sdk/python/tests/test_collections.py

   1 from __future__ import absolute_import
   2 # usage example:
   3 #
   4 # ARVADOS_API_TOKEN=abc ARVADOS_API_HOST=arvados.local python -m unittest discover
   5
   6 from builtins import object
   7 import arvados
   8 import copy
   9 import mock
  10 import os
  11 import pprint
  12 import re
  13 import tempfile
  14 import unittest
  15
  16 from . import run_test_server
  17 from arvados._ranges import Range, LocatorAndRange
  18 from arvados.collection import Collection, CollectionReader
  19 from . import arvados_testutil as tutil
  20
  21 class TestResumableWriter(arvados.ResumableCollectionWriter):
  22     KEEP_BLOCK_SIZE = 1024  # PUT to Keep every 1K.
  23
  24     def current_state(self):
  25         return self.dump_state(copy.deepcopy)
  26
  27
  28 class ArvadosCollectionsTest(run_test_server.TestCaseWithServers,
  29                              tutil.ArvadosBaseTestCase):
  30     MAIN_SERVER = {}
  31
  32     @classmethod
  33     def setUpClass(cls):
  34         super(ArvadosCollectionsTest, cls).setUpClass()
  35         run_test_server.authorize_with('active')
  36         cls.api_client = arvados.api('v1')
  37         cls.keep_client = arvados.KeepClient(api_client=cls.api_client,
  38                                              local_store=cls.local_store)
  39
  40     def write_foo_bar_baz(self):
  41         cw = arvados.CollectionWriter(self.api_client)
  42         self.assertEqual(cw.current_stream_name(), '.',
  43                          'current_stream_name() should be "." now')
  44         cw.set_current_file_name('foo.txt')
  45         cw.write(b'foo')
  46         self.assertEqual(cw.current_file_name(), 'foo.txt',
  47                          'current_file_name() should be foo.txt now')
  48         cw.start_new_file('bar.txt')
  49         cw.write(b'bar')
  50         cw.start_new_stream('baz')
  51         cw.write(b'baz')
  52         cw.set_current_file_name('baz.txt')
  53         self.assertEqual(cw.manifest_text(),
  54                          ". 3858f62230ac3c915f300c664312c63f+6 0:3:foo.txt 3:3:bar.txt\n" +
  55                          "./baz 73feffa4b7f6bb68e44cf984c85f6e88+3 0:3:baz.txt\n",
  56                          "wrong manifest: got {}".format(cw.manifest_text()))
  57         cw.finish()
  58         return cw.portable_data_hash()
  59
  60     def test_keep_local_store(self):
  61         self.assertEqual(self.keep_client.put(b'foo'), 'acbd18db4cc2f85cedef654fccc4a4d8+3', 'wrong md5 hash from Keep.put')
  62         self.assertEqual(self.keep_client.get('acbd18db4cc2f85cedef654fccc4a4d8+3'), b'foo', 'wrong data from Keep.get')
  63
  64     def test_local_collection_writer(self):
  65         self.assertEqual(self.write_foo_bar_baz(),
  66                          '23ca013983d6239e98931cc779e68426+114',
  67                          'wrong locator hash: ' + self.write_foo_bar_baz())
  68
  69     def test_local_collection_reader(self):
  70         foobarbaz = self.write_foo_bar_baz()
  71         cr = arvados.CollectionReader(
  72             foobarbaz + '+Xzizzle', self.api_client)
  73         got = []
  74         for s in cr.all_streams():
  75             for f in s.all_files():
  76                 got += [[f.size(), f.stream_name(), f.name(), f.read(2**26)]]
  77         expected = [[3, '.', 'foo.txt', b'foo'],
  78                     [3, '.', 'bar.txt', b'bar'],
  79                     [3, './baz', 'baz.txt', b'baz']]
  80         self.assertEqual(got,
  81                          expected)
  82         stream0 = cr.all_streams()[0]
  83         self.assertEqual(stream0.readfrom(0, 0),
  84                          b'',
  85                          'reading zero bytes should have returned empty string')
  86         self.assertEqual(stream0.readfrom(0, 2**26),
  87                          b'foobar',
  88                          'reading entire stream failed')
  89         self.assertEqual(stream0.readfrom(2**26, 0),
  90                          b'',
  91                          'reading zero bytes should have returned empty string')
  92
  93     def _test_subset(self, collection, expected):
  94         cr = arvados.CollectionReader(collection, self.api_client)
  95         for s in cr.all_streams():
  96             for ex in expected:
  97                 if ex[0] == s:
  98                     f = s.files()[ex[2]]
  99                     got = [f.size(), f.stream_name(), f.name(), "".join(f.readall(2**26))]
 100                     self.assertEqual(got,
 101                                      ex,
 102                                      'all_files|as_manifest did not preserve manifest contents: got %s expected %s' % (got, ex))
 103
 104     def test_collection_manifest_subset(self):
 105         foobarbaz = self.write_foo_bar_baz()
 106         self._test_subset(foobarbaz,
 107                           [[3, '.',     'bar.txt', b'bar'],
 108                            [3, '.',     'foo.txt', b'foo'],
 109                            [3, './baz', 'baz.txt', b'baz']])
 110         self._test_subset((". %s %s 0:3:foo.txt 3:3:bar.txt\n" %
 111                            (self.keep_client.put(b"foo"),
 112                             self.keep_client.put(b"bar"))),
 113                           [[3, '.', 'bar.txt', b'bar'],
 114                            [3, '.', 'foo.txt', b'foo']])
 115         self._test_subset((". %s %s 0:2:fo.txt 2:4:obar.txt\n" %
 116                            (self.keep_client.put(b"foo"),
 117                             self.keep_client.put(b"bar"))),
 118                           [[2, '.', 'fo.txt', b'fo'],
 119                            [4, '.', 'obar.txt', b'obar']])
 120         self._test_subset((". %s %s 0:2:fo.txt 2:0:zero.txt 2:2:ob.txt 4:2:ar.txt\n" %
 121                            (self.keep_client.put(b"foo"),
 122                             self.keep_client.put(b"bar"))),
 123                           [[2, '.', 'ar.txt', b'ar'],
 124                            [2, '.', 'fo.txt', b'fo'],
 125                            [2, '.', 'ob.txt', b'ob'],
 126                            [0, '.', 'zero.txt', b'']])
 127
 128     def test_collection_empty_file(self):
 129         cw = arvados.CollectionWriter(self.api_client)
 130         cw.start_new_file('zero.txt')
 131         cw.write(b'')
 132
 133         self.assertEqual(cw.manifest_text(), ". d41d8cd98f00b204e9800998ecf8427e+0 0:0:zero.txt\n")
 134         self.check_manifest_file_sizes(cw.manifest_text(), [0])
 135         cw = arvados.CollectionWriter(self.api_client)
 136         cw.start_new_file('zero.txt')
 137         cw.write(b'')
 138         cw.start_new_file('one.txt')
 139         cw.write(b'1')
 140         cw.start_new_stream('foo')
 141         cw.start_new_file('zero.txt')
 142         cw.write(b'')
 143         self.check_manifest_file_sizes(cw.manifest_text(), [0,1,0])
 144
 145     def test_no_implicit_normalize(self):
 146         cw = arvados.CollectionWriter(self.api_client)
 147         cw.start_new_file('b')
 148         cw.write(b'b')
 149         cw.start_new_file('a')
 150         cw.write(b'')
 151         self.check_manifest_file_sizes(cw.manifest_text(), [1,0])
 152         self.check_manifest_file_sizes(
 153             arvados.CollectionReader(
 154                 cw.manifest_text()).manifest_text(normalize=True),
 155             [0,1])
 156
 157     def check_manifest_file_sizes(self, manifest_text, expect_sizes):
 158         cr = arvados.CollectionReader(manifest_text, self.api_client)
 159         got_sizes = []
 160         for f in cr.all_files():
 161             got_sizes += [f.size()]
 162         self.assertEqual(got_sizes, expect_sizes, "got wrong file sizes %s, expected %s" % (got_sizes, expect_sizes))
 163
 164     def test_normalized_collection(self):
 165         m1 = """. 5348b82a029fd9e971a811ce1f71360b+43 0:43:md5sum.txt
 166 . 085c37f02916da1cad16f93c54d899b7+41 0:41:md5sum.txt
 167 . 8b22da26f9f433dea0a10e5ec66d73ba+43 0:43:md5sum.txt
 168 """
 169         self.assertEqual(arvados.CollectionReader(m1, self.api_client).manifest_text(normalize=True),
 170                          """. 5348b82a029fd9e971a811ce1f71360b+43 085c37f02916da1cad16f93c54d899b7+41 8b22da26f9f433dea0a10e5ec66d73ba+43 0:127:md5sum.txt
 171 """)
 172
 173         m2 = """. 204e43b8a1185621ca55a94839582e6f+67108864 b9677abbac956bd3e86b1deb28dfac03+67108864 fc15aff2a762b13f521baf042140acec+67108864 323d2a3ce20370c4ca1d3462a344f8fd+25885655 0:227212247:var-GS000016015-ASM.tsv.bz2
 174 """
 175         self.assertEqual(arvados.CollectionReader(m2, self.api_client).manifest_text(normalize=True), m2)
 176
 177         m3 = """. 5348b82a029fd9e971a811ce1f71360b+43 3:40:md5sum.txt
 178 . 085c37f02916da1cad16f93c54d899b7+41 0:41:md5sum.txt
 179 . 8b22da26f9f433dea0a10e5ec66d73ba+43 0:43:md5sum.txt
 180 """
 181         self.assertEqual(arvados.CollectionReader(m3, self.api_client).manifest_text(normalize=True),
 182                          """. 5348b82a029fd9e971a811ce1f71360b+43 085c37f02916da1cad16f93c54d899b7+41 8b22da26f9f433dea0a10e5ec66d73ba+43 3:124:md5sum.txt
 183 """)
 184
 185         m4 = """. 204e43b8a1185621ca55a94839582e6f+67108864 0:3:foo/bar
 186 ./zzz 204e43b8a1185621ca55a94839582e6f+67108864 0:999:zzz
 187 ./foo 323d2a3ce20370c4ca1d3462a344f8fd+25885655 0:3:bar
 188 """
 189         self.assertEqual(arvados.CollectionReader(m4, self.api_client).manifest_text(normalize=True),
 190                          """./foo 204e43b8a1185621ca55a94839582e6f+67108864 323d2a3ce20370c4ca1d3462a344f8fd+25885655 0:3:bar 67108864:3:bar
 191 ./zzz 204e43b8a1185621ca55a94839582e6f+67108864 0:999:zzz
 192 """)
 193
 194         m5 = """. 204e43b8a1185621ca55a94839582e6f+67108864 0:3:foo/bar
 195 ./zzz 204e43b8a1185621ca55a94839582e6f+67108864 0:999:zzz
 196 ./foo 204e43b8a1185621ca55a94839582e6f+67108864 3:3:bar
 197 """
 198         self.assertEqual(arvados.CollectionReader(m5, self.api_client).manifest_text(normalize=True),
 199                          """./foo 204e43b8a1185621ca55a94839582e6f+67108864 0:6:bar
 200 ./zzz 204e43b8a1185621ca55a94839582e6f+67108864 0:999:zzz
 201 """)
 202
 203         with self.data_file('1000G_ref_manifest') as f6:
 204             m6 = f6.read()
 205             self.assertEqual(arvados.CollectionReader(m6, self.api_client).manifest_text(normalize=True), m6)
 206
 207         with self.data_file('jlake_manifest') as f7:
 208             m7 = f7.read()
 209             self.assertEqual(arvados.CollectionReader(m7, self.api_client).manifest_text(normalize=True), m7)
 210
 211         m8 = """./a\\040b\\040c 59ca0efa9f5633cb0371bbc0355478d8+13 0:13:hello\\040world.txt
 212 """
 213         self.assertEqual(arvados.CollectionReader(m8, self.api_client).manifest_text(normalize=True), m8)
 214
 215     def test_locators_and_ranges(self):
 216         blocks2 = [Range('a', 0, 10),
 217                    Range('b', 10, 10),
 218                    Range('c', 20, 10),
 219                    Range('d', 30, 10),
 220                    Range('e', 40, 10),
 221                    Range('f', 50, 10)]
 222
 223         self.assertEqual(arvados.locators_and_ranges(blocks2,  2,  2), [LocatorAndRange('a', 10, 2, 2)])
 224         self.assertEqual(arvados.locators_and_ranges(blocks2, 12, 2), [LocatorAndRange('b', 10, 2, 2)])
 225         self.assertEqual(arvados.locators_and_ranges(blocks2, 22, 2), [LocatorAndRange('c', 10, 2, 2)])
 226         self.assertEqual(arvados.locators_and_ranges(blocks2, 32, 2), [LocatorAndRange('d', 10, 2, 2)])
 227         self.assertEqual(arvados.locators_and_ranges(blocks2, 42, 2), [LocatorAndRange('e', 10, 2, 2)])
 228         self.assertEqual(arvados.locators_and_ranges(blocks2, 52, 2), [LocatorAndRange('f', 10, 2, 2)])
 229         self.assertEqual(arvados.locators_and_ranges(blocks2, 62, 2), [])
 230         self.assertEqual(arvados.locators_and_ranges(blocks2, -2, 2), [])
 231
 232         self.assertEqual(arvados.locators_and_ranges(blocks2,  0,  2), [LocatorAndRange('a', 10, 0, 2)])
 233         self.assertEqual(arvados.locators_and_ranges(blocks2, 10, 2), [LocatorAndRange('b', 10, 0, 2)])
 234         self.assertEqual(arvados.locators_and_ranges(blocks2, 20, 2), [LocatorAndRange('c', 10, 0, 2)])
 235         self.assertEqual(arvados.locators_and_ranges(blocks2, 30, 2), [LocatorAndRange('d', 10, 0, 2)])
 236         self.assertEqual(arvados.locators_and_ranges(blocks2, 40, 2), [LocatorAndRange('e', 10, 0, 2)])
 237         self.assertEqual(arvados.locators_and_ranges(blocks2, 50, 2), [LocatorAndRange('f', 10, 0, 2)])
 238         self.assertEqual(arvados.locators_and_ranges(blocks2, 60, 2), [])
 239         self.assertEqual(arvados.locators_and_ranges(blocks2, -2, 2), [])
 240
 241         self.assertEqual(arvados.locators_and_ranges(blocks2,  9,  2), [LocatorAndRange('a', 10, 9, 1), LocatorAndRange('b', 10, 0, 1)])
 242         self.assertEqual(arvados.locators_and_ranges(blocks2, 19, 2), [LocatorAndRange('b', 10, 9, 1), LocatorAndRange('c', 10, 0, 1)])
 243         self.assertEqual(arvados.locators_and_ranges(blocks2, 29, 2), [LocatorAndRange('c', 10, 9, 1), LocatorAndRange('d', 10, 0, 1)])
 244         self.assertEqual(arvados.locators_and_ranges(blocks2, 39, 2), [LocatorAndRange('d', 10, 9, 1), LocatorAndRange('e', 10, 0, 1)])
 245         self.assertEqual(arvados.locators_and_ranges(blocks2, 49, 2), [LocatorAndRange('e', 10, 9, 1), LocatorAndRange('f', 10, 0, 1)])
 246         self.assertEqual(arvados.locators_and_ranges(blocks2, 59, 2), [LocatorAndRange('f', 10, 9, 1)])
 247
 248
 249         blocks3 = [Range('a', 0, 10),
 250                   Range('b', 10, 10),
 251                   Range('c', 20, 10),
 252                   Range('d', 30, 10),
 253                   Range('e', 40, 10),
 254                   Range('f', 50, 10),
 255                    Range('g', 60, 10)]
 256
 257         self.assertEqual(arvados.locators_and_ranges(blocks3,  2,  2), [LocatorAndRange('a', 10, 2, 2)])
 258         self.assertEqual(arvados.locators_and_ranges(blocks3, 12, 2), [LocatorAndRange('b', 10, 2, 2)])
 259         self.assertEqual(arvados.locators_and_ranges(blocks3, 22, 2), [LocatorAndRange('c', 10, 2, 2)])
 260         self.assertEqual(arvados.locators_and_ranges(blocks3, 32, 2), [LocatorAndRange('d', 10, 2, 2)])
 261         self.assertEqual(arvados.locators_and_ranges(blocks3, 42, 2), [LocatorAndRange('e', 10, 2, 2)])
 262         self.assertEqual(arvados.locators_and_ranges(blocks3, 52, 2), [LocatorAndRange('f', 10, 2, 2)])
 263         self.assertEqual(arvados.locators_and_ranges(blocks3, 62, 2), [LocatorAndRange('g', 10, 2, 2)])
 264
 265
 266         blocks = [Range('a', 0, 10),
 267                   Range('b', 10, 15),
 268                   Range('c', 25, 5)]
 269         self.assertEqual(arvados.locators_and_ranges(blocks, 1, 0), [])
 270         self.assertEqual(arvados.locators_and_ranges(blocks, 0, 5), [LocatorAndRange('a', 10, 0, 5)])
 271         self.assertEqual(arvados.locators_and_ranges(blocks, 3, 5), [LocatorAndRange('a', 10, 3, 5)])
 272         self.assertEqual(arvados.locators_and_ranges(blocks, 0, 10), [LocatorAndRange('a', 10, 0, 10)])
 273
 274         self.assertEqual(arvados.locators_and_ranges(blocks, 0, 11), [LocatorAndRange('a', 10, 0, 10),
 275                                                                       LocatorAndRange('b', 15, 0, 1)])
 276         self.assertEqual(arvados.locators_and_ranges(blocks, 1, 11), [LocatorAndRange('a', 10, 1, 9),
 277                                                                       LocatorAndRange('b', 15, 0, 2)])
 278         self.assertEqual(arvados.locators_and_ranges(blocks, 0, 25), [LocatorAndRange('a', 10, 0, 10),
 279                                                                       LocatorAndRange('b', 15, 0, 15)])
 280
 281         self.assertEqual(arvados.locators_and_ranges(blocks, 0, 30), [LocatorAndRange('a', 10, 0, 10),
 282                                                                       LocatorAndRange('b', 15, 0, 15),
 283                                                                       LocatorAndRange('c', 5, 0, 5)])
 284         self.assertEqual(arvados.locators_and_ranges(blocks, 1, 30), [LocatorAndRange('a', 10, 1, 9),
 285                                                                       LocatorAndRange('b', 15, 0, 15),
 286                                                                       LocatorAndRange('c', 5, 0, 5)])
 287         self.assertEqual(arvados.locators_and_ranges(blocks, 0, 31), [LocatorAndRange('a', 10, 0, 10),
 288                                                                       LocatorAndRange('b', 15, 0, 15),
 289                                                                       LocatorAndRange('c', 5, 0, 5)])
 290
 291         self.assertEqual(arvados.locators_and_ranges(blocks, 15, 5), [LocatorAndRange('b', 15, 5, 5)])
 292
 293         self.assertEqual(arvados.locators_and_ranges(blocks, 8, 17), [LocatorAndRange('a', 10, 8, 2),
 294                                                                       LocatorAndRange('b', 15, 0, 15)])
 295
 296         self.assertEqual(arvados.locators_and_ranges(blocks, 8, 20), [LocatorAndRange('a', 10, 8, 2),
 297                                                                       LocatorAndRange('b', 15, 0, 15),
 298                                                                       LocatorAndRange('c', 5, 0, 3)])
 299
 300         self.assertEqual(arvados.locators_and_ranges(blocks, 26, 2), [LocatorAndRange('c', 5, 1, 2)])
 301
 302         self.assertEqual(arvados.locators_and_ranges(blocks, 9, 15), [LocatorAndRange('a', 10, 9, 1),
 303                                                                       LocatorAndRange('b', 15, 0, 14)])
 304         self.assertEqual(arvados.locators_and_ranges(blocks, 10, 15), [LocatorAndRange('b', 15, 0, 15)])
 305         self.assertEqual(arvados.locators_and_ranges(blocks, 11, 15), [LocatorAndRange('b', 15, 1, 14),
 306                                                                        LocatorAndRange('c', 5, 0, 1)])
 307
 308     class MockKeep(object):
 309         def __init__(self, content, num_retries=0):
 310             self.content = content
 311
 312         def get(self, locator, num_retries=0):
 313             return self.content[locator]
 314
 315     def test_stream_reader(self):
 316         keepblocks = {
 317             'aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa+10': b'abcdefghij',
 318             'bbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbb+15': b'klmnopqrstuvwxy',
 319             'cccccccccccccccccccccccccccccccc+5': b'z0123',
 320         }
 321         mk = self.MockKeep(keepblocks)
 322
 323         sr = arvados.StreamReader([".", "aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa+10", "bbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbb+15", "cccccccccccccccccccccccccccccccc+5", "0:30:foo"], mk)
 324
 325         content = b'abcdefghijklmnopqrstuvwxyz0123456789'
 326
 327         self.assertEqual(sr.readfrom(0, 30), content[0:30])
 328         self.assertEqual(sr.readfrom(2, 30), content[2:30])
 329
 330         self.assertEqual(sr.readfrom(2, 8), content[2:10])
 331         self.assertEqual(sr.readfrom(0, 10), content[0:10])
 332
 333         self.assertEqual(sr.readfrom(0, 5), content[0:5])
 334         self.assertEqual(sr.readfrom(5, 5), content[5:10])
 335         self.assertEqual(sr.readfrom(10, 5), content[10:15])
 336         self.assertEqual(sr.readfrom(15, 5), content[15:20])
 337         self.assertEqual(sr.readfrom(20, 5), content[20:25])
 338         self.assertEqual(sr.readfrom(25, 5), content[25:30])
 339         self.assertEqual(sr.readfrom(30, 5), b'')
 340
 341     def test_extract_file(self):
 342         m1 = """. 5348b82a029fd9e971a811ce1f71360b+43 0:43:md5sum.txt
 343 . 085c37f02916da1cad16f93c54d899b7+41 0:41:md6sum.txt
 344 . 8b22da26f9f433dea0a10e5ec66d73ba+43 0:43:md7sum.txt
 345 . 085c37f02916da1cad16f93c54d899b7+41 5348b82a029fd9e971a811ce1f71360b+43 8b22da26f9f433dea0a10e5ec66d73ba+43 47:80:md8sum.txt
 346 . 085c37f02916da1cad16f93c54d899b7+41 5348b82a029fd9e971a811ce1f71360b+43 8b22da26f9f433dea0a10e5ec66d73ba+43 40:80:md9sum.txt
 347 """
 348
 349         m2 = arvados.CollectionReader(m1, self.api_client).manifest_text(normalize=True)
 350
 351         self.assertEqual(m2,
 352                          ". 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")
 353         files = arvados.CollectionReader(
 354             m2, self.api_client).all_streams()[0].files()
 355
 356         self.assertEqual(files['md5sum.txt'].as_manifest(),
 357                          ". 5348b82a029fd9e971a811ce1f71360b+43 0:43:md5sum.txt\n")
 358         self.assertEqual(files['md6sum.txt'].as_manifest(),
 359                          ". 085c37f02916da1cad16f93c54d899b7+41 0:41:md6sum.txt\n")
 360         self.assertEqual(files['md7sum.txt'].as_manifest(),
 361                          ". 8b22da26f9f433dea0a10e5ec66d73ba+43 0:43:md7sum.txt\n")
 362         self.assertEqual(files['md9sum.txt'].as_manifest(),
 363                          ". 085c37f02916da1cad16f93c54d899b7+41 5348b82a029fd9e971a811ce1f71360b+43 8b22da26f9f433dea0a10e5ec66d73ba+43 40:80:md9sum.txt\n")
 364
 365     def test_write_directory_tree(self):
 366         cwriter = arvados.CollectionWriter(self.api_client)
 367         cwriter.write_directory_tree(self.build_directory_tree(
 368                 ['basefile', 'subdir/subfile']))
 369         self.assertEqual(cwriter.manifest_text(),
 370                          """. c5110c5ac93202d8e0f9e381f22bac0f+8 0:8:basefile
 371 ./subdir 1ca4dec89403084bf282ad31e6cf7972+14 0:14:subfile\n""")
 372
 373     def test_write_named_directory_tree(self):
 374         cwriter = arvados.CollectionWriter(self.api_client)
 375         cwriter.write_directory_tree(self.build_directory_tree(
 376                 ['basefile', 'subdir/subfile']), 'root')
 377         self.assertEqual(
 378             cwriter.manifest_text(),
 379             """./root c5110c5ac93202d8e0f9e381f22bac0f+8 0:8:basefile
 380 ./root/subdir 1ca4dec89403084bf282ad31e6cf7972+14 0:14:subfile\n""")
 381
 382     def test_write_directory_tree_in_one_stream(self):
 383         cwriter = arvados.CollectionWriter(self.api_client)
 384         cwriter.write_directory_tree(self.build_directory_tree(
 385                 ['basefile', 'subdir/subfile']), max_manifest_depth=0)
 386         self.assertEqual(cwriter.manifest_text(),
 387                          """. 4ace875ffdc6824a04950f06858f4465+22 0:8:basefile 8:14:subdir/subfile\n""")
 388
 389     def test_write_directory_tree_with_limited_recursion(self):
 390         cwriter = arvados.CollectionWriter(self.api_client)
 391         cwriter.write_directory_tree(
 392             self.build_directory_tree(['f1', 'd1/f2', 'd1/d2/f3']),
 393             max_manifest_depth=1)
 394         self.assertEqual(cwriter.manifest_text(),
 395                          """. bd19836ddb62c11c55ab251ccaca5645+2 0:2:f1
 396 ./d1 50170217e5b04312024aa5cd42934494+13 0:8:d2/f3 8:5:f2\n""")
 397
 398     def test_write_directory_tree_with_zero_recursion(self):
 399         cwriter = arvados.CollectionWriter(self.api_client)
 400         content = 'd1/d2/f3d1/f2f1'
 401         blockhash = tutil.str_keep_locator(content)
 402         cwriter.write_directory_tree(
 403             self.build_directory_tree(['f1', 'd1/f2', 'd1/d2/f3']),
 404             max_manifest_depth=0)
 405         self.assertEqual(
 406             cwriter.manifest_text(),
 407             ". {} 0:8:d1/d2/f3 8:5:d1/f2 13:2:f1\n".format(blockhash))
 408
 409     def test_write_one_file(self):
 410         cwriter = arvados.CollectionWriter(self.api_client)
 411         with self.make_test_file() as testfile:
 412             cwriter.write_file(testfile.name)
 413             self.assertEqual(
 414                 cwriter.manifest_text(),
 415                 ". 098f6bcd4621d373cade4e832627b4f6+4 0:4:{}\n".format(
 416                     os.path.basename(testfile.name)))
 417
 418     def test_write_named_file(self):
 419         cwriter = arvados.CollectionWriter(self.api_client)
 420         with self.make_test_file() as testfile:
 421             cwriter.write_file(testfile.name, 'foo')
 422             self.assertEqual(cwriter.manifest_text(),
 423                              ". 098f6bcd4621d373cade4e832627b4f6+4 0:4:foo\n")
 424
 425     def test_write_multiple_files(self):
 426         cwriter = arvados.CollectionWriter(self.api_client)
 427         for letter in 'ABC':
 428             with self.make_test_file(letter.encode()) as testfile:
 429                 cwriter.write_file(testfile.name, letter)
 430         self.assertEqual(
 431             cwriter.manifest_text(),
 432             ". 902fbdd2b1df0c4f70b4a5d23525e932+3 0:1:A 1:1:B 2:1:C\n")
 433
 434     def test_basic_resume(self):
 435         cwriter = TestResumableWriter()
 436         with self.make_test_file() as testfile:
 437             cwriter.write_file(testfile.name, 'test')
 438             resumed = TestResumableWriter.from_state(cwriter.current_state())
 439         self.assertEqual(cwriter.manifest_text(), resumed.manifest_text(),
 440                           "resumed CollectionWriter had different manifest")
 441
 442     def test_resume_fails_when_missing_dependency(self):
 443         cwriter = TestResumableWriter()
 444         with self.make_test_file() as testfile:
 445             cwriter.write_file(testfile.name, 'test')
 446         self.assertRaises(arvados.errors.StaleWriterStateError,
 447                           TestResumableWriter.from_state,
 448                           cwriter.current_state())
 449
 450     def test_resume_fails_when_dependency_mtime_changed(self):
 451         cwriter = TestResumableWriter()
 452         with self.make_test_file() as testfile:
 453             cwriter.write_file(testfile.name, 'test')
 454             os.utime(testfile.name, (0, 0))
 455             self.assertRaises(arvados.errors.StaleWriterStateError,
 456                               TestResumableWriter.from_state,
 457                               cwriter.current_state())
 458
 459     def test_resume_fails_when_dependency_is_nonfile(self):
 460         cwriter = TestResumableWriter()
 461         cwriter.write_file('/dev/null', 'empty')
 462         self.assertRaises(arvados.errors.StaleWriterStateError,
 463                           TestResumableWriter.from_state,
 464                           cwriter.current_state())
 465
 466     def test_resume_fails_when_dependency_size_changed(self):
 467         cwriter = TestResumableWriter()
 468         with self.make_test_file() as testfile:
 469             cwriter.write_file(testfile.name, 'test')
 470             orig_mtime = os.fstat(testfile.fileno()).st_mtime
 471             testfile.write(b'extra')
 472             testfile.flush()
 473             os.utime(testfile.name, (orig_mtime, orig_mtime))
 474             self.assertRaises(arvados.errors.StaleWriterStateError,
 475                               TestResumableWriter.from_state,
 476                               cwriter.current_state())
 477
 478     def test_resume_fails_with_expired_locator(self):
 479         cwriter = TestResumableWriter()
 480         state = cwriter.current_state()
 481         # Add an expired locator to the state.
 482         state['_current_stream_locators'].append(''.join([
 483                     'a' * 32, '+1+A', 'b' * 40, '@', '10000000']))
 484         self.assertRaises(arvados.errors.StaleWriterStateError,
 485                           TestResumableWriter.from_state, state)
 486
 487     def test_arbitrary_objects_not_resumable(self):
 488         cwriter = TestResumableWriter()
 489         with open('/dev/null') as badfile:
 490             self.assertRaises(arvados.errors.AssertionError,
 491                               cwriter.write_file, badfile)
 492
 493     def test_arbitrary_writes_not_resumable(self):
 494         cwriter = TestResumableWriter()
 495         self.assertRaises(arvados.errors.AssertionError,
 496                           cwriter.write, "badtext")
 497
 498     def test_read_arbitrary_data_with_collection_reader(self):
 499         # arv-get relies on this to do "arv-get {keep-locator} -".
 500         self.write_foo_bar_baz()
 501         self.assertEqual(
 502             'foobar',
 503             arvados.CollectionReader(
 504                 '3858f62230ac3c915f300c664312c63f+6'
 505                 ).manifest_text())
 506
 507
 508 class CollectionTestMixin(tutil.ApiClientMock):
 509     API_COLLECTIONS = run_test_server.fixture('collections')
 510     DEFAULT_COLLECTION = API_COLLECTIONS['foo_file']
 511     DEFAULT_DATA_HASH = DEFAULT_COLLECTION['portable_data_hash']
 512     DEFAULT_MANIFEST = DEFAULT_COLLECTION['manifest_text']
 513     DEFAULT_UUID = DEFAULT_COLLECTION['uuid']
 514     ALT_COLLECTION = API_COLLECTIONS['bar_file']
 515     ALT_DATA_HASH = ALT_COLLECTION['portable_data_hash']
 516     ALT_MANIFEST = ALT_COLLECTION['manifest_text']
 517
 518     def api_client_mock(self, status=200):
 519         client = super(CollectionTestMixin, self).api_client_mock()
 520         self.mock_keep_services(client, status=status, service_type='proxy', count=1)
 521         return client
 522
 523
 524 @tutil.skip_sleep
 525 class CollectionReaderTestCase(unittest.TestCase, CollectionTestMixin):
 526     def mock_get_collection(self, api_mock, code, body):
 527         body = self.API_COLLECTIONS.get(body)
 528         self._mock_api_call(api_mock.collections().get, code, body)
 529
 530     def api_client_mock(self, status=200):
 531         client = super(CollectionReaderTestCase, self).api_client_mock()
 532         self.mock_get_collection(client, status, 'foo_file')
 533         return client
 534
 535     def test_init_no_default_retries(self):
 536         client = self.api_client_mock(200)
 537         reader = arvados.CollectionReader(self.DEFAULT_UUID, api_client=client)
 538         reader.manifest_text()
 539         client.collections().get().execute.assert_called_with(num_retries=0)
 540
 541     def test_uuid_init_success(self):
 542         client = self.api_client_mock(200)
 543         reader = arvados.CollectionReader(self.DEFAULT_UUID, api_client=client,
 544                                           num_retries=3)
 545         self.assertEqual(self.DEFAULT_COLLECTION['manifest_text'],
 546                          reader.manifest_text())
 547         client.collections().get().execute.assert_called_with(num_retries=3)
 548
 549     def test_uuid_init_failure_raises_api_error(self):
 550         client = self.api_client_mock(500)
 551         with self.assertRaises(arvados.errors.ApiError):
 552             reader = arvados.CollectionReader(self.DEFAULT_UUID, api_client=client)
 553
 554     def test_locator_init(self):
 555         client = self.api_client_mock(200)
 556         # Ensure Keep will not return anything if asked.
 557         with tutil.mock_keep_responses(None, 404):
 558             reader = arvados.CollectionReader(self.DEFAULT_DATA_HASH,
 559                                               api_client=client)
 560             self.assertEqual(self.DEFAULT_MANIFEST, reader.manifest_text())
 561
 562     def test_locator_init_fallback_to_keep(self):
 563         # crunch-job needs this to read manifests that have only ever
 564         # been written to Keep.
 565         client = self.api_client_mock(200)
 566         self.mock_get_collection(client, 404, None)
 567         with tutil.mock_keep_responses(self.DEFAULT_MANIFEST, 200):
 568             reader = arvados.CollectionReader(self.DEFAULT_DATA_HASH,
 569                                               api_client=client)
 570             self.assertEqual(self.DEFAULT_MANIFEST, reader.manifest_text())
 571
 572     def test_uuid_init_no_fallback_to_keep(self):
 573         # Do not look up a collection UUID in Keep.
 574         client = self.api_client_mock(404)
 575         with tutil.mock_keep_responses(self.DEFAULT_MANIFEST, 200):
 576             with self.assertRaises(arvados.errors.ApiError):
 577                 reader = arvados.CollectionReader(self.DEFAULT_UUID,
 578                                                   api_client=client)
 579
 580     def test_try_keep_first_if_permission_hint(self):
 581         # To verify that CollectionReader tries Keep first here, we
 582         # mock API server to return the wrong data.
 583         client = self.api_client_mock(200)
 584         with tutil.mock_keep_responses(self.ALT_MANIFEST, 200):
 585             self.assertEqual(
 586                 self.ALT_MANIFEST,
 587                 arvados.CollectionReader(
 588                     self.ALT_DATA_HASH + '+Affffffffffffffffffffffffffffffffffffffff@fedcba98',
 589                     api_client=client).manifest_text())
 590
 591     def test_init_num_retries_propagated(self):
 592         # More of an integration test...
 593         client = self.api_client_mock(200)
 594         reader = arvados.CollectionReader(self.DEFAULT_UUID, api_client=client,
 595                                           num_retries=3)
 596         with tutil.mock_keep_responses('foo', 500, 500, 200):
 597             self.assertEqual(b'foo',
 598                              b''.join(f.read(9) for f in reader.all_files()))
 599
 600     def test_read_nonnormalized_manifest_with_collection_reader(self):
 601         # client should be able to use CollectionReader on a manifest without normalizing it
 602         client = self.api_client_mock(500)
 603         nonnormal = ". acbd18db4cc2f85cedef654fccc4a4d8+3+Aabadbadbee@abeebdee 0:3:foo.txt 1:0:bar.txt 0:3:foo.txt\n"
 604         reader = arvados.CollectionReader(
 605             nonnormal,
 606             api_client=client, num_retries=0)
 607         # Ensure stripped_manifest() doesn't mangle our manifest in
 608         # any way other than stripping hints.
 609         self.assertEqual(
 610             re.sub('\+[^\d\s\+]+', '', nonnormal),
 611             reader.stripped_manifest())
 612         # Ensure stripped_manifest() didn't mutate our reader.
 613         self.assertEqual(nonnormal, reader.manifest_text())
 614         # Ensure the files appear in the order given in the manifest.
 615         self.assertEqual(
 616             [[6, '.', 'foo.txt'],
 617              [0, '.', 'bar.txt']],
 618             [[f.size(), f.stream_name(), f.name()]
 619              for f in reader.all_streams()[0].all_files()])
 620
 621     def test_read_empty_collection(self):
 622         client = self.api_client_mock(200)
 623         self.mock_get_collection(client, 200, 'empty')
 624         reader = arvados.CollectionReader('d41d8cd98f00b204e9800998ecf8427e+0',
 625                                           api_client=client)
 626         self.assertEqual('', reader.manifest_text())
 627
 628     def test_api_response(self):
 629         client = self.api_client_mock()
 630         reader = arvados.CollectionReader(self.DEFAULT_UUID, api_client=client)
 631         self.assertEqual(self.DEFAULT_COLLECTION, reader.api_response())
 632
 633     def test_api_response_with_collection_from_keep(self):
 634         client = self.api_client_mock()
 635         self.mock_get_collection(client, 404, 'foo')
 636         with tutil.mock_keep_responses(self.DEFAULT_MANIFEST, 200):
 637             reader = arvados.CollectionReader(self.DEFAULT_DATA_HASH,
 638                                               api_client=client)
 639             api_response = reader.api_response()
 640         self.assertIsNone(api_response)
 641
 642     def check_open_file(self, coll_file, stream_name, file_name, file_size):
 643         self.assertFalse(coll_file.closed, "returned file is not open")
 644         self.assertEqual(stream_name, coll_file.stream_name())
 645         self.assertEqual(file_name, coll_file.name)
 646         self.assertEqual(file_size, coll_file.size())
 647
 648     def test_open_collection_file_one_argument(self):
 649         client = self.api_client_mock(200)
 650         reader = arvados.CollectionReader(self.DEFAULT_UUID, api_client=client)
 651         cfile = reader.open('./foo')
 652         self.check_open_file(cfile, '.', 'foo', 3)
 653
 654     def test_open_deep_file(self):
 655         coll_name = 'collection_with_files_in_subdir'
 656         client = self.api_client_mock(200)
 657         self.mock_get_collection(client, 200, coll_name)
 658         reader = arvados.CollectionReader(
 659             self.API_COLLECTIONS[coll_name]['uuid'], api_client=client)
 660         cfile = reader.open('./subdir2/subdir3/file2_in_subdir3.txt')
 661         self.check_open_file(cfile, './subdir2/subdir3', 'file2_in_subdir3.txt',
 662                              32)
 663
 664     def test_open_nonexistent_stream(self):
 665         client = self.api_client_mock(200)
 666         reader = arvados.CollectionReader(self.DEFAULT_UUID, api_client=client)
 667         self.assertRaises(IOError, reader.open, './nonexistent/foo')
 668
 669     def test_open_nonexistent_file(self):
 670         client = self.api_client_mock(200)
 671         reader = arvados.CollectionReader(self.DEFAULT_UUID, api_client=client)
 672         self.assertRaises(IOError, reader.open, 'nonexistent')
 673
 674
 675 @tutil.skip_sleep
 676 class CollectionWriterTestCase(unittest.TestCase, CollectionTestMixin):
 677     def mock_keep(self, body, *codes, **headers):
 678         headers.setdefault('x-keep-replicas-stored', 2)
 679         return tutil.mock_keep_responses(body, *codes, **headers)
 680
 681     def foo_writer(self, **kwargs):
 682         kwargs.setdefault('api_client', self.api_client_mock())
 683         writer = arvados.CollectionWriter(**kwargs)
 684         writer.start_new_file('foo')
 685         writer.write(b'foo')
 686         return writer
 687
 688     def test_write_whole_collection(self):
 689         writer = self.foo_writer()
 690         with self.mock_keep(self.DEFAULT_DATA_HASH, 200, 200):
 691             self.assertEqual(self.DEFAULT_DATA_HASH, writer.finish())
 692
 693     def test_write_no_default(self):
 694         writer = self.foo_writer()
 695         with self.mock_keep(None, 500):
 696             with self.assertRaises(arvados.errors.KeepWriteError):
 697                 writer.finish()
 698
 699     def test_write_insufficient_replicas_via_proxy(self):
 700         writer = self.foo_writer(replication=3)
 701         with self.mock_keep(None, 200, **{'x-keep-replicas-stored': 2}):
 702             with self.assertRaises(arvados.errors.KeepWriteError):
 703                 writer.manifest_text()
 704
 705     def test_write_insufficient_replicas_via_disks(self):
 706         client = mock.MagicMock(name='api_client')
 707         with self.mock_keep(
 708                 None, 200, 200,
 709                 **{'x-keep-replicas-stored': 1}) as keepmock:
 710             self.mock_keep_services(client, status=200, service_type='disk', count=2)
 711             writer = self.foo_writer(api_client=client, replication=3)
 712             with self.assertRaises(arvados.errors.KeepWriteError):
 713                 writer.manifest_text()
 714
 715     def test_write_three_replicas(self):
 716         client = mock.MagicMock(name='api_client')
 717         with self.mock_keep(
 718                 "", 500, 500, 500, 200, 200, 200,
 719                 **{'x-keep-replicas-stored': 1}) as keepmock:
 720             self.mock_keep_services(client, status=200, service_type='disk', count=6)
 721             writer = self.foo_writer(api_client=client, replication=3)
 722             writer.manifest_text()
 723             self.assertEqual(6, keepmock.call_count)
 724
 725     def test_write_whole_collection_through_retries(self):
 726         writer = self.foo_writer(num_retries=2)
 727         with self.mock_keep(self.DEFAULT_DATA_HASH,
 728                             500, 500, 200, 500, 500, 200):
 729             self.assertEqual(self.DEFAULT_DATA_HASH, writer.finish())
 730
 731     def test_flush_data_retries(self):
 732         writer = self.foo_writer(num_retries=2)
 733         foo_hash = self.DEFAULT_MANIFEST.split()[1]
 734         with self.mock_keep(foo_hash, 500, 200):
 735             writer.flush_data()
 736         self.assertEqual(self.DEFAULT_MANIFEST, writer.manifest_text())
 737
 738     def test_one_open(self):
 739         client = self.api_client_mock()
 740         writer = arvados.CollectionWriter(client)
 741         with writer.open('out') as out_file:
 742             self.assertEqual('.', writer.current_stream_name())
 743             self.assertEqual('out', writer.current_file_name())
 744             out_file.write(b'test data')
 745             data_loc = tutil.str_keep_locator('test data')
 746         self.assertTrue(out_file.closed, "writer file not closed after context")
 747         self.assertRaises(ValueError, out_file.write, 'extra text')
 748         with self.mock_keep(data_loc, 200) as keep_mock:
 749             self.assertEqual(". {} 0:9:out\n".format(data_loc),
 750                              writer.manifest_text())
 751
 752     def test_open_writelines(self):
 753         client = self.api_client_mock()
 754         writer = arvados.CollectionWriter(client)
 755         with writer.open('six') as out_file:
 756             out_file.writelines(['12', '34', '56'])
 757             data_loc = tutil.str_keep_locator('123456')
 758         with self.mock_keep(data_loc, 200) as keep_mock:
 759             self.assertEqual(". {} 0:6:six\n".format(data_loc),
 760                              writer.manifest_text())
 761
 762     def test_open_flush(self):
 763         client = self.api_client_mock()
 764         data_loc1 = tutil.str_keep_locator('flush1')
 765         data_loc2 = tutil.str_keep_locator('flush2')
 766         with self.mock_keep((data_loc1, 200), (data_loc2, 200)) as keep_mock:
 767             writer = arvados.CollectionWriter(client)
 768             with writer.open('flush_test') as out_file:
 769                 out_file.write(b'flush1')
 770                 out_file.flush()
 771                 out_file.write(b'flush2')
 772             self.assertEqual(". {} {} 0:12:flush_test\n".format(data_loc1,
 773                                                                 data_loc2),
 774                              writer.manifest_text())
 775
 776     def test_two_opens_same_stream(self):
 777         client = self.api_client_mock()
 778         writer = arvados.CollectionWriter(client)
 779         with writer.open('.', '1') as out_file:
 780             out_file.write(b'1st')
 781         with writer.open('.', '2') as out_file:
 782             out_file.write(b'2nd')
 783         data_loc = tutil.str_keep_locator('1st2nd')
 784         with self.mock_keep(data_loc, 200) as keep_mock:
 785             self.assertEqual(". {} 0:3:1 3:3:2\n".format(data_loc),
 786                              writer.manifest_text())
 787
 788     def test_two_opens_two_streams(self):
 789         client = self.api_client_mock()
 790         data_loc1 = tutil.str_keep_locator('file')
 791         data_loc2 = tutil.str_keep_locator('indir')
 792         with self.mock_keep((data_loc1, 200), (data_loc2, 200)) as keep_mock:
 793             writer = arvados.CollectionWriter(client)
 794             with writer.open('file') as out_file:
 795                 out_file.write(b'file')
 796             with writer.open('./dir', 'indir') as out_file:
 797                 out_file.write(b'indir')
 798             expected = ". {} 0:4:file\n./dir {} 0:5:indir\n".format(
 799                 data_loc1, data_loc2)
 800             self.assertEqual(expected, writer.manifest_text())
 801
 802     def test_dup_open_fails(self):
 803         client = self.api_client_mock()
 804         writer = arvados.CollectionWriter(client)
 805         file1 = writer.open('one')
 806         self.assertRaises(arvados.errors.AssertionError, writer.open, 'two')
 807
 808
 809 class NewCollectionTestCase(unittest.TestCase, CollectionTestMixin):
 810
 811     def test_replication_desired_kept_on_load(self):
 812         m = '. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt 0:10:count2.txt\n'
 813         c1 = Collection(m, replication_desired=1)
 814         c1.save_new()
 815         loc = c1.manifest_locator()
 816         c2 = Collection(loc)
 817         self.assertEqual(c1.manifest_text, c2.manifest_text)
 818         self.assertEqual(c1.replication_desired, c2.replication_desired)
 819
 820     def test_replication_desired_not_loaded_if_provided(self):
 821         m = '. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt 0:10:count2.txt\n'
 822         c1 = Collection(m, replication_desired=1)
 823         c1.save_new()
 824         loc = c1.manifest_locator()
 825         c2 = Collection(loc, replication_desired=2)
 826         self.assertEqual(c1.manifest_text, c2.manifest_text)
 827         self.assertNotEqual(c1.replication_desired, c2.replication_desired)
 828
 829     def test_init_manifest(self):
 830         m1 = """. 5348b82a029fd9e971a811ce1f71360b+43 0:43:md5sum.txt
 831 . 085c37f02916da1cad16f93c54d899b7+41 0:41:md5sum.txt
 832 . 8b22da26f9f433dea0a10e5ec66d73ba+43 0:43:md5sum.txt
 833 """
 834         self.assertEqual(m1, CollectionReader(m1).manifest_text(normalize=False))
 835         self.assertEqual(". 5348b82a029fd9e971a811ce1f71360b+43 085c37f02916da1cad16f93c54d899b7+41 8b22da26f9f433dea0a10e5ec66d73ba+43 0:127:md5sum.txt\n", CollectionReader(m1).manifest_text(normalize=True))
 836
 837     def test_init_manifest_with_collision(self):
 838         m1 = """. 5348b82a029fd9e971a811ce1f71360b+43 0:43:md5sum.txt
 839 ./md5sum.txt 085c37f02916da1cad16f93c54d899b7+41 0:41:md5sum.txt
 840 """
 841         with self.assertRaises(arvados.errors.ArgumentError):
 842             self.assertEqual(m1, CollectionReader(m1))
 843
 844     def test_init_manifest_with_error(self):
 845         m1 = """. 0:43:md5sum.txt"""
 846         with self.assertRaises(arvados.errors.ArgumentError):
 847             self.assertEqual(m1, CollectionReader(m1))
 848
 849     def test_remove(self):
 850         c = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt 0:10:count2.txt\n')
 851         self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt 0:10:count2.txt\n", c.portable_manifest_text())
 852         self.assertIn("count1.txt", c)
 853         c.remove("count1.txt")
 854         self.assertNotIn("count1.txt", c)
 855         self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count2.txt\n", c.portable_manifest_text())
 856         with self.assertRaises(arvados.errors.ArgumentError):
 857             c.remove("")
 858
 859     def test_find(self):
 860         c = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt 0:10:count2.txt\n')
 861         self.assertIs(c.find("."), c)
 862         self.assertIs(c.find("./count1.txt"), c["count1.txt"])
 863         self.assertIs(c.find("count1.txt"), c["count1.txt"])
 864         with self.assertRaises(IOError):
 865             c.find("/.")
 866         with self.assertRaises(arvados.errors.ArgumentError):
 867             c.find("")
 868         self.assertIs(c.find("./nonexistant.txt"), None)
 869         self.assertIs(c.find("./nonexistantsubdir/nonexistant.txt"), None)
 870
 871     def test_remove_in_subdir(self):
 872         c = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n./foo 781e5e245d69b566979b86e28d23f2c7+10 0:10:count2.txt\n')
 873         c.remove("foo/count2.txt")
 874         self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n", c.portable_manifest_text())
 875
 876     def test_remove_empty_subdir(self):
 877         c = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n./foo 781e5e245d69b566979b86e28d23f2c7+10 0:10:count2.txt\n')
 878         c.remove("foo/count2.txt")
 879         c.remove("foo")
 880         self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n", c.portable_manifest_text())
 881
 882     def test_remove_nonempty_subdir(self):
 883         c = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n./foo 781e5e245d69b566979b86e28d23f2c7+10 0:10:count2.txt\n')
 884         with self.assertRaises(IOError):
 885             c.remove("foo")
 886         c.remove("foo", recursive=True)
 887         self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n", c.portable_manifest_text())
 888
 889     def test_copy_to_file_in_dir(self):
 890         c = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n')
 891         c.copy("count1.txt", "foo/count2.txt")
 892         self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n./foo 781e5e245d69b566979b86e28d23f2c7+10 0:10:count2.txt\n", c.portable_manifest_text())
 893
 894     def test_copy_file(self):
 895         c = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n')
 896         c.copy("count1.txt", "count2.txt")
 897         self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt 0:10:count2.txt\n", c.portable_manifest_text())
 898
 899     def test_copy_to_existing_dir(self):
 900         c = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n./foo 781e5e245d69b566979b86e28d23f2c7+10 0:10:count2.txt\n')
 901         c.copy("count1.txt", "foo")
 902         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())
 903
 904     def test_copy_to_new_dir(self):
 905         c = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n')
 906         c.copy("count1.txt", "foo/")
 907         self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n./foo 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n", c.portable_manifest_text())
 908
 909     def test_rename_file(self):
 910         c = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n')
 911         c.rename("count1.txt", "count2.txt")
 912         self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count2.txt\n", c.manifest_text())
 913
 914     def test_move_file_to_dir(self):
 915         c = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n')
 916         c.mkdirs("foo")
 917         c.rename("count1.txt", "foo/count2.txt")
 918         self.assertEqual("./foo 781e5e245d69b566979b86e28d23f2c7+10 0:10:count2.txt\n", c.manifest_text())
 919
 920     def test_move_file_to_other(self):
 921         c1 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n')
 922         c2 = Collection()
 923         c2.rename("count1.txt", "count2.txt", source_collection=c1)
 924         self.assertEqual("", c1.manifest_text())
 925         self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count2.txt\n", c2.manifest_text())
 926
 927     def test_clone(self):
 928         c = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n./foo 781e5e245d69b566979b86e28d23f2c7+10 0:10:count2.txt\n')
 929         cl = c.clone()
 930         self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n./foo 781e5e245d69b566979b86e28d23f2c7+10 0:10:count2.txt\n", cl.portable_manifest_text())
 931
 932     def test_diff_del_add(self):
 933         c1 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n')
 934         c2 = Collection('. 5348b82a029fd9e971a811ce1f71360b+43 0:10:count2.txt\n')
 935         d = c2.diff(c1)
 936         self.assertEqual(d, [('del', './count2.txt', c2["count2.txt"]),
 937                              ('add', './count1.txt', c1["count1.txt"])])
 938         d = c1.diff(c2)
 939         self.assertEqual(d, [('del', './count1.txt', c1["count1.txt"]),
 940                              ('add', './count2.txt', c2["count2.txt"])])
 941         self.assertNotEqual(c1.portable_manifest_text(), c2.portable_manifest_text())
 942         c1.apply(d)
 943         self.assertEqual(c1.portable_manifest_text(), c2.portable_manifest_text())
 944
 945     def test_diff_same(self):
 946         c1 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n')
 947         c2 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n')
 948         d = c2.diff(c1)
 949         self.assertEqual(d, [('tok', './count1.txt', c2["count1.txt"], c1["count1.txt"])])
 950         d = c1.diff(c2)
 951         self.assertEqual(d, [('tok', './count1.txt', c2["count1.txt"], c1["count1.txt"])])
 952
 953         self.assertEqual(c1.portable_manifest_text(), c2.portable_manifest_text())
 954         c1.apply(d)
 955         self.assertEqual(c1.portable_manifest_text(), c2.portable_manifest_text())
 956
 957     def test_diff_mod(self):
 958         c1 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n')
 959         c2 = Collection('. 5348b82a029fd9e971a811ce1f71360b+43 0:10:count1.txt\n')
 960         d = c2.diff(c1)
 961         self.assertEqual(d, [('mod', './count1.txt', c2["count1.txt"], c1["count1.txt"])])
 962         d = c1.diff(c2)
 963         self.assertEqual(d, [('mod', './count1.txt', c1["count1.txt"], c2["count1.txt"])])
 964
 965         self.assertNotEqual(c1.portable_manifest_text(), c2.portable_manifest_text())
 966         c1.apply(d)
 967         self.assertEqual(c1.portable_manifest_text(), c2.portable_manifest_text())
 968
 969     def test_diff_add(self):
 970         c1 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n')
 971         c2 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 5348b82a029fd9e971a811ce1f71360b+43 0:10:count1.txt 10:20:count2.txt\n')
 972         d = c2.diff(c1)
 973         self.assertEqual(d, [('del', './count2.txt', c2["count2.txt"]),
 974                              ('tok', './count1.txt', c2["count1.txt"], c1["count1.txt"])])
 975         d = c1.diff(c2)
 976         self.assertEqual(d, [('add', './count2.txt', c2["count2.txt"]),
 977                              ('tok', './count1.txt', c2["count1.txt"], c1["count1.txt"])])
 978
 979         self.assertNotEqual(c1.portable_manifest_text(), c2.portable_manifest_text())
 980         c1.apply(d)
 981         self.assertEqual(c1.portable_manifest_text(), c2.portable_manifest_text())
 982
 983     def test_diff_add_in_subcollection(self):
 984         c1 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n')
 985         c2 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n./foo 5348b82a029fd9e971a811ce1f71360b+43 0:10:count2.txt\n')
 986         d = c2.diff(c1)
 987         self.assertEqual(d, [('del', './foo', c2["foo"]),
 988                              ('tok', './count1.txt', c2["count1.txt"], c1["count1.txt"])])
 989         d = c1.diff(c2)
 990         self.assertEqual(d, [('add', './foo', c2["foo"]),
 991                              ('tok', './count1.txt', c2["count1.txt"], c1["count1.txt"])])
 992
 993         self.assertNotEqual(c1.portable_manifest_text(), c2.portable_manifest_text())
 994         c1.apply(d)
 995         self.assertEqual(c1.portable_manifest_text(), c2.portable_manifest_text())
 996
 997     def test_diff_del_add_in_subcollection(self):
 998         c1 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n./foo 5348b82a029fd9e971a811ce1f71360b+43 0:10:count2.txt\n')
 999         c2 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n./foo 5348b82a029fd9e971a811ce1f71360b+43 0:3:count3.txt\n')
1000
1001         d = c2.diff(c1)
1002         self.assertEqual(d, [('del', './foo/count3.txt', c2.find("foo/count3.txt")),
1003                              ('add', './foo/count2.txt', c1.find("foo/count2.txt")),
1004                              ('tok', './count1.txt', c2["count1.txt"], c1["count1.txt"])])
1005         d = c1.diff(c2)
1006         self.assertEqual(d, [('del', './foo/count2.txt', c1.find("foo/count2.txt")),
1007                              ('add', './foo/count3.txt', c2.find("foo/count3.txt")),
1008                              ('tok', './count1.txt', c2["count1.txt"], c1["count1.txt"])])
1009
1010         self.assertNotEqual(c1.portable_manifest_text(), c2.portable_manifest_text())
1011         c1.apply(d)
1012         self.assertEqual(c1.portable_manifest_text(), c2.portable_manifest_text())
1013
1014     def test_diff_mod_in_subcollection(self):
1015         c1 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n./foo 5348b82a029fd9e971a811ce1f71360b+43 0:10:count2.txt\n')
1016         c2 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt 0:3:foo\n')
1017         d = c2.diff(c1)
1018         self.assertEqual(d, [('mod', './foo', c2["foo"], c1["foo"]),
1019                              ('tok', './count1.txt', c2["count1.txt"], c1["count1.txt"])])
1020         d = c1.diff(c2)
1021         self.assertEqual(d, [('mod', './foo', c1["foo"], c2["foo"]),
1022                              ('tok', './count1.txt', c2["count1.txt"], c1["count1.txt"])])
1023
1024         self.assertNotEqual(c1.portable_manifest_text(), c2.portable_manifest_text())
1025         c1.apply(d)
1026         self.assertEqual(c1.portable_manifest_text(), c2.portable_manifest_text())
1027
1028     def test_conflict_keep_local_change(self):
1029         c1 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n')
1030         c2 = Collection('. 5348b82a029fd9e971a811ce1f71360b+43 0:10:count2.txt\n')
1031         d = c1.diff(c2)
1032         self.assertEqual(d, [('del', './count1.txt', c1["count1.txt"]),
1033                              ('add', './count2.txt', c2["count2.txt"])])
1034         f = c1.open("count1.txt", "w")
1035         f.write(b"zzzzz")
1036
1037         # c1 changed, so it should not be deleted.
1038         c1.apply(d)
1039         self.assertEqual(c1.portable_manifest_text(), ". 95ebc3c7b3b9f1d2c40fec14415d3cb8+5 5348b82a029fd9e971a811ce1f71360b+43 0:5:count1.txt 5:10:count2.txt\n")
1040
1041     def test_conflict_mod(self):
1042         c1 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt')
1043         c2 = Collection('. 5348b82a029fd9e971a811ce1f71360b+43 0:10:count1.txt')
1044         d = c1.diff(c2)
1045         self.assertEqual(d, [('mod', './count1.txt', c1["count1.txt"], c2["count1.txt"])])
1046         f = c1.open("count1.txt", "w")
1047         f.write(b"zzzzz")
1048
1049         # c1 changed, so c2 mod will go to a conflict file
1050         c1.apply(d)
1051         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~$")
1052
1053     def test_conflict_add(self):
1054         c1 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count2.txt\n')
1055         c2 = Collection('. 5348b82a029fd9e971a811ce1f71360b+43 0:10:count1.txt\n')
1056         d = c1.diff(c2)
1057         self.assertEqual(d, [('del', './count2.txt', c1["count2.txt"]),
1058                              ('add', './count1.txt', c2["count1.txt"])])
1059         f = c1.open("count1.txt", "w")
1060         f.write(b"zzzzz")
1061
1062         # c1 added count1.txt, so c2 add will go to a conflict file
1063         c1.apply(d)
1064         self.assertRegexpMatches(c1.portable_manifest_text(), r"\. 95ebc3c7b3b9f1d2c40fec14415d3cb8\+5 5348b82a029fd9e971a811ce1f71360b\+43 0:5:count1\.txt 5:10:count1\.txt~\d\d\d\d\d\d\d\d-\d\d\d\d\d\d~conflict~$")
1065
1066     def test_conflict_del(self):
1067         c1 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt')
1068         c2 = Collection('. 5348b82a029fd9e971a811ce1f71360b+43 0:10:count1.txt')
1069         d = c1.diff(c2)
1070         self.assertEqual(d, [('mod', './count1.txt', c1["count1.txt"], c2["count1.txt"])])
1071         c1.remove("count1.txt")
1072
1073         # c1 deleted, so c2 mod will go to a conflict file
1074         c1.apply(d)
1075         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~$")
1076
1077     def test_notify(self):
1078         c1 = Collection()
1079         events = []
1080         c1.subscribe(lambda event, collection, name, item: events.append((event, collection, name, item)))
1081         f = c1.open("foo.txt", "w")
1082         self.assertEqual(events[0], (arvados.collection.ADD, c1, "foo.txt", f.arvadosfile))
1083
1084     def test_open_w(self):
1085         c1 = Collection(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n")
1086         self.assertEqual(c1["count1.txt"].size(), 10)
1087         c1.open("count1.txt", "w").close()
1088         self.assertEqual(c1["count1.txt"].size(), 0)
1089
1090
1091 class NewCollectionTestCaseWithServers(run_test_server.TestCaseWithServers):
1092     def test_get_manifest_text_only_committed(self):
1093         c = Collection()
1094         with c.open("count.txt", "w") as f:
1095             # One file committed
1096             with c.open("foo.txt", "w") as foo:
1097                 foo.write(b"foo")
1098                 foo.flush() # Force block commit
1099             f.write(b"0123456789")
1100             # Other file not committed. Block not written to keep yet.
1101             self.assertEqual(
1102                 c._get_manifest_text(".",
1103                                      strip=False,
1104                                      normalize=False,
1105                                      only_committed=True),
1106                 '. acbd18db4cc2f85cedef654fccc4a4d8+3 0:0:count.txt 0:3:foo.txt\n')
1107             # And now with the file closed...
1108             f.flush() # Force block commit
1109         self.assertEqual(
1110             c._get_manifest_text(".",
1111                                  strip=False,
1112                                  normalize=False,
1113                                  only_committed=True),
1114             ". 781e5e245d69b566979b86e28d23f2c7+10 acbd18db4cc2f85cedef654fccc4a4d8+3 0:10:count.txt 10:3:foo.txt\n")
1115
1116     def test_only_small_blocks_are_packed_together(self):
1117         c = Collection()
1118         # Write a couple of small files,
1119         f = c.open("count.txt", "w")
1120         f.write(b"0123456789")
1121         f.close(flush=False)
1122         foo = c.open("foo.txt", "w")
1123         foo.write(b"foo")
1124         foo.close(flush=False)
1125         # Then, write a big file, it shouldn't be packed with the ones above
1126         big = c.open("bigfile.txt", "w")
1127         big.write(b"x" * 1024 * 1024 * 33) # 33 MB > KEEP_BLOCK_SIZE/2
1128         big.close(flush=False)
1129         self.assertEqual(
1130             c.manifest_text("."),
1131             '. 2d303c138c118af809f39319e5d507e9+34603008 a8430a058b8fbf408e1931b794dbd6fb+13 0:34603008:bigfile.txt 34603008:10:count.txt 34603018:3:foo.txt\n')
1132
1133
1134 class CollectionCreateUpdateTest(run_test_server.TestCaseWithServers):
1135     MAIN_SERVER = {}
1136     KEEP_SERVER = {}
1137
1138     def create_count_txt(self):
1139         # Create an empty collection, save it to the API server, then write a
1140         # file, but don't save it.
1141
1142         c = Collection()
1143         c.save_new("CollectionCreateUpdateTest", ensure_unique_name=True)
1144         self.assertEqual(c.portable_data_hash(), "d41d8cd98f00b204e9800998ecf8427e+0")
1145         self.assertEqual(c.api_response()["portable_data_hash"], "d41d8cd98f00b204e9800998ecf8427e+0" )
1146
1147         with c.open("count.txt", "w") as f:
1148             f.write(b"0123456789")
1149
1150         self.assertEqual(c.portable_manifest_text(), ". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n")
1151
1152         return c
1153
1154     def test_create_and_save(self):
1155         c = self.create_count_txt()
1156         c.save()
1157         self.assertRegexpMatches(c.manifest_text(), r"^\. 781e5e245d69b566979b86e28d23f2c7\+10\+A[a-f0-9]{40}@[a-f0-9]{8} 0:10:count\.txt$",)
1158
1159     def test_create_and_save_new(self):
1160         c = self.create_count_txt()
1161         c.save_new()
1162         self.assertRegexpMatches(c.manifest_text(), r"^\. 781e5e245d69b566979b86e28d23f2c7\+10\+A[a-f0-9]{40}@[a-f0-9]{8} 0:10:count\.txt$",)
1163
1164     def test_create_diff_apply(self):
1165         c1 = self.create_count_txt()
1166         c1.save()
1167
1168         c2 = Collection(c1.manifest_locator())
1169         with c2.open("count.txt", "w") as f:
1170             f.write(b"abcdefg")
1171
1172         diff = c1.diff(c2)
1173
1174         self.assertEqual(diff[0], (arvados.collection.MOD, u'./count.txt', c1["count.txt"], c2["count.txt"]))
1175
1176         c1.apply(diff)
1177         self.assertEqual(c1.portable_data_hash(), c2.portable_data_hash())
1178
1179     def test_diff_apply_with_token(self):
1180         baseline = CollectionReader(". 781e5e245d69b566979b86e28d23f2c7+10+A715fd31f8111894f717eb1003c1b0216799dd9ec@54f5dd1a 0:10:count.txt\n")
1181         c = Collection(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n")
1182         other = CollectionReader(". 7ac66c0f148de9519b8bd264312c4d64+7+A715fd31f8111894f717eb1003c1b0216799dd9ec@54f5dd1a 0:7:count.txt\n")
1183
1184         diff = baseline.diff(other)
1185         self.assertEqual(diff, [('mod', u'./count.txt', c["count.txt"], other["count.txt"])])
1186
1187         c.apply(diff)
1188
1189         self.assertEqual(c.manifest_text(), ". 7ac66c0f148de9519b8bd264312c4d64+7+A715fd31f8111894f717eb1003c1b0216799dd9ec@54f5dd1a 0:7:count.txt\n")
1190
1191
1192     def test_create_and_update(self):
1193         c1 = self.create_count_txt()
1194         c1.save()
1195
1196         c2 = arvados.collection.Collection(c1.manifest_locator())
1197         with c2.open("count.txt", "w") as f:
1198             f.write(b"abcdefg")
1199
1200         c2.save()
1201
1202         self.assertNotEqual(c1.portable_data_hash(), c2.portable_data_hash())
1203         c1.update()
1204         self.assertEqual(c1.portable_data_hash(), c2.portable_data_hash())
1205
1206
1207     def test_create_and_update_with_conflict(self):
1208         c1 = self.create_count_txt()
1209         c1.save()
1210
1211         with c1.open("count.txt", "w") as f:
1212             f.write(b"XYZ")
1213
1214         c2 = arvados.collection.Collection(c1.manifest_locator())
1215         with c2.open("count.txt", "w") as f:
1216             f.write(b"abcdefg")
1217
1218         c2.save()
1219
1220         c1.update()
1221         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~$")
1222
1223
1224 if __name__ == '__main__':
1225     unittest.main()