with Collection(api_client=api, keep_client=keep) as c:
 366             w1 = c.open("count1.txt", "w")
 367             w2 = c.open("count2.txt", "w")
 368             w1.write("01234567")
 369             w2.write("abcdefgh")
 370             self.assertEqual(w1.size(), 8)
 371             self.assertEqual(w2.size(), 8)
 372
 373             self.assertEqual(None, c.manifest_locator())
 374             self.assertEqual(True, c.modified())
 375             self.assertEqual(None, keep.get("2e9ec317e197819358fbc43afca7d837+8"))
 376             c.save_new("test_create_multiple")
 377             self.assertEqual("zzzzz-4zz18-mockcollection0", c.manifest_locator())
 378             self.assertEqual(False, c.modified())
 379             self.assertEqual("01234567", keep.get("2e9ec317e197819358fbc43afca7d837+8"))
 380
 381
 382 class ArvadosFileReaderTestCase(StreamFileReaderTestCase):
 383     class MockParent(object):
 384         class MockBlockMgr(object):
 385             def __init__(self, blocks, nocache):
 386                 self.blocks = blocks
 387                 self.nocache = nocache
 388
 389             def block_prefetch(self, loc):
 390                 pass
 391
 392             def get_block_contents(self, loc, num_retries=0, cache_only=False):
 393                 if self.nocache and cache_only:
 394                     return None
 395                 return self.blocks[loc]
 396
 397         def __init__(self, blocks, nocache):
 398             self.blocks = blocks
 399             self.nocache = nocache
 400             self.lock = arvados.arvfile.NoopLock()
 401
 402         def root_collection(self):
 403             return self
 404
 405         def _my_block_manager(self):
 406             return ArvadosFileReaderTestCase.MockParent.MockBlockMgr(self.blocks, self.nocache)
 407
 408
 409     def make_count_reader(self, nocache=False):
 410         stream = []
 411         n = 0
 412         blocks = {}
 413         for d in ['01234', '34567', '67890']:
 414             loc = '{}+{}'.format(hashlib.md5(d).hexdigest(), len(d))
 415             blocks[loc] = d
 416             stream.append(Range(loc, n, len(d)))
 417             n += len(d)
 418         af = ArvadosFile(ArvadosFileReaderTestCase.MockParent(blocks, nocache), stream=stream, segments=[Range(1, 0, 3), Range(6, 3, 3), Range(11, 6, 3)])
 419         return ArvadosFileReader(af, "count.txt")
 420
 421     def test_read_returns_first_block(self):
 422         # read() calls will be aligned on block boundaries - see #3663.
 423         sfile = self.make_count_reader(nocache=True)
 424         self.assertEqual('123', sfile.read(10))
 425
 426     def test_successive_reads(self):
 427         sfile = self.make_count_reader(nocache=True)
 428         for expect in ['123', '456', '789', '']:
 429             self.assertEqual(expect, sfile.read(10))
 430
 431     def test_tell_after_block_read(self):
 432         sfile = self.make_count_reader(nocache=True)
 433         sfile.read(5)
 434         self.assertEqual(3, sfile.tell())
 435
 436     def test_prefetch(self):
 437         keep = ArvadosFileWriterTestCase.MockKeep({"2e9ec317e197819358fbc43afca7d837+8": "01234567", "e8dc4081b13434b45189a720b77b6818+8": "abcdefgh"})
 438         with Collection(". 2e9ec317e197819358fbc43afca7d837+8 e8dc4081b13434b45189a720b77b6818+8 0:16:count.txt\n", keep_client=keep) as c:
 439             r = c.open("count.txt", "r")
 440             self.assertEqual("0123", r.read(4))
 441         self.assertIn("2e9ec317e197819358fbc43afca7d837+8", keep.requests)
 442         self.assertIn("e8dc4081b13434b45189a720b77b6818+8", keep.requests)
 443
 444     def test__eq__from_manifest(self):
 445         with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt') as c1:
 446             with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt') as c2:
 447                 self.assertTrue(c1["count1.txt"] == c2["count1.txt"])
 448                 self.assertFalse(c1["count1.txt"] != c2["count1.txt"])
 449
 450     def test__eq__from_writes(self):
 451         with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt') as c1:
 452             with Collection() as c2:
 453                 with c2.open("count1.txt", "w") as f:
 454                     f.write("0123456789")
 455
 456                 self.assertTrue(c1["count1.txt"] == c2["count1.txt"])
 457                 self.assertFalse(c1["count1.txt"] != c2["count1.txt"])
 458
 459     def test__ne__(self):
 460         with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt') as c1:
 461             with Collection() as c2:
 462                 with c2.open("count1.txt", "w") as f:
 463                     f.write("1234567890")
 464
 465                 self.assertTrue(c1["count1.txt"] != c2["count1.txt"])
 466                 self.assertFalse(c1["count1.txt"] == c2["count1.txt"])
 467
 468
 469 class ArvadosFileReadTestCase(unittest.TestCase, StreamRetryTestMixin):
 470     def reader_for(self, coll_name, **kwargs):
 471         stream = []
 472         segments = []
 473         n = 0
 474         for d in self.manifest_for(coll_name).split():
 475             try:
 476                 k = KeepLocator(d)
 477                 segments.append(Range(n, n, k.size))
 478                 stream.append(Range(d, n, k.size))
 479                 n += k.size
 480             except ValueError:
 481                 pass
 482
 483         blockmanager = arvados.arvfile._BlockManager(self.keep_client())
 484         blockmanager.prefetch_enabled = False
 485         col = Collection(keep_client=self.keep_client(), block_manager=blockmanager)
 486         af = ArvadosFile(col,
 487                          stream=stream,
 488                          segments=segments)
 489         return ArvadosFileReader(af, "test", **kwargs)
 490
 491     def read_for_test(self, reader, byte_count, **kwargs):
 492         return reader.read(byte_count, **kwargs)
 493
 494
 495 class ArvadosFileReadFromTestCase(ArvadosFileReadTestCase):
 496     def read_for_test(self, reader, byte_count, **kwargs):
 497         return reader.readfrom(0, byte_count, **kwargs)
 498
 499
 500 class ArvadosFileReadAllTestCase(ArvadosFileReadTestCase):
 501     def read_for_test(self, reader, byte_count, **kwargs):
 502         return ''.join(reader.readall(**kwargs))
 503
 504
 505 class ArvadosFileReadAllDecompressedTestCase(ArvadosFileReadTestCase):
 506     def read_for_test(self, reader, byte_count, **kwargs):
 507         return ''.join(reader.readall_decompressed(**kwargs))
 508
 509
 510 class ArvadosFileReadlinesTestCase(ArvadosFileReadTestCase):
 511     def read_for_test(self, reader, byte_count, **kwargs):
 512         return ''.join(reader.readlines(**kwargs))
 513
 514 class BlockManagerTest(unittest.TestCase):
 515     def test_bufferblock_append(self):
 516         keep = ArvadosFileWriterTestCase.MockKeep({})
 517         blockmanager = arvados.arvfile._BlockManager(keep)
 518         bufferblock = blockmanager.alloc_bufferblock()
 519         bufferblock.append("foo")
 520
 521         self.assertEqual(bufferblock.size(), 3)
 522         self.assertEqual(bufferblock.buffer_view[0:3], "foo")
 523         self.assertEqual(bufferblock.locator(), "acbd18db4cc2f85cedef654fccc4a4d8+3")
 524
 525         bufferblock.append("bar")
 526
 527         self.assertEqual(bufferblock.size(), 6)
 528         self.assertEqual(bufferblock.buffer_view[0:6], "foobar")
 529         self.assertEqual(bufferblock.locator(), "3858f62230ac3c915f300c664312c63f+6")
 530
 531         bufferblock.set_state(arvados.arvfile._BufferBlock.PENDING)
 532         with self.assertRaises(arvados.errors.AssertionError):
 533             bufferblock.append("bar")
 534
 535     def test_bufferblock_dup(self):
 536         keep = ArvadosFileWriterTestCase.MockKeep({})
 537         blockmanager = arvados.arvfile._BlockManager(keep)
 538         bufferblock = blockmanager.alloc_bufferblock()
 539         bufferblock.append("foo")
 540
 541         self.assertEqual(bufferblock.size(), 3)
 542         self.assertEqual(bufferblock.buffer_view[0:3], "foo")
 543         self.assertEqual(bufferblock.locator(), "acbd18db4cc2f85cedef654fccc4a4d8+3")
 544         bufferblock.set_state(arvados.arvfile._BufferBlock.PENDING)
 545
 546         bufferblock2 = blockmanager.dup_block(bufferblock, None)
 547         self.assertNotEqual(bufferblock.blockid, bufferblock2.blockid)
 548
 549         bufferblock2.append("bar")
 550
 551         self.assertEqual(bufferblock2.size(), 6)
 552         self.assertEqual(bufferblock2.buffer_view[0:6], "foobar")
 553         self.assertEqual(bufferblock2.locator(), "3858f62230ac3c915f300c664312c63f+6")
 554
 555         self.assertEqual(bufferblock.size(), 3)
 556         self.assertEqual(bufferblock.buffer_view[0:3], "foo")
 557         self.assertEqual(bufferblock.locator(), "acbd18db4cc2f85cedef654fccc4a4d8+3")
 558
 559     def test_bufferblock_get(self):
 560         keep = ArvadosFileWriterTestCase.MockKeep({"781e5e245d69b566979b86e28d23f2c7+10": "0123456789"})
 561         blockmanager = arvados.arvfile._BlockManager(keep)
 562         bufferblock = blockmanager.alloc_bufferblock()
 563         bufferblock.append("foo")
 564
 565         self.assertEqual(blockmanager.get_block_contents("781e5e245d69b566979b86e28d23f2c7+10", 1), "0123456789")
 566         self.assertEqual(blockmanager.get_block_contents(bufferblock.blockid, 1), "foo")
 567
 568     def test_bufferblock_commit(self):
 569         mockkeep = mock.MagicMock()
 570         blockmanager = arvados.arvfile._BlockManager(mockkeep)
 571         bufferblock = blockmanager.alloc_bufferblock()
 572         bufferblock.append("foo")
 573         blockmanager.commit_all()
 574         self.assertTrue(mockkeep.put.called)
 575         self.assertEqual(bufferblock.state(), arvados.arvfile._BufferBlock.COMMITTED)
 576         self.assertIsNone(bufferblock.buffer_view)
 577
 578
 579     def test_bufferblock_commit_with_error(self):
 580         mockkeep = mock.MagicMock()
 581         mockkeep.put.side_effect = arvados.errors.KeepWriteError("fail")
 582         blockmanager = arvados.arvfile._BlockManager(mockkeep)
 583         bufferblock = blockmanager.alloc_bufferblock()
 584         bufferblock.append("foo")
 585         with self.assertRaises(arvados.errors.KeepWriteError) as err:
 586             blockmanager.commit_all()
 587         self.assertEqual(str(err.exception), "Error writing some blocks: block acbd18db4cc2f85cedef654fccc4a4d8+3 raised KeepWriteError (fail)")
 588         self.assertEqual(bufferblock.state(), arvados.arvfile._BufferBlock.PENDING)