sdk/python/test_collections.py

   1 # usage example:
   2 #
   3 # ARVADOS_API_TOKEN=abc ARVADOS_API_HOST=arvados.local python -m unittest discover
   4
   5 import unittest
   6 import arvados
   7 import os
   8 import bz2
   9 import sys
  10 import subprocess
  11
  12 class KeepLocalStoreTest(unittest.TestCase):
  13     def setUp(self):
  14         os.environ['KEEP_LOCAL_STORE'] = '/tmp'
  15     def runTest(self):
  16         self.assertEqual(arvados.Keep.put('foo'), 'acbd18db4cc2f85cedef654fccc4a4d8+3', 'wrong md5 hash from Keep.put')
  17         self.assertEqual(arvados.Keep.get('acbd18db4cc2f85cedef654fccc4a4d8+3'), 'foo', 'wrong data from Keep.get')
  18
  19 class LocalCollectionWriterTest(unittest.TestCase):
  20     def setUp(self):
  21         os.environ['KEEP_LOCAL_STORE'] = '/tmp'
  22     def runTest(self):
  23         cw = arvados.CollectionWriter()
  24         self.assertEqual(cw.current_stream_name(), '.',
  25                          'current_stream_name() should be "." now')
  26         cw.set_current_file_name('foo.txt')
  27         cw.write('foo')
  28         self.assertEqual(cw.current_file_name(), 'foo.txt',
  29                          'current_file_name() should be foo.txt now')
  30         cw.start_new_file('bar.txt')
  31         cw.write('bar')
  32         cw.start_new_stream('baz')
  33         cw.write('baz')
  34         cw.set_current_file_name('baz.txt')
  35         hash = cw.finish()
  36         self.assertEqual(hash,
  37                          'd6c3b8e571f1b81ebb150a45ed06c884+114',
  38                          "resulting manifest hash was {0}, expecting d6c3b8e571f1b81ebb150a45ed06c884+114".format(hash))
  39
  40 class LocalCollectionReaderTest(unittest.TestCase):
  41     def setUp(self):
  42         os.environ['KEEP_LOCAL_STORE'] = '/tmp'
  43         LocalCollectionWriterTest().runTest()
  44     def runTest(self):
  45         cr = arvados.CollectionReader('d6c3b8e571f1b81ebb150a45ed06c884+114')
  46         got = []
  47         for s in cr.all_streams():
  48             for f in s.all_files():
  49                 got += [[f.size(), f.stream_name(), f.name(), f.read(2**26)]]
  50         expected = [[3, '.', 'bar.txt', 'bar'],
  51                     [3, '.', 'foo.txt', 'foo'],
  52                     [3, './baz', 'baz.txt', 'baz']]
  53         self.assertEqual(got,
  54                          expected)
  55         stream0 = cr.all_streams()[0]
  56         self.assertEqual(stream0.readfrom(0, 0),
  57                          '',
  58                          'reading zero bytes should have returned empty string')
  59         self.assertEqual(stream0.readfrom(0, 2**26),
  60                          'foobar',
  61                          'reading entire stream failed')
  62         self.assertEqual(stream0.readfrom(2**26, 0),
  63                          '',
  64                          'reading zero bytes should have returned empty string')
  65
  66 class LocalCollectionManifestSubsetTest(unittest.TestCase):
  67     def setUp(self):
  68         os.environ['KEEP_LOCAL_STORE'] = '/tmp'
  69         LocalCollectionWriterTest().runTest()
  70     def runTest(self):
  71         self._runTest('d6c3b8e571f1b81ebb150a45ed06c884+114',
  72                       [[3, '.',     'bar.txt', 'bar'],
  73                        [3, '.',     'foo.txt', 'foo'],
  74                        [3, './baz', 'baz.txt', 'baz']])
  75         self._runTest((". %s %s 0:3:foo.txt 3:3:bar.txt\n" %
  76                        (arvados.Keep.put("foo"),
  77                         arvados.Keep.put("bar"))),
  78                       [[3, '.', 'bar.txt', 'bar'],
  79                        [3, '.', 'foo.txt', 'foo']])
  80         self._runTest((". %s %s 0:2:fo.txt 2:4:obar.txt\n" %
  81                        (arvados.Keep.put("foo"),
  82                         arvados.Keep.put("bar"))),
  83                       [[2, '.', 'fo.txt', 'fo'],
  84                        [4, '.', 'obar.txt', 'obar']])
  85         self._runTest((". %s %s 0:2:fo.txt 2:0:zero.txt 2:2:ob.txt 4:2:ar.txt\n" %
  86                        (arvados.Keep.put("foo"),
  87                         arvados.Keep.put("bar"))),
  88                       [[2, '.', 'ar.txt', 'ar'],
  89                        [2, '.', 'fo.txt', 'fo'],
  90                        [2, '.', 'ob.txt', 'ob'],
  91                        [0, '.', 'zero.txt', '']])
  92
  93     def _runTest(self, collection, expected):
  94         cr = arvados.CollectionReader(collection)
  95         for s in cr.all_streams():
  96             for ex in expected:
  97                 if ex[0] == s:
  98                     f = s.files()[ex[2]]
  99                     got = [f.size(), f.stream_name(), f.name(), "".join(f.readall(2**26))]
 100                     self.assertEqual(got,
 101                                      ex,
 102                                      'all_files|as_manifest did not preserve manifest contents: got %s expected %s' % (got, ex))
 103
 104 class LocalCollectionReadlineTest(unittest.TestCase):
 105     def setUp(self):
 106         os.environ['KEEP_LOCAL_STORE'] = '/tmp'
 107     def _runTest(self, what_in, what_out):
 108         cw = arvados.CollectionWriter()
 109         cw.start_new_file('test.txt')
 110         cw.write(what_in)
 111         test1 = cw.finish()
 112         cr = arvados.CollectionReader(test1)
 113         got = []
 114         for x in list(cr.all_files())[0].readlines():
 115             got += [x]
 116         self.assertEqual(got,
 117                          what_out,
 118                          "readlines did not split lines correctly: %s" % got)
 119     def runTest(self):
 120         self._runTest("\na\nbcd\n\nefg\nz",
 121                       ["\n", "a\n", "bcd\n", "\n", "efg\n", "z"])
 122         self._runTest("ab\ncd\n",
 123                       ["ab\n", "cd\n"])
 124
 125 class LocalCollectionEmptyFileTest(unittest.TestCase):
 126     def setUp(self):
 127         os.environ['KEEP_LOCAL_STORE'] = '/tmp'
 128     def runTest(self):
 129         cw = arvados.CollectionWriter()
 130         cw.start_new_file('zero.txt')
 131         cw.write('')
 132
 133         self.assertEqual(cw.manifest_text(), ". 0:0:zero.txt\n")
 134         self.check_manifest_file_sizes(cw.manifest_text(), [0])
 135         cw = arvados.CollectionWriter()
 136         cw.start_new_file('zero.txt')
 137         cw.write('')
 138         cw.start_new_file('one.txt')
 139         cw.write('1')
 140         cw.start_new_stream('foo')
 141         cw.start_new_file('zero.txt')
 142         cw.write('')
 143         self.check_manifest_file_sizes(cw.manifest_text(), [1,0,0])
 144
 145     def check_manifest_file_sizes(self, manifest_text, expect_sizes):
 146         cr = arvados.CollectionReader(manifest_text)
 147         got_sizes = []
 148         for f in cr.all_files():
 149             got_sizes += [f.size()]
 150         self.assertEqual(got_sizes, expect_sizes, "got wrong file sizes %s, expected %s" % (got_sizes, expect_sizes))
 151
 152 class LocalCollectionBZ2DecompressionTest(unittest.TestCase):
 153     def setUp(self):
 154         os.environ['KEEP_LOCAL_STORE'] = '/tmp'
 155     def runTest(self):
 156         n_lines_in = 2**18
 157         data_in = "abc\n"
 158         for x in xrange(0, 18):
 159             data_in += data_in
 160         compressed_data_in = bz2.compress(data_in)
 161         cw = arvados.CollectionWriter()
 162         cw.start_new_file('test.bz2')
 163         cw.write(compressed_data_in)
 164         bz2_manifest = cw.manifest_text()
 165
 166         cr = arvados.CollectionReader(bz2_manifest)
 167
 168         got = 0
 169         for x in list(cr.all_files())[0].readlines():
 170             self.assertEqual(x, "abc\n", "decompression returned wrong data: %s" % x)
 171             got += 1
 172         self.assertEqual(got,
 173                          n_lines_in,
 174                          "decompression returned %d lines instead of %d" % (got, n_lines_in))
 175
 176 class LocalCollectionGzipDecompressionTest(unittest.TestCase):
 177     def setUp(self):
 178         os.environ['KEEP_LOCAL_STORE'] = '/tmp'
 179     def runTest(self):
 180         n_lines_in = 2**18
 181         data_in = "abc\n"
 182         for x in xrange(0, 18):
 183             data_in += data_in
 184         p = subprocess.Popen(["gzip", "-1cn"],
 185                              stdout=subprocess.PIPE,
 186                              stdin=subprocess.PIPE,
 187                              stderr=subprocess.PIPE,
 188                              shell=False, close_fds=True)
 189         compressed_data_in, stderrdata = p.communicate(data_in)
 190
 191         cw = arvados.CollectionWriter()
 192         cw.start_new_file('test.gz')
 193         cw.write(compressed_data_in)
 194         gzip_manifest = cw.manifest_text()
 195
 196         cr = arvados.CollectionReader(gzip_manifest)
 197         got = 0
 198         for x in list(cr.all_files())[0].readlines():
 199             self.assertEqual(x, "abc\n", "decompression returned wrong data: %s" % x)
 200             got += 1
 201         self.assertEqual(got,
 202                          n_lines_in,
 203                          "decompression returned %d lines instead of %d" % (got, n_lines_in))
 204
 205 class NormalizedCollectionTest(unittest.TestCase):
 206     def runTest(self):
 207         m1 = """. 5348b82a029fd9e971a811ce1f71360b+43 0:43:md5sum.txt
 208 . 085c37f02916da1cad16f93c54d899b7+41 0:41:md5sum.txt
 209 . 8b22da26f9f433dea0a10e5ec66d73ba+43 0:43:md5sum.txt"""
 210         self.assertEqual(arvados.CollectionReader(m1).manifest_text(),
 211                          """. 5348b82a029fd9e971a811ce1f71360b+43 085c37f02916da1cad16f93c54d899b7+41 8b22da26f9f433dea0a10e5ec66d73ba+43 0:127:md5sum.txt
 212 """)
 213
 214         m2 = """. 204e43b8a1185621ca55a94839582e6f+67108864 b9677abbac956bd3e86b1deb28dfac03+67108864 fc15aff2a762b13f521baf042140acec+67108864 323d2a3ce20370c4ca1d3462a344f8fd+25885655 0:227212247:var-GS000016015-ASM.tsv.bz2
 215 """
 216         self.assertEqual(arvados.CollectionReader(m2).manifest_text(), m2)
 217
 218         m3 = """. 5348b82a029fd9e971a811ce1f71360b+43 3:40:md5sum.txt
 219 . 085c37f02916da1cad16f93c54d899b7+41 0:41:md5sum.txt
 220 . 8b22da26f9f433dea0a10e5ec66d73ba+43 0:43:md5sum.txt"""
 221         self.assertEqual(arvados.CollectionReader(m3).manifest_text(),
 222                          """. 5348b82a029fd9e971a811ce1f71360b+43 085c37f02916da1cad16f93c54d899b7+41 8b22da26f9f433dea0a10e5ec66d73ba+43 3:124:md5sum.txt
 223 """)
 224
 225         m4 = """. 204e43b8a1185621ca55a94839582e6f+67108864 0:3:foo/bar
 226 ./zzz 204e43b8a1185621ca55a94839582e6f+67108864 0:999:zzz
 227 ./foo 323d2a3ce20370c4ca1d3462a344f8fd+25885655 0:3:bar"""
 228         self.assertEqual(arvados.CollectionReader(m4).manifest_text(),
 229                          """./foo 204e43b8a1185621ca55a94839582e6f+67108864 323d2a3ce20370c4ca1d3462a344f8fd+25885655 0:3:bar 67108864:3:bar
 230 ./zzz 204e43b8a1185621ca55a94839582e6f+67108864 0:999:zzz
 231 """)
 232
 233         m5 = """. 204e43b8a1185621ca55a94839582e6f+67108864 0:3:foo/bar
 234 ./zzz 204e43b8a1185621ca55a94839582e6f+67108864 0:999:zzz
 235 ./foo 204e43b8a1185621ca55a94839582e6f+67108864 3:3:bar"""
 236         self.assertEqual(arvados.CollectionReader(m5).manifest_text(),
 237                          """./foo 204e43b8a1185621ca55a94839582e6f+67108864 0:6:bar
 238 ./zzz 204e43b8a1185621ca55a94839582e6f+67108864 0:999:zzz
 239 """)
 240
 241         with open('testdata/1000G_ref_manifest') as f6:
 242             m6 = f6.read()
 243             self.assertEqual(arvados.CollectionReader(m6).manifest_text(), m6)
 244
 245         with open('testdata/jlake_manifest') as f7:
 246             m7 = f7.read()
 247             self.assertEqual(arvados.CollectionReader(m7).manifest_text(), m7)
 248
 249         m8 = """./a\\040b\\040c 59ca0efa9f5633cb0371bbc0355478d8+13 0:13:hello\\040world.txt
 250 """
 251         self.assertEqual(arvados.CollectionReader(m8).manifest_text(), m8)
 252
 253 class LocatorsAndRangesTest(unittest.TestCase):
 254     def runTest(self):
 255         blocks2 = [['a', 10, 0],
 256                   ['b', 10, 10],
 257                   ['c', 10, 20],
 258                   ['d', 10, 30],
 259                   ['e', 10, 40],
 260                   ['f', 10, 50]]
 261
 262         self.assertEqual(arvados.locators_and_ranges(blocks2,  2,  2), [['a', 10, 2, 2]])
 263         self.assertEqual(arvados.locators_and_ranges(blocks2, 12, 2), [['b', 10, 2, 2]])
 264         self.assertEqual(arvados.locators_and_ranges(blocks2, 22, 2), [['c', 10, 2, 2]])
 265         self.assertEqual(arvados.locators_and_ranges(blocks2, 32, 2), [['d', 10, 2, 2]])
 266         self.assertEqual(arvados.locators_and_ranges(blocks2, 42, 2), [['e', 10, 2, 2]])
 267         self.assertEqual(arvados.locators_and_ranges(blocks2, 52, 2), [['f', 10, 2, 2]])
 268         self.assertEqual(arvados.locators_and_ranges(blocks2, 62, 2), [])
 269         self.assertEqual(arvados.locators_and_ranges(blocks2, -2, 2), [])
 270
 271         self.assertEqual(arvados.locators_and_ranges(blocks2,  0,  2), [['a', 10, 0, 2]])
 272         self.assertEqual(arvados.locators_and_ranges(blocks2, 10, 2), [['b', 10, 0, 2]])
 273         self.assertEqual(arvados.locators_and_ranges(blocks2, 20, 2), [['c', 10, 0, 2]])
 274         self.assertEqual(arvados.locators_and_ranges(blocks2, 30, 2), [['d', 10, 0, 2]])
 275         self.assertEqual(arvados.locators_and_ranges(blocks2, 40, 2), [['e', 10, 0, 2]])
 276         self.assertEqual(arvados.locators_and_ranges(blocks2, 50, 2), [['f', 10, 0, 2]])
 277         self.assertEqual(arvados.locators_and_ranges(blocks2, 60, 2), [])
 278         self.assertEqual(arvados.locators_and_ranges(blocks2, -2, 2), [])
 279
 280         self.assertEqual(arvados.locators_and_ranges(blocks2,  9,  2), [['a', 10, 9, 1], ['b', 10, 0, 1]])
 281         self.assertEqual(arvados.locators_and_ranges(blocks2, 19, 2), [['b', 10, 9, 1], ['c', 10, 0, 1]])
 282         self.assertEqual(arvados.locators_and_ranges(blocks2, 29, 2), [['c', 10, 9, 1], ['d', 10, 0, 1]])
 283         self.assertEqual(arvados.locators_and_ranges(blocks2, 39, 2), [['d', 10, 9, 1], ['e', 10, 0, 1]])
 284         self.assertEqual(arvados.locators_and_ranges(blocks2, 49, 2), [['e', 10, 9, 1], ['f', 10, 0, 1]])
 285         self.assertEqual(arvados.locators_and_ranges(blocks2, 59, 2), [['f', 10, 9, 1]])
 286
 287
 288         blocks3 = [['a', 10, 0],
 289                   ['b', 10, 10],
 290                   ['c', 10, 20],
 291                   ['d', 10, 30],
 292                   ['e', 10, 40],
 293                   ['f', 10, 50],
 294                   ['g', 10, 60]]
 295
 296         self.assertEqual(arvados.locators_and_ranges(blocks3,  2,  2), [['a', 10, 2, 2]])
 297         self.assertEqual(arvados.locators_and_ranges(blocks3, 12, 2), [['b', 10, 2, 2]])
 298         self.assertEqual(arvados.locators_and_ranges(blocks3, 22, 2), [['c', 10, 2, 2]])
 299         self.assertEqual(arvados.locators_and_ranges(blocks3, 32, 2), [['d', 10, 2, 2]])
 300         self.assertEqual(arvados.locators_and_ranges(blocks3, 42, 2), [['e', 10, 2, 2]])
 301         self.assertEqual(arvados.locators_and_ranges(blocks3, 52, 2), [['f', 10, 2, 2]])
 302         self.assertEqual(arvados.locators_and_ranges(blocks3, 62, 2), [['g', 10, 2, 2]])
 303
 304
 305         blocks = [['a', 10, 0],
 306                   ['b', 15, 10],
 307                   ['c', 5, 25]]
 308         self.assertEqual(arvados.locators_and_ranges(blocks, 1, 0), [])
 309         self.assertEqual(arvados.locators_and_ranges(blocks, 0, 5), [['a', 10, 0, 5]])
 310         self.assertEqual(arvados.locators_and_ranges(blocks, 3, 5), [['a', 10, 3, 5]])
 311         self.assertEqual(arvados.locators_and_ranges(blocks, 0, 10), [['a', 10, 0, 10]])
 312
 313         self.assertEqual(arvados.locators_and_ranges(blocks, 0, 11), [['a', 10, 0, 10],
 314                                                                       ['b', 15, 0, 1]])
 315         self.assertEqual(arvados.locators_and_ranges(blocks, 1, 11), [['a', 10, 1, 9],
 316                                                                       ['b', 15, 0, 2]])
 317         self.assertEqual(arvados.locators_and_ranges(blocks, 0, 25), [['a', 10, 0, 10],
 318                                                                       ['b', 15, 0, 15]])
 319
 320         self.assertEqual(arvados.locators_and_ranges(blocks, 0, 30), [['a', 10, 0, 10],
 321                                                                       ['b', 15, 0, 15],
 322                                                                       ['c', 5, 0, 5]])
 323         self.assertEqual(arvados.locators_and_ranges(blocks, 1, 30), [['a', 10, 1, 9],
 324                                                                       ['b', 15, 0, 15],
 325                                                                       ['c', 5, 0, 5]])
 326         self.assertEqual(arvados.locators_and_ranges(blocks, 0, 31), [['a', 10, 0, 10],
 327                                                                       ['b', 15, 0, 15],
 328                                                                       ['c', 5, 0, 5]])
 329
 330         self.assertEqual(arvados.locators_and_ranges(blocks, 15, 5), [['b', 15, 5, 5]])
 331
 332         self.assertEqual(arvados.locators_and_ranges(blocks, 8, 17), [['a', 10, 8, 2],
 333                                                                       ['b', 15, 0, 15]])
 334
 335         self.assertEqual(arvados.locators_and_ranges(blocks, 8, 20), [['a', 10, 8, 2],
 336                                                                       ['b', 15, 0, 15],
 337                                                                       ['c', 5, 0, 3]])
 338
 339         self.assertEqual(arvados.locators_and_ranges(blocks, 26, 2), [['c', 5, 1, 2]])
 340
 341         self.assertEqual(arvados.locators_and_ranges(blocks, 9, 15), [['a', 10, 9, 1],
 342                                                                       ['b', 15, 0, 14]])
 343         self.assertEqual(arvados.locators_and_ranges(blocks, 10, 15), [['b', 15, 0, 15]])
 344         self.assertEqual(arvados.locators_and_ranges(blocks, 11, 15), [['b', 15, 1, 14],
 345                                                                        ['c', 5, 0, 1]])
 346
 347 class FileStreamTest(unittest.TestCase):
 348     class MockStreamReader(object):
 349         def __init__(self, content):
 350             self.content = content
 351
 352         def readfrom(self, start, size):
 353             return self.content[start:start+size]
 354
 355     def runTest(self):
 356         content = 'abcdefghijklmnopqrstuvwxyz0123456789'
 357         msr = FileStreamTest.MockStreamReader(content)
 358         segments = [[0, 10, 0],
 359                     [10, 15, 10],
 360                     [25, 5, 25]]
 361
 362         sfr = arvados.StreamFileReader(msr, segments, "test")
 363
 364         self.assertEqual(sfr.name(), "test")
 365         self.assertEqual(sfr.size(), 30)
 366
 367         self.assertEqual(sfr.readfrom(0, 30), content[0:30])
 368         self.assertEqual(sfr.readfrom(2, 30), content[2:30])
 369
 370         self.assertEqual(sfr.readfrom(2, 8), content[2:10])
 371         self.assertEqual(sfr.readfrom(0, 10), content[0:10])
 372
 373         self.assertEqual(sfr.tell(), 0)
 374         self.assertEqual(sfr.read(5), content[0:5])
 375         self.assertEqual(sfr.tell(), 5)
 376         self.assertEqual(sfr.read(5), content[5:10])
 377         self.assertEqual(sfr.tell(), 10)
 378         self.assertEqual(sfr.read(5), content[10:15])
 379         self.assertEqual(sfr.tell(), 15)
 380         self.assertEqual(sfr.read(5), content[15:20])
 381         self.assertEqual(sfr.tell(), 20)
 382         self.assertEqual(sfr.read(5), content[20:25])
 383         self.assertEqual(sfr.tell(), 25)
 384         self.assertEqual(sfr.read(5), content[25:30])
 385         self.assertEqual(sfr.tell(), 30)
 386         self.assertEqual(sfr.read(5), '')
 387         self.assertEqual(sfr.tell(), 30)
 388
 389         segments = [[26, 10, 0],
 390                     [0, 15, 10],
 391                     [15, 5, 25]]
 392
 393         sfr = arvados.StreamFileReader(msr, segments, "test")
 394
 395         self.assertEqual(sfr.size(), 30)
 396
 397         self.assertEqual(sfr.readfrom(0, 30), content[26:36] + content[0:20])
 398         self.assertEqual(sfr.readfrom(2, 30), content[28:36] + content[0:20])
 399
 400         self.assertEqual(sfr.readfrom(2, 8), content[28:36])
 401         self.assertEqual(sfr.readfrom(0, 10), content[26:36])
 402
 403         self.assertEqual(sfr.tell(), 0)
 404         self.assertEqual(sfr.read(5), content[26:31])
 405         self.assertEqual(sfr.tell(), 5)
 406         self.assertEqual(sfr.read(5), content[31:36])
 407         self.assertEqual(sfr.tell(), 10)
 408         self.assertEqual(sfr.read(5), content[0:5])
 409         self.assertEqual(sfr.tell(), 15)
 410         self.assertEqual(sfr.read(5), content[5:10])
 411         self.assertEqual(sfr.tell(), 20)
 412         self.assertEqual(sfr.read(5), content[10:15])
 413         self.assertEqual(sfr.tell(), 25)
 414         self.assertEqual(sfr.read(5), content[15:20])
 415         self.assertEqual(sfr.tell(), 30)
 416         self.assertEqual(sfr.read(5), '')
 417         self.assertEqual(sfr.tell(), 30)
 418
 419
 420 class StreamReaderTest(unittest.TestCase):
 421
 422     class MockKeep(object):
 423         def __init__(self, content):
 424             self.content = content
 425
 426         def get(self, locator):
 427             return self.content[locator]
 428
 429     def runTest(self):
 430         keepblocks = {'aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa+10': 'abcdefghij',
 431                       'bbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbb+15': 'klmnopqrstuvwxy',
 432                       'cccccccccccccccccccccccccccccccc+5': 'z0123'}
 433         mk = StreamReaderTest.MockKeep(keepblocks)
 434
 435         sr = arvados.StreamReader([".", "aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa+10", "bbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbb+15", "cccccccccccccccccccccccccccccccc+5", "0:30:foo"], mk)
 436
 437         content = 'abcdefghijklmnopqrstuvwxyz0123456789'
 438
 439         self.assertEqual(sr.readfrom(0, 30), content[0:30])
 440         self.assertEqual(sr.readfrom(2, 30), content[2:30])
 441
 442         self.assertEqual(sr.readfrom(2, 8), content[2:10])
 443         self.assertEqual(sr.readfrom(0, 10), content[0:10])
 444
 445         self.assertEqual(sr.readfrom(0, 5), content[0:5])
 446         self.assertEqual(sr.readfrom(5, 5), content[5:10])
 447         self.assertEqual(sr.readfrom(10, 5), content[10:15])
 448         self.assertEqual(sr.readfrom(15, 5), content[15:20])
 449         self.assertEqual(sr.readfrom(20, 5), content[20:25])
 450         self.assertEqual(sr.readfrom(25, 5), content[25:30])
 451         self.assertEqual(sr.readfrom(30, 5), '')
 452
 453 class ExtractFileTest(unittest.TestCase):
 454     def runTest(self):
 455         m1 = """. 5348b82a029fd9e971a811ce1f71360b+43 0:43:md5sum.txt
 456 . 085c37f02916da1cad16f93c54d899b7+41 0:41:md6sum.txt
 457 . 8b22da26f9f433dea0a10e5ec66d73ba+43 0:43:md7sum.txt
 458 . 085c37f02916da1cad16f93c54d899b7+41 5348b82a029fd9e971a811ce1f71360b+43 8b22da26f9f433dea0a10e5ec66d73ba+43 47:80:md8sum.txt
 459 . 085c37f02916da1cad16f93c54d899b7+41 5348b82a029fd9e971a811ce1f71360b+43 8b22da26f9f433dea0a10e5ec66d73ba+43 40:80:md9sum.txt"""
 460
 461         m2 = arvados.CollectionReader(m1)
 462
 463         self.assertEqual(m2.manifest_text(),
 464                          ". 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")
 465
 466         self.assertEqual(arvados.CollectionReader(m1).all_streams()[0].files()['md5sum.txt'].as_manifest(),
 467                          ". 5348b82a029fd9e971a811ce1f71360b+43 0:43:md5sum.txt\n")
 468         self.assertEqual(arvados.CollectionReader(m1).all_streams()[0].files()['md6sum.txt'].as_manifest(),
 469                          ". 085c37f02916da1cad16f93c54d899b7+41 0:41:md6sum.txt\n")
 470         self.assertEqual(arvados.CollectionReader(m1).all_streams()[0].files()['md7sum.txt'].as_manifest(),
 471                          ". 8b22da26f9f433dea0a10e5ec66d73ba+43 0:43:md7sum.txt\n")
 472         self.assertEqual(arvados.CollectionReader(m1).all_streams()[0].files()['md9sum.txt'].as_manifest(),
 473                          ". 085c37f02916da1cad16f93c54d899b7+41 5348b82a029fd9e971a811ce1f71360b+43 8b22da26f9f433dea0a10e5ec66d73ba+43 40:80:md9sum.txt\n")