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 errno
   9 import os
  10 import pprint
  11 import shutil
  12 import subprocess
  13 import sys
  14 import tempfile
  15 import unittest
  16
  17 class TestResumableWriter(arvados.ResumableCollectionWriter):
  18     KEEP_BLOCK_SIZE = 1024  # PUT to Keep every 1K.
  19
  20     def __init__(self):
  21         self.saved_states = []
  22         return super(TestResumableWriter, self).__init__()
  23
  24     def checkpoint_state(self):
  25         self.saved_states.append(self.dump_state(copy.deepcopy))
  26
  27     def last_state(self):
  28         assert self.saved_states, "resumable writer did not save any state"
  29         return self.saved_states[-1]
  30
  31
  32 class ArvadosCollectionsTest(unittest.TestCase):
  33     def _make_tmpdir(self):
  34         self._tempdirs.append(tempfile.mkdtemp())
  35         return self._tempdirs[-1]
  36
  37     def setUp(self):
  38         self._orig_keep_local_store = os.environ.get('KEEP_LOCAL_STORE')
  39         self._tempdirs = []
  40         os.environ['KEEP_LOCAL_STORE'] = self._make_tmpdir()
  41
  42     def tearDown(self):
  43         for workdir in self._tempdirs:
  44             shutil.rmtree(workdir, ignore_errors=True)
  45         if self._orig_keep_local_store is None:
  46             del os.environ['KEEP_LOCAL_STORE']
  47         else:
  48             os.environ['KEEP_LOCAL_STORE'] = self._orig_keep_local_store
  49
  50     def write_foo_bar_baz(self):
  51         cw = arvados.CollectionWriter()
  52         self.assertEqual(cw.current_stream_name(), '.',
  53                          'current_stream_name() should be "." now')
  54         cw.set_current_file_name('foo.txt')
  55         cw.write('foo')
  56         self.assertEqual(cw.current_file_name(), 'foo.txt',
  57                          'current_file_name() should be foo.txt now')
  58         cw.start_new_file('bar.txt')
  59         cw.write('bar')
  60         cw.start_new_stream('baz')
  61         cw.write('baz')
  62         cw.set_current_file_name('baz.txt')
  63         return cw.finish()
  64
  65     def test_keep_local_store(self):
  66         self.assertEqual(arvados.Keep.put('foo'), 'acbd18db4cc2f85cedef654fccc4a4d8+3', 'wrong md5 hash from Keep.put')
  67         self.assertEqual(arvados.Keep.get('acbd18db4cc2f85cedef654fccc4a4d8+3'), 'foo', 'wrong data from Keep.get')
  68
  69     def test_local_collection_writer(self):
  70         self.assertEqual(self.write_foo_bar_baz(),
  71                          'd6c3b8e571f1b81ebb150a45ed06c884+114',
  72                          "wrong locator hash for files foo, bar, baz")
  73
  74     def test_local_collection_reader(self):
  75         self.write_foo_bar_baz()
  76         cr = arvados.CollectionReader('d6c3b8e571f1b81ebb150a45ed06c884+114+Xzizzle')
  77         got = []
  78         for s in cr.all_streams():
  79             for f in s.all_files():
  80                 got += [[f.size(), f.stream_name(), f.name(), f.read(2**26)]]
  81         expected = [[3, '.', 'bar.txt', 'bar'],
  82                     [3, '.', 'foo.txt', 'foo'],
  83                     [3, './baz', 'baz.txt', 'baz']]
  84         self.assertEqual(got,
  85                          expected)
  86         stream0 = cr.all_streams()[0]
  87         self.assertEqual(stream0.readfrom(0, 0),
  88                          '',
  89                          'reading zero bytes should have returned empty string')
  90         self.assertEqual(stream0.readfrom(0, 2**26),
  91                          'foobar',
  92                          'reading entire stream failed')
  93         self.assertEqual(stream0.readfrom(2**26, 0),
  94                          '',
  95                          'reading zero bytes should have returned empty string')
  96
  97     def _test_subset(self, collection, expected):
  98         cr = arvados.CollectionReader(collection)
  99         for s in cr.all_streams():
 100             for ex in expected:
 101                 if ex[0] == s:
 102                     f = s.files()[ex[2]]
 103                     got = [f.size(), f.stream_name(), f.name(), "".join(f.readall(2**26))]
 104                     self.assertEqual(got,
 105                                      ex,
 106                                      'all_files|as_manifest did not preserve manifest contents: got %s expected %s' % (got, ex))
 107
 108     def test_collection_manifest_subset(self):
 109         self.write_foo_bar_baz()
 110         self._test_subset('d6c3b8e571f1b81ebb150a45ed06c884+114',
 111                           [[3, '.',     'bar.txt', 'bar'],
 112                            [3, '.',     'foo.txt', 'foo'],
 113                            [3, './baz', 'baz.txt', 'baz']])
 114         self._test_subset((". %s %s 0:3:foo.txt 3:3:bar.txt\n" %
 115                            (arvados.Keep.put("foo"),
 116                             arvados.Keep.put("bar"))),
 117                           [[3, '.', 'bar.txt', 'bar'],
 118                            [3, '.', 'foo.txt', 'foo']])
 119         self._test_subset((". %s %s 0:2:fo.txt 2:4:obar.txt\n" %
 120                            (arvados.Keep.put("foo"),
 121                             arvados.Keep.put("bar"))),
 122                           [[2, '.', 'fo.txt', 'fo'],
 123                            [4, '.', 'obar.txt', 'obar']])
 124         self._test_subset((". %s %s 0:2:fo.txt 2:0:zero.txt 2:2:ob.txt 4:2:ar.txt\n" %
 125                            (arvados.Keep.put("foo"),
 126                             arvados.Keep.put("bar"))),
 127                           [[2, '.', 'ar.txt', 'ar'],
 128                            [2, '.', 'fo.txt', 'fo'],
 129                            [2, '.', 'ob.txt', 'ob'],
 130                            [0, '.', 'zero.txt', '']])
 131
 132     def _test_readline(self, what_in, what_out):
 133         cw = arvados.CollectionWriter()
 134         cw.start_new_file('test.txt')
 135         cw.write(what_in)
 136         test1 = cw.finish()
 137         cr = arvados.CollectionReader(test1)
 138         got = []
 139         for x in list(cr.all_files())[0].readlines():
 140             got += [x]
 141         self.assertEqual(got,
 142                          what_out,
 143                          "readlines did not split lines correctly: %s" % got)
 144
 145     def test_collection_readline(self):
 146         self._test_readline("\na\nbcd\n\nefg\nz",
 147                             ["\n", "a\n", "bcd\n", "\n", "efg\n", "z"])
 148         self._test_readline("ab\ncd\n",
 149                             ["ab\n", "cd\n"])
 150
 151     def test_collection_empty_file(self):
 152         cw = arvados.CollectionWriter()
 153         cw.start_new_file('zero.txt')
 154         cw.write('')
 155
 156         self.assertEqual(cw.manifest_text(), ". 0:0:zero.txt\n")
 157         self.check_manifest_file_sizes(cw.manifest_text(), [0])
 158         cw = arvados.CollectionWriter()
 159         cw.start_new_file('zero.txt')
 160         cw.write('')
 161         cw.start_new_file('one.txt')
 162         cw.write('1')
 163         cw.start_new_stream('foo')
 164         cw.start_new_file('zero.txt')
 165         cw.write('')
 166         self.check_manifest_file_sizes(cw.manifest_text(), [1,0,0])
 167
 168     def check_manifest_file_sizes(self, manifest_text, expect_sizes):
 169         cr = arvados.CollectionReader(manifest_text)
 170         got_sizes = []
 171         for f in cr.all_files():
 172             got_sizes += [f.size()]
 173         self.assertEqual(got_sizes, expect_sizes, "got wrong file sizes %s, expected %s" % (got_sizes, expect_sizes))
 174
 175     def test_collection_bz2_decompression(self):
 176         n_lines_in = 2**18
 177         data_in = "abc\n"
 178         for x in xrange(0, 18):
 179             data_in += data_in
 180         compressed_data_in = bz2.compress(data_in)
 181         cw = arvados.CollectionWriter()
 182         cw.start_new_file('test.bz2')
 183         cw.write(compressed_data_in)
 184         bz2_manifest = cw.manifest_text()
 185
 186         cr = arvados.CollectionReader(bz2_manifest)
 187
 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_collection_gzip_decompression(self):
 197         n_lines_in = 2**18
 198         data_in = "abc\n"
 199         for x in xrange(0, 18):
 200             data_in += data_in
 201         p = subprocess.Popen(["gzip", "-1cn"],
 202                              stdout=subprocess.PIPE,
 203                              stdin=subprocess.PIPE,
 204                              stderr=subprocess.PIPE,
 205                              shell=False, close_fds=True)
 206         compressed_data_in, stderrdata = p.communicate(data_in)
 207
 208         cw = arvados.CollectionWriter()
 209         cw.start_new_file('test.gz')
 210         cw.write(compressed_data_in)
 211         gzip_manifest = cw.manifest_text()
 212
 213         cr = arvados.CollectionReader(gzip_manifest)
 214         got = 0
 215         for x in list(cr.all_files())[0].readlines():
 216             self.assertEqual(x, "abc\n", "decompression returned wrong data: %s" % x)
 217             got += 1
 218         self.assertEqual(got,
 219                          n_lines_in,
 220                          "decompression returned %d lines instead of %d" % (got, n_lines_in))
 221
 222     def data_file(self, filename):
 223         try:
 224             basedir = os.path.dirname(__file__)
 225         except NameError:
 226             basedir = '.'
 227         return open(os.path.join(basedir, 'data', filename))
 228
 229     def test_normalized_collection(self):
 230         m1 = """. 5348b82a029fd9e971a811ce1f71360b+43 0:43:md5sum.txt
 231 . 085c37f02916da1cad16f93c54d899b7+41 0:41:md5sum.txt
 232 . 8b22da26f9f433dea0a10e5ec66d73ba+43 0:43:md5sum.txt"""
 233         self.assertEqual(arvados.CollectionReader(m1).manifest_text(),
 234                          """. 5348b82a029fd9e971a811ce1f71360b+43 085c37f02916da1cad16f93c54d899b7+41 8b22da26f9f433dea0a10e5ec66d73ba+43 0:127:md5sum.txt
 235 """)
 236
 237         m2 = """. 204e43b8a1185621ca55a94839582e6f+67108864 b9677abbac956bd3e86b1deb28dfac03+67108864 fc15aff2a762b13f521baf042140acec+67108864 323d2a3ce20370c4ca1d3462a344f8fd+25885655 0:227212247:var-GS000016015-ASM.tsv.bz2
 238 """
 239         self.assertEqual(arvados.CollectionReader(m2).manifest_text(), m2)
 240
 241         m3 = """. 5348b82a029fd9e971a811ce1f71360b+43 3:40:md5sum.txt
 242 . 085c37f02916da1cad16f93c54d899b7+41 0:41:md5sum.txt
 243 . 8b22da26f9f433dea0a10e5ec66d73ba+43 0:43:md5sum.txt"""
 244         self.assertEqual(arvados.CollectionReader(m3).manifest_text(),
 245                          """. 5348b82a029fd9e971a811ce1f71360b+43 085c37f02916da1cad16f93c54d899b7+41 8b22da26f9f433dea0a10e5ec66d73ba+43 3:124:md5sum.txt
 246 """)
 247
 248         m4 = """. 204e43b8a1185621ca55a94839582e6f+67108864 0:3:foo/bar
 249 ./zzz 204e43b8a1185621ca55a94839582e6f+67108864 0:999:zzz
 250 ./foo 323d2a3ce20370c4ca1d3462a344f8fd+25885655 0:3:bar"""
 251         self.assertEqual(arvados.CollectionReader(m4).manifest_text(),
 252                          """./foo 204e43b8a1185621ca55a94839582e6f+67108864 323d2a3ce20370c4ca1d3462a344f8fd+25885655 0:3:bar 67108864:3:bar
 253 ./zzz 204e43b8a1185621ca55a94839582e6f+67108864 0:999:zzz
 254 """)
 255
 256         m5 = """. 204e43b8a1185621ca55a94839582e6f+67108864 0:3:foo/bar
 257 ./zzz 204e43b8a1185621ca55a94839582e6f+67108864 0:999:zzz
 258 ./foo 204e43b8a1185621ca55a94839582e6f+67108864 3:3:bar"""
 259         self.assertEqual(arvados.CollectionReader(m5).manifest_text(),
 260                          """./foo 204e43b8a1185621ca55a94839582e6f+67108864 0:6:bar
 261 ./zzz 204e43b8a1185621ca55a94839582e6f+67108864 0:999:zzz
 262 """)
 263
 264         with self.data_file('1000G_ref_manifest') as f6:
 265             m6 = f6.read()
 266             self.assertEqual(arvados.CollectionReader(m6).manifest_text(), m6)
 267
 268         with self.data_file('jlake_manifest') as f7:
 269             m7 = f7.read()
 270             self.assertEqual(arvados.CollectionReader(m7).manifest_text(), m7)
 271
 272         m8 = """./a\\040b\\040c 59ca0efa9f5633cb0371bbc0355478d8+13 0:13:hello\\040world.txt
 273 """
 274         self.assertEqual(arvados.CollectionReader(m8).manifest_text(), m8)
 275
 276     def test_locators_and_ranges(self):
 277         blocks2 = [['a', 10, 0],
 278                   ['b', 10, 10],
 279                   ['c', 10, 20],
 280                   ['d', 10, 30],
 281                   ['e', 10, 40],
 282                   ['f', 10, 50]]
 283
 284         self.assertEqual(arvados.locators_and_ranges(blocks2,  2,  2), [['a', 10, 2, 2]])
 285         self.assertEqual(arvados.locators_and_ranges(blocks2, 12, 2), [['b', 10, 2, 2]])
 286         self.assertEqual(arvados.locators_and_ranges(blocks2, 22, 2), [['c', 10, 2, 2]])
 287         self.assertEqual(arvados.locators_and_ranges(blocks2, 32, 2), [['d', 10, 2, 2]])
 288         self.assertEqual(arvados.locators_and_ranges(blocks2, 42, 2), [['e', 10, 2, 2]])
 289         self.assertEqual(arvados.locators_and_ranges(blocks2, 52, 2), [['f', 10, 2, 2]])
 290         self.assertEqual(arvados.locators_and_ranges(blocks2, 62, 2), [])
 291         self.assertEqual(arvados.locators_and_ranges(blocks2, -2, 2), [])
 292
 293         self.assertEqual(arvados.locators_and_ranges(blocks2,  0,  2), [['a', 10, 0, 2]])
 294         self.assertEqual(arvados.locators_and_ranges(blocks2, 10, 2), [['b', 10, 0, 2]])
 295         self.assertEqual(arvados.locators_and_ranges(blocks2, 20, 2), [['c', 10, 0, 2]])
 296         self.assertEqual(arvados.locators_and_ranges(blocks2, 30, 2), [['d', 10, 0, 2]])
 297         self.assertEqual(arvados.locators_and_ranges(blocks2, 40, 2), [['e', 10, 0, 2]])
 298         self.assertEqual(arvados.locators_and_ranges(blocks2, 50, 2), [['f', 10, 0, 2]])
 299         self.assertEqual(arvados.locators_and_ranges(blocks2, 60, 2), [])
 300         self.assertEqual(arvados.locators_and_ranges(blocks2, -2, 2), [])
 301
 302         self.assertEqual(arvados.locators_and_ranges(blocks2,  9,  2), [['a', 10, 9, 1], ['b', 10, 0, 1]])
 303         self.assertEqual(arvados.locators_and_ranges(blocks2, 19, 2), [['b', 10, 9, 1], ['c', 10, 0, 1]])
 304         self.assertEqual(arvados.locators_and_ranges(blocks2, 29, 2), [['c', 10, 9, 1], ['d', 10, 0, 1]])
 305         self.assertEqual(arvados.locators_and_ranges(blocks2, 39, 2), [['d', 10, 9, 1], ['e', 10, 0, 1]])
 306         self.assertEqual(arvados.locators_and_ranges(blocks2, 49, 2), [['e', 10, 9, 1], ['f', 10, 0, 1]])
 307         self.assertEqual(arvados.locators_and_ranges(blocks2, 59, 2), [['f', 10, 9, 1]])
 308
 309
 310         blocks3 = [['a', 10, 0],
 311                   ['b', 10, 10],
 312                   ['c', 10, 20],
 313                   ['d', 10, 30],
 314                   ['e', 10, 40],
 315                   ['f', 10, 50],
 316                   ['g', 10, 60]]
 317
 318         self.assertEqual(arvados.locators_and_ranges(blocks3,  2,  2), [['a', 10, 2, 2]])
 319         self.assertEqual(arvados.locators_and_ranges(blocks3, 12, 2), [['b', 10, 2, 2]])
 320         self.assertEqual(arvados.locators_and_ranges(blocks3, 22, 2), [['c', 10, 2, 2]])
 321         self.assertEqual(arvados.locators_and_ranges(blocks3, 32, 2), [['d', 10, 2, 2]])
 322         self.assertEqual(arvados.locators_and_ranges(blocks3, 42, 2), [['e', 10, 2, 2]])
 323         self.assertEqual(arvados.locators_and_ranges(blocks3, 52, 2), [['f', 10, 2, 2]])
 324         self.assertEqual(arvados.locators_and_ranges(blocks3, 62, 2), [['g', 10, 2, 2]])
 325
 326
 327         blocks = [['a', 10, 0],
 328                   ['b', 15, 10],
 329                   ['c', 5, 25]]
 330         self.assertEqual(arvados.locators_and_ranges(blocks, 1, 0), [])
 331         self.assertEqual(arvados.locators_and_ranges(blocks, 0, 5), [['a', 10, 0, 5]])
 332         self.assertEqual(arvados.locators_and_ranges(blocks, 3, 5), [['a', 10, 3, 5]])
 333         self.assertEqual(arvados.locators_and_ranges(blocks, 0, 10), [['a', 10, 0, 10]])
 334
 335         self.assertEqual(arvados.locators_and_ranges(blocks, 0, 11), [['a', 10, 0, 10],
 336                                                                       ['b', 15, 0, 1]])
 337         self.assertEqual(arvados.locators_and_ranges(blocks, 1, 11), [['a', 10, 1, 9],
 338                                                                       ['b', 15, 0, 2]])
 339         self.assertEqual(arvados.locators_and_ranges(blocks, 0, 25), [['a', 10, 0, 10],
 340                                                                       ['b', 15, 0, 15]])
 341
 342         self.assertEqual(arvados.locators_and_ranges(blocks, 0, 30), [['a', 10, 0, 10],
 343                                                                       ['b', 15, 0, 15],
 344                                                                       ['c', 5, 0, 5]])
 345         self.assertEqual(arvados.locators_and_ranges(blocks, 1, 30), [['a', 10, 1, 9],
 346                                                                       ['b', 15, 0, 15],
 347                                                                       ['c', 5, 0, 5]])
 348         self.assertEqual(arvados.locators_and_ranges(blocks, 0, 31), [['a', 10, 0, 10],
 349                                                                       ['b', 15, 0, 15],
 350                                                                       ['c', 5, 0, 5]])
 351
 352         self.assertEqual(arvados.locators_and_ranges(blocks, 15, 5), [['b', 15, 5, 5]])
 353
 354         self.assertEqual(arvados.locators_and_ranges(blocks, 8, 17), [['a', 10, 8, 2],
 355                                                                       ['b', 15, 0, 15]])
 356
 357         self.assertEqual(arvados.locators_and_ranges(blocks, 8, 20), [['a', 10, 8, 2],
 358                                                                       ['b', 15, 0, 15],
 359                                                                       ['c', 5, 0, 3]])
 360
 361         self.assertEqual(arvados.locators_and_ranges(blocks, 26, 2), [['c', 5, 1, 2]])
 362
 363         self.assertEqual(arvados.locators_and_ranges(blocks, 9, 15), [['a', 10, 9, 1],
 364                                                                       ['b', 15, 0, 14]])
 365         self.assertEqual(arvados.locators_and_ranges(blocks, 10, 15), [['b', 15, 0, 15]])
 366         self.assertEqual(arvados.locators_and_ranges(blocks, 11, 15), [['b', 15, 1, 14],
 367                                                                        ['c', 5, 0, 1]])
 368
 369     class MockStreamReader(object):
 370         def __init__(self, content):
 371             self.content = content
 372
 373         def readfrom(self, start, size):
 374             return self.content[start:start+size]
 375
 376     def test_file_stream(self):
 377         content = 'abcdefghijklmnopqrstuvwxyz0123456789'
 378         msr = self.MockStreamReader(content)
 379         segments = [[0, 10, 0],
 380                     [10, 15, 10],
 381                     [25, 5, 25]]
 382
 383         sfr = arvados.StreamFileReader(msr, segments, "test")
 384
 385         self.assertEqual(sfr.name(), "test")
 386         self.assertEqual(sfr.size(), 30)
 387
 388         self.assertEqual(sfr.readfrom(0, 30), content[0:30])
 389         self.assertEqual(sfr.readfrom(2, 30), content[2:30])
 390
 391         self.assertEqual(sfr.readfrom(2, 8), content[2:10])
 392         self.assertEqual(sfr.readfrom(0, 10), content[0:10])
 393
 394         self.assertEqual(sfr.tell(), 0)
 395         self.assertEqual(sfr.read(5), content[0:5])
 396         self.assertEqual(sfr.tell(), 5)
 397         self.assertEqual(sfr.read(5), content[5:10])
 398         self.assertEqual(sfr.tell(), 10)
 399         self.assertEqual(sfr.read(5), content[10:15])
 400         self.assertEqual(sfr.tell(), 15)
 401         self.assertEqual(sfr.read(5), content[15:20])
 402         self.assertEqual(sfr.tell(), 20)
 403         self.assertEqual(sfr.read(5), content[20:25])
 404         self.assertEqual(sfr.tell(), 25)
 405         self.assertEqual(sfr.read(5), content[25:30])
 406         self.assertEqual(sfr.tell(), 30)
 407         self.assertEqual(sfr.read(5), '')
 408         self.assertEqual(sfr.tell(), 30)
 409
 410         segments = [[26, 10, 0],
 411                     [0, 15, 10],
 412                     [15, 5, 25]]
 413
 414         sfr = arvados.StreamFileReader(msr, segments, "test")
 415
 416         self.assertEqual(sfr.size(), 30)
 417
 418         self.assertEqual(sfr.readfrom(0, 30), content[26:36] + content[0:20])
 419         self.assertEqual(sfr.readfrom(2, 30), content[28:36] + content[0:20])
 420
 421         self.assertEqual(sfr.readfrom(2, 8), content[28:36])
 422         self.assertEqual(sfr.readfrom(0, 10), content[26:36])
 423
 424         self.assertEqual(sfr.tell(), 0)
 425         self.assertEqual(sfr.read(5), content[26:31])
 426         self.assertEqual(sfr.tell(), 5)
 427         self.assertEqual(sfr.read(5), content[31:36])
 428         self.assertEqual(sfr.tell(), 10)
 429         self.assertEqual(sfr.read(5), content[0:5])
 430         self.assertEqual(sfr.tell(), 15)
 431         self.assertEqual(sfr.read(5), content[5:10])
 432         self.assertEqual(sfr.tell(), 20)
 433         self.assertEqual(sfr.read(5), content[10:15])
 434         self.assertEqual(sfr.tell(), 25)
 435         self.assertEqual(sfr.read(5), content[15:20])
 436         self.assertEqual(sfr.tell(), 30)
 437         self.assertEqual(sfr.read(5), '')
 438         self.assertEqual(sfr.tell(), 30)
 439
 440
 441     class MockKeep(object):
 442         def __init__(self, content):
 443             self.content = content
 444
 445         def get(self, locator):
 446             return self.content[locator]
 447
 448     def test_stream_reader(self):
 449         keepblocks = {'aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa+10': 'abcdefghij',
 450                       'bbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbb+15': 'klmnopqrstuvwxy',
 451                       'cccccccccccccccccccccccccccccccc+5': 'z0123'}
 452         mk = self.MockKeep(keepblocks)
 453
 454         sr = arvados.StreamReader([".", "aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa+10", "bbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbb+15", "cccccccccccccccccccccccccccccccc+5", "0:30:foo"], mk)
 455
 456         content = 'abcdefghijklmnopqrstuvwxyz0123456789'
 457
 458         self.assertEqual(sr.readfrom(0, 30), content[0:30])
 459         self.assertEqual(sr.readfrom(2, 30), content[2:30])
 460
 461         self.assertEqual(sr.readfrom(2, 8), content[2:10])
 462         self.assertEqual(sr.readfrom(0, 10), content[0:10])
 463
 464         self.assertEqual(sr.readfrom(0, 5), content[0:5])
 465         self.assertEqual(sr.readfrom(5, 5), content[5:10])
 466         self.assertEqual(sr.readfrom(10, 5), content[10:15])
 467         self.assertEqual(sr.readfrom(15, 5), content[15:20])
 468         self.assertEqual(sr.readfrom(20, 5), content[20:25])
 469         self.assertEqual(sr.readfrom(25, 5), content[25:30])
 470         self.assertEqual(sr.readfrom(30, 5), '')
 471
 472     def test_extract_file(self):
 473         m1 = """. 5348b82a029fd9e971a811ce1f71360b+43 0:43:md5sum.txt
 474 . 085c37f02916da1cad16f93c54d899b7+41 0:41:md6sum.txt
 475 . 8b22da26f9f433dea0a10e5ec66d73ba+43 0:43:md7sum.txt
 476 . 085c37f02916da1cad16f93c54d899b7+41 5348b82a029fd9e971a811ce1f71360b+43 8b22da26f9f433dea0a10e5ec66d73ba+43 47:80:md8sum.txt
 477 . 085c37f02916da1cad16f93c54d899b7+41 5348b82a029fd9e971a811ce1f71360b+43 8b22da26f9f433dea0a10e5ec66d73ba+43 40:80:md9sum.txt"""
 478
 479         m2 = arvados.CollectionReader(m1)
 480
 481         self.assertEqual(m2.manifest_text(),
 482                          ". 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")
 483
 484         self.assertEqual(arvados.CollectionReader(m1).all_streams()[0].files()['md5sum.txt'].as_manifest(),
 485                          ". 5348b82a029fd9e971a811ce1f71360b+43 0:43:md5sum.txt\n")
 486         self.assertEqual(arvados.CollectionReader(m1).all_streams()[0].files()['md6sum.txt'].as_manifest(),
 487                          ". 085c37f02916da1cad16f93c54d899b7+41 0:41:md6sum.txt\n")
 488         self.assertEqual(arvados.CollectionReader(m1).all_streams()[0].files()['md7sum.txt'].as_manifest(),
 489                          ". 8b22da26f9f433dea0a10e5ec66d73ba+43 0:43:md7sum.txt\n")
 490         self.assertEqual(arvados.CollectionReader(m1).all_streams()[0].files()['md9sum.txt'].as_manifest(),
 491                          ". 085c37f02916da1cad16f93c54d899b7+41 5348b82a029fd9e971a811ce1f71360b+43 8b22da26f9f433dea0a10e5ec66d73ba+43 40:80:md9sum.txt\n")
 492
 493     def build_directory_tree(self, tree=['basefile', 'subdir/subfile']):
 494         tree_root = self._make_tmpdir()
 495         for leaf in tree:
 496             path = os.path.join(tree_root, leaf)
 497             try:
 498                 os.makedirs(os.path.dirname(path))
 499             except OSError as error:
 500                 if error.errno != errno.EEXIST:
 501                     raise
 502             with open(path, 'w') as tmpfile:
 503                 tmpfile.write(leaf)
 504         return tree_root
 505
 506     def test_write_directory_tree(self):
 507         cwriter = arvados.CollectionWriter()
 508         cwriter.write_directory_tree(self.build_directory_tree())
 509         self.assertEqual(cwriter.manifest_text(),
 510                          """. c5110c5ac93202d8e0f9e381f22bac0f+8 0:8:basefile
 511 ./subdir 1ca4dec89403084bf282ad31e6cf7972+14 0:14:subfile\n""")
 512
 513     def test_write_named_directory_tree(self):
 514         cwriter = arvados.CollectionWriter()
 515         cwriter.write_directory_tree(self.build_directory_tree(), 'root')
 516         self.assertEqual(
 517             cwriter.manifest_text(),
 518             """./root c5110c5ac93202d8e0f9e381f22bac0f+8 0:8:basefile
 519 ./root/subdir 1ca4dec89403084bf282ad31e6cf7972+14 0:14:subfile\n""")
 520
 521     def test_write_directory_tree_in_one_stream(self):
 522         cwriter = arvados.CollectionWriter()
 523         cwriter.write_directory_tree(self.build_directory_tree(),
 524                                      max_manifest_depth=0)
 525         self.assertEqual(cwriter.manifest_text(),
 526                          """. 4ace875ffdc6824a04950f06858f4465+22 0:8:basefile
 527 ./subdir 4ace875ffdc6824a04950f06858f4465+22 8:14:subfile\n""")
 528
 529     def test_write_directory_tree_with_limited_recursion(self):
 530         cwriter = arvados.CollectionWriter()
 531         cwriter.write_directory_tree(
 532             self.build_directory_tree(['f1', 'd1/f2', 'd1/d2/f3']),
 533             max_manifest_depth=1)
 534         self.assertEqual(cwriter.manifest_text(),
 535                          """. bd19836ddb62c11c55ab251ccaca5645+2 0:2:f1
 536 ./d1 50170217e5b04312024aa5cd42934494+13 8:5:f2
 537 ./d1/d2 50170217e5b04312024aa5cd42934494+13 0:8:f3\n""")
 538
 539     def make_test_file(self, text="test"):
 540         testfile = tempfile.NamedTemporaryFile()
 541         testfile.write(text)
 542         testfile.flush()
 543         return testfile
 544
 545     def test_write_one_file(self):
 546         cwriter = arvados.CollectionWriter()
 547         with self.make_test_file() as testfile:
 548             cwriter.write_file(testfile.name)
 549             self.assertEqual(
 550                 cwriter.manifest_text(),
 551                 ". 098f6bcd4621d373cade4e832627b4f6+4 0:4:{}\n".format(
 552                     os.path.basename(testfile.name)))
 553
 554     def test_write_named_file(self):
 555         cwriter = arvados.CollectionWriter()
 556         with self.make_test_file() as testfile:
 557             cwriter.write_file(testfile.name, 'foo')
 558             self.assertEqual(cwriter.manifest_text(),
 559                              ". 098f6bcd4621d373cade4e832627b4f6+4 0:4:foo\n")
 560
 561     def test_write_multiple_files(self):
 562         cwriter = arvados.CollectionWriter()
 563         for letter in 'ABC':
 564             with self.make_test_file(letter) as testfile:
 565                 cwriter.write_file(testfile.name, letter)
 566         self.assertEqual(
 567             cwriter.manifest_text(),
 568             ". 902fbdd2b1df0c4f70b4a5d23525e932+3 0:1:A 1:1:B 2:1:C\n")
 569
 570     def test_checkpoint_after_put(self):
 571         cwriter = TestResumableWriter()
 572         with self.make_test_file(
 573               't' * (cwriter.KEEP_BLOCK_SIZE + 10)) as testfile:
 574             testpath = os.path.realpath(testfile.name)
 575             cwriter.write_file(testpath, 'test')
 576         for state in cwriter.saved_states:
 577             if state.get('_current_file') == (testpath,
 578                                               cwriter.KEEP_BLOCK_SIZE):
 579                 break
 580         else:
 581             self.fail("can't find state immediately after PUT to Keep")
 582         self.assertIn('d45107e93f9052fa88a82fc08bb1d316+1024',  # 't' * 1024
 583                       state['_current_stream_locators'])
 584
 585     def test_basic_resume(self):
 586         cwriter = TestResumableWriter()
 587         with self.make_test_file() as testfile:
 588             cwriter.write_file(testfile.name, 'test')
 589             last_state = cwriter.last_state()
 590             resumed = TestResumableWriter.from_state(last_state)
 591         self.assertEquals(cwriter.manifest_text(), resumed.manifest_text(),
 592                           "resumed CollectionWriter had different manifest")
 593
 594     def test_resume_fails_when_missing_dependency(self):
 595         cwriter = TestResumableWriter()
 596         with self.make_test_file() as testfile:
 597             cwriter.write_file(testfile.name, 'test')
 598         self.assertRaises(arvados.errors.StaleWriterStateError,
 599                           TestResumableWriter.from_state,
 600                           cwriter.last_state())
 601
 602     def test_resume_fails_when_dependency_mtime_changed(self):
 603         cwriter = TestResumableWriter()
 604         with self.make_test_file() as testfile:
 605             cwriter.write_file(testfile.name, 'test')
 606             os.utime(testfile.name, (0, 0))
 607             self.assertRaises(arvados.errors.StaleWriterStateError,
 608                               TestResumableWriter.from_state,
 609                               cwriter.last_state())
 610
 611     def test_resume_fails_when_dependency_is_nonfile(self):
 612         cwriter = TestResumableWriter()
 613         cwriter.write_file('/dev/null', 'empty')
 614         self.assertRaises(arvados.errors.StaleWriterStateError,
 615                           TestResumableWriter.from_state,
 616                           cwriter.last_state())
 617
 618     def test_resume_fails_when_dependency_size_changed(self):
 619         cwriter = TestResumableWriter()
 620         with self.make_test_file() as testfile:
 621             cwriter.write_file(testfile.name, 'test')
 622             orig_mtime = os.fstat(testfile.fileno()).st_mtime
 623             testfile.write('extra')
 624             testfile.flush()
 625             os.utime(testfile.name, (orig_mtime, orig_mtime))
 626             self.assertRaises(arvados.errors.StaleWriterStateError,
 627                               TestResumableWriter.from_state,
 628                               cwriter.last_state())
 629
 630     def test_successful_resumes(self):
 631         # FIXME: This is more of an integration test than a unit test.
 632         cwriter = TestResumableWriter()
 633         source_tree = self.build_directory_tree()
 634         with open(os.path.join(source_tree, 'long'), 'w') as longfile:
 635             longfile.write('t' * (cwriter.KEEP_BLOCK_SIZE + 10))
 636         cwriter.write_directory_tree(source_tree)
 637         # A state for each file, plus a fourth for mid-longfile.
 638         self.assertGreater(len(cwriter.saved_states), 3,
 639                            "CollectionWriter didn't save enough states to test")
 640
 641         for state in cwriter.saved_states:
 642             new_writer = TestResumableWriter.from_state(state)
 643             manifests = [writer.manifest_text()
 644                          for writer in (cwriter, new_writer)]
 645             self.assertEquals(
 646                 manifests[0], manifests[1],
 647                 "\n".join(["manifest mismatch after resuming from state:",
 648                            pprint.pformat(state), ""] + manifests))
 649
 650     def test_arbitrary_objects_not_resumable(self):
 651         cwriter = TestResumableWriter()
 652         with open('/dev/null') as badfile:
 653             self.assertRaises(arvados.errors.AssertionError,
 654                               cwriter.write_file, badfile)
 655
 656     def test_arbitrary_writes_not_resumable(self):
 657         cwriter = TestResumableWriter()
 658         self.assertRaises(arvados.errors.AssertionError,
 659                           cwriter.write, "badtext")
 660
 661
 662 if __name__ == '__main__':
 663     unittest.main()