sdk/python/tests/test_collections.py

   1 # Copyright (C) The Arvados Authors. All rights reserved.
   2 #
   3 # SPDX-License-Identifier: Apache-2.0
   4
   5 from __future__ import absolute_import
   6
   7 from builtins import object
   8 import arvados
   9 import copy
  10 import mock
  11 import os
  12 import pprint
  13 import random
  14 import re
  15 import sys
  16 import tempfile
  17 import time
  18 import unittest
  19
  20 from . import run_test_server
  21 from arvados._ranges import Range, LocatorAndRange
  22 from arvados.collection import Collection, CollectionReader
  23 from . import arvados_testutil as tutil
  24
  25 class TestResumableWriter(arvados.ResumableCollectionWriter):
  26     KEEP_BLOCK_SIZE = 1024  # PUT to Keep every 1K.
  27
  28     def current_state(self):
  29         return self.dump_state(copy.deepcopy)
  30
  31
  32 class ArvadosCollectionsTest(run_test_server.TestCaseWithServers,
  33                              tutil.ArvadosBaseTestCase):
  34     MAIN_SERVER = {}
  35
  36     @classmethod
  37     def setUpClass(cls):
  38         super(ArvadosCollectionsTest, cls).setUpClass()
  39         # need admin privileges to make collections with unsigned blocks
  40         run_test_server.authorize_with('admin')
  41         cls.api_client = arvados.api('v1')
  42         cls.keep_client = arvados.KeepClient(api_client=cls.api_client,
  43                                              local_store=cls.local_store)
  44
  45     def write_foo_bar_baz(self):
  46         cw = arvados.CollectionWriter(self.api_client)
  47         self.assertEqual(cw.current_stream_name(), '.',
  48                          'current_stream_name() should be "." now')
  49         cw.set_current_file_name('foo.txt')
  50         cw.write(b'foo')
  51         self.assertEqual(cw.current_file_name(), 'foo.txt',
  52                          'current_file_name() should be foo.txt now')
  53         cw.start_new_file('bar.txt')
  54         cw.write(b'bar')
  55         cw.start_new_stream('baz')
  56         cw.write(b'baz')
  57         cw.set_current_file_name('baz.txt')
  58         self.assertEqual(cw.manifest_text(),
  59                          ". 3858f62230ac3c915f300c664312c63f+6 0:3:foo.txt 3:3:bar.txt\n" +
  60                          "./baz 73feffa4b7f6bb68e44cf984c85f6e88+3 0:3:baz.txt\n",
  61                          "wrong manifest: got {}".format(cw.manifest_text()))
  62         cw.save_new()
  63         return cw.portable_data_hash()
  64
  65     def test_pdh_is_native_str(self):
  66         pdh = self.write_foo_bar_baz()
  67         self.assertEqual(type(''), type(pdh))
  68
  69     def test_keep_local_store(self):
  70         self.assertEqual(self.keep_client.put(b'foo'), 'acbd18db4cc2f85cedef654fccc4a4d8+3', 'wrong md5 hash from Keep.put')
  71         self.assertEqual(self.keep_client.get('acbd18db4cc2f85cedef654fccc4a4d8+3'), b'foo', 'wrong data from Keep.get')
  72
  73     def test_local_collection_writer(self):
  74         self.assertEqual(self.write_foo_bar_baz(),
  75                          '23ca013983d6239e98931cc779e68426+114',
  76                          'wrong locator hash: ' + self.write_foo_bar_baz())
  77
  78     def test_local_collection_reader(self):
  79         foobarbaz = self.write_foo_bar_baz()
  80         cr = arvados.CollectionReader(
  81             foobarbaz + '+Xzizzle', self.api_client)
  82         got = []
  83         for s in cr.all_streams():
  84             for f in s.all_files():
  85                 got += [[f.size(), f.stream_name(), f.name(), f.read(2**26)]]
  86         expected = [[3, '.', 'foo.txt', b'foo'],
  87                     [3, '.', 'bar.txt', b'bar'],
  88                     [3, './baz', 'baz.txt', b'baz']]
  89         self.assertEqual(got,
  90                          expected)
  91         stream0 = cr.all_streams()[0]
  92         self.assertEqual(stream0.readfrom(0, 0),
  93                          b'',
  94                          'reading zero bytes should have returned empty string')
  95         self.assertEqual(stream0.readfrom(0, 2**26),
  96                          b'foobar',
  97                          'reading entire stream failed')
  98         self.assertEqual(stream0.readfrom(2**26, 0),
  99                          b'',
 100                          'reading zero bytes should have returned empty string')
 101         self.assertEqual(3, len(cr))
 102         self.assertTrue(cr)
 103
 104     def _test_subset(self, collection, expected):
 105         cr = arvados.CollectionReader(collection, self.api_client)
 106         for s in cr.all_streams():
 107             for ex in expected:
 108                 if ex[0] == s:
 109                     f = s.files()[ex[2]]
 110                     got = [f.size(), f.stream_name(), f.name(), "".join(f.readall(2**26))]
 111                     self.assertEqual(got,
 112                                      ex,
 113                                      'all_files|as_manifest did not preserve manifest contents: got %s expected %s' % (got, ex))
 114
 115     def test_collection_manifest_subset(self):
 116         foobarbaz = self.write_foo_bar_baz()
 117         self._test_subset(foobarbaz,
 118                           [[3, '.',     'bar.txt', b'bar'],
 119                            [3, '.',     'foo.txt', b'foo'],
 120                            [3, './baz', 'baz.txt', b'baz']])
 121         self._test_subset((". %s %s 0:3:foo.txt 3:3:bar.txt\n" %
 122                            (self.keep_client.put(b"foo"),
 123                             self.keep_client.put(b"bar"))),
 124                           [[3, '.', 'bar.txt', b'bar'],
 125                            [3, '.', 'foo.txt', b'foo']])
 126         self._test_subset((". %s %s 0:2:fo.txt 2:4:obar.txt\n" %
 127                            (self.keep_client.put(b"foo"),
 128                             self.keep_client.put(b"bar"))),
 129                           [[2, '.', 'fo.txt', b'fo'],
 130                            [4, '.', 'obar.txt', b'obar']])
 131         self._test_subset((". %s %s 0:2:fo.txt 2:0:zero.txt 2:2:ob.txt 4:2:ar.txt\n" %
 132                            (self.keep_client.put(b"foo"),
 133                             self.keep_client.put(b"bar"))),
 134                           [[2, '.', 'ar.txt', b'ar'],
 135                            [2, '.', 'fo.txt', b'fo'],
 136                            [2, '.', 'ob.txt', b'ob'],
 137                            [0, '.', 'zero.txt', b'']])
 138
 139     def test_collection_empty_file(self):
 140         cw = arvados.CollectionWriter(self.api_client)
 141         cw.start_new_file('zero.txt')
 142         cw.write(b'')
 143
 144         self.assertEqual(cw.manifest_text(), ". d41d8cd98f00b204e9800998ecf8427e+0 0:0:zero.txt\n")
 145         self.check_manifest_file_sizes(cw.manifest_text(), [0])
 146         cw = arvados.CollectionWriter(self.api_client)
 147         cw.start_new_file('zero.txt')
 148         cw.write(b'')
 149         cw.start_new_file('one.txt')
 150         cw.write(b'1')
 151         cw.start_new_stream('foo')
 152         cw.start_new_file('zero.txt')
 153         cw.write(b'')
 154         self.check_manifest_file_sizes(cw.manifest_text(), [0,1,0])
 155
 156     def test_no_implicit_normalize(self):
 157         cw = arvados.CollectionWriter(self.api_client)
 158         cw.start_new_file('b')
 159         cw.write(b'b')
 160         cw.start_new_file('a')
 161         cw.write(b'')
 162         self.check_manifest_file_sizes(cw.manifest_text(), [1,0])
 163         self.check_manifest_file_sizes(
 164             arvados.CollectionReader(
 165                 cw.manifest_text()).manifest_text(normalize=True),
 166             [0,1])
 167
 168     def check_manifest_file_sizes(self, manifest_text, expect_sizes):
 169         cr = arvados.CollectionReader(manifest_text, self.api_client)
 170         got_sizes = []
 171         for f in cr.all_files():
 172             got_sizes += [f.size()]
 173         self.assertEqual(got_sizes, expect_sizes, "got wrong file sizes %s, expected %s" % (got_sizes, expect_sizes))
 174
 175     def test_normalized_collection(self):
 176         m1 = """. 5348b82a029fd9e971a811ce1f71360b+43 0:43:md5sum.txt
 177 . 085c37f02916da1cad16f93c54d899b7+41 0:41:md5sum.txt
 178 . 8b22da26f9f433dea0a10e5ec66d73ba+43 0:43:md5sum.txt
 179 """
 180         self.assertEqual(arvados.CollectionReader(m1, self.api_client).manifest_text(normalize=True),
 181                          """. 5348b82a029fd9e971a811ce1f71360b+43 085c37f02916da1cad16f93c54d899b7+41 8b22da26f9f433dea0a10e5ec66d73ba+43 0:127:md5sum.txt
 182 """)
 183
 184         m2 = """. 204e43b8a1185621ca55a94839582e6f+67108864 b9677abbac956bd3e86b1deb28dfac03+67108864 fc15aff2a762b13f521baf042140acec+67108864 323d2a3ce20370c4ca1d3462a344f8fd+25885655 0:227212247:var-GS000016015-ASM.tsv.bz2
 185 """
 186         self.assertEqual(arvados.CollectionReader(m2, self.api_client).manifest_text(normalize=True), m2)
 187
 188         m3 = """. 5348b82a029fd9e971a811ce1f71360b+43 3:40:md5sum.txt
 189 . 085c37f02916da1cad16f93c54d899b7+41 0:41:md5sum.txt
 190 . 8b22da26f9f433dea0a10e5ec66d73ba+43 0:43:md5sum.txt
 191 """
 192         self.assertEqual(arvados.CollectionReader(m3, self.api_client).manifest_text(normalize=True),
 193                          """. 5348b82a029fd9e971a811ce1f71360b+43 085c37f02916da1cad16f93c54d899b7+41 8b22da26f9f433dea0a10e5ec66d73ba+43 3:124:md5sum.txt
 194 """)
 195
 196         m4 = """. 204e43b8a1185621ca55a94839582e6f+67108864 0:3:foo/bar
 197 ./zzz 204e43b8a1185621ca55a94839582e6f+67108864 0:999:zzz
 198 ./foo 323d2a3ce20370c4ca1d3462a344f8fd+25885655 0:3:bar
 199 """
 200         self.assertEqual(arvados.CollectionReader(m4, self.api_client).manifest_text(normalize=True),
 201                          """./foo 204e43b8a1185621ca55a94839582e6f+67108864 323d2a3ce20370c4ca1d3462a344f8fd+25885655 0:3:bar 67108864:3:bar
 202 ./zzz 204e43b8a1185621ca55a94839582e6f+67108864 0:999:zzz
 203 """)
 204
 205         m5 = """. 204e43b8a1185621ca55a94839582e6f+67108864 0:3:foo/bar
 206 ./zzz 204e43b8a1185621ca55a94839582e6f+67108864 0:999:zzz
 207 ./foo 204e43b8a1185621ca55a94839582e6f+67108864 3:3:bar
 208 """
 209         self.assertEqual(arvados.CollectionReader(m5, self.api_client).manifest_text(normalize=True),
 210                          """./foo 204e43b8a1185621ca55a94839582e6f+67108864 0:6:bar
 211 ./zzz 204e43b8a1185621ca55a94839582e6f+67108864 0:999:zzz
 212 """)
 213
 214         with self.data_file('1000G_ref_manifest') as f6:
 215             m6 = f6.read()
 216             self.assertEqual(arvados.CollectionReader(m6, self.api_client).manifest_text(normalize=True), m6)
 217
 218         with self.data_file('jlake_manifest') as f7:
 219             m7 = f7.read()
 220             self.assertEqual(arvados.CollectionReader(m7, self.api_client).manifest_text(normalize=True), m7)
 221
 222         m8 = """./a\\040b\\040c 59ca0efa9f5633cb0371bbc0355478d8+13 0:13:hello\\040world.txt
 223 """
 224         self.assertEqual(arvados.CollectionReader(m8, self.api_client).manifest_text(normalize=True), m8)
 225
 226     def test_locators_and_ranges(self):
 227         blocks2 = [Range('a', 0, 10),
 228                    Range('b', 10, 10),
 229                    Range('c', 20, 10),
 230                    Range('d', 30, 10),
 231                    Range('e', 40, 10),
 232                    Range('f', 50, 10)]
 233
 234         self.assertEqual(arvados.locators_and_ranges(blocks2,  2,  2), [LocatorAndRange('a', 10, 2, 2)])
 235         self.assertEqual(arvados.locators_and_ranges(blocks2, 12, 2), [LocatorAndRange('b', 10, 2, 2)])
 236         self.assertEqual(arvados.locators_and_ranges(blocks2, 22, 2), [LocatorAndRange('c', 10, 2, 2)])
 237         self.assertEqual(arvados.locators_and_ranges(blocks2, 32, 2), [LocatorAndRange('d', 10, 2, 2)])
 238         self.assertEqual(arvados.locators_and_ranges(blocks2, 42, 2), [LocatorAndRange('e', 10, 2, 2)])
 239         self.assertEqual(arvados.locators_and_ranges(blocks2, 52, 2), [LocatorAndRange('f', 10, 2, 2)])
 240         self.assertEqual(arvados.locators_and_ranges(blocks2, 62, 2), [])
 241         self.assertEqual(arvados.locators_and_ranges(blocks2, -2, 2), [])
 242
 243         self.assertEqual(arvados.locators_and_ranges(blocks2,  0,  2), [LocatorAndRange('a', 10, 0, 2)])
 244         self.assertEqual(arvados.locators_and_ranges(blocks2, 10, 2), [LocatorAndRange('b', 10, 0, 2)])
 245         self.assertEqual(arvados.locators_and_ranges(blocks2, 20, 2), [LocatorAndRange('c', 10, 0, 2)])
 246         self.assertEqual(arvados.locators_and_ranges(blocks2, 30, 2), [LocatorAndRange('d', 10, 0, 2)])
 247         self.assertEqual(arvados.locators_and_ranges(blocks2, 40, 2), [LocatorAndRange('e', 10, 0, 2)])
 248         self.assertEqual(arvados.locators_and_ranges(blocks2, 50, 2), [LocatorAndRange('f', 10, 0, 2)])
 249         self.assertEqual(arvados.locators_and_ranges(blocks2, 60, 2), [])
 250         self.assertEqual(arvados.locators_and_ranges(blocks2, -2, 2), [])
 251
 252         self.assertEqual(arvados.locators_and_ranges(blocks2,  9,  2), [LocatorAndRange('a', 10, 9, 1), LocatorAndRange('b', 10, 0, 1)])
 253         self.assertEqual(arvados.locators_and_ranges(blocks2, 19, 2), [LocatorAndRange('b', 10, 9, 1), LocatorAndRange('c', 10, 0, 1)])
 254         self.assertEqual(arvados.locators_and_ranges(blocks2, 29, 2), [LocatorAndRange('c', 10, 9, 1), LocatorAndRange('d', 10, 0, 1)])
 255         self.assertEqual(arvados.locators_and_ranges(blocks2, 39, 2), [LocatorAndRange('d', 10, 9, 1), LocatorAndRange('e', 10, 0, 1)])
 256         self.assertEqual(arvados.locators_and_ranges(blocks2, 49, 2), [LocatorAndRange('e', 10, 9, 1), LocatorAndRange('f', 10, 0, 1)])
 257         self.assertEqual(arvados.locators_and_ranges(blocks2, 59, 2), [LocatorAndRange('f', 10, 9, 1)])
 258
 259
 260         blocks3 = [Range('a', 0, 10),
 261                   Range('b', 10, 10),
 262                   Range('c', 20, 10),
 263                   Range('d', 30, 10),
 264                   Range('e', 40, 10),
 265                   Range('f', 50, 10),
 266                    Range('g', 60, 10)]
 267
 268         self.assertEqual(arvados.locators_and_ranges(blocks3,  2,  2), [LocatorAndRange('a', 10, 2, 2)])
 269         self.assertEqual(arvados.locators_and_ranges(blocks3, 12, 2), [LocatorAndRange('b', 10, 2, 2)])
 270         self.assertEqual(arvados.locators_and_ranges(blocks3, 22, 2), [LocatorAndRange('c', 10, 2, 2)])
 271         self.assertEqual(arvados.locators_and_ranges(blocks3, 32, 2), [LocatorAndRange('d', 10, 2, 2)])
 272         self.assertEqual(arvados.locators_and_ranges(blocks3, 42, 2), [LocatorAndRange('e', 10, 2, 2)])
 273         self.assertEqual(arvados.locators_and_ranges(blocks3, 52, 2), [LocatorAndRange('f', 10, 2, 2)])
 274         self.assertEqual(arvados.locators_and_ranges(blocks3, 62, 2), [LocatorAndRange('g', 10, 2, 2)])
 275
 276
 277         blocks = [Range('a', 0, 10),
 278                   Range('b', 10, 15),
 279                   Range('c', 25, 5)]
 280         self.assertEqual(arvados.locators_and_ranges(blocks, 1, 0), [])
 281         self.assertEqual(arvados.locators_and_ranges(blocks, 0, 5), [LocatorAndRange('a', 10, 0, 5)])
 282         self.assertEqual(arvados.locators_and_ranges(blocks, 3, 5), [LocatorAndRange('a', 10, 3, 5)])
 283         self.assertEqual(arvados.locators_and_ranges(blocks, 0, 10), [LocatorAndRange('a', 10, 0, 10)])
 284
 285         self.assertEqual(arvados.locators_and_ranges(blocks, 0, 11), [LocatorAndRange('a', 10, 0, 10),
 286                                                                       LocatorAndRange('b', 15, 0, 1)])
 287         self.assertEqual(arvados.locators_and_ranges(blocks, 1, 11), [LocatorAndRange('a', 10, 1, 9),
 288                                                                       LocatorAndRange('b', 15, 0, 2)])
 289         self.assertEqual(arvados.locators_and_ranges(blocks, 0, 25), [LocatorAndRange('a', 10, 0, 10),
 290                                                                       LocatorAndRange('b', 15, 0, 15)])
 291
 292         self.assertEqual(arvados.locators_and_ranges(blocks, 0, 30), [LocatorAndRange('a', 10, 0, 10),
 293                                                                       LocatorAndRange('b', 15, 0, 15),
 294                                                                       LocatorAndRange('c', 5, 0, 5)])
 295         self.assertEqual(arvados.locators_and_ranges(blocks, 1, 30), [LocatorAndRange('a', 10, 1, 9),
 296                                                                       LocatorAndRange('b', 15, 0, 15),
 297                                                                       LocatorAndRange('c', 5, 0, 5)])
 298         self.assertEqual(arvados.locators_and_ranges(blocks, 0, 31), [LocatorAndRange('a', 10, 0, 10),
 299                                                                       LocatorAndRange('b', 15, 0, 15),
 300                                                                       LocatorAndRange('c', 5, 0, 5)])
 301
 302         self.assertEqual(arvados.locators_and_ranges(blocks, 15, 5), [LocatorAndRange('b', 15, 5, 5)])
 303
 304         self.assertEqual(arvados.locators_and_ranges(blocks, 8, 17), [LocatorAndRange('a', 10, 8, 2),
 305                                                                       LocatorAndRange('b', 15, 0, 15)])
 306
 307         self.assertEqual(arvados.locators_and_ranges(blocks, 8, 20), [LocatorAndRange('a', 10, 8, 2),
 308                                                                       LocatorAndRange('b', 15, 0, 15),
 309                                                                       LocatorAndRange('c', 5, 0, 3)])
 310
 311         self.assertEqual(arvados.locators_and_ranges(blocks, 26, 2), [LocatorAndRange('c', 5, 1, 2)])
 312
 313         self.assertEqual(arvados.locators_and_ranges(blocks, 9, 15), [LocatorAndRange('a', 10, 9, 1),
 314                                                                       LocatorAndRange('b', 15, 0, 14)])
 315         self.assertEqual(arvados.locators_and_ranges(blocks, 10, 15), [LocatorAndRange('b', 15, 0, 15)])
 316         self.assertEqual(arvados.locators_and_ranges(blocks, 11, 15), [LocatorAndRange('b', 15, 1, 14),
 317                                                                        LocatorAndRange('c', 5, 0, 1)])
 318
 319     class MockKeep(object):
 320         def __init__(self, content, num_retries=0):
 321             self.content = content
 322
 323         def get(self, locator, num_retries=0):
 324             return self.content[locator]
 325
 326     def test_stream_reader(self):
 327         keepblocks = {
 328             'aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa+10': b'abcdefghij',
 329             'bbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbb+15': b'klmnopqrstuvwxy',
 330             'cccccccccccccccccccccccccccccccc+5': b'z0123',
 331         }
 332         mk = self.MockKeep(keepblocks)
 333
 334         sr = arvados.StreamReader([".", "aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa+10", "bbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbb+15", "cccccccccccccccccccccccccccccccc+5", "0:30:foo"], mk)
 335
 336         content = b'abcdefghijklmnopqrstuvwxyz0123456789'
 337
 338         self.assertEqual(sr.readfrom(0, 30), content[0:30])
 339         self.assertEqual(sr.readfrom(2, 30), content[2:30])
 340
 341         self.assertEqual(sr.readfrom(2, 8), content[2:10])
 342         self.assertEqual(sr.readfrom(0, 10), content[0:10])
 343
 344         self.assertEqual(sr.readfrom(0, 5), content[0:5])
 345         self.assertEqual(sr.readfrom(5, 5), content[5:10])
 346         self.assertEqual(sr.readfrom(10, 5), content[10:15])
 347         self.assertEqual(sr.readfrom(15, 5), content[15:20])
 348         self.assertEqual(sr.readfrom(20, 5), content[20:25])
 349         self.assertEqual(sr.readfrom(25, 5), content[25:30])
 350         self.assertEqual(sr.readfrom(30, 5), b'')
 351
 352     def test_extract_file(self):
 353         m1 = """. 5348b82a029fd9e971a811ce1f71360b+43 0:43:md5sum.txt
 354 . 085c37f02916da1cad16f93c54d899b7+41 0:41:md6sum.txt
 355 . 8b22da26f9f433dea0a10e5ec66d73ba+43 0:43:md7sum.txt
 356 . 085c37f02916da1cad16f93c54d899b7+41 5348b82a029fd9e971a811ce1f71360b+43 8b22da26f9f433dea0a10e5ec66d73ba+43 47:80:md8sum.txt
 357 . 085c37f02916da1cad16f93c54d899b7+41 5348b82a029fd9e971a811ce1f71360b+43 8b22da26f9f433dea0a10e5ec66d73ba+43 40:80:md9sum.txt
 358 """
 359
 360         m2 = arvados.CollectionReader(m1, self.api_client).manifest_text(normalize=True)
 361
 362         self.assertEqual(m2,
 363                          ". 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")
 364         files = arvados.CollectionReader(
 365             m2, self.api_client).all_streams()[0].files()
 366
 367         self.assertEqual(files['md5sum.txt'].as_manifest(),
 368                          ". 5348b82a029fd9e971a811ce1f71360b+43 0:43:md5sum.txt\n")
 369         self.assertEqual(files['md6sum.txt'].as_manifest(),
 370                          ". 085c37f02916da1cad16f93c54d899b7+41 0:41:md6sum.txt\n")
 371         self.assertEqual(files['md7sum.txt'].as_manifest(),
 372                          ". 8b22da26f9f433dea0a10e5ec66d73ba+43 0:43:md7sum.txt\n")
 373         self.assertEqual(files['md9sum.txt'].as_manifest(),
 374                          ". 085c37f02916da1cad16f93c54d899b7+41 5348b82a029fd9e971a811ce1f71360b+43 8b22da26f9f433dea0a10e5ec66d73ba+43 40:80:md9sum.txt\n")
 375
 376     def test_write_directory_tree(self):
 377         cwriter = arvados.CollectionWriter(self.api_client)
 378         cwriter.write_directory_tree(self.build_directory_tree(
 379                 ['basefile', 'subdir/subfile']))
 380         self.assertEqual(cwriter.manifest_text(),
 381                          """. c5110c5ac93202d8e0f9e381f22bac0f+8 0:8:basefile
 382 ./subdir 1ca4dec89403084bf282ad31e6cf7972+14 0:14:subfile\n""")
 383
 384     def test_write_named_directory_tree(self):
 385         cwriter = arvados.CollectionWriter(self.api_client)
 386         cwriter.write_directory_tree(self.build_directory_tree(
 387                 ['basefile', 'subdir/subfile']), 'root')
 388         self.assertEqual(
 389             cwriter.manifest_text(),
 390             """./root c5110c5ac93202d8e0f9e381f22bac0f+8 0:8:basefile
 391 ./root/subdir 1ca4dec89403084bf282ad31e6cf7972+14 0:14:subfile\n""")
 392
 393     def test_write_directory_tree_in_one_stream(self):
 394         cwriter = arvados.CollectionWriter(self.api_client)
 395         cwriter.write_directory_tree(self.build_directory_tree(
 396                 ['basefile', 'subdir/subfile']), max_manifest_depth=0)
 397         self.assertEqual(cwriter.manifest_text(),
 398                          """. 4ace875ffdc6824a04950f06858f4465+22 0:8:basefile 8:14:subdir/subfile\n""")
 399
 400     def test_write_directory_tree_with_limited_recursion(self):
 401         cwriter = arvados.CollectionWriter(self.api_client)
 402         cwriter.write_directory_tree(
 403             self.build_directory_tree(['f1', 'd1/f2', 'd1/d2/f3']),
 404             max_manifest_depth=1)
 405         self.assertEqual(cwriter.manifest_text(),
 406                          """. bd19836ddb62c11c55ab251ccaca5645+2 0:2:f1
 407 ./d1 50170217e5b04312024aa5cd42934494+13 0:8:d2/f3 8:5:f2\n""")
 408
 409     def test_write_directory_tree_with_zero_recursion(self):
 410         cwriter = arvados.CollectionWriter(self.api_client)
 411         content = 'd1/d2/f3d1/f2f1'
 412         blockhash = tutil.str_keep_locator(content)
 413         cwriter.write_directory_tree(
 414             self.build_directory_tree(['f1', 'd1/f2', 'd1/d2/f3']),
 415             max_manifest_depth=0)
 416         self.assertEqual(
 417             cwriter.manifest_text(),
 418             ". {} 0:8:d1/d2/f3 8:5:d1/f2 13:2:f1\n".format(blockhash))
 419
 420     def test_write_one_file(self):
 421         cwriter = arvados.CollectionWriter(self.api_client)
 422         with self.make_test_file() as testfile:
 423             cwriter.write_file(testfile.name)
 424             self.assertEqual(
 425                 cwriter.manifest_text(),
 426                 ". 098f6bcd4621d373cade4e832627b4f6+4 0:4:{}\n".format(
 427                     os.path.basename(testfile.name)))
 428
 429     def test_write_named_file(self):
 430         cwriter = arvados.CollectionWriter(self.api_client)
 431         with self.make_test_file() as testfile:
 432             cwriter.write_file(testfile.name, 'foo')
 433             self.assertEqual(cwriter.manifest_text(),
 434                              ". 098f6bcd4621d373cade4e832627b4f6+4 0:4:foo\n")
 435
 436     def test_write_multiple_files(self):
 437         cwriter = arvados.CollectionWriter(self.api_client)
 438         for letter in 'ABC':
 439             with self.make_test_file(letter.encode()) as testfile:
 440                 cwriter.write_file(testfile.name, letter)
 441         self.assertEqual(
 442             cwriter.manifest_text(),
 443             ". 902fbdd2b1df0c4f70b4a5d23525e932+3 0:1:A 1:1:B 2:1:C\n")
 444
 445     def test_basic_resume(self):
 446         cwriter = TestResumableWriter()
 447         with self.make_test_file() as testfile:
 448             cwriter.write_file(testfile.name, 'test')
 449             resumed = TestResumableWriter.from_state(cwriter.current_state())
 450         self.assertEqual(cwriter.manifest_text(), resumed.manifest_text(),
 451                           "resumed CollectionWriter had different manifest")
 452
 453     def test_resume_fails_when_missing_dependency(self):
 454         cwriter = TestResumableWriter()
 455         with self.make_test_file() as testfile:
 456             cwriter.write_file(testfile.name, 'test')
 457         self.assertRaises(arvados.errors.StaleWriterStateError,
 458                           TestResumableWriter.from_state,
 459                           cwriter.current_state())
 460
 461     def test_resume_fails_when_dependency_mtime_changed(self):
 462         cwriter = TestResumableWriter()
 463         with self.make_test_file() as testfile:
 464             cwriter.write_file(testfile.name, 'test')
 465             os.utime(testfile.name, (0, 0))
 466             self.assertRaises(arvados.errors.StaleWriterStateError,
 467                               TestResumableWriter.from_state,
 468                               cwriter.current_state())
 469
 470     def test_resume_fails_when_dependency_is_nonfile(self):
 471         cwriter = TestResumableWriter()
 472         cwriter.write_file('/dev/null', 'empty')
 473         self.assertRaises(arvados.errors.StaleWriterStateError,
 474                           TestResumableWriter.from_state,
 475                           cwriter.current_state())
 476
 477     def test_resume_fails_when_dependency_size_changed(self):
 478         cwriter = TestResumableWriter()
 479         with self.make_test_file() as testfile:
 480             cwriter.write_file(testfile.name, 'test')
 481             orig_mtime = os.fstat(testfile.fileno()).st_mtime
 482             testfile.write(b'extra')
 483             testfile.flush()
 484             os.utime(testfile.name, (orig_mtime, orig_mtime))
 485             self.assertRaises(arvados.errors.StaleWriterStateError,
 486                               TestResumableWriter.from_state,
 487                               cwriter.current_state())
 488
 489     def test_resume_fails_with_expired_locator(self):
 490         cwriter = TestResumableWriter()
 491         state = cwriter.current_state()
 492         # Add an expired locator to the state.
 493         state['_current_stream_locators'].append(''.join([
 494                     'a' * 32, '+1+A', 'b' * 40, '@', '10000000']))
 495         self.assertRaises(arvados.errors.StaleWriterStateError,
 496                           TestResumableWriter.from_state, state)
 497
 498     def test_arbitrary_objects_not_resumable(self):
 499         cwriter = TestResumableWriter()
 500         with open('/dev/null') as badfile:
 501             self.assertRaises(arvados.errors.AssertionError,
 502                               cwriter.write_file, badfile)
 503
 504     def test_arbitrary_writes_not_resumable(self):
 505         cwriter = TestResumableWriter()
 506         self.assertRaises(arvados.errors.AssertionError,
 507                           cwriter.write, "badtext")
 508
 509
 510 class CollectionTestMixin(tutil.ApiClientMock):
 511     API_COLLECTIONS = run_test_server.fixture('collections')
 512     DEFAULT_COLLECTION = API_COLLECTIONS['foo_file']
 513     DEFAULT_DATA_HASH = DEFAULT_COLLECTION['portable_data_hash']
 514     DEFAULT_MANIFEST = DEFAULT_COLLECTION['manifest_text']
 515     DEFAULT_UUID = DEFAULT_COLLECTION['uuid']
 516     ALT_COLLECTION = API_COLLECTIONS['bar_file']
 517     ALT_DATA_HASH = ALT_COLLECTION['portable_data_hash']
 518     ALT_MANIFEST = ALT_COLLECTION['manifest_text']
 519
 520     def api_client_mock(self, status=200):
 521         client = super(CollectionTestMixin, self).api_client_mock()
 522         self.mock_keep_services(client, status=status, service_type='proxy', count=1)
 523         return client
 524
 525
 526 @tutil.skip_sleep
 527 class CollectionReaderTestCase(unittest.TestCase, CollectionTestMixin):
 528     def mock_get_collection(self, api_mock, code, fixturename):
 529         body = self.API_COLLECTIONS.get(fixturename)
 530         self._mock_api_call(api_mock.collections().get, code, body)
 531
 532     def api_client_mock(self, status=200):
 533         client = super(CollectionReaderTestCase, self).api_client_mock()
 534         self.mock_get_collection(client, status, 'foo_file')
 535         return client
 536
 537     def test_init_no_default_retries(self):
 538         client = self.api_client_mock(200)
 539         reader = arvados.CollectionReader(self.DEFAULT_UUID, api_client=client)
 540         reader.manifest_text()
 541         client.collections().get().execute.assert_called_with(num_retries=0)
 542
 543     def test_uuid_init_success(self):
 544         client = self.api_client_mock(200)
 545         reader = arvados.CollectionReader(self.DEFAULT_UUID, api_client=client,
 546                                           num_retries=3)
 547         self.assertEqual(self.DEFAULT_COLLECTION['manifest_text'],
 548                          reader.manifest_text())
 549         client.collections().get().execute.assert_called_with(num_retries=3)
 550
 551     def test_uuid_init_failure_raises_api_error(self):
 552         client = self.api_client_mock(500)
 553         with self.assertRaises(arvados.errors.ApiError):
 554             reader = arvados.CollectionReader(self.DEFAULT_UUID, api_client=client)
 555
 556     def test_locator_init(self):
 557         client = self.api_client_mock(200)
 558         # Ensure Keep will not return anything if asked.
 559         with tutil.mock_keep_responses(None, 404):
 560             reader = arvados.CollectionReader(self.DEFAULT_DATA_HASH,
 561                                               api_client=client)
 562             self.assertEqual(self.DEFAULT_MANIFEST, reader.manifest_text())
 563
 564     def test_init_no_fallback_to_keep(self):
 565         # Do not look up a collection UUID or PDH in Keep.
 566         for key in [self.DEFAULT_UUID, self.DEFAULT_DATA_HASH]:
 567             client = self.api_client_mock(404)
 568             with tutil.mock_keep_responses(self.DEFAULT_MANIFEST, 200):
 569                 with self.assertRaises(arvados.errors.ApiError):
 570                     reader = arvados.CollectionReader(key, api_client=client)
 571
 572     def test_init_num_retries_propagated(self):
 573         # More of an integration test...
 574         client = self.api_client_mock(200)
 575         reader = arvados.CollectionReader(self.DEFAULT_UUID, api_client=client,
 576                                           num_retries=3)
 577         with tutil.mock_keep_responses('foo', 500, 500, 200):
 578             self.assertEqual(b'foo',
 579                              b''.join(f.read(9) for f in reader.all_files()))
 580
 581     def test_read_nonnormalized_manifest_with_collection_reader(self):
 582         # client should be able to use CollectionReader on a manifest without normalizing it
 583         client = self.api_client_mock(500)
 584         nonnormal = ". acbd18db4cc2f85cedef654fccc4a4d8+3+Aabadbadbee@abeebdee 0:3:foo.txt 1:0:bar.txt 0:3:foo.txt\n"
 585         reader = arvados.CollectionReader(
 586             nonnormal,
 587             api_client=client, num_retries=0)
 588         # Ensure stripped_manifest() doesn't mangle our manifest in
 589         # any way other than stripping hints.
 590         self.assertEqual(
 591             re.sub('\+[^\d\s\+]+', '', nonnormal),
 592             reader.stripped_manifest())
 593         # Ensure stripped_manifest() didn't mutate our reader.
 594         self.assertEqual(nonnormal, reader.manifest_text())
 595         # Ensure the files appear in the order given in the manifest.
 596         self.assertEqual(
 597             [[6, '.', 'foo.txt'],
 598              [0, '.', 'bar.txt']],
 599             [[f.size(), f.stream_name(), f.name()]
 600              for f in reader.all_streams()[0].all_files()])
 601
 602     def test_read_empty_collection(self):
 603         client = self.api_client_mock(200)
 604         self.mock_get_collection(client, 200, 'empty')
 605         reader = arvados.CollectionReader('d41d8cd98f00b204e9800998ecf8427e+0',
 606                                           api_client=client)
 607         self.assertEqual('', reader.manifest_text())
 608         self.assertEqual(0, len(reader))
 609         self.assertFalse(reader)
 610
 611     def test_api_response(self):
 612         client = self.api_client_mock()
 613         reader = arvados.CollectionReader(self.DEFAULT_UUID, api_client=client)
 614         self.assertEqual(self.DEFAULT_COLLECTION, reader.api_response())
 615
 616     def check_open_file(self, coll_file, stream_name, file_name, file_size):
 617         self.assertFalse(coll_file.closed, "returned file is not open")
 618         self.assertEqual(stream_name, coll_file.stream_name())
 619         self.assertEqual(file_name, coll_file.name)
 620         self.assertEqual(file_size, coll_file.size())
 621
 622     def test_open_collection_file_one_argument(self):
 623         client = self.api_client_mock(200)
 624         reader = arvados.CollectionReader(self.DEFAULT_UUID, api_client=client)
 625         cfile = reader.open('./foo', 'rb')
 626         self.check_open_file(cfile, '.', 'foo', 3)
 627
 628     def test_open_deep_file(self):
 629         coll_name = 'collection_with_files_in_subdir'
 630         client = self.api_client_mock(200)
 631         self.mock_get_collection(client, 200, coll_name)
 632         reader = arvados.CollectionReader(
 633             self.API_COLLECTIONS[coll_name]['uuid'], api_client=client)
 634         cfile = reader.open('./subdir2/subdir3/file2_in_subdir3.txt', 'rb')
 635         self.check_open_file(cfile, './subdir2/subdir3', 'file2_in_subdir3.txt',
 636                              32)
 637
 638     def test_open_nonexistent_stream(self):
 639         client = self.api_client_mock(200)
 640         reader = arvados.CollectionReader(self.DEFAULT_UUID, api_client=client)
 641         self.assertRaises(IOError, reader.open, './nonexistent/foo')
 642
 643     def test_open_nonexistent_file(self):
 644         client = self.api_client_mock(200)
 645         reader = arvados.CollectionReader(self.DEFAULT_UUID, api_client=client)
 646         self.assertRaises(IOError, reader.open, 'nonexistent')
 647
 648
 649 @tutil.skip_sleep
 650 class CollectionWriterTestCase(unittest.TestCase, CollectionTestMixin):
 651     def mock_keep(self, body, *codes, **headers):
 652         headers.setdefault('x-keep-replicas-stored', 2)
 653         return tutil.mock_keep_responses(body, *codes, **headers)
 654
 655     def foo_writer(self, **kwargs):
 656         kwargs.setdefault('api_client', self.api_client_mock())
 657         writer = arvados.CollectionWriter(**kwargs)
 658         writer.start_new_file('foo')
 659         writer.write(b'foo')
 660         return writer
 661
 662     def test_write_whole_collection(self):
 663         writer = self.foo_writer()
 664         with self.mock_keep(self.DEFAULT_DATA_HASH, 200, 200):
 665             self.assertEqual(self.DEFAULT_DATA_HASH, writer.finish())
 666
 667     def test_write_no_default(self):
 668         writer = self.foo_writer()
 669         with self.mock_keep(None, 500):
 670             with self.assertRaises(arvados.errors.KeepWriteError):
 671                 writer.finish()
 672
 673     def test_write_insufficient_replicas_via_proxy(self):
 674         writer = self.foo_writer(replication=3)
 675         with self.mock_keep(None, 200, **{'x-keep-replicas-stored': 2}):
 676             with self.assertRaises(arvados.errors.KeepWriteError):
 677                 writer.manifest_text()
 678
 679     def test_write_insufficient_replicas_via_disks(self):
 680         client = mock.MagicMock(name='api_client')
 681         with self.mock_keep(
 682                 None, 200, 200,
 683                 **{'x-keep-replicas-stored': 1}) as keepmock:
 684             self.mock_keep_services(client, status=200, service_type='disk', count=2)
 685             writer = self.foo_writer(api_client=client, replication=3)
 686             with self.assertRaises(arvados.errors.KeepWriteError):
 687                 writer.manifest_text()
 688
 689     def test_write_three_replicas(self):
 690         client = mock.MagicMock(name='api_client')
 691         with self.mock_keep(
 692                 "", 500, 500, 500, 200, 200, 200,
 693                 **{'x-keep-replicas-stored': 1}) as keepmock:
 694             self.mock_keep_services(client, status=200, service_type='disk', count=6)
 695             writer = self.foo_writer(api_client=client, replication=3)
 696             writer.manifest_text()
 697             self.assertEqual(6, keepmock.call_count)
 698
 699     def test_write_whole_collection_through_retries(self):
 700         writer = self.foo_writer(num_retries=2)
 701         with self.mock_keep(self.DEFAULT_DATA_HASH,
 702                             500, 500, 200, 500, 500, 200):
 703             self.assertEqual(self.DEFAULT_DATA_HASH, writer.finish())
 704
 705     def test_flush_data_retries(self):
 706         writer = self.foo_writer(num_retries=2)
 707         foo_hash = self.DEFAULT_MANIFEST.split()[1]
 708         with self.mock_keep(foo_hash, 500, 200):
 709             writer.flush_data()
 710         self.assertEqual(self.DEFAULT_MANIFEST, writer.manifest_text())
 711
 712     def test_one_open(self):
 713         client = self.api_client_mock()
 714         writer = arvados.CollectionWriter(client)
 715         with writer.open('out') as out_file:
 716             self.assertEqual('.', writer.current_stream_name())
 717             self.assertEqual('out', writer.current_file_name())
 718             out_file.write(b'test data')
 719             data_loc = tutil.str_keep_locator('test data')
 720         self.assertTrue(out_file.closed, "writer file not closed after context")
 721         self.assertRaises(ValueError, out_file.write, 'extra text')
 722         with self.mock_keep(data_loc, 200) as keep_mock:
 723             self.assertEqual(". {} 0:9:out\n".format(data_loc),
 724                              writer.manifest_text())
 725
 726     def test_open_writelines(self):
 727         client = self.api_client_mock()
 728         writer = arvados.CollectionWriter(client)
 729         with writer.open('six') as out_file:
 730             out_file.writelines(['12', '34', '56'])
 731             data_loc = tutil.str_keep_locator('123456')
 732         with self.mock_keep(data_loc, 200) as keep_mock:
 733             self.assertEqual(". {} 0:6:six\n".format(data_loc),
 734                              writer.manifest_text())
 735
 736     def test_open_flush(self):
 737         client = self.api_client_mock()
 738         data_loc1 = tutil.str_keep_locator('flush1')
 739         data_loc2 = tutil.str_keep_locator('flush2')
 740         with self.mock_keep((data_loc1, 200), (data_loc2, 200)) as keep_mock:
 741             writer = arvados.CollectionWriter(client)
 742             with writer.open('flush_test') as out_file:
 743                 out_file.write(b'flush1')
 744                 out_file.flush()
 745                 out_file.write(b'flush2')
 746             self.assertEqual(". {} {} 0:12:flush_test\n".format(data_loc1,
 747                                                                 data_loc2),
 748                              writer.manifest_text())
 749
 750     def test_two_opens_same_stream(self):
 751         client = self.api_client_mock()
 752         writer = arvados.CollectionWriter(client)
 753         with writer.open('.', '1') as out_file:
 754             out_file.write(b'1st')
 755         with writer.open('.', '2') as out_file:
 756             out_file.write(b'2nd')
 757         data_loc = tutil.str_keep_locator('1st2nd')
 758         with self.mock_keep(data_loc, 200) as keep_mock:
 759             self.assertEqual(". {} 0:3:1 3:3:2\n".format(data_loc),
 760                              writer.manifest_text())
 761
 762     def test_two_opens_two_streams(self):
 763         client = self.api_client_mock()
 764         data_loc1 = tutil.str_keep_locator('file')
 765         data_loc2 = tutil.str_keep_locator('indir')
 766         with self.mock_keep((data_loc1, 200), (data_loc2, 200)) as keep_mock:
 767             writer = arvados.CollectionWriter(client)
 768             with writer.open('file') as out_file:
 769                 out_file.write(b'file')
 770             with writer.open('./dir', 'indir') as out_file:
 771                 out_file.write(b'indir')
 772             expected = ". {} 0:4:file\n./dir {} 0:5:indir\n".format(
 773                 data_loc1, data_loc2)
 774             self.assertEqual(expected, writer.manifest_text())
 775
 776     def test_dup_open_fails(self):
 777         client = self.api_client_mock()
 778         writer = arvados.CollectionWriter(client)
 779         file1 = writer.open('one')
 780         self.assertRaises(arvados.errors.AssertionError, writer.open, 'two')
 781
 782
 783 class CollectionMethods(run_test_server.TestCaseWithServers):
 784
 785     def test_keys_values_items_support_indexing(self):
 786         c = Collection()
 787         with c.open('foo', 'wb') as f:
 788             f.write(b'foo')
 789         with c.open('bar', 'wb') as f:
 790             f.write(b'bar')
 791         self.assertEqual(2, len(c.keys()))
 792         if sys.version_info < (3, 0):
 793             # keys() supports indexing only for python2 callers.
 794             fn0 = c.keys()[0]
 795             fn1 = c.keys()[1]
 796         else:
 797             fn0, fn1 = c.keys()
 798         self.assertEqual(2, len(c.values()))
 799         f0 = c.values()[0]
 800         f1 = c.values()[1]
 801         self.assertEqual(2, len(c.items()))
 802         self.assertEqual(fn0, c.items()[0][0])
 803         self.assertEqual(fn1, c.items()[1][0])
 804
 805
 806 class CollectionOpenModes(run_test_server.TestCaseWithServers):
 807
 808     def test_open_binary_modes(self):
 809         c = Collection()
 810         for mode in ['wb', 'wb+', 'ab', 'ab+']:
 811             with c.open('foo', mode) as f:
 812                 f.write(b'foo')
 813
 814     def test_open_invalid_modes(self):
 815         c = Collection()
 816         for mode in ['+r', 'aa', '++', 'r+b', 'beer', '', None]:
 817             with self.assertRaises(Exception):
 818                 c.open('foo', mode)
 819
 820     def test_open_text_modes(self):
 821         c = Collection()
 822         with c.open('foo', 'wb') as f:
 823             f.write('foo')
 824         for mode in ['r', 'rt', 'r+', 'rt+', 'w', 'wt', 'a', 'at']:
 825             if sys.version_info >= (3, 0):
 826                 with self.assertRaises(NotImplementedError):
 827                     c.open('foo', mode)
 828             else:
 829                 with c.open('foo', mode) as f:
 830                     if mode[0] == 'r' and '+' not in mode:
 831                         self.assertEqual('foo', f.read(3))
 832                     else:
 833                         f.write('bar')
 834                         f.seek(-3, os.SEEK_CUR)
 835                         self.assertEqual('bar', f.read(3))
 836
 837
 838 class NewCollectionTestCase(unittest.TestCase, CollectionTestMixin):
 839
 840     def test_replication_desired_kept_on_load(self):
 841         m = '. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt 0:10:count2.txt\n'
 842         c1 = Collection(m, replication_desired=1)
 843         c1.save_new()
 844         loc = c1.manifest_locator()
 845         c2 = Collection(loc)
 846         self.assertEqual(c1.manifest_text, c2.manifest_text)
 847         self.assertEqual(c1.replication_desired, c2.replication_desired)
 848
 849     def test_replication_desired_not_loaded_if_provided(self):
 850         m = '. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt 0:10:count2.txt\n'
 851         c1 = Collection(m, replication_desired=1)
 852         c1.save_new()
 853         loc = c1.manifest_locator()
 854         c2 = Collection(loc, replication_desired=2)
 855         self.assertEqual(c1.manifest_text, c2.manifest_text)
 856         self.assertNotEqual(c1.replication_desired, c2.replication_desired)
 857
 858     def test_init_manifest(self):
 859         m1 = """. 5348b82a029fd9e971a811ce1f71360b+43 0:43:md5sum.txt
 860 . 085c37f02916da1cad16f93c54d899b7+41 0:41:md5sum.txt
 861 . 8b22da26f9f433dea0a10e5ec66d73ba+43 0:43:md5sum.txt
 862 """
 863         self.assertEqual(m1, CollectionReader(m1).manifest_text(normalize=False))
 864         self.assertEqual(". 5348b82a029fd9e971a811ce1f71360b+43 085c37f02916da1cad16f93c54d899b7+41 8b22da26f9f433dea0a10e5ec66d73ba+43 0:127:md5sum.txt\n", CollectionReader(m1).manifest_text(normalize=True))
 865
 866     def test_init_manifest_with_collision(self):
 867         m1 = """. 5348b82a029fd9e971a811ce1f71360b+43 0:43:md5sum.txt
 868 ./md5sum.txt 085c37f02916da1cad16f93c54d899b7+41 0:41:md5sum.txt
 869 """
 870         with self.assertRaises(arvados.errors.ArgumentError):
 871             self.assertEqual(m1, CollectionReader(m1))
 872
 873     def test_init_manifest_with_error(self):
 874         m1 = """. 0:43:md5sum.txt"""
 875         with self.assertRaises(arvados.errors.ArgumentError):
 876             self.assertEqual(m1, CollectionReader(m1))
 877
 878     def test_remove(self):
 879         c = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt 0:10:count2.txt\n')
 880         self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt 0:10:count2.txt\n", c.portable_manifest_text())
 881         self.assertIn("count1.txt", c)
 882         c.remove("count1.txt")
 883         self.assertNotIn("count1.txt", c)
 884         self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count2.txt\n", c.portable_manifest_text())
 885         with self.assertRaises(arvados.errors.ArgumentError):
 886             c.remove("")
 887
 888     def test_find(self):
 889         c = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt 0:10:count2.txt\n')
 890         self.assertIs(c.find("."), c)
 891         self.assertIs(c.find("./count1.txt"), c["count1.txt"])
 892         self.assertIs(c.find("count1.txt"), c["count1.txt"])
 893         with self.assertRaises(IOError):
 894             c.find("/.")
 895         with self.assertRaises(arvados.errors.ArgumentError):
 896             c.find("")
 897         self.assertIs(c.find("./nonexistant.txt"), None)
 898         self.assertIs(c.find("./nonexistantsubdir/nonexistant.txt"), None)
 899
 900     def test_remove_in_subdir(self):
 901         c = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n./foo 781e5e245d69b566979b86e28d23f2c7+10 0:10:count2.txt\n')
 902         c.remove("foo/count2.txt")
 903         self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n", c.portable_manifest_text())
 904
 905     def test_remove_empty_subdir(self):
 906         c = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n./foo 781e5e245d69b566979b86e28d23f2c7+10 0:10:count2.txt\n')
 907         c.remove("foo/count2.txt")
 908         c.remove("foo")
 909         self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n", c.portable_manifest_text())
 910
 911     def test_remove_nonempty_subdir(self):
 912         c = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n./foo 781e5e245d69b566979b86e28d23f2c7+10 0:10:count2.txt\n')
 913         with self.assertRaises(IOError):
 914             c.remove("foo")
 915         c.remove("foo", recursive=True)
 916         self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n", c.portable_manifest_text())
 917
 918     def test_copy_to_file_in_dir(self):
 919         c = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n')
 920         c.copy("count1.txt", "foo/count2.txt")
 921         self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n./foo 781e5e245d69b566979b86e28d23f2c7+10 0:10:count2.txt\n", c.portable_manifest_text())
 922
 923     def test_copy_file(self):
 924         c = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n')
 925         c.copy("count1.txt", "count2.txt")
 926         self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt 0:10:count2.txt\n", c.portable_manifest_text())
 927
 928     def test_copy_to_existing_dir(self):
 929         c = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n./foo 781e5e245d69b566979b86e28d23f2c7+10 0:10:count2.txt\n')
 930         c.copy("count1.txt", "foo")
 931         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())
 932
 933     def test_copy_to_new_dir(self):
 934         c = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n')
 935         c.copy("count1.txt", "foo/")
 936         self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n./foo 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n", c.portable_manifest_text())
 937
 938     def test_rename_file(self):
 939         c = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n')
 940         c.rename("count1.txt", "count2.txt")
 941         self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count2.txt\n", c.manifest_text())
 942
 943     def test_move_file_to_dir(self):
 944         c = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n')
 945         c.mkdirs("foo")
 946         c.rename("count1.txt", "foo/count2.txt")
 947         self.assertEqual("./foo 781e5e245d69b566979b86e28d23f2c7+10 0:10:count2.txt\n", c.manifest_text())
 948
 949     def test_move_file_to_other(self):
 950         c1 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n')
 951         c2 = Collection()
 952         c2.rename("count1.txt", "count2.txt", source_collection=c1)
 953         self.assertEqual("", c1.manifest_text())
 954         self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count2.txt\n", c2.manifest_text())
 955
 956     def test_clone(self):
 957         c = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n./foo 781e5e245d69b566979b86e28d23f2c7+10 0:10:count2.txt\n')
 958         cl = c.clone()
 959         self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n./foo 781e5e245d69b566979b86e28d23f2c7+10 0:10:count2.txt\n", cl.portable_manifest_text())
 960
 961     def test_diff_del_add(self):
 962         c1 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n')
 963         c2 = Collection('. 5348b82a029fd9e971a811ce1f71360b+43 0:10:count2.txt\n')
 964         d = c2.diff(c1)
 965         self.assertEqual(sorted(d), [
 966             ('add', './count1.txt', c1["count1.txt"]),
 967             ('del', './count2.txt', c2["count2.txt"]),
 968         ])
 969         d = c1.diff(c2)
 970         self.assertEqual(sorted(d), [
 971             ('add', './count2.txt', c2["count2.txt"]),
 972             ('del', './count1.txt', c1["count1.txt"]),
 973         ])
 974         self.assertNotEqual(c1.portable_manifest_text(), c2.portable_manifest_text())
 975         c1.apply(d)
 976         self.assertEqual(c1.portable_manifest_text(), c2.portable_manifest_text())
 977
 978     def test_diff_same(self):
 979         c1 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n')
 980         c2 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n')
 981         d = c2.diff(c1)
 982         self.assertEqual(d, [('tok', './count1.txt', c2["count1.txt"], c1["count1.txt"])])
 983         d = c1.diff(c2)
 984         self.assertEqual(d, [('tok', './count1.txt', c2["count1.txt"], c1["count1.txt"])])
 985
 986         self.assertEqual(c1.portable_manifest_text(), c2.portable_manifest_text())
 987         c1.apply(d)
 988         self.assertEqual(c1.portable_manifest_text(), c2.portable_manifest_text())
 989
 990     def test_diff_mod(self):
 991         c1 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n')
 992         c2 = Collection('. 5348b82a029fd9e971a811ce1f71360b+43 0:10:count1.txt\n')
 993         d = c2.diff(c1)
 994         self.assertEqual(d, [('mod', './count1.txt', c2["count1.txt"], c1["count1.txt"])])
 995         d = c1.diff(c2)
 996         self.assertEqual(d, [('mod', './count1.txt', c1["count1.txt"], c2["count1.txt"])])
 997
 998         self.assertNotEqual(c1.portable_manifest_text(), c2.portable_manifest_text())
 999         c1.apply(d)
1000         self.assertEqual(c1.portable_manifest_text(), c2.portable_manifest_text())
1001
1002     def test_diff_add(self):
1003         c1 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n')
1004         c2 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 5348b82a029fd9e971a811ce1f71360b+43 0:10:count1.txt 10:20:count2.txt\n')
1005         d = c2.diff(c1)
1006         self.assertEqual(sorted(d), [
1007             ('del', './count2.txt', c2["count2.txt"]),
1008             ('tok', './count1.txt', c2["count1.txt"], c1["count1.txt"]),
1009         ])
1010         d = c1.diff(c2)
1011         self.assertEqual(sorted(d), [
1012             ('add', './count2.txt', c2["count2.txt"]),
1013             ('tok', './count1.txt', c2["count1.txt"], c1["count1.txt"]),
1014         ])
1015
1016         self.assertNotEqual(c1.portable_manifest_text(), c2.portable_manifest_text())
1017         c1.apply(d)
1018         self.assertEqual(c1.portable_manifest_text(), c2.portable_manifest_text())
1019
1020     def test_diff_add_in_subcollection(self):
1021         c1 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n')
1022         c2 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n./foo 5348b82a029fd9e971a811ce1f71360b+43 0:10:count2.txt\n')
1023         d = c2.diff(c1)
1024         self.assertEqual(sorted(d), [
1025             ('del', './foo', c2["foo"]),
1026             ('tok', './count1.txt', c2["count1.txt"], c1["count1.txt"]),
1027         ])
1028         d = c1.diff(c2)
1029         self.assertEqual(sorted(d), [
1030             ('add', './foo', c2["foo"]),
1031             ('tok', './count1.txt', c2["count1.txt"], c1["count1.txt"]),
1032         ])
1033         self.assertNotEqual(c1.portable_manifest_text(), c2.portable_manifest_text())
1034         c1.apply(d)
1035         self.assertEqual(c1.portable_manifest_text(), c2.portable_manifest_text())
1036
1037     def test_diff_del_add_in_subcollection(self):
1038         c1 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n./foo 5348b82a029fd9e971a811ce1f71360b+43 0:10:count2.txt\n')
1039         c2 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n./foo 5348b82a029fd9e971a811ce1f71360b+43 0:3:count3.txt\n')
1040         d = c2.diff(c1)
1041         self.assertEqual(sorted(d), [
1042             ('add', './foo/count2.txt', c1.find("foo/count2.txt")),
1043             ('del', './foo/count3.txt', c2.find("foo/count3.txt")),
1044             ('tok', './count1.txt', c2["count1.txt"], c1["count1.txt"]),
1045         ])
1046         d = c1.diff(c2)
1047         self.assertEqual(sorted(d), [
1048             ('add', './foo/count3.txt', c2.find("foo/count3.txt")),
1049             ('del', './foo/count2.txt', c1.find("foo/count2.txt")),
1050             ('tok', './count1.txt', c2["count1.txt"], c1["count1.txt"]),
1051         ])
1052
1053         self.assertNotEqual(c1.portable_manifest_text(), c2.portable_manifest_text())
1054         c1.apply(d)
1055         self.assertEqual(c1.portable_manifest_text(), c2.portable_manifest_text())
1056
1057     def test_diff_mod_in_subcollection(self):
1058         c1 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n./foo 5348b82a029fd9e971a811ce1f71360b+43 0:10:count2.txt\n')
1059         c2 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt 0:3:foo\n')
1060         d = c2.diff(c1)
1061         self.assertEqual(sorted(d), [
1062             ('mod', './foo', c2["foo"], c1["foo"]),
1063             ('tok', './count1.txt', c2["count1.txt"], c1["count1.txt"]),
1064         ])
1065         d = c1.diff(c2)
1066         self.assertEqual(sorted(d), [
1067             ('mod', './foo', c1["foo"], c2["foo"]),
1068             ('tok', './count1.txt', c2["count1.txt"], c1["count1.txt"]),
1069         ])
1070
1071         self.assertNotEqual(c1.portable_manifest_text(), c2.portable_manifest_text())
1072         c1.apply(d)
1073         self.assertEqual(c1.portable_manifest_text(), c2.portable_manifest_text())
1074
1075     def test_conflict_keep_local_change(self):
1076         c1 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n')
1077         c2 = Collection('. 5348b82a029fd9e971a811ce1f71360b+43 0:10:count2.txt\n')
1078         d = c1.diff(c2)
1079         self.assertEqual(sorted(d), [
1080             ('add', './count2.txt', c2["count2.txt"]),
1081             ('del', './count1.txt', c1["count1.txt"]),
1082         ])
1083         f = c1.open("count1.txt", "wb")
1084         f.write(b"zzzzz")
1085
1086         # c1 changed, so it should not be deleted.
1087         c1.apply(d)
1088         self.assertEqual(c1.portable_manifest_text(), ". 95ebc3c7b3b9f1d2c40fec14415d3cb8+5 5348b82a029fd9e971a811ce1f71360b+43 0:5:count1.txt 5:10:count2.txt\n")
1089
1090     def test_conflict_mod(self):
1091         c1 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt')
1092         c2 = Collection('. 5348b82a029fd9e971a811ce1f71360b+43 0:10:count1.txt')
1093         d = c1.diff(c2)
1094         self.assertEqual(d, [('mod', './count1.txt', c1["count1.txt"], c2["count1.txt"])])
1095         f = c1.open("count1.txt", "wb")
1096         f.write(b"zzzzz")
1097
1098         # c1 changed, so c2 mod will go to a conflict file
1099         c1.apply(d)
1100         self.assertRegex(
1101             c1.portable_manifest_text(),
1102             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~$")
1103
1104     def test_conflict_add(self):
1105         c1 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count2.txt\n')
1106         c2 = Collection('. 5348b82a029fd9e971a811ce1f71360b+43 0:10:count1.txt\n')
1107         d = c1.diff(c2)
1108         self.assertEqual(sorted(d), [
1109             ('add', './count1.txt', c2["count1.txt"]),
1110             ('del', './count2.txt', c1["count2.txt"]),
1111         ])
1112         f = c1.open("count1.txt", "wb")
1113         f.write(b"zzzzz")
1114
1115         # c1 added count1.txt, so c2 add will go to a conflict file
1116         c1.apply(d)
1117         self.assertRegex(
1118             c1.portable_manifest_text(),
1119             r"\. 95ebc3c7b3b9f1d2c40fec14415d3cb8\+5 5348b82a029fd9e971a811ce1f71360b\+43 0:5:count1\.txt 5:10:count1\.txt~\d\d\d\d\d\d\d\d-\d\d\d\d\d\d~conflict~$")
1120
1121     def test_conflict_del(self):
1122         c1 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt')
1123         c2 = Collection('. 5348b82a029fd9e971a811ce1f71360b+43 0:10:count1.txt')
1124         d = c1.diff(c2)
1125         self.assertEqual(d, [('mod', './count1.txt', c1["count1.txt"], c2["count1.txt"])])
1126         c1.remove("count1.txt")
1127
1128         # c1 deleted, so c2 mod will go to a conflict file
1129         c1.apply(d)
1130         self.assertRegex(
1131             c1.portable_manifest_text(),
1132             r"\. 5348b82a029fd9e971a811ce1f71360b\+43 0:10:count1\.txt~\d\d\d\d\d\d\d\d-\d\d\d\d\d\d~conflict~$")
1133
1134     def test_notify(self):
1135         c1 = Collection()
1136         events = []
1137         c1.subscribe(lambda event, collection, name, item: events.append((event, collection, name, item)))
1138         f = c1.open("foo.txt", "wb")
1139         self.assertEqual(events[0], (arvados.collection.ADD, c1, "foo.txt", f.arvadosfile))
1140
1141     def test_open_w(self):
1142         c1 = Collection(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n")
1143         self.assertEqual(c1["count1.txt"].size(), 10)
1144         c1.open("count1.txt", "wb").close()
1145         self.assertEqual(c1["count1.txt"].size(), 0)
1146
1147
1148 class NewCollectionTestCaseWithServersAndTokens(run_test_server.TestCaseWithServers):
1149     MAIN_SERVER = {}
1150     KEEP_SERVER = {}
1151
1152     def setUp(self):
1153         self.keep_put = getattr(arvados.keep.KeepClient, 'put')
1154
1155     def test_repacked_block_submission_get_permission_token(self):
1156         '''
1157         Make sure that those blocks that are committed after repacking small ones,
1158         get their permission tokens assigned on the collection manifest.
1159         '''
1160         def wrapped_keep_put(*args, **kwargs):
1161             # Simulate slow put operations
1162             time.sleep(1)
1163             return self.keep_put(*args, **kwargs)
1164
1165         re_locator = "[0-9a-f]{32}\+\d+\+A[a-f0-9]{40}@[a-f0-9]{8}"
1166
1167         with mock.patch('arvados.keep.KeepClient.put', autospec=True) as mocked_put:
1168             mocked_put.side_effect = wrapped_keep_put
1169             c = Collection()
1170             # Write 70 files ~1MiB each so we force to produce 1 big block by repacking
1171             # small ones before finishing the upload.
1172             for i in range(70):
1173                 f = c.open("file_{}.txt".format(i), 'wb')
1174                 f.write(random.choice('abcdefghijklmnopqrstuvwxyz') * (2**20+i))
1175                 f.close(flush=False)
1176             # We should get 2 blocks with their tokens
1177             self.assertEqual(len(re.findall(re_locator, c.manifest_text())), 2)
1178
1179
1180 class NewCollectionTestCaseWithServers(run_test_server.TestCaseWithServers):
1181     def test_get_manifest_text_only_committed(self):
1182         c = Collection()
1183         with c.open("count.txt", "wb") as f:
1184             # One file committed
1185             with c.open("foo.txt", "wb") as foo:
1186                 foo.write(b"foo")
1187                 foo.flush() # Force block commit
1188             f.write(b"0123456789")
1189             # Other file not committed. Block not written to keep yet.
1190             self.assertEqual(
1191                 c._get_manifest_text(".",
1192                                      strip=False,
1193                                      normalize=False,
1194                                      only_committed=True),
1195                 '. acbd18db4cc2f85cedef654fccc4a4d8+3 0:0:count.txt 0:3:foo.txt\n')
1196             # And now with the file closed...
1197             f.flush() # Force block commit
1198         self.assertEqual(
1199             c._get_manifest_text(".",
1200                                  strip=False,
1201                                  normalize=False,
1202                                  only_committed=True),
1203             ". 781e5e245d69b566979b86e28d23f2c7+10 acbd18db4cc2f85cedef654fccc4a4d8+3 0:10:count.txt 10:3:foo.txt\n")
1204
1205     def test_only_small_blocks_are_packed_together(self):
1206         c = Collection()
1207         # Write a couple of small files,
1208         f = c.open("count.txt", "wb")
1209         f.write(b"0123456789")
1210         f.close(flush=False)
1211         foo = c.open("foo.txt", "wb")
1212         foo.write(b"foo")
1213         foo.close(flush=False)
1214         # Then, write a big file, it shouldn't be packed with the ones above
1215         big = c.open("bigfile.txt", "wb")
1216         big.write(b"x" * 1024 * 1024 * 33) # 33 MB > KEEP_BLOCK_SIZE/2
1217         big.close(flush=False)
1218         self.assertEqual(
1219             c.manifest_text("."),
1220             '. 2d303c138c118af809f39319e5d507e9+34603008 a8430a058b8fbf408e1931b794dbd6fb+13 0:34603008:bigfile.txt 34603008:10:count.txt 34603018:3:foo.txt\n')
1221
1222     def test_flush_after_small_block_packing(self):
1223         c = Collection()
1224         # Write a couple of small files,
1225         f = c.open("count.txt", "wb")
1226         f.write(b"0123456789")
1227         f.close(flush=False)
1228         foo = c.open("foo.txt", "wb")
1229         foo.write(b"foo")
1230         foo.close(flush=False)
1231
1232         self.assertEqual(
1233             c.manifest_text(),
1234             '. a8430a058b8fbf408e1931b794dbd6fb+13 0:10:count.txt 10:3:foo.txt\n')
1235
1236         f = c.open("count.txt", "rb+")
1237         f.close(flush=True)
1238
1239         self.assertEqual(
1240             c.manifest_text(),
1241             '. a8430a058b8fbf408e1931b794dbd6fb+13 0:10:count.txt 10:3:foo.txt\n')
1242
1243     def test_write_after_small_block_packing2(self):
1244         c = Collection()
1245         # Write a couple of small files,
1246         f = c.open("count.txt", "wb")
1247         f.write(b"0123456789")
1248         f.close(flush=False)
1249         foo = c.open("foo.txt", "wb")
1250         foo.write(b"foo")
1251         foo.close(flush=False)
1252
1253         self.assertEqual(
1254             c.manifest_text(),
1255             '. a8430a058b8fbf408e1931b794dbd6fb+13 0:10:count.txt 10:3:foo.txt\n')
1256
1257         f = c.open("count.txt", "rb+")
1258         f.write(b"abc")
1259         f.close(flush=False)
1260
1261         self.assertEqual(
1262             c.manifest_text(),
1263             '. 900150983cd24fb0d6963f7d28e17f72+3 a8430a058b8fbf408e1931b794dbd6fb+13 0:3:count.txt 6:7:count.txt 13:3:foo.txt\n')
1264
1265
1266     def test_small_block_packing_with_overwrite(self):
1267         c = Collection()
1268         c.open("b1", "wb").close()
1269         c["b1"].writeto(0, b"b1", 0)
1270
1271         c.open("b2", "wb").close()
1272         c["b2"].writeto(0, b"b2", 0)
1273
1274         c["b1"].writeto(0, b"1b", 0)
1275
1276         self.assertEqual(c.manifest_text(), ". ed4f3f67c70b02b29c50ce1ea26666bd+4 0:2:b1 2:2:b2\n")
1277         self.assertEqual(c["b1"].manifest_text(), ". ed4f3f67c70b02b29c50ce1ea26666bd+4 0:2:b1\n")
1278         self.assertEqual(c["b2"].manifest_text(), ". ed4f3f67c70b02b29c50ce1ea26666bd+4 2:2:b2\n")
1279
1280
1281 class CollectionCreateUpdateTest(run_test_server.TestCaseWithServers):
1282     MAIN_SERVER = {}
1283     KEEP_SERVER = {}
1284
1285     def create_count_txt(self):
1286         # Create an empty collection, save it to the API server, then write a
1287         # file, but don't save it.
1288
1289         c = Collection()
1290         c.save_new("CollectionCreateUpdateTest", ensure_unique_name=True)
1291         self.assertEqual(c.portable_data_hash(), "d41d8cd98f00b204e9800998ecf8427e+0")
1292         self.assertEqual(c.api_response()["portable_data_hash"], "d41d8cd98f00b204e9800998ecf8427e+0" )
1293
1294         with c.open("count.txt", "wb") as f:
1295             f.write(b"0123456789")
1296
1297         self.assertEqual(c.portable_manifest_text(), ". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n")
1298
1299         return c
1300
1301     def test_create_and_save(self):
1302         c = self.create_count_txt()
1303         c.save(storage_classes=['archive'])
1304
1305         self.assertRegex(
1306             c.manifest_text(),
1307             r"^\. 781e5e245d69b566979b86e28d23f2c7\+10\+A[a-f0-9]{40}@[a-f0-9]{8} 0:10:count\.txt$",)
1308         self.assertEqual(c.api_response()["storage_classes_desired"], ['archive'])
1309
1310
1311     def test_create_and_save_new(self):
1312         c = self.create_count_txt()
1313         c.save_new(storage_classes=['archive'])
1314
1315         self.assertRegex(
1316             c.manifest_text(),
1317             r"^\. 781e5e245d69b566979b86e28d23f2c7\+10\+A[a-f0-9]{40}@[a-f0-9]{8} 0:10:count\.txt$",)
1318         self.assertEqual(c.api_response()["storage_classes_desired"], ['archive'])
1319
1320     def test_create_diff_apply(self):
1321         c1 = self.create_count_txt()
1322         c1.save()
1323
1324         c2 = Collection(c1.manifest_locator())
1325         with c2.open("count.txt", "wb") as f:
1326             f.write(b"abcdefg")
1327
1328         diff = c1.diff(c2)
1329
1330         self.assertEqual(diff[0], (arvados.collection.MOD, u'./count.txt', c1["count.txt"], c2["count.txt"]))
1331
1332         c1.apply(diff)
1333         self.assertEqual(c1.portable_data_hash(), c2.portable_data_hash())
1334
1335     def test_diff_apply_with_token(self):
1336         baseline = CollectionReader(". 781e5e245d69b566979b86e28d23f2c7+10+A715fd31f8111894f717eb1003c1b0216799dd9ec@54f5dd1a 0:10:count.txt\n")
1337         c = Collection(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n")
1338         other = CollectionReader(". 7ac66c0f148de9519b8bd264312c4d64+7+A715fd31f8111894f717eb1003c1b0216799dd9ec@54f5dd1a 0:7:count.txt\n")
1339
1340         diff = baseline.diff(other)
1341         self.assertEqual(diff, [('mod', u'./count.txt', c["count.txt"], other["count.txt"])])
1342
1343         c.apply(diff)
1344
1345         self.assertEqual(c.manifest_text(), ". 7ac66c0f148de9519b8bd264312c4d64+7+A715fd31f8111894f717eb1003c1b0216799dd9ec@54f5dd1a 0:7:count.txt\n")
1346
1347
1348     def test_create_and_update(self):
1349         c1 = self.create_count_txt()
1350         c1.save()
1351
1352         c2 = arvados.collection.Collection(c1.manifest_locator())
1353         with c2.open("count.txt", "wb") as f:
1354             f.write(b"abcdefg")
1355
1356         c2.save()
1357
1358         self.assertNotEqual(c1.portable_data_hash(), c2.portable_data_hash())
1359         c1.update()
1360         self.assertEqual(c1.portable_data_hash(), c2.portable_data_hash())
1361
1362
1363     def test_create_and_update_with_conflict(self):
1364         c1 = self.create_count_txt()
1365         c1.save()
1366
1367         with c1.open("count.txt", "wb") as f:
1368             f.write(b"XYZ")
1369
1370         c2 = arvados.collection.Collection(c1.manifest_locator())
1371         with c2.open("count.txt", "wb") as f:
1372             f.write(b"abcdefg")
1373
1374         c2.save()
1375
1376         c1.update()
1377         self.assertRegex(
1378             c1.manifest_text(),
1379             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~$")
1380
1381     def test_pdh_is_native_str(self):
1382         c1 = self.create_count_txt()
1383         pdh = c1.portable_data_hash()
1384         self.assertEqual(type(''), type(pdh))
1385
1386
1387 if __name__ == '__main__':
1388     unittest.main()