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 copy
   7 import hashlib
   8 import mock
   9 import os
  10 import pprint
  11 import re
  12 import tempfile
  13 import unittest
  14
  15 import run_test_server
  16 from arvados._ranges import Range, LocatorAndRange
  17 from arvados.collection import Collection, CollectionReader
  18 from arvados.arvfile import SYNC_EXPLICIT
  19 import arvados_testutil as tutil
  20
  21 class TestResumableWriter(arvados.ResumableCollectionWriter):
  22     KEEP_BLOCK_SIZE = 1024  # PUT to Keep every 1K.
  23
  24     def current_state(self):
  25         return self.dump_state(copy.deepcopy)
  26
  27
  28 class ArvadosCollectionsTest(run_test_server.TestCaseWithServers,
  29                              tutil.ArvadosBaseTestCase):
  30     MAIN_SERVER = {}
  31
  32     @classmethod
  33     def setUpClass(cls):
  34         super(ArvadosCollectionsTest, cls).setUpClass()
  35         run_test_server.authorize_with('active')
  36         cls.api_client = arvados.api('v1')
  37         cls.keep_client = arvados.KeepClient(api_client=cls.api_client,
  38                                              local_store=cls.local_store)
  39
  40     def write_foo_bar_baz(self):
  41         cw = arvados.CollectionWriter(self.api_client)
  42         self.assertEqual(cw.current_stream_name(), '.',
  43                          'current_stream_name() should be "." now')
  44         cw.set_current_file_name('foo.txt')
  45         cw.write('foo')
  46         self.assertEqual(cw.current_file_name(), 'foo.txt',
  47                          'current_file_name() should be foo.txt now')
  48         cw.start_new_file('bar.txt')
  49         cw.write('bar')
  50         cw.start_new_stream('baz')
  51         cw.write('baz')
  52         cw.set_current_file_name('baz.txt')
  53         self.assertEqual(cw.manifest_text(),
  54                          ". 3858f62230ac3c915f300c664312c63f+6 0:3:foo.txt 3:3:bar.txt\n" +
  55                          "./baz 73feffa4b7f6bb68e44cf984c85f6e88+3 0:3:baz.txt\n",
  56                          "wrong manifest: got {}".format(cw.manifest_text()))
  57         cw.finish()
  58         return cw.portable_data_hash()
  59
  60     def test_keep_local_store(self):
  61         self.assertEqual(self.keep_client.put('foo'), 'acbd18db4cc2f85cedef654fccc4a4d8+3', 'wrong md5 hash from Keep.put')
  62         self.assertEqual(self.keep_client.get('acbd18db4cc2f85cedef654fccc4a4d8+3'), 'foo', 'wrong data from Keep.get')
  63
  64     def test_local_collection_writer(self):
  65         self.assertEqual(self.write_foo_bar_baz(),
  66                          '23ca013983d6239e98931cc779e68426+114',
  67                          'wrong locator hash: ' + self.write_foo_bar_baz())
  68
  69     def test_local_collection_reader(self):
  70         foobarbaz = self.write_foo_bar_baz()
  71         cr = arvados.CollectionReader(
  72             foobarbaz + '+Xzizzle', self.api_client)
  73         got = []
  74         for s in cr.all_streams():
  75             for f in s.all_files():
  76                 got += [[f.size(), f.stream_name(), f.name(), f.read(2**26)]]
  77         expected = [[3, '.', 'foo.txt', 'foo'],
  78                     [3, '.', 'bar.txt', 'bar'],
  79                     [3, './baz', 'baz.txt', 'baz']]
  80         self.assertEqual(got,
  81                          expected)
  82         stream0 = cr.all_streams()[0]
  83         self.assertEqual(stream0.readfrom(0, 0),
  84                          '',
  85                          'reading zero bytes should have returned empty string')
  86         self.assertEqual(stream0.readfrom(0, 2**26),
  87                          'foobar',
  88                          'reading entire stream failed')
  89         self.assertEqual(stream0.readfrom(2**26, 0),
  90                          '',
  91                          'reading zero bytes should have returned empty string')
  92
  93     def _test_subset(self, collection, expected):
  94         cr = arvados.CollectionReader(collection, self.api_client)
  95         for s in cr.all_streams():
  96             for ex in expected:
  97                 if ex[0] == s:
  98                     f = s.files()[ex[2]]
  99                     got = [f.size(), f.stream_name(), f.name(), "".join(f.readall(2**26))]
 100                     self.assertEqual(got,
 101                                      ex,
 102                                      'all_files|as_manifest did not preserve manifest contents: got %s expected %s' % (got, ex))
 103
 104     def test_collection_manifest_subset(self):
 105         foobarbaz = self.write_foo_bar_baz()
 106         self._test_subset(foobarbaz,
 107                           [[3, '.',     'bar.txt', 'bar'],
 108                            [3, '.',     'foo.txt', 'foo'],
 109                            [3, './baz', 'baz.txt', 'baz']])
 110         self._test_subset((". %s %s 0:3:foo.txt 3:3:bar.txt\n" %
 111                            (self.keep_client.put("foo"),
 112                             self.keep_client.put("bar"))),
 113                           [[3, '.', 'bar.txt', 'bar'],
 114                            [3, '.', 'foo.txt', 'foo']])
 115         self._test_subset((". %s %s 0:2:fo.txt 2:4:obar.txt\n" %
 116                            (self.keep_client.put("foo"),
 117                             self.keep_client.put("bar"))),
 118                           [[2, '.', 'fo.txt', 'fo'],
 119                            [4, '.', 'obar.txt', 'obar']])
 120         self._test_subset((". %s %s 0:2:fo.txt 2:0:zero.txt 2:2:ob.txt 4:2:ar.txt\n" %
 121                            (self.keep_client.put("foo"),
 122                             self.keep_client.put("bar"))),
 123                           [[2, '.', 'ar.txt', 'ar'],
 124                            [2, '.', 'fo.txt', 'fo'],
 125                            [2, '.', 'ob.txt', 'ob'],
 126                            [0, '.', 'zero.txt', '']])
 127
 128     def test_collection_empty_file(self):
 129         cw = arvados.CollectionWriter(self.api_client)
 130         cw.start_new_file('zero.txt')
 131         cw.write('')
 132
 133         self.assertEqual(cw.manifest_text(), ". d41d8cd98f00b204e9800998ecf8427e+0 0:0:zero.txt\n")
 134         self.check_manifest_file_sizes(cw.manifest_text(), [0])
 135         cw = arvados.CollectionWriter(self.api_client)
 136         cw.start_new_file('zero.txt')
 137         cw.write('')
 138         cw.start_new_file('one.txt')
 139         cw.write('1')
 140         cw.start_new_stream('foo')
 141         cw.start_new_file('zero.txt')
 142         cw.write('')
 143         self.check_manifest_file_sizes(cw.manifest_text(), [0,1,0])
 144
 145     def test_no_implicit_normalize(self):
 146         cw = arvados.CollectionWriter(self.api_client)
 147         cw.start_new_file('b')
 148         cw.write('b')
 149         cw.start_new_file('a')
 150         cw.write('')
 151         self.check_manifest_file_sizes(cw.manifest_text(), [1,0])
 152         self.check_manifest_file_sizes(
 153             arvados.CollectionReader(
 154                 cw.manifest_text()).manifest_text(normalize=True),
 155             [0,1])
 156
 157     def check_manifest_file_sizes(self, manifest_text, expect_sizes):
 158         cr = arvados.CollectionReader(manifest_text, self.api_client)
 159         got_sizes = []
 160         for f in cr.all_files():
 161             got_sizes += [f.size()]
 162         self.assertEqual(got_sizes, expect_sizes, "got wrong file sizes %s, expected %s" % (got_sizes, expect_sizes))
 163
 164     def test_normalized_collection(self):
 165         m1 = """. 5348b82a029fd9e971a811ce1f71360b+43 0:43:md5sum.txt
 166 . 085c37f02916da1cad16f93c54d899b7+41 0:41:md5sum.txt
 167 . 8b22da26f9f433dea0a10e5ec66d73ba+43 0:43:md5sum.txt
 168 """
 169         self.assertEqual(arvados.CollectionReader(m1, self.api_client).manifest_text(normalize=True),
 170                          """. 5348b82a029fd9e971a811ce1f71360b+43 085c37f02916da1cad16f93c54d899b7+41 8b22da26f9f433dea0a10e5ec66d73ba+43 0:127:md5sum.txt
 171 """)
 172
 173         m2 = """. 204e43b8a1185621ca55a94839582e6f+67108864 b9677abbac956bd3e86b1deb28dfac03+67108864 fc15aff2a762b13f521baf042140acec+67108864 323d2a3ce20370c4ca1d3462a344f8fd+25885655 0:227212247:var-GS000016015-ASM.tsv.bz2
 174 """
 175         self.assertEqual(arvados.CollectionReader(m2, self.api_client).manifest_text(normalize=True), m2)
 176
 177         m3 = """. 5348b82a029fd9e971a811ce1f71360b+43 3:40:md5sum.txt
 178 . 085c37f02916da1cad16f93c54d899b7+41 0:41:md5sum.txt
 179 . 8b22da26f9f433dea0a10e5ec66d73ba+43 0:43:md5sum.txt
 180 """
 181         self.assertEqual(arvados.CollectionReader(m3, self.api_client).manifest_text(normalize=True),
 182                          """. 5348b82a029fd9e971a811ce1f71360b+43 085c37f02916da1cad16f93c54d899b7+41 8b22da26f9f433dea0a10e5ec66d73ba+43 3:124:md5sum.txt
 183 """)
 184
 185         m4 = """. 204e43b8a1185621ca55a94839582e6f+67108864 0:3:foo/bar
 186 ./zzz 204e43b8a1185621ca55a94839582e6f+67108864 0:999:zzz
 187 ./foo 323d2a3ce20370c4ca1d3462a344f8fd+25885655 0:3:bar
 188 """
 189         self.assertEqual(arvados.CollectionReader(m4, self.api_client).manifest_text(normalize=True),
 190                          """./foo 204e43b8a1185621ca55a94839582e6f+67108864 323d2a3ce20370c4ca1d3462a344f8fd+25885655 0:3:bar 67108864:3:bar
 191 ./zzz 204e43b8a1185621ca55a94839582e6f+67108864 0:999:zzz
 192 """)
 193
 194         m5 = """. 204e43b8a1185621ca55a94839582e6f+67108864 0:3:foo/bar
 195 ./zzz 204e43b8a1185621ca55a94839582e6f+67108864 0:999:zzz
 196 ./foo 204e43b8a1185621ca55a94839582e6f+67108864 3:3:bar
 197 """
 198         self.assertEqual(arvados.CollectionReader(m5, self.api_client).manifest_text(normalize=True),
 199                          """./foo 204e43b8a1185621ca55a94839582e6f+67108864 0:6:bar
 200 ./zzz 204e43b8a1185621ca55a94839582e6f+67108864 0:999:zzz
 201 """)
 202
 203         with self.data_file('1000G_ref_manifest') as f6:
 204             m6 = f6.read()
 205             self.assertEqual(arvados.CollectionReader(m6, self.api_client).manifest_text(normalize=True), m6)
 206
 207         with self.data_file('jlake_manifest') as f7:
 208             m7 = f7.read()
 209             self.assertEqual(arvados.CollectionReader(m7, self.api_client).manifest_text(normalize=True), m7)
 210
 211         m8 = """./a\\040b\\040c 59ca0efa9f5633cb0371bbc0355478d8+13 0:13:hello\\040world.txt
 212 """
 213         self.assertEqual(arvados.CollectionReader(m8, self.api_client).manifest_text(normalize=True), m8)
 214
 215     def test_locators_and_ranges(self):
 216         blocks2 = [Range('a', 0, 10),
 217                    Range('b', 10, 10),
 218                    Range('c', 20, 10),
 219                    Range('d', 30, 10),
 220                    Range('e', 40, 10),
 221                    Range('f', 50, 10)]
 222
 223         self.assertEqual(arvados.locators_and_ranges(blocks2,  2,  2), [LocatorAndRange('a', 10, 2, 2)])
 224         self.assertEqual(arvados.locators_and_ranges(blocks2, 12, 2), [LocatorAndRange('b', 10, 2, 2)])
 225         self.assertEqual(arvados.locators_and_ranges(blocks2, 22, 2), [LocatorAndRange('c', 10, 2, 2)])
 226         self.assertEqual(arvados.locators_and_ranges(blocks2, 32, 2), [LocatorAndRange('d', 10, 2, 2)])
 227         self.assertEqual(arvados.locators_and_ranges(blocks2, 42, 2), [LocatorAndRange('e', 10, 2, 2)])
 228         self.assertEqual(arvados.locators_and_ranges(blocks2, 52, 2), [LocatorAndRange('f', 10, 2, 2)])
 229         self.assertEqual(arvados.locators_and_ranges(blocks2, 62, 2), [])
 230         self.assertEqual(arvados.locators_and_ranges(blocks2, -2, 2), [])
 231
 232         self.assertEqual(arvados.locators_and_ranges(blocks2,  0,  2), [LocatorAndRange('a', 10, 0, 2)])
 233         self.assertEqual(arvados.locators_and_ranges(blocks2, 10, 2), [LocatorAndRange('b', 10, 0, 2)])
 234         self.assertEqual(arvados.locators_and_ranges(blocks2, 20, 2), [LocatorAndRange('c', 10, 0, 2)])
 235         self.assertEqual(arvados.locators_and_ranges(blocks2, 30, 2), [LocatorAndRange('d', 10, 0, 2)])
 236         self.assertEqual(arvados.locators_and_ranges(blocks2, 40, 2), [LocatorAndRange('e', 10, 0, 2)])
 237         self.assertEqual(arvados.locators_and_ranges(blocks2, 50, 2), [LocatorAndRange('f', 10, 0, 2)])
 238         self.assertEqual(arvados.locators_and_ranges(blocks2, 60, 2), [])
 239         self.assertEqual(arvados.locators_and_ranges(blocks2, -2, 2), [])
 240
 241         self.assertEqual(arvados.locators_and_ranges(blocks2,  9,  2), [LocatorAndRange('a', 10, 9, 1), LocatorAndRange('b', 10, 0, 1)])
 242         self.assertEqual(arvados.locators_and_ranges(blocks2, 19, 2), [LocatorAndRange('b', 10, 9, 1), LocatorAndRange('c', 10, 0, 1)])
 243         self.assertEqual(arvados.locators_and_ranges(blocks2, 29, 2), [LocatorAndRange('c', 10, 9, 1), LocatorAndRange('d', 10, 0, 1)])
 244         self.assertEqual(arvados.locators_and_ranges(blocks2, 39, 2), [LocatorAndRange('d', 10, 9, 1), LocatorAndRange('e', 10, 0, 1)])
 245         self.assertEqual(arvados.locators_and_ranges(blocks2, 49, 2), [LocatorAndRange('e', 10, 9, 1), LocatorAndRange('f', 10, 0, 1)])
 246         self.assertEqual(arvados.locators_and_ranges(blocks2, 59, 2), [LocatorAndRange('f', 10, 9, 1)])
 247
 248
 249         blocks3 = [Range('a', 0, 10),
 250                   Range('b', 10, 10),
 251                   Range('c', 20, 10),
 252                   Range('d', 30, 10),
 253                   Range('e', 40, 10),
 254                   Range('f', 50, 10),
 255                    Range('g', 60, 10)]
 256
 257         self.assertEqual(arvados.locators_and_ranges(blocks3,  2,  2), [LocatorAndRange('a', 10, 2, 2)])
 258         self.assertEqual(arvados.locators_and_ranges(blocks3, 12, 2), [LocatorAndRange('b', 10, 2, 2)])
 259         self.assertEqual(arvados.locators_and_ranges(blocks3, 22, 2), [LocatorAndRange('c', 10, 2, 2)])
 260         self.assertEqual(arvados.locators_and_ranges(blocks3, 32, 2), [LocatorAndRange('d', 10, 2, 2)])
 261         self.assertEqual(arvados.locators_and_ranges(blocks3, 42, 2), [LocatorAndRange('e', 10, 2, 2)])
 262         self.assertEqual(arvados.locators_and_ranges(blocks3, 52, 2), [LocatorAndRange('f', 10, 2, 2)])
 263         self.assertEqual(arvados.locators_and_ranges(blocks3, 62, 2), [LocatorAndRange('g', 10, 2, 2)])
 264
 265
 266         blocks = [Range('a', 0, 10),
 267                   Range('b', 10, 15),
 268                   Range('c', 25, 5)]
 269         self.assertEqual(arvados.locators_and_ranges(blocks, 1, 0), [])
 270         self.assertEqual(arvados.locators_and_ranges(blocks, 0, 5), [LocatorAndRange('a', 10, 0, 5)])
 271         self.assertEqual(arvados.locators_and_ranges(blocks, 3, 5), [LocatorAndRange('a', 10, 3, 5)])
 272         self.assertEqual(arvados.locators_and_ranges(blocks, 0, 10), [LocatorAndRange('a', 10, 0, 10)])
 273
 274         self.assertEqual(arvados.locators_and_ranges(blocks, 0, 11), [LocatorAndRange('a', 10, 0, 10),
 275                                                                       LocatorAndRange('b', 15, 0, 1)])
 276         self.assertEqual(arvados.locators_and_ranges(blocks, 1, 11), [LocatorAndRange('a', 10, 1, 9),
 277                                                                       LocatorAndRange('b', 15, 0, 2)])
 278         self.assertEqual(arvados.locators_and_ranges(blocks, 0, 25), [LocatorAndRange('a', 10, 0, 10),
 279                                                                       LocatorAndRange('b', 15, 0, 15)])
 280
 281         self.assertEqual(arvados.locators_and_ranges(blocks, 0, 30), [LocatorAndRange('a', 10, 0, 10),
 282                                                                       LocatorAndRange('b', 15, 0, 15),
 283                                                                       LocatorAndRange('c', 5, 0, 5)])
 284         self.assertEqual(arvados.locators_and_ranges(blocks, 1, 30), [LocatorAndRange('a', 10, 1, 9),
 285                                                                       LocatorAndRange('b', 15, 0, 15),
 286                                                                       LocatorAndRange('c', 5, 0, 5)])
 287         self.assertEqual(arvados.locators_and_ranges(blocks, 0, 31), [LocatorAndRange('a', 10, 0, 10),
 288                                                                       LocatorAndRange('b', 15, 0, 15),
 289                                                                       LocatorAndRange('c', 5, 0, 5)])
 290
 291         self.assertEqual(arvados.locators_and_ranges(blocks, 15, 5), [LocatorAndRange('b', 15, 5, 5)])
 292
 293         self.assertEqual(arvados.locators_and_ranges(blocks, 8, 17), [LocatorAndRange('a', 10, 8, 2),
 294                                                                       LocatorAndRange('b', 15, 0, 15)])
 295
 296         self.assertEqual(arvados.locators_and_ranges(blocks, 8, 20), [LocatorAndRange('a', 10, 8, 2),
 297                                                                       LocatorAndRange('b', 15, 0, 15),
 298                                                                       LocatorAndRange('c', 5, 0, 3)])
 299
 300         self.assertEqual(arvados.locators_and_ranges(blocks, 26, 2), [LocatorAndRange('c', 5, 1, 2)])
 301
 302         self.assertEqual(arvados.locators_and_ranges(blocks, 9, 15), [LocatorAndRange('a', 10, 9, 1),
 303                                                                       LocatorAndRange('b', 15, 0, 14)])
 304         self.assertEqual(arvados.locators_and_ranges(blocks, 10, 15), [LocatorAndRange('b', 15, 0, 15)])
 305         self.assertEqual(arvados.locators_and_ranges(blocks, 11, 15), [LocatorAndRange('b', 15, 1, 14),
 306                                                                        LocatorAndRange('c', 5, 0, 1)])
 307
 308     class MockKeep(object):
 309         def __init__(self, content, num_retries=0):
 310             self.content = content
 311
 312         def get(self, locator, num_retries=0):
 313             return self.content[locator]
 314
 315     def test_stream_reader(self):
 316         keepblocks = {'aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa+10': 'abcdefghij',
 317                       'bbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbb+15': 'klmnopqrstuvwxy',
 318                       'cccccccccccccccccccccccccccccccc+5': 'z0123'}
 319         mk = self.MockKeep(keepblocks)
 320
 321         sr = arvados.StreamReader([".", "aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa+10", "bbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbb+15", "cccccccccccccccccccccccccccccccc+5", "0:30:foo"], mk)
 322
 323         content = 'abcdefghijklmnopqrstuvwxyz0123456789'
 324
 325         self.assertEqual(sr.readfrom(0, 30), content[0:30])
 326         self.assertEqual(sr.readfrom(2, 30), content[2:30])
 327
 328         self.assertEqual(sr.readfrom(2, 8), content[2:10])
 329         self.assertEqual(sr.readfrom(0, 10), content[0:10])
 330
 331         self.assertEqual(sr.readfrom(0, 5), content[0:5])
 332         self.assertEqual(sr.readfrom(5, 5), content[5:10])
 333         self.assertEqual(sr.readfrom(10, 5), content[10:15])
 334         self.assertEqual(sr.readfrom(15, 5), content[15:20])
 335         self.assertEqual(sr.readfrom(20, 5), content[20:25])
 336         self.assertEqual(sr.readfrom(25, 5), content[25:30])
 337         self.assertEqual(sr.readfrom(30, 5), '')
 338
 339     def test_extract_file(self):
 340         m1 = """. 5348b82a029fd9e971a811ce1f71360b+43 0:43:md5sum.txt
 341 . 085c37f02916da1cad16f93c54d899b7+41 0:41:md6sum.txt
 342 . 8b22da26f9f433dea0a10e5ec66d73ba+43 0:43:md7sum.txt
 343 . 085c37f02916da1cad16f93c54d899b7+41 5348b82a029fd9e971a811ce1f71360b+43 8b22da26f9f433dea0a10e5ec66d73ba+43 47:80:md8sum.txt
 344 . 085c37f02916da1cad16f93c54d899b7+41 5348b82a029fd9e971a811ce1f71360b+43 8b22da26f9f433dea0a10e5ec66d73ba+43 40:80:md9sum.txt
 345 """
 346
 347         m2 = arvados.CollectionReader(m1, self.api_client).manifest_text(normalize=True)
 348
 349         self.assertEqual(m2,
 350                          ". 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")
 351         files = arvados.CollectionReader(
 352             m2, self.api_client).all_streams()[0].files()
 353
 354         self.assertEqual(files['md5sum.txt'].as_manifest(),
 355                          ". 5348b82a029fd9e971a811ce1f71360b+43 0:43:md5sum.txt\n")
 356         self.assertEqual(files['md6sum.txt'].as_manifest(),
 357                          ". 085c37f02916da1cad16f93c54d899b7+41 0:41:md6sum.txt\n")
 358         self.assertEqual(files['md7sum.txt'].as_manifest(),
 359                          ". 8b22da26f9f433dea0a10e5ec66d73ba+43 0:43:md7sum.txt\n")
 360         self.assertEqual(files['md9sum.txt'].as_manifest(),
 361                          ". 085c37f02916da1cad16f93c54d899b7+41 5348b82a029fd9e971a811ce1f71360b+43 8b22da26f9f433dea0a10e5ec66d73ba+43 40:80:md9sum.txt\n")
 362
 363     def test_write_directory_tree(self):
 364         cwriter = arvados.CollectionWriter(self.api_client)
 365         cwriter.write_directory_tree(self.build_directory_tree(
 366                 ['basefile', 'subdir/subfile']))
 367         self.assertEqual(cwriter.manifest_text(),
 368                          """. c5110c5ac93202d8e0f9e381f22bac0f+8 0:8:basefile
 369 ./subdir 1ca4dec89403084bf282ad31e6cf7972+14 0:14:subfile\n""")
 370
 371     def test_write_named_directory_tree(self):
 372         cwriter = arvados.CollectionWriter(self.api_client)
 373         cwriter.write_directory_tree(self.build_directory_tree(
 374                 ['basefile', 'subdir/subfile']), 'root')
 375         self.assertEqual(
 376             cwriter.manifest_text(),
 377             """./root c5110c5ac93202d8e0f9e381f22bac0f+8 0:8:basefile
 378 ./root/subdir 1ca4dec89403084bf282ad31e6cf7972+14 0:14:subfile\n""")
 379
 380     def test_write_directory_tree_in_one_stream(self):
 381         cwriter = arvados.CollectionWriter(self.api_client)
 382         cwriter.write_directory_tree(self.build_directory_tree(
 383                 ['basefile', 'subdir/subfile']), max_manifest_depth=0)
 384         self.assertEqual(cwriter.manifest_text(),
 385                          """. 4ace875ffdc6824a04950f06858f4465+22 0:8:basefile 8:14:subdir/subfile\n""")
 386
 387     def test_write_directory_tree_with_limited_recursion(self):
 388         cwriter = arvados.CollectionWriter(self.api_client)
 389         cwriter.write_directory_tree(
 390             self.build_directory_tree(['f1', 'd1/f2', 'd1/d2/f3']),
 391             max_manifest_depth=1)
 392         self.assertEqual(cwriter.manifest_text(),
 393                          """. bd19836ddb62c11c55ab251ccaca5645+2 0:2:f1
 394 ./d1 50170217e5b04312024aa5cd42934494+13 0:8:d2/f3 8:5:f2\n""")
 395
 396     def test_write_directory_tree_with_zero_recursion(self):
 397         cwriter = arvados.CollectionWriter(self.api_client)
 398         content = 'd1/d2/f3d1/f2f1'
 399         blockhash = hashlib.md5(content).hexdigest() + '+' + str(len(content))
 400         cwriter.write_directory_tree(
 401             self.build_directory_tree(['f1', 'd1/f2', 'd1/d2/f3']),
 402             max_manifest_depth=0)
 403         self.assertEqual(
 404             cwriter.manifest_text(),
 405             ". {} 0:8:d1/d2/f3 8:5:d1/f2 13:2:f1\n".format(blockhash))
 406
 407     def test_write_one_file(self):
 408         cwriter = arvados.CollectionWriter(self.api_client)
 409         with self.make_test_file() as testfile:
 410             cwriter.write_file(testfile.name)
 411             self.assertEqual(
 412                 cwriter.manifest_text(),
 413                 ". 098f6bcd4621d373cade4e832627b4f6+4 0:4:{}\n".format(
 414                     os.path.basename(testfile.name)))
 415
 416     def test_write_named_file(self):
 417         cwriter = arvados.CollectionWriter(self.api_client)
 418         with self.make_test_file() as testfile:
 419             cwriter.write_file(testfile.name, 'foo')
 420             self.assertEqual(cwriter.manifest_text(),
 421                              ". 098f6bcd4621d373cade4e832627b4f6+4 0:4:foo\n")
 422
 423     def test_write_multiple_files(self):
 424         cwriter = arvados.CollectionWriter(self.api_client)
 425         for letter in 'ABC':
 426             with self.make_test_file(letter) as testfile:
 427                 cwriter.write_file(testfile.name, letter)
 428         self.assertEqual(
 429             cwriter.manifest_text(),
 430             ". 902fbdd2b1df0c4f70b4a5d23525e932+3 0:1:A 1:1:B 2:1:C\n")
 431
 432     def test_basic_resume(self):
 433         cwriter = TestResumableWriter()
 434         with self.make_test_file() as testfile:
 435             cwriter.write_file(testfile.name, 'test')
 436             resumed = TestResumableWriter.from_state(cwriter.current_state())
 437         self.assertEquals(cwriter.manifest_text(), resumed.manifest_text(),
 438                           "resumed CollectionWriter had different manifest")
 439
 440     def test_resume_fails_when_missing_dependency(self):
 441         cwriter = TestResumableWriter()
 442         with self.make_test_file() as testfile:
 443             cwriter.write_file(testfile.name, 'test')
 444         self.assertRaises(arvados.errors.StaleWriterStateError,
 445                           TestResumableWriter.from_state,
 446                           cwriter.current_state())
 447
 448     def test_resume_fails_when_dependency_mtime_changed(self):
 449         cwriter = TestResumableWriter()
 450         with self.make_test_file() as testfile:
 451             cwriter.write_file(testfile.name, 'test')
 452             os.utime(testfile.name, (0, 0))
 453             self.assertRaises(arvados.errors.StaleWriterStateError,
 454                               TestResumableWriter.from_state,
 455                               cwriter.current_state())
 456
 457     def test_resume_fails_when_dependency_is_nonfile(self):
 458         cwriter = TestResumableWriter()
 459         cwriter.write_file('/dev/null', 'empty')
 460         self.assertRaises(arvados.errors.StaleWriterStateError,
 461                           TestResumableWriter.from_state,
 462                           cwriter.current_state())
 463
 464     def test_resume_fails_when_dependency_size_changed(self):
 465         cwriter = TestResumableWriter()
 466         with self.make_test_file() as testfile:
 467             cwriter.write_file(testfile.name, 'test')
 468             orig_mtime = os.fstat(testfile.fileno()).st_mtime
 469             testfile.write('extra')
 470             testfile.flush()
 471             os.utime(testfile.name, (orig_mtime, orig_mtime))
 472             self.assertRaises(arvados.errors.StaleWriterStateError,
 473                               TestResumableWriter.from_state,
 474                               cwriter.current_state())
 475
 476     def test_resume_fails_with_expired_locator(self):
 477         cwriter = TestResumableWriter()
 478         state = cwriter.current_state()
 479         # Add an expired locator to the state.
 480         state['_current_stream_locators'].append(''.join([
 481                     'a' * 32, '+1+A', 'b' * 40, '@', '10000000']))
 482         self.assertRaises(arvados.errors.StaleWriterStateError,
 483                           TestResumableWriter.from_state, state)
 484
 485     def test_arbitrary_objects_not_resumable(self):
 486         cwriter = TestResumableWriter()
 487         with open('/dev/null') as badfile:
 488             self.assertRaises(arvados.errors.AssertionError,
 489                               cwriter.write_file, badfile)
 490
 491     def test_arbitrary_writes_not_resumable(self):
 492         cwriter = TestResumableWriter()
 493         self.assertRaises(arvados.errors.AssertionError,
 494                           cwriter.write, "badtext")
 495
 496     def test_read_arbitrary_data_with_collection_reader(self):
 497         # arv-get relies on this to do "arv-get {keep-locator} -".
 498         self.write_foo_bar_baz()
 499         self.assertEqual(
 500             'foobar',
 501             arvados.CollectionReader(
 502                 '3858f62230ac3c915f300c664312c63f+6'
 503                 ).manifest_text())
 504
 505
 506 class CollectionTestMixin(tutil.ApiClientMock):
 507     API_COLLECTIONS = run_test_server.fixture('collections')
 508     DEFAULT_COLLECTION = API_COLLECTIONS['foo_file']
 509     DEFAULT_DATA_HASH = DEFAULT_COLLECTION['portable_data_hash']
 510     DEFAULT_MANIFEST = DEFAULT_COLLECTION['manifest_text']
 511     DEFAULT_UUID = DEFAULT_COLLECTION['uuid']
 512     ALT_COLLECTION = API_COLLECTIONS['bar_file']
 513     ALT_DATA_HASH = ALT_COLLECTION['portable_data_hash']
 514     ALT_MANIFEST = ALT_COLLECTION['manifest_text']
 515
 516     def api_client_mock(self, status=200):
 517         client = super(CollectionTestMixin, self).api_client_mock()
 518         self.mock_keep_services(client, status=status, service_type='proxy', count=1)
 519         return client
 520
 521
 522 @tutil.skip_sleep
 523 class CollectionReaderTestCase(unittest.TestCase, CollectionTestMixin):
 524     def mock_get_collection(self, api_mock, code, body):
 525         body = self.API_COLLECTIONS.get(body)
 526         self._mock_api_call(api_mock.collections().get, code, body)
 527
 528     def api_client_mock(self, status=200):
 529         client = super(CollectionReaderTestCase, self).api_client_mock()
 530         self.mock_get_collection(client, status, 'foo_file')
 531         return client
 532
 533     def test_init_no_default_retries(self):
 534         client = self.api_client_mock(200)
 535         reader = arvados.CollectionReader(self.DEFAULT_UUID, api_client=client)
 536         reader.manifest_text()
 537         client.collections().get().execute.assert_called_with(num_retries=0)
 538
 539     def test_uuid_init_success(self):
 540         client = self.api_client_mock(200)
 541         reader = arvados.CollectionReader(self.DEFAULT_UUID, api_client=client,
 542                                           num_retries=3)
 543         self.assertEqual(self.DEFAULT_COLLECTION['manifest_text'],
 544                          reader.manifest_text())
 545         client.collections().get().execute.assert_called_with(num_retries=3)
 546
 547     def test_uuid_init_failure_raises_api_error(self):
 548         client = self.api_client_mock(500)
 549         with self.assertRaises(arvados.errors.ApiError):
 550             reader = arvados.CollectionReader(self.DEFAULT_UUID, api_client=client)
 551
 552     def test_locator_init(self):
 553         client = self.api_client_mock(200)
 554         # Ensure Keep will not return anything if asked.
 555         with tutil.mock_get_responses(None, 404):
 556             reader = arvados.CollectionReader(self.DEFAULT_DATA_HASH,
 557                                               api_client=client)
 558             self.assertEqual(self.DEFAULT_MANIFEST, reader.manifest_text())
 559
 560     def test_locator_init_fallback_to_keep(self):
 561         # crunch-job needs this to read manifests that have only ever
 562         # been written to Keep.
 563         client = self.api_client_mock(200)
 564         self.mock_get_collection(client, 404, None)
 565         with tutil.mock_get_responses(self.DEFAULT_MANIFEST, 200):
 566             reader = arvados.CollectionReader(self.DEFAULT_DATA_HASH,
 567                                               api_client=client)
 568             self.assertEqual(self.DEFAULT_MANIFEST, reader.manifest_text())
 569
 570     def test_uuid_init_no_fallback_to_keep(self):
 571         # Do not look up a collection UUID in Keep.
 572         client = self.api_client_mock(404)
 573         with tutil.mock_get_responses(self.DEFAULT_MANIFEST, 200):
 574             with self.assertRaises(arvados.errors.ApiError):
 575                 reader = arvados.CollectionReader(self.DEFAULT_UUID,
 576                                                   api_client=client)
 577
 578     def test_try_keep_first_if_permission_hint(self):
 579         # To verify that CollectionReader tries Keep first here, we
 580         # mock API server to return the wrong data.
 581         client = self.api_client_mock(200)
 582         with tutil.mock_get_responses(self.ALT_MANIFEST, 200):
 583             self.assertEqual(
 584                 self.ALT_MANIFEST,
 585                 arvados.CollectionReader(
 586                     self.ALT_DATA_HASH + '+Affffffffffffffffffffffffffffffffffffffff@fedcba98',
 587                     api_client=client).manifest_text())
 588
 589     def test_init_num_retries_propagated(self):
 590         # More of an integration test...
 591         client = self.api_client_mock(200)
 592         reader = arvados.CollectionReader(self.DEFAULT_UUID, api_client=client,
 593                                           num_retries=3)
 594         with tutil.mock_get_responses('foo', 500, 500, 200):
 595             self.assertEqual('foo',
 596                              ''.join(f.read(9) for f in reader.all_files()))
 597
 598     def test_read_nonnormalized_manifest_with_collection_reader(self):
 599         # client should be able to use CollectionReader on a manifest without normalizing it
 600         client = self.api_client_mock(500)
 601         nonnormal = ". acbd18db4cc2f85cedef654fccc4a4d8+3+Aabadbadbee@abeebdee 0:3:foo.txt 1:0:bar.txt 0:3:foo.txt\n"
 602         reader = arvados.CollectionReader(
 603             nonnormal,
 604             api_client=client, num_retries=0)
 605         # Ensure stripped_manifest() doesn't mangle our manifest in
 606         # any way other than stripping hints.
 607         self.assertEqual(
 608             re.sub('\+[^\d\s\+]+', '', nonnormal),
 609             reader.stripped_manifest())
 610         # Ensure stripped_manifest() didn't mutate our reader.
 611         self.assertEqual(nonnormal, reader.manifest_text())
 612         # Ensure the files appear in the order given in the manifest.
 613         self.assertEqual(
 614             [[6, '.', 'foo.txt'],
 615              [0, '.', 'bar.txt']],
 616             [[f.size(), f.stream_name(), f.name()]
 617              for f in reader.all_streams()[0].all_files()])
 618
 619     def test_read_empty_collection(self):
 620         client = self.api_client_mock(200)
 621         self.mock_get_collection(client, 200, 'empty')
 622         reader = arvados.CollectionReader('d41d8cd98f00b204e9800998ecf8427e+0',
 623                                           api_client=client)
 624         self.assertEqual('', reader.manifest_text())
 625
 626     def test_api_response(self):
 627         client = self.api_client_mock()
 628         reader = arvados.CollectionReader(self.DEFAULT_UUID, api_client=client)
 629         self.assertEqual(self.DEFAULT_COLLECTION, reader.api_response())
 630
 631     def test_api_response_with_collection_from_keep(self):
 632         client = self.api_client_mock()
 633         self.mock_get_collection(client, 404, 'foo')
 634         with tutil.mock_get_responses(self.DEFAULT_MANIFEST, 200):
 635             reader = arvados.CollectionReader(self.DEFAULT_DATA_HASH,
 636                                               api_client=client)
 637             api_response = reader.api_response()
 638         self.assertIsNone(api_response)
 639
 640     def check_open_file(self, coll_file, stream_name, file_name, file_size):
 641         self.assertFalse(coll_file.closed, "returned file is not open")
 642         self.assertEqual(stream_name, coll_file.stream_name())
 643         self.assertEqual(file_name, coll_file.name())
 644         self.assertEqual(file_size, coll_file.size())
 645
 646     def test_open_collection_file_one_argument(self):
 647         client = self.api_client_mock(200)
 648         reader = arvados.CollectionReader(self.DEFAULT_UUID, api_client=client)
 649         cfile = reader.open('./foo')
 650         self.check_open_file(cfile, '.', 'foo', 3)
 651
 652     def test_open_deep_file(self):
 653         coll_name = 'collection_with_files_in_subdir'
 654         client = self.api_client_mock(200)
 655         self.mock_get_collection(client, 200, coll_name)
 656         reader = arvados.CollectionReader(
 657             self.API_COLLECTIONS[coll_name]['uuid'], api_client=client)
 658         cfile = reader.open('./subdir2/subdir3/file2_in_subdir3.txt')
 659         self.check_open_file(cfile, './subdir2/subdir3', 'file2_in_subdir3.txt',
 660                              32)
 661
 662     def test_open_nonexistent_stream(self):
 663         client = self.api_client_mock(200)
 664         reader = arvados.CollectionReader(self.DEFAULT_UUID, api_client=client)
 665         self.assertRaises(IOError, reader.open, './nonexistent/foo')
 666
 667     def test_open_nonexistent_file(self):
 668         client = self.api_client_mock(200)
 669         reader = arvados.CollectionReader(self.DEFAULT_UUID, api_client=client)
 670         self.assertRaises(IOError, reader.open, 'nonexistent')
 671
 672
 673 @tutil.skip_sleep
 674 class CollectionWriterTestCase(unittest.TestCase, CollectionTestMixin):
 675     def mock_keep(self, body, *codes, **headers):
 676         headers.setdefault('x-keep-replicas-stored', 2)
 677         return tutil.mock_put_responses(body, *codes, **headers)
 678
 679     def foo_writer(self, **kwargs):
 680         kwargs.setdefault('api_client', self.api_client_mock())
 681         writer = arvados.CollectionWriter(**kwargs)
 682         writer.start_new_file('foo')
 683         writer.write('foo')
 684         return writer
 685
 686     def test_write_whole_collection(self):
 687         writer = self.foo_writer()
 688         with self.mock_keep(self.DEFAULT_DATA_HASH, 200, 200):
 689             self.assertEqual(self.DEFAULT_DATA_HASH, writer.finish())
 690
 691     def test_write_no_default(self):
 692         writer = self.foo_writer()
 693         with self.mock_keep(None, 500):
 694             with self.assertRaises(arvados.errors.KeepWriteError):
 695                 writer.finish()
 696
 697     def test_write_insufficient_replicas_via_proxy(self):
 698         writer = self.foo_writer(replication=3)
 699         with self.mock_keep(None, 200, headers={'x-keep-replicas-stored': 2}):
 700             with self.assertRaises(arvados.errors.KeepWriteError):
 701                 writer.manifest_text()
 702
 703     def test_write_insufficient_replicas_via_disks(self):
 704         client = mock.MagicMock(name='api_client')
 705         with self.mock_keep(
 706                 None, 200, 200,
 707                 **{'x-keep-replicas-stored': 1}) as keepmock:
 708             self.mock_keep_services(client, status=200, service_type='disk', count=2)
 709             writer = self.foo_writer(api_client=client, replication=3)
 710             with self.assertRaises(arvados.errors.KeepWriteError):
 711                 writer.manifest_text()
 712
 713     def test_write_three_replicas(self):
 714         client = mock.MagicMock(name='api_client')
 715         with self.mock_keep(
 716                 None, 500, 500, 500, 200, 200, 200,
 717                 **{'x-keep-replicas-stored': 1}) as keepmock:
 718             self.mock_keep_services(client, status=200, service_type='disk', count=6)
 719             writer = self.foo_writer(api_client=client, replication=3)
 720             writer.manifest_text()
 721             # keepmock is the mock session constructor; keepmock.return_value
 722             # is the mock session object, and keepmock.return_value.put is the
 723             # actual mock method of interest.
 724             self.assertEqual(6, keepmock.return_value.put.call_count)
 725
 726     def test_write_whole_collection_through_retries(self):
 727         writer = self.foo_writer(num_retries=2)
 728         with self.mock_keep(self.DEFAULT_DATA_HASH,
 729                             500, 500, 200, 500, 500, 200):
 730             self.assertEqual(self.DEFAULT_DATA_HASH, writer.finish())
 731
 732     def test_flush_data_retries(self):
 733         writer = self.foo_writer(num_retries=2)
 734         foo_hash = self.DEFAULT_MANIFEST.split()[1]
 735         with self.mock_keep(foo_hash, 500, 200):
 736             writer.flush_data()
 737         self.assertEqual(self.DEFAULT_MANIFEST, writer.manifest_text())
 738
 739     def test_one_open(self):
 740         client = self.api_client_mock()
 741         writer = arvados.CollectionWriter(client)
 742         with writer.open('out') as out_file:
 743             self.assertEqual('.', writer.current_stream_name())
 744             self.assertEqual('out', writer.current_file_name())
 745             out_file.write('test data')
 746             data_loc = hashlib.md5('test data').hexdigest() + '+9'
 747         self.assertTrue(out_file.closed, "writer file not closed after context")
 748         self.assertRaises(ValueError, out_file.write, 'extra text')
 749         with self.mock_keep(data_loc, 200) as keep_mock:
 750             self.assertEqual(". {} 0:9:out\n".format(data_loc),
 751                              writer.manifest_text())
 752
 753     def test_open_writelines(self):
 754         client = self.api_client_mock()
 755         writer = arvados.CollectionWriter(client)
 756         with writer.open('six') as out_file:
 757             out_file.writelines(['12', '34', '56'])
 758             data_loc = hashlib.md5('123456').hexdigest() + '+6'
 759         with self.mock_keep(data_loc, 200) as keep_mock:
 760             self.assertEqual(". {} 0:6:six\n".format(data_loc),
 761                              writer.manifest_text())
 762
 763     def test_open_flush(self):
 764         client = self.api_client_mock()
 765         data_loc1 = hashlib.md5('flush1').hexdigest() + '+6'
 766         data_loc2 = hashlib.md5('flush2').hexdigest() + '+6'
 767         with self.mock_keep((data_loc1, 200), (data_loc2, 200)) as keep_mock:
 768             writer = arvados.CollectionWriter(client)
 769             with writer.open('flush_test') as out_file:
 770                 out_file.write('flush1')
 771                 out_file.flush()
 772                 out_file.write('flush2')
 773             self.assertEqual(". {} {} 0:12:flush_test\n".format(data_loc1,
 774                                                                 data_loc2),
 775                              writer.manifest_text())
 776
 777     def test_two_opens_same_stream(self):
 778         client = self.api_client_mock()
 779         writer = arvados.CollectionWriter(client)
 780         with writer.open('.', '1') as out_file:
 781             out_file.write('1st')
 782         with writer.open('.', '2') as out_file:
 783             out_file.write('2nd')
 784         data_loc = hashlib.md5('1st2nd').hexdigest() + '+6'
 785         with self.mock_keep(data_loc, 200) as keep_mock:
 786             self.assertEqual(". {} 0:3:1 3:3:2\n".format(data_loc),
 787                              writer.manifest_text())
 788
 789     def test_two_opens_two_streams(self):
 790         client = self.api_client_mock()
 791         data_loc1 = hashlib.md5('file').hexdigest() + '+4'
 792         data_loc2 = hashlib.md5('indir').hexdigest() + '+5'
 793         with self.mock_keep((data_loc1, 200), (data_loc2, 200)) as keep_mock:
 794             writer = arvados.CollectionWriter(client)
 795             with writer.open('file') as out_file:
 796                 out_file.write('file')
 797             with writer.open('./dir', 'indir') as out_file:
 798                 out_file.write('indir')
 799             expected = ". {} 0:4:file\n./dir {} 0:5:indir\n".format(
 800                 data_loc1, data_loc2)
 801             self.assertEqual(expected, writer.manifest_text())
 802
 803     def test_dup_open_fails(self):
 804         client = self.api_client_mock()
 805         writer = arvados.CollectionWriter(client)
 806         file1 = writer.open('one')
 807         self.assertRaises(arvados.errors.AssertionError, writer.open, 'two')
 808
 809
 810 class NewCollectionTestCase(unittest.TestCase, CollectionTestMixin):
 811
 812     def test_init_manifest(self):
 813         m1 = """. 5348b82a029fd9e971a811ce1f71360b+43 0:43:md5sum.txt
 814 . 085c37f02916da1cad16f93c54d899b7+41 0:41:md5sum.txt
 815 . 8b22da26f9f433dea0a10e5ec66d73ba+43 0:43:md5sum.txt
 816 """
 817         self.assertEqual(m1, CollectionReader(m1).manifest_text(normalize=False))
 818         self.assertEqual(". 5348b82a029fd9e971a811ce1f71360b+43 085c37f02916da1cad16f93c54d899b7+41 8b22da26f9f433dea0a10e5ec66d73ba+43 0:127:md5sum.txt\n", CollectionReader(m1).manifest_text(normalize=True))
 819
 820
 821     def test_remove(self):
 822         c = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt 0:10:count2.txt\n')
 823         self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt 0:10:count2.txt\n", c.manifest_text())
 824         self.assertIn("count1.txt", c)
 825         c.remove("count1.txt")
 826         self.assertNotIn("count1.txt", c)
 827         self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count2.txt\n", c.manifest_text())
 828
 829     def test_remove_in_subdir(self):
 830         c = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n./foo 781e5e245d69b566979b86e28d23f2c7+10 0:10:count2.txt\n')
 831         c.remove("foo/count2.txt")
 832         self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n", c.manifest_text())
 833
 834     def test_remove_empty_subdir(self):
 835         c = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n./foo 781e5e245d69b566979b86e28d23f2c7+10 0:10:count2.txt\n')
 836         c.remove("foo/count2.txt")
 837         c.remove("foo")
 838         self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n", c.manifest_text())
 839
 840     def test_remove_nonempty_subdir(self):
 841         c = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n./foo 781e5e245d69b566979b86e28d23f2c7+10 0:10:count2.txt\n')
 842         with self.assertRaises(IOError):
 843             c.remove("foo")
 844         c.remove("foo", recursive=True)
 845         self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n", c.manifest_text())
 846
 847     def test_copy_to_file_in_dir(self):
 848         c = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n')
 849         c.copy("count1.txt", "foo/count2.txt")
 850         self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n./foo 781e5e245d69b566979b86e28d23f2c7+10 0:10:count2.txt\n", c.manifest_text())
 851
 852     def test_copy_file(self):
 853         c = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n')
 854         c.copy("count1.txt", "count2.txt")
 855         self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt 0:10:count2.txt\n", c.manifest_text())
 856
 857     def test_copy_to_existing_dir(self):
 858         c = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n./foo 781e5e245d69b566979b86e28d23f2c7+10 0:10:count2.txt\n')
 859         c.copy("count1.txt", "foo")
 860         self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n./foo 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt 0:10:count2.txt\n", c.manifest_text())
 861
 862     def test_copy_to_new_dir(self):
 863         c = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n')
 864         c.copy("count1.txt", "foo/")
 865         self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n./foo 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n", c.manifest_text())
 866
 867     def test_clone(self):
 868         c = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n./foo 781e5e245d69b566979b86e28d23f2c7+10 0:10:count2.txt\n')
 869         cl = c.clone()
 870         self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n./foo 781e5e245d69b566979b86e28d23f2c7+10 0:10:count2.txt\n", cl.manifest_text())
 871
 872     def test_diff_del_add(self):
 873         c1 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n')
 874         c2 = Collection('. 5348b82a029fd9e971a811ce1f71360b+43 0:10:count2.txt\n')
 875         d = c2.diff(c1)
 876         self.assertEqual(d, [('del', './count2.txt', c2["count2.txt"]),
 877                              ('add', './count1.txt', c1["count1.txt"])])
 878         d = c1.diff(c2)
 879         self.assertEqual(d, [('del', './count1.txt', c1["count1.txt"]),
 880                              ('add', './count2.txt', c2["count2.txt"])])
 881         self.assertNotEqual(c1.manifest_text(), c2.manifest_text())
 882         c1.apply(d)
 883         self.assertEqual(c1.manifest_text(), c2.manifest_text())
 884
 885     def test_diff_same(self):
 886         c1 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n')
 887         c2 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n')
 888         d = c2.diff(c1)
 889         self.assertEqual(d, [])
 890         d = c1.diff(c2)
 891         self.assertEqual(d, [])
 892
 893         self.assertEqual(c1.manifest_text(), c2.manifest_text())
 894         c1.apply(d)
 895         self.assertEqual(c1.manifest_text(), c2.manifest_text())
 896
 897     def test_diff_mod(self):
 898         c1 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n')
 899         c2 = Collection('. 5348b82a029fd9e971a811ce1f71360b+43 0:10:count1.txt\n')
 900         d = c2.diff(c1)
 901         self.assertEqual(d, [('mod', './count1.txt', c2["count1.txt"], c1["count1.txt"])])
 902         d = c1.diff(c2)
 903         self.assertEqual(d, [('mod', './count1.txt', c1["count1.txt"], c2["count1.txt"])])
 904
 905         self.assertNotEqual(c1.manifest_text(), c2.manifest_text())
 906         c1.apply(d)
 907         self.assertEqual(c1.manifest_text(), c2.manifest_text())
 908
 909     def test_diff_add(self):
 910         c1 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n')
 911         c2 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 5348b82a029fd9e971a811ce1f71360b+43 0:10:count1.txt 10:20:count2.txt\n')
 912         d = c2.diff(c1)
 913         self.assertEqual(d, [('del', './count2.txt', c2["count2.txt"])])
 914         d = c1.diff(c2)
 915         self.assertEqual(d, [('add', './count2.txt', c2["count2.txt"])])
 916
 917         self.assertNotEqual(c1.manifest_text(), c2.manifest_text())
 918         c1.apply(d)
 919         self.assertEqual(c1.manifest_text(), c2.manifest_text())
 920
 921     def test_diff_add_in_subcollection(self):
 922         c1 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n')
 923         c2 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n./foo 5348b82a029fd9e971a811ce1f71360b+43 0:10:count2.txt\n')
 924         d = c2.diff(c1)
 925         self.assertEqual(d, [('del', './foo', c2["foo"])])
 926         d = c1.diff(c2)
 927         self.assertEqual(d, [('add', './foo', c2["foo"])])
 928
 929         self.assertNotEqual(c1.manifest_text(), c2.manifest_text())
 930         c1.apply(d)
 931         self.assertEqual(c1.manifest_text(), c2.manifest_text())
 932
 933     def test_diff_del_add_in_subcollection(self):
 934         c1 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n./foo 5348b82a029fd9e971a811ce1f71360b+43 0:10:count2.txt\n')
 935         c2 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n./foo 5348b82a029fd9e971a811ce1f71360b+43 0:3:count3.txt\n')
 936
 937         d = c2.diff(c1)
 938         self.assertEqual(d, [('del', './foo/count3.txt', c2.find("foo/count3.txt")),
 939                              ('add', './foo/count2.txt', c1.find("foo/count2.txt"))])
 940         d = c1.diff(c2)
 941         self.assertEqual(d, [('del', './foo/count2.txt', c1.find("foo/count2.txt")),
 942                              ('add', './foo/count3.txt', c2.find("foo/count3.txt"))])
 943
 944         self.assertNotEqual(c1.manifest_text(), c2.manifest_text())
 945         c1.apply(d)
 946         self.assertEqual(c1.manifest_text(), c2.manifest_text())
 947
 948     def test_diff_mod_in_subcollection(self):
 949         c1 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n./foo 5348b82a029fd9e971a811ce1f71360b+43 0:10:count2.txt\n')
 950         c2 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt 0:3:foo\n')
 951         d = c2.diff(c1)
 952         self.assertEqual(d, [('mod', './foo', c2["foo"], c1["foo"])])
 953         d = c1.diff(c2)
 954         self.assertEqual(d, [('mod', './foo', c1["foo"], c2["foo"])])
 955
 956         self.assertNotEqual(c1.manifest_text(), c2.manifest_text())
 957         c1.apply(d)
 958         self.assertEqual(c1.manifest_text(), c2.manifest_text())
 959
 960     def test_conflict_keep_local_change(self):
 961         c1 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n')
 962         c2 = Collection('. 5348b82a029fd9e971a811ce1f71360b+43 0:10:count2.txt\n')
 963         d = c1.diff(c2)
 964         self.assertEqual(d, [('del', './count1.txt', c1["count1.txt"]),
 965                              ('add', './count2.txt', c2["count2.txt"])])
 966         with c1.open("count1.txt", "w") as f:
 967             f.write("zzzzz")
 968
 969         # c1 changed, so it should not be deleted.
 970         c1.apply(d)
 971         self.assertEqual(c1.manifest_text(), ". 95ebc3c7b3b9f1d2c40fec14415d3cb8+5 5348b82a029fd9e971a811ce1f71360b+43 0:5:count1.txt 5:10:count2.txt\n")
 972
 973     def test_conflict_mod(self):
 974         c1 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt')
 975         c2 = Collection('. 5348b82a029fd9e971a811ce1f71360b+43 0:10:count1.txt')
 976         d = c1.diff(c2)
 977         self.assertEqual(d, [('mod', './count1.txt', c1["count1.txt"], c2["count1.txt"])])
 978         with c1.open("count1.txt", "w") as f:
 979             f.write("zzzzz")
 980
 981         # c1 changed, so c2 mod will go to a conflict file
 982         c1.apply(d)
 983         self.assertTrue(re.match(r"\. 95ebc3c7b3b9f1d2c40fec14415d3cb8\+5 5348b82a029fd9e971a811ce1f71360b\+43 0:5:count1.txt 5:10:count1.txt~conflict-\d\d\d\d-\d\d-\d\d-\d\d:\d\d:\d\d~$",
 984                                  c1.manifest_text()))
 985
 986     def test_conflict_add(self):
 987         c1 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count2.txt\n')
 988         c2 = Collection('. 5348b82a029fd9e971a811ce1f71360b+43 0:10:count1.txt\n')
 989         d = c1.diff(c2)
 990         self.assertEqual(d, [('del', './count2.txt', c1["count2.txt"]),
 991                              ('add', './count1.txt', c2["count1.txt"])])
 992         with c1.open("count1.txt", "w") as f:
 993             f.write("zzzzz")
 994
 995         # c1 added count1.txt, so c2 add will go to a conflict file
 996         c1.apply(d)
 997         self.assertTrue(re.match(r"\. 95ebc3c7b3b9f1d2c40fec14415d3cb8\+5 5348b82a029fd9e971a811ce1f71360b\+43 0:5:count1.txt 5:10:count1.txt~conflict-\d\d\d\d-\d\d-\d\d-\d\d:\d\d:\d\d~$",
 998                                  c1.manifest_text()))
 999
1000     def test_conflict_del(self):
1001         c1 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt')
1002         c2 = Collection('. 5348b82a029fd9e971a811ce1f71360b+43 0:10:count1.txt')
1003         d = c1.diff(c2)
1004         self.assertEqual(d, [('mod', './count1.txt', c1["count1.txt"], c2["count1.txt"])])
1005         c1.remove("count1.txt")
1006
1007         # c1 deleted, so c2 mod will go to a conflict file
1008         c1.apply(d)
1009         self.assertTrue(re.match(r"\. 5348b82a029fd9e971a811ce1f71360b\+43 0:10:count1.txt~conflict-\d\d\d\d-\d\d-\d\d-\d\d:\d\d:\d\d~$",
1010                                  c1.manifest_text()))
1011
1012     def test_notify(self):
1013         c1 = Collection()
1014         events = []
1015         c1.subscribe(lambda event, collection, name, item: events.append((event, collection, name, item)))
1016         f = c1.open("foo.txt", "w")
1017         self.assertEqual(events[0], (arvados.collection.ADD, c1, "foo.txt", f.arvadosfile))
1018
1019     def test_open_w(self):
1020         c1 = Collection(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n")
1021         self.assertEqual(c1["count1.txt"].size(), 10)
1022         c1.open("count1.txt", "w").close()
1023         self.assertEqual(c1["count1.txt"].size(), 0)
1024
1025
1026 class CollectionCreateUpdateTest(run_test_server.TestCaseWithServers):
1027     MAIN_SERVER = {}
1028     KEEP_SERVER = {}
1029
1030     def create_count_txt(self):
1031         # Create an empty collection, save it to the API server, then write a
1032         # file, but don't save it.
1033
1034         c = Collection()
1035         c.save_new("CollectionCreateUpdateTest", ensure_unique_name=True)
1036         self.assertEquals(c.portable_data_hash(), "d41d8cd98f00b204e9800998ecf8427e+0")
1037         self.assertEquals(c.api_response()["portable_data_hash"], "d41d8cd98f00b204e9800998ecf8427e+0" )
1038
1039         with c.open("count.txt", "w") as f:
1040             f.write("0123456789")
1041
1042         self.assertEquals(c.manifest_text(), ". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n")
1043
1044         return c
1045
1046     def test_create_and_save(self):
1047         c = self.create_count_txt()
1048         c.save()
1049         self.assertTrue(re.match(r"^\. 781e5e245d69b566979b86e28d23f2c7\+10\+A[a-f0-9]{40}@[a-f0-9]{8} 0:10:count.txt$",
1050                                  c.manifest_text()))
1051
1052
1053     def test_create_and_save_new(self):
1054         c = self.create_count_txt()
1055         c.save_new()
1056         self.assertTrue(re.match(r"^\. 781e5e245d69b566979b86e28d23f2c7\+10\+A[a-f0-9]{40}@[a-f0-9]{8} 0:10:count.txt$",
1057                                  c.manifest_text()))
1058
1059     def test_create_diff_apply(self):
1060         c1 = self.create_count_txt()
1061         c1.save()
1062
1063         c2 = Collection(c1._manifest_locator)
1064         with c2.open("count.txt", "w") as f:
1065             f.write("abcdefg")
1066
1067         diff = c1.diff(c2)
1068
1069         self.assertEqual(diff[0], (arvados.collection.MOD, u'./count.txt', c1["count.txt"], c2["count.txt"]))
1070
1071         c1.apply(diff)
1072         self.assertEqual(c1.portable_data_hash(), c2.portable_data_hash())
1073
1074     def test_diff_apply_with_token(self):
1075         baseline = CollectionReader(". 781e5e245d69b566979b86e28d23f2c7+10+A715fd31f8111894f717eb1003c1b0216799dd9ec@54f5dd1a 0:10:count.txt\n")
1076         c = Collection(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n")
1077         other = CollectionReader(". 7ac66c0f148de9519b8bd264312c4d64+7+A715fd31f8111894f717eb1003c1b0216799dd9ec@54f5dd1a 0:7:count.txt\n")
1078
1079         diff = baseline.diff(other)
1080         self.assertEqual(diff, [('mod', u'./count.txt', c["count.txt"], other["count.txt"])])
1081
1082         c.apply(diff)
1083
1084         self.assertEqual(c.manifest_text(), ". 7ac66c0f148de9519b8bd264312c4d64+7+A715fd31f8111894f717eb1003c1b0216799dd9ec@54f5dd1a 0:7:count.txt\n")
1085
1086
1087     def test_create_and_update(self):
1088         c1 = self.create_count_txt()
1089         c1.save()
1090
1091         c2 = arvados.collection.Collection(c1._manifest_locator)
1092         with c2.open("count.txt", "w") as f:
1093             f.write("abcdefg")
1094
1095         c2.save()
1096
1097         self.assertNotEqual(c1.portable_data_hash(), c2.portable_data_hash())
1098         c1.update()
1099         self.assertEqual(c1.portable_data_hash(), c2.portable_data_hash())
1100
1101
1102     def test_create_and_update_with_conflict(self):
1103         c1 = self.create_count_txt()
1104         c1.save()
1105
1106         with c1.open("count.txt", "w") as f:
1107             f.write("XYZ")
1108
1109         c2 = arvados.collection.Collection(c1._manifest_locator)
1110         with c2.open("count.txt", "w") as f:
1111             f.write("abcdefg")
1112
1113         c2.save()
1114
1115         c1.update()
1116         self.assertTrue(re.match(r"\. e65075d550f9b5bf9992fa1d71a131be\+3 7ac66c0f148de9519b8bd264312c4d64\+7\+A[a-f0-9]{40}@[a-f0-9]{8} 0:3:count.txt 3:7:count.txt~conflict-\d\d\d\d-\d\d-\d\d-\d\d:\d\d:\d\d~$", c1.manifest_text()))
1117
1118
1119 if __name__ == '__main__':
1120     unittest.main()