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 import arvados_testutil as tutil
  17 from arvados.ranges import Range, LocatorAndRange
  18 from arvados import import_manifest, export_manifest
  19 from arvados.arvfile import SYNC_EXPLICIT
  20 from arvados.collection import Collection
  21
  22 class TestResumableWriter(arvados.ResumableCollectionWriter):
  23     KEEP_BLOCK_SIZE = 1024  # PUT to Keep every 1K.
  24
  25     def current_state(self):
  26         return self.dump_state(copy.deepcopy)
  27
  28
  29 class ArvadosCollectionsTest(run_test_server.TestCaseWithServers,
  30                              tutil.ArvadosBaseTestCase):
  31     MAIN_SERVER = {}
  32
  33     @classmethod
  34     def setUpClass(cls):
  35         super(ArvadosCollectionsTest, cls).setUpClass()
  36         run_test_server.authorize_with('active')
  37         cls.api_client = arvados.api('v1')
  38         cls.keep_client = arvados.KeepClient(api_client=cls.api_client,
  39                                              local_store=cls.local_store)
  40
  41     def write_foo_bar_baz(self):
  42         cw = arvados.CollectionWriter(self.api_client)
  43         self.assertEqual(cw.current_stream_name(), '.',
  44                          'current_stream_name() should be "." now')
  45         cw.set_current_file_name('foo.txt')
  46         cw.write('foo')
  47         self.assertEqual(cw.current_file_name(), 'foo.txt',
  48                          'current_file_name() should be foo.txt now')
  49         cw.start_new_file('bar.txt')
  50         cw.write('bar')
  51         cw.start_new_stream('baz')
  52         cw.write('baz')
  53         cw.set_current_file_name('baz.txt')
  54         self.assertEqual(cw.manifest_text(),
  55                          ". 3858f62230ac3c915f300c664312c63f+6 0:3:foo.txt 3:3:bar.txt\n" +
  56                          "./baz 73feffa4b7f6bb68e44cf984c85f6e88+3 0:3:baz.txt\n",
  57                          "wrong manifest: got {}".format(cw.manifest_text()))
  58         cw.finish()
  59         return cw.portable_data_hash()
  60
  61     def test_keep_local_store(self):
  62         self.assertEqual(self.keep_client.put('foo'), 'acbd18db4cc2f85cedef654fccc4a4d8+3', 'wrong md5 hash from Keep.put')
  63         self.assertEqual(self.keep_client.get('acbd18db4cc2f85cedef654fccc4a4d8+3'), 'foo', 'wrong data from Keep.get')
  64
  65     def test_local_collection_writer(self):
  66         self.assertEqual(self.write_foo_bar_baz(),
  67                          '23ca013983d6239e98931cc779e68426+114',
  68                          'wrong locator hash: ' + self.write_foo_bar_baz())
  69
  70     def test_local_collection_reader(self):
  71         foobarbaz = self.write_foo_bar_baz()
  72         cr = arvados.CollectionReader(
  73             foobarbaz + '+Xzizzle', self.api_client)
  74         got = []
  75         for s in cr.all_streams():
  76             for f in s.all_files():
  77                 got += [[f.size(), f.stream_name(), f.name(), f.read(2**26)]]
  78         expected = [[3, '.', 'foo.txt', 'foo'],
  79                     [3, '.', 'bar.txt', 'bar'],
  80                     [3, './baz', 'baz.txt', 'baz']]
  81         self.assertEqual(got,
  82                          expected)
  83         stream0 = cr.all_streams()[0]
  84         self.assertEqual(stream0.readfrom(0, 0),
  85                          '',
  86                          'reading zero bytes should have returned empty string')
  87         self.assertEqual(stream0.readfrom(0, 2**26),
  88                          'foobar',
  89                          'reading entire stream failed')
  90         self.assertEqual(stream0.readfrom(2**26, 0),
  91                          '',
  92                          'reading zero bytes should have returned empty string')
  93
  94     def _test_subset(self, collection, expected):
  95         cr = arvados.CollectionReader(collection, self.api_client)
  96         for s in cr.all_streams():
  97             for ex in expected:
  98                 if ex[0] == s:
  99                     f = s.files()[ex[2]]
 100                     got = [f.size(), f.stream_name(), f.name(), "".join(f.readall(2**26))]
 101                     self.assertEqual(got,
 102                                      ex,
 103                                      'all_files|as_manifest did not preserve manifest contents: got %s expected %s' % (got, ex))
 104
 105     def test_collection_manifest_subset(self):
 106         foobarbaz = self.write_foo_bar_baz()
 107         self._test_subset(foobarbaz,
 108                           [[3, '.',     'bar.txt', 'bar'],
 109                            [3, '.',     'foo.txt', 'foo'],
 110                            [3, './baz', 'baz.txt', 'baz']])
 111         self._test_subset((". %s %s 0:3:foo.txt 3:3:bar.txt\n" %
 112                            (self.keep_client.put("foo"),
 113                             self.keep_client.put("bar"))),
 114                           [[3, '.', 'bar.txt', 'bar'],
 115                            [3, '.', 'foo.txt', 'foo']])
 116         self._test_subset((". %s %s 0:2:fo.txt 2:4:obar.txt\n" %
 117                            (self.keep_client.put("foo"),
 118                             self.keep_client.put("bar"))),
 119                           [[2, '.', 'fo.txt', 'fo'],
 120                            [4, '.', 'obar.txt', 'obar']])
 121         self._test_subset((". %s %s 0:2:fo.txt 2:0:zero.txt 2:2:ob.txt 4:2:ar.txt\n" %
 122                            (self.keep_client.put("foo"),
 123                             self.keep_client.put("bar"))),
 124                           [[2, '.', 'ar.txt', 'ar'],
 125                            [2, '.', 'fo.txt', 'fo'],
 126                            [2, '.', 'ob.txt', 'ob'],
 127                            [0, '.', 'zero.txt', '']])
 128
 129     def test_collection_empty_file(self):
 130         cw = arvados.CollectionWriter(self.api_client)
 131         cw.start_new_file('zero.txt')
 132         cw.write('')
 133
 134         self.assertEqual(cw.manifest_text(), ". d41d8cd98f00b204e9800998ecf8427e+0 0:0:zero.txt\n")
 135         self.check_manifest_file_sizes(cw.manifest_text(), [0])
 136         cw = arvados.CollectionWriter(self.api_client)
 137         cw.start_new_file('zero.txt')
 138         cw.write('')
 139         cw.start_new_file('one.txt')
 140         cw.write('1')
 141         cw.start_new_stream('foo')
 142         cw.start_new_file('zero.txt')
 143         cw.write('')
 144         self.check_manifest_file_sizes(cw.manifest_text(), [0,1,0])
 145
 146     def test_no_implicit_normalize(self):
 147         cw = arvados.CollectionWriter(self.api_client)
 148         cw.start_new_file('b')
 149         cw.write('b')
 150         cw.start_new_file('a')
 151         cw.write('')
 152         self.check_manifest_file_sizes(cw.manifest_text(), [1,0])
 153         self.check_manifest_file_sizes(
 154             arvados.CollectionReader(
 155                 cw.manifest_text()).manifest_text(normalize=True),
 156             [0,1])
 157
 158     def check_manifest_file_sizes(self, manifest_text, expect_sizes):
 159         cr = arvados.CollectionReader(manifest_text, self.api_client)
 160         got_sizes = []
 161         for f in cr.all_files():
 162             got_sizes += [f.size()]
 163         self.assertEqual(got_sizes, expect_sizes, "got wrong file sizes %s, expected %s" % (got_sizes, expect_sizes))
 164
 165     def test_normalized_collection(self):
 166         m1 = """. 5348b82a029fd9e971a811ce1f71360b+43 0:43:md5sum.txt
 167 . 085c37f02916da1cad16f93c54d899b7+41 0:41:md5sum.txt
 168 . 8b22da26f9f433dea0a10e5ec66d73ba+43 0:43:md5sum.txt
 169 """
 170         self.assertEqual(arvados.CollectionReader(m1, self.api_client).manifest_text(normalize=True),
 171                          """. 5348b82a029fd9e971a811ce1f71360b+43 085c37f02916da1cad16f93c54d899b7+41 8b22da26f9f433dea0a10e5ec66d73ba+43 0:127:md5sum.txt
 172 """)
 173
 174         m2 = """. 204e43b8a1185621ca55a94839582e6f+67108864 b9677abbac956bd3e86b1deb28dfac03+67108864 fc15aff2a762b13f521baf042140acec+67108864 323d2a3ce20370c4ca1d3462a344f8fd+25885655 0:227212247:var-GS000016015-ASM.tsv.bz2
 175 """
 176         self.assertEqual(arvados.CollectionReader(m2, self.api_client).manifest_text(normalize=True), m2)
 177
 178         m3 = """. 5348b82a029fd9e971a811ce1f71360b+43 3:40:md5sum.txt
 179 . 085c37f02916da1cad16f93c54d899b7+41 0:41:md5sum.txt
 180 . 8b22da26f9f433dea0a10e5ec66d73ba+43 0:43:md5sum.txt
 181 """
 182         self.assertEqual(arvados.CollectionReader(m3, self.api_client).manifest_text(normalize=True),
 183                          """. 5348b82a029fd9e971a811ce1f71360b+43 085c37f02916da1cad16f93c54d899b7+41 8b22da26f9f433dea0a10e5ec66d73ba+43 3:124:md5sum.txt
 184 """)
 185
 186         m4 = """. 204e43b8a1185621ca55a94839582e6f+67108864 0:3:foo/bar
 187 ./zzz 204e43b8a1185621ca55a94839582e6f+67108864 0:999:zzz
 188 ./foo 323d2a3ce20370c4ca1d3462a344f8fd+25885655 0:3:bar
 189 """
 190         self.assertEqual(arvados.CollectionReader(m4, self.api_client).manifest_text(normalize=True),
 191                          """./foo 204e43b8a1185621ca55a94839582e6f+67108864 323d2a3ce20370c4ca1d3462a344f8fd+25885655 0:3:bar 67108864:3:bar
 192 ./zzz 204e43b8a1185621ca55a94839582e6f+67108864 0:999:zzz
 193 """)
 194
 195         m5 = """. 204e43b8a1185621ca55a94839582e6f+67108864 0:3:foo/bar
 196 ./zzz 204e43b8a1185621ca55a94839582e6f+67108864 0:999:zzz
 197 ./foo 204e43b8a1185621ca55a94839582e6f+67108864 3:3:bar
 198 """
 199         self.assertEqual(arvados.CollectionReader(m5, self.api_client).manifest_text(normalize=True),
 200                          """./foo 204e43b8a1185621ca55a94839582e6f+67108864 0:6:bar
 201 ./zzz 204e43b8a1185621ca55a94839582e6f+67108864 0:999:zzz
 202 """)
 203
 204         with self.data_file('1000G_ref_manifest') as f6:
 205             m6 = f6.read()
 206             self.assertEqual(arvados.CollectionReader(m6, self.api_client).manifest_text(normalize=True), m6)
 207
 208         with self.data_file('jlake_manifest') as f7:
 209             m7 = f7.read()
 210             self.assertEqual(arvados.CollectionReader(m7, self.api_client).manifest_text(normalize=True), m7)
 211
 212         m8 = """./a\\040b\\040c 59ca0efa9f5633cb0371bbc0355478d8+13 0:13:hello\\040world.txt
 213 """
 214         self.assertEqual(arvados.CollectionReader(m8, self.api_client).manifest_text(normalize=True), m8)
 215
 216     def test_locators_and_ranges(self):
 217         blocks2 = [Range('a', 0, 10),
 218                    Range('b', 10, 10),
 219                    Range('c', 20, 10),
 220                    Range('d', 30, 10),
 221                    Range('e', 40, 10),
 222                    Range('f', 50, 10)]
 223
 224         self.assertEqual(arvados.locators_and_ranges(blocks2,  2,  2), [LocatorAndRange('a', 10, 2, 2)])
 225         self.assertEqual(arvados.locators_and_ranges(blocks2, 12, 2), [LocatorAndRange('b', 10, 2, 2)])
 226         self.assertEqual(arvados.locators_and_ranges(blocks2, 22, 2), [LocatorAndRange('c', 10, 2, 2)])
 227         self.assertEqual(arvados.locators_and_ranges(blocks2, 32, 2), [LocatorAndRange('d', 10, 2, 2)])
 228         self.assertEqual(arvados.locators_and_ranges(blocks2, 42, 2), [LocatorAndRange('e', 10, 2, 2)])
 229         self.assertEqual(arvados.locators_and_ranges(blocks2, 52, 2), [LocatorAndRange('f', 10, 2, 2)])
 230         self.assertEqual(arvados.locators_and_ranges(blocks2, 62, 2), [])
 231         self.assertEqual(arvados.locators_and_ranges(blocks2, -2, 2), [])
 232
 233         self.assertEqual(arvados.locators_and_ranges(blocks2,  0,  2), [LocatorAndRange('a', 10, 0, 2)])
 234         self.assertEqual(arvados.locators_and_ranges(blocks2, 10, 2), [LocatorAndRange('b', 10, 0, 2)])
 235         self.assertEqual(arvados.locators_and_ranges(blocks2, 20, 2), [LocatorAndRange('c', 10, 0, 2)])
 236         self.assertEqual(arvados.locators_and_ranges(blocks2, 30, 2), [LocatorAndRange('d', 10, 0, 2)])
 237         self.assertEqual(arvados.locators_and_ranges(blocks2, 40, 2), [LocatorAndRange('e', 10, 0, 2)])
 238         self.assertEqual(arvados.locators_and_ranges(blocks2, 50, 2), [LocatorAndRange('f', 10, 0, 2)])
 239         self.assertEqual(arvados.locators_and_ranges(blocks2, 60, 2), [])
 240         self.assertEqual(arvados.locators_and_ranges(blocks2, -2, 2), [])
 241
 242         self.assertEqual(arvados.locators_and_ranges(blocks2,  9,  2), [LocatorAndRange('a', 10, 9, 1), LocatorAndRange('b', 10, 0, 1)])
 243         self.assertEqual(arvados.locators_and_ranges(blocks2, 19, 2), [LocatorAndRange('b', 10, 9, 1), LocatorAndRange('c', 10, 0, 1)])
 244         self.assertEqual(arvados.locators_and_ranges(blocks2, 29, 2), [LocatorAndRange('c', 10, 9, 1), LocatorAndRange('d', 10, 0, 1)])
 245         self.assertEqual(arvados.locators_and_ranges(blocks2, 39, 2), [LocatorAndRange('d', 10, 9, 1), LocatorAndRange('e', 10, 0, 1)])
 246         self.assertEqual(arvados.locators_and_ranges(blocks2, 49, 2), [LocatorAndRange('e', 10, 9, 1), LocatorAndRange('f', 10, 0, 1)])
 247         self.assertEqual(arvados.locators_and_ranges(blocks2, 59, 2), [LocatorAndRange('f', 10, 9, 1)])
 248
 249
 250         blocks3 = [Range('a', 0, 10),
 251                   Range('b', 10, 10),
 252                   Range('c', 20, 10),
 253                   Range('d', 30, 10),
 254                   Range('e', 40, 10),
 255                   Range('f', 50, 10),
 256                    Range('g', 60, 10)]
 257
 258         self.assertEqual(arvados.locators_and_ranges(blocks3,  2,  2), [LocatorAndRange('a', 10, 2, 2)])
 259         self.assertEqual(arvados.locators_and_ranges(blocks3, 12, 2), [LocatorAndRange('b', 10, 2, 2)])
 260         self.assertEqual(arvados.locators_and_ranges(blocks3, 22, 2), [LocatorAndRange('c', 10, 2, 2)])
 261         self.assertEqual(arvados.locators_and_ranges(blocks3, 32, 2), [LocatorAndRange('d', 10, 2, 2)])
 262         self.assertEqual(arvados.locators_and_ranges(blocks3, 42, 2), [LocatorAndRange('e', 10, 2, 2)])
 263         self.assertEqual(arvados.locators_and_ranges(blocks3, 52, 2), [LocatorAndRange('f', 10, 2, 2)])
 264         self.assertEqual(arvados.locators_and_ranges(blocks3, 62, 2), [LocatorAndRange('g', 10, 2, 2)])
 265
 266
 267         blocks = [Range('a', 0, 10),
 268                   Range('b', 10, 15),
 269                   Range('c', 25, 5)]
 270         self.assertEqual(arvados.locators_and_ranges(blocks, 1, 0), [])
 271         self.assertEqual(arvados.locators_and_ranges(blocks, 0, 5), [LocatorAndRange('a', 10, 0, 5)])
 272         self.assertEqual(arvados.locators_and_ranges(blocks, 3, 5), [LocatorAndRange('a', 10, 3, 5)])
 273         self.assertEqual(arvados.locators_and_ranges(blocks, 0, 10), [LocatorAndRange('a', 10, 0, 10)])
 274
 275         self.assertEqual(arvados.locators_and_ranges(blocks, 0, 11), [LocatorAndRange('a', 10, 0, 10),
 276                                                                       LocatorAndRange('b', 15, 0, 1)])
 277         self.assertEqual(arvados.locators_and_ranges(blocks, 1, 11), [LocatorAndRange('a', 10, 1, 9),
 278                                                                       LocatorAndRange('b', 15, 0, 2)])
 279         self.assertEqual(arvados.locators_and_ranges(blocks, 0, 25), [LocatorAndRange('a', 10, 0, 10),
 280                                                                       LocatorAndRange('b', 15, 0, 15)])
 281
 282         self.assertEqual(arvados.locators_and_ranges(blocks, 0, 30), [LocatorAndRange('a', 10, 0, 10),
 283                                                                       LocatorAndRange('b', 15, 0, 15),
 284                                                                       LocatorAndRange('c', 5, 0, 5)])
 285         self.assertEqual(arvados.locators_and_ranges(blocks, 1, 30), [LocatorAndRange('a', 10, 1, 9),
 286                                                                       LocatorAndRange('b', 15, 0, 15),
 287                                                                       LocatorAndRange('c', 5, 0, 5)])
 288         self.assertEqual(arvados.locators_and_ranges(blocks, 0, 31), [LocatorAndRange('a', 10, 0, 10),
 289                                                                       LocatorAndRange('b', 15, 0, 15),
 290                                                                       LocatorAndRange('c', 5, 0, 5)])
 291
 292         self.assertEqual(arvados.locators_and_ranges(blocks, 15, 5), [LocatorAndRange('b', 15, 5, 5)])
 293
 294         self.assertEqual(arvados.locators_and_ranges(blocks, 8, 17), [LocatorAndRange('a', 10, 8, 2),
 295                                                                       LocatorAndRange('b', 15, 0, 15)])
 296
 297         self.assertEqual(arvados.locators_and_ranges(blocks, 8, 20), [LocatorAndRange('a', 10, 8, 2),
 298                                                                       LocatorAndRange('b', 15, 0, 15),
 299                                                                       LocatorAndRange('c', 5, 0, 3)])
 300
 301         self.assertEqual(arvados.locators_and_ranges(blocks, 26, 2), [LocatorAndRange('c', 5, 1, 2)])
 302
 303         self.assertEqual(arvados.locators_and_ranges(blocks, 9, 15), [LocatorAndRange('a', 10, 9, 1),
 304                                                                       LocatorAndRange('b', 15, 0, 14)])
 305         self.assertEqual(arvados.locators_and_ranges(blocks, 10, 15), [LocatorAndRange('b', 15, 0, 15)])
 306         self.assertEqual(arvados.locators_and_ranges(blocks, 11, 15), [LocatorAndRange('b', 15, 1, 14),
 307                                                                        LocatorAndRange('c', 5, 0, 1)])
 308
 309     class MockKeep(object):
 310         def __init__(self, content, num_retries=0):
 311             self.content = content
 312
 313         def get(self, locator, num_retries=0):
 314             return self.content[locator]
 315
 316     def test_stream_reader(self):
 317         keepblocks = {'aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa+10': 'abcdefghij',
 318                       'bbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbb+15': 'klmnopqrstuvwxy',
 319                       'cccccccccccccccccccccccccccccccc+5': 'z0123'}
 320         mk = self.MockKeep(keepblocks)
 321
 322         sr = arvados.StreamReader([".", "aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa+10", "bbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbb+15", "cccccccccccccccccccccccccccccccc+5", "0:30:foo"], mk)
 323
 324         content = 'abcdefghijklmnopqrstuvwxyz0123456789'
 325
 326         self.assertEqual(sr.readfrom(0, 30), content[0:30])
 327         self.assertEqual(sr.readfrom(2, 30), content[2:30])
 328
 329         self.assertEqual(sr.readfrom(2, 8), content[2:10])
 330         self.assertEqual(sr.readfrom(0, 10), content[0:10])
 331
 332         self.assertEqual(sr.readfrom(0, 5), content[0:5])
 333         self.assertEqual(sr.readfrom(5, 5), content[5:10])
 334         self.assertEqual(sr.readfrom(10, 5), content[10:15])
 335         self.assertEqual(sr.readfrom(15, 5), content[15:20])
 336         self.assertEqual(sr.readfrom(20, 5), content[20:25])
 337         self.assertEqual(sr.readfrom(25, 5), content[25:30])
 338         self.assertEqual(sr.readfrom(30, 5), '')
 339
 340     def test_extract_file(self):
 341         m1 = """. 5348b82a029fd9e971a811ce1f71360b+43 0:43:md5sum.txt
 342 . 085c37f02916da1cad16f93c54d899b7+41 0:41:md6sum.txt
 343 . 8b22da26f9f433dea0a10e5ec66d73ba+43 0:43:md7sum.txt
 344 . 085c37f02916da1cad16f93c54d899b7+41 5348b82a029fd9e971a811ce1f71360b+43 8b22da26f9f433dea0a10e5ec66d73ba+43 47:80:md8sum.txt
 345 . 085c37f02916da1cad16f93c54d899b7+41 5348b82a029fd9e971a811ce1f71360b+43 8b22da26f9f433dea0a10e5ec66d73ba+43 40:80:md9sum.txt
 346 """
 347
 348         m2 = arvados.CollectionReader(m1, self.api_client).manifest_text(normalize=True)
 349
 350         self.assertEqual(m2,
 351                          ". 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")
 352         files = arvados.CollectionReader(
 353             m2, self.api_client).all_streams()[0].files()
 354
 355         self.assertEqual(files['md5sum.txt'].as_manifest(),
 356                          ". 5348b82a029fd9e971a811ce1f71360b+43 0:43:md5sum.txt\n")
 357         self.assertEqual(files['md6sum.txt'].as_manifest(),
 358                          ". 085c37f02916da1cad16f93c54d899b7+41 0:41:md6sum.txt\n")
 359         self.assertEqual(files['md7sum.txt'].as_manifest(),
 360                          ". 8b22da26f9f433dea0a10e5ec66d73ba+43 0:43:md7sum.txt\n")
 361         self.assertEqual(files['md9sum.txt'].as_manifest(),
 362                          ". 085c37f02916da1cad16f93c54d899b7+41 5348b82a029fd9e971a811ce1f71360b+43 8b22da26f9f433dea0a10e5ec66d73ba+43 40:80:md9sum.txt\n")
 363
 364     def test_write_directory_tree(self):
 365         cwriter = arvados.CollectionWriter(self.api_client)
 366         cwriter.write_directory_tree(self.build_directory_tree(
 367                 ['basefile', 'subdir/subfile']))
 368         self.assertEqual(cwriter.manifest_text(),
 369                          """. c5110c5ac93202d8e0f9e381f22bac0f+8 0:8:basefile
 370 ./subdir 1ca4dec89403084bf282ad31e6cf7972+14 0:14:subfile\n""")
 371
 372     def test_write_named_directory_tree(self):
 373         cwriter = arvados.CollectionWriter(self.api_client)
 374         cwriter.write_directory_tree(self.build_directory_tree(
 375                 ['basefile', 'subdir/subfile']), 'root')
 376         self.assertEqual(
 377             cwriter.manifest_text(),
 378             """./root c5110c5ac93202d8e0f9e381f22bac0f+8 0:8:basefile
 379 ./root/subdir 1ca4dec89403084bf282ad31e6cf7972+14 0:14:subfile\n""")
 380
 381     def test_write_directory_tree_in_one_stream(self):
 382         cwriter = arvados.CollectionWriter(self.api_client)
 383         cwriter.write_directory_tree(self.build_directory_tree(
 384                 ['basefile', 'subdir/subfile']), max_manifest_depth=0)
 385         self.assertEqual(cwriter.manifest_text(),
 386                          """. 4ace875ffdc6824a04950f06858f4465+22 0:8:basefile 8:14:subdir/subfile\n""")
 387
 388     def test_write_directory_tree_with_limited_recursion(self):
 389         cwriter = arvados.CollectionWriter(self.api_client)
 390         cwriter.write_directory_tree(
 391             self.build_directory_tree(['f1', 'd1/f2', 'd1/d2/f3']),
 392             max_manifest_depth=1)
 393         self.assertEqual(cwriter.manifest_text(),
 394                          """. bd19836ddb62c11c55ab251ccaca5645+2 0:2:f1
 395 ./d1 50170217e5b04312024aa5cd42934494+13 0:8:d2/f3 8:5:f2\n""")
 396
 397     def test_write_directory_tree_with_zero_recursion(self):
 398         cwriter = arvados.CollectionWriter(self.api_client)
 399         content = 'd1/d2/f3d1/f2f1'
 400         blockhash = hashlib.md5(content).hexdigest() + '+' + str(len(content))
 401         cwriter.write_directory_tree(
 402             self.build_directory_tree(['f1', 'd1/f2', 'd1/d2/f3']),
 403             max_manifest_depth=0)
 404         self.assertEqual(
 405             cwriter.manifest_text(),
 406             ". {} 0:8:d1/d2/f3 8:5:d1/f2 13:2:f1\n".format(blockhash))
 407
 408     def test_write_one_file(self):
 409         cwriter = arvados.CollectionWriter(self.api_client)
 410         with self.make_test_file() as testfile:
 411             cwriter.write_file(testfile.name)
 412             self.assertEqual(
 413                 cwriter.manifest_text(),
 414                 ". 098f6bcd4621d373cade4e832627b4f6+4 0:4:{}\n".format(
 415                     os.path.basename(testfile.name)))
 416
 417     def test_write_named_file(self):
 418         cwriter = arvados.CollectionWriter(self.api_client)
 419         with self.make_test_file() as testfile:
 420             cwriter.write_file(testfile.name, 'foo')
 421             self.assertEqual(cwriter.manifest_text(),
 422                              ". 098f6bcd4621d373cade4e832627b4f6+4 0:4:foo\n")
 423
 424     def test_write_multiple_files(self):
 425         cwriter = arvados.CollectionWriter(self.api_client)
 426         for letter in 'ABC':
 427             with self.make_test_file(letter) as testfile:
 428                 cwriter.write_file(testfile.name, letter)
 429         self.assertEqual(
 430             cwriter.manifest_text(),
 431             ". 902fbdd2b1df0c4f70b4a5d23525e932+3 0:1:A 1:1:B 2:1:C\n")
 432
 433     def test_basic_resume(self):
 434         cwriter = TestResumableWriter()
 435         with self.make_test_file() as testfile:
 436             cwriter.write_file(testfile.name, 'test')
 437             resumed = TestResumableWriter.from_state(cwriter.current_state())
 438         self.assertEquals(cwriter.manifest_text(), resumed.manifest_text(),
 439                           "resumed CollectionWriter had different manifest")
 440
 441     def test_resume_fails_when_missing_dependency(self):
 442         cwriter = TestResumableWriter()
 443         with self.make_test_file() as testfile:
 444             cwriter.write_file(testfile.name, 'test')
 445         self.assertRaises(arvados.errors.StaleWriterStateError,
 446                           TestResumableWriter.from_state,
 447                           cwriter.current_state())
 448
 449     def test_resume_fails_when_dependency_mtime_changed(self):
 450         cwriter = TestResumableWriter()
 451         with self.make_test_file() as testfile:
 452             cwriter.write_file(testfile.name, 'test')
 453             os.utime(testfile.name, (0, 0))
 454             self.assertRaises(arvados.errors.StaleWriterStateError,
 455                               TestResumableWriter.from_state,
 456                               cwriter.current_state())
 457
 458     def test_resume_fails_when_dependency_is_nonfile(self):
 459         cwriter = TestResumableWriter()
 460         cwriter.write_file('/dev/null', 'empty')
 461         self.assertRaises(arvados.errors.StaleWriterStateError,
 462                           TestResumableWriter.from_state,
 463                           cwriter.current_state())
 464
 465     def test_resume_fails_when_dependency_size_changed(self):
 466         cwriter = TestResumableWriter()
 467         with self.make_test_file() as testfile:
 468             cwriter.write_file(testfile.name, 'test')
 469             orig_mtime = os.fstat(testfile.fileno()).st_mtime
 470             testfile.write('extra')
 471             testfile.flush()
 472             os.utime(testfile.name, (orig_mtime, orig_mtime))
 473             self.assertRaises(arvados.errors.StaleWriterStateError,
 474                               TestResumableWriter.from_state,
 475                               cwriter.current_state())
 476
 477     def test_resume_fails_with_expired_locator(self):
 478         cwriter = TestResumableWriter()
 479         state = cwriter.current_state()
 480         # Add an expired locator to the state.
 481         state['_current_stream_locators'].append(''.join([
 482                     'a' * 32, '+1+A', 'b' * 40, '@', '10000000']))
 483         self.assertRaises(arvados.errors.StaleWriterStateError,
 484                           TestResumableWriter.from_state, state)
 485
 486     def test_arbitrary_objects_not_resumable(self):
 487         cwriter = TestResumableWriter()
 488         with open('/dev/null') as badfile:
 489             self.assertRaises(arvados.errors.AssertionError,
 490                               cwriter.write_file, badfile)
 491
 492     def test_arbitrary_writes_not_resumable(self):
 493         cwriter = TestResumableWriter()
 494         self.assertRaises(arvados.errors.AssertionError,
 495                           cwriter.write, "badtext")
 496
 497     def test_read_arbitrary_data_with_collection_reader(self):
 498         # arv-get relies on this to do "arv-get {keep-locator} -".
 499         self.write_foo_bar_baz()
 500         self.assertEqual(
 501             'foobar',
 502             arvados.CollectionReader(
 503                 '3858f62230ac3c915f300c664312c63f+6'
 504                 ).manifest_text())
 505
 506
 507 class CollectionTestMixin(object):
 508     PROXY_RESPONSE = {
 509         'items_available': 1,
 510         'items': [{
 511                 'uuid': 'zzzzz-bi6l4-mockproxy012345',
 512                 'owner_uuid': 'zzzzz-tpzed-mockowner012345',
 513                 'service_host': tutil.TEST_HOST,
 514                 'service_port': 65535,
 515                 'service_ssl_flag': True,
 516                 'service_type': 'proxy',
 517                 }]}
 518     API_COLLECTIONS = run_test_server.fixture('collections')
 519     DEFAULT_COLLECTION = API_COLLECTIONS['foo_file']
 520     DEFAULT_DATA_HASH = DEFAULT_COLLECTION['portable_data_hash']
 521     DEFAULT_MANIFEST = DEFAULT_COLLECTION['manifest_text']
 522     DEFAULT_UUID = DEFAULT_COLLECTION['uuid']
 523     ALT_COLLECTION = API_COLLECTIONS['bar_file']
 524     ALT_DATA_HASH = ALT_COLLECTION['portable_data_hash']
 525     ALT_MANIFEST = ALT_COLLECTION['manifest_text']
 526
 527     def _mock_api_call(self, mock_method, code, body):
 528         mock_method = mock_method().execute
 529         if code == 200:
 530             mock_method.return_value = body
 531         else:
 532             mock_method.side_effect = arvados.errors.ApiError(
 533                 tutil.fake_httplib2_response(code), "{}")
 534
 535     def mock_keep_services(self, api_mock, code, body):
 536         self._mock_api_call(api_mock.keep_services().accessible, code, body)
 537
 538     def api_client_mock(self, code=200):
 539         client = mock.MagicMock(name='api_client')
 540         self.mock_keep_services(client, code, self.PROXY_RESPONSE)
 541         return client
 542
 543
 544 @tutil.skip_sleep
 545 class CollectionReaderTestCase(unittest.TestCase, CollectionTestMixin):
 546     def mock_get_collection(self, api_mock, code, body):
 547         body = self.API_COLLECTIONS.get(body)
 548         self._mock_api_call(api_mock.collections().get, code, body)
 549
 550     def api_client_mock(self, code=200):
 551         client = super(CollectionReaderTestCase, self).api_client_mock(code)
 552         self.mock_get_collection(client, code, 'foo_file')
 553         return client
 554
 555     def test_init_no_default_retries(self):
 556         client = self.api_client_mock(200)
 557         reader = arvados.CollectionReader(self.DEFAULT_UUID, api_client=client)
 558         reader.manifest_text()
 559         client.collections().get().execute.assert_called_with(num_retries=0)
 560
 561     def test_uuid_init_success(self):
 562         client = self.api_client_mock(200)
 563         reader = arvados.CollectionReader(self.DEFAULT_UUID, api_client=client,
 564                                           num_retries=3)
 565         self.assertEqual(self.DEFAULT_COLLECTION['manifest_text'],
 566                          reader.manifest_text())
 567         client.collections().get().execute.assert_called_with(num_retries=3)
 568
 569     def test_uuid_init_failure_raises_api_error(self):
 570         client = self.api_client_mock(500)
 571         reader = arvados.CollectionReader(self.DEFAULT_UUID, api_client=client)
 572         with self.assertRaises(arvados.errors.ApiError):
 573             reader.manifest_text()
 574
 575     def test_locator_init(self):
 576         client = self.api_client_mock(200)
 577         # Ensure Keep will not return anything if asked.
 578         with tutil.mock_get_responses(None, 404):
 579             reader = arvados.CollectionReader(self.DEFAULT_DATA_HASH,
 580                                               api_client=client)
 581             self.assertEqual(self.DEFAULT_MANIFEST, reader.manifest_text())
 582
 583     def test_locator_init_fallback_to_keep(self):
 584         # crunch-job needs this to read manifests that have only ever
 585         # been written to Keep.
 586         client = self.api_client_mock(200)
 587         self.mock_get_collection(client, 404, None)
 588         with tutil.mock_get_responses(self.DEFAULT_MANIFEST, 200):
 589             reader = arvados.CollectionReader(self.DEFAULT_DATA_HASH,
 590                                               api_client=client)
 591             self.assertEqual(self.DEFAULT_MANIFEST, reader.manifest_text())
 592
 593     def test_uuid_init_no_fallback_to_keep(self):
 594         # Do not look up a collection UUID in Keep.
 595         client = self.api_client_mock(404)
 596         reader = arvados.CollectionReader(self.DEFAULT_UUID,
 597                                           api_client=client)
 598         with tutil.mock_get_responses(self.DEFAULT_MANIFEST, 200):
 599             with self.assertRaises(arvados.errors.ApiError):
 600                 reader.manifest_text()
 601
 602     def test_try_keep_first_if_permission_hint(self):
 603         # To verify that CollectionReader tries Keep first here, we
 604         # mock API server to return the wrong data.
 605         client = self.api_client_mock(200)
 606         with tutil.mock_get_responses(self.ALT_MANIFEST, 200):
 607             self.assertEqual(
 608                 self.ALT_MANIFEST,
 609                 arvados.CollectionReader(
 610                     self.ALT_DATA_HASH + '+Affffffffffffffffffffffffffffffffffffffff@fedcba98',
 611                     api_client=client).manifest_text())
 612
 613     def test_init_num_retries_propagated(self):
 614         # More of an integration test...
 615         client = self.api_client_mock(200)
 616         reader = arvados.CollectionReader(self.DEFAULT_UUID, api_client=client,
 617                                           num_retries=3)
 618         with tutil.mock_get_responses('foo', 500, 500, 200):
 619             self.assertEqual('foo',
 620                              ''.join(f.read(9) for f in reader.all_files()))
 621
 622     def test_read_nonnormalized_manifest_with_collection_reader(self):
 623         # client should be able to use CollectionReader on a manifest without normalizing it
 624         client = self.api_client_mock(500)
 625         nonnormal = ". acbd18db4cc2f85cedef654fccc4a4d8+3+Aabadbadbee@abeebdee 0:3:foo.txt 1:0:bar.txt 0:3:foo.txt\n"
 626         reader = arvados.CollectionReader(
 627             nonnormal,
 628             api_client=client, num_retries=0)
 629         # Ensure stripped_manifest() doesn't mangle our manifest in
 630         # any way other than stripping hints.
 631         self.assertEqual(
 632             re.sub('\+[^\d\s\+]+', '', nonnormal),
 633             reader.stripped_manifest())
 634         # Ensure stripped_manifest() didn't mutate our reader.
 635         self.assertEqual(nonnormal, reader.manifest_text())
 636         # Ensure the files appear in the order given in the manifest.
 637         self.assertEqual(
 638             [[6, '.', 'foo.txt'],
 639              [0, '.', 'bar.txt']],
 640             [[f.size(), f.stream_name(), f.name()]
 641              for f in reader.all_streams()[0].all_files()])
 642
 643     def test_read_empty_collection(self):
 644         client = self.api_client_mock(200)
 645         self.mock_get_collection(client, 200, 'empty')
 646         reader = arvados.CollectionReader('d41d8cd98f00b204e9800998ecf8427e+0',
 647                                           api_client=client)
 648         self.assertEqual('', reader.manifest_text())
 649
 650     def test_api_response(self):
 651         client = self.api_client_mock()
 652         reader = arvados.CollectionReader(self.DEFAULT_UUID, api_client=client)
 653         self.assertEqual(self.DEFAULT_COLLECTION, reader.api_response())
 654
 655     def test_api_response_with_collection_from_keep(self):
 656         client = self.api_client_mock()
 657         self.mock_get_collection(client, 404, 'foo')
 658         with tutil.mock_get_responses(self.DEFAULT_MANIFEST, 200):
 659             reader = arvados.CollectionReader(self.DEFAULT_DATA_HASH,
 660                                               api_client=client)
 661             api_response = reader.api_response()
 662         self.assertIsNone(api_response)
 663
 664     def check_open_file(self, coll_file, stream_name, file_name, file_size):
 665         self.assertFalse(coll_file.closed, "returned file is not open")
 666         self.assertEqual(stream_name, coll_file.stream_name())
 667         self.assertEqual(file_name, coll_file.name())
 668         self.assertEqual(file_size, coll_file.size())
 669
 670     def test_open_collection_file_one_argument(self):
 671         client = self.api_client_mock(200)
 672         reader = arvados.CollectionReader(self.DEFAULT_UUID, api_client=client)
 673         cfile = reader.open('./foo')
 674         self.check_open_file(cfile, '.', 'foo', 3)
 675
 676     def test_open_collection_file_two_arguments(self):
 677         client = self.api_client_mock(200)
 678         reader = arvados.CollectionReader(self.DEFAULT_UUID, api_client=client)
 679         cfile = reader.open('.', 'foo')
 680         self.check_open_file(cfile, '.', 'foo', 3)
 681
 682     def test_open_deep_file(self):
 683         coll_name = 'collection_with_files_in_subdir'
 684         client = self.api_client_mock(200)
 685         self.mock_get_collection(client, 200, coll_name)
 686         reader = arvados.CollectionReader(
 687             self.API_COLLECTIONS[coll_name]['uuid'], api_client=client)
 688         cfile = reader.open('./subdir2/subdir3/file2_in_subdir3.txt')
 689         self.check_open_file(cfile, './subdir2/subdir3', 'file2_in_subdir3.txt',
 690                              32)
 691
 692     def test_open_nonexistent_stream(self):
 693         client = self.api_client_mock(200)
 694         reader = arvados.CollectionReader(self.DEFAULT_UUID, api_client=client)
 695         self.assertRaises(ValueError, reader.open, './nonexistent', 'foo')
 696
 697     def test_open_nonexistent_file(self):
 698         client = self.api_client_mock(200)
 699         reader = arvados.CollectionReader(self.DEFAULT_UUID, api_client=client)
 700         self.assertRaises(ValueError, reader.open, '.', 'nonexistent')
 701
 702
 703 @tutil.skip_sleep
 704 class CollectionWriterTestCase(unittest.TestCase, CollectionTestMixin):
 705     def mock_keep(self, body, *codes, **headers):
 706         headers.setdefault('x-keep-replicas-stored', 2)
 707         return tutil.mock_put_responses(body, *codes, **headers)
 708
 709     def foo_writer(self, **kwargs):
 710         api_client = self.api_client_mock()
 711         writer = arvados.CollectionWriter(api_client, **kwargs)
 712         writer.start_new_file('foo')
 713         writer.write('foo')
 714         return writer
 715
 716     def test_write_whole_collection(self):
 717         writer = self.foo_writer()
 718         with self.mock_keep(self.DEFAULT_DATA_HASH, 200, 200):
 719             self.assertEqual(self.DEFAULT_DATA_HASH, writer.finish())
 720
 721     def test_write_no_default(self):
 722         writer = self.foo_writer()
 723         with self.mock_keep(None, 500):
 724             with self.assertRaises(arvados.errors.KeepWriteError):
 725                 writer.finish()
 726
 727     def test_write_whole_collection_through_retries(self):
 728         writer = self.foo_writer(num_retries=2)
 729         with self.mock_keep(self.DEFAULT_DATA_HASH,
 730                             500, 500, 200, 500, 500, 200):
 731             self.assertEqual(self.DEFAULT_DATA_HASH, writer.finish())
 732
 733     def test_flush_data_retries(self):
 734         writer = self.foo_writer(num_retries=2)
 735         foo_hash = self.DEFAULT_MANIFEST.split()[1]
 736         with self.mock_keep(foo_hash, 500, 200):
 737             writer.flush_data()
 738         self.assertEqual(self.DEFAULT_MANIFEST, writer.manifest_text())
 739
 740     def test_one_open(self):
 741         client = self.api_client_mock()
 742         writer = arvados.CollectionWriter(client)
 743         with writer.open('out') as out_file:
 744             self.assertEqual('.', writer.current_stream_name())
 745             self.assertEqual('out', writer.current_file_name())
 746             out_file.write('test data')
 747             data_loc = hashlib.md5('test data').hexdigest() + '+9'
 748         self.assertTrue(out_file.closed, "writer file not closed after context")
 749         self.assertRaises(ValueError, out_file.write, 'extra text')
 750         with self.mock_keep(data_loc, 200) as keep_mock:
 751             self.assertEqual(". {} 0:9:out\n".format(data_loc),
 752                              writer.manifest_text())
 753
 754     def test_open_writelines(self):
 755         client = self.api_client_mock()
 756         writer = arvados.CollectionWriter(client)
 757         with writer.open('six') as out_file:
 758             out_file.writelines(['12', '34', '56'])
 759             data_loc = hashlib.md5('123456').hexdigest() + '+6'
 760         with self.mock_keep(data_loc, 200) as keep_mock:
 761             self.assertEqual(". {} 0:6:six\n".format(data_loc),
 762                              writer.manifest_text())
 763
 764     def test_open_flush(self):
 765         client = self.api_client_mock()
 766         data_loc1 = hashlib.md5('flush1').hexdigest() + '+6'
 767         data_loc2 = hashlib.md5('flush2').hexdigest() + '+6'
 768         with self.mock_keep((data_loc1, 200), (data_loc2, 200)) as keep_mock:
 769             writer = arvados.CollectionWriter(client)
 770             with writer.open('flush_test') as out_file:
 771                 out_file.write('flush1')
 772                 out_file.flush()
 773                 out_file.write('flush2')
 774             self.assertEqual(". {} {} 0:12:flush_test\n".format(data_loc1,
 775                                                                 data_loc2),
 776                              writer.manifest_text())
 777
 778     def test_two_opens_same_stream(self):
 779         client = self.api_client_mock()
 780         writer = arvados.CollectionWriter(client)
 781         with writer.open('.', '1') as out_file:
 782             out_file.write('1st')
 783         with writer.open('.', '2') as out_file:
 784             out_file.write('2nd')
 785         data_loc = hashlib.md5('1st2nd').hexdigest() + '+6'
 786         with self.mock_keep(data_loc, 200) as keep_mock:
 787             self.assertEqual(". {} 0:3:1 3:3:2\n".format(data_loc),
 788                              writer.manifest_text())
 789
 790     def test_two_opens_two_streams(self):
 791         client = self.api_client_mock()
 792         data_loc1 = hashlib.md5('file').hexdigest() + '+4'
 793         data_loc2 = hashlib.md5('indir').hexdigest() + '+5'
 794         with self.mock_keep((data_loc1, 200), (data_loc2, 200)) as keep_mock:
 795             writer = arvados.CollectionWriter(client)
 796             with writer.open('file') as out_file:
 797                 out_file.write('file')
 798             with writer.open('./dir', 'indir') as out_file:
 799                 out_file.write('indir')
 800             expected = ". {} 0:4:file\n./dir {} 0:5:indir\n".format(
 801                 data_loc1, data_loc2)
 802             self.assertEqual(expected, writer.manifest_text())
 803
 804     def test_dup_open_fails(self):
 805         client = self.api_client_mock()
 806         writer = arvados.CollectionWriter(client)
 807         file1 = writer.open('one')
 808         self.assertRaises(arvados.errors.AssertionError, writer.open, 'two')
 809
 810
 811 class NewCollectionTestCase(unittest.TestCase, CollectionTestMixin):
 812     def test_import_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(". 5348b82a029fd9e971a811ce1f71360b+43 085c37f02916da1cad16f93c54d899b7+41 8b22da26f9f433dea0a10e5ec66d73ba+43 0:127:md5sum.txt\n", arvados.export_manifest(Collection(m1)))
 818
 819     def test_init_manifest(self):
 820         m1 = """. 5348b82a029fd9e971a811ce1f71360b+43 0:43:md5sum.txt
 821 . 085c37f02916da1cad16f93c54d899b7+41 0:41:md5sum.txt
 822 . 8b22da26f9f433dea0a10e5ec66d73ba+43 0:43:md5sum.txt
 823 """
 824         self.assertEqual(". 5348b82a029fd9e971a811ce1f71360b+43 085c37f02916da1cad16f93c54d899b7+41 8b22da26f9f433dea0a10e5ec66d73ba+43 0:127:md5sum.txt\n", arvados.export_manifest(arvados.Collection(m1)))
 825
 826
 827     def test_remove(self):
 828         c = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt 0:10:count2.txt\n', sync=SYNC_EXPLICIT)
 829         self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt 0:10:count2.txt\n", export_manifest(c))
 830         self.assertIn("count1.txt", c)
 831         c.remove("count1.txt")
 832         self.assertNotIn("count1.txt", c)
 833         self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count2.txt\n", export_manifest(c))
 834
 835     def test_remove_in_subdir(self):
 836         c = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n./foo 781e5e245d69b566979b86e28d23f2c7+10 0:10:count2.txt\n', sync=SYNC_EXPLICIT)
 837         c.remove("foo/count2.txt")
 838         self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n", export_manifest(c))
 839
 840     def test_remove_empty_subdir(self):
 841         c = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n./foo 781e5e245d69b566979b86e28d23f2c7+10 0:10:count2.txt\n', sync=SYNC_EXPLICIT)
 842         c.remove("foo/count2.txt")
 843         c.remove("foo")
 844         self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n", export_manifest(c))
 845
 846     def test_remove_nonempty_subdir(self):
 847         c = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n./foo 781e5e245d69b566979b86e28d23f2c7+10 0:10:count2.txt\n', sync=SYNC_EXPLICIT)
 848         with self.assertRaises(IOError):
 849             c.remove("foo")
 850         c.remove("foo", rm_r=True)
 851         self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n", export_manifest(c))
 852
 853     def test_copy_to_dir1(self):
 854         c = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n', sync=SYNC_EXPLICIT)
 855         c.copy("count1.txt", "foo/count2.txt")
 856         self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n./foo 781e5e245d69b566979b86e28d23f2c7+10 0:10:count2.txt\n", export_manifest(c))
 857
 858     def test_copy_to_dir2(self):
 859         c = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n', sync=SYNC_EXPLICIT)
 860         c.copy("count1.txt", "foo")
 861         self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n./foo 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n", export_manifest(c))
 862
 863     def test_copy_to_dir2(self):
 864         c = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n', sync=SYNC_EXPLICIT)
 865         c.copy("count1.txt", "foo/")
 866         self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n./foo 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n", export_manifest(c))
 867
 868     def test_copy_file(self):
 869         c = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n', sync=SYNC_EXPLICIT)
 870         c.copy("count1.txt", "count2.txt")
 871         self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt 0:10:count2.txt\n", c.manifest_text())
 872
 873     def test_clone(self):
 874         c = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n./foo 781e5e245d69b566979b86e28d23f2c7+10 0:10:count2.txt\n', sync=SYNC_EXPLICIT)
 875         cl = c.clone()
 876         self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n./foo 781e5e245d69b566979b86e28d23f2c7+10 0:10:count2.txt\n", export_manifest(cl))
 877
 878     def test_diff1(self):
 879         c1 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n', sync=SYNC_EXPLICIT)
 880         c2 = Collection('. 5348b82a029fd9e971a811ce1f71360b+43 0:10:count2.txt\n')
 881         d = c2.diff(c1)
 882         self.assertEqual(d, [('del', './count2.txt', c2["count2.txt"]),
 883                              ('add', './count1.txt', c1["count1.txt"])])
 884         d = c1.diff(c2)
 885         self.assertEqual(d, [('del', './count1.txt', c1["count1.txt"]),
 886                              ('add', './count2.txt', c2["count2.txt"])])
 887         self.assertNotEqual(c1.manifest_text(), c2.manifest_text())
 888         c1.apply(d)
 889         self.assertEqual(c1.manifest_text(), c2.manifest_text())
 890
 891     def test_diff2(self):
 892         c1 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n', sync=SYNC_EXPLICIT)
 893         c2 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n')
 894         d = c2.diff(c1)
 895         self.assertEqual(d, [])
 896         d = c1.diff(c2)
 897         self.assertEqual(d, [])
 898
 899         self.assertEqual(c1.manifest_text(), c2.manifest_text())
 900         c1.apply(d)
 901         self.assertEqual(c1.manifest_text(), c2.manifest_text())
 902
 903     def test_diff3(self):
 904         c1 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n', sync=SYNC_EXPLICIT)
 905         c2 = Collection('. 5348b82a029fd9e971a811ce1f71360b+43 0:10:count1.txt\n')
 906         d = c2.diff(c1)
 907         self.assertEqual(d, [('mod', './count1.txt', c2["count1.txt"], c1["count1.txt"])])
 908         d = c1.diff(c2)
 909         self.assertEqual(d, [('mod', './count1.txt', c1["count1.txt"], c2["count1.txt"])])
 910
 911         self.assertNotEqual(c1.manifest_text(), c2.manifest_text())
 912         c1.apply(d)
 913         self.assertEqual(c1.manifest_text(), c2.manifest_text())
 914
 915     def test_diff4(self):
 916         c1 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n', sync=SYNC_EXPLICIT)
 917         c2 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 5348b82a029fd9e971a811ce1f71360b+43 0:10:count1.txt 10:20:count2.txt\n')
 918         d = c2.diff(c1)
 919         self.assertEqual(d, [('del', './count2.txt', c2["count2.txt"])])
 920         d = c1.diff(c2)
 921         self.assertEqual(d, [('add', './count2.txt', c2["count2.txt"])])
 922
 923         self.assertNotEqual(c1.manifest_text(), c2.manifest_text())
 924         c1.apply(d)
 925         self.assertEqual(c1.manifest_text(), c2.manifest_text())
 926
 927     def test_diff5(self):
 928         c1 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n', sync=SYNC_EXPLICIT)
 929         c2 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n./foo 5348b82a029fd9e971a811ce1f71360b+43 0:10:count2.txt\n')
 930         d = c2.diff(c1)
 931         self.assertEqual(d, [('del', './foo', c2["foo"])])
 932         d = c1.diff(c2)
 933         self.assertEqual(d, [('add', './foo', c2["foo"])])
 934
 935         self.assertNotEqual(c1.manifest_text(), c2.manifest_text())
 936         c1.apply(d)
 937         self.assertEqual(c1.manifest_text(), c2.manifest_text())
 938
 939     def test_diff6(self):
 940         c1 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n./foo 5348b82a029fd9e971a811ce1f71360b+43 0:10:count2.txt\n', sync=SYNC_EXPLICIT)
 941         c2 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n./foo 5348b82a029fd9e971a811ce1f71360b+43 0:3:count3.txt\n')
 942
 943         d = c2.diff(c1)
 944         self.assertEqual(d, [('del', './foo/count3.txt', c2.find("foo/count3.txt")),
 945                              ('add', './foo/count2.txt', c1.find("foo/count2.txt"))])
 946         d = c1.diff(c2)
 947         self.assertEqual(d, [('del', './foo/count2.txt', c1.find("foo/count2.txt")),
 948                              ('add', './foo/count3.txt', c2.find("foo/count3.txt"))])
 949
 950         self.assertNotEqual(c1.manifest_text(), c2.manifest_text())
 951         c1.apply(d)
 952         self.assertEqual(c1.manifest_text(), c2.manifest_text())
 953
 954     def test_diff7(self):
 955         c1 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n./foo 5348b82a029fd9e971a811ce1f71360b+43 0:10:count2.txt\n', sync=SYNC_EXPLICIT)
 956         c2 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt 0:3:foo\n')
 957         d = c2.diff(c1)
 958         self.assertEqual(d, [('mod', './foo', c2["foo"], c1["foo"])])
 959         d = c1.diff(c2)
 960         self.assertEqual(d, [('mod', './foo', c1["foo"], c2["foo"])])
 961
 962         self.assertNotEqual(c1.manifest_text(), c2.manifest_text())
 963         c1.apply(d)
 964         self.assertEqual(c1.manifest_text(), c2.manifest_text())
 965
 966     def test_conflict1(self):
 967         c1 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n', sync=SYNC_EXPLICIT)
 968         c2 = Collection('. 5348b82a029fd9e971a811ce1f71360b+43 0:10:count2.txt\n')
 969         d = c1.diff(c2)
 970         self.assertEqual(d, [('del', './count1.txt', c1["count1.txt"]),
 971                              ('add', './count2.txt', c2["count2.txt"])])
 972         with c1.open("count1.txt", "w") as f:
 973             f.write("zzzzz")
 974
 975         # c1 changed, so it should not be deleted.
 976         c1.apply(d)
 977         self.assertEqual(c1.manifest_text(), ". 95ebc3c7b3b9f1d2c40fec14415d3cb8+5 5348b82a029fd9e971a811ce1f71360b+43 0:5:count1.txt 5:10:count2.txt\n")
 978
 979     def test_conflict2(self):
 980         c1 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt', sync=SYNC_EXPLICIT)
 981         c2 = Collection('. 5348b82a029fd9e971a811ce1f71360b+43 0:10:count1.txt')
 982         d = c1.diff(c2)
 983         self.assertEqual(d, [('mod', './count1.txt', c1["count1.txt"], c2["count1.txt"])])
 984         with c1.open("count1.txt", "w") as f:
 985             f.write("zzzzz")
 986
 987         # c1 changed, so c2 mod will go to a conflict file
 988         c1.apply(d)
 989         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~$",
 990                                  c1.manifest_text()))
 991
 992     def test_conflict3(self):
 993         c1 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count2.txt\n', sync=SYNC_EXPLICIT)
 994         c2 = Collection('. 5348b82a029fd9e971a811ce1f71360b+43 0:10:count1.txt\n')
 995         d = c1.diff(c2)
 996         self.assertEqual(d, [('del', './count2.txt', c1["count2.txt"]),
 997                              ('add', './count1.txt', c2["count1.txt"])])
 998         with c1.open("count1.txt", "w") as f:
 999             f.write("zzzzz")
1000
1001         # c1 added count1.txt, so c2 add will go to a conflict file
1002         c1.apply(d)
1003         self.assertTrue(re.match(r"\. 95ebc3c7b3b9f1d2c40fec14415d3cb8\+5 5348b82a029fd9e971a811ce1f71360b\+43 0:5:count1.txt 5:10:count1.txt~conflict-\d\d\d\d-\d\d-\d\d-\d\d:\d\d:\d\d~$",
1004                                  c1.manifest_text()))
1005
1006     def test_conflict4(self):
1007         c1 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt', sync=SYNC_EXPLICIT)
1008         c2 = Collection('. 5348b82a029fd9e971a811ce1f71360b+43 0:10:count1.txt')
1009         d = c1.diff(c2)
1010         self.assertEqual(d, [('mod', './count1.txt', c1["count1.txt"], c2["count1.txt"])])
1011         c1.remove("count1.txt")
1012
1013         # c1 deleted, so c2 mod will go to a conflict file
1014         c1.apply(d)
1015         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~$",
1016                                  c1.manifest_text()))
1017
1018     def test_notify1(self):
1019         c1 = Collection(sync=SYNC_EXPLICIT)
1020         events = []
1021         c1.subscribe(lambda event, collection, name, item: events.append((event, collection, name, item)))
1022         c1.find("")
1023
1024 if __name__ == '__main__':
1025     unittest.main()