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         cls.api_client = arvados.api('v1')
  33         cls.keep_client = arvados.KeepClient(api_client=cls.api_client,
  34                                              local_store=cls.local_store)
  35
  36     def write_foo_bar_baz(self):
  37         cw = arvados.CollectionWriter(self.api_client)
  38         self.assertEqual(cw.current_stream_name(), '.',
  39                          'current_stream_name() should be "." now')
  40         cw.set_current_file_name('foo.txt')
  41         cw.write('foo')
  42         self.assertEqual(cw.current_file_name(), 'foo.txt',
  43                          'current_file_name() should be foo.txt now')
  44         cw.start_new_file('bar.txt')
  45         cw.write('bar')
  46         cw.start_new_stream('baz')
  47         cw.write('baz')
  48         cw.set_current_file_name('baz.txt')
  49         return cw.finish()
  50
  51     def test_keep_local_store(self):
  52         self.assertEqual(self.keep_client.put('foo'), 'acbd18db4cc2f85cedef654fccc4a4d8+3', 'wrong md5 hash from Keep.put')
  53         self.assertEqual(self.keep_client.get('acbd18db4cc2f85cedef654fccc4a4d8+3'), 'foo', 'wrong data from Keep.get')
  54
  55     def test_local_collection_writer(self):
  56         self.assertEqual(self.write_foo_bar_baz(),
  57                          'd6c3b8e571f1b81ebb150a45ed06c884+114',
  58                          "wrong locator hash for files foo, bar, baz")
  59
  60     def test_local_collection_reader(self):
  61         self.write_foo_bar_baz()
  62         cr = arvados.CollectionReader(
  63             'd6c3b8e571f1b81ebb150a45ed06c884+114+Xzizzle', self.api_client)
  64         got = []
  65         for s in cr.all_streams():
  66             for f in s.all_files():
  67                 got += [[f.size(), f.stream_name(), f.name(), f.read(2**26)]]
  68         expected = [[3, '.', 'bar.txt', 'bar'],
  69                     [3, '.', 'foo.txt', 'foo'],
  70                     [3, './baz', 'baz.txt', 'baz']]
  71         self.assertEqual(got,
  72                          expected)
  73         stream0 = cr.all_streams()[0]
  74         self.assertEqual(stream0.readfrom(0, 0),
  75                          '',
  76                          'reading zero bytes should have returned empty string')
  77         self.assertEqual(stream0.readfrom(0, 2**26),
  78                          'foobar',
  79                          'reading entire stream failed')
  80         self.assertEqual(stream0.readfrom(2**26, 0),
  81                          '',
  82                          'reading zero bytes should have returned empty string')
  83
  84     def _test_subset(self, collection, expected):
  85         cr = arvados.CollectionReader(collection, self.api_client)
  86         for s in cr.all_streams():
  87             for ex in expected:
  88                 if ex[0] == s:
  89                     f = s.files()[ex[2]]
  90                     got = [f.size(), f.stream_name(), f.name(), "".join(f.readall(2**26))]
  91                     self.assertEqual(got,
  92                                      ex,
  93                                      'all_files|as_manifest did not preserve manifest contents: got %s expected %s' % (got, ex))
  94
  95     def test_collection_manifest_subset(self):
  96         self.write_foo_bar_baz()
  97         self._test_subset('d6c3b8e571f1b81ebb150a45ed06c884+114',
  98                           [[3, '.',     'bar.txt', 'bar'],
  99                            [3, '.',     'foo.txt', 'foo'],
 100                            [3, './baz', 'baz.txt', 'baz']])
 101         self._test_subset((". %s %s 0:3:foo.txt 3:3:bar.txt\n" %
 102                            (self.keep_client.put("foo"),
 103                             self.keep_client.put("bar"))),
 104                           [[3, '.', 'bar.txt', 'bar'],
 105                            [3, '.', 'foo.txt', 'foo']])
 106         self._test_subset((". %s %s 0:2:fo.txt 2:4:obar.txt\n" %
 107                            (self.keep_client.put("foo"),
 108                             self.keep_client.put("bar"))),
 109                           [[2, '.', 'fo.txt', 'fo'],
 110                            [4, '.', 'obar.txt', 'obar']])
 111         self._test_subset((". %s %s 0:2:fo.txt 2:0:zero.txt 2:2:ob.txt 4:2:ar.txt\n" %
 112                            (self.keep_client.put("foo"),
 113                             self.keep_client.put("bar"))),
 114                           [[2, '.', 'ar.txt', 'ar'],
 115                            [2, '.', 'fo.txt', 'fo'],
 116                            [2, '.', 'ob.txt', 'ob'],
 117                            [0, '.', 'zero.txt', '']])
 118
 119     def _test_readline(self, what_in, what_out):
 120         cw = arvados.CollectionWriter(self.api_client)
 121         cw.start_new_file('test.txt')
 122         cw.write(what_in)
 123         test1 = cw.finish()
 124         cr = arvados.CollectionReader(test1, self.api_client)
 125         got = []
 126         for x in list(cr.all_files())[0].readlines():
 127             got += [x]
 128         self.assertEqual(got,
 129                          what_out,
 130                          "readlines did not split lines correctly: %s" % got)
 131
 132     def test_collection_readline(self):
 133         self._test_readline("\na\nbcd\n\nefg\nz",
 134                             ["\n", "a\n", "bcd\n", "\n", "efg\n", "z"])
 135         self._test_readline("ab\ncd\n",
 136                             ["ab\n", "cd\n"])
 137
 138     def test_collection_empty_file(self):
 139         cw = arvados.CollectionWriter(self.api_client)
 140         cw.start_new_file('zero.txt')
 141         cw.write('')
 142
 143         self.assertEqual(cw.manifest_text(), ". d41d8cd98f00b204e9800998ecf8427e+0 0:0:zero.txt\n")
 144         self.check_manifest_file_sizes(cw.manifest_text(), [0])
 145         cw = arvados.CollectionWriter(self.api_client)
 146         cw.start_new_file('zero.txt')
 147         cw.write('')
 148         cw.start_new_file('one.txt')
 149         cw.write('1')
 150         cw.start_new_stream('foo')
 151         cw.start_new_file('zero.txt')
 152         cw.write('')
 153         self.check_manifest_file_sizes(cw.manifest_text(), [1,0,0])
 154
 155     def check_manifest_file_sizes(self, manifest_text, expect_sizes):
 156         cr = arvados.CollectionReader(manifest_text, self.api_client)
 157         got_sizes = []
 158         for f in cr.all_files():
 159             got_sizes += [f.size()]
 160         self.assertEqual(got_sizes, expect_sizes, "got wrong file sizes %s, expected %s" % (got_sizes, expect_sizes))
 161
 162     def test_collection_bz2_decompression(self):
 163         n_lines_in = 2**18
 164         data_in = "abc\n"
 165         for x in xrange(0, 18):
 166             data_in += data_in
 167         compressed_data_in = bz2.compress(data_in)
 168         cw = arvados.CollectionWriter(self.api_client)
 169         cw.start_new_file('test.bz2')
 170         cw.write(compressed_data_in)
 171         bz2_manifest = cw.manifest_text()
 172
 173         cr = arvados.CollectionReader(bz2_manifest, self.api_client)
 174
 175         got = 0
 176         for x in list(cr.all_files())[0].readlines():
 177             self.assertEqual(x, "abc\n", "decompression returned wrong data: %s" % x)
 178             got += 1
 179         self.assertEqual(got,
 180                          n_lines_in,
 181                          "decompression returned %d lines instead of %d" % (got, n_lines_in))
 182
 183     def test_collection_gzip_decompression(self):
 184         n_lines_in = 2**18
 185         data_in = "abc\n"
 186         for x in xrange(0, 18):
 187             data_in += data_in
 188         p = subprocess.Popen(["gzip", "-1cn"],
 189                              stdout=subprocess.PIPE,
 190                              stdin=subprocess.PIPE,
 191                              stderr=subprocess.PIPE,
 192                              shell=False, close_fds=True)
 193         compressed_data_in, stderrdata = p.communicate(data_in)
 194
 195         cw = arvados.CollectionWriter(self.api_client)
 196         cw.start_new_file('test.gz')
 197         cw.write(compressed_data_in)
 198         gzip_manifest = cw.manifest_text()
 199
 200         cr = arvados.CollectionReader(gzip_manifest, self.api_client)
 201         got = 0
 202         for x in list(cr.all_files())[0].readlines():
 203             self.assertEqual(x, "abc\n", "decompression returned wrong data: %s" % x)
 204             got += 1
 205         self.assertEqual(got,
 206                          n_lines_in,
 207                          "decompression returned %d lines instead of %d" % (got, n_lines_in))
 208
 209     def test_normalized_collection(self):
 210         m1 = """. 5348b82a029fd9e971a811ce1f71360b+43 0:43:md5sum.txt
 211 . 085c37f02916da1cad16f93c54d899b7+41 0:41:md5sum.txt
 212 . 8b22da26f9f433dea0a10e5ec66d73ba+43 0:43:md5sum.txt"""
 213         self.assertEqual(arvados.CollectionReader(m1, self.api_client).manifest_text(),
 214                          """. 5348b82a029fd9e971a811ce1f71360b+43 085c37f02916da1cad16f93c54d899b7+41 8b22da26f9f433dea0a10e5ec66d73ba+43 0:127:md5sum.txt
 215 """)
 216
 217         m2 = """. 204e43b8a1185621ca55a94839582e6f+67108864 b9677abbac956bd3e86b1deb28dfac03+67108864 fc15aff2a762b13f521baf042140acec+67108864 323d2a3ce20370c4ca1d3462a344f8fd+25885655 0:227212247:var-GS000016015-ASM.tsv.bz2
 218 """
 219         self.assertEqual(arvados.CollectionReader(m2, self.api_client).manifest_text(), m2)
 220
 221         m3 = """. 5348b82a029fd9e971a811ce1f71360b+43 3:40:md5sum.txt
 222 . 085c37f02916da1cad16f93c54d899b7+41 0:41:md5sum.txt
 223 . 8b22da26f9f433dea0a10e5ec66d73ba+43 0:43:md5sum.txt"""
 224         self.assertEqual(arvados.CollectionReader(m3, self.api_client).manifest_text(),
 225                          """. 5348b82a029fd9e971a811ce1f71360b+43 085c37f02916da1cad16f93c54d899b7+41 8b22da26f9f433dea0a10e5ec66d73ba+43 3:124:md5sum.txt
 226 """)
 227
 228         m4 = """. 204e43b8a1185621ca55a94839582e6f+67108864 0:3:foo/bar
 229 ./zzz 204e43b8a1185621ca55a94839582e6f+67108864 0:999:zzz
 230 ./foo 323d2a3ce20370c4ca1d3462a344f8fd+25885655 0:3:bar"""
 231         self.assertEqual(arvados.CollectionReader(m4, self.api_client).manifest_text(),
 232                          """./foo 204e43b8a1185621ca55a94839582e6f+67108864 323d2a3ce20370c4ca1d3462a344f8fd+25885655 0:3:bar 67108864:3:bar
 233 ./zzz 204e43b8a1185621ca55a94839582e6f+67108864 0:999:zzz
 234 """)
 235
 236         m5 = """. 204e43b8a1185621ca55a94839582e6f+67108864 0:3:foo/bar
 237 ./zzz 204e43b8a1185621ca55a94839582e6f+67108864 0:999:zzz
 238 ./foo 204e43b8a1185621ca55a94839582e6f+67108864 3:3:bar"""
 239         self.assertEqual(arvados.CollectionReader(m5, self.api_client).manifest_text(),
 240                          """./foo 204e43b8a1185621ca55a94839582e6f+67108864 0:6:bar
 241 ./zzz 204e43b8a1185621ca55a94839582e6f+67108864 0:999:zzz
 242 """)
 243
 244         with self.data_file('1000G_ref_manifest') as f6:
 245             m6 = f6.read()
 246             self.assertEqual(arvados.CollectionReader(m6, self.api_client).manifest_text(), m6)
 247
 248         with self.data_file('jlake_manifest') as f7:
 249             m7 = f7.read()
 250             self.assertEqual(arvados.CollectionReader(m7, self.api_client).manifest_text(), m7)
 251
 252         m8 = """./a\\040b\\040c 59ca0efa9f5633cb0371bbc0355478d8+13 0:13:hello\\040world.txt
 253 """
 254         self.assertEqual(arvados.CollectionReader(m8, self.api_client).manifest_text(), m8)
 255
 256     def test_locators_and_ranges(self):
 257         blocks2 = [['a', 10, 0],
 258                   ['b', 10, 10],
 259                   ['c', 10, 20],
 260                   ['d', 10, 30],
 261                   ['e', 10, 40],
 262                   ['f', 10, 50]]
 263
 264         self.assertEqual(arvados.locators_and_ranges(blocks2,  2,  2), [['a', 10, 2, 2]])
 265         self.assertEqual(arvados.locators_and_ranges(blocks2, 12, 2), [['b', 10, 2, 2]])
 266         self.assertEqual(arvados.locators_and_ranges(blocks2, 22, 2), [['c', 10, 2, 2]])
 267         self.assertEqual(arvados.locators_and_ranges(blocks2, 32, 2), [['d', 10, 2, 2]])
 268         self.assertEqual(arvados.locators_and_ranges(blocks2, 42, 2), [['e', 10, 2, 2]])
 269         self.assertEqual(arvados.locators_and_ranges(blocks2, 52, 2), [['f', 10, 2, 2]])
 270         self.assertEqual(arvados.locators_and_ranges(blocks2, 62, 2), [])
 271         self.assertEqual(arvados.locators_and_ranges(blocks2, -2, 2), [])
 272
 273         self.assertEqual(arvados.locators_and_ranges(blocks2,  0,  2), [['a', 10, 0, 2]])
 274         self.assertEqual(arvados.locators_and_ranges(blocks2, 10, 2), [['b', 10, 0, 2]])
 275         self.assertEqual(arvados.locators_and_ranges(blocks2, 20, 2), [['c', 10, 0, 2]])
 276         self.assertEqual(arvados.locators_and_ranges(blocks2, 30, 2), [['d', 10, 0, 2]])
 277         self.assertEqual(arvados.locators_and_ranges(blocks2, 40, 2), [['e', 10, 0, 2]])
 278         self.assertEqual(arvados.locators_and_ranges(blocks2, 50, 2), [['f', 10, 0, 2]])
 279         self.assertEqual(arvados.locators_and_ranges(blocks2, 60, 2), [])
 280         self.assertEqual(arvados.locators_and_ranges(blocks2, -2, 2), [])
 281
 282         self.assertEqual(arvados.locators_and_ranges(blocks2,  9,  2), [['a', 10, 9, 1], ['b', 10, 0, 1]])
 283         self.assertEqual(arvados.locators_and_ranges(blocks2, 19, 2), [['b', 10, 9, 1], ['c', 10, 0, 1]])
 284         self.assertEqual(arvados.locators_and_ranges(blocks2, 29, 2), [['c', 10, 9, 1], ['d', 10, 0, 1]])
 285         self.assertEqual(arvados.locators_and_ranges(blocks2, 39, 2), [['d', 10, 9, 1], ['e', 10, 0, 1]])
 286         self.assertEqual(arvados.locators_and_ranges(blocks2, 49, 2), [['e', 10, 9, 1], ['f', 10, 0, 1]])
 287         self.assertEqual(arvados.locators_and_ranges(blocks2, 59, 2), [['f', 10, 9, 1]])
 288
 289
 290         blocks3 = [['a', 10, 0],
 291                   ['b', 10, 10],
 292                   ['c', 10, 20],
 293                   ['d', 10, 30],
 294                   ['e', 10, 40],
 295                   ['f', 10, 50],
 296                   ['g', 10, 60]]
 297
 298         self.assertEqual(arvados.locators_and_ranges(blocks3,  2,  2), [['a', 10, 2, 2]])
 299         self.assertEqual(arvados.locators_and_ranges(blocks3, 12, 2), [['b', 10, 2, 2]])
 300         self.assertEqual(arvados.locators_and_ranges(blocks3, 22, 2), [['c', 10, 2, 2]])
 301         self.assertEqual(arvados.locators_and_ranges(blocks3, 32, 2), [['d', 10, 2, 2]])
 302         self.assertEqual(arvados.locators_and_ranges(blocks3, 42, 2), [['e', 10, 2, 2]])
 303         self.assertEqual(arvados.locators_and_ranges(blocks3, 52, 2), [['f', 10, 2, 2]])
 304         self.assertEqual(arvados.locators_and_ranges(blocks3, 62, 2), [['g', 10, 2, 2]])
 305
 306
 307         blocks = [['a', 10, 0],
 308                   ['b', 15, 10],
 309                   ['c', 5, 25]]
 310         self.assertEqual(arvados.locators_and_ranges(blocks, 1, 0), [])
 311         self.assertEqual(arvados.locators_and_ranges(blocks, 0, 5), [['a', 10, 0, 5]])
 312         self.assertEqual(arvados.locators_and_ranges(blocks, 3, 5), [['a', 10, 3, 5]])
 313         self.assertEqual(arvados.locators_and_ranges(blocks, 0, 10), [['a', 10, 0, 10]])
 314
 315         self.assertEqual(arvados.locators_and_ranges(blocks, 0, 11), [['a', 10, 0, 10],
 316                                                                       ['b', 15, 0, 1]])
 317         self.assertEqual(arvados.locators_and_ranges(blocks, 1, 11), [['a', 10, 1, 9],
 318                                                                       ['b', 15, 0, 2]])
 319         self.assertEqual(arvados.locators_and_ranges(blocks, 0, 25), [['a', 10, 0, 10],
 320                                                                       ['b', 15, 0, 15]])
 321
 322         self.assertEqual(arvados.locators_and_ranges(blocks, 0, 30), [['a', 10, 0, 10],
 323                                                                       ['b', 15, 0, 15],
 324                                                                       ['c', 5, 0, 5]])
 325         self.assertEqual(arvados.locators_and_ranges(blocks, 1, 30), [['a', 10, 1, 9],
 326                                                                       ['b', 15, 0, 15],
 327                                                                       ['c', 5, 0, 5]])
 328         self.assertEqual(arvados.locators_and_ranges(blocks, 0, 31), [['a', 10, 0, 10],
 329                                                                       ['b', 15, 0, 15],
 330                                                                       ['c', 5, 0, 5]])
 331
 332         self.assertEqual(arvados.locators_and_ranges(blocks, 15, 5), [['b', 15, 5, 5]])
 333
 334         self.assertEqual(arvados.locators_and_ranges(blocks, 8, 17), [['a', 10, 8, 2],
 335                                                                       ['b', 15, 0, 15]])
 336
 337         self.assertEqual(arvados.locators_and_ranges(blocks, 8, 20), [['a', 10, 8, 2],
 338                                                                       ['b', 15, 0, 15],
 339                                                                       ['c', 5, 0, 3]])
 340
 341         self.assertEqual(arvados.locators_and_ranges(blocks, 26, 2), [['c', 5, 1, 2]])
 342
 343         self.assertEqual(arvados.locators_and_ranges(blocks, 9, 15), [['a', 10, 9, 1],
 344                                                                       ['b', 15, 0, 14]])
 345         self.assertEqual(arvados.locators_and_ranges(blocks, 10, 15), [['b', 15, 0, 15]])
 346         self.assertEqual(arvados.locators_and_ranges(blocks, 11, 15), [['b', 15, 1, 14],
 347                                                                        ['c', 5, 0, 1]])
 348
 349     class MockStreamReader(object):
 350         def __init__(self, content):
 351             self.content = content
 352
 353         def readfrom(self, start, size):
 354             return self.content[start:start+size]
 355
 356     def test_file_stream(self):
 357         content = 'abcdefghijklmnopqrstuvwxyz0123456789'
 358         msr = self.MockStreamReader(content)
 359         segments = [[0, 10, 0],
 360                     [10, 15, 10],
 361                     [25, 5, 25]]
 362
 363         sfr = arvados.StreamFileReader(msr, segments, "test")
 364
 365         self.assertEqual(sfr.name(), "test")
 366         self.assertEqual(sfr.size(), 30)
 367
 368         self.assertEqual(sfr.readfrom(0, 30), content[0:30])
 369         self.assertEqual(sfr.readfrom(2, 30), content[2:30])
 370
 371         self.assertEqual(sfr.readfrom(2, 8), content[2:10])
 372         self.assertEqual(sfr.readfrom(0, 10), content[0:10])
 373
 374         self.assertEqual(sfr.tell(), 0)
 375         self.assertEqual(sfr.read(5), content[0:5])
 376         self.assertEqual(sfr.tell(), 5)
 377         self.assertEqual(sfr.read(5), content[5:10])
 378         self.assertEqual(sfr.tell(), 10)
 379         self.assertEqual(sfr.read(5), content[10:15])
 380         self.assertEqual(sfr.tell(), 15)
 381         self.assertEqual(sfr.read(5), content[15:20])
 382         self.assertEqual(sfr.tell(), 20)
 383         self.assertEqual(sfr.read(5), content[20:25])
 384         self.assertEqual(sfr.tell(), 25)
 385         self.assertEqual(sfr.read(5), content[25:30])
 386         self.assertEqual(sfr.tell(), 30)
 387         self.assertEqual(sfr.read(5), '')
 388         self.assertEqual(sfr.tell(), 30)
 389
 390         segments = [[26, 10, 0],
 391                     [0, 15, 10],
 392                     [15, 5, 25]]
 393
 394         sfr = arvados.StreamFileReader(msr, segments, "test")
 395
 396         self.assertEqual(sfr.size(), 30)
 397
 398         self.assertEqual(sfr.readfrom(0, 30), content[26:36] + content[0:20])
 399         self.assertEqual(sfr.readfrom(2, 30), content[28:36] + content[0:20])
 400
 401         self.assertEqual(sfr.readfrom(2, 8), content[28:36])
 402         self.assertEqual(sfr.readfrom(0, 10), content[26:36])
 403
 404         self.assertEqual(sfr.tell(), 0)
 405         self.assertEqual(sfr.read(5), content[26:31])
 406         self.assertEqual(sfr.tell(), 5)
 407         self.assertEqual(sfr.read(5), content[31:36])
 408         self.assertEqual(sfr.tell(), 10)
 409         self.assertEqual(sfr.read(5), content[0:5])
 410         self.assertEqual(sfr.tell(), 15)
 411         self.assertEqual(sfr.read(5), content[5:10])
 412         self.assertEqual(sfr.tell(), 20)
 413         self.assertEqual(sfr.read(5), content[10:15])
 414         self.assertEqual(sfr.tell(), 25)
 415         self.assertEqual(sfr.read(5), content[15:20])
 416         self.assertEqual(sfr.tell(), 30)
 417         self.assertEqual(sfr.read(5), '')
 418         self.assertEqual(sfr.tell(), 30)
 419
 420
 421     class MockKeep(object):
 422         def __init__(self, content):
 423             self.content = content
 424
 425         def get(self, locator):
 426             return self.content[locator]
 427
 428     def test_stream_reader(self):
 429         keepblocks = {'aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa+10': 'abcdefghij',
 430                       'bbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbb+15': 'klmnopqrstuvwxy',
 431                       'cccccccccccccccccccccccccccccccc+5': 'z0123'}
 432         mk = self.MockKeep(keepblocks)
 433
 434         sr = arvados.StreamReader([".", "aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa+10", "bbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbb+15", "cccccccccccccccccccccccccccccccc+5", "0:30:foo"], mk)
 435
 436         content = 'abcdefghijklmnopqrstuvwxyz0123456789'
 437
 438         self.assertEqual(sr.readfrom(0, 30), content[0:30])
 439         self.assertEqual(sr.readfrom(2, 30), content[2:30])
 440
 441         self.assertEqual(sr.readfrom(2, 8), content[2:10])
 442         self.assertEqual(sr.readfrom(0, 10), content[0:10])
 443
 444         self.assertEqual(sr.readfrom(0, 5), content[0:5])
 445         self.assertEqual(sr.readfrom(5, 5), content[5:10])
 446         self.assertEqual(sr.readfrom(10, 5), content[10:15])
 447         self.assertEqual(sr.readfrom(15, 5), content[15:20])
 448         self.assertEqual(sr.readfrom(20, 5), content[20:25])
 449         self.assertEqual(sr.readfrom(25, 5), content[25:30])
 450         self.assertEqual(sr.readfrom(30, 5), '')
 451
 452     def test_file_reader(self):
 453         keepblocks = {'aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa+10': 'abcdefghij',
 454                       'bbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbb+15': 'klmnopqrstuvwxy',
 455                       'cccccccccccccccccccccccccccccccc+5': 'z0123'}
 456         mk = self.MockKeep(keepblocks)
 457
 458         sr = arvados.StreamReader([".", "aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa+10", "bbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbb+15", "cccccccccccccccccccccccccccccccc+5", "0:10:foo", "15:10:foo"], mk)
 459
 460         content = 'abcdefghijpqrstuvwxy'
 461
 462         f = sr.files()["foo"]
 463
 464         # f.read() calls will be aligned on block boundaries (as a
 465         # result of ticket #3663).
 466
 467         f.seek(0)
 468         self.assertEqual(f.read(20), content[0:10])
 469
 470         f.seek(0)
 471         self.assertEqual(f.read(6), content[0:6])
 472         self.assertEqual(f.read(6), content[6:10])
 473         self.assertEqual(f.read(6), content[10:16])
 474         self.assertEqual(f.read(6), content[16:20])
 475
 476     def test_extract_file(self):
 477         m1 = """. 5348b82a029fd9e971a811ce1f71360b+43 0:43:md5sum.txt
 478 . 085c37f02916da1cad16f93c54d899b7+41 0:41:md6sum.txt
 479 . 8b22da26f9f433dea0a10e5ec66d73ba+43 0:43:md7sum.txt
 480 . 085c37f02916da1cad16f93c54d899b7+41 5348b82a029fd9e971a811ce1f71360b+43 8b22da26f9f433dea0a10e5ec66d73ba+43 47:80:md8sum.txt
 481 . 085c37f02916da1cad16f93c54d899b7+41 5348b82a029fd9e971a811ce1f71360b+43 8b22da26f9f433dea0a10e5ec66d73ba+43 40:80:md9sum.txt"""
 482
 483         m2 = arvados.CollectionReader(m1, self.api_client)
 484
 485         self.assertEqual(m2.manifest_text(),
 486                          ". 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")
 487
 488         self.assertEqual(arvados.CollectionReader(m1, self.api_client).all_streams()[0].files()['md5sum.txt'].as_manifest(),
 489                          ". 5348b82a029fd9e971a811ce1f71360b+43 0:43:md5sum.txt\n")
 490         self.assertEqual(arvados.CollectionReader(m1, self.api_client).all_streams()[0].files()['md6sum.txt'].as_manifest(),
 491                          ". 085c37f02916da1cad16f93c54d899b7+41 0:41:md6sum.txt\n")
 492         self.assertEqual(arvados.CollectionReader(m1, self.api_client).all_streams()[0].files()['md7sum.txt'].as_manifest(),
 493                          ". 8b22da26f9f433dea0a10e5ec66d73ba+43 0:43:md7sum.txt\n")
 494         self.assertEqual(arvados.CollectionReader(m1, self.api_client).all_streams()[0].files()['md9sum.txt'].as_manifest(),
 495                          ". 085c37f02916da1cad16f93c54d899b7+41 5348b82a029fd9e971a811ce1f71360b+43 8b22da26f9f433dea0a10e5ec66d73ba+43 40:80:md9sum.txt\n")
 496
 497     def test_write_directory_tree(self):
 498         cwriter = arvados.CollectionWriter(self.api_client)
 499         cwriter.write_directory_tree(self.build_directory_tree(
 500                 ['basefile', 'subdir/subfile']))
 501         self.assertEqual(cwriter.manifest_text(),
 502                          """. c5110c5ac93202d8e0f9e381f22bac0f+8 0:8:basefile
 503 ./subdir 1ca4dec89403084bf282ad31e6cf7972+14 0:14:subfile\n""")
 504
 505     def test_write_named_directory_tree(self):
 506         cwriter = arvados.CollectionWriter(self.api_client)
 507         cwriter.write_directory_tree(self.build_directory_tree(
 508                 ['basefile', 'subdir/subfile']), 'root')
 509         self.assertEqual(
 510             cwriter.manifest_text(),
 511             """./root c5110c5ac93202d8e0f9e381f22bac0f+8 0:8:basefile
 512 ./root/subdir 1ca4dec89403084bf282ad31e6cf7972+14 0:14:subfile\n""")
 513
 514     def test_write_directory_tree_in_one_stream(self):
 515         cwriter = arvados.CollectionWriter(self.api_client)
 516         cwriter.write_directory_tree(self.build_directory_tree(
 517                 ['basefile', 'subdir/subfile']), max_manifest_depth=0)
 518         self.assertEqual(cwriter.manifest_text(),
 519                          """. 4ace875ffdc6824a04950f06858f4465+22 0:8:basefile
 520 ./subdir 4ace875ffdc6824a04950f06858f4465+22 8:14:subfile\n""")
 521
 522     def test_write_directory_tree_with_limited_recursion(self):
 523         cwriter = arvados.CollectionWriter(self.api_client)
 524         cwriter.write_directory_tree(
 525             self.build_directory_tree(['f1', 'd1/f2', 'd1/d2/f3']),
 526             max_manifest_depth=1)
 527         self.assertEqual(cwriter.manifest_text(),
 528                          """. bd19836ddb62c11c55ab251ccaca5645+2 0:2:f1
 529 ./d1 50170217e5b04312024aa5cd42934494+13 8:5:f2
 530 ./d1/d2 50170217e5b04312024aa5cd42934494+13 0:8:f3\n""")
 531
 532     def test_write_one_file(self):
 533         cwriter = arvados.CollectionWriter(self.api_client)
 534         with self.make_test_file() as testfile:
 535             cwriter.write_file(testfile.name)
 536             self.assertEqual(
 537                 cwriter.manifest_text(),
 538                 ". 098f6bcd4621d373cade4e832627b4f6+4 0:4:{}\n".format(
 539                     os.path.basename(testfile.name)))
 540
 541     def test_write_named_file(self):
 542         cwriter = arvados.CollectionWriter(self.api_client)
 543         with self.make_test_file() as testfile:
 544             cwriter.write_file(testfile.name, 'foo')
 545             self.assertEqual(cwriter.manifest_text(),
 546                              ". 098f6bcd4621d373cade4e832627b4f6+4 0:4:foo\n")
 547
 548     def test_write_multiple_files(self):
 549         cwriter = arvados.CollectionWriter(self.api_client)
 550         for letter in 'ABC':
 551             with self.make_test_file(letter) as testfile:
 552                 cwriter.write_file(testfile.name, letter)
 553         self.assertEqual(
 554             cwriter.manifest_text(),
 555             ". 902fbdd2b1df0c4f70b4a5d23525e932+3 0:1:A 1:1:B 2:1:C\n")
 556
 557     def test_basic_resume(self):
 558         cwriter = TestResumableWriter()
 559         with self.make_test_file() as testfile:
 560             cwriter.write_file(testfile.name, 'test')
 561             resumed = TestResumableWriter.from_state(cwriter.current_state())
 562         self.assertEquals(cwriter.manifest_text(), resumed.manifest_text(),
 563                           "resumed CollectionWriter had different manifest")
 564
 565     def test_resume_fails_when_missing_dependency(self):
 566         cwriter = TestResumableWriter()
 567         with self.make_test_file() as testfile:
 568             cwriter.write_file(testfile.name, 'test')
 569         self.assertRaises(arvados.errors.StaleWriterStateError,
 570                           TestResumableWriter.from_state,
 571                           cwriter.current_state())
 572
 573     def test_resume_fails_when_dependency_mtime_changed(self):
 574         cwriter = TestResumableWriter()
 575         with self.make_test_file() as testfile:
 576             cwriter.write_file(testfile.name, 'test')
 577             os.utime(testfile.name, (0, 0))
 578             self.assertRaises(arvados.errors.StaleWriterStateError,
 579                               TestResumableWriter.from_state,
 580                               cwriter.current_state())
 581
 582     def test_resume_fails_when_dependency_is_nonfile(self):
 583         cwriter = TestResumableWriter()
 584         cwriter.write_file('/dev/null', 'empty')
 585         self.assertRaises(arvados.errors.StaleWriterStateError,
 586                           TestResumableWriter.from_state,
 587                           cwriter.current_state())
 588
 589     def test_resume_fails_when_dependency_size_changed(self):
 590         cwriter = TestResumableWriter()
 591         with self.make_test_file() as testfile:
 592             cwriter.write_file(testfile.name, 'test')
 593             orig_mtime = os.fstat(testfile.fileno()).st_mtime
 594             testfile.write('extra')
 595             testfile.flush()
 596             os.utime(testfile.name, (orig_mtime, orig_mtime))
 597             self.assertRaises(arvados.errors.StaleWriterStateError,
 598                               TestResumableWriter.from_state,
 599                               cwriter.current_state())
 600
 601     def test_resume_fails_with_expired_locator(self):
 602         cwriter = TestResumableWriter()
 603         state = cwriter.current_state()
 604         # Add an expired locator to the state.
 605         state['_current_stream_locators'].append(''.join([
 606                     'a' * 32, '+1+A', 'b' * 40, '@', '10000000']))
 607         self.assertRaises(arvados.errors.StaleWriterStateError,
 608                           TestResumableWriter.from_state, state)
 609
 610     def test_arbitrary_objects_not_resumable(self):
 611         cwriter = TestResumableWriter()
 612         with open('/dev/null') as badfile:
 613             self.assertRaises(arvados.errors.AssertionError,
 614                               cwriter.write_file, badfile)
 615
 616     def test_arbitrary_writes_not_resumable(self):
 617         cwriter = TestResumableWriter()
 618         self.assertRaises(arvados.errors.AssertionError,
 619                           cwriter.write, "badtext")
 620
 621
 622 if __name__ == '__main__':
 623     unittest.main()