sdk/python/tests/test_collections.py

   1 # usage example:
   2 #
   3 # ARVADOS_API_TOKEN=abc ARVADOS_API_HOST=arvados.local python -m unittest discover
   4
   5 import arvados
   6 import bz2
   7 import copy
   8 import os
   9 import pprint
  10 import subprocess
  11 import tempfile
  12 import unittest
  13
  14 from arvados_testutil import ArvadosKeepLocalStoreTestCase
  15
  16 class TestResumableWriter(arvados.ResumableCollectionWriter):
  17     KEEP_BLOCK_SIZE = 1024  # PUT to Keep every 1K.
  18
  19     def current_state(self):
  20         return self.dump_state(copy.deepcopy)
  21
  22
  23 class ArvadosCollectionsTest(ArvadosKeepLocalStoreTestCase):
  24     def write_foo_bar_baz(self):
  25         cw = arvados.CollectionWriter()
  26         self.assertEqual(cw.current_stream_name(), '.',
  27                          'current_stream_name() should be "." now')
  28         cw.set_current_file_name('foo.txt')
  29         cw.write('foo')
  30         self.assertEqual(cw.current_file_name(), 'foo.txt',
  31                          'current_file_name() should be foo.txt now')
  32         cw.start_new_file('bar.txt')
  33         cw.write('bar')
  34         cw.start_new_stream('baz')
  35         cw.write('baz')
  36         cw.set_current_file_name('baz.txt')
  37         return cw.finish()
  38
  39     def test_keep_local_store(self):
  40         self.assertEqual(arvados.Keep.put('foo'), 'acbd18db4cc2f85cedef654fccc4a4d8+3', 'wrong md5 hash from Keep.put')
  41         self.assertEqual(arvados.Keep.get('acbd18db4cc2f85cedef654fccc4a4d8+3'), 'foo', 'wrong data from Keep.get')
  42
  43     def test_local_collection_writer(self):
  44         self.assertEqual(self.write_foo_bar_baz(),
  45                          'd6c3b8e571f1b81ebb150a45ed06c884+114',
  46                          "wrong locator hash for files foo, bar, baz")
  47
  48     def test_local_collection_reader(self):
  49         self.write_foo_bar_baz()
  50         cr = arvados.CollectionReader('d6c3b8e571f1b81ebb150a45ed06c884+114+Xzizzle')
  51         got = []
  52         for s in cr.all_streams():
  53             for f in s.all_files():
  54                 got += [[f.size(), f.stream_name(), f.name(), f.read(2**26)]]
  55         expected = [[3, '.', 'bar.txt', 'bar'],
  56                     [3, '.', 'foo.txt', 'foo'],
  57                     [3, './baz', 'baz.txt', 'baz']]
  58         self.assertEqual(got,
  59                          expected)
  60         stream0 = cr.all_streams()[0]
  61         self.assertEqual(stream0.readfrom(0, 0),
  62                          '',
  63                          'reading zero bytes should have returned empty string')
  64         self.assertEqual(stream0.readfrom(0, 2**26),
  65                          'foobar',
  66                          'reading entire stream failed')
  67         self.assertEqual(stream0.readfrom(2**26, 0),
  68                          '',
  69                          'reading zero bytes should have returned empty string')
  70
  71     def _test_subset(self, collection, expected):
  72         cr = arvados.CollectionReader(collection)
  73         for s in cr.all_streams():
  74             for ex in expected:
  75                 if ex[0] == s:
  76                     f = s.files()[ex[2]]
  77                     got = [f.size(), f.stream_name(), f.name(), "".join(f.readall(2**26))]
  78                     self.assertEqual(got,
  79                                      ex,
  80                                      'all_files|as_manifest did not preserve manifest contents: got %s expected %s' % (got, ex))
  81
  82     def test_collection_manifest_subset(self):
  83         self.write_foo_bar_baz()
  84         self._test_subset('d6c3b8e571f1b81ebb150a45ed06c884+114',
  85                           [[3, '.',     'bar.txt', 'bar'],
  86                            [3, '.',     'foo.txt', 'foo'],
  87                            [3, './baz', 'baz.txt', 'baz']])
  88         self._test_subset((". %s %s 0:3:foo.txt 3:3:bar.txt\n" %
  89                            (arvados.Keep.put("foo"),
  90                             arvados.Keep.put("bar"))),
  91                           [[3, '.', 'bar.txt', 'bar'],
  92                            [3, '.', 'foo.txt', 'foo']])
  93         self._test_subset((". %s %s 0:2:fo.txt 2:4:obar.txt\n" %
  94                            (arvados.Keep.put("foo"),
  95                             arvados.Keep.put("bar"))),
  96                           [[2, '.', 'fo.txt', 'fo'],
  97                            [4, '.', 'obar.txt', 'obar']])
  98         self._test_subset((". %s %s 0:2:fo.txt 2:0:zero.txt 2:2:ob.txt 4:2:ar.txt\n" %
  99                            (arvados.Keep.put("foo"),
 100                             arvados.Keep.put("bar"))),
 101                           [[2, '.', 'ar.txt', 'ar'],
 102                            [2, '.', 'fo.txt', 'fo'],
 103                            [2, '.', 'ob.txt', 'ob'],
 104                            [0, '.', 'zero.txt', '']])
 105
 106     def _test_readline(self, what_in, what_out):
 107         cw = arvados.CollectionWriter()
 108         cw.start_new_file('test.txt')
 109         cw.write(what_in)
 110         test1 = cw.finish()
 111         cr = arvados.CollectionReader(test1)
 112         got = []
 113         for x in list(cr.all_files())[0].readlines():
 114             got += [x]
 115         self.assertEqual(got,
 116                          what_out,
 117                          "readlines did not split lines correctly: %s" % got)
 118
 119     def test_collection_readline(self):
 120         self._test_readline("\na\nbcd\n\nefg\nz",
 121                             ["\n", "a\n", "bcd\n", "\n", "efg\n", "z"])
 122         self._test_readline("ab\ncd\n",
 123                             ["ab\n", "cd\n"])
 124
 125     def test_collection_empty_file(self):
 126         cw = arvados.CollectionWriter()
 127         cw.start_new_file('zero.txt')
 128         cw.write('')
 129
 130         self.assertEqual(cw.manifest_text(), ". 0:0:zero.txt\n")
 131         self.check_manifest_file_sizes(cw.manifest_text(), [0])
 132         cw = arvados.CollectionWriter()
 133         cw.start_new_file('zero.txt')
 134         cw.write('')
 135         cw.start_new_file('one.txt')
 136         cw.write('1')
 137         cw.start_new_stream('foo')
 138         cw.start_new_file('zero.txt')
 139         cw.write('')
 140         self.check_manifest_file_sizes(cw.manifest_text(), [1,0,0])
 141
 142     def check_manifest_file_sizes(self, manifest_text, expect_sizes):
 143         cr = arvados.CollectionReader(manifest_text)
 144         got_sizes = []
 145         for f in cr.all_files():
 146             got_sizes += [f.size()]
 147         self.assertEqual(got_sizes, expect_sizes, "got wrong file sizes %s, expected %s" % (got_sizes, expect_sizes))
 148
 149     def test_collection_bz2_decompression(self):
 150         n_lines_in = 2**18
 151         data_in = "abc\n"
 152         for x in xrange(0, 18):
 153             data_in += data_in
 154         compressed_data_in = bz2.compress(data_in)
 155         cw = arvados.CollectionWriter()
 156         cw.start_new_file('test.bz2')
 157         cw.write(compressed_data_in)
 158         bz2_manifest = cw.manifest_text()
 159
 160         cr = arvados.CollectionReader(bz2_manifest)
 161
 162         got = 0
 163         for x in list(cr.all_files())[0].readlines():
 164             self.assertEqual(x, "abc\n", "decompression returned wrong data: %s" % x)
 165             got += 1
 166         self.assertEqual(got,
 167                          n_lines_in,
 168                          "decompression returned %d lines instead of %d" % (got, n_lines_in))
 169
 170     def test_collection_gzip_decompression(self):
 171         n_lines_in = 2**18
 172         data_in = "abc\n"
 173         for x in xrange(0, 18):
 174             data_in += data_in
 175         p = subprocess.Popen(["gzip", "-1cn"],
 176                              stdout=subprocess.PIPE,
 177                              stdin=subprocess.PIPE,
 178                              stderr=subprocess.PIPE,
 179                              shell=False, close_fds=True)
 180         compressed_data_in, stderrdata = p.communicate(data_in)
 181
 182         cw = arvados.CollectionWriter()
 183         cw.start_new_file('test.gz')
 184         cw.write(compressed_data_in)
 185         gzip_manifest = cw.manifest_text()
 186
 187         cr = arvados.CollectionReader(gzip_manifest)
 188         got = 0
 189         for x in list(cr.all_files())[0].readlines():
 190             self.assertEqual(x, "abc\n", "decompression returned wrong data: %s" % x)
 191             got += 1
 192         self.assertEqual(got,
 193                          n_lines_in,
 194                          "decompression returned %d lines instead of %d" % (got, n_lines_in))
 195
 196     def test_normalized_collection(self):
 197         m1 = """. 5348b82a029fd9e971a811ce1f71360b+43 0:43:md5sum.txt
 198 . 085c37f02916da1cad16f93c54d899b7+41 0:41:md5sum.txt
 199 . 8b22da26f9f433dea0a10e5ec66d73ba+43 0:43:md5sum.txt"""
 200         self.assertEqual(arvados.CollectionReader(m1).manifest_text(),
 201                          """. 5348b82a029fd9e971a811ce1f71360b+43 085c37f02916da1cad16f93c54d899b7+41 8b22da26f9f433dea0a10e5ec66d73ba+43 0:127:md5sum.txt
 202 """)
 203
 204         m2 = """. 204e43b8a1185621ca55a94839582e6f+67108864 b9677abbac956bd3e86b1deb28dfac03+67108864 fc15aff2a762b13f521baf042140acec+67108864 323d2a3ce20370c4ca1d3462a344f8fd+25885655 0:227212247:var-GS000016015-ASM.tsv.bz2
 205 """
 206         self.assertEqual(arvados.CollectionReader(m2).manifest_text(), m2)
 207
 208         m3 = """. 5348b82a029fd9e971a811ce1f71360b+43 3:40:md5sum.txt
 209 . 085c37f02916da1cad16f93c54d899b7+41 0:41:md5sum.txt
 210 . 8b22da26f9f433dea0a10e5ec66d73ba+43 0:43:md5sum.txt"""
 211         self.assertEqual(arvados.CollectionReader(m3).manifest_text(),
 212                          """. 5348b82a029fd9e971a811ce1f71360b+43 085c37f02916da1cad16f93c54d899b7+41 8b22da26f9f433dea0a10e5ec66d73ba+43 3:124:md5sum.txt
 213 """)
 214
 215         m4 = """. 204e43b8a1185621ca55a94839582e6f+67108864 0:3:foo/bar
 216 ./zzz 204e43b8a1185621ca55a94839582e6f+67108864 0:999:zzz
 217 ./foo 323d2a3ce20370c4ca1d3462a344f8fd+25885655 0:3:bar"""
 218         self.assertEqual(arvados.CollectionReader(m4).manifest_text(),
 219                          """./foo 204e43b8a1185621ca55a94839582e6f+67108864 323d2a3ce20370c4ca1d3462a344f8fd+25885655 0:3:bar 67108864:3:bar
 220 ./zzz 204e43b8a1185621ca55a94839582e6f+67108864 0:999:zzz
 221 """)
 222
 223         m5 = """. 204e43b8a1185621ca55a94839582e6f+67108864 0:3:foo/bar
 224 ./zzz 204e43b8a1185621ca55a94839582e6f+67108864 0:999:zzz
 225 ./foo 204e43b8a1185621ca55a94839582e6f+67108864 3:3:bar"""
 226         self.assertEqual(arvados.CollectionReader(m5).manifest_text(),
 227                          """./foo 204e43b8a1185621ca55a94839582e6f+67108864 0:6:bar
 228 ./zzz 204e43b8a1185621ca55a94839582e6f+67108864 0:999:zzz
 229 """)
 230
 231         with self.data_file('1000G_ref_manifest') as f6:
 232             m6 = f6.read()
 233             self.assertEqual(arvados.CollectionReader(m6).manifest_text(), m6)
 234
 235         with self.data_file('jlake_manifest') as f7:
 236             m7 = f7.read()
 237             self.assertEqual(arvados.CollectionReader(m7).manifest_text(), m7)
 238
 239         m8 = """./a\\040b\\040c 59ca0efa9f5633cb0371bbc0355478d8+13 0:13:hello\\040world.txt
 240 """
 241         self.assertEqual(arvados.CollectionReader(m8).manifest_text(), m8)
 242
 243     def test_locators_and_ranges(self):
 244         blocks2 = [['a', 10, 0],
 245                   ['b', 10, 10],
 246                   ['c', 10, 20],
 247                   ['d', 10, 30],
 248                   ['e', 10, 40],
 249                   ['f', 10, 50]]
 250
 251         self.assertEqual(arvados.locators_and_ranges(blocks2,  2,  2), [['a', 10, 2, 2]])
 252         self.assertEqual(arvados.locators_and_ranges(blocks2, 12, 2), [['b', 10, 2, 2]])
 253         self.assertEqual(arvados.locators_and_ranges(blocks2, 22, 2), [['c', 10, 2, 2]])
 254         self.assertEqual(arvados.locators_and_ranges(blocks2, 32, 2), [['d', 10, 2, 2]])
 255         self.assertEqual(arvados.locators_and_ranges(blocks2, 42, 2), [['e', 10, 2, 2]])
 256         self.assertEqual(arvados.locators_and_ranges(blocks2, 52, 2), [['f', 10, 2, 2]])
 257         self.assertEqual(arvados.locators_and_ranges(blocks2, 62, 2), [])
 258         self.assertEqual(arvados.locators_and_ranges(blocks2, -2, 2), [])
 259
 260         self.assertEqual(arvados.locators_and_ranges(blocks2,  0,  2), [['a', 10, 0, 2]])
 261         self.assertEqual(arvados.locators_and_ranges(blocks2, 10, 2), [['b', 10, 0, 2]])
 262         self.assertEqual(arvados.locators_and_ranges(blocks2, 20, 2), [['c', 10, 0, 2]])
 263         self.assertEqual(arvados.locators_and_ranges(blocks2, 30, 2), [['d', 10, 0, 2]])
 264         self.assertEqual(arvados.locators_and_ranges(blocks2, 40, 2), [['e', 10, 0, 2]])
 265         self.assertEqual(arvados.locators_and_ranges(blocks2, 50, 2), [['f', 10, 0, 2]])
 266         self.assertEqual(arvados.locators_and_ranges(blocks2, 60, 2), [])
 267         self.assertEqual(arvados.locators_and_ranges(blocks2, -2, 2), [])
 268
 269         self.assertEqual(arvados.locators_and_ranges(blocks2,  9,  2), [['a', 10, 9, 1], ['b', 10, 0, 1]])
 270         self.assertEqual(arvados.locators_and_ranges(blocks2, 19, 2), [['b', 10, 9, 1], ['c', 10, 0, 1]])
 271         self.assertEqual(arvados.locators_and_ranges(blocks2, 29, 2), [['c', 10, 9, 1], ['d', 10, 0, 1]])
 272         self.assertEqual(arvados.locators_and_ranges(blocks2, 39, 2), [['d', 10, 9, 1], ['e', 10, 0, 1]])
 273         self.assertEqual(arvados.locators_and_ranges(blocks2, 49, 2), [['e', 10, 9, 1], ['f', 10, 0, 1]])
 274         self.assertEqual(arvados.locators_and_ranges(blocks2, 59, 2), [['f', 10, 9, 1]])
 275
 276
 277         blocks3 = [['a', 10, 0],
 278                   ['b', 10, 10],
 279                   ['c', 10, 20],
 280                   ['d', 10, 30],
 281                   ['e', 10, 40],
 282                   ['f', 10, 50],
 283                   ['g', 10, 60]]
 284
 285         self.assertEqual(arvados.locators_and_ranges(blocks3,  2,  2), [['a', 10, 2, 2]])
 286         self.assertEqual(arvados.locators_and_ranges(blocks3, 12, 2), [['b', 10, 2, 2]])
 287         self.assertEqual(arvados.locators_and_ranges(blocks3, 22, 2), [['c', 10, 2, 2]])
 288         self.assertEqual(arvados.locators_and_ranges(blocks3, 32, 2), [['d', 10, 2, 2]])
 289         self.assertEqual(arvados.locators_and_ranges(blocks3, 42, 2), [['e', 10, 2, 2]])
 290         self.assertEqual(arvados.locators_and_ranges(blocks3, 52, 2), [['f', 10, 2, 2]])
 291         self.assertEqual(arvados.locators_and_ranges(blocks3, 62, 2), [['g', 10, 2, 2]])
 292
 293
 294         blocks = [['a', 10, 0],
 295                   ['b', 15, 10],
 296                   ['c', 5, 25]]
 297         self.assertEqual(arvados.locators_and_ranges(blocks, 1, 0), [])
 298         self.assertEqual(arvados.locators_and_ranges(blocks, 0, 5), [['a', 10, 0, 5]])
 299         self.assertEqual(arvados.locators_and_ranges(blocks, 3, 5), [['a', 10, 3, 5]])
 300         self.assertEqual(arvados.locators_and_ranges(blocks, 0, 10), [['a', 10, 0, 10]])
 301
 302         self.assertEqual(arvados.locators_and_ranges(blocks, 0, 11), [['a', 10, 0, 10],
 303                                                                       ['b', 15, 0, 1]])
 304         self.assertEqual(arvados.locators_and_ranges(blocks, 1, 11), [['a', 10, 1, 9],
 305                                                                       ['b', 15, 0, 2]])
 306         self.assertEqual(arvados.locators_and_ranges(blocks, 0, 25), [['a', 10, 0, 10],
 307                                                                       ['b', 15, 0, 15]])
 308
 309         self.assertEqual(arvados.locators_and_ranges(blocks, 0, 30), [['a', 10, 0, 10],
 310                                                                       ['b', 15, 0, 15],
 311                                                                       ['c', 5, 0, 5]])
 312         self.assertEqual(arvados.locators_and_ranges(blocks, 1, 30), [['a', 10, 1, 9],
 313                                                                       ['b', 15, 0, 15],
 314                                                                       ['c', 5, 0, 5]])
 315         self.assertEqual(arvados.locators_and_ranges(blocks, 0, 31), [['a', 10, 0, 10],
 316                                                                       ['b', 15, 0, 15],
 317                                                                       ['c', 5, 0, 5]])
 318
 319         self.assertEqual(arvados.locators_and_ranges(blocks, 15, 5), [['b', 15, 5, 5]])
 320
 321         self.assertEqual(arvados.locators_and_ranges(blocks, 8, 17), [['a', 10, 8, 2],
 322                                                                       ['b', 15, 0, 15]])
 323
 324         self.assertEqual(arvados.locators_and_ranges(blocks, 8, 20), [['a', 10, 8, 2],
 325                                                                       ['b', 15, 0, 15],
 326                                                                       ['c', 5, 0, 3]])
 327
 328         self.assertEqual(arvados.locators_and_ranges(blocks, 26, 2), [['c', 5, 1, 2]])
 329
 330         self.assertEqual(arvados.locators_and_ranges(blocks, 9, 15), [['a', 10, 9, 1],
 331                                                                       ['b', 15, 0, 14]])
 332         self.assertEqual(arvados.locators_and_ranges(blocks, 10, 15), [['b', 15, 0, 15]])
 333         self.assertEqual(arvados.locators_and_ranges(blocks, 11, 15), [['b', 15, 1, 14],
 334                                                                        ['c', 5, 0, 1]])
 335
 336     class MockStreamReader(object):
 337         def __init__(self, content):
 338             self.content = content
 339
 340         def readfrom(self, start, size):
 341             return self.content[start:start+size]
 342
 343     def test_file_stream(self):
 344         content = 'abcdefghijklmnopqrstuvwxyz0123456789'
 345         msr = self.MockStreamReader(content)
 346         segments = [[0, 10, 0],
 347                     [10, 15, 10],
 348                     [25, 5, 25]]
 349
 350         sfr = arvados.StreamFileReader(msr, segments, "test")
 351
 352         self.assertEqual(sfr.name(), "test")
 353         self.assertEqual(sfr.size(), 30)
 354
 355         self.assertEqual(sfr.readfrom(0, 30), content[0:30])
 356         self.assertEqual(sfr.readfrom(2, 30), content[2:30])
 357
 358         self.assertEqual(sfr.readfrom(2, 8), content[2:10])
 359         self.assertEqual(sfr.readfrom(0, 10), content[0:10])
 360
 361         self.assertEqual(sfr.tell(), 0)
 362         self.assertEqual(sfr.read(5), content[0:5])
 363         self.assertEqual(sfr.tell(), 5)
 364         self.assertEqual(sfr.read(5), content[5:10])
 365         self.assertEqual(sfr.tell(), 10)
 366         self.assertEqual(sfr.read(5), content[10:15])
 367         self.assertEqual(sfr.tell(), 15)
 368         self.assertEqual(sfr.read(5), content[15:20])
 369         self.assertEqual(sfr.tell(), 20)
 370         self.assertEqual(sfr.read(5), content[20:25])
 371         self.assertEqual(sfr.tell(), 25)
 372         self.assertEqual(sfr.read(5), content[25:30])
 373         self.assertEqual(sfr.tell(), 30)
 374         self.assertEqual(sfr.read(5), '')
 375         self.assertEqual(sfr.tell(), 30)
 376
 377         segments = [[26, 10, 0],
 378                     [0, 15, 10],
 379                     [15, 5, 25]]
 380
 381         sfr = arvados.StreamFileReader(msr, segments, "test")
 382
 383         self.assertEqual(sfr.size(), 30)
 384
 385         self.assertEqual(sfr.readfrom(0, 30), content[26:36] + content[0:20])
 386         self.assertEqual(sfr.readfrom(2, 30), content[28:36] + content[0:20])
 387
 388         self.assertEqual(sfr.readfrom(2, 8), content[28:36])
 389         self.assertEqual(sfr.readfrom(0, 10), content[26:36])
 390
 391         self.assertEqual(sfr.tell(), 0)
 392         self.assertEqual(sfr.read(5), content[26:31])
 393         self.assertEqual(sfr.tell(), 5)
 394         self.assertEqual(sfr.read(5), content[31:36])
 395         self.assertEqual(sfr.tell(), 10)
 396         self.assertEqual(sfr.read(5), content[0:5])
 397         self.assertEqual(sfr.tell(), 15)
 398         self.assertEqual(sfr.read(5), content[5:10])
 399         self.assertEqual(sfr.tell(), 20)
 400         self.assertEqual(sfr.read(5), content[10:15])
 401         self.assertEqual(sfr.tell(), 25)
 402         self.assertEqual(sfr.read(5), content[15:20])
 403         self.assertEqual(sfr.tell(), 30)
 404         self.assertEqual(sfr.read(5), '')
 405         self.assertEqual(sfr.tell(), 30)
 406
 407
 408     class MockKeep(object):
 409         def __init__(self, content):
 410             self.content = content
 411
 412         def get(self, locator):
 413             return self.content[locator]
 414
 415     def test_stream_reader(self):
 416         keepblocks = {'aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa+10': 'abcdefghij',
 417                       'bbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbb+15': 'klmnopqrstuvwxy',
 418                       'cccccccccccccccccccccccccccccccc+5': 'z0123'}
 419         mk = self.MockKeep(keepblocks)
 420
 421         sr = arvados.StreamReader([".", "aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa+10", "bbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbb+15", "cccccccccccccccccccccccccccccccc+5", "0:30:foo"], mk)
 422
 423         content = 'abcdefghijklmnopqrstuvwxyz0123456789'
 424
 425         self.assertEqual(sr.readfrom(0, 30), content[0:30])
 426         self.assertEqual(sr.readfrom(2, 30), content[2:30])
 427
 428         self.assertEqual(sr.readfrom(2, 8), content[2:10])
 429         self.assertEqual(sr.readfrom(0, 10), content[0:10])
 430
 431         self.assertEqual(sr.readfrom(0, 5), content[0:5])
 432         self.assertEqual(sr.readfrom(5, 5), content[5:10])
 433         self.assertEqual(sr.readfrom(10, 5), content[10:15])
 434         self.assertEqual(sr.readfrom(15, 5), content[15:20])
 435         self.assertEqual(sr.readfrom(20, 5), content[20:25])
 436         self.assertEqual(sr.readfrom(25, 5), content[25:30])
 437         self.assertEqual(sr.readfrom(30, 5), '')
 438
 439     def test_extract_file(self):
 440         m1 = """. 5348b82a029fd9e971a811ce1f71360b+43 0:43:md5sum.txt
 441 . 085c37f02916da1cad16f93c54d899b7+41 0:41:md6sum.txt
 442 . 8b22da26f9f433dea0a10e5ec66d73ba+43 0:43:md7sum.txt
 443 . 085c37f02916da1cad16f93c54d899b7+41 5348b82a029fd9e971a811ce1f71360b+43 8b22da26f9f433dea0a10e5ec66d73ba+43 47:80:md8sum.txt
 444 . 085c37f02916da1cad16f93c54d899b7+41 5348b82a029fd9e971a811ce1f71360b+43 8b22da26f9f433dea0a10e5ec66d73ba+43 40:80:md9sum.txt"""
 445
 446         m2 = arvados.CollectionReader(m1)
 447
 448         self.assertEqual(m2.manifest_text(),
 449                          ". 5348b82a029fd9e971a811ce1f71360b+43 085c37f02916da1cad16f93c54d899b7+41 8b22da26f9f433dea0a10e5ec66d73ba+43 0:43:md5sum.txt 43:41:md6sum.txt 84:43:md7sum.txt 6:37:md8sum.txt 84:43:md8sum.txt 83:1:md9sum.txt 0:43:md9sum.txt 84:36:md9sum.txt\n")
 450
 451         self.assertEqual(arvados.CollectionReader(m1).all_streams()[0].files()['md5sum.txt'].as_manifest(),
 452                          ". 5348b82a029fd9e971a811ce1f71360b+43 0:43:md5sum.txt\n")
 453         self.assertEqual(arvados.CollectionReader(m1).all_streams()[0].files()['md6sum.txt'].as_manifest(),
 454                          ". 085c37f02916da1cad16f93c54d899b7+41 0:41:md6sum.txt\n")
 455         self.assertEqual(arvados.CollectionReader(m1).all_streams()[0].files()['md7sum.txt'].as_manifest(),
 456                          ". 8b22da26f9f433dea0a10e5ec66d73ba+43 0:43:md7sum.txt\n")
 457         self.assertEqual(arvados.CollectionReader(m1).all_streams()[0].files()['md9sum.txt'].as_manifest(),
 458                          ". 085c37f02916da1cad16f93c54d899b7+41 5348b82a029fd9e971a811ce1f71360b+43 8b22da26f9f433dea0a10e5ec66d73ba+43 40:80:md9sum.txt\n")
 459
 460     def test_write_directory_tree(self):
 461         cwriter = arvados.CollectionWriter()
 462         cwriter.write_directory_tree(self.build_directory_tree(
 463                 ['basefile', 'subdir/subfile']))
 464         self.assertEqual(cwriter.manifest_text(),
 465                          """. c5110c5ac93202d8e0f9e381f22bac0f+8 0:8:basefile
 466 ./subdir 1ca4dec89403084bf282ad31e6cf7972+14 0:14:subfile\n""")
 467
 468     def test_write_named_directory_tree(self):
 469         cwriter = arvados.CollectionWriter()
 470         cwriter.write_directory_tree(self.build_directory_tree(
 471                 ['basefile', 'subdir/subfile']), 'root')
 472         self.assertEqual(
 473             cwriter.manifest_text(),
 474             """./root c5110c5ac93202d8e0f9e381f22bac0f+8 0:8:basefile
 475 ./root/subdir 1ca4dec89403084bf282ad31e6cf7972+14 0:14:subfile\n""")
 476
 477     def test_write_directory_tree_in_one_stream(self):
 478         cwriter = arvados.CollectionWriter()
 479         cwriter.write_directory_tree(self.build_directory_tree(
 480                 ['basefile', 'subdir/subfile']), max_manifest_depth=0)
 481         self.assertEqual(cwriter.manifest_text(),
 482                          """. 4ace875ffdc6824a04950f06858f4465+22 0:8:basefile
 483 ./subdir 4ace875ffdc6824a04950f06858f4465+22 8:14:subfile\n""")
 484
 485     def test_write_directory_tree_with_limited_recursion(self):
 486         cwriter = arvados.CollectionWriter()
 487         cwriter.write_directory_tree(
 488             self.build_directory_tree(['f1', 'd1/f2', 'd1/d2/f3']),
 489             max_manifest_depth=1)
 490         self.assertEqual(cwriter.manifest_text(),
 491                          """. bd19836ddb62c11c55ab251ccaca5645+2 0:2:f1
 492 ./d1 50170217e5b04312024aa5cd42934494+13 8:5:f2
 493 ./d1/d2 50170217e5b04312024aa5cd42934494+13 0:8:f3\n""")
 494
 495     def test_write_one_file(self):
 496         cwriter = arvados.CollectionWriter()
 497         with self.make_test_file() as testfile:
 498             cwriter.write_file(testfile.name)
 499             self.assertEqual(
 500                 cwriter.manifest_text(),
 501                 ". 098f6bcd4621d373cade4e832627b4f6+4 0:4:{}\n".format(
 502                     os.path.basename(testfile.name)))
 503
 504     def test_write_named_file(self):
 505         cwriter = arvados.CollectionWriter()
 506         with self.make_test_file() as testfile:
 507             cwriter.write_file(testfile.name, 'foo')
 508             self.assertEqual(cwriter.manifest_text(),
 509                              ". 098f6bcd4621d373cade4e832627b4f6+4 0:4:foo\n")
 510
 511     def test_write_multiple_files(self):
 512         cwriter = arvados.CollectionWriter()
 513         for letter in 'ABC':
 514             with self.make_test_file(letter) as testfile:
 515                 cwriter.write_file(testfile.name, letter)
 516         self.assertEqual(
 517             cwriter.manifest_text(),
 518             ". 902fbdd2b1df0c4f70b4a5d23525e932+3 0:1:A 1:1:B 2:1:C\n")
 519
 520     def test_basic_resume(self):
 521         cwriter = TestResumableWriter()
 522         with self.make_test_file() as testfile:
 523             cwriter.write_file(testfile.name, 'test')
 524             resumed = TestResumableWriter.from_state(cwriter.current_state())
 525         self.assertEquals(cwriter.manifest_text(), resumed.manifest_text(),
 526                           "resumed CollectionWriter had different manifest")
 527
 528     def test_resume_fails_when_missing_dependency(self):
 529         cwriter = TestResumableWriter()
 530         with self.make_test_file() as testfile:
 531             cwriter.write_file(testfile.name, 'test')
 532         self.assertRaises(arvados.errors.StaleWriterStateError,
 533                           TestResumableWriter.from_state,
 534                           cwriter.current_state())
 535
 536     def test_resume_fails_when_dependency_mtime_changed(self):
 537         cwriter = TestResumableWriter()
 538         with self.make_test_file() as testfile:
 539             cwriter.write_file(testfile.name, 'test')
 540             os.utime(testfile.name, (0, 0))
 541             self.assertRaises(arvados.errors.StaleWriterStateError,
 542                               TestResumableWriter.from_state,
 543                               cwriter.current_state())
 544
 545     def test_resume_fails_when_dependency_is_nonfile(self):
 546         cwriter = TestResumableWriter()
 547         cwriter.write_file('/dev/null', 'empty')
 548         self.assertRaises(arvados.errors.StaleWriterStateError,
 549                           TestResumableWriter.from_state,
 550                           cwriter.current_state())
 551
 552     def test_resume_fails_when_dependency_size_changed(self):
 553         cwriter = TestResumableWriter()
 554         with self.make_test_file() as testfile:
 555             cwriter.write_file(testfile.name, 'test')
 556             orig_mtime = os.fstat(testfile.fileno()).st_mtime
 557             testfile.write('extra')
 558             testfile.flush()
 559             os.utime(testfile.name, (orig_mtime, orig_mtime))
 560             self.assertRaises(arvados.errors.StaleWriterStateError,
 561                               TestResumableWriter.from_state,
 562                               cwriter.current_state())
 563
 564     def test_resume_fails_with_expired_locator(self):
 565         cwriter = TestResumableWriter()
 566         state = cwriter.current_state()
 567         # Add an expired locator to the state.
 568         state['_current_stream_locators'].append(''.join([
 569                     'a' * 32, '+A', 'b' * 40, '@', '10000000']))
 570         self.assertRaises(arvados.errors.StaleWriterStateError,
 571                           TestResumableWriter.from_state, state)
 572
 573     def test_arbitrary_objects_not_resumable(self):
 574         cwriter = TestResumableWriter()
 575         with open('/dev/null') as badfile:
 576             self.assertRaises(arvados.errors.AssertionError,
 577                               cwriter.write_file, badfile)
 578
 579     def test_arbitrary_writes_not_resumable(self):
 580         cwriter = TestResumableWriter()
 581         self.assertRaises(arvados.errors.AssertionError,
 582                           cwriter.write, "badtext")
 583
 584
 585 if __name__ == '__main__':
 586     unittest.main()