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 mock
   8 import os
   9 import pprint
  10 import re
  11 import tempfile
  12 import unittest
  13
  14 import run_test_server
  15 from arvados._ranges import Range, LocatorAndRange
  16 from arvados.collection import Collection, CollectionReader
  17 import arvados_testutil as tutil
  18
  19 class TestResumableWriter(arvados.ResumableCollectionWriter):
  20     KEEP_BLOCK_SIZE = 1024  # PUT to Keep every 1K.
  21
  22     def current_state(self):
  23         return self.dump_state(copy.deepcopy)
  24
  25
  26 class ArvadosCollectionsTest(run_test_server.TestCaseWithServers,
  27                              tutil.ArvadosBaseTestCase):
  28     MAIN_SERVER = {}
  29
  30     @classmethod
  31     def setUpClass(cls):
  32         super(ArvadosCollectionsTest, cls).setUpClass()
  33         run_test_server.authorize_with('active')
  34         cls.api_client = arvados.api('v1')
  35         cls.keep_client = arvados.KeepClient(api_client=cls.api_client,
  36                                              local_store=cls.local_store)
  37
  38     def write_foo_bar_baz(self):
  39         cw = arvados.CollectionWriter(self.api_client)
  40         self.assertEqual(cw.current_stream_name(), '.',
  41                          'current_stream_name() should be "." now')
  42         cw.set_current_file_name('foo.txt')
  43         cw.write('foo')
  44         self.assertEqual(cw.current_file_name(), 'foo.txt',
  45                          'current_file_name() should be foo.txt now')
  46         cw.start_new_file('bar.txt')
  47         cw.write('bar')
  48         cw.start_new_stream('baz')
  49         cw.write('baz')
  50         cw.set_current_file_name('baz.txt')
  51         self.assertEqual(cw.manifest_text(),
  52                          ". 3858f62230ac3c915f300c664312c63f+6 0:3:foo.txt 3:3:bar.txt\n" +
  53                          "./baz 73feffa4b7f6bb68e44cf984c85f6e88+3 0:3:baz.txt\n",
  54                          "wrong manifest: got {}".format(cw.manifest_text()))
  55         cw.finish()
  56         return cw.portable_data_hash()
  57
  58     def test_keep_local_store(self):
  59         self.assertEqual(self.keep_client.put('foo'), 'acbd18db4cc2f85cedef654fccc4a4d8+3', 'wrong md5 hash from Keep.put')
  60         self.assertEqual(self.keep_client.get('acbd18db4cc2f85cedef654fccc4a4d8+3'), 'foo', 'wrong data from Keep.get')
  61
  62     def test_local_collection_writer(self):
  63         self.assertEqual(self.write_foo_bar_baz(),
  64                          '23ca013983d6239e98931cc779e68426+114',
  65                          'wrong locator hash: ' + self.write_foo_bar_baz())
  66
  67     def test_local_collection_reader(self):
  68         foobarbaz = self.write_foo_bar_baz()
  69         cr = arvados.CollectionReader(
  70             foobarbaz + '+Xzizzle', self.api_client)
  71         got = []
  72         for s in cr.all_streams():
  73             for f in s.all_files():
  74                 got += [[f.size(), f.stream_name(), f.name(), f.read(2**26)]]
  75         expected = [[3, '.', 'foo.txt', 'foo'],
  76                     [3, '.', 'bar.txt', 'bar'],
  77                     [3, './baz', 'baz.txt', 'baz']]
  78         self.assertEqual(got,
  79                          expected)
  80         stream0 = cr.all_streams()[0]
  81         self.assertEqual(stream0.readfrom(0, 0),
  82                          '',
  83                          'reading zero bytes should have returned empty string')
  84         self.assertEqual(stream0.readfrom(0, 2**26),
  85                          'foobar',
  86                          'reading entire stream failed')
  87         self.assertEqual(stream0.readfrom(2**26, 0),
  88                          '',
  89                          'reading zero bytes should have returned empty string')
  90
  91     def _test_subset(self, collection, expected):
  92         cr = arvados.CollectionReader(collection, self.api_client)
  93         for s in cr.all_streams():
  94             for ex in expected:
  95                 if ex[0] == s:
  96                     f = s.files()[ex[2]]
  97                     got = [f.size(), f.stream_name(), f.name(), "".join(f.readall(2**26))]
  98                     self.assertEqual(got,
  99                                      ex,
 100                                      'all_files|as_manifest did not preserve manifest contents: got %s expected %s' % (got, ex))
 101
 102     def test_collection_manifest_subset(self):
 103         foobarbaz = self.write_foo_bar_baz()
 104         self._test_subset(foobarbaz,
 105                           [[3, '.',     'bar.txt', 'bar'],
 106                            [3, '.',     'foo.txt', 'foo'],
 107                            [3, './baz', 'baz.txt', 'baz']])
 108         self._test_subset((". %s %s 0:3:foo.txt 3:3:bar.txt\n" %
 109                            (self.keep_client.put("foo"),
 110                             self.keep_client.put("bar"))),
 111                           [[3, '.', 'bar.txt', 'bar'],
 112                            [3, '.', 'foo.txt', 'foo']])
 113         self._test_subset((". %s %s 0:2:fo.txt 2:4:obar.txt\n" %
 114                            (self.keep_client.put("foo"),
 115                             self.keep_client.put("bar"))),
 116                           [[2, '.', 'fo.txt', 'fo'],
 117                            [4, '.', 'obar.txt', 'obar']])
 118         self._test_subset((". %s %s 0:2:fo.txt 2:0:zero.txt 2:2:ob.txt 4:2:ar.txt\n" %
 119                            (self.keep_client.put("foo"),
 120                             self.keep_client.put("bar"))),
 121                           [[2, '.', 'ar.txt', 'ar'],
 122                            [2, '.', 'fo.txt', 'fo'],
 123                            [2, '.', 'ob.txt', 'ob'],
 124                            [0, '.', 'zero.txt', '']])
 125
 126     def test_collection_empty_file(self):
 127         cw = arvados.CollectionWriter(self.api_client)
 128         cw.start_new_file('zero.txt')
 129         cw.write('')
 130
 131         self.assertEqual(cw.manifest_text(), ". d41d8cd98f00b204e9800998ecf8427e+0 0:0:zero.txt\n")
 132         self.check_manifest_file_sizes(cw.manifest_text(), [0])
 133         cw = arvados.CollectionWriter(self.api_client)
 134         cw.start_new_file('zero.txt')
 135         cw.write('')
 136         cw.start_new_file('one.txt')
 137         cw.write('1')
 138         cw.start_new_stream('foo')
 139         cw.start_new_file('zero.txt')
 140         cw.write('')
 141         self.check_manifest_file_sizes(cw.manifest_text(), [0,1,0])
 142
 143     def test_no_implicit_normalize(self):
 144         cw = arvados.CollectionWriter(self.api_client)
 145         cw.start_new_file('b')
 146         cw.write('b')
 147         cw.start_new_file('a')
 148         cw.write('')
 149         self.check_manifest_file_sizes(cw.manifest_text(), [1,0])
 150         self.check_manifest_file_sizes(
 151             arvados.CollectionReader(
 152                 cw.manifest_text()).manifest_text(normalize=True),
 153             [0,1])
 154
 155     def check_manifest_file_sizes(self, manifest_text, expect_sizes):
 156         cr = arvados.CollectionReader(manifest_text, self.api_client)
 157         got_sizes = []
 158         for f in cr.all_files():
 159             got_sizes += [f.size()]
 160         self.assertEqual(got_sizes, expect_sizes, "got wrong file sizes %s, expected %s" % (got_sizes, expect_sizes))
 161
 162     def test_normalized_collection(self):
 163         m1 = """. 5348b82a029fd9e971a811ce1f71360b+43 0:43:md5sum.txt
 164 . 085c37f02916da1cad16f93c54d899b7+41 0:41:md5sum.txt
 165 . 8b22da26f9f433dea0a10e5ec66d73ba+43 0:43:md5sum.txt
 166 """
 167         self.assertEqual(arvados.CollectionReader(m1, self.api_client).manifest_text(normalize=True),
 168                          """. 5348b82a029fd9e971a811ce1f71360b+43 085c37f02916da1cad16f93c54d899b7+41 8b22da26f9f433dea0a10e5ec66d73ba+43 0:127:md5sum.txt
 169 """)
 170
 171         m2 = """. 204e43b8a1185621ca55a94839582e6f+67108864 b9677abbac956bd3e86b1deb28dfac03+67108864 fc15aff2a762b13f521baf042140acec+67108864 323d2a3ce20370c4ca1d3462a344f8fd+25885655 0:227212247:var-GS000016015-ASM.tsv.bz2
 172 """
 173         self.assertEqual(arvados.CollectionReader(m2, self.api_client).manifest_text(normalize=True), m2)
 174
 175         m3 = """. 5348b82a029fd9e971a811ce1f71360b+43 3:40:md5sum.txt
 176 . 085c37f02916da1cad16f93c54d899b7+41 0:41:md5sum.txt
 177 . 8b22da26f9f433dea0a10e5ec66d73ba+43 0:43:md5sum.txt
 178 """
 179         self.assertEqual(arvados.CollectionReader(m3, self.api_client).manifest_text(normalize=True),
 180                          """. 5348b82a029fd9e971a811ce1f71360b+43 085c37f02916da1cad16f93c54d899b7+41 8b22da26f9f433dea0a10e5ec66d73ba+43 3:124:md5sum.txt
 181 """)
 182
 183         m4 = """. 204e43b8a1185621ca55a94839582e6f+67108864 0:3:foo/bar
 184 ./zzz 204e43b8a1185621ca55a94839582e6f+67108864 0:999:zzz
 185 ./foo 323d2a3ce20370c4ca1d3462a344f8fd+25885655 0:3:bar
 186 """
 187         self.assertEqual(arvados.CollectionReader(m4, self.api_client).manifest_text(normalize=True),
 188                          """./foo 204e43b8a1185621ca55a94839582e6f+67108864 323d2a3ce20370c4ca1d3462a344f8fd+25885655 0:3:bar 67108864:3:bar
 189 ./zzz 204e43b8a1185621ca55a94839582e6f+67108864 0:999:zzz
 190 """)
 191
 192         m5 = """. 204e43b8a1185621ca55a94839582e6f+67108864 0:3:foo/bar
 193 ./zzz 204e43b8a1185621ca55a94839582e6f+67108864 0:999:zzz
 194 ./foo 204e43b8a1185621ca55a94839582e6f+67108864 3:3:bar
 195 """
 196         self.assertEqual(arvados.CollectionReader(m5, self.api_client).manifest_text(normalize=True),
 197                          """./foo 204e43b8a1185621ca55a94839582e6f+67108864 0:6:bar
 198 ./zzz 204e43b8a1185621ca55a94839582e6f+67108864 0:999:zzz
 199 """)
 200
 201         with self.data_file('1000G_ref_manifest') as f6:
 202             m6 = f6.read()
 203             self.assertEqual(arvados.CollectionReader(m6, self.api_client).manifest_text(normalize=True), m6)
 204
 205         with self.data_file('jlake_manifest') as f7:
 206             m7 = f7.read()
 207             self.assertEqual(arvados.CollectionReader(m7, self.api_client).manifest_text(normalize=True), m7)
 208
 209         m8 = """./a\\040b\\040c 59ca0efa9f5633cb0371bbc0355478d8+13 0:13:hello\\040world.txt
 210 """
 211         self.assertEqual(arvados.CollectionReader(m8, self.api_client).manifest_text(normalize=True), m8)
 212
 213     def test_locators_and_ranges(self):
 214         blocks2 = [Range('a', 0, 10),
 215                    Range('b', 10, 10),
 216                    Range('c', 20, 10),
 217                    Range('d', 30, 10),
 218                    Range('e', 40, 10),
 219                    Range('f', 50, 10)]
 220
 221         self.assertEqual(arvados.locators_and_ranges(blocks2,  2,  2), [LocatorAndRange('a', 10, 2, 2)])
 222         self.assertEqual(arvados.locators_and_ranges(blocks2, 12, 2), [LocatorAndRange('b', 10, 2, 2)])
 223         self.assertEqual(arvados.locators_and_ranges(blocks2, 22, 2), [LocatorAndRange('c', 10, 2, 2)])
 224         self.assertEqual(arvados.locators_and_ranges(blocks2, 32, 2), [LocatorAndRange('d', 10, 2, 2)])
 225         self.assertEqual(arvados.locators_and_ranges(blocks2, 42, 2), [LocatorAndRange('e', 10, 2, 2)])
 226         self.assertEqual(arvados.locators_and_ranges(blocks2, 52, 2), [LocatorAndRange('f', 10, 2, 2)])
 227         self.assertEqual(arvados.locators_and_ranges(blocks2, 62, 2), [])
 228         self.assertEqual(arvados.locators_and_ranges(blocks2, -2, 2), [])
 229
 230         self.assertEqual(arvados.locators_and_ranges(blocks2,  0,  2), [LocatorAndRange('a', 10, 0, 2)])
 231         self.assertEqual(arvados.locators_and_ranges(blocks2, 10, 2), [LocatorAndRange('b', 10, 0, 2)])
 232         self.assertEqual(arvados.locators_and_ranges(blocks2, 20, 2), [LocatorAndRange('c', 10, 0, 2)])
 233         self.assertEqual(arvados.locators_and_ranges(blocks2, 30, 2), [LocatorAndRange('d', 10, 0, 2)])
 234         self.assertEqual(arvados.locators_and_ranges(blocks2, 40, 2), [LocatorAndRange('e', 10, 0, 2)])
 235         self.assertEqual(arvados.locators_and_ranges(blocks2, 50, 2), [LocatorAndRange('f', 10, 0, 2)])
 236         self.assertEqual(arvados.locators_and_ranges(blocks2, 60, 2), [])
 237         self.assertEqual(arvados.locators_and_ranges(blocks2, -2, 2), [])
 238
 239         self.assertEqual(arvados.locators_and_ranges(blocks2,  9,  2), [LocatorAndRange('a', 10, 9, 1), LocatorAndRange('b', 10, 0, 1)])
 240         self.assertEqual(arvados.locators_and_ranges(blocks2, 19, 2), [LocatorAndRange('b', 10, 9, 1), LocatorAndRange('c', 10, 0, 1)])
 241         self.assertEqual(arvados.locators_and_ranges(blocks2, 29, 2), [LocatorAndRange('c', 10, 9, 1), LocatorAndRange('d', 10, 0, 1)])
 242         self.assertEqual(arvados.locators_and_ranges(blocks2, 39, 2), [LocatorAndRange('d', 10, 9, 1), LocatorAndRange('e', 10, 0, 1)])
 243         self.assertEqual(arvados.locators_and_ranges(blocks2, 49, 2), [LocatorAndRange('e', 10, 9, 1), LocatorAndRange('f', 10, 0, 1)])
 244         self.assertEqual(arvados.locators_and_ranges(blocks2, 59, 2), [LocatorAndRange('f', 10, 9, 1)])
 245
 246
 247         blocks3 = [Range('a', 0, 10),
 248                   Range('b', 10, 10),
 249                   Range('c', 20, 10),
 250                   Range('d', 30, 10),
 251                   Range('e', 40, 10),
 252                   Range('f', 50, 10),
 253                    Range('g', 60, 10)]
 254
 255         self.assertEqual(arvados.locators_and_ranges(blocks3,  2,  2), [LocatorAndRange('a', 10, 2, 2)])
 256         self.assertEqual(arvados.locators_and_ranges(blocks3, 12, 2), [LocatorAndRange('b', 10, 2, 2)])
 257         self.assertEqual(arvados.locators_and_ranges(blocks3, 22, 2), [LocatorAndRange('c', 10, 2, 2)])
 258         self.assertEqual(arvados.locators_and_ranges(blocks3, 32, 2), [LocatorAndRange('d', 10, 2, 2)])
 259         self.assertEqual(arvados.locators_and_ranges(blocks3, 42, 2), [LocatorAndRange('e', 10, 2, 2)])
 260         self.assertEqual(arvados.locators_and_ranges(blocks3, 52, 2), [LocatorAndRange('f', 10, 2, 2)])
 261         self.assertEqual(arvados.locators_and_ranges(blocks3, 62, 2), [LocatorAndRange('g', 10, 2, 2)])
 262
 263
 264         blocks = [Range('a', 0, 10),
 265                   Range('b', 10, 15),
 266                   Range('c', 25, 5)]
 267         self.assertEqual(arvados.locators_and_ranges(blocks, 1, 0), [])
 268         self.assertEqual(arvados.locators_and_ranges(blocks, 0, 5), [LocatorAndRange('a', 10, 0, 5)])
 269         self.assertEqual(arvados.locators_and_ranges(blocks, 3, 5), [LocatorAndRange('a', 10, 3, 5)])
 270         self.assertEqual(arvados.locators_and_ranges(blocks, 0, 10), [LocatorAndRange('a', 10, 0, 10)])
 271
 272         self.assertEqual(arvados.locators_and_ranges(blocks, 0, 11), [LocatorAndRange('a', 10, 0, 10),
 273                                                                       LocatorAndRange('b', 15, 0, 1)])
 274         self.assertEqual(arvados.locators_and_ranges(blocks, 1, 11), [LocatorAndRange('a', 10, 1, 9),
 275                                                                       LocatorAndRange('b', 15, 0, 2)])
 276         self.assertEqual(arvados.locators_and_ranges(blocks, 0, 25), [LocatorAndRange('a', 10, 0, 10),
 277                                                                       LocatorAndRange('b', 15, 0, 15)])
 278
 279         self.assertEqual(arvados.locators_and_ranges(blocks, 0, 30), [LocatorAndRange('a', 10, 0, 10),
 280                                                                       LocatorAndRange('b', 15, 0, 15),
 281                                                                       LocatorAndRange('c', 5, 0, 5)])
 282         self.assertEqual(arvados.locators_and_ranges(blocks, 1, 30), [LocatorAndRange('a', 10, 1, 9),
 283                                                                       LocatorAndRange('b', 15, 0, 15),
 284                                                                       LocatorAndRange('c', 5, 0, 5)])
 285         self.assertEqual(arvados.locators_and_ranges(blocks, 0, 31), [LocatorAndRange('a', 10, 0, 10),
 286                                                                       LocatorAndRange('b', 15, 0, 15),
 287                                                                       LocatorAndRange('c', 5, 0, 5)])
 288
 289         self.assertEqual(arvados.locators_and_ranges(blocks, 15, 5), [LocatorAndRange('b', 15, 5, 5)])
 290
 291         self.assertEqual(arvados.locators_and_ranges(blocks, 8, 17), [LocatorAndRange('a', 10, 8, 2),
 292                                                                       LocatorAndRange('b', 15, 0, 15)])
 293
 294         self.assertEqual(arvados.locators_and_ranges(blocks, 8, 20), [LocatorAndRange('a', 10, 8, 2),
 295                                                                       LocatorAndRange('b', 15, 0, 15),
 296                                                                       LocatorAndRange('c', 5, 0, 3)])
 297
 298         self.assertEqual(arvados.locators_and_ranges(blocks, 26, 2), [LocatorAndRange('c', 5, 1, 2)])
 299
 300         self.assertEqual(arvados.locators_and_ranges(blocks, 9, 15), [LocatorAndRange('a', 10, 9, 1),
 301                                                                       LocatorAndRange('b', 15, 0, 14)])
 302         self.assertEqual(arvados.locators_and_ranges(blocks, 10, 15), [LocatorAndRange('b', 15, 0, 15)])
 303         self.assertEqual(arvados.locators_and_ranges(blocks, 11, 15), [LocatorAndRange('b', 15, 1, 14),
 304                                                                        LocatorAndRange('c', 5, 0, 1)])
 305
 306     class MockKeep(object):
 307         def __init__(self, content, num_retries=0):
 308             self.content = content
 309
 310         def get(self, locator, num_retries=0):
 311             return self.content[locator]
 312
 313     def test_stream_reader(self):
 314         keepblocks = {'aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa+10': 'abcdefghij',
 315                       'bbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbb+15': 'klmnopqrstuvwxy',
 316                       'cccccccccccccccccccccccccccccccc+5': 'z0123'}
 317         mk = self.MockKeep(keepblocks)
 318
 319         sr = arvados.StreamReader([".", "aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa+10", "bbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbb+15", "cccccccccccccccccccccccccccccccc+5", "0:30:foo"], mk)
 320
 321         content = 'abcdefghijklmnopqrstuvwxyz0123456789'
 322
 323         self.assertEqual(sr.readfrom(0, 30), content[0:30])
 324         self.assertEqual(sr.readfrom(2, 30), content[2:30])
 325
 326         self.assertEqual(sr.readfrom(2, 8), content[2:10])
 327         self.assertEqual(sr.readfrom(0, 10), content[0:10])
 328
 329         self.assertEqual(sr.readfrom(0, 5), content[0:5])
 330         self.assertEqual(sr.readfrom(5, 5), content[5:10])
 331         self.assertEqual(sr.readfrom(10, 5), content[10:15])
 332         self.assertEqual(sr.readfrom(15, 5), content[15:20])
 333         self.assertEqual(sr.readfrom(20, 5), content[20:25])
 334         self.assertEqual(sr.readfrom(25, 5), content[25:30])
 335         self.assertEqual(sr.readfrom(30, 5), '')
 336
 337     def test_extract_file(self):
 338         m1 = """. 5348b82a029fd9e971a811ce1f71360b+43 0:43:md5sum.txt
 339 . 085c37f02916da1cad16f93c54d899b7+41 0:41:md6sum.txt
 340 . 8b22da26f9f433dea0a10e5ec66d73ba+43 0:43:md7sum.txt
 341 . 085c37f02916da1cad16f93c54d899b7+41 5348b82a029fd9e971a811ce1f71360b+43 8b22da26f9f433dea0a10e5ec66d73ba+43 47:80:md8sum.txt
 342 . 085c37f02916da1cad16f93c54d899b7+41 5348b82a029fd9e971a811ce1f71360b+43 8b22da26f9f433dea0a10e5ec66d73ba+43 40:80:md9sum.txt
 343 """
 344
 345         m2 = arvados.CollectionReader(m1, self.api_client).manifest_text(normalize=True)
 346
 347         self.assertEqual(m2,
 348                          ". 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")
 349         files = arvados.CollectionReader(
 350             m2, self.api_client).all_streams()[0].files()
 351
 352         self.assertEqual(files['md5sum.txt'].as_manifest(),
 353                          ". 5348b82a029fd9e971a811ce1f71360b+43 0:43:md5sum.txt\n")
 354         self.assertEqual(files['md6sum.txt'].as_manifest(),
 355                          ". 085c37f02916da1cad16f93c54d899b7+41 0:41:md6sum.txt\n")
 356         self.assertEqual(files['md7sum.txt'].as_manifest(),
 357                          ". 8b22da26f9f433dea0a10e5ec66d73ba+43 0:43:md7sum.txt\n")
 358         self.assertEqual(files['md9sum.txt'].as_manifest(),
 359                          ". 085c37f02916da1cad16f93c54d899b7+41 5348b82a029fd9e971a811ce1f71360b+43 8b22da26f9f433dea0a10e5ec66d73ba+43 40:80:md9sum.txt\n")
 360
 361     def test_write_directory_tree(self):
 362         cwriter = arvados.CollectionWriter(self.api_client)
 363         cwriter.write_directory_tree(self.build_directory_tree(
 364                 ['basefile', 'subdir/subfile']))
 365         self.assertEqual(cwriter.manifest_text(),
 366                          """. c5110c5ac93202d8e0f9e381f22bac0f+8 0:8:basefile
 367 ./subdir 1ca4dec89403084bf282ad31e6cf7972+14 0:14:subfile\n""")
 368
 369     def test_write_named_directory_tree(self):
 370         cwriter = arvados.CollectionWriter(self.api_client)
 371         cwriter.write_directory_tree(self.build_directory_tree(
 372                 ['basefile', 'subdir/subfile']), 'root')
 373         self.assertEqual(
 374             cwriter.manifest_text(),
 375             """./root c5110c5ac93202d8e0f9e381f22bac0f+8 0:8:basefile
 376 ./root/subdir 1ca4dec89403084bf282ad31e6cf7972+14 0:14:subfile\n""")
 377
 378     def test_write_directory_tree_in_one_stream(self):
 379         cwriter = arvados.CollectionWriter(self.api_client)
 380         cwriter.write_directory_tree(self.build_directory_tree(
 381                 ['basefile', 'subdir/subfile']), max_manifest_depth=0)
 382         self.assertEqual(cwriter.manifest_text(),
 383                          """. 4ace875ffdc6824a04950f06858f4465+22 0:8:basefile 8:14:subdir/subfile\n""")
 384
 385     def test_write_directory_tree_with_limited_recursion(self):
 386         cwriter = arvados.CollectionWriter(self.api_client)
 387         cwriter.write_directory_tree(
 388             self.build_directory_tree(['f1', 'd1/f2', 'd1/d2/f3']),
 389             max_manifest_depth=1)
 390         self.assertEqual(cwriter.manifest_text(),
 391                          """. bd19836ddb62c11c55ab251ccaca5645+2 0:2:f1
 392 ./d1 50170217e5b04312024aa5cd42934494+13 0:8:d2/f3 8:5:f2\n""")
 393
 394     def test_write_directory_tree_with_zero_recursion(self):
 395         cwriter = arvados.CollectionWriter(self.api_client)
 396         content = 'd1/d2/f3d1/f2f1'
 397         blockhash = tutil.str_keep_locator(content)
 398         cwriter.write_directory_tree(
 399             self.build_directory_tree(['f1', 'd1/f2', 'd1/d2/f3']),
 400             max_manifest_depth=0)
 401         self.assertEqual(
 402             cwriter.manifest_text(),
 403             ". {} 0:8:d1/d2/f3 8:5:d1/f2 13:2:f1\n".format(blockhash))
 404
 405     def test_write_one_file(self):
 406         cwriter = arvados.CollectionWriter(self.api_client)
 407         with self.make_test_file() as testfile:
 408             cwriter.write_file(testfile.name)
 409             self.assertEqual(
 410                 cwriter.manifest_text(),
 411                 ". 098f6bcd4621d373cade4e832627b4f6+4 0:4:{}\n".format(
 412                     os.path.basename(testfile.name)))
 413
 414     def test_write_named_file(self):
 415         cwriter = arvados.CollectionWriter(self.api_client)
 416         with self.make_test_file() as testfile:
 417             cwriter.write_file(testfile.name, 'foo')
 418             self.assertEqual(cwriter.manifest_text(),
 419                              ". 098f6bcd4621d373cade4e832627b4f6+4 0:4:foo\n")
 420
 421     def test_write_multiple_files(self):
 422         cwriter = arvados.CollectionWriter(self.api_client)
 423         for letter in 'ABC':
 424             with self.make_test_file(letter) as testfile:
 425                 cwriter.write_file(testfile.name, letter)
 426         self.assertEqual(
 427             cwriter.manifest_text(),
 428             ". 902fbdd2b1df0c4f70b4a5d23525e932+3 0:1:A 1:1:B 2:1:C\n")
 429
 430     def test_basic_resume(self):
 431         cwriter = TestResumableWriter()
 432         with self.make_test_file() as testfile:
 433             cwriter.write_file(testfile.name, 'test')
 434             resumed = TestResumableWriter.from_state(cwriter.current_state())
 435         self.assertEqual(cwriter.manifest_text(), resumed.manifest_text(),
 436                           "resumed CollectionWriter had different manifest")
 437
 438     def test_resume_fails_when_missing_dependency(self):
 439         cwriter = TestResumableWriter()
 440         with self.make_test_file() as testfile:
 441             cwriter.write_file(testfile.name, 'test')
 442         self.assertRaises(arvados.errors.StaleWriterStateError,
 443                           TestResumableWriter.from_state,
 444                           cwriter.current_state())
 445
 446     def test_resume_fails_when_dependency_mtime_changed(self):
 447         cwriter = TestResumableWriter()
 448         with self.make_test_file() as testfile:
 449             cwriter.write_file(testfile.name, 'test')
 450             os.utime(testfile.name, (0, 0))
 451             self.assertRaises(arvados.errors.StaleWriterStateError,
 452                               TestResumableWriter.from_state,
 453                               cwriter.current_state())
 454
 455     def test_resume_fails_when_dependency_is_nonfile(self):
 456         cwriter = TestResumableWriter()
 457         cwriter.write_file('/dev/null', 'empty')
 458         self.assertRaises(arvados.errors.StaleWriterStateError,
 459                           TestResumableWriter.from_state,
 460                           cwriter.current_state())
 461
 462     def test_resume_fails_when_dependency_size_changed(self):
 463         cwriter = TestResumableWriter()
 464         with self.make_test_file() as testfile:
 465             cwriter.write_file(testfile.name, 'test')
 466             orig_mtime = os.fstat(testfile.fileno()).st_mtime
 467             testfile.write('extra')
 468             testfile.flush()
 469             os.utime(testfile.name, (orig_mtime, orig_mtime))
 470             self.assertRaises(arvados.errors.StaleWriterStateError,
 471                               TestResumableWriter.from_state,
 472                               cwriter.current_state())
 473
 474     def test_resume_fails_with_expired_locator(self):
 475         cwriter = TestResumableWriter()
 476         state = cwriter.current_state()
 477         # Add an expired locator to the state.
 478         state['_current_stream_locators'].append(''.join([
 479                     'a' * 32, '+1+A', 'b' * 40, '@', '10000000']))
 480         self.assertRaises(arvados.errors.StaleWriterStateError,
 481                           TestResumableWriter.from_state, state)
 482
 483     def test_arbitrary_objects_not_resumable(self):
 484         cwriter = TestResumableWriter()
 485         with open('/dev/null') as badfile:
 486             self.assertRaises(arvados.errors.AssertionError,
 487                               cwriter.write_file, badfile)
 488
 489     def test_arbitrary_writes_not_resumable(self):
 490         cwriter = TestResumableWriter()
 491         self.assertRaises(arvados.errors.AssertionError,
 492                           cwriter.write, "badtext")
 493
 494     def test_read_arbitrary_data_with_collection_reader(self):
 495         # arv-get relies on this to do "arv-get {keep-locator} -".
 496         self.write_foo_bar_baz()
 497         self.assertEqual(
 498             'foobar',
 499             arvados.CollectionReader(
 500                 '3858f62230ac3c915f300c664312c63f+6'
 501                 ).manifest_text())
 502
 503
 504 class CollectionTestMixin(tutil.ApiClientMock):
 505     API_COLLECTIONS = run_test_server.fixture('collections')
 506     DEFAULT_COLLECTION = API_COLLECTIONS['foo_file']
 507     DEFAULT_DATA_HASH = DEFAULT_COLLECTION['portable_data_hash']
 508     DEFAULT_MANIFEST = DEFAULT_COLLECTION['manifest_text']
 509     DEFAULT_UUID = DEFAULT_COLLECTION['uuid']
 510     ALT_COLLECTION = API_COLLECTIONS['bar_file']
 511     ALT_DATA_HASH = ALT_COLLECTION['portable_data_hash']
 512     ALT_MANIFEST = ALT_COLLECTION['manifest_text']
 513
 514     def api_client_mock(self, status=200):
 515         client = super(CollectionTestMixin, self).api_client_mock()
 516         self.mock_keep_services(client, status=status, service_type='proxy', count=1)
 517         return client
 518
 519
 520 @tutil.skip_sleep
 521 class CollectionReaderTestCase(unittest.TestCase, CollectionTestMixin):
 522     def mock_get_collection(self, api_mock, code, body):
 523         body = self.API_COLLECTIONS.get(body)
 524         self._mock_api_call(api_mock.collections().get, code, body)
 525
 526     def api_client_mock(self, status=200):
 527         client = super(CollectionReaderTestCase, self).api_client_mock()
 528         self.mock_get_collection(client, status, 'foo_file')
 529         return client
 530
 531     def test_init_no_default_retries(self):
 532         client = self.api_client_mock(200)
 533         reader = arvados.CollectionReader(self.DEFAULT_UUID, api_client=client)
 534         reader.manifest_text()
 535         client.collections().get().execute.assert_called_with(num_retries=0)
 536
 537     def test_uuid_init_success(self):
 538         client = self.api_client_mock(200)
 539         reader = arvados.CollectionReader(self.DEFAULT_UUID, api_client=client,
 540                                           num_retries=3)
 541         self.assertEqual(self.DEFAULT_COLLECTION['manifest_text'],
 542                          reader.manifest_text())
 543         client.collections().get().execute.assert_called_with(num_retries=3)
 544
 545     def test_uuid_init_failure_raises_api_error(self):
 546         client = self.api_client_mock(500)
 547         with self.assertRaises(arvados.errors.ApiError):
 548             reader = arvados.CollectionReader(self.DEFAULT_UUID, api_client=client)
 549
 550     def test_locator_init(self):
 551         client = self.api_client_mock(200)
 552         # Ensure Keep will not return anything if asked.
 553         with tutil.mock_keep_responses(None, 404):
 554             reader = arvados.CollectionReader(self.DEFAULT_DATA_HASH,
 555                                               api_client=client)
 556             self.assertEqual(self.DEFAULT_MANIFEST, reader.manifest_text())
 557
 558     def test_locator_init_fallback_to_keep(self):
 559         # crunch-job needs this to read manifests that have only ever
 560         # been written to Keep.
 561         client = self.api_client_mock(200)
 562         self.mock_get_collection(client, 404, None)
 563         with tutil.mock_keep_responses(self.DEFAULT_MANIFEST, 200):
 564             reader = arvados.CollectionReader(self.DEFAULT_DATA_HASH,
 565                                               api_client=client)
 566             self.assertEqual(self.DEFAULT_MANIFEST, reader.manifest_text())
 567
 568     def test_uuid_init_no_fallback_to_keep(self):
 569         # Do not look up a collection UUID in Keep.
 570         client = self.api_client_mock(404)
 571         with tutil.mock_keep_responses(self.DEFAULT_MANIFEST, 200):
 572             with self.assertRaises(arvados.errors.ApiError):
 573                 reader = arvados.CollectionReader(self.DEFAULT_UUID,
 574                                                   api_client=client)
 575
 576     def test_try_keep_first_if_permission_hint(self):
 577         # To verify that CollectionReader tries Keep first here, we
 578         # mock API server to return the wrong data.
 579         client = self.api_client_mock(200)
 580         with tutil.mock_keep_responses(self.ALT_MANIFEST, 200):
 581             self.assertEqual(
 582                 self.ALT_MANIFEST,
 583                 arvados.CollectionReader(
 584                     self.ALT_DATA_HASH + '+Affffffffffffffffffffffffffffffffffffffff@fedcba98',
 585                     api_client=client).manifest_text())
 586
 587     def test_init_num_retries_propagated(self):
 588         # More of an integration test...
 589         client = self.api_client_mock(200)
 590         reader = arvados.CollectionReader(self.DEFAULT_UUID, api_client=client,
 591                                           num_retries=3)
 592         with tutil.mock_keep_responses('foo', 500, 500, 200):
 593             self.assertEqual('foo',
 594                              ''.join(f.read(9) for f in reader.all_files()))
 595
 596     def test_read_nonnormalized_manifest_with_collection_reader(self):
 597         # client should be able to use CollectionReader on a manifest without normalizing it
 598         client = self.api_client_mock(500)
 599         nonnormal = ". acbd18db4cc2f85cedef654fccc4a4d8+3+Aabadbadbee@abeebdee 0:3:foo.txt 1:0:bar.txt 0:3:foo.txt\n"
 600         reader = arvados.CollectionReader(
 601             nonnormal,
 602             api_client=client, num_retries=0)
 603         # Ensure stripped_manifest() doesn't mangle our manifest in
 604         # any way other than stripping hints.
 605         self.assertEqual(
 606             re.sub('\+[^\d\s\+]+', '', nonnormal),
 607             reader.stripped_manifest())
 608         # Ensure stripped_manifest() didn't mutate our reader.
 609         self.assertEqual(nonnormal, reader.manifest_text())
 610         # Ensure the files appear in the order given in the manifest.
 611         self.assertEqual(
 612             [[6, '.', 'foo.txt'],
 613              [0, '.', 'bar.txt']],
 614             [[f.size(), f.stream_name(), f.name()]
 615              for f in reader.all_streams()[0].all_files()])
 616
 617     def test_read_empty_collection(self):
 618         client = self.api_client_mock(200)
 619         self.mock_get_collection(client, 200, 'empty')
 620         reader = arvados.CollectionReader('d41d8cd98f00b204e9800998ecf8427e+0',
 621                                           api_client=client)
 622         self.assertEqual('', reader.manifest_text())
 623
 624     def test_api_response(self):
 625         client = self.api_client_mock()
 626         reader = arvados.CollectionReader(self.DEFAULT_UUID, api_client=client)
 627         self.assertEqual(self.DEFAULT_COLLECTION, reader.api_response())
 628
 629     def test_api_response_with_collection_from_keep(self):
 630         client = self.api_client_mock()
 631         self.mock_get_collection(client, 404, 'foo')
 632         with tutil.mock_keep_responses(self.DEFAULT_MANIFEST, 200):
 633             reader = arvados.CollectionReader(self.DEFAULT_DATA_HASH,
 634                                               api_client=client)
 635             api_response = reader.api_response()
 636         self.assertIsNone(api_response)
 637
 638     def check_open_file(self, coll_file, stream_name, file_name, file_size):
 639         self.assertFalse(coll_file.closed, "returned file is not open")
 640         self.assertEqual(stream_name, coll_file.stream_name())
 641         self.assertEqual(file_name, coll_file.name)
 642         self.assertEqual(file_size, coll_file.size())
 643
 644     def test_open_collection_file_one_argument(self):
 645         client = self.api_client_mock(200)
 646         reader = arvados.CollectionReader(self.DEFAULT_UUID, api_client=client)
 647         cfile = reader.open('./foo')
 648         self.check_open_file(cfile, '.', 'foo', 3)
 649
 650     def test_open_deep_file(self):
 651         coll_name = 'collection_with_files_in_subdir'
 652         client = self.api_client_mock(200)
 653         self.mock_get_collection(client, 200, coll_name)
 654         reader = arvados.CollectionReader(
 655             self.API_COLLECTIONS[coll_name]['uuid'], api_client=client)
 656         cfile = reader.open('./subdir2/subdir3/file2_in_subdir3.txt')
 657         self.check_open_file(cfile, './subdir2/subdir3', 'file2_in_subdir3.txt',
 658                              32)
 659
 660     def test_open_nonexistent_stream(self):
 661         client = self.api_client_mock(200)
 662         reader = arvados.CollectionReader(self.DEFAULT_UUID, api_client=client)
 663         self.assertRaises(IOError, reader.open, './nonexistent/foo')
 664
 665     def test_open_nonexistent_file(self):
 666         client = self.api_client_mock(200)
 667         reader = arvados.CollectionReader(self.DEFAULT_UUID, api_client=client)
 668         self.assertRaises(IOError, reader.open, 'nonexistent')
 669
 670
 671 @tutil.skip_sleep
 672 class CollectionWriterTestCase(unittest.TestCase, CollectionTestMixin):
 673     def mock_keep(self, body, *codes, **headers):
 674         headers.setdefault('x-keep-replicas-stored', 2)
 675         return tutil.mock_keep_responses(body, *codes, **headers)
 676
 677     def foo_writer(self, **kwargs):
 678         kwargs.setdefault('api_client', self.api_client_mock())
 679         writer = arvados.CollectionWriter(**kwargs)
 680         writer.start_new_file('foo')
 681         writer.write('foo')
 682         return writer
 683
 684     def test_write_whole_collection(self):
 685         writer = self.foo_writer()
 686         with self.mock_keep(self.DEFAULT_DATA_HASH, 200, 200):
 687             self.assertEqual(self.DEFAULT_DATA_HASH, writer.finish())
 688
 689     def test_write_no_default(self):
 690         writer = self.foo_writer()
 691         with self.mock_keep(None, 500):
 692             with self.assertRaises(arvados.errors.KeepWriteError):
 693                 writer.finish()
 694
 695     def test_write_insufficient_replicas_via_proxy(self):
 696         writer = self.foo_writer(replication=3)
 697         with self.mock_keep(None, 200, **{'x-keep-replicas-stored': 2}):
 698             with self.assertRaises(arvados.errors.KeepWriteError):
 699                 writer.manifest_text()
 700
 701     def test_write_insufficient_replicas_via_disks(self):
 702         client = mock.MagicMock(name='api_client')
 703         with self.mock_keep(
 704                 None, 200, 200,
 705                 **{'x-keep-replicas-stored': 1}) as keepmock:
 706             self.mock_keep_services(client, status=200, service_type='disk', count=2)
 707             writer = self.foo_writer(api_client=client, replication=3)
 708             with self.assertRaises(arvados.errors.KeepWriteError):
 709                 writer.manifest_text()
 710
 711     def test_write_three_replicas(self):
 712         client = mock.MagicMock(name='api_client')
 713         with self.mock_keep(
 714                 "", 500, 500, 500, 200, 200, 200,
 715                 **{'x-keep-replicas-stored': 1}) as keepmock:
 716             self.mock_keep_services(client, status=200, service_type='disk', count=6)
 717             writer = self.foo_writer(api_client=client, replication=3)
 718             writer.manifest_text()
 719             self.assertEqual(6, keepmock.call_count)
 720
 721     def test_write_whole_collection_through_retries(self):
 722         writer = self.foo_writer(num_retries=2)
 723         with self.mock_keep(self.DEFAULT_DATA_HASH,
 724                             500, 500, 200, 500, 500, 200):
 725             self.assertEqual(self.DEFAULT_DATA_HASH, writer.finish())
 726
 727     def test_flush_data_retries(self):
 728         writer = self.foo_writer(num_retries=2)
 729         foo_hash = self.DEFAULT_MANIFEST.split()[1]
 730         with self.mock_keep(foo_hash, 500, 200):
 731             writer.flush_data()
 732         self.assertEqual(self.DEFAULT_MANIFEST, writer.manifest_text())
 733
 734     def test_one_open(self):
 735         client = self.api_client_mock()
 736         writer = arvados.CollectionWriter(client)
 737         with writer.open('out') as out_file:
 738             self.assertEqual('.', writer.current_stream_name())
 739             self.assertEqual('out', writer.current_file_name())
 740             out_file.write('test data')
 741             data_loc = tutil.str_keep_locator('test data')
 742         self.assertTrue(out_file.closed, "writer file not closed after context")
 743         self.assertRaises(ValueError, out_file.write, 'extra text')
 744         with self.mock_keep(data_loc, 200) as keep_mock:
 745             self.assertEqual(". {} 0:9:out\n".format(data_loc),
 746                              writer.manifest_text())
 747
 748     def test_open_writelines(self):
 749         client = self.api_client_mock()
 750         writer = arvados.CollectionWriter(client)
 751         with writer.open('six') as out_file:
 752             out_file.writelines(['12', '34', '56'])
 753             data_loc = tutil.str_keep_locator('123456')
 754         with self.mock_keep(data_loc, 200) as keep_mock:
 755             self.assertEqual(". {} 0:6:six\n".format(data_loc),
 756                              writer.manifest_text())
 757
 758     def test_open_flush(self):
 759         client = self.api_client_mock()
 760         data_loc1 = tutil.str_keep_locator('flush1')
 761         data_loc2 = tutil.str_keep_locator('flush2')
 762         with self.mock_keep((data_loc1, 200), (data_loc2, 200)) as keep_mock:
 763             writer = arvados.CollectionWriter(client)
 764             with writer.open('flush_test') as out_file:
 765                 out_file.write('flush1')
 766                 out_file.flush()
 767                 out_file.write('flush2')
 768             self.assertEqual(". {} {} 0:12:flush_test\n".format(data_loc1,
 769                                                                 data_loc2),
 770                              writer.manifest_text())
 771
 772     def test_two_opens_same_stream(self):
 773         client = self.api_client_mock()
 774         writer = arvados.CollectionWriter(client)
 775         with writer.open('.', '1') as out_file:
 776             out_file.write('1st')
 777         with writer.open('.', '2') as out_file:
 778             out_file.write('2nd')
 779         data_loc = tutil.str_keep_locator('1st2nd')
 780         with self.mock_keep(data_loc, 200) as keep_mock:
 781             self.assertEqual(". {} 0:3:1 3:3:2\n".format(data_loc),
 782                              writer.manifest_text())
 783
 784     def test_two_opens_two_streams(self):
 785         client = self.api_client_mock()
 786         data_loc1 = tutil.str_keep_locator('file')
 787         data_loc2 = tutil.str_keep_locator('indir')
 788         with self.mock_keep((data_loc1, 200), (data_loc2, 200)) as keep_mock:
 789             writer = arvados.CollectionWriter(client)
 790             with writer.open('file') as out_file:
 791                 out_file.write('file')
 792             with writer.open('./dir', 'indir') as out_file:
 793                 out_file.write('indir')
 794             expected = ". {} 0:4:file\n./dir {} 0:5:indir\n".format(
 795                 data_loc1, data_loc2)
 796             self.assertEqual(expected, writer.manifest_text())
 797
 798     def test_dup_open_fails(self):
 799         client = self.api_client_mock()
 800         writer = arvados.CollectionWriter(client)
 801         file1 = writer.open('one')
 802         self.assertRaises(arvados.errors.AssertionError, writer.open, 'two')
 803
 804
 805 class NewCollectionTestCase(unittest.TestCase, CollectionTestMixin):
 806
 807     def test_replication_desired_kept_on_load(self):
 808         m = '. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt 0:10:count2.txt\n'
 809         c1 = Collection(m, replication_desired=1)
 810         c1.save_new()
 811         loc = c1.manifest_locator()
 812         c2 = Collection(loc)
 813         self.assertEqual(c1.manifest_text, c2.manifest_text)
 814         self.assertEqual(c1.replication_desired, c2.replication_desired)
 815
 816     def test_replication_desired_not_loaded_if_provided(self):
 817         m = '. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt 0:10:count2.txt\n'
 818         c1 = Collection(m, replication_desired=1)
 819         c1.save_new()
 820         loc = c1.manifest_locator()
 821         c2 = Collection(loc, replication_desired=2)
 822         self.assertEqual(c1.manifest_text, c2.manifest_text)
 823         self.assertNotEqual(c1.replication_desired, c2.replication_desired)
 824
 825     def test_init_manifest(self):
 826         m1 = """. 5348b82a029fd9e971a811ce1f71360b+43 0:43:md5sum.txt
 827 . 085c37f02916da1cad16f93c54d899b7+41 0:41:md5sum.txt
 828 . 8b22da26f9f433dea0a10e5ec66d73ba+43 0:43:md5sum.txt
 829 """
 830         self.assertEqual(m1, CollectionReader(m1).manifest_text(normalize=False))
 831         self.assertEqual(". 5348b82a029fd9e971a811ce1f71360b+43 085c37f02916da1cad16f93c54d899b7+41 8b22da26f9f433dea0a10e5ec66d73ba+43 0:127:md5sum.txt\n", CollectionReader(m1).manifest_text(normalize=True))
 832
 833     def test_init_manifest_with_collision(self):
 834         m1 = """. 5348b82a029fd9e971a811ce1f71360b+43 0:43:md5sum.txt
 835 ./md5sum.txt 085c37f02916da1cad16f93c54d899b7+41 0:41:md5sum.txt
 836 """
 837         with self.assertRaises(arvados.errors.ArgumentError):
 838             self.assertEqual(m1, CollectionReader(m1))
 839
 840     def test_init_manifest_with_error(self):
 841         m1 = """. 0:43:md5sum.txt"""
 842         with self.assertRaises(arvados.errors.ArgumentError):
 843             self.assertEqual(m1, CollectionReader(m1))
 844
 845     def test_remove(self):
 846         c = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt 0:10:count2.txt\n')
 847         self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt 0:10:count2.txt\n", c.portable_manifest_text())
 848         self.assertIn("count1.txt", c)
 849         c.remove("count1.txt")
 850         self.assertNotIn("count1.txt", c)
 851         self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count2.txt\n", c.portable_manifest_text())
 852         with self.assertRaises(arvados.errors.ArgumentError):
 853             c.remove("")
 854
 855     def test_find(self):
 856         c = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt 0:10:count2.txt\n')
 857         self.assertIs(c.find("."), c)
 858         self.assertIs(c.find("./count1.txt"), c["count1.txt"])
 859         self.assertIs(c.find("count1.txt"), c["count1.txt"])
 860         with self.assertRaises(IOError):
 861             c.find("/.")
 862         with self.assertRaises(arvados.errors.ArgumentError):
 863             c.find("")
 864
 865     def test_remove_in_subdir(self):
 866         c = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n./foo 781e5e245d69b566979b86e28d23f2c7+10 0:10:count2.txt\n')
 867         c.remove("foo/count2.txt")
 868         self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n", c.portable_manifest_text())
 869
 870     def test_remove_empty_subdir(self):
 871         c = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n./foo 781e5e245d69b566979b86e28d23f2c7+10 0:10:count2.txt\n')
 872         c.remove("foo/count2.txt")
 873         c.remove("foo")
 874         self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n", c.portable_manifest_text())
 875
 876     def test_remove_nonempty_subdir(self):
 877         c = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n./foo 781e5e245d69b566979b86e28d23f2c7+10 0:10:count2.txt\n')
 878         with self.assertRaises(IOError):
 879             c.remove("foo")
 880         c.remove("foo", recursive=True)
 881         self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n", c.portable_manifest_text())
 882
 883     def test_copy_to_file_in_dir(self):
 884         c = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n')
 885         c.copy("count1.txt", "foo/count2.txt")
 886         self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n./foo 781e5e245d69b566979b86e28d23f2c7+10 0:10:count2.txt\n", c.portable_manifest_text())
 887
 888     def test_copy_file(self):
 889         c = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n')
 890         c.copy("count1.txt", "count2.txt")
 891         self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt 0:10:count2.txt\n", c.portable_manifest_text())
 892
 893     def test_copy_to_existing_dir(self):
 894         c = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n./foo 781e5e245d69b566979b86e28d23f2c7+10 0:10:count2.txt\n')
 895         c.copy("count1.txt", "foo")
 896         self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n./foo 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt 0:10:count2.txt\n", c.portable_manifest_text())
 897
 898     def test_copy_to_new_dir(self):
 899         c = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n')
 900         c.copy("count1.txt", "foo/")
 901         self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n./foo 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n", c.portable_manifest_text())
 902
 903     def test_rename_file(self):
 904         c = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n')
 905         c.rename("count1.txt", "count2.txt")
 906         self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count2.txt\n", c.manifest_text())
 907
 908     def test_move_file_to_dir(self):
 909         c = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n')
 910         c.mkdirs("foo")
 911         c.rename("count1.txt", "foo/count2.txt")
 912         self.assertEqual("./foo 781e5e245d69b566979b86e28d23f2c7+10 0:10:count2.txt\n", c.manifest_text())
 913
 914     def test_move_file_to_other(self):
 915         c1 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n')
 916         c2 = Collection()
 917         c2.rename("count1.txt", "count2.txt", source_collection=c1)
 918         self.assertEqual("", c1.manifest_text())
 919         self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count2.txt\n", c2.manifest_text())
 920
 921     def test_clone(self):
 922         c = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n./foo 781e5e245d69b566979b86e28d23f2c7+10 0:10:count2.txt\n')
 923         cl = c.clone()
 924         self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n./foo 781e5e245d69b566979b86e28d23f2c7+10 0:10:count2.txt\n", cl.portable_manifest_text())
 925
 926     def test_diff_del_add(self):
 927         c1 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n')
 928         c2 = Collection('. 5348b82a029fd9e971a811ce1f71360b+43 0:10:count2.txt\n')
 929         d = c2.diff(c1)
 930         self.assertEqual(d, [('del', './count2.txt', c2["count2.txt"]),
 931                              ('add', './count1.txt', c1["count1.txt"])])
 932         d = c1.diff(c2)
 933         self.assertEqual(d, [('del', './count1.txt', c1["count1.txt"]),
 934                              ('add', './count2.txt', c2["count2.txt"])])
 935         self.assertNotEqual(c1.portable_manifest_text(), c2.portable_manifest_text())
 936         c1.apply(d)
 937         self.assertEqual(c1.portable_manifest_text(), c2.portable_manifest_text())
 938
 939     def test_diff_same(self):
 940         c1 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n')
 941         c2 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n')
 942         d = c2.diff(c1)
 943         self.assertEqual(d, [('tok', './count1.txt', c2["count1.txt"], c1["count1.txt"])])
 944         d = c1.diff(c2)
 945         self.assertEqual(d, [('tok', './count1.txt', c2["count1.txt"], c1["count1.txt"])])
 946
 947         self.assertEqual(c1.portable_manifest_text(), c2.portable_manifest_text())
 948         c1.apply(d)
 949         self.assertEqual(c1.portable_manifest_text(), c2.portable_manifest_text())
 950
 951     def test_diff_mod(self):
 952         c1 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n')
 953         c2 = Collection('. 5348b82a029fd9e971a811ce1f71360b+43 0:10:count1.txt\n')
 954         d = c2.diff(c1)
 955         self.assertEqual(d, [('mod', './count1.txt', c2["count1.txt"], c1["count1.txt"])])
 956         d = c1.diff(c2)
 957         self.assertEqual(d, [('mod', './count1.txt', c1["count1.txt"], c2["count1.txt"])])
 958
 959         self.assertNotEqual(c1.portable_manifest_text(), c2.portable_manifest_text())
 960         c1.apply(d)
 961         self.assertEqual(c1.portable_manifest_text(), c2.portable_manifest_text())
 962
 963     def test_diff_add(self):
 964         c1 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n')
 965         c2 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 5348b82a029fd9e971a811ce1f71360b+43 0:10:count1.txt 10:20:count2.txt\n')
 966         d = c2.diff(c1)
 967         self.assertEqual(d, [('del', './count2.txt', c2["count2.txt"]),
 968                              ('tok', './count1.txt', c2["count1.txt"], c1["count1.txt"])])
 969         d = c1.diff(c2)
 970         self.assertEqual(d, [('add', './count2.txt', c2["count2.txt"]),
 971                              ('tok', './count1.txt', c2["count1.txt"], c1["count1.txt"])])
 972
 973         self.assertNotEqual(c1.portable_manifest_text(), c2.portable_manifest_text())
 974         c1.apply(d)
 975         self.assertEqual(c1.portable_manifest_text(), c2.portable_manifest_text())
 976
 977     def test_diff_add_in_subcollection(self):
 978         c1 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n')
 979         c2 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n./foo 5348b82a029fd9e971a811ce1f71360b+43 0:10:count2.txt\n')
 980         d = c2.diff(c1)
 981         self.assertEqual(d, [('del', './foo', c2["foo"]),
 982                              ('tok', './count1.txt', c2["count1.txt"], c1["count1.txt"])])
 983         d = c1.diff(c2)
 984         self.assertEqual(d, [('add', './foo', c2["foo"]),
 985                              ('tok', './count1.txt', c2["count1.txt"], c1["count1.txt"])])
 986
 987         self.assertNotEqual(c1.portable_manifest_text(), c2.portable_manifest_text())
 988         c1.apply(d)
 989         self.assertEqual(c1.portable_manifest_text(), c2.portable_manifest_text())
 990
 991     def test_diff_del_add_in_subcollection(self):
 992         c1 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n./foo 5348b82a029fd9e971a811ce1f71360b+43 0:10:count2.txt\n')
 993         c2 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n./foo 5348b82a029fd9e971a811ce1f71360b+43 0:3:count3.txt\n')
 994
 995         d = c2.diff(c1)
 996         self.assertEqual(d, [('del', './foo/count3.txt', c2.find("foo/count3.txt")),
 997                              ('add', './foo/count2.txt', c1.find("foo/count2.txt")),
 998                              ('tok', './count1.txt', c2["count1.txt"], c1["count1.txt"])])
 999         d = c1.diff(c2)
1000         self.assertEqual(d, [('del', './foo/count2.txt', c1.find("foo/count2.txt")),
1001                              ('add', './foo/count3.txt', c2.find("foo/count3.txt")),
1002                              ('tok', './count1.txt', c2["count1.txt"], c1["count1.txt"])])
1003
1004         self.assertNotEqual(c1.portable_manifest_text(), c2.portable_manifest_text())
1005         c1.apply(d)
1006         self.assertEqual(c1.portable_manifest_text(), c2.portable_manifest_text())
1007
1008     def test_diff_mod_in_subcollection(self):
1009         c1 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n./foo 5348b82a029fd9e971a811ce1f71360b+43 0:10:count2.txt\n')
1010         c2 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt 0:3:foo\n')
1011         d = c2.diff(c1)
1012         self.assertEqual(d, [('mod', './foo', c2["foo"], c1["foo"]),
1013                              ('tok', './count1.txt', c2["count1.txt"], c1["count1.txt"])])
1014         d = c1.diff(c2)
1015         self.assertEqual(d, [('mod', './foo', c1["foo"], c2["foo"]),
1016                              ('tok', './count1.txt', c2["count1.txt"], c1["count1.txt"])])
1017
1018         self.assertNotEqual(c1.portable_manifest_text(), c2.portable_manifest_text())
1019         c1.apply(d)
1020         self.assertEqual(c1.portable_manifest_text(), c2.portable_manifest_text())
1021
1022     def test_conflict_keep_local_change(self):
1023         c1 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n')
1024         c2 = Collection('. 5348b82a029fd9e971a811ce1f71360b+43 0:10:count2.txt\n')
1025         d = c1.diff(c2)
1026         self.assertEqual(d, [('del', './count1.txt', c1["count1.txt"]),
1027                              ('add', './count2.txt', c2["count2.txt"])])
1028         f = c1.open("count1.txt", "w")
1029         f.write("zzzzz")
1030
1031         # c1 changed, so it should not be deleted.
1032         c1.apply(d)
1033         self.assertEqual(c1.portable_manifest_text(), ". 95ebc3c7b3b9f1d2c40fec14415d3cb8+5 5348b82a029fd9e971a811ce1f71360b+43 0:5:count1.txt 5:10:count2.txt\n")
1034
1035     def test_conflict_mod(self):
1036         c1 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt')
1037         c2 = Collection('. 5348b82a029fd9e971a811ce1f71360b+43 0:10:count1.txt')
1038         d = c1.diff(c2)
1039         self.assertEqual(d, [('mod', './count1.txt', c1["count1.txt"], c2["count1.txt"])])
1040         f = c1.open("count1.txt", "w")
1041         f.write("zzzzz")
1042
1043         # c1 changed, so c2 mod will go to a conflict file
1044         c1.apply(d)
1045         self.assertRegexpMatches(c1.portable_manifest_text(), r"\. 95ebc3c7b3b9f1d2c40fec14415d3cb8\+5 5348b82a029fd9e971a811ce1f71360b\+43 0:5:count1\.txt 5:10:count1\.txt~\d\d\d\d\d\d\d\d-\d\d\d\d\d\d~conflict~$")
1046
1047     def test_conflict_add(self):
1048         c1 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count2.txt\n')
1049         c2 = Collection('. 5348b82a029fd9e971a811ce1f71360b+43 0:10:count1.txt\n')
1050         d = c1.diff(c2)
1051         self.assertEqual(d, [('del', './count2.txt', c1["count2.txt"]),
1052                              ('add', './count1.txt', c2["count1.txt"])])
1053         f = c1.open("count1.txt", "w")
1054         f.write("zzzzz")
1055
1056         # c1 added count1.txt, so c2 add will go to a conflict file
1057         c1.apply(d)
1058         self.assertRegexpMatches(c1.portable_manifest_text(), r"\. 95ebc3c7b3b9f1d2c40fec14415d3cb8\+5 5348b82a029fd9e971a811ce1f71360b\+43 0:5:count1\.txt 5:10:count1\.txt~\d\d\d\d\d\d\d\d-\d\d\d\d\d\d~conflict~$")
1059
1060     def test_conflict_del(self):
1061         c1 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt')
1062         c2 = Collection('. 5348b82a029fd9e971a811ce1f71360b+43 0:10:count1.txt')
1063         d = c1.diff(c2)
1064         self.assertEqual(d, [('mod', './count1.txt', c1["count1.txt"], c2["count1.txt"])])
1065         c1.remove("count1.txt")
1066
1067         # c1 deleted, so c2 mod will go to a conflict file
1068         c1.apply(d)
1069         self.assertRegexpMatches(c1.portable_manifest_text(), r"\. 5348b82a029fd9e971a811ce1f71360b\+43 0:10:count1\.txt~\d\d\d\d\d\d\d\d-\d\d\d\d\d\d~conflict~$")
1070
1071     def test_notify(self):
1072         c1 = Collection()
1073         events = []
1074         c1.subscribe(lambda event, collection, name, item: events.append((event, collection, name, item)))
1075         f = c1.open("foo.txt", "w")
1076         self.assertEqual(events[0], (arvados.collection.ADD, c1, "foo.txt", f.arvadosfile))
1077
1078     def test_open_w(self):
1079         c1 = Collection(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n")
1080         self.assertEqual(c1["count1.txt"].size(), 10)
1081         c1.open("count1.txt", "w").close()
1082         self.assertEqual(c1["count1.txt"].size(), 0)
1083
1084
1085 class NewCollectionTestCaseWithServers(run_test_server.TestCaseWithServers):
1086     def test_get_manifest_text_only_committed(self):
1087         c = Collection()
1088         with c.open("count.txt", "w") as f:
1089             # One file committed
1090             with c.open("foo.txt", "w") as foo:
1091                 foo.write("foo")
1092                 foo.flush() # Force block commit
1093             f.write("0123456789")
1094             # Other file not committed. Block not written to keep yet.
1095             self.assertEqual(
1096                 c._get_manifest_text(".",
1097                                      strip=False,
1098                                      normalize=False,
1099                                      only_committed=True),
1100                 '. acbd18db4cc2f85cedef654fccc4a4d8+3 0:0:count.txt 0:3:foo.txt\n')
1101             # And now with the file closed...
1102             f.flush() # Force block commit
1103         self.assertEqual(
1104             c._get_manifest_text(".",
1105                                  strip=False,
1106                                  normalize=False,
1107                                  only_committed=True),
1108             ". 781e5e245d69b566979b86e28d23f2c7+10 acbd18db4cc2f85cedef654fccc4a4d8+3 0:10:count.txt 10:3:foo.txt\n")
1109
1110     def test_only_small_blocks_are_packed_together(self):
1111         c = Collection()
1112         # Write a couple of small files,
1113         f = c.open("count.txt", "w")
1114         f.write("0123456789")
1115         f.close(flush=False)
1116         foo = c.open("foo.txt", "w")
1117         foo.write("foo")
1118         foo.close(flush=False)
1119         # Then, write a big file, it shouldn't be packed with the ones above
1120         big = c.open("bigfile.txt", "w")
1121         big.write("x" * 1024 * 1024 * 33) # 33 MB > KEEP_BLOCK_SIZE/2
1122         big.close(flush=False)
1123         self.assertEqual(
1124             c.manifest_text("."),
1125             '. 2d303c138c118af809f39319e5d507e9+34603008 a8430a058b8fbf408e1931b794dbd6fb+13 0:34603008:bigfile.txt 34603008:10:count.txt 34603018:3:foo.txt\n')
1126
1127
1128 class CollectionCreateUpdateTest(run_test_server.TestCaseWithServers):
1129     MAIN_SERVER = {}
1130     KEEP_SERVER = {}
1131
1132     def create_count_txt(self):
1133         # Create an empty collection, save it to the API server, then write a
1134         # file, but don't save it.
1135
1136         c = Collection()
1137         c.save_new("CollectionCreateUpdateTest", ensure_unique_name=True)
1138         self.assertEqual(c.portable_data_hash(), "d41d8cd98f00b204e9800998ecf8427e+0")
1139         self.assertEqual(c.api_response()["portable_data_hash"], "d41d8cd98f00b204e9800998ecf8427e+0" )
1140
1141         with c.open("count.txt", "w") as f:
1142             f.write("0123456789")
1143
1144         self.assertEqual(c.portable_manifest_text(), ". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n")
1145
1146         return c
1147
1148     def test_create_and_save(self):
1149         c = self.create_count_txt()
1150         c.save()
1151         self.assertRegexpMatches(c.manifest_text(), r"^\. 781e5e245d69b566979b86e28d23f2c7\+10\+A[a-f0-9]{40}@[a-f0-9]{8} 0:10:count\.txt$",)
1152
1153     def test_create_and_save_new(self):
1154         c = self.create_count_txt()
1155         c.save_new()
1156         self.assertRegexpMatches(c.manifest_text(), r"^\. 781e5e245d69b566979b86e28d23f2c7\+10\+A[a-f0-9]{40}@[a-f0-9]{8} 0:10:count\.txt$",)
1157
1158     def test_create_diff_apply(self):
1159         c1 = self.create_count_txt()
1160         c1.save()
1161
1162         c2 = Collection(c1.manifest_locator())
1163         with c2.open("count.txt", "w") as f:
1164             f.write("abcdefg")
1165
1166         diff = c1.diff(c2)
1167
1168         self.assertEqual(diff[0], (arvados.collection.MOD, u'./count.txt', c1["count.txt"], c2["count.txt"]))
1169
1170         c1.apply(diff)
1171         self.assertEqual(c1.portable_data_hash(), c2.portable_data_hash())
1172
1173     def test_diff_apply_with_token(self):
1174         baseline = CollectionReader(". 781e5e245d69b566979b86e28d23f2c7+10+A715fd31f8111894f717eb1003c1b0216799dd9ec@54f5dd1a 0:10:count.txt\n")
1175         c = Collection(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n")
1176         other = CollectionReader(". 7ac66c0f148de9519b8bd264312c4d64+7+A715fd31f8111894f717eb1003c1b0216799dd9ec@54f5dd1a 0:7:count.txt\n")
1177
1178         diff = baseline.diff(other)
1179         self.assertEqual(diff, [('mod', u'./count.txt', c["count.txt"], other["count.txt"])])
1180
1181         c.apply(diff)
1182
1183         self.assertEqual(c.manifest_text(), ". 7ac66c0f148de9519b8bd264312c4d64+7+A715fd31f8111894f717eb1003c1b0216799dd9ec@54f5dd1a 0:7:count.txt\n")
1184
1185
1186     def test_create_and_update(self):
1187         c1 = self.create_count_txt()
1188         c1.save()
1189
1190         c2 = arvados.collection.Collection(c1.manifest_locator())
1191         with c2.open("count.txt", "w") as f:
1192             f.write("abcdefg")
1193
1194         c2.save()
1195
1196         self.assertNotEqual(c1.portable_data_hash(), c2.portable_data_hash())
1197         c1.update()
1198         self.assertEqual(c1.portable_data_hash(), c2.portable_data_hash())
1199
1200
1201     def test_create_and_update_with_conflict(self):
1202         c1 = self.create_count_txt()
1203         c1.save()
1204
1205         with c1.open("count.txt", "w") as f:
1206             f.write("XYZ")
1207
1208         c2 = arvados.collection.Collection(c1.manifest_locator())
1209         with c2.open("count.txt", "w") as f:
1210             f.write("abcdefg")
1211
1212         c2.save()
1213
1214         c1.update()
1215         self.assertRegexpMatches(c1.manifest_text(), r"\. e65075d550f9b5bf9992fa1d71a131be\+3\S* 7ac66c0f148de9519b8bd264312c4d64\+7\S* 0:3:count\.txt 3:7:count\.txt~\d\d\d\d\d\d\d\d-\d\d\d\d\d\d~conflict~$")
1216
1217
1218 if __name__ == '__main__':
1219     unittest.main()