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 hashlib
   9 import mock
  10 import os
  11 import pprint
  12 import re
  13 import subprocess
  14 import tempfile
  15 import unittest
  16
  17 import run_test_server
  18 import arvados_testutil as tutil
  19
  20 class TestResumableWriter(arvados.ResumableCollectionWriter):
  21     KEEP_BLOCK_SIZE = 1024  # PUT to Keep every 1K.
  22
  23     def current_state(self):
  24         return self.dump_state(copy.deepcopy)
  25
  26
  27 class ArvadosCollectionsTest(run_test_server.TestCaseWithServers,
  28                              tutil.ArvadosBaseTestCase):
  29     MAIN_SERVER = {}
  30
  31     @classmethod
  32     def setUpClass(cls):
  33         super(ArvadosCollectionsTest, cls).setUpClass()
  34         run_test_server.authorize_with('active')
  35         cls.api_client = arvados.api('v1')
  36         cls.keep_client = arvados.KeepClient(api_client=cls.api_client,
  37                                              local_store=cls.local_store)
  38
  39     def write_foo_bar_baz(self):
  40         cw = arvados.CollectionWriter(self.api_client)
  41         self.assertEqual(cw.current_stream_name(), '.',
  42                          'current_stream_name() should be "." now')
  43         cw.set_current_file_name('foo.txt')
  44         cw.write('foo')
  45         self.assertEqual(cw.current_file_name(), 'foo.txt',
  46                          'current_file_name() should be foo.txt now')
  47         cw.start_new_file('bar.txt')
  48         cw.write('bar')
  49         cw.start_new_stream('baz')
  50         cw.write('baz')
  51         cw.set_current_file_name('baz.txt')
  52         self.assertEqual(cw.manifest_text(),
  53                          ". 3858f62230ac3c915f300c664312c63f+6 0:3:foo.txt 3:3:bar.txt\n" +
  54                          "./baz 73feffa4b7f6bb68e44cf984c85f6e88+3 0:3:baz.txt\n",
  55                          "wrong manifest: got {}".format(cw.manifest_text()))
  56         cw.finish()
  57         return cw.portable_data_hash()
  58
  59     def test_keep_local_store(self):
  60         self.assertEqual(self.keep_client.put('foo'), 'acbd18db4cc2f85cedef654fccc4a4d8+3', 'wrong md5 hash from Keep.put')
  61         self.assertEqual(self.keep_client.get('acbd18db4cc2f85cedef654fccc4a4d8+3'), 'foo', 'wrong data from Keep.get')
  62
  63     def test_local_collection_writer(self):
  64         self.assertEqual(self.write_foo_bar_baz(),
  65                          '23ca013983d6239e98931cc779e68426+114',
  66                          'wrong locator hash: ' + self.write_foo_bar_baz())
  67
  68     def test_local_collection_reader(self):
  69         foobarbaz = self.write_foo_bar_baz()
  70         cr = arvados.CollectionReader(
  71             foobarbaz + '+Xzizzle', self.api_client)
  72         got = []
  73         for s in cr.all_streams():
  74             for f in s.all_files():
  75                 got += [[f.size(), f.stream_name(), f.name(), f.read(2**26)]]
  76         expected = [[3, '.', 'foo.txt', 'foo'],
  77                     [3, '.', 'bar.txt', 'bar'],
  78                     [3, './baz', 'baz.txt', 'baz']]
  79         self.assertEqual(got,
  80                          expected)
  81         stream0 = cr.all_streams()[0]
  82         self.assertEqual(stream0.readfrom(0, 0),
  83                          '',
  84                          'reading zero bytes should have returned empty string')
  85         self.assertEqual(stream0.readfrom(0, 2**26),
  86                          'foobar',
  87                          'reading entire stream failed')
  88         self.assertEqual(stream0.readfrom(2**26, 0),
  89                          '',
  90                          'reading zero bytes should have returned empty string')
  91
  92     def _test_subset(self, collection, expected):
  93         cr = arvados.CollectionReader(collection, self.api_client)
  94         for s in cr.all_streams():
  95             for ex in expected:
  96                 if ex[0] == s:
  97                     f = s.files()[ex[2]]
  98                     got = [f.size(), f.stream_name(), f.name(), "".join(f.readall(2**26))]
  99                     self.assertEqual(got,
 100                                      ex,
 101                                      'all_files|as_manifest did not preserve manifest contents: got %s expected %s' % (got, ex))
 102
 103     def test_collection_manifest_subset(self):
 104         foobarbaz = self.write_foo_bar_baz()
 105         self._test_subset(foobarbaz,
 106                           [[3, '.',     'bar.txt', 'bar'],
 107                            [3, '.',     'foo.txt', 'foo'],
 108                            [3, './baz', 'baz.txt', 'baz']])
 109         self._test_subset((". %s %s 0:3:foo.txt 3:3:bar.txt\n" %
 110                            (self.keep_client.put("foo"),
 111                             self.keep_client.put("bar"))),
 112                           [[3, '.', 'bar.txt', 'bar'],
 113                            [3, '.', 'foo.txt', 'foo']])
 114         self._test_subset((". %s %s 0:2:fo.txt 2:4:obar.txt\n" %
 115                            (self.keep_client.put("foo"),
 116                             self.keep_client.put("bar"))),
 117                           [[2, '.', 'fo.txt', 'fo'],
 118                            [4, '.', 'obar.txt', 'obar']])
 119         self._test_subset((". %s %s 0:2:fo.txt 2:0:zero.txt 2:2:ob.txt 4:2:ar.txt\n" %
 120                            (self.keep_client.put("foo"),
 121                             self.keep_client.put("bar"))),
 122                           [[2, '.', 'ar.txt', 'ar'],
 123                            [2, '.', 'fo.txt', 'fo'],
 124                            [2, '.', 'ob.txt', 'ob'],
 125                            [0, '.', 'zero.txt', '']])
 126
 127     def _test_readline(self, what_in, what_out):
 128         cw = arvados.CollectionWriter(self.api_client)
 129         cw.start_new_file('test.txt')
 130         cw.write(what_in)
 131         test1 = cw.finish()
 132         cr = arvados.CollectionReader(test1, self.api_client)
 133         got = []
 134         for x in list(cr.all_files())[0].readlines():
 135             got += [x]
 136         self.assertEqual(got,
 137                          what_out,
 138                          "readlines did not split lines correctly: %s" % got)
 139
 140     def test_collection_readline(self):
 141         self._test_readline("\na\nbcd\n\nefg\nz",
 142                             ["\n", "a\n", "bcd\n", "\n", "efg\n", "z"])
 143         self._test_readline("ab\ncd\n",
 144                             ["ab\n", "cd\n"])
 145
 146     def test_collection_empty_file(self):
 147         cw = arvados.CollectionWriter(self.api_client)
 148         cw.start_new_file('zero.txt')
 149         cw.write('')
 150
 151         self.assertEqual(cw.manifest_text(), ". d41d8cd98f00b204e9800998ecf8427e+0 0:0:zero.txt\n")
 152         self.check_manifest_file_sizes(cw.manifest_text(), [0])
 153         cw = arvados.CollectionWriter(self.api_client)
 154         cw.start_new_file('zero.txt')
 155         cw.write('')
 156         cw.start_new_file('one.txt')
 157         cw.write('1')
 158         cw.start_new_stream('foo')
 159         cw.start_new_file('zero.txt')
 160         cw.write('')
 161         self.check_manifest_file_sizes(cw.manifest_text(), [0,1,0])
 162
 163     def test_no_implicit_normalize(self):
 164         cw = arvados.CollectionWriter(self.api_client)
 165         cw.start_new_file('b')
 166         cw.write('b')
 167         cw.start_new_file('a')
 168         cw.write('')
 169         self.check_manifest_file_sizes(cw.manifest_text(), [1,0])
 170         self.check_manifest_file_sizes(
 171             arvados.CollectionReader(
 172                 cw.manifest_text()).manifest_text(normalize=True),
 173             [0,1])
 174
 175     def check_manifest_file_sizes(self, manifest_text, expect_sizes):
 176         cr = arvados.CollectionReader(manifest_text, self.api_client)
 177         got_sizes = []
 178         for f in cr.all_files():
 179             got_sizes += [f.size()]
 180         self.assertEqual(got_sizes, expect_sizes, "got wrong file sizes %s, expected %s" % (got_sizes, expect_sizes))
 181
 182     def test_collection_bz2_decompression(self):
 183         n_lines_in = 2**18
 184         data_in = "abc\n"
 185         for x in xrange(0, 18):
 186             data_in += data_in
 187         compressed_data_in = bz2.compress(data_in)
 188         cw = arvados.CollectionWriter(self.api_client)
 189         cw.start_new_file('test.bz2')
 190         cw.write(compressed_data_in)
 191         bz2_manifest = cw.manifest_text()
 192
 193         cr = arvados.CollectionReader(bz2_manifest, self.api_client)
 194
 195         got = 0
 196         for x in list(cr.all_files())[0].readlines():
 197             self.assertEqual(x, "abc\n", "decompression returned wrong data: %s" % x)
 198             got += 1
 199         self.assertEqual(got,
 200                          n_lines_in,
 201                          "decompression returned %d lines instead of %d" % (got, n_lines_in))
 202
 203     def test_collection_gzip_decompression(self):
 204         n_lines_in = 2**18
 205         data_in = "abc\n"
 206         for x in xrange(0, 18):
 207             data_in += data_in
 208         p = subprocess.Popen(["gzip", "-1cn"],
 209                              stdout=subprocess.PIPE,
 210                              stdin=subprocess.PIPE,
 211                              stderr=subprocess.PIPE,
 212                              shell=False, close_fds=True)
 213         compressed_data_in, stderrdata = p.communicate(data_in)
 214
 215         cw = arvados.CollectionWriter(self.api_client)
 216         cw.start_new_file('test.gz')
 217         cw.write(compressed_data_in)
 218         gzip_manifest = cw.manifest_text()
 219
 220         cr = arvados.CollectionReader(gzip_manifest, self.api_client)
 221         got = 0
 222         for x in list(cr.all_files())[0].readlines():
 223             self.assertEqual(x, "abc\n", "decompression returned wrong data: %s" % x)
 224             got += 1
 225         self.assertEqual(got,
 226                          n_lines_in,
 227                          "decompression returned %d lines instead of %d" % (got, n_lines_in))
 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 """
 234         self.assertEqual(arvados.CollectionReader(m1, self.api_client).manifest_text(normalize=True),
 235                          """. 5348b82a029fd9e971a811ce1f71360b+43 085c37f02916da1cad16f93c54d899b7+41 8b22da26f9f433dea0a10e5ec66d73ba+43 0:127:md5sum.txt
 236 """)
 237
 238         m2 = """. 204e43b8a1185621ca55a94839582e6f+67108864 b9677abbac956bd3e86b1deb28dfac03+67108864 fc15aff2a762b13f521baf042140acec+67108864 323d2a3ce20370c4ca1d3462a344f8fd+25885655 0:227212247:var-GS000016015-ASM.tsv.bz2
 239 """
 240         self.assertEqual(arvados.CollectionReader(m2, self.api_client).manifest_text(normalize=True), m2)
 241
 242         m3 = """. 5348b82a029fd9e971a811ce1f71360b+43 3:40:md5sum.txt
 243 . 085c37f02916da1cad16f93c54d899b7+41 0:41:md5sum.txt
 244 . 8b22da26f9f433dea0a10e5ec66d73ba+43 0:43:md5sum.txt
 245 """
 246         self.assertEqual(arvados.CollectionReader(m3, self.api_client).manifest_text(normalize=True),
 247                          """. 5348b82a029fd9e971a811ce1f71360b+43 085c37f02916da1cad16f93c54d899b7+41 8b22da26f9f433dea0a10e5ec66d73ba+43 3:124:md5sum.txt
 248 """)
 249
 250         m4 = """. 204e43b8a1185621ca55a94839582e6f+67108864 0:3:foo/bar
 251 ./zzz 204e43b8a1185621ca55a94839582e6f+67108864 0:999:zzz
 252 ./foo 323d2a3ce20370c4ca1d3462a344f8fd+25885655 0:3:bar
 253 """
 254         self.assertEqual(arvados.CollectionReader(m4, self.api_client).manifest_text(normalize=True),
 255                          """./foo 204e43b8a1185621ca55a94839582e6f+67108864 323d2a3ce20370c4ca1d3462a344f8fd+25885655 0:3:bar 67108864:3:bar
 256 ./zzz 204e43b8a1185621ca55a94839582e6f+67108864 0:999:zzz
 257 """)
 258
 259         m5 = """. 204e43b8a1185621ca55a94839582e6f+67108864 0:3:foo/bar
 260 ./zzz 204e43b8a1185621ca55a94839582e6f+67108864 0:999:zzz
 261 ./foo 204e43b8a1185621ca55a94839582e6f+67108864 3:3:bar
 262 """
 263         self.assertEqual(arvados.CollectionReader(m5, self.api_client).manifest_text(normalize=True),
 264                          """./foo 204e43b8a1185621ca55a94839582e6f+67108864 0:6:bar
 265 ./zzz 204e43b8a1185621ca55a94839582e6f+67108864 0:999:zzz
 266 """)
 267
 268         with self.data_file('1000G_ref_manifest') as f6:
 269             m6 = f6.read()
 270             self.assertEqual(arvados.CollectionReader(m6, self.api_client).manifest_text(normalize=True), m6)
 271
 272         with self.data_file('jlake_manifest') as f7:
 273             m7 = f7.read()
 274             self.assertEqual(arvados.CollectionReader(m7, self.api_client).manifest_text(normalize=True), m7)
 275
 276         m8 = """./a\\040b\\040c 59ca0efa9f5633cb0371bbc0355478d8+13 0:13:hello\\040world.txt
 277 """
 278         self.assertEqual(arvados.CollectionReader(m8, self.api_client).manifest_text(normalize=True), m8)
 279
 280     def test_locators_and_ranges(self):
 281         blocks2 = [['a', 10, 0],
 282                   ['b', 10, 10],
 283                   ['c', 10, 20],
 284                   ['d', 10, 30],
 285                   ['e', 10, 40],
 286                   ['f', 10, 50]]
 287
 288         self.assertEqual(arvados.locators_and_ranges(blocks2,  2,  2), [['a', 10, 2, 2]])
 289         self.assertEqual(arvados.locators_and_ranges(blocks2, 12, 2), [['b', 10, 2, 2]])
 290         self.assertEqual(arvados.locators_and_ranges(blocks2, 22, 2), [['c', 10, 2, 2]])
 291         self.assertEqual(arvados.locators_and_ranges(blocks2, 32, 2), [['d', 10, 2, 2]])
 292         self.assertEqual(arvados.locators_and_ranges(blocks2, 42, 2), [['e', 10, 2, 2]])
 293         self.assertEqual(arvados.locators_and_ranges(blocks2, 52, 2), [['f', 10, 2, 2]])
 294         self.assertEqual(arvados.locators_and_ranges(blocks2, 62, 2), [])
 295         self.assertEqual(arvados.locators_and_ranges(blocks2, -2, 2), [])
 296
 297         self.assertEqual(arvados.locators_and_ranges(blocks2,  0,  2), [['a', 10, 0, 2]])
 298         self.assertEqual(arvados.locators_and_ranges(blocks2, 10, 2), [['b', 10, 0, 2]])
 299         self.assertEqual(arvados.locators_and_ranges(blocks2, 20, 2), [['c', 10, 0, 2]])
 300         self.assertEqual(arvados.locators_and_ranges(blocks2, 30, 2), [['d', 10, 0, 2]])
 301         self.assertEqual(arvados.locators_and_ranges(blocks2, 40, 2), [['e', 10, 0, 2]])
 302         self.assertEqual(arvados.locators_and_ranges(blocks2, 50, 2), [['f', 10, 0, 2]])
 303         self.assertEqual(arvados.locators_and_ranges(blocks2, 60, 2), [])
 304         self.assertEqual(arvados.locators_and_ranges(blocks2, -2, 2), [])
 305
 306         self.assertEqual(arvados.locators_and_ranges(blocks2,  9,  2), [['a', 10, 9, 1], ['b', 10, 0, 1]])
 307         self.assertEqual(arvados.locators_and_ranges(blocks2, 19, 2), [['b', 10, 9, 1], ['c', 10, 0, 1]])
 308         self.assertEqual(arvados.locators_and_ranges(blocks2, 29, 2), [['c', 10, 9, 1], ['d', 10, 0, 1]])
 309         self.assertEqual(arvados.locators_and_ranges(blocks2, 39, 2), [['d', 10, 9, 1], ['e', 10, 0, 1]])
 310         self.assertEqual(arvados.locators_and_ranges(blocks2, 49, 2), [['e', 10, 9, 1], ['f', 10, 0, 1]])
 311         self.assertEqual(arvados.locators_and_ranges(blocks2, 59, 2), [['f', 10, 9, 1]])
 312
 313
 314         blocks3 = [['a', 10, 0],
 315                   ['b', 10, 10],
 316                   ['c', 10, 20],
 317                   ['d', 10, 30],
 318                   ['e', 10, 40],
 319                   ['f', 10, 50],
 320                   ['g', 10, 60]]
 321
 322         self.assertEqual(arvados.locators_and_ranges(blocks3,  2,  2), [['a', 10, 2, 2]])
 323         self.assertEqual(arvados.locators_and_ranges(blocks3, 12, 2), [['b', 10, 2, 2]])
 324         self.assertEqual(arvados.locators_and_ranges(blocks3, 22, 2), [['c', 10, 2, 2]])
 325         self.assertEqual(arvados.locators_and_ranges(blocks3, 32, 2), [['d', 10, 2, 2]])
 326         self.assertEqual(arvados.locators_and_ranges(blocks3, 42, 2), [['e', 10, 2, 2]])
 327         self.assertEqual(arvados.locators_and_ranges(blocks3, 52, 2), [['f', 10, 2, 2]])
 328         self.assertEqual(arvados.locators_and_ranges(blocks3, 62, 2), [['g', 10, 2, 2]])
 329
 330
 331         blocks = [['a', 10, 0],
 332                   ['b', 15, 10],
 333                   ['c', 5, 25]]
 334         self.assertEqual(arvados.locators_and_ranges(blocks, 1, 0), [])
 335         self.assertEqual(arvados.locators_and_ranges(blocks, 0, 5), [['a', 10, 0, 5]])
 336         self.assertEqual(arvados.locators_and_ranges(blocks, 3, 5), [['a', 10, 3, 5]])
 337         self.assertEqual(arvados.locators_and_ranges(blocks, 0, 10), [['a', 10, 0, 10]])
 338
 339         self.assertEqual(arvados.locators_and_ranges(blocks, 0, 11), [['a', 10, 0, 10],
 340                                                                       ['b', 15, 0, 1]])
 341         self.assertEqual(arvados.locators_and_ranges(blocks, 1, 11), [['a', 10, 1, 9],
 342                                                                       ['b', 15, 0, 2]])
 343         self.assertEqual(arvados.locators_and_ranges(blocks, 0, 25), [['a', 10, 0, 10],
 344                                                                       ['b', 15, 0, 15]])
 345
 346         self.assertEqual(arvados.locators_and_ranges(blocks, 0, 30), [['a', 10, 0, 10],
 347                                                                       ['b', 15, 0, 15],
 348                                                                       ['c', 5, 0, 5]])
 349         self.assertEqual(arvados.locators_and_ranges(blocks, 1, 30), [['a', 10, 1, 9],
 350                                                                       ['b', 15, 0, 15],
 351                                                                       ['c', 5, 0, 5]])
 352         self.assertEqual(arvados.locators_and_ranges(blocks, 0, 31), [['a', 10, 0, 10],
 353                                                                       ['b', 15, 0, 15],
 354                                                                       ['c', 5, 0, 5]])
 355
 356         self.assertEqual(arvados.locators_and_ranges(blocks, 15, 5), [['b', 15, 5, 5]])
 357
 358         self.assertEqual(arvados.locators_and_ranges(blocks, 8, 17), [['a', 10, 8, 2],
 359                                                                       ['b', 15, 0, 15]])
 360
 361         self.assertEqual(arvados.locators_and_ranges(blocks, 8, 20), [['a', 10, 8, 2],
 362                                                                       ['b', 15, 0, 15],
 363                                                                       ['c', 5, 0, 3]])
 364
 365         self.assertEqual(arvados.locators_and_ranges(blocks, 26, 2), [['c', 5, 1, 2]])
 366
 367         self.assertEqual(arvados.locators_and_ranges(blocks, 9, 15), [['a', 10, 9, 1],
 368                                                                       ['b', 15, 0, 14]])
 369         self.assertEqual(arvados.locators_and_ranges(blocks, 10, 15), [['b', 15, 0, 15]])
 370         self.assertEqual(arvados.locators_and_ranges(blocks, 11, 15), [['b', 15, 1, 14],
 371                                                                        ['c', 5, 0, 1]])
 372
 373     class MockStreamReader(object):
 374         def __init__(self, content):
 375             self.content = content
 376             self.num_retries = 0
 377
 378         def readfrom(self, start, size, num_retries=0):
 379             return self.content[start:start+size]
 380
 381     def test_file_stream(self):
 382         content = 'abcdefghijklmnopqrstuvwxyz0123456789'
 383         msr = self.MockStreamReader(content)
 384         segments = [[0, 10, 0],
 385                     [10, 15, 10],
 386                     [25, 5, 25]]
 387
 388         sfr = arvados.StreamFileReader(msr, segments, "test")
 389
 390         self.assertEqual(sfr.name(), "test")
 391         self.assertEqual(sfr.size(), 30)
 392
 393         self.assertEqual(sfr.readfrom(0, 30), content[0:30])
 394         self.assertEqual(sfr.readfrom(2, 30), content[2:30])
 395
 396         self.assertEqual(sfr.readfrom(2, 8), content[2:10])
 397         self.assertEqual(sfr.readfrom(0, 10), content[0:10])
 398
 399         self.assertEqual(sfr.tell(), 0)
 400         self.assertEqual(sfr.read(5), content[0:5])
 401         self.assertEqual(sfr.tell(), 5)
 402         self.assertEqual(sfr.read(5), content[5:10])
 403         self.assertEqual(sfr.tell(), 10)
 404         self.assertEqual(sfr.read(5), content[10:15])
 405         self.assertEqual(sfr.tell(), 15)
 406         self.assertEqual(sfr.read(5), content[15:20])
 407         self.assertEqual(sfr.tell(), 20)
 408         self.assertEqual(sfr.read(5), content[20:25])
 409         self.assertEqual(sfr.tell(), 25)
 410         self.assertEqual(sfr.read(5), content[25:30])
 411         self.assertEqual(sfr.tell(), 30)
 412         self.assertEqual(sfr.read(5), '')
 413         self.assertEqual(sfr.tell(), 30)
 414
 415         segments = [[26, 10, 0],
 416                     [0, 15, 10],
 417                     [15, 5, 25]]
 418
 419         sfr = arvados.StreamFileReader(msr, segments, "test")
 420
 421         self.assertEqual(sfr.size(), 30)
 422
 423         self.assertEqual(sfr.readfrom(0, 30), content[26:36] + content[0:20])
 424         self.assertEqual(sfr.readfrom(2, 30), content[28:36] + content[0:20])
 425
 426         self.assertEqual(sfr.readfrom(2, 8), content[28:36])
 427         self.assertEqual(sfr.readfrom(0, 10), content[26:36])
 428
 429         self.assertEqual(sfr.tell(), 0)
 430         self.assertEqual(sfr.read(5), content[26:31])
 431         self.assertEqual(sfr.tell(), 5)
 432         self.assertEqual(sfr.read(5), content[31:36])
 433         self.assertEqual(sfr.tell(), 10)
 434         self.assertEqual(sfr.read(5), content[0:5])
 435         self.assertEqual(sfr.tell(), 15)
 436         self.assertEqual(sfr.read(5), content[5:10])
 437         self.assertEqual(sfr.tell(), 20)
 438         self.assertEqual(sfr.read(5), content[10:15])
 439         self.assertEqual(sfr.tell(), 25)
 440         self.assertEqual(sfr.read(5), content[15:20])
 441         self.assertEqual(sfr.tell(), 30)
 442         self.assertEqual(sfr.read(5), '')
 443         self.assertEqual(sfr.tell(), 30)
 444
 445
 446     class MockKeep(object):
 447         def __init__(self, content, num_retries=0):
 448             self.content = content
 449
 450         def get(self, locator, num_retries=0):
 451             return self.content[locator]
 452
 453     def test_stream_reader(self):
 454         keepblocks = {'aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa+10': 'abcdefghij',
 455                       'bbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbb+15': 'klmnopqrstuvwxy',
 456                       'cccccccccccccccccccccccccccccccc+5': 'z0123'}
 457         mk = self.MockKeep(keepblocks)
 458
 459         sr = arvados.StreamReader([".", "aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa+10", "bbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbb+15", "cccccccccccccccccccccccccccccccc+5", "0:30:foo"], mk)
 460
 461         content = 'abcdefghijklmnopqrstuvwxyz0123456789'
 462
 463         self.assertEqual(sr.readfrom(0, 30), content[0:30])
 464         self.assertEqual(sr.readfrom(2, 30), content[2:30])
 465
 466         self.assertEqual(sr.readfrom(2, 8), content[2:10])
 467         self.assertEqual(sr.readfrom(0, 10), content[0:10])
 468
 469         self.assertEqual(sr.readfrom(0, 5), content[0:5])
 470         self.assertEqual(sr.readfrom(5, 5), content[5:10])
 471         self.assertEqual(sr.readfrom(10, 5), content[10:15])
 472         self.assertEqual(sr.readfrom(15, 5), content[15:20])
 473         self.assertEqual(sr.readfrom(20, 5), content[20:25])
 474         self.assertEqual(sr.readfrom(25, 5), content[25:30])
 475         self.assertEqual(sr.readfrom(30, 5), '')
 476
 477     def test_file_reader(self):
 478         keepblocks = {'aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa+10': 'abcdefghij',
 479                       'bbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbb+15': 'klmnopqrstuvwxy',
 480                       'cccccccccccccccccccccccccccccccc+5': 'z0123'}
 481         mk = self.MockKeep(keepblocks)
 482
 483         sr = arvados.StreamReader([".", "aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa+10", "bbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbb+15", "cccccccccccccccccccccccccccccccc+5", "0:10:foo", "15:10:foo"], mk)
 484
 485         content = 'abcdefghijpqrstuvwxy'
 486
 487         f = sr.files()["foo"]
 488
 489         # f.read() calls will be aligned on block boundaries (as a
 490         # result of ticket #3663).
 491
 492         f.seek(0)
 493         self.assertEqual(f.read(20), content[0:10])
 494
 495         f.seek(0)
 496         self.assertEqual(f.read(6), content[0:6])
 497         self.assertEqual(f.read(6), content[6:10])
 498         self.assertEqual(f.read(6), content[10:16])
 499         self.assertEqual(f.read(6), content[16:20])
 500
 501     def test_extract_file(self):
 502         m1 = """. 5348b82a029fd9e971a811ce1f71360b+43 0:43:md5sum.txt
 503 . 085c37f02916da1cad16f93c54d899b7+41 0:41:md6sum.txt
 504 . 8b22da26f9f433dea0a10e5ec66d73ba+43 0:43:md7sum.txt
 505 . 085c37f02916da1cad16f93c54d899b7+41 5348b82a029fd9e971a811ce1f71360b+43 8b22da26f9f433dea0a10e5ec66d73ba+43 47:80:md8sum.txt
 506 . 085c37f02916da1cad16f93c54d899b7+41 5348b82a029fd9e971a811ce1f71360b+43 8b22da26f9f433dea0a10e5ec66d73ba+43 40:80:md9sum.txt
 507 """
 508
 509         m2 = arvados.CollectionReader(m1, self.api_client).manifest_text(normalize=True)
 510
 511         self.assertEqual(m2,
 512                          ". 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")
 513         files = arvados.CollectionReader(
 514             m2, self.api_client).all_streams()[0].files()
 515
 516         self.assertEqual(files['md5sum.txt'].as_manifest(),
 517                          ". 5348b82a029fd9e971a811ce1f71360b+43 0:43:md5sum.txt\n")
 518         self.assertEqual(files['md6sum.txt'].as_manifest(),
 519                          ". 085c37f02916da1cad16f93c54d899b7+41 0:41:md6sum.txt\n")
 520         self.assertEqual(files['md7sum.txt'].as_manifest(),
 521                          ". 8b22da26f9f433dea0a10e5ec66d73ba+43 0:43:md7sum.txt\n")
 522         self.assertEqual(files['md9sum.txt'].as_manifest(),
 523                          ". 085c37f02916da1cad16f93c54d899b7+41 5348b82a029fd9e971a811ce1f71360b+43 8b22da26f9f433dea0a10e5ec66d73ba+43 40:80:md9sum.txt\n")
 524
 525     def test_write_directory_tree(self):
 526         cwriter = arvados.CollectionWriter(self.api_client)
 527         cwriter.write_directory_tree(self.build_directory_tree(
 528                 ['basefile', 'subdir/subfile']))
 529         self.assertEqual(cwriter.manifest_text(),
 530                          """. c5110c5ac93202d8e0f9e381f22bac0f+8 0:8:basefile
 531 ./subdir 1ca4dec89403084bf282ad31e6cf7972+14 0:14:subfile\n""")
 532
 533     def test_write_named_directory_tree(self):
 534         cwriter = arvados.CollectionWriter(self.api_client)
 535         cwriter.write_directory_tree(self.build_directory_tree(
 536                 ['basefile', 'subdir/subfile']), 'root')
 537         self.assertEqual(
 538             cwriter.manifest_text(),
 539             """./root c5110c5ac93202d8e0f9e381f22bac0f+8 0:8:basefile
 540 ./root/subdir 1ca4dec89403084bf282ad31e6cf7972+14 0:14:subfile\n""")
 541
 542     def test_write_directory_tree_in_one_stream(self):
 543         cwriter = arvados.CollectionWriter(self.api_client)
 544         cwriter.write_directory_tree(self.build_directory_tree(
 545                 ['basefile', 'subdir/subfile']), max_manifest_depth=0)
 546         self.assertEqual(cwriter.manifest_text(),
 547                          """. 4ace875ffdc6824a04950f06858f4465+22 0:8:basefile 8:14:subdir/subfile\n""")
 548
 549     def test_write_directory_tree_with_limited_recursion(self):
 550         cwriter = arvados.CollectionWriter(self.api_client)
 551         cwriter.write_directory_tree(
 552             self.build_directory_tree(['f1', 'd1/f2', 'd1/d2/f3']),
 553             max_manifest_depth=1)
 554         self.assertEqual(cwriter.manifest_text(),
 555                          """. bd19836ddb62c11c55ab251ccaca5645+2 0:2:f1
 556 ./d1 50170217e5b04312024aa5cd42934494+13 0:8:d2/f3 8:5:f2\n""")
 557
 558     def test_write_directory_tree_with_zero_recursion(self):
 559         cwriter = arvados.CollectionWriter(self.api_client)
 560         content = 'd1/d2/f3d1/f2f1'
 561         blockhash = hashlib.md5(content).hexdigest() + '+' + str(len(content))
 562         cwriter.write_directory_tree(
 563             self.build_directory_tree(['f1', 'd1/f2', 'd1/d2/f3']),
 564             max_manifest_depth=0)
 565         self.assertEqual(
 566             cwriter.manifest_text(),
 567             ". {} 0:8:d1/d2/f3 8:5:d1/f2 13:2:f1\n".format(blockhash))
 568
 569     def test_write_one_file(self):
 570         cwriter = arvados.CollectionWriter(self.api_client)
 571         with self.make_test_file() as testfile:
 572             cwriter.write_file(testfile.name)
 573             self.assertEqual(
 574                 cwriter.manifest_text(),
 575                 ". 098f6bcd4621d373cade4e832627b4f6+4 0:4:{}\n".format(
 576                     os.path.basename(testfile.name)))
 577
 578     def test_write_named_file(self):
 579         cwriter = arvados.CollectionWriter(self.api_client)
 580         with self.make_test_file() as testfile:
 581             cwriter.write_file(testfile.name, 'foo')
 582             self.assertEqual(cwriter.manifest_text(),
 583                              ". 098f6bcd4621d373cade4e832627b4f6+4 0:4:foo\n")
 584
 585     def test_write_multiple_files(self):
 586         cwriter = arvados.CollectionWriter(self.api_client)
 587         for letter in 'ABC':
 588             with self.make_test_file(letter) as testfile:
 589                 cwriter.write_file(testfile.name, letter)
 590         self.assertEqual(
 591             cwriter.manifest_text(),
 592             ". 902fbdd2b1df0c4f70b4a5d23525e932+3 0:1:A 1:1:B 2:1:C\n")
 593
 594     def test_basic_resume(self):
 595         cwriter = TestResumableWriter()
 596         with self.make_test_file() as testfile:
 597             cwriter.write_file(testfile.name, 'test')
 598             resumed = TestResumableWriter.from_state(cwriter.current_state())
 599         self.assertEquals(cwriter.manifest_text(), resumed.manifest_text(),
 600                           "resumed CollectionWriter had different manifest")
 601
 602     def test_resume_fails_when_missing_dependency(self):
 603         cwriter = TestResumableWriter()
 604         with self.make_test_file() as testfile:
 605             cwriter.write_file(testfile.name, 'test')
 606         self.assertRaises(arvados.errors.StaleWriterStateError,
 607                           TestResumableWriter.from_state,
 608                           cwriter.current_state())
 609
 610     def test_resume_fails_when_dependency_mtime_changed(self):
 611         cwriter = TestResumableWriter()
 612         with self.make_test_file() as testfile:
 613             cwriter.write_file(testfile.name, 'test')
 614             os.utime(testfile.name, (0, 0))
 615             self.assertRaises(arvados.errors.StaleWriterStateError,
 616                               TestResumableWriter.from_state,
 617                               cwriter.current_state())
 618
 619     def test_resume_fails_when_dependency_is_nonfile(self):
 620         cwriter = TestResumableWriter()
 621         cwriter.write_file('/dev/null', 'empty')
 622         self.assertRaises(arvados.errors.StaleWriterStateError,
 623                           TestResumableWriter.from_state,
 624                           cwriter.current_state())
 625
 626     def test_resume_fails_when_dependency_size_changed(self):
 627         cwriter = TestResumableWriter()
 628         with self.make_test_file() as testfile:
 629             cwriter.write_file(testfile.name, 'test')
 630             orig_mtime = os.fstat(testfile.fileno()).st_mtime
 631             testfile.write('extra')
 632             testfile.flush()
 633             os.utime(testfile.name, (orig_mtime, orig_mtime))
 634             self.assertRaises(arvados.errors.StaleWriterStateError,
 635                               TestResumableWriter.from_state,
 636                               cwriter.current_state())
 637
 638     def test_resume_fails_with_expired_locator(self):
 639         cwriter = TestResumableWriter()
 640         state = cwriter.current_state()
 641         # Add an expired locator to the state.
 642         state['_current_stream_locators'].append(''.join([
 643                     'a' * 32, '+1+A', 'b' * 40, '@', '10000000']))
 644         self.assertRaises(arvados.errors.StaleWriterStateError,
 645                           TestResumableWriter.from_state, state)
 646
 647     def test_arbitrary_objects_not_resumable(self):
 648         cwriter = TestResumableWriter()
 649         with open('/dev/null') as badfile:
 650             self.assertRaises(arvados.errors.AssertionError,
 651                               cwriter.write_file, badfile)
 652
 653     def test_arbitrary_writes_not_resumable(self):
 654         cwriter = TestResumableWriter()
 655         self.assertRaises(arvados.errors.AssertionError,
 656                           cwriter.write, "badtext")
 657
 658     def test_read_arbitrary_data_with_collection_reader(self):
 659         # arv-get relies on this to do "arv-get {keep-locator} -".
 660         self.write_foo_bar_baz()
 661         self.assertEqual(
 662             'foobar',
 663             arvados.CollectionReader(
 664                 '3858f62230ac3c915f300c664312c63f+6'
 665                 ).manifest_text())
 666
 667
 668 class CollectionTestMixin(object):
 669     PROXY_RESPONSE = {
 670         'items_available': 1,
 671         'items': [{
 672                 'uuid': 'zzzzz-bi6l4-mockproxy012345',
 673                 'owner_uuid': 'zzzzz-tpzed-mockowner012345',
 674                 'service_host': tutil.TEST_HOST,
 675                 'service_port': 65535,
 676                 'service_ssl_flag': True,
 677                 'service_type': 'proxy',
 678                 }]}
 679     API_COLLECTIONS = run_test_server.fixture('collections')
 680     DEFAULT_COLLECTION = API_COLLECTIONS['foo_file']
 681     DEFAULT_DATA_HASH = DEFAULT_COLLECTION['portable_data_hash']
 682     DEFAULT_MANIFEST = DEFAULT_COLLECTION['manifest_text']
 683     DEFAULT_UUID = DEFAULT_COLLECTION['uuid']
 684     ALT_COLLECTION = API_COLLECTIONS['bar_file']
 685     ALT_DATA_HASH = ALT_COLLECTION['portable_data_hash']
 686     ALT_MANIFEST = ALT_COLLECTION['manifest_text']
 687
 688     def _mock_api_call(self, mock_method, code, body):
 689         mock_method = mock_method().execute
 690         if code == 200:
 691             mock_method.return_value = body
 692         else:
 693             mock_method.side_effect = arvados.errors.ApiError(
 694                 tutil.fake_requests_response(code, None), "{}")
 695
 696     def mock_keep_services(self, api_mock, code, body):
 697         self._mock_api_call(api_mock.keep_services().accessible, code, body)
 698
 699     def api_client_mock(self, code=200):
 700         client = mock.MagicMock(name='api_client')
 701         self.mock_keep_services(client, code, self.PROXY_RESPONSE)
 702         return client
 703
 704
 705 @tutil.skip_sleep
 706 class CollectionReaderTestCase(unittest.TestCase, CollectionTestMixin):
 707     def mock_get_collection(self, api_mock, code, body):
 708         body = self.API_COLLECTIONS.get(body)
 709         self._mock_api_call(api_mock.collections().get, code, body)
 710
 711     def api_client_mock(self, code=200):
 712         client = super(CollectionReaderTestCase, self).api_client_mock(code)
 713         self.mock_get_collection(client, code, 'foo_file')
 714         return client
 715
 716     def test_init_no_default_retries(self):
 717         client = self.api_client_mock(200)
 718         reader = arvados.CollectionReader(self.DEFAULT_UUID, api_client=client)
 719         reader.manifest_text()
 720         client.collections().get().execute.assert_called_with(num_retries=0)
 721
 722     def test_uuid_init_success(self):
 723         client = self.api_client_mock(200)
 724         reader = arvados.CollectionReader(self.DEFAULT_UUID, api_client=client,
 725                                           num_retries=3)
 726         self.assertEqual(self.DEFAULT_COLLECTION['manifest_text'],
 727                          reader.manifest_text())
 728         client.collections().get().execute.assert_called_with(num_retries=3)
 729
 730     def test_uuid_init_failure_raises_api_error(self):
 731         client = self.api_client_mock(500)
 732         reader = arvados.CollectionReader(self.DEFAULT_UUID, api_client=client)
 733         with self.assertRaises(arvados.errors.ApiError):
 734             reader.manifest_text()
 735
 736     def test_locator_init(self):
 737         client = self.api_client_mock(200)
 738         # Ensure Keep will not return anything if asked.
 739         with tutil.mock_get_responses(None, 404):
 740             reader = arvados.CollectionReader(self.DEFAULT_DATA_HASH,
 741                                               api_client=client)
 742             self.assertEqual(self.DEFAULT_MANIFEST, reader.manifest_text())
 743
 744     def test_locator_init_fallback_to_keep(self):
 745         # crunch-job needs this to read manifests that have only ever
 746         # been written to Keep.
 747         client = self.api_client_mock(200)
 748         self.mock_get_collection(client, 404, None)
 749         with tutil.mock_get_responses(self.DEFAULT_MANIFEST, 200):
 750             reader = arvados.CollectionReader(self.DEFAULT_DATA_HASH,
 751                                               api_client=client)
 752             self.assertEqual(self.DEFAULT_MANIFEST, reader.manifest_text())
 753
 754     def test_uuid_init_no_fallback_to_keep(self):
 755         # Do not look up a collection UUID in Keep.
 756         client = self.api_client_mock(404)
 757         reader = arvados.CollectionReader(self.DEFAULT_UUID,
 758                                           api_client=client)
 759         with tutil.mock_get_responses(self.DEFAULT_MANIFEST, 200):
 760             with self.assertRaises(arvados.errors.ApiError):
 761                 reader.manifest_text()
 762
 763     def test_try_keep_first_if_permission_hint(self):
 764         # To verify that CollectionReader tries Keep first here, we
 765         # mock API server to return the wrong data.
 766         client = self.api_client_mock(200)
 767         with tutil.mock_get_responses(self.ALT_MANIFEST, 200):
 768             self.assertEqual(
 769                 self.ALT_MANIFEST,
 770                 arvados.CollectionReader(
 771                     self.ALT_DATA_HASH + '+Affffffffffffffffffffffffffffffffffffffff@fedcba98',
 772                     api_client=client).manifest_text())
 773
 774     def test_init_num_retries_propagated(self):
 775         # More of an integration test...
 776         client = self.api_client_mock(200)
 777         reader = arvados.CollectionReader(self.DEFAULT_UUID, api_client=client,
 778                                           num_retries=3)
 779         with tutil.mock_get_responses('foo', 500, 500, 200):
 780             self.assertEqual('foo',
 781                              ''.join(f.read(9) for f in reader.all_files()))
 782
 783     def test_read_nonnormalized_manifest_with_collection_reader(self):
 784         # client should be able to use CollectionReader on a manifest without normalizing it
 785         client = self.api_client_mock(500)
 786         nonnormal = ". acbd18db4cc2f85cedef654fccc4a4d8+3+Aabadbadbee@abeebdee 0:3:foo.txt 1:0:bar.txt 0:3:foo.txt\n"
 787         reader = arvados.CollectionReader(
 788             nonnormal,
 789             api_client=client, num_retries=0)
 790         # Ensure stripped_manifest() doesn't mangle our manifest in
 791         # any way other than stripping hints.
 792         self.assertEqual(
 793             re.sub('\+[^\d\s\+]+', '', nonnormal),
 794             reader.stripped_manifest())
 795         # Ensure stripped_manifest() didn't mutate our reader.
 796         self.assertEqual(nonnormal, reader.manifest_text())
 797         # Ensure the files appear in the order given in the manifest.
 798         self.assertEqual(
 799             [[6, '.', 'foo.txt'],
 800              [0, '.', 'bar.txt']],
 801             [[f.size(), f.stream_name(), f.name()]
 802              for f in reader.all_streams()[0].all_files()])
 803
 804
 805 @tutil.skip_sleep
 806 class CollectionWriterTestCase(unittest.TestCase, CollectionTestMixin):
 807     def mock_keep(self, body, *codes, **headers):
 808         headers.setdefault('x-keep-replicas-stored', 2)
 809         return tutil.mock_put_responses(body, *codes, **headers)
 810
 811     def foo_writer(self, **kwargs):
 812         api_client = self.api_client_mock()
 813         writer = arvados.CollectionWriter(api_client, **kwargs)
 814         writer.start_new_file('foo')
 815         writer.write('foo')
 816         return writer
 817
 818     def test_write_whole_collection(self):
 819         writer = self.foo_writer()
 820         with self.mock_keep(self.DEFAULT_DATA_HASH, 200, 200):
 821             self.assertEqual(self.DEFAULT_DATA_HASH, writer.finish())
 822
 823     def test_write_no_default(self):
 824         writer = self.foo_writer()
 825         with self.mock_keep(None, 500):
 826             with self.assertRaises(arvados.errors.KeepWriteError):
 827                 writer.finish()
 828
 829     def test_write_whole_collection_through_retries(self):
 830         writer = self.foo_writer(num_retries=2)
 831         with self.mock_keep(self.DEFAULT_DATA_HASH,
 832                             500, 500, 200, 500, 500, 200):
 833             self.assertEqual(self.DEFAULT_DATA_HASH, writer.finish())
 834
 835     def test_flush_data_retries(self):
 836         writer = self.foo_writer(num_retries=2)
 837         foo_hash = self.DEFAULT_MANIFEST.split()[1]
 838         with self.mock_keep(foo_hash, 500, 200):
 839             writer.flush_data()
 840         self.assertEqual(self.DEFAULT_MANIFEST, writer.manifest_text())
 841
 842
 843 if __name__ == '__main__':
 844     unittest.main()