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 random
  13 import re
  14 import sys
  15 import datetime
  16 import ciso8601
  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     def test_get_properties(self):
 806         c = Collection()
 807         self.assertEqual(c.get_properties(), {})
 808         c.save_new(properties={"foo":"bar"})
 809         self.assertEqual(c.get_properties(), {"foo":"bar"})
 810
 811     def test_get_trash_at(self):
 812         c = Collection()
 813         self.assertEqual(c.get_trash_at(), None)
 814         c.save_new(trash_at=datetime.datetime(2111, 1, 1, 11, 11, 11, 111111))
 815         self.assertEqual(c.get_trash_at(), ciso8601.parse_datetime('2111-01-01T11:11:11.111111000Z'))
 816
 817
 818 class CollectionOpenModes(run_test_server.TestCaseWithServers):
 819
 820     def test_open_binary_modes(self):
 821         c = Collection()
 822         for mode in ['wb', 'wb+', 'ab', 'ab+']:
 823             with c.open('foo', mode) as f:
 824                 f.write(b'foo')
 825
 826     def test_open_invalid_modes(self):
 827         c = Collection()
 828         for mode in ['+r', 'aa', '++', 'r+b', 'beer', '', None]:
 829             with self.assertRaises(Exception):
 830                 c.open('foo', mode)
 831
 832     def test_open_text_modes(self):
 833         c = Collection()
 834         with c.open('foo', 'wb') as f:
 835             f.write('foo')
 836         for mode in ['r', 'rt', 'r+', 'rt+', 'w', 'wt', 'a', 'at']:
 837             with c.open('foo', mode) as f:
 838                 if mode[0] == 'r' and '+' not in mode:
 839                     self.assertEqual('foo', f.read(3))
 840                 else:
 841                     f.write('bar')
 842                     f.seek(0, os.SEEK_SET)
 843                     self.assertEqual('bar', f.read(3))
 844
 845
 846 class TextModes(run_test_server.TestCaseWithServers):
 847
 848     def setUp(self):
 849         arvados.config.KEEP_BLOCK_SIZE = 4
 850         if sys.version_info < (3, 0):
 851             import unicodedata
 852             self.sailboat = unicodedata.lookup('SAILBOAT')
 853             self.snowman = unicodedata.lookup('SNOWMAN')
 854         else:
 855             self.sailboat = '\N{SAILBOAT}'
 856             self.snowman = '\N{SNOWMAN}'
 857
 858     def tearDown(self):
 859         arvados.config.KEEP_BLOCK_SIZE = 2 ** 26
 860
 861     def test_read_sailboat_across_block_boundary(self):
 862         c = Collection()
 863         f = c.open('sailboats', 'wb')
 864         data = self.sailboat.encode('utf-8')
 865         f.write(data)
 866         f.write(data[:1])
 867         f.write(data[1:])
 868         f.write(b'\n')
 869         f.close()
 870         self.assertRegex(c.portable_manifest_text(), r'\+4 .*\+3 ')
 871
 872         f = c.open('sailboats', 'r')
 873         string = f.readline()
 874         self.assertEqual(string, self.sailboat+self.sailboat+'\n')
 875         f.close()
 876
 877     def test_write_snowman_across_block_boundary(self):
 878         c = Collection()
 879         f = c.open('snowmany', 'w')
 880         data = self.snowman
 881         f.write(data+data+'\n'+data+'\n')
 882         f.close()
 883         self.assertRegex(c.portable_manifest_text(), r'\+4 .*\+4 .*\+3 ')
 884
 885         f = c.open('snowmany', 'r')
 886         self.assertEqual(f.readline(), self.snowman+self.snowman+'\n')
 887         self.assertEqual(f.readline(), self.snowman+'\n')
 888         f.close()
 889
 890
 891 class NewCollectionTestCase(unittest.TestCase, CollectionTestMixin):
 892
 893     def test_replication_desired_kept_on_load(self):
 894         m = '. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt 0:10:count2.txt\n'
 895         c1 = Collection(m, replication_desired=1)
 896         c1.save_new()
 897         loc = c1.manifest_locator()
 898         c2 = Collection(loc)
 899         self.assertEqual(c1.manifest_text, c2.manifest_text)
 900         self.assertEqual(c1.replication_desired, c2.replication_desired)
 901
 902     def test_replication_desired_not_loaded_if_provided(self):
 903         m = '. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt 0:10:count2.txt\n'
 904         c1 = Collection(m, replication_desired=1)
 905         c1.save_new()
 906         loc = c1.manifest_locator()
 907         c2 = Collection(loc, replication_desired=2)
 908         self.assertEqual(c1.manifest_text, c2.manifest_text)
 909         self.assertNotEqual(c1.replication_desired, c2.replication_desired)
 910
 911     def test_storage_classes_desired_kept_on_load(self):
 912         m = '. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt 0:10:count2.txt\n'
 913         c1 = Collection(m, storage_classes_desired=['archival'])
 914         c1.save_new()
 915         loc = c1.manifest_locator()
 916         c2 = Collection(loc)
 917         self.assertEqual(c1.manifest_text, c2.manifest_text)
 918         self.assertEqual(c1.storage_classes_desired(), c2.storage_classes_desired())
 919
 920     def test_storage_classes_change_after_save(self):
 921         m = '. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt 0:10:count2.txt\n'
 922         c1 = Collection(m, storage_classes_desired=['archival'])
 923         c1.save_new()
 924         loc = c1.manifest_locator()
 925         c2 = Collection(loc)
 926         self.assertEqual(['archival'], c2.storage_classes_desired())
 927         c2.save(storage_classes=['highIO'])
 928         self.assertEqual(['highIO'], c2.storage_classes_desired())
 929         c3 = Collection(loc)
 930         self.assertEqual(c1.manifest_text, c3.manifest_text)
 931         self.assertEqual(['highIO'], c3.storage_classes_desired())
 932
 933     def test_storage_classes_desired_not_loaded_if_provided(self):
 934         m = '. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt 0:10:count2.txt\n'
 935         c1 = Collection(m, storage_classes_desired=['archival'])
 936         c1.save_new()
 937         loc = c1.manifest_locator()
 938         c2 = Collection(loc, storage_classes_desired=['default'])
 939         self.assertEqual(c1.manifest_text, c2.manifest_text)
 940         self.assertNotEqual(c1.storage_classes_desired(), c2.storage_classes_desired())
 941
 942     def test_init_manifest(self):
 943         m1 = """. 5348b82a029fd9e971a811ce1f71360b+43 0:43:md5sum.txt
 944 . 085c37f02916da1cad16f93c54d899b7+41 0:41:md5sum.txt
 945 . 8b22da26f9f433dea0a10e5ec66d73ba+43 0:43:md5sum.txt
 946 """
 947         self.assertEqual(m1, CollectionReader(m1).manifest_text(normalize=False))
 948         self.assertEqual(". 5348b82a029fd9e971a811ce1f71360b+43 085c37f02916da1cad16f93c54d899b7+41 8b22da26f9f433dea0a10e5ec66d73ba+43 0:127:md5sum.txt\n", CollectionReader(m1).manifest_text(normalize=True))
 949
 950     def test_init_manifest_with_collision(self):
 951         m1 = """. 5348b82a029fd9e971a811ce1f71360b+43 0:43:md5sum.txt
 952 ./md5sum.txt 085c37f02916da1cad16f93c54d899b7+41 0:41:md5sum.txt
 953 """
 954         with self.assertRaises(arvados.errors.ArgumentError):
 955             self.assertEqual(m1, CollectionReader(m1))
 956
 957     def test_init_manifest_with_error(self):
 958         m1 = """. 0:43:md5sum.txt"""
 959         with self.assertRaises(arvados.errors.ArgumentError):
 960             self.assertEqual(m1, CollectionReader(m1))
 961
 962     def test_remove(self):
 963         c = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt 0:10:count2.txt\n')
 964         self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt 0:10:count2.txt\n", c.portable_manifest_text())
 965         self.assertIn("count1.txt", c)
 966         c.remove("count1.txt")
 967         self.assertNotIn("count1.txt", c)
 968         self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count2.txt\n", c.portable_manifest_text())
 969         with self.assertRaises(arvados.errors.ArgumentError):
 970             c.remove("")
 971
 972     def test_find(self):
 973         c = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt 0:10:count2.txt\n')
 974         self.assertIs(c.find("."), c)
 975         self.assertIs(c.find("./count1.txt"), c["count1.txt"])
 976         self.assertIs(c.find("count1.txt"), c["count1.txt"])
 977         with self.assertRaises(IOError):
 978             c.find("/.")
 979         with self.assertRaises(arvados.errors.ArgumentError):
 980             c.find("")
 981         self.assertIs(c.find("./nonexistant.txt"), None)
 982         self.assertIs(c.find("./nonexistantsubdir/nonexistant.txt"), None)
 983
 984     def test_escaped_paths_dont_get_unescaped_on_manifest(self):
 985         # Dir & file names are literally '\056' (escaped form: \134056)
 986         manifest = './\\134056\\040Test d41d8cd98f00b204e9800998ecf8427e+0 0:0:\\134056\n'
 987         c = Collection(manifest)
 988         self.assertEqual(c.portable_manifest_text(), manifest)
 989
 990     def test_other_special_chars_on_file_token(self):
 991         cases = [
 992             ('\\000', '\0'),
 993             ('\\011', '\t'),
 994             ('\\012', '\n'),
 995             ('\\072', ':'),
 996             ('\\134400', '\\400'),
 997         ]
 998         for encoded, decoded in cases:
 999             manifest = '. d41d8cd98f00b204e9800998ecf8427e+0 0:0:some%sfile.txt\n' % encoded
1000             c = Collection(manifest)
1001             self.assertEqual(c.portable_manifest_text(), manifest)
1002             self.assertIn('some%sfile.txt' % decoded, c.keys())
1003
1004     def test_escaped_paths_do_get_unescaped_on_listing(self):
1005         # Dir & file names are literally '\056' (escaped form: \134056)
1006         manifest = './\\134056\\040Test d41d8cd98f00b204e9800998ecf8427e+0 0:0:\\134056\n'
1007         c = Collection(manifest)
1008         self.assertIn('\\056 Test', c.keys())
1009         self.assertIn('\\056', c['\\056 Test'].keys())
1010
1011     def test_make_empty_dir_with_escaped_chars(self):
1012         c = Collection()
1013         c.mkdirs('./Empty\\056Dir')
1014         self.assertEqual(c.portable_manifest_text(),
1015                          './Empty\\134056Dir d41d8cd98f00b204e9800998ecf8427e+0 0:0:\\056\n')
1016
1017     def test_make_empty_dir_with_spaces(self):
1018         c = Collection()
1019         c.mkdirs('./foo bar/baz waz')
1020         self.assertEqual(c.portable_manifest_text(),
1021                          './foo\\040bar/baz\\040waz d41d8cd98f00b204e9800998ecf8427e+0 0:0:\\056\n')
1022
1023     def test_remove_in_subdir(self):
1024         c = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n./foo 781e5e245d69b566979b86e28d23f2c7+10 0:10:count2.txt\n')
1025         c.remove("foo/count2.txt")
1026         self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n./foo d41d8cd98f00b204e9800998ecf8427e+0 0:0:\\056\n", c.portable_manifest_text())
1027
1028     def test_remove_empty_subdir(self):
1029         c = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n./foo 781e5e245d69b566979b86e28d23f2c7+10 0:10:count2.txt\n')
1030         c.remove("foo/count2.txt")
1031         c.remove("foo")
1032         self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n", c.portable_manifest_text())
1033
1034     def test_remove_nonempty_subdir(self):
1035         c = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n./foo 781e5e245d69b566979b86e28d23f2c7+10 0:10:count2.txt\n')
1036         with self.assertRaises(IOError):
1037             c.remove("foo")
1038         c.remove("foo", recursive=True)
1039         self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n", c.portable_manifest_text())
1040
1041     def test_copy_to_file_in_dir(self):
1042         c = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n')
1043         c.copy("count1.txt", "foo/count2.txt")
1044         self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n./foo 781e5e245d69b566979b86e28d23f2c7+10 0:10:count2.txt\n", c.portable_manifest_text())
1045
1046     def test_copy_file(self):
1047         c = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n')
1048         c.copy("count1.txt", "count2.txt")
1049         self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt 0:10:count2.txt\n", c.portable_manifest_text())
1050
1051     def test_copy_to_existing_dir(self):
1052         c = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n./foo 781e5e245d69b566979b86e28d23f2c7+10 0:10:count2.txt\n')
1053         c.copy("count1.txt", "foo")
1054         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())
1055
1056     def test_copy_to_new_dir(self):
1057         c = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n')
1058         c.copy("count1.txt", "foo/")
1059         self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n./foo 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n", c.portable_manifest_text())
1060
1061     def test_rename_file(self):
1062         c = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n')
1063         c.rename("count1.txt", "count2.txt")
1064         self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count2.txt\n", c.manifest_text())
1065
1066     def test_move_file_to_dir(self):
1067         c = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n')
1068         c.mkdirs("foo")
1069         c.rename("count1.txt", "foo/count2.txt")
1070         self.assertEqual("./foo 781e5e245d69b566979b86e28d23f2c7+10 0:10:count2.txt\n", c.manifest_text())
1071
1072     def test_move_file_to_other(self):
1073         c1 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n')
1074         c2 = Collection()
1075         c2.rename("count1.txt", "count2.txt", source_collection=c1)
1076         self.assertEqual("", c1.manifest_text())
1077         self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count2.txt\n", c2.manifest_text())
1078
1079     def test_clone(self):
1080         c = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n./foo 781e5e245d69b566979b86e28d23f2c7+10 0:10:count2.txt\n')
1081         cl = c.clone()
1082         self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n./foo 781e5e245d69b566979b86e28d23f2c7+10 0:10:count2.txt\n", cl.portable_manifest_text())
1083
1084     def test_diff_del_add(self):
1085         c1 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n')
1086         c2 = Collection('. 5348b82a029fd9e971a811ce1f71360b+43 0:10:count2.txt\n')
1087         d = c2.diff(c1)
1088         self.assertEqual(sorted(d), [
1089             ('add', './count1.txt', c1["count1.txt"]),
1090             ('del', './count2.txt', c2["count2.txt"]),
1091         ])
1092         d = c1.diff(c2)
1093         self.assertEqual(sorted(d), [
1094             ('add', './count2.txt', c2["count2.txt"]),
1095             ('del', './count1.txt', c1["count1.txt"]),
1096         ])
1097         self.assertNotEqual(c1.portable_manifest_text(), c2.portable_manifest_text())
1098         c1.apply(d)
1099         self.assertEqual(c1.portable_manifest_text(), c2.portable_manifest_text())
1100
1101     def test_diff_same(self):
1102         c1 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n')
1103         c2 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n')
1104         d = c2.diff(c1)
1105         self.assertEqual(d, [('tok', './count1.txt', c2["count1.txt"], c1["count1.txt"])])
1106         d = c1.diff(c2)
1107         self.assertEqual(d, [('tok', './count1.txt', c2["count1.txt"], c1["count1.txt"])])
1108
1109         self.assertEqual(c1.portable_manifest_text(), c2.portable_manifest_text())
1110         c1.apply(d)
1111         self.assertEqual(c1.portable_manifest_text(), c2.portable_manifest_text())
1112
1113     def test_diff_mod(self):
1114         c1 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n')
1115         c2 = Collection('. 5348b82a029fd9e971a811ce1f71360b+43 0:10:count1.txt\n')
1116         d = c2.diff(c1)
1117         self.assertEqual(d, [('mod', './count1.txt', c2["count1.txt"], c1["count1.txt"])])
1118         d = c1.diff(c2)
1119         self.assertEqual(d, [('mod', './count1.txt', c1["count1.txt"], c2["count1.txt"])])
1120
1121         self.assertNotEqual(c1.portable_manifest_text(), c2.portable_manifest_text())
1122         c1.apply(d)
1123         self.assertEqual(c1.portable_manifest_text(), c2.portable_manifest_text())
1124
1125     def test_diff_add(self):
1126         c1 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n')
1127         c2 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 5348b82a029fd9e971a811ce1f71360b+43 0:10:count1.txt 10:20:count2.txt\n')
1128         d = c2.diff(c1)
1129         self.assertEqual(sorted(d), [
1130             ('del', './count2.txt', c2["count2.txt"]),
1131             ('tok', './count1.txt', c2["count1.txt"], c1["count1.txt"]),
1132         ])
1133         d = c1.diff(c2)
1134         self.assertEqual(sorted(d), [
1135             ('add', './count2.txt', c2["count2.txt"]),
1136             ('tok', './count1.txt', c2["count1.txt"], c1["count1.txt"]),
1137         ])
1138
1139         self.assertNotEqual(c1.portable_manifest_text(), c2.portable_manifest_text())
1140         c1.apply(d)
1141         self.assertEqual(c1.portable_manifest_text(), c2.portable_manifest_text())
1142
1143     def test_diff_add_in_subcollection(self):
1144         c1 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n')
1145         c2 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n./foo 5348b82a029fd9e971a811ce1f71360b+43 0:10:count2.txt\n')
1146         d = c2.diff(c1)
1147         self.assertEqual(sorted(d), [
1148             ('del', './foo', c2["foo"]),
1149             ('tok', './count1.txt', c2["count1.txt"], c1["count1.txt"]),
1150         ])
1151         d = c1.diff(c2)
1152         self.assertEqual(sorted(d), [
1153             ('add', './foo', c2["foo"]),
1154             ('tok', './count1.txt', c2["count1.txt"], c1["count1.txt"]),
1155         ])
1156         self.assertNotEqual(c1.portable_manifest_text(), c2.portable_manifest_text())
1157         c1.apply(d)
1158         self.assertEqual(c1.portable_manifest_text(), c2.portable_manifest_text())
1159
1160     def test_diff_del_add_in_subcollection(self):
1161         c1 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n./foo 5348b82a029fd9e971a811ce1f71360b+43 0:10:count2.txt\n')
1162         c2 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n./foo 5348b82a029fd9e971a811ce1f71360b+43 0:3:count3.txt\n')
1163         d = c2.diff(c1)
1164         self.assertEqual(sorted(d), [
1165             ('add', './foo/count2.txt', c1.find("foo/count2.txt")),
1166             ('del', './foo/count3.txt', c2.find("foo/count3.txt")),
1167             ('tok', './count1.txt', c2["count1.txt"], c1["count1.txt"]),
1168         ])
1169         d = c1.diff(c2)
1170         self.assertEqual(sorted(d), [
1171             ('add', './foo/count3.txt', c2.find("foo/count3.txt")),
1172             ('del', './foo/count2.txt', c1.find("foo/count2.txt")),
1173             ('tok', './count1.txt', c2["count1.txt"], c1["count1.txt"]),
1174         ])
1175
1176         self.assertNotEqual(c1.portable_manifest_text(), c2.portable_manifest_text())
1177         c1.apply(d)
1178         self.assertEqual(c1.portable_manifest_text(), c2.portable_manifest_text())
1179
1180     def test_diff_mod_in_subcollection(self):
1181         c1 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n./foo 5348b82a029fd9e971a811ce1f71360b+43 0:10:count2.txt\n')
1182         c2 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt 0:3:foo\n')
1183         d = c2.diff(c1)
1184         self.assertEqual(sorted(d), [
1185             ('mod', './foo', c2["foo"], c1["foo"]),
1186             ('tok', './count1.txt', c2["count1.txt"], c1["count1.txt"]),
1187         ])
1188         d = c1.diff(c2)
1189         self.assertEqual(sorted(d), [
1190             ('mod', './foo', c1["foo"], c2["foo"]),
1191             ('tok', './count1.txt', c2["count1.txt"], c1["count1.txt"]),
1192         ])
1193
1194         self.assertNotEqual(c1.portable_manifest_text(), c2.portable_manifest_text())
1195         c1.apply(d)
1196         self.assertEqual(c1.portable_manifest_text(), c2.portable_manifest_text())
1197
1198     def test_conflict_keep_local_change(self):
1199         c1 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n')
1200         c2 = Collection('. 5348b82a029fd9e971a811ce1f71360b+43 0:10:count2.txt\n')
1201         d = c1.diff(c2)
1202         self.assertEqual(sorted(d), [
1203             ('add', './count2.txt', c2["count2.txt"]),
1204             ('del', './count1.txt', c1["count1.txt"]),
1205         ])
1206         f = c1.open("count1.txt", "wb")
1207         f.write(b"zzzzz")
1208
1209         # c1 changed, so it should not be deleted.
1210         c1.apply(d)
1211         self.assertEqual(c1.portable_manifest_text(), ". 95ebc3c7b3b9f1d2c40fec14415d3cb8+5 5348b82a029fd9e971a811ce1f71360b+43 0:5:count1.txt 5:10:count2.txt\n")
1212
1213     def test_conflict_mod(self):
1214         c1 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt')
1215         c2 = Collection('. 5348b82a029fd9e971a811ce1f71360b+43 0:10:count1.txt')
1216         d = c1.diff(c2)
1217         self.assertEqual(d, [('mod', './count1.txt', c1["count1.txt"], c2["count1.txt"])])
1218         f = c1.open("count1.txt", "wb")
1219         f.write(b"zzzzz")
1220
1221         # c1 changed, so c2 mod will go to a conflict file
1222         c1.apply(d)
1223         self.assertRegex(
1224             c1.portable_manifest_text(),
1225             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~$")
1226
1227     def test_conflict_add(self):
1228         c1 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count2.txt\n')
1229         c2 = Collection('. 5348b82a029fd9e971a811ce1f71360b+43 0:10:count1.txt\n')
1230         d = c1.diff(c2)
1231         self.assertEqual(sorted(d), [
1232             ('add', './count1.txt', c2["count1.txt"]),
1233             ('del', './count2.txt', c1["count2.txt"]),
1234         ])
1235         f = c1.open("count1.txt", "wb")
1236         f.write(b"zzzzz")
1237
1238         # c1 added count1.txt, so c2 add will go to a conflict file
1239         c1.apply(d)
1240         self.assertRegex(
1241             c1.portable_manifest_text(),
1242             r"\. 95ebc3c7b3b9f1d2c40fec14415d3cb8\+5 5348b82a029fd9e971a811ce1f71360b\+43 0:5:count1\.txt 5:10:count1\.txt~\d\d\d\d\d\d\d\d-\d\d\d\d\d\d~conflict~$")
1243
1244     def test_conflict_del(self):
1245         c1 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt')
1246         c2 = Collection('. 5348b82a029fd9e971a811ce1f71360b+43 0:10:count1.txt')
1247         d = c1.diff(c2)
1248         self.assertEqual(d, [('mod', './count1.txt', c1["count1.txt"], c2["count1.txt"])])
1249         c1.remove("count1.txt")
1250
1251         # c1 deleted, so c2 mod will go to a conflict file
1252         c1.apply(d)
1253         self.assertRegex(
1254             c1.portable_manifest_text(),
1255             r"\. 5348b82a029fd9e971a811ce1f71360b\+43 0:10:count1\.txt~\d\d\d\d\d\d\d\d-\d\d\d\d\d\d~conflict~$")
1256
1257     def test_notify(self):
1258         c1 = Collection()
1259         events = []
1260         c1.subscribe(lambda event, collection, name, item: events.append((event, collection, name, item)))
1261         f = c1.open("foo.txt", "wb")
1262         self.assertEqual(events[0], (arvados.collection.ADD, c1, "foo.txt", f.arvadosfile))
1263
1264     def test_open_w(self):
1265         c1 = Collection(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n")
1266         self.assertEqual(c1["count1.txt"].size(), 10)
1267         c1.open("count1.txt", "wb").close()
1268         self.assertEqual(c1["count1.txt"].size(), 0)
1269
1270
1271 class NewCollectionTestCaseWithServersAndTokens(run_test_server.TestCaseWithServers):
1272     MAIN_SERVER = {}
1273     KEEP_SERVER = {}
1274     local_locator_re = r"[0-9a-f]{32}\+\d+\+A[a-f0-9]{40}@[a-f0-9]{8}"
1275     remote_locator_re = r"[0-9a-f]{32}\+\d+\+R[a-z]{5}-[a-f0-9]{40}@[a-f0-9]{8}"
1276
1277     def setUp(self):
1278         self.keep_put = getattr(arvados.keep.KeepClient, 'put')
1279
1280     @mock.patch('arvados.keep.KeepClient.put', autospec=True)
1281     def test_storage_classes_desired(self, put_mock):
1282         put_mock.side_effect = self.keep_put
1283         c = Collection(storage_classes_desired=['default'])
1284         with c.open("file.txt", 'wb') as f:
1285             f.write('content')
1286         c.save_new()
1287         _, kwargs = put_mock.call_args
1288         self.assertEqual(['default'], kwargs['classes'])
1289
1290     @mock.patch('arvados.keep.KeepClient.put', autospec=True)
1291     def test_repacked_block_submission_get_permission_token(self, mocked_put):
1292         '''
1293         Make sure that those blocks that are committed after repacking small ones,
1294         get their permission tokens assigned on the collection manifest.
1295         '''
1296         def wrapped_keep_put(*args, **kwargs):
1297             # Simulate slow put operations
1298             time.sleep(1)
1299             return self.keep_put(*args, **kwargs)
1300
1301         mocked_put.side_effect = wrapped_keep_put
1302         c = Collection()
1303         # Write 70 files ~1MiB each so we force to produce 1 big block by repacking
1304         # small ones before finishing the upload.
1305         for i in range(70):
1306             f = c.open("file_{}.txt".format(i), 'wb')
1307             f.write(random.choice('abcdefghijklmnopqrstuvwxyz') * (2**20+i))
1308             f.close(flush=False)
1309         # We should get 2 blocks with their tokens
1310         self.assertEqual(len(re.findall(self.local_locator_re, c.manifest_text())), 2)
1311
1312     @mock.patch('arvados.keep.KeepClient.refresh_signature')
1313     def test_copy_remote_blocks_on_save_new(self, rs_mock):
1314         remote_block_loc = "acbd18db4cc2f85cedef654fccc4a4d8+3+Remote-" + "a" * 40 + "@abcdef01"
1315         local_block_loc = "acbd18db4cc2f85cedef654fccc4a4d8+3+A" + "b" * 40 + "@abcdef01"
1316         rs_mock.return_value = local_block_loc
1317         c = Collection(". " + remote_block_loc + " 0:3:foofile.txt\n")
1318         self.assertEqual(
1319             len(re.findall(self.remote_locator_re, c.manifest_text())), 1)
1320         self.assertEqual(
1321             len(re.findall(self.local_locator_re, c.manifest_text())), 0)
1322         c.save_new()
1323         rs_mock.assert_called()
1324         self.assertEqual(
1325             len(re.findall(self.remote_locator_re, c.manifest_text())), 0)
1326         self.assertEqual(
1327             len(re.findall(self.local_locator_re, c.manifest_text())), 1)
1328
1329     @mock.patch('arvados.keep.KeepClient.refresh_signature')
1330     def test_copy_remote_blocks_on_save(self, rs_mock):
1331         remote_block_loc = "acbd18db4cc2f85cedef654fccc4a4d8+3+Remote-" + "a" * 40 + "@abcdef01"
1332         local_block_loc = "acbd18db4cc2f85cedef654fccc4a4d8+3+A" + "b" * 40 + "@abcdef01"
1333         rs_mock.return_value = local_block_loc
1334         # Remote collection
1335         remote_c = Collection(". " + remote_block_loc + " 0:3:foofile.txt\n")
1336         self.assertEqual(
1337             len(re.findall(self.remote_locator_re, remote_c.manifest_text())), 1)
1338         # Local collection
1339         local_c = Collection()
1340         with local_c.open('barfile.txt', 'wb') as f:
1341             f.write('bar')
1342         local_c.save_new()
1343         self.assertEqual(
1344             len(re.findall(self.local_locator_re, local_c.manifest_text())), 1)
1345         self.assertEqual(
1346             len(re.findall(self.remote_locator_re, local_c.manifest_text())), 0)
1347         # Copy remote file to local collection
1348         local_c.copy('./foofile.txt', './copied/foofile.txt', remote_c)
1349         self.assertEqual(
1350             len(re.findall(self.local_locator_re, local_c.manifest_text())), 1)
1351         self.assertEqual(
1352             len(re.findall(self.remote_locator_re, local_c.manifest_text())), 1)
1353         # Save local collection: remote block should be copied
1354         local_c.save()
1355         rs_mock.assert_called()
1356         self.assertEqual(
1357             len(re.findall(self.local_locator_re, local_c.manifest_text())), 2)
1358         self.assertEqual(
1359             len(re.findall(self.remote_locator_re, local_c.manifest_text())), 0)
1360
1361
1362 class NewCollectionTestCaseWithServers(run_test_server.TestCaseWithServers):
1363     def test_get_manifest_text_only_committed(self):
1364         c = Collection()
1365         with c.open("count.txt", "wb") as f:
1366             # One file committed
1367             with c.open("foo.txt", "wb") as foo:
1368                 foo.write(b"foo")
1369                 foo.flush() # Force block commit
1370             f.write(b"0123456789")
1371             # Other file not committed. Block not written to keep yet.
1372             self.assertEqual(
1373                 c._get_manifest_text(".",
1374                                      strip=False,
1375                                      normalize=False,
1376                                      only_committed=True),
1377                 '. acbd18db4cc2f85cedef654fccc4a4d8+3 0:0:count.txt 0:3:foo.txt\n')
1378             # And now with the file closed...
1379             f.flush() # Force block commit
1380         self.assertEqual(
1381             c._get_manifest_text(".",
1382                                  strip=False,
1383                                  normalize=False,
1384                                  only_committed=True),
1385             ". 781e5e245d69b566979b86e28d23f2c7+10 acbd18db4cc2f85cedef654fccc4a4d8+3 0:10:count.txt 10:3:foo.txt\n")
1386
1387     def test_only_small_blocks_are_packed_together(self):
1388         c = Collection()
1389         # Write a couple of small files,
1390         f = c.open("count.txt", "wb")
1391         f.write(b"0123456789")
1392         f.close(flush=False)
1393         foo = c.open("foo.txt", "wb")
1394         foo.write(b"foo")
1395         foo.close(flush=False)
1396         # Then, write a big file, it shouldn't be packed with the ones above
1397         big = c.open("bigfile.txt", "wb")
1398         big.write(b"x" * 1024 * 1024 * 33) # 33 MB > KEEP_BLOCK_SIZE/2
1399         big.close(flush=False)
1400         self.assertEqual(
1401             c.manifest_text("."),
1402             '. 2d303c138c118af809f39319e5d507e9+34603008 a8430a058b8fbf408e1931b794dbd6fb+13 0:34603008:bigfile.txt 34603008:10:count.txt 34603018:3:foo.txt\n')
1403
1404     def test_flush_after_small_block_packing(self):
1405         c = Collection()
1406         # Write a couple of small files,
1407         f = c.open("count.txt", "wb")
1408         f.write(b"0123456789")
1409         f.close(flush=False)
1410         foo = c.open("foo.txt", "wb")
1411         foo.write(b"foo")
1412         foo.close(flush=False)
1413
1414         self.assertEqual(
1415             c.manifest_text(),
1416             '. a8430a058b8fbf408e1931b794dbd6fb+13 0:10:count.txt 10:3:foo.txt\n')
1417
1418         f = c.open("count.txt", "rb+")
1419         f.close(flush=True)
1420
1421         self.assertEqual(
1422             c.manifest_text(),
1423             '. a8430a058b8fbf408e1931b794dbd6fb+13 0:10:count.txt 10:3:foo.txt\n')
1424
1425     def test_write_after_small_block_packing2(self):
1426         c = Collection()
1427         # Write a couple of small files,
1428         f = c.open("count.txt", "wb")
1429         f.write(b"0123456789")
1430         f.close(flush=False)
1431         foo = c.open("foo.txt", "wb")
1432         foo.write(b"foo")
1433         foo.close(flush=False)
1434
1435         self.assertEqual(
1436             c.manifest_text(),
1437             '. a8430a058b8fbf408e1931b794dbd6fb+13 0:10:count.txt 10:3:foo.txt\n')
1438
1439         f = c.open("count.txt", "rb+")
1440         f.write(b"abc")
1441         f.close(flush=False)
1442
1443         self.assertEqual(
1444             c.manifest_text(),
1445             '. 900150983cd24fb0d6963f7d28e17f72+3 a8430a058b8fbf408e1931b794dbd6fb+13 0:3:count.txt 6:7:count.txt 13:3:foo.txt\n')
1446
1447
1448     def test_small_block_packing_with_overwrite(self):
1449         c = Collection()
1450         c.open("b1", "wb").close()
1451         c["b1"].writeto(0, b"b1", 0)
1452
1453         c.open("b2", "wb").close()
1454         c["b2"].writeto(0, b"b2", 0)
1455
1456         c["b1"].writeto(0, b"1b", 0)
1457
1458         self.assertEqual(c.manifest_text(), ". ed4f3f67c70b02b29c50ce1ea26666bd+4 0:2:b1 2:2:b2\n")
1459         self.assertEqual(c["b1"].manifest_text(), ". ed4f3f67c70b02b29c50ce1ea26666bd+4 0:2:b1\n")
1460         self.assertEqual(c["b2"].manifest_text(), ". ed4f3f67c70b02b29c50ce1ea26666bd+4 2:2:b2\n")
1461
1462
1463 class CollectionCreateUpdateTest(run_test_server.TestCaseWithServers):
1464     MAIN_SERVER = {}
1465     KEEP_SERVER = {}
1466
1467     def create_count_txt(self):
1468         # Create an empty collection, save it to the API server, then write a
1469         # file, but don't save it.
1470
1471         c = Collection()
1472         c.save_new("CollectionCreateUpdateTest", ensure_unique_name=True)
1473         self.assertEqual(c.portable_data_hash(), "d41d8cd98f00b204e9800998ecf8427e+0")
1474         self.assertEqual(c.api_response()["portable_data_hash"], "d41d8cd98f00b204e9800998ecf8427e+0" )
1475
1476         with c.open("count.txt", "wb") as f:
1477             f.write(b"0123456789")
1478
1479         self.assertEqual(c.portable_manifest_text(), ". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n")
1480
1481         return c
1482
1483     def test_create_and_save(self):
1484         c = self.create_count_txt()
1485         c.save(properties={'type' : 'Intermediate'},
1486                storage_classes=['archive'],
1487                trash_at=datetime.datetime(2111, 1, 1, 11, 11, 11, 111111))
1488
1489         self.assertRegex(
1490             c.manifest_text(),
1491             r"^\. 781e5e245d69b566979b86e28d23f2c7\+10\+A[a-f0-9]{40}@[a-f0-9]{8} 0:10:count\.txt$",)
1492         self.assertEqual(c.api_response()["storage_classes_desired"], ['archive'])
1493         self.assertEqual(c.api_response()["properties"], {'type' : 'Intermediate'})
1494         self.assertEqual(c.api_response()["trash_at"], '2111-01-01T11:11:11.111111000Z')
1495
1496
1497     def test_create_and_save_new(self):
1498         c = self.create_count_txt()
1499         c.save_new(properties={'type' : 'Intermediate'},
1500                    storage_classes=['archive'],
1501                    trash_at=datetime.datetime(2111, 1, 1, 11, 11, 11, 111111))
1502
1503         self.assertRegex(
1504             c.manifest_text(),
1505             r"^\. 781e5e245d69b566979b86e28d23f2c7\+10\+A[a-f0-9]{40}@[a-f0-9]{8} 0:10:count\.txt$",)
1506         self.assertEqual(c.api_response()["storage_classes_desired"], ['archive'])
1507         self.assertEqual(c.api_response()["properties"], {'type' : 'Intermediate'})
1508         self.assertEqual(c.api_response()["trash_at"], '2111-01-01T11:11:11.111111000Z')
1509
1510     def test_create_and_save_after_commiting(self):
1511         c = self.create_count_txt()
1512         c.save(properties={'type' : 'Intermediate'},
1513                storage_classes=['hot'],
1514                trash_at=datetime.datetime(2111, 1, 1, 11, 11, 11, 111111))
1515         c.save(properties={'type' : 'Output'},
1516                storage_classes=['cold'],
1517                trash_at=datetime.datetime(2222, 2, 2, 22, 22, 22, 222222))
1518
1519         self.assertEqual(c.api_response()["storage_classes_desired"], ['cold'])
1520         self.assertEqual(c.api_response()["properties"], {'type' : 'Output'})
1521         self.assertEqual(c.api_response()["trash_at"], '2222-02-02T22:22:22.222222000Z')
1522
1523     def test_create_diff_apply(self):
1524         c1 = self.create_count_txt()
1525         c1.save()
1526
1527         c2 = Collection(c1.manifest_locator())
1528         with c2.open("count.txt", "wb") as f:
1529             f.write(b"abcdefg")
1530
1531         diff = c1.diff(c2)
1532
1533         self.assertEqual(diff[0], (arvados.collection.MOD, u'./count.txt', c1["count.txt"], c2["count.txt"]))
1534
1535         c1.apply(diff)
1536         self.assertEqual(c1.portable_data_hash(), c2.portable_data_hash())
1537
1538     def test_diff_apply_with_token(self):
1539         baseline = CollectionReader(". 781e5e245d69b566979b86e28d23f2c7+10+A715fd31f8111894f717eb1003c1b0216799dd9ec@54f5dd1a 0:10:count.txt\n")
1540         c = Collection(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n")
1541         other = CollectionReader(". 7ac66c0f148de9519b8bd264312c4d64+7+A715fd31f8111894f717eb1003c1b0216799dd9ec@54f5dd1a 0:7:count.txt\n")
1542
1543         diff = baseline.diff(other)
1544         self.assertEqual(diff, [('mod', u'./count.txt', c["count.txt"], other["count.txt"])])
1545
1546         c.apply(diff)
1547
1548         self.assertEqual(c.manifest_text(), ". 7ac66c0f148de9519b8bd264312c4d64+7+A715fd31f8111894f717eb1003c1b0216799dd9ec@54f5dd1a 0:7:count.txt\n")
1549
1550
1551     def test_create_and_update(self):
1552         c1 = self.create_count_txt()
1553         c1.save()
1554
1555         c2 = arvados.collection.Collection(c1.manifest_locator())
1556         with c2.open("count.txt", "wb") as f:
1557             f.write(b"abcdefg")
1558
1559         c2.save()
1560
1561         self.assertNotEqual(c1.portable_data_hash(), c2.portable_data_hash())
1562         c1.update()
1563         self.assertEqual(c1.portable_data_hash(), c2.portable_data_hash())
1564
1565
1566     def test_create_and_update_with_conflict(self):
1567         c1 = self.create_count_txt()
1568         c1.save()
1569
1570         with c1.open("count.txt", "wb") as f:
1571             f.write(b"XYZ")
1572
1573         c2 = arvados.collection.Collection(c1.manifest_locator())
1574         with c2.open("count.txt", "wb") as f:
1575             f.write(b"abcdefg")
1576
1577         c2.save()
1578
1579         c1.update()
1580         self.assertRegex(
1581             c1.manifest_text(),
1582             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~$")
1583
1584     def test_pdh_is_native_str(self):
1585         c1 = self.create_count_txt()
1586         pdh = c1.portable_data_hash()
1587         self.assertEqual(type(''), type(pdh))
1588
1589
1590 if __name__ == '__main__':
1591     unittest.main()