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 __init__(self):
  20         self.saved_states = []
  21         return super(TestResumableWriter, self).__init__()
  22
  23     def checkpoint_state(self):
  24         self.saved_states.append(self.dump_state(copy.deepcopy))
  25
  26     def last_state(self):
  27         assert self.saved_states, "resumable writer did not save any state"
  28         return self.saved_states[-1]
  29
  30
  31 class ArvadosCollectionsTest(ArvadosKeepLocalStoreTestCase):
  32     def write_foo_bar_baz(self):
  33         cw = arvados.CollectionWriter()
  34         self.assertEqual(cw.current_stream_name(), '.',
  35                          'current_stream_name() should be "." now')
  36         cw.set_current_file_name('foo.txt')
  37         cw.write('foo')
  38         self.assertEqual(cw.current_file_name(), 'foo.txt',
  39                          'current_file_name() should be foo.txt now')
  40         cw.start_new_file('bar.txt')
  41         cw.write('bar')
  42         cw.start_new_stream('baz')
  43         cw.write('baz')
  44         cw.set_current_file_name('baz.txt')
  45         return cw.finish()
  46
  47     def test_keep_local_store(self):
  48         self.assertEqual(arvados.Keep.put('foo'), 'acbd18db4cc2f85cedef654fccc4a4d8+3', 'wrong md5 hash from Keep.put')
  49         self.assertEqual(arvados.Keep.get('acbd18db4cc2f85cedef654fccc4a4d8+3'), 'foo', 'wrong data from Keep.get')
  50
  51     def test_local_collection_writer(self):
  52         self.assertEqual(self.write_foo_bar_baz(),
  53                          'd6c3b8e571f1b81ebb150a45ed06c884+114',
  54                          "wrong locator hash for files foo, bar, baz")
  55
  56     def test_local_collection_reader(self):
  57         self.write_foo_bar_baz()
  58         cr = arvados.CollectionReader('d6c3b8e571f1b81ebb150a45ed06c884+114+Xzizzle')
  59         got = []
  60         for s in cr.all_streams():
  61             for f in s.all_files():
  62                 got += [[f.size(), f.stream_name(), f.name(), f.read(2**26)]]
  63         expected = [[3, '.', 'bar.txt', 'bar'],
  64                     [3, '.', 'foo.txt', 'foo'],
  65                     [3, './baz', 'baz.txt', 'baz']]
  66         self.assertEqual(got,
  67                          expected)
  68         stream0 = cr.all_streams()[0]
  69         self.assertEqual(stream0.readfrom(0, 0),
  70                          '',
  71                          'reading zero bytes should have returned empty string')
  72         self.assertEqual(stream0.readfrom(0, 2**26),
  73                          'foobar',
  74                          'reading entire stream failed')
  75         self.assertEqual(stream0.readfrom(2**26, 0),
  76                          '',
  77                          'reading zero bytes should have returned empty string')
  78
  79     def _test_subset(self, collection, expected):
  80         cr = arvados.CollectionReader(collection)
  81         for s in cr.all_streams():
  82             for ex in expected:
  83                 if ex[0] == s:
  84                     f = s.files()[ex[2]]
  85                     got = [f.size(), f.stream_name(), f.name(), "".join(f.readall(2**26))]
  86                     self.assertEqual(got,
  87                                      ex,
  88                                      'all_files|as_manifest did not preserve manifest contents: got %s expected %s' % (got, ex))
  89
  90     def test_collection_manifest_subset(self):
  91         self.write_foo_bar_baz()
  92         self._test_subset('d6c3b8e571f1b81ebb150a45ed06c884+114',
  93                           [[3, '.',     'bar.txt', 'bar'],
  94                            [3, '.',     'foo.txt', 'foo'],
  95                            [3, './baz', 'baz.txt', 'baz']])
  96         self._test_subset((". %s %s 0:3:foo.txt 3:3:bar.txt\n" %
  97                            (arvados.Keep.put("foo"),
  98                             arvados.Keep.put("bar"))),
  99                           [[3, '.', 'bar.txt', 'bar'],
 100                            [3, '.', 'foo.txt', 'foo']])
 101         self._test_subset((". %s %s 0:2:fo.txt 2:4:obar.txt\n" %
 102                            (arvados.Keep.put("foo"),
 103                             arvados.Keep.put("bar"))),
 104                           [[2, '.', 'fo.txt', 'fo'],
 105                            [4, '.', 'obar.txt', 'obar']])
 106         self._test_subset((". %s %s 0:2:fo.txt 2:0:zero.txt 2:2:ob.txt 4:2:ar.txt\n" %
 107                            (arvados.Keep.put("foo"),
 108                             arvados.Keep.put("bar"))),
 109                           [[2, '.', 'ar.txt', 'ar'],
 110                            [2, '.', 'fo.txt', 'fo'],
 111                            [2, '.', 'ob.txt', 'ob'],
 112                            [0, '.', 'zero.txt', '']])
 113
 114     def _test_readline(self, what_in, what_out):
 115         cw = arvados.CollectionWriter()
 116         cw.start_new_file('test.txt')
 117         cw.write(what_in)
 118         test1 = cw.finish()
 119         cr = arvados.CollectionReader(test1)
 120         got = []
 121         for x in list(cr.all_files())[0].readlines():
 122             got += [x]
 123         self.assertEqual(got,
 124                          what_out,
 125                          "readlines did not split lines correctly: %s" % got)
 126
 127     def test_collection_readline(self):
 128         self._test_readline("\na\nbcd\n\nefg\nz",
 129                             ["\n", "a\n", "bcd\n", "\n", "efg\n", "z"])
 130         self._test_readline("ab\ncd\n",
 131                             ["ab\n", "cd\n"])
 132
 133     def test_collection_empty_file(self):
 134         cw = arvados.CollectionWriter()
 135         cw.start_new_file('zero.txt')
 136         cw.write('')
 137
 138         self.assertEqual(cw.manifest_text(), ". 0:0:zero.txt\n")
 139         self.check_manifest_file_sizes(cw.manifest_text(), [0])
 140         cw = arvados.CollectionWriter()
 141         cw.start_new_file('zero.txt')
 142         cw.write('')
 143         cw.start_new_file('one.txt')
 144         cw.write('1')
 145         cw.start_new_stream('foo')
 146         cw.start_new_file('zero.txt')
 147         cw.write('')
 148         self.check_manifest_file_sizes(cw.manifest_text(), [1,0,0])
 149
 150     def check_manifest_file_sizes(self, manifest_text, expect_sizes):
 151         cr = arvados.CollectionReader(manifest_text)
 152         got_sizes = []
 153         for f in cr.all_files():
 154             got_sizes += [f.size()]
 155         self.assertEqual(got_sizes, expect_sizes, "got wrong file sizes %s, expected %s" % (got_sizes, expect_sizes))
 156
 157     def test_collection_bz2_decompression(self):
 158         n_lines_in = 2**18
 159         data_in = "abc\n"
 160         for x in xrange(0, 18):
 161             data_in += data_in
 162         compressed_data_in = bz2.compress(data_in)
 163         cw = arvados.CollectionWriter()
 164         cw.start_new_file('test.bz2')
 165         cw.write(compressed_data_in)
 166         bz2_manifest = cw.manifest_text()
 167
 168         cr = arvados.CollectionReader(bz2_manifest)
 169
 170         got = 0
 171         for x in list(cr.all_files())[0].readlines():
 172             self.assertEqual(x, "abc\n", "decompression returned wrong data: %s" % x)
 173             got += 1
 174         self.assertEqual(got,
 175                          n_lines_in,
 176                          "decompression returned %d lines instead of %d" % (got, n_lines_in))
 177
 178     def test_collection_gzip_decompression(self):
 179         n_lines_in = 2**18
 180         data_in = "abc\n"
 181         for x in xrange(0, 18):
 182             data_in += data_in
 183         p = subprocess.Popen(["gzip", "-1cn"],
 184                              stdout=subprocess.PIPE,
 185                              stdin=subprocess.PIPE,
 186                              stderr=subprocess.PIPE,
 187                              shell=False, close_fds=True)
 188         compressed_data_in, stderrdata = p.communicate(data_in)
 189
 190         cw = arvados.CollectionWriter()
 191         cw.start_new_file('test.gz')
 192         cw.write(compressed_data_in)
 193         gzip_manifest = cw.manifest_text()
 194
 195         cr = arvados.CollectionReader(gzip_manifest)
 196         got = 0
 197         for x in list(cr.all_files())[0].readlines():
 198             self.assertEqual(x, "abc\n", "decompression returned wrong data: %s" % x)
 199             got += 1
 200         self.assertEqual(got,
 201                          n_lines_in,
 202                          "decompression returned %d lines instead of %d" % (got, n_lines_in))
 203
 204     def test_normalized_collection(self):
 205         m1 = """. 5348b82a029fd9e971a811ce1f71360b+43 0:43:md5sum.txt
 206 . 085c37f02916da1cad16f93c54d899b7+41 0:41:md5sum.txt
 207 . 8b22da26f9f433dea0a10e5ec66d73ba+43 0:43:md5sum.txt"""
 208         self.assertEqual(arvados.CollectionReader(m1).manifest_text(),
 209                          """. 5348b82a029fd9e971a811ce1f71360b+43 085c37f02916da1cad16f93c54d899b7+41 8b22da26f9f433dea0a10e5ec66d73ba+43 0:127:md5sum.txt
 210 """)
 211
 212         m2 = """. 204e43b8a1185621ca55a94839582e6f+67108864 b9677abbac956bd3e86b1deb28dfac03+67108864 fc15aff2a762b13f521baf042140acec+67108864 323d2a3ce20370c4ca1d3462a344f8fd+25885655 0:227212247:var-GS000016015-ASM.tsv.bz2
 213 """
 214         self.assertEqual(arvados.CollectionReader(m2).manifest_text(), m2)
 215
 216         m3 = """. 5348b82a029fd9e971a811ce1f71360b+43 3:40:md5sum.txt
 217 . 085c37f02916da1cad16f93c54d899b7+41 0:41:md5sum.txt
 218 . 8b22da26f9f433dea0a10e5ec66d73ba+43 0:43:md5sum.txt"""
 219         self.assertEqual(arvados.CollectionReader(m3).manifest_text(),
 220                          """. 5348b82a029fd9e971a811ce1f71360b+43 085c37f02916da1cad16f93c54d899b7+41 8b22da26f9f433dea0a10e5ec66d73ba+43 3:124:md5sum.txt
 221 """)
 222
 223         m4 = """. 204e43b8a1185621ca55a94839582e6f+67108864 0:3:foo/bar
 224 ./zzz 204e43b8a1185621ca55a94839582e6f+67108864 0:999:zzz
 225 ./foo 323d2a3ce20370c4ca1d3462a344f8fd+25885655 0:3:bar"""
 226         self.assertEqual(arvados.CollectionReader(m4).manifest_text(),
 227                          """./foo 204e43b8a1185621ca55a94839582e6f+67108864 323d2a3ce20370c4ca1d3462a344f8fd+25885655 0:3:bar 67108864:3:bar
 228 ./zzz 204e43b8a1185621ca55a94839582e6f+67108864 0:999:zzz
 229 """)
 230
 231         m5 = """. 204e43b8a1185621ca55a94839582e6f+67108864 0:3:foo/bar
 232 ./zzz 204e43b8a1185621ca55a94839582e6f+67108864 0:999:zzz
 233 ./foo 204e43b8a1185621ca55a94839582e6f+67108864 3:3:bar"""
 234         self.assertEqual(arvados.CollectionReader(m5).manifest_text(),
 235                          """./foo 204e43b8a1185621ca55a94839582e6f+67108864 0:6:bar
 236 ./zzz 204e43b8a1185621ca55a94839582e6f+67108864 0:999:zzz
 237 """)
 238
 239         with self.data_file('1000G_ref_manifest') as f6:
 240             m6 = f6.read()
 241             self.assertEqual(arvados.CollectionReader(m6).manifest_text(), m6)
 242
 243         with self.data_file('jlake_manifest') as f7:
 244             m7 = f7.read()
 245             self.assertEqual(arvados.CollectionReader(m7).manifest_text(), m7)
 246
 247         m8 = """./a\\040b\\040c 59ca0efa9f5633cb0371bbc0355478d8+13 0:13:hello\\040world.txt
 248 """
 249         self.assertEqual(arvados.CollectionReader(m8).manifest_text(), m8)
 250
 251     def test_locators_and_ranges(self):
 252         blocks2 = [['a', 10, 0],
 253                   ['b', 10, 10],
 254                   ['c', 10, 20],
 255                   ['d', 10, 30],
 256                   ['e', 10, 40],
 257                   ['f', 10, 50]]
 258
 259         self.assertEqual(arvados.locators_and_ranges(blocks2,  2,  2), [['a', 10, 2, 2]])
 260         self.assertEqual(arvados.locators_and_ranges(blocks2, 12, 2), [['b', 10, 2, 2]])
 261         self.assertEqual(arvados.locators_and_ranges(blocks2, 22, 2), [['c', 10, 2, 2]])
 262         self.assertEqual(arvados.locators_and_ranges(blocks2, 32, 2), [['d', 10, 2, 2]])
 263         self.assertEqual(arvados.locators_and_ranges(blocks2, 42, 2), [['e', 10, 2, 2]])
 264         self.assertEqual(arvados.locators_and_ranges(blocks2, 52, 2), [['f', 10, 2, 2]])
 265         self.assertEqual(arvados.locators_and_ranges(blocks2, 62, 2), [])
 266         self.assertEqual(arvados.locators_and_ranges(blocks2, -2, 2), [])
 267
 268         self.assertEqual(arvados.locators_and_ranges(blocks2,  0,  2), [['a', 10, 0, 2]])
 269         self.assertEqual(arvados.locators_and_ranges(blocks2, 10, 2), [['b', 10, 0, 2]])
 270         self.assertEqual(arvados.locators_and_ranges(blocks2, 20, 2), [['c', 10, 0, 2]])
 271         self.assertEqual(arvados.locators_and_ranges(blocks2, 30, 2), [['d', 10, 0, 2]])
 272         self.assertEqual(arvados.locators_and_ranges(blocks2, 40, 2), [['e', 10, 0, 2]])
 273         self.assertEqual(arvados.locators_and_ranges(blocks2, 50, 2), [['f', 10, 0, 2]])
 274         self.assertEqual(arvados.locators_and_ranges(blocks2, 60, 2), [])
 275         self.assertEqual(arvados.locators_and_ranges(blocks2, -2, 2), [])
 276
 277         self.assertEqual(arvados.locators_and_ranges(blocks2,  9,  2), [['a', 10, 9, 1], ['b', 10, 0, 1]])
 278         self.assertEqual(arvados.locators_and_ranges(blocks2, 19, 2), [['b', 10, 9, 1], ['c', 10, 0, 1]])
 279         self.assertEqual(arvados.locators_and_ranges(blocks2, 29, 2), [['c', 10, 9, 1], ['d', 10, 0, 1]])
 280         self.assertEqual(arvados.locators_and_ranges(blocks2, 39, 2), [['d', 10, 9, 1], ['e', 10, 0, 1]])
 281         self.assertEqual(arvados.locators_and_ranges(blocks2, 49, 2), [['e', 10, 9, 1], ['f', 10, 0, 1]])
 282         self.assertEqual(arvados.locators_and_ranges(blocks2, 59, 2), [['f', 10, 9, 1]])
 283
 284
 285         blocks3 = [['a', 10, 0],
 286                   ['b', 10, 10],
 287                   ['c', 10, 20],
 288                   ['d', 10, 30],
 289                   ['e', 10, 40],
 290                   ['f', 10, 50],
 291                   ['g', 10, 60]]
 292
 293         self.assertEqual(arvados.locators_and_ranges(blocks3,  2,  2), [['a', 10, 2, 2]])
 294         self.assertEqual(arvados.locators_and_ranges(blocks3, 12, 2), [['b', 10, 2, 2]])
 295         self.assertEqual(arvados.locators_and_ranges(blocks3, 22, 2), [['c', 10, 2, 2]])
 296         self.assertEqual(arvados.locators_and_ranges(blocks3, 32, 2), [['d', 10, 2, 2]])
 297         self.assertEqual(arvados.locators_and_ranges(blocks3, 42, 2), [['e', 10, 2, 2]])
 298         self.assertEqual(arvados.locators_and_ranges(blocks3, 52, 2), [['f', 10, 2, 2]])
 299         self.assertEqual(arvados.locators_and_ranges(blocks3, 62, 2), [['g', 10, 2, 2]])
 300
 301
 302         blocks = [['a', 10, 0],
 303                   ['b', 15, 10],
 304                   ['c', 5, 25]]
 305         self.assertEqual(arvados.locators_and_ranges(blocks, 1, 0), [])
 306         self.assertEqual(arvados.locators_and_ranges(blocks, 0, 5), [['a', 10, 0, 5]])
 307         self.assertEqual(arvados.locators_and_ranges(blocks, 3, 5), [['a', 10, 3, 5]])
 308         self.assertEqual(arvados.locators_and_ranges(blocks, 0, 10), [['a', 10, 0, 10]])
 309
 310         self.assertEqual(arvados.locators_and_ranges(blocks, 0, 11), [['a', 10, 0, 10],
 311                                                                       ['b', 15, 0, 1]])
 312         self.assertEqual(arvados.locators_and_ranges(blocks, 1, 11), [['a', 10, 1, 9],
 313                                                                       ['b', 15, 0, 2]])
 314         self.assertEqual(arvados.locators_and_ranges(blocks, 0, 25), [['a', 10, 0, 10],
 315                                                                       ['b', 15, 0, 15]])
 316
 317         self.assertEqual(arvados.locators_and_ranges(blocks, 0, 30), [['a', 10, 0, 10],
 318                                                                       ['b', 15, 0, 15],
 319                                                                       ['c', 5, 0, 5]])
 320         self.assertEqual(arvados.locators_and_ranges(blocks, 1, 30), [['a', 10, 1, 9],
 321                                                                       ['b', 15, 0, 15],
 322                                                                       ['c', 5, 0, 5]])
 323         self.assertEqual(arvados.locators_and_ranges(blocks, 0, 31), [['a', 10, 0, 10],
 324                                                                       ['b', 15, 0, 15],
 325                                                                       ['c', 5, 0, 5]])
 326
 327         self.assertEqual(arvados.locators_and_ranges(blocks, 15, 5), [['b', 15, 5, 5]])
 328
 329         self.assertEqual(arvados.locators_and_ranges(blocks, 8, 17), [['a', 10, 8, 2],
 330                                                                       ['b', 15, 0, 15]])
 331
 332         self.assertEqual(arvados.locators_and_ranges(blocks, 8, 20), [['a', 10, 8, 2],
 333                                                                       ['b', 15, 0, 15],
 334                                                                       ['c', 5, 0, 3]])
 335
 336         self.assertEqual(arvados.locators_and_ranges(blocks, 26, 2), [['c', 5, 1, 2]])
 337
 338         self.assertEqual(arvados.locators_and_ranges(blocks, 9, 15), [['a', 10, 9, 1],
 339                                                                       ['b', 15, 0, 14]])
 340         self.assertEqual(arvados.locators_and_ranges(blocks, 10, 15), [['b', 15, 0, 15]])
 341         self.assertEqual(arvados.locators_and_ranges(blocks, 11, 15), [['b', 15, 1, 14],
 342                                                                        ['c', 5, 0, 1]])
 343
 344     class MockStreamReader(object):
 345         def __init__(self, content):
 346             self.content = content
 347
 348         def readfrom(self, start, size):
 349             return self.content[start:start+size]
 350
 351     def test_file_stream(self):
 352         content = 'abcdefghijklmnopqrstuvwxyz0123456789'
 353         msr = self.MockStreamReader(content)
 354         segments = [[0, 10, 0],
 355                     [10, 15, 10],
 356                     [25, 5, 25]]
 357
 358         sfr = arvados.StreamFileReader(msr, segments, "test")
 359
 360         self.assertEqual(sfr.name(), "test")
 361         self.assertEqual(sfr.size(), 30)
 362
 363         self.assertEqual(sfr.readfrom(0, 30), content[0:30])
 364         self.assertEqual(sfr.readfrom(2, 30), content[2:30])
 365
 366         self.assertEqual(sfr.readfrom(2, 8), content[2:10])
 367         self.assertEqual(sfr.readfrom(0, 10), content[0:10])
 368
 369         self.assertEqual(sfr.tell(), 0)
 370         self.assertEqual(sfr.read(5), content[0:5])
 371         self.assertEqual(sfr.tell(), 5)
 372         self.assertEqual(sfr.read(5), content[5:10])
 373         self.assertEqual(sfr.tell(), 10)
 374         self.assertEqual(sfr.read(5), content[10:15])
 375         self.assertEqual(sfr.tell(), 15)
 376         self.assertEqual(sfr.read(5), content[15:20])
 377         self.assertEqual(sfr.tell(), 20)
 378         self.assertEqual(sfr.read(5), content[20:25])
 379         self.assertEqual(sfr.tell(), 25)
 380         self.assertEqual(sfr.read(5), content[25:30])
 381         self.assertEqual(sfr.tell(), 30)
 382         self.assertEqual(sfr.read(5), '')
 383         self.assertEqual(sfr.tell(), 30)
 384
 385         segments = [[26, 10, 0],
 386                     [0, 15, 10],
 387                     [15, 5, 25]]
 388
 389         sfr = arvados.StreamFileReader(msr, segments, "test")
 390
 391         self.assertEqual(sfr.size(), 30)
 392
 393         self.assertEqual(sfr.readfrom(0, 30), content[26:36] + content[0:20])
 394         self.assertEqual(sfr.readfrom(2, 30), content[28:36] + content[0:20])
 395
 396         self.assertEqual(sfr.readfrom(2, 8), content[28:36])
 397         self.assertEqual(sfr.readfrom(0, 10), content[26:36])
 398
 399         self.assertEqual(sfr.tell(), 0)
 400         self.assertEqual(sfr.read(5), content[26:31])
 401         self.assertEqual(sfr.tell(), 5)
 402         self.assertEqual(sfr.read(5), content[31:36])
 403         self.assertEqual(sfr.tell(), 10)
 404         self.assertEqual(sfr.read(5), content[0:5])
 405         self.assertEqual(sfr.tell(), 15)
 406         self.assertEqual(sfr.read(5), content[5:10])
 407         self.assertEqual(sfr.tell(), 20)
 408         self.assertEqual(sfr.read(5), content[10:15])
 409         self.assertEqual(sfr.tell(), 25)
 410         self.assertEqual(sfr.read(5), content[15:20])
 411         self.assertEqual(sfr.tell(), 30)
 412         self.assertEqual(sfr.read(5), '')
 413         self.assertEqual(sfr.tell(), 30)
 414
 415
 416     class MockKeep(object):
 417         def __init__(self, content):
 418             self.content = content
 419
 420         def get(self, locator):
 421             return self.content[locator]
 422
 423     def test_stream_reader(self):
 424         keepblocks = {'aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa+10': 'abcdefghij',
 425                       'bbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbb+15': 'klmnopqrstuvwxy',
 426                       'cccccccccccccccccccccccccccccccc+5': 'z0123'}
 427         mk = self.MockKeep(keepblocks)
 428
 429         sr = arvados.StreamReader([".", "aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa+10", "bbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbb+15", "cccccccccccccccccccccccccccccccc+5", "0:30:foo"], mk)
 430
 431         content = 'abcdefghijklmnopqrstuvwxyz0123456789'
 432
 433         self.assertEqual(sr.readfrom(0, 30), content[0:30])
 434         self.assertEqual(sr.readfrom(2, 30), content[2:30])
 435
 436         self.assertEqual(sr.readfrom(2, 8), content[2:10])
 437         self.assertEqual(sr.readfrom(0, 10), content[0:10])
 438
 439         self.assertEqual(sr.readfrom(0, 5), content[0:5])
 440         self.assertEqual(sr.readfrom(5, 5), content[5:10])
 441         self.assertEqual(sr.readfrom(10, 5), content[10:15])
 442         self.assertEqual(sr.readfrom(15, 5), content[15:20])
 443         self.assertEqual(sr.readfrom(20, 5), content[20:25])
 444         self.assertEqual(sr.readfrom(25, 5), content[25:30])
 445         self.assertEqual(sr.readfrom(30, 5), '')
 446
 447     def test_extract_file(self):
 448         m1 = """. 5348b82a029fd9e971a811ce1f71360b+43 0:43:md5sum.txt
 449 . 085c37f02916da1cad16f93c54d899b7+41 0:41:md6sum.txt
 450 . 8b22da26f9f433dea0a10e5ec66d73ba+43 0:43:md7sum.txt
 451 . 085c37f02916da1cad16f93c54d899b7+41 5348b82a029fd9e971a811ce1f71360b+43 8b22da26f9f433dea0a10e5ec66d73ba+43 47:80:md8sum.txt
 452 . 085c37f02916da1cad16f93c54d899b7+41 5348b82a029fd9e971a811ce1f71360b+43 8b22da26f9f433dea0a10e5ec66d73ba+43 40:80:md9sum.txt"""
 453
 454         m2 = arvados.CollectionReader(m1)
 455
 456         self.assertEqual(m2.manifest_text(),
 457                          ". 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")
 458
 459         self.assertEqual(arvados.CollectionReader(m1).all_streams()[0].files()['md5sum.txt'].as_manifest(),
 460                          ". 5348b82a029fd9e971a811ce1f71360b+43 0:43:md5sum.txt\n")
 461         self.assertEqual(arvados.CollectionReader(m1).all_streams()[0].files()['md6sum.txt'].as_manifest(),
 462                          ". 085c37f02916da1cad16f93c54d899b7+41 0:41:md6sum.txt\n")
 463         self.assertEqual(arvados.CollectionReader(m1).all_streams()[0].files()['md7sum.txt'].as_manifest(),
 464                          ". 8b22da26f9f433dea0a10e5ec66d73ba+43 0:43:md7sum.txt\n")
 465         self.assertEqual(arvados.CollectionReader(m1).all_streams()[0].files()['md9sum.txt'].as_manifest(),
 466                          ". 085c37f02916da1cad16f93c54d899b7+41 5348b82a029fd9e971a811ce1f71360b+43 8b22da26f9f433dea0a10e5ec66d73ba+43 40:80:md9sum.txt\n")
 467
 468     def test_write_directory_tree(self):
 469         cwriter = arvados.CollectionWriter()
 470         cwriter.write_directory_tree(self.build_directory_tree(
 471                 ['basefile', 'subdir/subfile']))
 472         self.assertEqual(cwriter.manifest_text(),
 473                          """. c5110c5ac93202d8e0f9e381f22bac0f+8 0:8:basefile
 474 ./subdir 1ca4dec89403084bf282ad31e6cf7972+14 0:14:subfile\n""")
 475
 476     def test_write_named_directory_tree(self):
 477         cwriter = arvados.CollectionWriter()
 478         cwriter.write_directory_tree(self.build_directory_tree(
 479                 ['basefile', 'subdir/subfile']), 'root')
 480         self.assertEqual(
 481             cwriter.manifest_text(),
 482             """./root c5110c5ac93202d8e0f9e381f22bac0f+8 0:8:basefile
 483 ./root/subdir 1ca4dec89403084bf282ad31e6cf7972+14 0:14:subfile\n""")
 484
 485     def test_write_directory_tree_in_one_stream(self):
 486         cwriter = arvados.CollectionWriter()
 487         cwriter.write_directory_tree(self.build_directory_tree(
 488                 ['basefile', 'subdir/subfile']), max_manifest_depth=0)
 489         self.assertEqual(cwriter.manifest_text(),
 490                          """. 4ace875ffdc6824a04950f06858f4465+22 0:8:basefile
 491 ./subdir 4ace875ffdc6824a04950f06858f4465+22 8:14:subfile\n""")
 492
 493     def test_write_directory_tree_with_limited_recursion(self):
 494         cwriter = arvados.CollectionWriter()
 495         cwriter.write_directory_tree(
 496             self.build_directory_tree(['f1', 'd1/f2', 'd1/d2/f3']),
 497             max_manifest_depth=1)
 498         self.assertEqual(cwriter.manifest_text(),
 499                          """. bd19836ddb62c11c55ab251ccaca5645+2 0:2:f1
 500 ./d1 50170217e5b04312024aa5cd42934494+13 8:5:f2
 501 ./d1/d2 50170217e5b04312024aa5cd42934494+13 0:8:f3\n""")
 502
 503     def test_write_one_file(self):
 504         cwriter = arvados.CollectionWriter()
 505         with self.make_test_file() as testfile:
 506             cwriter.write_file(testfile.name)
 507             self.assertEqual(
 508                 cwriter.manifest_text(),
 509                 ". 098f6bcd4621d373cade4e832627b4f6+4 0:4:{}\n".format(
 510                     os.path.basename(testfile.name)))
 511
 512     def test_write_named_file(self):
 513         cwriter = arvados.CollectionWriter()
 514         with self.make_test_file() as testfile:
 515             cwriter.write_file(testfile.name, 'foo')
 516             self.assertEqual(cwriter.manifest_text(),
 517                              ". 098f6bcd4621d373cade4e832627b4f6+4 0:4:foo\n")
 518
 519     def test_write_multiple_files(self):
 520         cwriter = arvados.CollectionWriter()
 521         for letter in 'ABC':
 522             with self.make_test_file(letter) as testfile:
 523                 cwriter.write_file(testfile.name, letter)
 524         self.assertEqual(
 525             cwriter.manifest_text(),
 526             ". 902fbdd2b1df0c4f70b4a5d23525e932+3 0:1:A 1:1:B 2:1:C\n")
 527
 528     def test_checkpoint_after_put(self):
 529         cwriter = TestResumableWriter()
 530         with self.make_test_file(
 531               't' * (cwriter.KEEP_BLOCK_SIZE + 10)) as testfile:
 532             testpath = os.path.realpath(testfile.name)
 533             cwriter.write_file(testpath, 'test')
 534         for state in cwriter.saved_states:
 535             if state.get('_current_file') == (testpath,
 536                                               cwriter.KEEP_BLOCK_SIZE):
 537                 break
 538         else:
 539             self.fail("can't find state immediately after PUT to Keep")
 540         self.assertIn('d45107e93f9052fa88a82fc08bb1d316+1024',  # 't' * 1024
 541                       state['_current_stream_locators'])
 542
 543     def test_basic_resume(self):
 544         cwriter = TestResumableWriter()
 545         with self.make_test_file() as testfile:
 546             cwriter.write_file(testfile.name, 'test')
 547             last_state = cwriter.last_state()
 548             resumed = TestResumableWriter.from_state(last_state)
 549         self.assertEquals(cwriter.manifest_text(), resumed.manifest_text(),
 550                           "resumed CollectionWriter had different manifest")
 551
 552     def test_resume_fails_when_missing_dependency(self):
 553         cwriter = TestResumableWriter()
 554         with self.make_test_file() as testfile:
 555             cwriter.write_file(testfile.name, 'test')
 556         self.assertRaises(arvados.errors.StaleWriterStateError,
 557                           TestResumableWriter.from_state,
 558                           cwriter.last_state())
 559
 560     def test_resume_fails_when_dependency_mtime_changed(self):
 561         cwriter = TestResumableWriter()
 562         with self.make_test_file() as testfile:
 563             cwriter.write_file(testfile.name, 'test')
 564             os.utime(testfile.name, (0, 0))
 565             self.assertRaises(arvados.errors.StaleWriterStateError,
 566                               TestResumableWriter.from_state,
 567                               cwriter.last_state())
 568
 569     def test_resume_fails_when_dependency_is_nonfile(self):
 570         cwriter = TestResumableWriter()
 571         cwriter.write_file('/dev/null', 'empty')
 572         self.assertRaises(arvados.errors.StaleWriterStateError,
 573                           TestResumableWriter.from_state,
 574                           cwriter.last_state())
 575
 576     def test_resume_fails_when_dependency_size_changed(self):
 577         cwriter = TestResumableWriter()
 578         with self.make_test_file() as testfile:
 579             cwriter.write_file(testfile.name, 'test')
 580             orig_mtime = os.fstat(testfile.fileno()).st_mtime
 581             testfile.write('extra')
 582             testfile.flush()
 583             os.utime(testfile.name, (orig_mtime, orig_mtime))
 584             self.assertRaises(arvados.errors.StaleWriterStateError,
 585                               TestResumableWriter.from_state,
 586                               cwriter.last_state())
 587
 588     def test_successful_resumes(self):
 589         # FIXME: This is more of an integration test than a unit test.
 590         cwriter = TestResumableWriter()
 591         source_tree = self.build_directory_tree(['basefile', 'subdir/subfile'])
 592         with open(os.path.join(source_tree, 'long'), 'w') as longfile:
 593             longfile.write('t' * (cwriter.KEEP_BLOCK_SIZE + 10))
 594         cwriter.write_directory_tree(source_tree)
 595         # A state for each file, plus a fourth for mid-longfile.
 596         self.assertGreater(len(cwriter.saved_states), 3,
 597                            "CollectionWriter didn't save enough states to test")
 598
 599         for state in cwriter.saved_states:
 600             new_writer = TestResumableWriter.from_state(state)
 601             manifests = [writer.manifest_text()
 602                          for writer in (cwriter, new_writer)]
 603             self.assertEquals(
 604                 manifests[0], manifests[1],
 605                 "\n".join(["manifest mismatch after resuming from state:",
 606                            pprint.pformat(state), ""] + manifests))
 607
 608     def test_arbitrary_objects_not_resumable(self):
 609         cwriter = TestResumableWriter()
 610         with open('/dev/null') as badfile:
 611             self.assertRaises(arvados.errors.AssertionError,
 612                               cwriter.write_file, badfile)
 613
 614     def test_arbitrary_writes_not_resumable(self):
 615         cwriter = TestResumableWriter()
 616         self.assertRaises(arvados.errors.AssertionError,
 617                           cwriter.write, "badtext")
 618
 619
 620 if __name__ == '__main__':
 621     unittest.main()