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 import run_test_server
  15 from arvados_testutil import ArvadosBaseTestCase
  16
  17 class TestResumableWriter(arvados.ResumableCollectionWriter):
  18     KEEP_BLOCK_SIZE = 1024  # PUT to Keep every 1K.
  19
  20     def current_state(self):
  21         return self.dump_state(copy.deepcopy)
  22
  23
  24 class ArvadosCollectionsTest(run_test_server.TestCaseWithServers,
  25                              ArvadosBaseTestCase):
  26     MAIN_SERVER = {}
  27
  28     @classmethod
  29     def setUpClass(cls):
  30         super(ArvadosCollectionsTest, cls).setUpClass()
  31         run_test_server.authorize_with('active')
  32
  33     def write_foo_bar_baz(self):
  34         cw = arvados.CollectionWriter()
  35         self.assertEqual(cw.current_stream_name(), '.',
  36                          'current_stream_name() should be "." now')
  37         cw.set_current_file_name('foo.txt')
  38         cw.write('foo')
  39         self.assertEqual(cw.current_file_name(), 'foo.txt',
  40                          'current_file_name() should be foo.txt now')
  41         cw.start_new_file('bar.txt')
  42         cw.write('bar')
  43         cw.start_new_stream('baz')
  44         cw.write('baz')
  45         cw.set_current_file_name('baz.txt')
  46         return cw.finish()
  47
  48     def test_keep_local_store(self):
  49         self.assertEqual(arvados.Keep.put('foo'), 'acbd18db4cc2f85cedef654fccc4a4d8+3', 'wrong md5 hash from Keep.put')
  50         self.assertEqual(arvados.Keep.get('acbd18db4cc2f85cedef654fccc4a4d8+3'), 'foo', 'wrong data from Keep.get')
  51
  52     def test_local_collection_writer(self):
  53         self.assertEqual(self.write_foo_bar_baz(),
  54                          'd6c3b8e571f1b81ebb150a45ed06c884+114',
  55                          "wrong locator hash for files foo, bar, baz")
  56
  57     def test_local_collection_reader(self):
  58         self.write_foo_bar_baz()
  59         cr = arvados.CollectionReader('d6c3b8e571f1b81ebb150a45ed06c884+114+Xzizzle')
  60         got = []
  61         for s in cr.all_streams():
  62             for f in s.all_files():
  63                 got += [[f.size(), f.stream_name(), f.name(), f.read(2**26)]]
  64         expected = [[3, '.', 'bar.txt', 'bar'],
  65                     [3, '.', 'foo.txt', 'foo'],
  66                     [3, './baz', 'baz.txt', 'baz']]
  67         self.assertEqual(got,
  68                          expected)
  69         stream0 = cr.all_streams()[0]
  70         self.assertEqual(stream0.readfrom(0, 0),
  71                          '',
  72                          'reading zero bytes should have returned empty string')
  73         self.assertEqual(stream0.readfrom(0, 2**26),
  74                          'foobar',
  75                          'reading entire stream failed')
  76         self.assertEqual(stream0.readfrom(2**26, 0),
  77                          '',
  78                          'reading zero bytes should have returned empty string')
  79
  80     def _test_subset(self, collection, expected):
  81         cr = arvados.CollectionReader(collection)
  82         for s in cr.all_streams():
  83             for ex in expected:
  84                 if ex[0] == s:
  85                     f = s.files()[ex[2]]
  86                     got = [f.size(), f.stream_name(), f.name(), "".join(f.readall(2**26))]
  87                     self.assertEqual(got,
  88                                      ex,
  89                                      'all_files|as_manifest did not preserve manifest contents: got %s expected %s' % (got, ex))
  90
  91     def test_collection_manifest_subset(self):
  92         self.write_foo_bar_baz()
  93         self._test_subset('d6c3b8e571f1b81ebb150a45ed06c884+114',
  94                           [[3, '.',     'bar.txt', 'bar'],
  95                            [3, '.',     'foo.txt', 'foo'],
  96                            [3, './baz', 'baz.txt', 'baz']])
  97         self._test_subset((". %s %s 0:3:foo.txt 3:3:bar.txt\n" %
  98                            (arvados.Keep.put("foo"),
  99                             arvados.Keep.put("bar"))),
 100                           [[3, '.', 'bar.txt', 'bar'],
 101                            [3, '.', 'foo.txt', 'foo']])
 102         self._test_subset((". %s %s 0:2:fo.txt 2:4:obar.txt\n" %
 103                            (arvados.Keep.put("foo"),
 104                             arvados.Keep.put("bar"))),
 105                           [[2, '.', 'fo.txt', 'fo'],
 106                            [4, '.', 'obar.txt', 'obar']])
 107         self._test_subset((". %s %s 0:2:fo.txt 2:0:zero.txt 2:2:ob.txt 4:2:ar.txt\n" %
 108                            (arvados.Keep.put("foo"),
 109                             arvados.Keep.put("bar"))),
 110                           [[2, '.', 'ar.txt', 'ar'],
 111                            [2, '.', 'fo.txt', 'fo'],
 112                            [2, '.', 'ob.txt', 'ob'],
 113                            [0, '.', 'zero.txt', '']])
 114
 115     def _test_readline(self, what_in, what_out):
 116         cw = arvados.CollectionWriter()
 117         cw.start_new_file('test.txt')
 118         cw.write(what_in)
 119         test1 = cw.finish()
 120         cr = arvados.CollectionReader(test1)
 121         got = []
 122         for x in list(cr.all_files())[0].readlines():
 123             got += [x]
 124         self.assertEqual(got,
 125                          what_out,
 126                          "readlines did not split lines correctly: %s" % got)
 127
 128     def test_collection_readline(self):
 129         self._test_readline("\na\nbcd\n\nefg\nz",
 130                             ["\n", "a\n", "bcd\n", "\n", "efg\n", "z"])
 131         self._test_readline("ab\ncd\n",
 132                             ["ab\n", "cd\n"])
 133
 134     def test_collection_empty_file(self):
 135         cw = arvados.CollectionWriter()
 136         cw.start_new_file('zero.txt')
 137         cw.write('')
 138
 139         self.assertEqual(cw.manifest_text(), ". d41d8cd98f00b204e9800998ecf8427e+0 0:0:zero.txt\n")
 140         self.check_manifest_file_sizes(cw.manifest_text(), [0])
 141         cw = arvados.CollectionWriter()
 142         cw.start_new_file('zero.txt')
 143         cw.write('')
 144         cw.start_new_file('one.txt')
 145         cw.write('1')
 146         cw.start_new_stream('foo')
 147         cw.start_new_file('zero.txt')
 148         cw.write('')
 149         self.check_manifest_file_sizes(cw.manifest_text(), [1,0,0])
 150
 151     def check_manifest_file_sizes(self, manifest_text, expect_sizes):
 152         cr = arvados.CollectionReader(manifest_text)
 153         got_sizes = []
 154         for f in cr.all_files():
 155             got_sizes += [f.size()]
 156         self.assertEqual(got_sizes, expect_sizes, "got wrong file sizes %s, expected %s" % (got_sizes, expect_sizes))
 157
 158     def test_collection_bz2_decompression(self):
 159         n_lines_in = 2**18
 160         data_in = "abc\n"
 161         for x in xrange(0, 18):
 162             data_in += data_in
 163         compressed_data_in = bz2.compress(data_in)
 164         cw = arvados.CollectionWriter()
 165         cw.start_new_file('test.bz2')
 166         cw.write(compressed_data_in)
 167         bz2_manifest = cw.manifest_text()
 168
 169         cr = arvados.CollectionReader(bz2_manifest)
 170
 171         got = 0
 172         for x in list(cr.all_files())[0].readlines():
 173             self.assertEqual(x, "abc\n", "decompression returned wrong data: %s" % x)
 174             got += 1
 175         self.assertEqual(got,
 176                          n_lines_in,
 177                          "decompression returned %d lines instead of %d" % (got, n_lines_in))
 178
 179     def test_collection_gzip_decompression(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     def test_normalized_collection(self):
 206         m1 = """. 5348b82a029fd9e971a811ce1f71360b+43 0:43:md5sum.txt
 207 . 085c37f02916da1cad16f93c54d899b7+41 0:41:md5sum.txt
 208 . 8b22da26f9f433dea0a10e5ec66d73ba+43 0:43:md5sum.txt"""
 209         self.assertEqual(arvados.CollectionReader(m1).manifest_text(),
 210                          """. 5348b82a029fd9e971a811ce1f71360b+43 085c37f02916da1cad16f93c54d899b7+41 8b22da26f9f433dea0a10e5ec66d73ba+43 0:127:md5sum.txt
 211 """)
 212
 213         m2 = """. 204e43b8a1185621ca55a94839582e6f+67108864 b9677abbac956bd3e86b1deb28dfac03+67108864 fc15aff2a762b13f521baf042140acec+67108864 323d2a3ce20370c4ca1d3462a344f8fd+25885655 0:227212247:var-GS000016015-ASM.tsv.bz2
 214 """
 215         self.assertEqual(arvados.CollectionReader(m2).manifest_text(), m2)
 216
 217         m3 = """. 5348b82a029fd9e971a811ce1f71360b+43 3:40:md5sum.txt
 218 . 085c37f02916da1cad16f93c54d899b7+41 0:41:md5sum.txt
 219 . 8b22da26f9f433dea0a10e5ec66d73ba+43 0:43:md5sum.txt"""
 220         self.assertEqual(arvados.CollectionReader(m3).manifest_text(),
 221                          """. 5348b82a029fd9e971a811ce1f71360b+43 085c37f02916da1cad16f93c54d899b7+41 8b22da26f9f433dea0a10e5ec66d73ba+43 3:124:md5sum.txt
 222 """)
 223
 224         m4 = """. 204e43b8a1185621ca55a94839582e6f+67108864 0:3:foo/bar
 225 ./zzz 204e43b8a1185621ca55a94839582e6f+67108864 0:999:zzz
 226 ./foo 323d2a3ce20370c4ca1d3462a344f8fd+25885655 0:3:bar"""
 227         self.assertEqual(arvados.CollectionReader(m4).manifest_text(),
 228                          """./foo 204e43b8a1185621ca55a94839582e6f+67108864 323d2a3ce20370c4ca1d3462a344f8fd+25885655 0:3:bar 67108864:3:bar
 229 ./zzz 204e43b8a1185621ca55a94839582e6f+67108864 0:999:zzz
 230 """)
 231
 232         m5 = """. 204e43b8a1185621ca55a94839582e6f+67108864 0:3:foo/bar
 233 ./zzz 204e43b8a1185621ca55a94839582e6f+67108864 0:999:zzz
 234 ./foo 204e43b8a1185621ca55a94839582e6f+67108864 3:3:bar"""
 235         self.assertEqual(arvados.CollectionReader(m5).manifest_text(),
 236                          """./foo 204e43b8a1185621ca55a94839582e6f+67108864 0:6:bar
 237 ./zzz 204e43b8a1185621ca55a94839582e6f+67108864 0:999:zzz
 238 """)
 239
 240         with self.data_file('1000G_ref_manifest') as f6:
 241             m6 = f6.read()
 242             self.assertEqual(arvados.CollectionReader(m6).manifest_text(), m6)
 243
 244         with self.data_file('jlake_manifest') as f7:
 245             m7 = f7.read()
 246             self.assertEqual(arvados.CollectionReader(m7).manifest_text(), m7)
 247
 248         m8 = """./a\\040b\\040c 59ca0efa9f5633cb0371bbc0355478d8+13 0:13:hello\\040world.txt
 249 """
 250         self.assertEqual(arvados.CollectionReader(m8).manifest_text(), m8)
 251
 252     def test_locators_and_ranges(self):
 253         blocks2 = [['a', 10, 0],
 254                   ['b', 10, 10],
 255                   ['c', 10, 20],
 256                   ['d', 10, 30],
 257                   ['e', 10, 40],
 258                   ['f', 10, 50]]
 259
 260         self.assertEqual(arvados.locators_and_ranges(blocks2,  2,  2), [['a', 10, 2, 2]])
 261         self.assertEqual(arvados.locators_and_ranges(blocks2, 12, 2), [['b', 10, 2, 2]])
 262         self.assertEqual(arvados.locators_and_ranges(blocks2, 22, 2), [['c', 10, 2, 2]])
 263         self.assertEqual(arvados.locators_and_ranges(blocks2, 32, 2), [['d', 10, 2, 2]])
 264         self.assertEqual(arvados.locators_and_ranges(blocks2, 42, 2), [['e', 10, 2, 2]])
 265         self.assertEqual(arvados.locators_and_ranges(blocks2, 52, 2), [['f', 10, 2, 2]])
 266         self.assertEqual(arvados.locators_and_ranges(blocks2, 62, 2), [])
 267         self.assertEqual(arvados.locators_and_ranges(blocks2, -2, 2), [])
 268
 269         self.assertEqual(arvados.locators_and_ranges(blocks2,  0,  2), [['a', 10, 0, 2]])
 270         self.assertEqual(arvados.locators_and_ranges(blocks2, 10, 2), [['b', 10, 0, 2]])
 271         self.assertEqual(arvados.locators_and_ranges(blocks2, 20, 2), [['c', 10, 0, 2]])
 272         self.assertEqual(arvados.locators_and_ranges(blocks2, 30, 2), [['d', 10, 0, 2]])
 273         self.assertEqual(arvados.locators_and_ranges(blocks2, 40, 2), [['e', 10, 0, 2]])
 274         self.assertEqual(arvados.locators_and_ranges(blocks2, 50, 2), [['f', 10, 0, 2]])
 275         self.assertEqual(arvados.locators_and_ranges(blocks2, 60, 2), [])
 276         self.assertEqual(arvados.locators_and_ranges(blocks2, -2, 2), [])
 277
 278         self.assertEqual(arvados.locators_and_ranges(blocks2,  9,  2), [['a', 10, 9, 1], ['b', 10, 0, 1]])
 279         self.assertEqual(arvados.locators_and_ranges(blocks2, 19, 2), [['b', 10, 9, 1], ['c', 10, 0, 1]])
 280         self.assertEqual(arvados.locators_and_ranges(blocks2, 29, 2), [['c', 10, 9, 1], ['d', 10, 0, 1]])
 281         self.assertEqual(arvados.locators_and_ranges(blocks2, 39, 2), [['d', 10, 9, 1], ['e', 10, 0, 1]])
 282         self.assertEqual(arvados.locators_and_ranges(blocks2, 49, 2), [['e', 10, 9, 1], ['f', 10, 0, 1]])
 283         self.assertEqual(arvados.locators_and_ranges(blocks2, 59, 2), [['f', 10, 9, 1]])
 284
 285
 286         blocks3 = [['a', 10, 0],
 287                   ['b', 10, 10],
 288                   ['c', 10, 20],
 289                   ['d', 10, 30],
 290                   ['e', 10, 40],
 291                   ['f', 10, 50],
 292                   ['g', 10, 60]]
 293
 294         self.assertEqual(arvados.locators_and_ranges(blocks3,  2,  2), [['a', 10, 2, 2]])
 295         self.assertEqual(arvados.locators_and_ranges(blocks3, 12, 2), [['b', 10, 2, 2]])
 296         self.assertEqual(arvados.locators_and_ranges(blocks3, 22, 2), [['c', 10, 2, 2]])
 297         self.assertEqual(arvados.locators_and_ranges(blocks3, 32, 2), [['d', 10, 2, 2]])
 298         self.assertEqual(arvados.locators_and_ranges(blocks3, 42, 2), [['e', 10, 2, 2]])
 299         self.assertEqual(arvados.locators_and_ranges(blocks3, 52, 2), [['f', 10, 2, 2]])
 300         self.assertEqual(arvados.locators_and_ranges(blocks3, 62, 2), [['g', 10, 2, 2]])
 301
 302
 303         blocks = [['a', 10, 0],
 304                   ['b', 15, 10],
 305                   ['c', 5, 25]]
 306         self.assertEqual(arvados.locators_and_ranges(blocks, 1, 0), [])
 307         self.assertEqual(arvados.locators_and_ranges(blocks, 0, 5), [['a', 10, 0, 5]])
 308         self.assertEqual(arvados.locators_and_ranges(blocks, 3, 5), [['a', 10, 3, 5]])
 309         self.assertEqual(arvados.locators_and_ranges(blocks, 0, 10), [['a', 10, 0, 10]])
 310
 311         self.assertEqual(arvados.locators_and_ranges(blocks, 0, 11), [['a', 10, 0, 10],
 312                                                                       ['b', 15, 0, 1]])
 313         self.assertEqual(arvados.locators_and_ranges(blocks, 1, 11), [['a', 10, 1, 9],
 314                                                                       ['b', 15, 0, 2]])
 315         self.assertEqual(arvados.locators_and_ranges(blocks, 0, 25), [['a', 10, 0, 10],
 316                                                                       ['b', 15, 0, 15]])
 317
 318         self.assertEqual(arvados.locators_and_ranges(blocks, 0, 30), [['a', 10, 0, 10],
 319                                                                       ['b', 15, 0, 15],
 320                                                                       ['c', 5, 0, 5]])
 321         self.assertEqual(arvados.locators_and_ranges(blocks, 1, 30), [['a', 10, 1, 9],
 322                                                                       ['b', 15, 0, 15],
 323                                                                       ['c', 5, 0, 5]])
 324         self.assertEqual(arvados.locators_and_ranges(blocks, 0, 31), [['a', 10, 0, 10],
 325                                                                       ['b', 15, 0, 15],
 326                                                                       ['c', 5, 0, 5]])
 327
 328         self.assertEqual(arvados.locators_and_ranges(blocks, 15, 5), [['b', 15, 5, 5]])
 329
 330         self.assertEqual(arvados.locators_and_ranges(blocks, 8, 17), [['a', 10, 8, 2],
 331                                                                       ['b', 15, 0, 15]])
 332
 333         self.assertEqual(arvados.locators_and_ranges(blocks, 8, 20), [['a', 10, 8, 2],
 334                                                                       ['b', 15, 0, 15],
 335                                                                       ['c', 5, 0, 3]])
 336
 337         self.assertEqual(arvados.locators_and_ranges(blocks, 26, 2), [['c', 5, 1, 2]])
 338
 339         self.assertEqual(arvados.locators_and_ranges(blocks, 9, 15), [['a', 10, 9, 1],
 340                                                                       ['b', 15, 0, 14]])
 341         self.assertEqual(arvados.locators_and_ranges(blocks, 10, 15), [['b', 15, 0, 15]])
 342         self.assertEqual(arvados.locators_and_ranges(blocks, 11, 15), [['b', 15, 1, 14],
 343                                                                        ['c', 5, 0, 1]])
 344
 345     class MockStreamReader(object):
 346         def __init__(self, content):
 347             self.content = content
 348
 349         def readfrom(self, start, size):
 350             return self.content[start:start+size]
 351
 352     def test_file_stream(self):
 353         content = 'abcdefghijklmnopqrstuvwxyz0123456789'
 354         msr = self.MockStreamReader(content)
 355         segments = [[0, 10, 0],
 356                     [10, 15, 10],
 357                     [25, 5, 25]]
 358
 359         sfr = arvados.StreamFileReader(msr, segments, "test")
 360
 361         self.assertEqual(sfr.name(), "test")
 362         self.assertEqual(sfr.size(), 30)
 363
 364         self.assertEqual(sfr.readfrom(0, 30), content[0:30])
 365         self.assertEqual(sfr.readfrom(2, 30), content[2:30])
 366
 367         self.assertEqual(sfr.readfrom(2, 8), content[2:10])
 368         self.assertEqual(sfr.readfrom(0, 10), content[0:10])
 369
 370         self.assertEqual(sfr.tell(), 0)
 371         self.assertEqual(sfr.read(5), content[0:5])
 372         self.assertEqual(sfr.tell(), 5)
 373         self.assertEqual(sfr.read(5), content[5:10])
 374         self.assertEqual(sfr.tell(), 10)
 375         self.assertEqual(sfr.read(5), content[10:15])
 376         self.assertEqual(sfr.tell(), 15)
 377         self.assertEqual(sfr.read(5), content[15:20])
 378         self.assertEqual(sfr.tell(), 20)
 379         self.assertEqual(sfr.read(5), content[20:25])
 380         self.assertEqual(sfr.tell(), 25)
 381         self.assertEqual(sfr.read(5), content[25:30])
 382         self.assertEqual(sfr.tell(), 30)
 383         self.assertEqual(sfr.read(5), '')
 384         self.assertEqual(sfr.tell(), 30)
 385
 386         segments = [[26, 10, 0],
 387                     [0, 15, 10],
 388                     [15, 5, 25]]
 389
 390         sfr = arvados.StreamFileReader(msr, segments, "test")
 391
 392         self.assertEqual(sfr.size(), 30)
 393
 394         self.assertEqual(sfr.readfrom(0, 30), content[26:36] + content[0:20])
 395         self.assertEqual(sfr.readfrom(2, 30), content[28:36] + content[0:20])
 396
 397         self.assertEqual(sfr.readfrom(2, 8), content[28:36])
 398         self.assertEqual(sfr.readfrom(0, 10), content[26:36])
 399
 400         self.assertEqual(sfr.tell(), 0)
 401         self.assertEqual(sfr.read(5), content[26:31])
 402         self.assertEqual(sfr.tell(), 5)
 403         self.assertEqual(sfr.read(5), content[31:36])
 404         self.assertEqual(sfr.tell(), 10)
 405         self.assertEqual(sfr.read(5), content[0:5])
 406         self.assertEqual(sfr.tell(), 15)
 407         self.assertEqual(sfr.read(5), content[5:10])
 408         self.assertEqual(sfr.tell(), 20)
 409         self.assertEqual(sfr.read(5), content[10:15])
 410         self.assertEqual(sfr.tell(), 25)
 411         self.assertEqual(sfr.read(5), content[15:20])
 412         self.assertEqual(sfr.tell(), 30)
 413         self.assertEqual(sfr.read(5), '')
 414         self.assertEqual(sfr.tell(), 30)
 415
 416
 417     class MockKeep(object):
 418         def __init__(self, content):
 419             self.content = content
 420
 421         def get(self, locator):
 422             return self.content[locator]
 423
 424     def test_stream_reader(self):
 425         keepblocks = {'aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa+10': 'abcdefghij',
 426                       'bbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbb+15': 'klmnopqrstuvwxy',
 427                       'cccccccccccccccccccccccccccccccc+5': 'z0123'}
 428         mk = self.MockKeep(keepblocks)
 429
 430         sr = arvados.StreamReader([".", "aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa+10", "bbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbb+15", "cccccccccccccccccccccccccccccccc+5", "0:30:foo"], mk)
 431
 432         content = 'abcdefghijklmnopqrstuvwxyz0123456789'
 433
 434         self.assertEqual(sr.readfrom(0, 30), content[0:30])
 435         self.assertEqual(sr.readfrom(2, 30), content[2:30])
 436
 437         self.assertEqual(sr.readfrom(2, 8), content[2:10])
 438         self.assertEqual(sr.readfrom(0, 10), content[0:10])
 439
 440         self.assertEqual(sr.readfrom(0, 5), content[0:5])
 441         self.assertEqual(sr.readfrom(5, 5), content[5:10])
 442         self.assertEqual(sr.readfrom(10, 5), content[10:15])
 443         self.assertEqual(sr.readfrom(15, 5), content[15:20])
 444         self.assertEqual(sr.readfrom(20, 5), content[20:25])
 445         self.assertEqual(sr.readfrom(25, 5), content[25:30])
 446         self.assertEqual(sr.readfrom(30, 5), '')
 447
 448     def test_file_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", "0:10:foo", "15:10:foo"], mk)
 455
 456         content = 'abcdefghijpqrstuvwxy'
 457
 458         f = sr.files()["foo"]
 459
 460         f.seek(0)
 461         self.assertEqual(f.read(20), content[0:20])
 462
 463         f.seek(0)
 464         self.assertEqual(f.read(6), content[0:6])
 465         self.assertEqual(f.read(6), content[6:12])
 466         self.assertEqual(f.read(6), content[12:18])
 467
 468     def test_extract_file(self):
 469         m1 = """. 5348b82a029fd9e971a811ce1f71360b+43 0:43:md5sum.txt
 470 . 085c37f02916da1cad16f93c54d899b7+41 0:41:md6sum.txt
 471 . 8b22da26f9f433dea0a10e5ec66d73ba+43 0:43:md7sum.txt
 472 . 085c37f02916da1cad16f93c54d899b7+41 5348b82a029fd9e971a811ce1f71360b+43 8b22da26f9f433dea0a10e5ec66d73ba+43 47:80:md8sum.txt
 473 . 085c37f02916da1cad16f93c54d899b7+41 5348b82a029fd9e971a811ce1f71360b+43 8b22da26f9f433dea0a10e5ec66d73ba+43 40:80:md9sum.txt"""
 474
 475         m2 = arvados.CollectionReader(m1)
 476
 477         self.assertEqual(m2.manifest_text(),
 478                          ". 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")
 479
 480         self.assertEqual(arvados.CollectionReader(m1).all_streams()[0].files()['md5sum.txt'].as_manifest(),
 481                          ". 5348b82a029fd9e971a811ce1f71360b+43 0:43:md5sum.txt\n")
 482         self.assertEqual(arvados.CollectionReader(m1).all_streams()[0].files()['md6sum.txt'].as_manifest(),
 483                          ". 085c37f02916da1cad16f93c54d899b7+41 0:41:md6sum.txt\n")
 484         self.assertEqual(arvados.CollectionReader(m1).all_streams()[0].files()['md7sum.txt'].as_manifest(),
 485                          ". 8b22da26f9f433dea0a10e5ec66d73ba+43 0:43:md7sum.txt\n")
 486         self.assertEqual(arvados.CollectionReader(m1).all_streams()[0].files()['md9sum.txt'].as_manifest(),
 487                          ". 085c37f02916da1cad16f93c54d899b7+41 5348b82a029fd9e971a811ce1f71360b+43 8b22da26f9f433dea0a10e5ec66d73ba+43 40:80:md9sum.txt\n")
 488
 489     def test_write_directory_tree(self):
 490         cwriter = arvados.CollectionWriter()
 491         cwriter.write_directory_tree(self.build_directory_tree(
 492                 ['basefile', 'subdir/subfile']))
 493         self.assertEqual(cwriter.manifest_text(),
 494                          """. c5110c5ac93202d8e0f9e381f22bac0f+8 0:8:basefile
 495 ./subdir 1ca4dec89403084bf282ad31e6cf7972+14 0:14:subfile\n""")
 496
 497     def test_write_named_directory_tree(self):
 498         cwriter = arvados.CollectionWriter()
 499         cwriter.write_directory_tree(self.build_directory_tree(
 500                 ['basefile', 'subdir/subfile']), 'root')
 501         self.assertEqual(
 502             cwriter.manifest_text(),
 503             """./root c5110c5ac93202d8e0f9e381f22bac0f+8 0:8:basefile
 504 ./root/subdir 1ca4dec89403084bf282ad31e6cf7972+14 0:14:subfile\n""")
 505
 506     def test_write_directory_tree_in_one_stream(self):
 507         cwriter = arvados.CollectionWriter()
 508         cwriter.write_directory_tree(self.build_directory_tree(
 509                 ['basefile', 'subdir/subfile']), max_manifest_depth=0)
 510         self.assertEqual(cwriter.manifest_text(),
 511                          """. 4ace875ffdc6824a04950f06858f4465+22 0:8:basefile
 512 ./subdir 4ace875ffdc6824a04950f06858f4465+22 8:14:subfile\n""")
 513
 514     def test_write_directory_tree_with_limited_recursion(self):
 515         cwriter = arvados.CollectionWriter()
 516         cwriter.write_directory_tree(
 517             self.build_directory_tree(['f1', 'd1/f2', 'd1/d2/f3']),
 518             max_manifest_depth=1)
 519         self.assertEqual(cwriter.manifest_text(),
 520                          """. bd19836ddb62c11c55ab251ccaca5645+2 0:2:f1
 521 ./d1 50170217e5b04312024aa5cd42934494+13 8:5:f2
 522 ./d1/d2 50170217e5b04312024aa5cd42934494+13 0:8:f3\n""")
 523
 524     def test_write_one_file(self):
 525         cwriter = arvados.CollectionWriter()
 526         with self.make_test_file() as testfile:
 527             cwriter.write_file(testfile.name)
 528             self.assertEqual(
 529                 cwriter.manifest_text(),
 530                 ". 098f6bcd4621d373cade4e832627b4f6+4 0:4:{}\n".format(
 531                     os.path.basename(testfile.name)))
 532
 533     def test_write_named_file(self):
 534         cwriter = arvados.CollectionWriter()
 535         with self.make_test_file() as testfile:
 536             cwriter.write_file(testfile.name, 'foo')
 537             self.assertEqual(cwriter.manifest_text(),
 538                              ". 098f6bcd4621d373cade4e832627b4f6+4 0:4:foo\n")
 539
 540     def test_write_multiple_files(self):
 541         cwriter = arvados.CollectionWriter()
 542         for letter in 'ABC':
 543             with self.make_test_file(letter) as testfile:
 544                 cwriter.write_file(testfile.name, letter)
 545         self.assertEqual(
 546             cwriter.manifest_text(),
 547             ". 902fbdd2b1df0c4f70b4a5d23525e932+3 0:1:A 1:1:B 2:1:C\n")
 548
 549     def test_basic_resume(self):
 550         cwriter = TestResumableWriter()
 551         with self.make_test_file() as testfile:
 552             cwriter.write_file(testfile.name, 'test')
 553             resumed = TestResumableWriter.from_state(cwriter.current_state())
 554         self.assertEquals(cwriter.manifest_text(), resumed.manifest_text(),
 555                           "resumed CollectionWriter had different manifest")
 556
 557     def test_resume_fails_when_missing_dependency(self):
 558         cwriter = TestResumableWriter()
 559         with self.make_test_file() as testfile:
 560             cwriter.write_file(testfile.name, 'test')
 561         self.assertRaises(arvados.errors.StaleWriterStateError,
 562                           TestResumableWriter.from_state,
 563                           cwriter.current_state())
 564
 565     def test_resume_fails_when_dependency_mtime_changed(self):
 566         cwriter = TestResumableWriter()
 567         with self.make_test_file() as testfile:
 568             cwriter.write_file(testfile.name, 'test')
 569             os.utime(testfile.name, (0, 0))
 570             self.assertRaises(arvados.errors.StaleWriterStateError,
 571                               TestResumableWriter.from_state,
 572                               cwriter.current_state())
 573
 574     def test_resume_fails_when_dependency_is_nonfile(self):
 575         cwriter = TestResumableWriter()
 576         cwriter.write_file('/dev/null', 'empty')
 577         self.assertRaises(arvados.errors.StaleWriterStateError,
 578                           TestResumableWriter.from_state,
 579                           cwriter.current_state())
 580
 581     def test_resume_fails_when_dependency_size_changed(self):
 582         cwriter = TestResumableWriter()
 583         with self.make_test_file() as testfile:
 584             cwriter.write_file(testfile.name, 'test')
 585             orig_mtime = os.fstat(testfile.fileno()).st_mtime
 586             testfile.write('extra')
 587             testfile.flush()
 588             os.utime(testfile.name, (orig_mtime, orig_mtime))
 589             self.assertRaises(arvados.errors.StaleWriterStateError,
 590                               TestResumableWriter.from_state,
 591                               cwriter.current_state())
 592
 593     def test_resume_fails_with_expired_locator(self):
 594         cwriter = TestResumableWriter()
 595         state = cwriter.current_state()
 596         # Add an expired locator to the state.
 597         state['_current_stream_locators'].append(''.join([
 598                     'a' * 32, '+1+A', 'b' * 40, '@', '10000000']))
 599         self.assertRaises(arvados.errors.StaleWriterStateError,
 600                           TestResumableWriter.from_state, state)
 601
 602     def test_arbitrary_objects_not_resumable(self):
 603         cwriter = TestResumableWriter()
 604         with open('/dev/null') as badfile:
 605             self.assertRaises(arvados.errors.AssertionError,
 606                               cwriter.write_file, badfile)
 607
 608     def test_arbitrary_writes_not_resumable(self):
 609         cwriter = TestResumableWriter()
 610         self.assertRaises(arvados.errors.AssertionError,
 611                           cwriter.write, "badtext")
 612
 613
 614 if __name__ == '__main__':
 615     unittest.main()