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 import arvados
   6 import copy
   7 import os
   8 import random
   9 import re
  10 import sys
  11 import datetime
  12 import ciso8601
  13 import time
  14 import unittest
  15 import parameterized
  16
  17 from unittest import mock
  18
  19 from . import run_test_server
  20 from arvados._ranges import Range, LocatorAndRange
  21 from arvados.collection import Collection, CollectionReader
  22 from . import arvados_testutil as tutil
  23 from .arvados_testutil import make_block_cache
  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 @parameterized.parameterized_class([{"disk_cache": True}, {"disk_cache": False}])
  32 class ArvadosCollectionsTest(run_test_server.TestCaseWithServers,
  33                              tutil.ArvadosBaseTestCase):
  34     disk_cache = False
  35     MAIN_SERVER = {}
  36
  37     @classmethod
  38     def setUpClass(cls):
  39         super(ArvadosCollectionsTest, cls).setUpClass()
  40         # need admin privileges to make collections with unsigned blocks
  41         run_test_server.authorize_with('admin')
  42         cls.api_client = arvados.api('v1')
  43         cls.keep_client = arvados.KeepClient(api_client=cls.api_client,
  44                                              local_store=cls.local_store,
  45                                              block_cache=make_block_cache(cls.disk_cache))
  46
  47     def write_foo_bar_baz(self):
  48         cw = arvados.CollectionWriter(self.api_client)
  49         self.assertEqual(cw.current_stream_name(), '.',
  50                          'current_stream_name() should be "." now')
  51         cw.set_current_file_name('foo.txt')
  52         cw.write(b'foo')
  53         self.assertEqual(cw.current_file_name(), 'foo.txt',
  54                          'current_file_name() should be foo.txt now')
  55         cw.start_new_file('bar.txt')
  56         cw.write(b'bar')
  57         cw.start_new_stream('baz')
  58         cw.write(b'baz')
  59         cw.set_current_file_name('baz.txt')
  60         self.assertEqual(cw.manifest_text(),
  61                          ". 3858f62230ac3c915f300c664312c63f+6 0:3:foo.txt 3:3:bar.txt\n" +
  62                          "./baz 73feffa4b7f6bb68e44cf984c85f6e88+3 0:3:baz.txt\n",
  63                          "wrong manifest: got {}".format(cw.manifest_text()))
  64         cw.save_new()
  65         return cw.portable_data_hash()
  66
  67     def test_pdh_is_native_str(self):
  68         pdh = self.write_foo_bar_baz()
  69         self.assertEqual(type(''), type(pdh))
  70
  71     def test_keep_local_store(self):
  72         self.assertEqual(self.keep_client.put(b'foo'), 'acbd18db4cc2f85cedef654fccc4a4d8+3', 'wrong md5 hash from Keep.put')
  73         self.assertEqual(self.keep_client.get('acbd18db4cc2f85cedef654fccc4a4d8+3'), b'foo', 'wrong data from Keep.get')
  74
  75     def test_local_collection_writer(self):
  76         self.assertEqual(self.write_foo_bar_baz(),
  77                          '23ca013983d6239e98931cc779e68426+114',
  78                          'wrong locator hash: ' + self.write_foo_bar_baz())
  79
  80     def test_local_collection_reader(self):
  81         foobarbaz = self.write_foo_bar_baz()
  82         cr = arvados.CollectionReader(
  83             foobarbaz + '+Xzizzle', self.api_client)
  84         got = []
  85         for s in cr.all_streams():
  86             for f in s.all_files():
  87                 got += [[f.size(), f.stream_name(), f.name(), f.read(2**26)]]
  88         expected = [[3, '.', 'foo.txt', b'foo'],
  89                     [3, '.', 'bar.txt', b'bar'],
  90                     [3, './baz', 'baz.txt', b'baz']]
  91         self.assertEqual(got,
  92                          expected)
  93         stream0 = cr.all_streams()[0]
  94         self.assertEqual(stream0.readfrom(0, 0),
  95                          b'',
  96                          'reading zero bytes should have returned empty string')
  97         self.assertEqual(stream0.readfrom(0, 2**26),
  98                          b'foobar',
  99                          'reading entire stream failed')
 100         self.assertEqual(stream0.readfrom(2**26, 0),
 101                          b'',
 102                          'reading zero bytes should have returned empty string')
 103         self.assertEqual(3, len(cr))
 104         self.assertTrue(cr)
 105
 106     def _test_subset(self, collection, expected):
 107         cr = arvados.CollectionReader(collection, self.api_client)
 108         for s in cr.all_streams():
 109             for ex in expected:
 110                 if ex[0] == s:
 111                     f = s.files()[ex[2]]
 112                     got = [f.size(), f.stream_name(), f.name(), "".join(f.readall(2**26))]
 113                     self.assertEqual(got,
 114                                      ex,
 115                                      'all_files|as_manifest did not preserve manifest contents: got %s expected %s' % (got, ex))
 116
 117     def test_collection_manifest_subset(self):
 118         foobarbaz = self.write_foo_bar_baz()
 119         self._test_subset(foobarbaz,
 120                           [[3, '.',     'bar.txt', b'bar'],
 121                            [3, '.',     'foo.txt', b'foo'],
 122                            [3, './baz', 'baz.txt', b'baz']])
 123         self._test_subset((". %s %s 0:3:foo.txt 3:3:bar.txt\n" %
 124                            (self.keep_client.put(b"foo"),
 125                             self.keep_client.put(b"bar"))),
 126                           [[3, '.', 'bar.txt', b'bar'],
 127                            [3, '.', 'foo.txt', b'foo']])
 128         self._test_subset((". %s %s 0:2:fo.txt 2:4:obar.txt\n" %
 129                            (self.keep_client.put(b"foo"),
 130                             self.keep_client.put(b"bar"))),
 131                           [[2, '.', 'fo.txt', b'fo'],
 132                            [4, '.', 'obar.txt', b'obar']])
 133         self._test_subset((". %s %s 0:2:fo.txt 2:0:zero.txt 2:2:ob.txt 4:2:ar.txt\n" %
 134                            (self.keep_client.put(b"foo"),
 135                             self.keep_client.put(b"bar"))),
 136                           [[2, '.', 'ar.txt', b'ar'],
 137                            [2, '.', 'fo.txt', b'fo'],
 138                            [2, '.', 'ob.txt', b'ob'],
 139                            [0, '.', 'zero.txt', b'']])
 140
 141     def test_collection_empty_file(self):
 142         cw = arvados.CollectionWriter(self.api_client)
 143         cw.start_new_file('zero.txt')
 144         cw.write(b'')
 145
 146         self.assertEqual(cw.manifest_text(), ". d41d8cd98f00b204e9800998ecf8427e+0 0:0:zero.txt\n")
 147         self.check_manifest_file_sizes(cw.manifest_text(), [0])
 148         cw = arvados.CollectionWriter(self.api_client)
 149         cw.start_new_file('zero.txt')
 150         cw.write(b'')
 151         cw.start_new_file('one.txt')
 152         cw.write(b'1')
 153         cw.start_new_stream('foo')
 154         cw.start_new_file('zero.txt')
 155         cw.write(b'')
 156         self.check_manifest_file_sizes(cw.manifest_text(), [0,1,0])
 157
 158     def test_no_implicit_normalize(self):
 159         cw = arvados.CollectionWriter(self.api_client)
 160         cw.start_new_file('b')
 161         cw.write(b'b')
 162         cw.start_new_file('a')
 163         cw.write(b'')
 164         self.check_manifest_file_sizes(cw.manifest_text(), [1,0])
 165         self.check_manifest_file_sizes(
 166             arvados.CollectionReader(
 167                 cw.manifest_text()).manifest_text(normalize=True),
 168             [0,1])
 169
 170     def check_manifest_file_sizes(self, manifest_text, expect_sizes):
 171         cr = arvados.CollectionReader(manifest_text, self.api_client)
 172         got_sizes = []
 173         for f in cr.all_files():
 174             got_sizes += [f.size()]
 175         self.assertEqual(got_sizes, expect_sizes, "got wrong file sizes %s, expected %s" % (got_sizes, expect_sizes))
 176
 177     def test_normalized_collection(self):
 178         m1 = """. 5348b82a029fd9e971a811ce1f71360b+43 0:43:md5sum.txt
 179 . 085c37f02916da1cad16f93c54d899b7+41 0:41:md5sum.txt
 180 . 8b22da26f9f433dea0a10e5ec66d73ba+43 0:43:md5sum.txt
 181 """
 182         self.assertEqual(arvados.CollectionReader(m1, self.api_client).manifest_text(normalize=True),
 183                          """. 5348b82a029fd9e971a811ce1f71360b+43 085c37f02916da1cad16f93c54d899b7+41 8b22da26f9f433dea0a10e5ec66d73ba+43 0:127:md5sum.txt
 184 """)
 185
 186         m2 = """. 204e43b8a1185621ca55a94839582e6f+67108864 b9677abbac956bd3e86b1deb28dfac03+67108864 fc15aff2a762b13f521baf042140acec+67108864 323d2a3ce20370c4ca1d3462a344f8fd+25885655 0:227212247:var-GS000016015-ASM.tsv.bz2
 187 """
 188         self.assertEqual(arvados.CollectionReader(m2, self.api_client).manifest_text(normalize=True), m2)
 189
 190         m3 = """. 5348b82a029fd9e971a811ce1f71360b+43 3:40:md5sum.txt
 191 . 085c37f02916da1cad16f93c54d899b7+41 0:41:md5sum.txt
 192 . 8b22da26f9f433dea0a10e5ec66d73ba+43 0:43:md5sum.txt
 193 """
 194         self.assertEqual(arvados.CollectionReader(m3, self.api_client).manifest_text(normalize=True),
 195                          """. 5348b82a029fd9e971a811ce1f71360b+43 085c37f02916da1cad16f93c54d899b7+41 8b22da26f9f433dea0a10e5ec66d73ba+43 3:124:md5sum.txt
 196 """)
 197
 198         m4 = """. 204e43b8a1185621ca55a94839582e6f+67108864 0:3:foo/bar
 199 ./zzz 204e43b8a1185621ca55a94839582e6f+67108864 0:999:zzz
 200 ./foo 323d2a3ce20370c4ca1d3462a344f8fd+25885655 0:3:bar
 201 """
 202         self.assertEqual(arvados.CollectionReader(m4, self.api_client).manifest_text(normalize=True),
 203                          """./foo 204e43b8a1185621ca55a94839582e6f+67108864 323d2a3ce20370c4ca1d3462a344f8fd+25885655 0:3:bar 67108864:3:bar
 204 ./zzz 204e43b8a1185621ca55a94839582e6f+67108864 0:999:zzz
 205 """)
 206
 207         m5 = """. 204e43b8a1185621ca55a94839582e6f+67108864 0:3:foo/bar
 208 ./zzz 204e43b8a1185621ca55a94839582e6f+67108864 0:999:zzz
 209 ./foo 204e43b8a1185621ca55a94839582e6f+67108864 3:3:bar
 210 """
 211         self.assertEqual(arvados.CollectionReader(m5, self.api_client).manifest_text(normalize=True),
 212                          """./foo 204e43b8a1185621ca55a94839582e6f+67108864 0:6:bar
 213 ./zzz 204e43b8a1185621ca55a94839582e6f+67108864 0:999:zzz
 214 """)
 215
 216         with self.data_file('1000G_ref_manifest') as f6:
 217             m6 = f6.read()
 218             self.assertEqual(arvados.CollectionReader(m6, self.api_client).manifest_text(normalize=True), m6)
 219
 220         with self.data_file('jlake_manifest') as f7:
 221             m7 = f7.read()
 222             self.assertEqual(arvados.CollectionReader(m7, self.api_client).manifest_text(normalize=True), m7)
 223
 224         m8 = """./a\\040b\\040c 59ca0efa9f5633cb0371bbc0355478d8+13 0:13:hello\\040world.txt
 225 """
 226         self.assertEqual(arvados.CollectionReader(m8, self.api_client).manifest_text(normalize=True), m8)
 227
 228     def test_locators_and_ranges(self):
 229         blocks2 = [Range('a', 0, 10),
 230                    Range('b', 10, 10),
 231                    Range('c', 20, 10),
 232                    Range('d', 30, 10),
 233                    Range('e', 40, 10),
 234                    Range('f', 50, 10)]
 235
 236         self.assertEqual(arvados.locators_and_ranges(blocks2,  2,  2), [LocatorAndRange('a', 10, 2, 2)])
 237         self.assertEqual(arvados.locators_and_ranges(blocks2, 12, 2), [LocatorAndRange('b', 10, 2, 2)])
 238         self.assertEqual(arvados.locators_and_ranges(blocks2, 22, 2), [LocatorAndRange('c', 10, 2, 2)])
 239         self.assertEqual(arvados.locators_and_ranges(blocks2, 32, 2), [LocatorAndRange('d', 10, 2, 2)])
 240         self.assertEqual(arvados.locators_and_ranges(blocks2, 42, 2), [LocatorAndRange('e', 10, 2, 2)])
 241         self.assertEqual(arvados.locators_and_ranges(blocks2, 52, 2), [LocatorAndRange('f', 10, 2, 2)])
 242         self.assertEqual(arvados.locators_and_ranges(blocks2, 62, 2), [])
 243         self.assertEqual(arvados.locators_and_ranges(blocks2, -2, 2), [])
 244
 245         self.assertEqual(arvados.locators_and_ranges(blocks2,  0,  2), [LocatorAndRange('a', 10, 0, 2)])
 246         self.assertEqual(arvados.locators_and_ranges(blocks2, 10, 2), [LocatorAndRange('b', 10, 0, 2)])
 247         self.assertEqual(arvados.locators_and_ranges(blocks2, 20, 2), [LocatorAndRange('c', 10, 0, 2)])
 248         self.assertEqual(arvados.locators_and_ranges(blocks2, 30, 2), [LocatorAndRange('d', 10, 0, 2)])
 249         self.assertEqual(arvados.locators_and_ranges(blocks2, 40, 2), [LocatorAndRange('e', 10, 0, 2)])
 250         self.assertEqual(arvados.locators_and_ranges(blocks2, 50, 2), [LocatorAndRange('f', 10, 0, 2)])
 251         self.assertEqual(arvados.locators_and_ranges(blocks2, 60, 2), [])
 252         self.assertEqual(arvados.locators_and_ranges(blocks2, -2, 2), [])
 253
 254         self.assertEqual(arvados.locators_and_ranges(blocks2,  9,  2), [LocatorAndRange('a', 10, 9, 1), LocatorAndRange('b', 10, 0, 1)])
 255         self.assertEqual(arvados.locators_and_ranges(blocks2, 19, 2), [LocatorAndRange('b', 10, 9, 1), LocatorAndRange('c', 10, 0, 1)])
 256         self.assertEqual(arvados.locators_and_ranges(blocks2, 29, 2), [LocatorAndRange('c', 10, 9, 1), LocatorAndRange('d', 10, 0, 1)])
 257         self.assertEqual(arvados.locators_and_ranges(blocks2, 39, 2), [LocatorAndRange('d', 10, 9, 1), LocatorAndRange('e', 10, 0, 1)])
 258         self.assertEqual(arvados.locators_and_ranges(blocks2, 49, 2), [LocatorAndRange('e', 10, 9, 1), LocatorAndRange('f', 10, 0, 1)])
 259         self.assertEqual(arvados.locators_and_ranges(blocks2, 59, 2), [LocatorAndRange('f', 10, 9, 1)])
 260
 261
 262         blocks3 = [Range('a', 0, 10),
 263                   Range('b', 10, 10),
 264                   Range('c', 20, 10),
 265                   Range('d', 30, 10),
 266                   Range('e', 40, 10),
 267                   Range('f', 50, 10),
 268                    Range('g', 60, 10)]
 269
 270         self.assertEqual(arvados.locators_and_ranges(blocks3,  2,  2), [LocatorAndRange('a', 10, 2, 2)])
 271         self.assertEqual(arvados.locators_and_ranges(blocks3, 12, 2), [LocatorAndRange('b', 10, 2, 2)])
 272         self.assertEqual(arvados.locators_and_ranges(blocks3, 22, 2), [LocatorAndRange('c', 10, 2, 2)])
 273         self.assertEqual(arvados.locators_and_ranges(blocks3, 32, 2), [LocatorAndRange('d', 10, 2, 2)])
 274         self.assertEqual(arvados.locators_and_ranges(blocks3, 42, 2), [LocatorAndRange('e', 10, 2, 2)])
 275         self.assertEqual(arvados.locators_and_ranges(blocks3, 52, 2), [LocatorAndRange('f', 10, 2, 2)])
 276         self.assertEqual(arvados.locators_and_ranges(blocks3, 62, 2), [LocatorAndRange('g', 10, 2, 2)])
 277
 278
 279         blocks = [Range('a', 0, 10),
 280                   Range('b', 10, 15),
 281                   Range('c', 25, 5)]
 282         self.assertEqual(arvados.locators_and_ranges(blocks, 1, 0), [])
 283         self.assertEqual(arvados.locators_and_ranges(blocks, 0, 5), [LocatorAndRange('a', 10, 0, 5)])
 284         self.assertEqual(arvados.locators_and_ranges(blocks, 3, 5), [LocatorAndRange('a', 10, 3, 5)])
 285         self.assertEqual(arvados.locators_and_ranges(blocks, 0, 10), [LocatorAndRange('a', 10, 0, 10)])
 286
 287         self.assertEqual(arvados.locators_and_ranges(blocks, 0, 11), [LocatorAndRange('a', 10, 0, 10),
 288                                                                       LocatorAndRange('b', 15, 0, 1)])
 289         self.assertEqual(arvados.locators_and_ranges(blocks, 1, 11), [LocatorAndRange('a', 10, 1, 9),
 290                                                                       LocatorAndRange('b', 15, 0, 2)])
 291         self.assertEqual(arvados.locators_and_ranges(blocks, 0, 25), [LocatorAndRange('a', 10, 0, 10),
 292                                                                       LocatorAndRange('b', 15, 0, 15)])
 293
 294         self.assertEqual(arvados.locators_and_ranges(blocks, 0, 30), [LocatorAndRange('a', 10, 0, 10),
 295                                                                       LocatorAndRange('b', 15, 0, 15),
 296                                                                       LocatorAndRange('c', 5, 0, 5)])
 297         self.assertEqual(arvados.locators_and_ranges(blocks, 1, 30), [LocatorAndRange('a', 10, 1, 9),
 298                                                                       LocatorAndRange('b', 15, 0, 15),
 299                                                                       LocatorAndRange('c', 5, 0, 5)])
 300         self.assertEqual(arvados.locators_and_ranges(blocks, 0, 31), [LocatorAndRange('a', 10, 0, 10),
 301                                                                       LocatorAndRange('b', 15, 0, 15),
 302                                                                       LocatorAndRange('c', 5, 0, 5)])
 303
 304         self.assertEqual(arvados.locators_and_ranges(blocks, 15, 5), [LocatorAndRange('b', 15, 5, 5)])
 305
 306         self.assertEqual(arvados.locators_and_ranges(blocks, 8, 17), [LocatorAndRange('a', 10, 8, 2),
 307                                                                       LocatorAndRange('b', 15, 0, 15)])
 308
 309         self.assertEqual(arvados.locators_and_ranges(blocks, 8, 20), [LocatorAndRange('a', 10, 8, 2),
 310                                                                       LocatorAndRange('b', 15, 0, 15),
 311                                                                       LocatorAndRange('c', 5, 0, 3)])
 312
 313         self.assertEqual(arvados.locators_and_ranges(blocks, 26, 2), [LocatorAndRange('c', 5, 1, 2)])
 314
 315         self.assertEqual(arvados.locators_and_ranges(blocks, 9, 15), [LocatorAndRange('a', 10, 9, 1),
 316                                                                       LocatorAndRange('b', 15, 0, 14)])
 317         self.assertEqual(arvados.locators_and_ranges(blocks, 10, 15), [LocatorAndRange('b', 15, 0, 15)])
 318         self.assertEqual(arvados.locators_and_ranges(blocks, 11, 15), [LocatorAndRange('b', 15, 1, 14),
 319                                                                        LocatorAndRange('c', 5, 0, 1)])
 320
 321     class MockKeep(object):
 322         def __init__(self, content, num_retries=0):
 323             self.content = content
 324             self.num_prefetch_threads = 1
 325
 326         def get(self, locator, num_retries=0, prefetch=False):
 327             return self.content[locator]
 328
 329     def test_stream_reader(self):
 330         keepblocks = {
 331             'aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa+10': b'abcdefghij',
 332             'bbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbb+15': b'klmnopqrstuvwxy',
 333             'cccccccccccccccccccccccccccccccc+5': b'z0123',
 334         }
 335         mk = self.MockKeep(keepblocks)
 336
 337         sr = arvados.StreamReader([".", "aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa+10", "bbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbb+15", "cccccccccccccccccccccccccccccccc+5", "0:30:foo"], mk)
 338
 339         content = b'abcdefghijklmnopqrstuvwxyz0123456789'
 340
 341         self.assertEqual(sr.readfrom(0, 30), content[0:30])
 342         self.assertEqual(sr.readfrom(2, 30), content[2:30])
 343
 344         self.assertEqual(sr.readfrom(2, 8), content[2:10])
 345         self.assertEqual(sr.readfrom(0, 10), content[0:10])
 346
 347         self.assertEqual(sr.readfrom(0, 5), content[0:5])
 348         self.assertEqual(sr.readfrom(5, 5), content[5:10])
 349         self.assertEqual(sr.readfrom(10, 5), content[10:15])
 350         self.assertEqual(sr.readfrom(15, 5), content[15:20])
 351         self.assertEqual(sr.readfrom(20, 5), content[20:25])
 352         self.assertEqual(sr.readfrom(25, 5), content[25:30])
 353         self.assertEqual(sr.readfrom(30, 5), b'')
 354
 355     def test_extract_file(self):
 356         m1 = """. 5348b82a029fd9e971a811ce1f71360b+43 0:43:md5sum.txt
 357 . 085c37f02916da1cad16f93c54d899b7+41 0:41:md6sum.txt
 358 . 8b22da26f9f433dea0a10e5ec66d73ba+43 0:43:md7sum.txt
 359 . 085c37f02916da1cad16f93c54d899b7+41 5348b82a029fd9e971a811ce1f71360b+43 8b22da26f9f433dea0a10e5ec66d73ba+43 47:80:md8sum.txt
 360 . 085c37f02916da1cad16f93c54d899b7+41 5348b82a029fd9e971a811ce1f71360b+43 8b22da26f9f433dea0a10e5ec66d73ba+43 40:80:md9sum.txt
 361 """
 362
 363         m2 = arvados.CollectionReader(m1, self.api_client).manifest_text(normalize=True)
 364
 365         self.assertEqual(m2,
 366                          ". 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")
 367         files = arvados.CollectionReader(
 368             m2, self.api_client).all_streams()[0].files()
 369
 370         self.assertEqual(files['md5sum.txt'].as_manifest(),
 371                          ". 5348b82a029fd9e971a811ce1f71360b+43 0:43:md5sum.txt\n")
 372         self.assertEqual(files['md6sum.txt'].as_manifest(),
 373                          ". 085c37f02916da1cad16f93c54d899b7+41 0:41:md6sum.txt\n")
 374         self.assertEqual(files['md7sum.txt'].as_manifest(),
 375                          ". 8b22da26f9f433dea0a10e5ec66d73ba+43 0:43:md7sum.txt\n")
 376         self.assertEqual(files['md9sum.txt'].as_manifest(),
 377                          ". 085c37f02916da1cad16f93c54d899b7+41 5348b82a029fd9e971a811ce1f71360b+43 8b22da26f9f433dea0a10e5ec66d73ba+43 40:80:md9sum.txt\n")
 378
 379     def test_write_directory_tree(self):
 380         cwriter = arvados.CollectionWriter(self.api_client)
 381         cwriter.write_directory_tree(self.build_directory_tree(
 382                 ['basefile', 'subdir/subfile']))
 383         self.assertEqual(cwriter.manifest_text(),
 384                          """. c5110c5ac93202d8e0f9e381f22bac0f+8 0:8:basefile
 385 ./subdir 1ca4dec89403084bf282ad31e6cf7972+14 0:14:subfile\n""")
 386
 387     def test_write_named_directory_tree(self):
 388         cwriter = arvados.CollectionWriter(self.api_client)
 389         cwriter.write_directory_tree(self.build_directory_tree(
 390                 ['basefile', 'subdir/subfile']), 'root')
 391         self.assertEqual(
 392             cwriter.manifest_text(),
 393             """./root c5110c5ac93202d8e0f9e381f22bac0f+8 0:8:basefile
 394 ./root/subdir 1ca4dec89403084bf282ad31e6cf7972+14 0:14:subfile\n""")
 395
 396     def test_write_directory_tree_in_one_stream(self):
 397         cwriter = arvados.CollectionWriter(self.api_client)
 398         cwriter.write_directory_tree(self.build_directory_tree(
 399                 ['basefile', 'subdir/subfile']), max_manifest_depth=0)
 400         self.assertEqual(cwriter.manifest_text(),
 401                          """. 4ace875ffdc6824a04950f06858f4465+22 0:8:basefile 8:14:subdir/subfile\n""")
 402
 403     def test_write_directory_tree_with_limited_recursion(self):
 404         cwriter = arvados.CollectionWriter(self.api_client)
 405         cwriter.write_directory_tree(
 406             self.build_directory_tree(['f1', 'd1/f2', 'd1/d2/f3']),
 407             max_manifest_depth=1)
 408         self.assertEqual(cwriter.manifest_text(),
 409                          """. bd19836ddb62c11c55ab251ccaca5645+2 0:2:f1
 410 ./d1 50170217e5b04312024aa5cd42934494+13 0:8:d2/f3 8:5:f2\n""")
 411
 412     def test_write_directory_tree_with_zero_recursion(self):
 413         cwriter = arvados.CollectionWriter(self.api_client)
 414         content = 'd1/d2/f3d1/f2f1'
 415         blockhash = tutil.str_keep_locator(content)
 416         cwriter.write_directory_tree(
 417             self.build_directory_tree(['f1', 'd1/f2', 'd1/d2/f3']),
 418             max_manifest_depth=0)
 419         self.assertEqual(
 420             cwriter.manifest_text(),
 421             ". {} 0:8:d1/d2/f3 8:5:d1/f2 13:2:f1\n".format(blockhash))
 422
 423     def test_write_one_file(self):
 424         cwriter = arvados.CollectionWriter(self.api_client)
 425         with self.make_test_file() as testfile:
 426             cwriter.write_file(testfile.name)
 427             self.assertEqual(
 428                 cwriter.manifest_text(),
 429                 ". 098f6bcd4621d373cade4e832627b4f6+4 0:4:{}\n".format(
 430                     os.path.basename(testfile.name)))
 431
 432     def test_write_named_file(self):
 433         cwriter = arvados.CollectionWriter(self.api_client)
 434         with self.make_test_file() as testfile:
 435             cwriter.write_file(testfile.name, 'foo')
 436             self.assertEqual(cwriter.manifest_text(),
 437                              ". 098f6bcd4621d373cade4e832627b4f6+4 0:4:foo\n")
 438
 439     def test_write_multiple_files(self):
 440         cwriter = arvados.CollectionWriter(self.api_client)
 441         for letter in 'ABC':
 442             with self.make_test_file(letter.encode()) as testfile:
 443                 cwriter.write_file(testfile.name, letter)
 444         self.assertEqual(
 445             cwriter.manifest_text(),
 446             ". 902fbdd2b1df0c4f70b4a5d23525e932+3 0:1:A 1:1:B 2:1:C\n")
 447
 448     def test_basic_resume(self):
 449         cwriter = TestResumableWriter()
 450         with self.make_test_file() as testfile:
 451             cwriter.write_file(testfile.name, 'test')
 452             resumed = TestResumableWriter.from_state(cwriter.current_state())
 453         self.assertEqual(cwriter.manifest_text(), resumed.manifest_text(),
 454                           "resumed CollectionWriter had different manifest")
 455
 456     def test_resume_fails_when_missing_dependency(self):
 457         cwriter = TestResumableWriter()
 458         with self.make_test_file() as testfile:
 459             cwriter.write_file(testfile.name, 'test')
 460         self.assertRaises(arvados.errors.StaleWriterStateError,
 461                           TestResumableWriter.from_state,
 462                           cwriter.current_state())
 463
 464     def test_resume_fails_when_dependency_mtime_changed(self):
 465         cwriter = TestResumableWriter()
 466         with self.make_test_file() as testfile:
 467             cwriter.write_file(testfile.name, 'test')
 468             os.utime(testfile.name, (0, 0))
 469             self.assertRaises(arvados.errors.StaleWriterStateError,
 470                               TestResumableWriter.from_state,
 471                               cwriter.current_state())
 472
 473     def test_resume_fails_when_dependency_is_nonfile(self):
 474         cwriter = TestResumableWriter()
 475         cwriter.write_file('/dev/null', 'empty')
 476         self.assertRaises(arvados.errors.StaleWriterStateError,
 477                           TestResumableWriter.from_state,
 478                           cwriter.current_state())
 479
 480     def test_resume_fails_when_dependency_size_changed(self):
 481         cwriter = TestResumableWriter()
 482         with self.make_test_file() as testfile:
 483             cwriter.write_file(testfile.name, 'test')
 484             orig_mtime = os.fstat(testfile.fileno()).st_mtime
 485             testfile.write(b'extra')
 486             testfile.flush()
 487             os.utime(testfile.name, (orig_mtime, orig_mtime))
 488             self.assertRaises(arvados.errors.StaleWriterStateError,
 489                               TestResumableWriter.from_state,
 490                               cwriter.current_state())
 491
 492     def test_resume_fails_with_expired_locator(self):
 493         cwriter = TestResumableWriter()
 494         state = cwriter.current_state()
 495         # Add an expired locator to the state.
 496         state['_current_stream_locators'].append(''.join([
 497                     'a' * 32, '+1+A', 'b' * 40, '@', '10000000']))
 498         self.assertRaises(arvados.errors.StaleWriterStateError,
 499                           TestResumableWriter.from_state, state)
 500
 501     def test_arbitrary_objects_not_resumable(self):
 502         cwriter = TestResumableWriter()
 503         with open('/dev/null') as badfile:
 504             self.assertRaises(arvados.errors.AssertionError,
 505                               cwriter.write_file, badfile)
 506
 507     def test_arbitrary_writes_not_resumable(self):
 508         cwriter = TestResumableWriter()
 509         self.assertRaises(arvados.errors.AssertionError,
 510                           cwriter.write, "badtext")
 511
 512
 513 class CollectionTestMixin(tutil.ApiClientMock):
 514     API_COLLECTIONS = run_test_server.fixture('collections')
 515     DEFAULT_COLLECTION = API_COLLECTIONS['foo_file']
 516     DEFAULT_DATA_HASH = DEFAULT_COLLECTION['portable_data_hash']
 517     DEFAULT_MANIFEST = DEFAULT_COLLECTION['manifest_text']
 518     DEFAULT_UUID = DEFAULT_COLLECTION['uuid']
 519     ALT_COLLECTION = API_COLLECTIONS['bar_file']
 520     ALT_DATA_HASH = ALT_COLLECTION['portable_data_hash']
 521     ALT_MANIFEST = ALT_COLLECTION['manifest_text']
 522
 523     def api_client_mock(self, status=200):
 524         client = super(CollectionTestMixin, self).api_client_mock()
 525         self.mock_keep_services(client, status=status, service_type='proxy', count=1)
 526         return client
 527
 528
 529 @tutil.skip_sleep
 530 class CollectionReaderTestCase(unittest.TestCase, CollectionTestMixin):
 531     def mock_get_collection(self, api_mock, code, fixturename):
 532         body = self.API_COLLECTIONS.get(fixturename)
 533         self._mock_api_call(api_mock.collections().get, code, body)
 534
 535     def api_client_mock(self, status=200):
 536         client = super(CollectionReaderTestCase, self).api_client_mock()
 537         self.mock_get_collection(client, status, 'foo_file')
 538         return client
 539
 540     def test_init_default_retries(self):
 541         client = self.api_client_mock(200)
 542         reader = arvados.CollectionReader(self.DEFAULT_UUID, api_client=client)
 543         reader.manifest_text()
 544         client.collections().get().execute.assert_called_with(num_retries=10)
 545
 546     def test_uuid_init_success(self):
 547         client = self.api_client_mock(200)
 548         reader = arvados.CollectionReader(self.DEFAULT_UUID, api_client=client,
 549                                           num_retries=3)
 550         self.assertEqual(self.DEFAULT_COLLECTION['manifest_text'],
 551                          reader.manifest_text())
 552         client.collections().get().execute.assert_called_with(num_retries=3)
 553
 554     def test_uuid_init_failure_raises_api_error(self):
 555         client = self.api_client_mock(500)
 556         with self.assertRaises(arvados.errors.ApiError):
 557             reader = arvados.CollectionReader(self.DEFAULT_UUID, api_client=client)
 558
 559     def test_locator_init(self):
 560         client = self.api_client_mock(200)
 561         # Ensure Keep will not return anything if asked.
 562         with tutil.mock_keep_responses(None, 404):
 563             reader = arvados.CollectionReader(self.DEFAULT_DATA_HASH,
 564                                               api_client=client)
 565             self.assertEqual(self.DEFAULT_MANIFEST, reader.manifest_text())
 566
 567     def test_init_no_fallback_to_keep(self):
 568         # Do not look up a collection UUID or PDH in Keep.
 569         for key in [self.DEFAULT_UUID, self.DEFAULT_DATA_HASH]:
 570             client = self.api_client_mock(404)
 571             with tutil.mock_keep_responses(self.DEFAULT_MANIFEST, 200):
 572                 with self.assertRaises(arvados.errors.ApiError):
 573                     reader = arvados.CollectionReader(key, api_client=client)
 574
 575     def test_init_num_retries_propagated(self):
 576         # More of an integration test...
 577         client = self.api_client_mock(200)
 578         reader = arvados.CollectionReader(self.DEFAULT_UUID, api_client=client,
 579                                           num_retries=3)
 580         with tutil.mock_keep_responses('foo', 500, 500, 200):
 581             self.assertEqual(b'foo',
 582                              b''.join(f.read(9) for f in reader.all_files()))
 583
 584     def test_read_nonnormalized_manifest_with_collection_reader(self):
 585         # client should be able to use CollectionReader on a manifest without normalizing it
 586         client = self.api_client_mock(500)
 587         nonnormal = ". acbd18db4cc2f85cedef654fccc4a4d8+3+Aabadbadbee@abeebdee 0:3:foo.txt 1:0:bar.txt 0:3:foo.txt\n"
 588         reader = arvados.CollectionReader(
 589             nonnormal,
 590             api_client=client, num_retries=0)
 591         # Ensure stripped_manifest() doesn't mangle our manifest in
 592         # any way other than stripping hints.
 593         self.assertEqual(
 594             re.sub(r'\+[^\d\s\+]+', '', nonnormal),
 595             reader.stripped_manifest())
 596         # Ensure stripped_manifest() didn't mutate our reader.
 597         self.assertEqual(nonnormal, reader.manifest_text())
 598         # Ensure the files appear in the order given in the manifest.
 599         self.assertEqual(
 600             [[6, '.', 'foo.txt'],
 601              [0, '.', 'bar.txt']],
 602             [[f.size(), f.stream_name(), f.name()]
 603              for f in reader.all_streams()[0].all_files()])
 604
 605     def test_read_empty_collection(self):
 606         client = self.api_client_mock(200)
 607         self.mock_get_collection(client, 200, 'empty')
 608         reader = arvados.CollectionReader('d41d8cd98f00b204e9800998ecf8427e+0',
 609                                           api_client=client)
 610         self.assertEqual('', reader.manifest_text())
 611         self.assertEqual(0, len(reader))
 612         self.assertFalse(reader)
 613
 614     def test_api_response(self):
 615         client = self.api_client_mock()
 616         reader = arvados.CollectionReader(self.DEFAULT_UUID, api_client=client)
 617         self.assertEqual(self.DEFAULT_COLLECTION, reader.api_response())
 618
 619     def check_open_file(self, coll_file, stream_name, file_name, file_size):
 620         self.assertFalse(coll_file.closed, "returned file is not open")
 621         self.assertEqual(stream_name, coll_file.stream_name())
 622         self.assertEqual(file_name, coll_file.name)
 623         self.assertEqual(file_size, coll_file.size())
 624
 625     def test_open_collection_file_one_argument(self):
 626         client = self.api_client_mock(200)
 627         reader = arvados.CollectionReader(self.DEFAULT_UUID, api_client=client)
 628         cfile = reader.open('./foo', 'rb')
 629         self.check_open_file(cfile, '.', 'foo', 3)
 630
 631     def test_open_deep_file(self):
 632         coll_name = 'collection_with_files_in_subdir'
 633         client = self.api_client_mock(200)
 634         self.mock_get_collection(client, 200, coll_name)
 635         reader = arvados.CollectionReader(
 636             self.API_COLLECTIONS[coll_name]['uuid'], api_client=client)
 637         cfile = reader.open('./subdir2/subdir3/file2_in_subdir3.txt', 'rb')
 638         self.check_open_file(cfile, './subdir2/subdir3', 'file2_in_subdir3.txt',
 639                              32)
 640
 641     def test_open_nonexistent_stream(self):
 642         client = self.api_client_mock(200)
 643         reader = arvados.CollectionReader(self.DEFAULT_UUID, api_client=client)
 644         self.assertRaises(IOError, reader.open, './nonexistent/foo')
 645
 646     def test_open_nonexistent_file(self):
 647         client = self.api_client_mock(200)
 648         reader = arvados.CollectionReader(self.DEFAULT_UUID, api_client=client)
 649         self.assertRaises(IOError, reader.open, 'nonexistent')
 650
 651
 652 @tutil.skip_sleep
 653 class CollectionWriterTestCase(unittest.TestCase, CollectionTestMixin):
 654     def mock_keep(self, body, *codes, **headers):
 655         headers.setdefault('x-keep-replicas-stored', 2)
 656         return tutil.mock_keep_responses(body, *codes, **headers)
 657
 658     def foo_writer(self, **kwargs):
 659         kwargs.setdefault('api_client', self.api_client_mock())
 660         writer = arvados.CollectionWriter(**kwargs)
 661         writer.start_new_file('foo')
 662         writer.write(b'foo')
 663         return writer
 664
 665     def test_write_whole_collection(self):
 666         writer = self.foo_writer()
 667         with self.mock_keep(self.DEFAULT_DATA_HASH, 200, 200):
 668             self.assertEqual(self.DEFAULT_DATA_HASH, writer.finish())
 669
 670     def test_write_no_default(self):
 671         writer = self.foo_writer()
 672         with self.mock_keep(None, 500):
 673             with self.assertRaises(arvados.errors.KeepWriteError):
 674                 writer.finish()
 675
 676     def test_write_insufficient_replicas_via_proxy(self):
 677         writer = self.foo_writer(replication=3)
 678         with self.mock_keep(None, 200, **{'x-keep-replicas-stored': 2}):
 679             with self.assertRaises(arvados.errors.KeepWriteError):
 680                 writer.manifest_text()
 681
 682     def test_write_insufficient_replicas_via_disks(self):
 683         client = mock.MagicMock(name='api_client')
 684         with self.mock_keep(
 685                 None, 200, 200,
 686                 **{'x-keep-replicas-stored': 1}) as keepmock:
 687             self.mock_keep_services(client, status=200, service_type='disk', count=2)
 688             writer = self.foo_writer(api_client=client, replication=3)
 689             with self.assertRaises(arvados.errors.KeepWriteError):
 690                 writer.manifest_text()
 691
 692     def test_write_three_replicas(self):
 693         client = mock.MagicMock(name='api_client')
 694         with self.mock_keep(
 695                 "", 500, 500, 500, 200, 200, 200,
 696                 **{'x-keep-replicas-stored': 1}) as keepmock:
 697             self.mock_keep_services(client, status=200, service_type='disk', count=6)
 698             writer = self.foo_writer(api_client=client, replication=3)
 699             writer.manifest_text()
 700             self.assertEqual(6, keepmock.call_count)
 701
 702     def test_write_whole_collection_through_retries(self):
 703         writer = self.foo_writer(num_retries=2)
 704         with self.mock_keep(self.DEFAULT_DATA_HASH,
 705                             500, 500, 200, 500, 500, 200):
 706             self.assertEqual(self.DEFAULT_DATA_HASH, writer.finish())
 707
 708     def test_flush_data_retries(self):
 709         writer = self.foo_writer(num_retries=2)
 710         foo_hash = self.DEFAULT_MANIFEST.split()[1]
 711         with self.mock_keep(foo_hash, 500, 200):
 712             writer.flush_data()
 713         self.assertEqual(self.DEFAULT_MANIFEST, writer.manifest_text())
 714
 715     def test_one_open(self):
 716         client = self.api_client_mock()
 717         writer = arvados.CollectionWriter(client)
 718         with writer.open('out') as out_file:
 719             self.assertEqual('.', writer.current_stream_name())
 720             self.assertEqual('out', writer.current_file_name())
 721             out_file.write(b'test data')
 722             data_loc = tutil.str_keep_locator('test data')
 723         self.assertTrue(out_file.closed, "writer file not closed after context")
 724         self.assertRaises(ValueError, out_file.write, 'extra text')
 725         with self.mock_keep(data_loc, 200) as keep_mock:
 726             self.assertEqual(". {} 0:9:out\n".format(data_loc),
 727                              writer.manifest_text())
 728
 729     def test_open_writelines(self):
 730         client = self.api_client_mock()
 731         writer = arvados.CollectionWriter(client)
 732         with writer.open('six') as out_file:
 733             out_file.writelines(['12', '34', '56'])
 734             data_loc = tutil.str_keep_locator('123456')
 735         with self.mock_keep(data_loc, 200) as keep_mock:
 736             self.assertEqual(". {} 0:6:six\n".format(data_loc),
 737                              writer.manifest_text())
 738
 739     def test_open_flush(self):
 740         client = self.api_client_mock()
 741         data_loc1 = tutil.str_keep_locator('flush1')
 742         data_loc2 = tutil.str_keep_locator('flush2')
 743         with self.mock_keep((data_loc1, 200), (data_loc2, 200)) as keep_mock:
 744             writer = arvados.CollectionWriter(client)
 745             with writer.open('flush_test') as out_file:
 746                 out_file.write(b'flush1')
 747                 out_file.flush()
 748                 out_file.write(b'flush2')
 749             self.assertEqual(". {} {} 0:12:flush_test\n".format(data_loc1,
 750                                                                 data_loc2),
 751                              writer.manifest_text())
 752
 753     def test_two_opens_same_stream(self):
 754         client = self.api_client_mock()
 755         writer = arvados.CollectionWriter(client)
 756         with writer.open('.', '1') as out_file:
 757             out_file.write(b'1st')
 758         with writer.open('.', '2') as out_file:
 759             out_file.write(b'2nd')
 760         data_loc = tutil.str_keep_locator('1st2nd')
 761         with self.mock_keep(data_loc, 200) as keep_mock:
 762             self.assertEqual(". {} 0:3:1 3:3:2\n".format(data_loc),
 763                              writer.manifest_text())
 764
 765     def test_two_opens_two_streams(self):
 766         client = self.api_client_mock()
 767         data_loc1 = tutil.str_keep_locator('file')
 768         data_loc2 = tutil.str_keep_locator('indir')
 769         with self.mock_keep((data_loc1, 200), (data_loc2, 200)) as keep_mock:
 770             writer = arvados.CollectionWriter(client)
 771             with writer.open('file') as out_file:
 772                 out_file.write(b'file')
 773             with writer.open('./dir', 'indir') as out_file:
 774                 out_file.write(b'indir')
 775             expected = ". {} 0:4:file\n./dir {} 0:5:indir\n".format(
 776                 data_loc1, data_loc2)
 777             self.assertEqual(expected, writer.manifest_text())
 778
 779     def test_dup_open_fails(self):
 780         client = self.api_client_mock()
 781         writer = arvados.CollectionWriter(client)
 782         file1 = writer.open('one')
 783         self.assertRaises(arvados.errors.AssertionError, writer.open, 'two')
 784
 785
 786 class CollectionMethods(run_test_server.TestCaseWithServers):
 787
 788     def test_keys_values_items_support_indexing(self):
 789         c = Collection()
 790         with c.open('foo', 'wb') as f:
 791             f.write(b'foo')
 792         with c.open('bar', 'wb') as f:
 793             f.write(b'bar')
 794         self.assertEqual(2, len(c.keys()))
 795         if sys.version_info < (3, 0):
 796             # keys() supports indexing only for python2 callers.
 797             fn0 = c.keys()[0]
 798             fn1 = c.keys()[1]
 799         else:
 800             fn0, fn1 = c.keys()
 801         self.assertEqual(2, len(c.values()))
 802         f0 = c.values()[0]
 803         f1 = c.values()[1]
 804         self.assertEqual(2, len(c.items()))
 805         self.assertEqual(fn0, c.items()[0][0])
 806         self.assertEqual(fn1, c.items()[1][0])
 807
 808     def test_get_properties(self):
 809         c = Collection()
 810         self.assertEqual(c.get_properties(), {})
 811         c.save_new(properties={"foo":"bar"})
 812         self.assertEqual(c.get_properties(), {"foo":"bar"})
 813
 814     def test_get_trash_at(self):
 815         c = Collection()
 816         self.assertEqual(c.get_trash_at(), None)
 817         c.save_new(trash_at=datetime.datetime(2111, 1, 1, 11, 11, 11, 111111))
 818         self.assertEqual(c.get_trash_at(), ciso8601.parse_datetime('2111-01-01T11:11:11.111111000Z'))
 819
 820
 821 class CollectionOpenModes(run_test_server.TestCaseWithServers):
 822
 823     def test_open_binary_modes(self):
 824         c = Collection()
 825         for mode in ['wb', 'wb+', 'ab', 'ab+']:
 826             with c.open('foo', mode) as f:
 827                 f.write(b'foo')
 828
 829     def test_open_invalid_modes(self):
 830         c = Collection()
 831         for mode in ['+r', 'aa', '++', 'r+b', 'beer', '', None]:
 832             with self.assertRaises(Exception):
 833                 c.open('foo', mode)
 834
 835     def test_open_text_modes(self):
 836         c = Collection()
 837         with c.open('foo', 'wb') as f:
 838             f.write('foo')
 839         for mode in ['r', 'rt', 'r+', 'rt+', 'w', 'wt', 'a', 'at']:
 840             with c.open('foo', mode) as f:
 841                 if mode[0] == 'r' and '+' not in mode:
 842                     self.assertEqual('foo', f.read(3))
 843                 else:
 844                     f.write('bar')
 845                     f.seek(0, os.SEEK_SET)
 846                     self.assertEqual('bar', f.read(3))
 847
 848
 849 class TextModes(run_test_server.TestCaseWithServers):
 850
 851     def setUp(self):
 852         arvados.config.KEEP_BLOCK_SIZE = 4
 853         if sys.version_info < (3, 0):
 854             import unicodedata
 855             self.sailboat = unicodedata.lookup('SAILBOAT')
 856             self.snowman = unicodedata.lookup('SNOWMAN')
 857         else:
 858             self.sailboat = '\N{SAILBOAT}'
 859             self.snowman = '\N{SNOWMAN}'
 860
 861     def tearDown(self):
 862         arvados.config.KEEP_BLOCK_SIZE = 2 ** 26
 863
 864     def test_read_sailboat_across_block_boundary(self):
 865         c = Collection()
 866         f = c.open('sailboats', 'wb')
 867         data = self.sailboat.encode('utf-8')
 868         f.write(data)
 869         f.write(data[:1])
 870         f.write(data[1:])
 871         f.write(b'\n')
 872         f.close()
 873         self.assertRegex(c.portable_manifest_text(), r'\+4 .*\+3 ')
 874
 875         f = c.open('sailboats', 'r')
 876         string = f.readline()
 877         self.assertEqual(string, self.sailboat+self.sailboat+'\n')
 878         f.close()
 879
 880     def test_write_snowman_across_block_boundary(self):
 881         c = Collection()
 882         f = c.open('snowmany', 'w')
 883         data = self.snowman
 884         f.write(data+data+'\n'+data+'\n')
 885         f.close()
 886         self.assertRegex(c.portable_manifest_text(), r'\+4 .*\+4 .*\+3 ')
 887
 888         f = c.open('snowmany', 'r')
 889         self.assertEqual(f.readline(), self.snowman+self.snowman+'\n')
 890         self.assertEqual(f.readline(), self.snowman+'\n')
 891         f.close()
 892
 893
 894 class NewCollectionTestCase(unittest.TestCase, CollectionTestMixin):
 895
 896     def test_replication_desired_kept_on_load(self):
 897         m = '. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt 0:10:count2.txt\n'
 898         c1 = Collection(m, replication_desired=1)
 899         c1.save_new()
 900         loc = c1.manifest_locator()
 901         c2 = Collection(loc)
 902         self.assertEqual(c1.manifest_text(strip=True), c2.manifest_text(strip=True))
 903         self.assertEqual(c1.replication_desired, c2.replication_desired)
 904
 905     def test_replication_desired_not_loaded_if_provided(self):
 906         m = '. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt 0:10:count2.txt\n'
 907         c1 = Collection(m, replication_desired=1)
 908         c1.save_new()
 909         loc = c1.manifest_locator()
 910         c2 = Collection(loc, replication_desired=2)
 911         self.assertEqual(c1.manifest_text(strip=True), c2.manifest_text(strip=True))
 912         self.assertNotEqual(c1.replication_desired, c2.replication_desired)
 913
 914     def test_storage_classes_desired_kept_on_load(self):
 915         m = '. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt 0:10:count2.txt\n'
 916         c1 = Collection(m, storage_classes_desired=['archival'])
 917         c1.save_new()
 918         loc = c1.manifest_locator()
 919         c2 = Collection(loc)
 920         self.assertEqual(c1.manifest_text(strip=True), c2.manifest_text(strip=True))
 921         self.assertEqual(c1.storage_classes_desired(), c2.storage_classes_desired())
 922
 923     def test_storage_classes_change_after_save(self):
 924         m = '. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt 0:10:count2.txt\n'
 925         c1 = Collection(m, storage_classes_desired=['archival'])
 926         c1.save_new()
 927         loc = c1.manifest_locator()
 928         c2 = Collection(loc)
 929         self.assertEqual(['archival'], c2.storage_classes_desired())
 930         c2.save(storage_classes=['highIO'])
 931         self.assertEqual(['highIO'], c2.storage_classes_desired())
 932         c3 = Collection(loc)
 933         self.assertEqual(c1.manifest_text(strip=True), c3.manifest_text(strip=True))
 934         self.assertEqual(['highIO'], c3.storage_classes_desired())
 935
 936     def test_storage_classes_desired_not_loaded_if_provided(self):
 937         m = '. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt 0:10:count2.txt\n'
 938         c1 = Collection(m, storage_classes_desired=['archival'])
 939         c1.save_new()
 940         loc = c1.manifest_locator()
 941         c2 = Collection(loc, storage_classes_desired=['default'])
 942         self.assertEqual(c1.manifest_text(strip=True), c2.manifest_text(strip=True))
 943         self.assertNotEqual(c1.storage_classes_desired(), c2.storage_classes_desired())
 944
 945     def test_init_manifest(self):
 946         m1 = """. 5348b82a029fd9e971a811ce1f71360b+43 0:43:md5sum.txt
 947 . 085c37f02916da1cad16f93c54d899b7+41 0:41:md5sum.txt
 948 . 8b22da26f9f433dea0a10e5ec66d73ba+43 0:43:md5sum.txt
 949 """
 950         self.assertEqual(m1, CollectionReader(m1).manifest_text(normalize=False))
 951         self.assertEqual(". 5348b82a029fd9e971a811ce1f71360b+43 085c37f02916da1cad16f93c54d899b7+41 8b22da26f9f433dea0a10e5ec66d73ba+43 0:127:md5sum.txt\n", CollectionReader(m1).manifest_text(normalize=True))
 952
 953     def test_init_manifest_with_collision(self):
 954         m1 = """. 5348b82a029fd9e971a811ce1f71360b+43 0:43:md5sum.txt
 955 ./md5sum.txt 085c37f02916da1cad16f93c54d899b7+41 0:41:md5sum.txt
 956 """
 957         with self.assertRaises(arvados.errors.ArgumentError):
 958             self.assertEqual(m1, CollectionReader(m1))
 959
 960     def test_init_manifest_with_error(self):
 961         m1 = """. 0:43:md5sum.txt"""
 962         with self.assertRaises(arvados.errors.ArgumentError):
 963             self.assertEqual(m1, CollectionReader(m1))
 964
 965     def test_remove(self):
 966         c = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt 0:10:count2.txt\n')
 967         self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt 0:10:count2.txt\n", c.portable_manifest_text())
 968         self.assertIn("count1.txt", c)
 969         c.remove("count1.txt")
 970         self.assertNotIn("count1.txt", c)
 971         self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count2.txt\n", c.portable_manifest_text())
 972         with self.assertRaises(arvados.errors.ArgumentError):
 973             c.remove("")
 974
 975     def test_remove_recursive(self):
 976         c = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:a/b/c/d/efg.txt 0:10:xyz.txt\n')
 977         self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:xyz.txt\n./a/b/c/d 781e5e245d69b566979b86e28d23f2c7+10 0:10:efg.txt\n", c.portable_manifest_text())
 978         self.assertIn("a", c)
 979         self.assertEqual(1, len(c["a"].keys()))
 980         # cannot remove non-empty directory with default recursive=False
 981         with self.assertRaises(OSError):
 982             c.remove("a/b")
 983         with self.assertRaises(OSError):
 984             c.remove("a/b/c/d")
 985         c.remove("a/b", recursive=True)
 986         self.assertEqual(0, len(c["a"].keys()))
 987         self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:xyz.txt\n./a d41d8cd98f00b204e9800998ecf8427e+0 0:0:\\056\n", c.portable_manifest_text())
 988
 989     def test_find(self):
 990         c = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt 0:10:count2.txt\n')
 991         self.assertIs(c.find("."), c)
 992         self.assertIs(c.find("./count1.txt"), c["count1.txt"])
 993         self.assertIs(c.find("count1.txt"), c["count1.txt"])
 994         with self.assertRaises(IOError):
 995             c.find("/.")
 996         with self.assertRaises(arvados.errors.ArgumentError):
 997             c.find("")
 998         self.assertIs(c.find("./nonexistant.txt"), None)
 999         self.assertIs(c.find("./nonexistantsubdir/nonexistant.txt"), None)
1000
1001     def test_escaped_paths_dont_get_unescaped_on_manifest(self):
1002         # Dir & file names are literally '\056' (escaped form: \134056)
1003         manifest = './\\134056\\040Test d41d8cd98f00b204e9800998ecf8427e+0 0:0:\\134056\n'
1004         c = Collection(manifest)
1005         self.assertEqual(c.portable_manifest_text(), manifest)
1006
1007     def test_other_special_chars_on_file_token(self):
1008         cases = [
1009             ('\\000', '\0'),
1010             ('\\011', '\t'),
1011             ('\\012', '\n'),
1012             ('\\072', ':'),
1013             ('\\134400', '\\400'),
1014         ]
1015         for encoded, decoded in cases:
1016             manifest = '. d41d8cd98f00b204e9800998ecf8427e+0 0:0:some%sfile.txt\n' % encoded
1017             c = Collection(manifest)
1018             self.assertEqual(c.portable_manifest_text(), manifest)
1019             self.assertIn('some%sfile.txt' % decoded, c.keys())
1020
1021     def test_escaped_paths_do_get_unescaped_on_listing(self):
1022         # Dir & file names are literally '\056' (escaped form: \134056)
1023         manifest = './\\134056\\040Test d41d8cd98f00b204e9800998ecf8427e+0 0:0:\\134056\n'
1024         c = Collection(manifest)
1025         self.assertIn('\\056 Test', c.keys())
1026         self.assertIn('\\056', c['\\056 Test'].keys())
1027
1028     def test_make_empty_dir_with_escaped_chars(self):
1029         c = Collection()
1030         c.mkdirs('./Empty\\056Dir')
1031         self.assertEqual(c.portable_manifest_text(),
1032                          './Empty\\134056Dir d41d8cd98f00b204e9800998ecf8427e+0 0:0:\\056\n')
1033
1034     def test_make_empty_dir_with_spaces(self):
1035         c = Collection()
1036         c.mkdirs('./foo bar/baz waz')
1037         self.assertEqual(c.portable_manifest_text(),
1038                          './foo\\040bar/baz\\040waz d41d8cd98f00b204e9800998ecf8427e+0 0:0:\\056\n')
1039
1040     def test_remove_in_subdir(self):
1041         c = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n./foo 781e5e245d69b566979b86e28d23f2c7+10 0:10:count2.txt\n')
1042         c.remove("foo/count2.txt")
1043         self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n./foo d41d8cd98f00b204e9800998ecf8427e+0 0:0:\\056\n", c.portable_manifest_text())
1044
1045     def test_remove_empty_subdir(self):
1046         c = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n./foo 781e5e245d69b566979b86e28d23f2c7+10 0:10:count2.txt\n')
1047         c.remove("foo/count2.txt")
1048         c.remove("foo")
1049         self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n", c.portable_manifest_text())
1050
1051     def test_remove_nonempty_subdir(self):
1052         c = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n./foo 781e5e245d69b566979b86e28d23f2c7+10 0:10:count2.txt\n')
1053         with self.assertRaises(IOError):
1054             c.remove("foo")
1055         c.remove("foo", recursive=True)
1056         self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n", c.portable_manifest_text())
1057
1058     def test_copy_to_file_in_dir(self):
1059         c = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n')
1060         c.copy("count1.txt", "foo/count2.txt")
1061         self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n./foo 781e5e245d69b566979b86e28d23f2c7+10 0:10:count2.txt\n", c.portable_manifest_text())
1062
1063     def test_copy_file(self):
1064         c = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n')
1065         c.copy("count1.txt", "count2.txt")
1066         self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt 0:10:count2.txt\n", c.portable_manifest_text())
1067
1068     def test_copy_to_existing_dir(self):
1069         c = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n./foo 781e5e245d69b566979b86e28d23f2c7+10 0:10:count2.txt\n')
1070         c.copy("count1.txt", "foo")
1071         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())
1072
1073     def test_copy_to_new_dir(self):
1074         c = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n')
1075         c.copy("count1.txt", "foo/")
1076         self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n./foo 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n", c.portable_manifest_text())
1077
1078     def test_rename_file(self):
1079         c = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n')
1080         c.rename("count1.txt", "count2.txt")
1081         self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count2.txt\n", c.manifest_text())
1082
1083     def test_move_file_to_dir(self):
1084         c = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n')
1085         c.mkdirs("foo")
1086         c.rename("count1.txt", "foo/count2.txt")
1087         self.assertEqual("./foo 781e5e245d69b566979b86e28d23f2c7+10 0:10:count2.txt\n", c.manifest_text())
1088
1089     def test_move_file_to_other(self):
1090         c1 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n')
1091         c2 = Collection()
1092         c2.rename("count1.txt", "count2.txt", source_collection=c1)
1093         self.assertEqual("", c1.manifest_text())
1094         self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count2.txt\n", c2.manifest_text())
1095
1096     def test_clone(self):
1097         c = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n./foo 781e5e245d69b566979b86e28d23f2c7+10 0:10:count2.txt\n')
1098         cl = c.clone()
1099         self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n./foo 781e5e245d69b566979b86e28d23f2c7+10 0:10:count2.txt\n", cl.portable_manifest_text())
1100
1101     def test_diff_del_add(self):
1102         c1 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n')
1103         c2 = Collection('. 5348b82a029fd9e971a811ce1f71360b+43 0:10:count2.txt\n')
1104         d = c2.diff(c1)
1105         self.assertEqual(sorted(d), [
1106             ('add', './count1.txt', c1["count1.txt"]),
1107             ('del', './count2.txt', c2["count2.txt"]),
1108         ])
1109         d = c1.diff(c2)
1110         self.assertEqual(sorted(d), [
1111             ('add', './count2.txt', c2["count2.txt"]),
1112             ('del', './count1.txt', c1["count1.txt"]),
1113         ])
1114         self.assertNotEqual(c1.portable_manifest_text(), c2.portable_manifest_text())
1115         c1.apply(d)
1116         self.assertEqual(c1.portable_manifest_text(), c2.portable_manifest_text())
1117
1118     def test_diff_same(self):
1119         c1 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n')
1120         c2 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n')
1121         d = c2.diff(c1)
1122         self.assertEqual(d, [('tok', './count1.txt', c2["count1.txt"], c1["count1.txt"])])
1123         d = c1.diff(c2)
1124         self.assertEqual(d, [('tok', './count1.txt', c2["count1.txt"], c1["count1.txt"])])
1125
1126         self.assertEqual(c1.portable_manifest_text(), c2.portable_manifest_text())
1127         c1.apply(d)
1128         self.assertEqual(c1.portable_manifest_text(), c2.portable_manifest_text())
1129
1130     def test_diff_mod(self):
1131         c1 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n')
1132         c2 = Collection('. 5348b82a029fd9e971a811ce1f71360b+43 0:10:count1.txt\n')
1133         d = c2.diff(c1)
1134         self.assertEqual(d, [('mod', './count1.txt', c2["count1.txt"], c1["count1.txt"])])
1135         d = c1.diff(c2)
1136         self.assertEqual(d, [('mod', './count1.txt', c1["count1.txt"], c2["count1.txt"])])
1137
1138         self.assertNotEqual(c1.portable_manifest_text(), c2.portable_manifest_text())
1139         c1.apply(d)
1140         self.assertEqual(c1.portable_manifest_text(), c2.portable_manifest_text())
1141
1142     def test_diff_add(self):
1143         c1 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n')
1144         c2 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 5348b82a029fd9e971a811ce1f71360b+43 0:10:count1.txt 10:20:count2.txt\n')
1145         d = c2.diff(c1)
1146         self.assertEqual(sorted(d), [
1147             ('del', './count2.txt', c2["count2.txt"]),
1148             ('tok', './count1.txt', c2["count1.txt"], c1["count1.txt"]),
1149         ])
1150         d = c1.diff(c2)
1151         self.assertEqual(sorted(d), [
1152             ('add', './count2.txt', c2["count2.txt"]),
1153             ('tok', './count1.txt', c2["count1.txt"], c1["count1.txt"]),
1154         ])
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_add_in_subcollection(self):
1161         c1 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n')
1162         c2 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n./foo 5348b82a029fd9e971a811ce1f71360b+43 0:10:count2.txt\n')
1163         d = c2.diff(c1)
1164         self.assertEqual(sorted(d), [
1165             ('del', './foo', c2["foo"]),
1166             ('tok', './count1.txt', c2["count1.txt"], c1["count1.txt"]),
1167         ])
1168         d = c1.diff(c2)
1169         self.assertEqual(sorted(d), [
1170             ('add', './foo', c2["foo"]),
1171             ('tok', './count1.txt', c2["count1.txt"], c1["count1.txt"]),
1172         ])
1173         self.assertNotEqual(c1.portable_manifest_text(), c2.portable_manifest_text())
1174         c1.apply(d)
1175         self.assertEqual(c1.portable_manifest_text(), c2.portable_manifest_text())
1176
1177     def test_diff_del_add_in_subcollection(self):
1178         c1 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n./foo 5348b82a029fd9e971a811ce1f71360b+43 0:10:count2.txt\n')
1179         c2 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n./foo 5348b82a029fd9e971a811ce1f71360b+43 0:3:count3.txt\n')
1180         d = c2.diff(c1)
1181         self.assertEqual(sorted(d), [
1182             ('add', './foo/count2.txt', c1.find("foo/count2.txt")),
1183             ('del', './foo/count3.txt', c2.find("foo/count3.txt")),
1184             ('tok', './count1.txt', c2["count1.txt"], c1["count1.txt"]),
1185         ])
1186         d = c1.diff(c2)
1187         self.assertEqual(sorted(d), [
1188             ('add', './foo/count3.txt', c2.find("foo/count3.txt")),
1189             ('del', './foo/count2.txt', c1.find("foo/count2.txt")),
1190             ('tok', './count1.txt', c2["count1.txt"], c1["count1.txt"]),
1191         ])
1192
1193         self.assertNotEqual(c1.portable_manifest_text(), c2.portable_manifest_text())
1194         c1.apply(d)
1195         self.assertEqual(c1.portable_manifest_text(), c2.portable_manifest_text())
1196
1197     def test_diff_mod_in_subcollection(self):
1198         c1 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n./foo 5348b82a029fd9e971a811ce1f71360b+43 0:10:count2.txt\n')
1199         c2 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt 0:3:foo\n')
1200         d = c2.diff(c1)
1201         self.assertEqual(sorted(d), [
1202             ('mod', './foo', c2["foo"], c1["foo"]),
1203             ('tok', './count1.txt', c2["count1.txt"], c1["count1.txt"]),
1204         ])
1205         d = c1.diff(c2)
1206         self.assertEqual(sorted(d), [
1207             ('mod', './foo', c1["foo"], c2["foo"]),
1208             ('tok', './count1.txt', c2["count1.txt"], c1["count1.txt"]),
1209         ])
1210
1211         self.assertNotEqual(c1.portable_manifest_text(), c2.portable_manifest_text())
1212         c1.apply(d)
1213         self.assertEqual(c1.portable_manifest_text(), c2.portable_manifest_text())
1214
1215     def test_conflict_keep_local_change(self):
1216         c1 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n')
1217         c2 = Collection('. 5348b82a029fd9e971a811ce1f71360b+43 0:10:count2.txt\n')
1218         d = c1.diff(c2)
1219         self.assertEqual(sorted(d), [
1220             ('add', './count2.txt', c2["count2.txt"]),
1221             ('del', './count1.txt', c1["count1.txt"]),
1222         ])
1223         f = c1.open("count1.txt", "wb")
1224         f.write(b"zzzzz")
1225
1226         # c1 changed, so it should not be deleted.
1227         c1.apply(d)
1228         self.assertEqual(c1.portable_manifest_text(), ". 95ebc3c7b3b9f1d2c40fec14415d3cb8+5 5348b82a029fd9e971a811ce1f71360b+43 0:5:count1.txt 5:10:count2.txt\n")
1229
1230     def test_conflict_mod(self):
1231         c1 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt')
1232         c2 = Collection('. 5348b82a029fd9e971a811ce1f71360b+43 0:10:count1.txt')
1233         d = c1.diff(c2)
1234         self.assertEqual(d, [('mod', './count1.txt', c1["count1.txt"], c2["count1.txt"])])
1235         f = c1.open("count1.txt", "wb")
1236         f.write(b"zzzzz")
1237
1238         # c1 changed, so c2 mod 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~$")
1243
1244     def test_conflict_add(self):
1245         c1 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count2.txt\n')
1246         c2 = Collection('. 5348b82a029fd9e971a811ce1f71360b+43 0:10:count1.txt\n')
1247         d = c1.diff(c2)
1248         self.assertEqual(sorted(d), [
1249             ('add', './count1.txt', c2["count1.txt"]),
1250             ('del', './count2.txt', c1["count2.txt"]),
1251         ])
1252         f = c1.open("count1.txt", "wb")
1253         f.write(b"zzzzz")
1254
1255         # c1 added count1.txt, so c2 add will go to a conflict file
1256         c1.apply(d)
1257         self.assertRegex(
1258             c1.portable_manifest_text(),
1259             r"\. 95ebc3c7b3b9f1d2c40fec14415d3cb8\+5 5348b82a029fd9e971a811ce1f71360b\+43 0:5:count1\.txt 5:10:count1\.txt~\d\d\d\d\d\d\d\d-\d\d\d\d\d\d~conflict~$")
1260
1261     def test_conflict_del(self):
1262         c1 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt')
1263         c2 = Collection('. 5348b82a029fd9e971a811ce1f71360b+43 0:10:count1.txt')
1264         d = c1.diff(c2)
1265         self.assertEqual(d, [('mod', './count1.txt', c1["count1.txt"], c2["count1.txt"])])
1266         c1.remove("count1.txt")
1267
1268         # c1 deleted, so c2 mod will go to a conflict file
1269         c1.apply(d)
1270         self.assertRegex(
1271             c1.portable_manifest_text(),
1272             r"\. 5348b82a029fd9e971a811ce1f71360b\+43 0:10:count1\.txt~\d\d\d\d\d\d\d\d-\d\d\d\d\d\d~conflict~$")
1273
1274     def test_notify(self):
1275         c1 = Collection()
1276         events = []
1277         c1.subscribe(lambda event, collection, name, item: events.append((event, collection, name, item)))
1278         f = c1.open("foo.txt", "wb")
1279         self.assertEqual(events[0], (arvados.collection.ADD, c1, "foo.txt", f.arvadosfile))
1280
1281     def test_open_w(self):
1282         c1 = Collection(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n")
1283         self.assertEqual(c1["count1.txt"].size(), 10)
1284         c1.open("count1.txt", "wb").close()
1285         self.assertEqual(c1["count1.txt"].size(), 0)
1286
1287
1288 class NewCollectionTestCaseWithServersAndTokens(run_test_server.TestCaseWithServers):
1289     MAIN_SERVER = {}
1290     KEEP_SERVER = {}
1291     local_locator_re = r"[0-9a-f]{32}\+\d+\+A[a-f0-9]{40}@[a-f0-9]{8}"
1292     remote_locator_re = r"[0-9a-f]{32}\+\d+\+R[a-z]{5}-[a-f0-9]{40}@[a-f0-9]{8}"
1293
1294     def setUp(self):
1295         self.keep_put = getattr(arvados.keep.KeepClient, 'put')
1296
1297     @mock.patch('arvados.keep.KeepClient.put', autospec=True)
1298     def test_storage_classes_desired(self, put_mock):
1299         put_mock.side_effect = self.keep_put
1300         c = Collection(storage_classes_desired=['default'])
1301         with c.open("file.txt", 'wb') as f:
1302             f.write('content')
1303         c.save_new()
1304         _, kwargs = put_mock.call_args
1305         self.assertEqual(['default'], kwargs['classes'])
1306
1307     @mock.patch('arvados.keep.KeepClient.put', autospec=True)
1308     def test_repacked_block_submission_get_permission_token(self, mocked_put):
1309         '''
1310         Make sure that those blocks that are committed after repacking small ones,
1311         get their permission tokens assigned on the collection manifest.
1312         '''
1313         def wrapped_keep_put(*args, **kwargs):
1314             # Simulate slow put operations
1315             time.sleep(1)
1316             return self.keep_put(*args, **kwargs)
1317
1318         mocked_put.side_effect = wrapped_keep_put
1319         c = Collection()
1320         # Write 70 files ~1MiB each so we force to produce 1 big block by repacking
1321         # small ones before finishing the upload.
1322         for i in range(70):
1323             f = c.open("file_{}.txt".format(i), 'wb')
1324             f.write(random.choice('abcdefghijklmnopqrstuvwxyz') * (2**20+i))
1325             f.close(flush=False)
1326         # We should get 2 blocks with their tokens
1327         self.assertEqual(len(re.findall(self.local_locator_re, c.manifest_text())), 2)
1328
1329     @mock.patch('arvados.keep.KeepClient.refresh_signature')
1330     def test_copy_remote_blocks_on_save_new(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         c = Collection(". " + remote_block_loc + " 0:3:foofile.txt\n")
1335         self.assertEqual(
1336             len(re.findall(self.remote_locator_re, c.manifest_text())), 1)
1337         self.assertEqual(
1338             len(re.findall(self.local_locator_re, c.manifest_text())), 0)
1339         c.save_new()
1340         rs_mock.assert_called()
1341         self.assertEqual(
1342             len(re.findall(self.remote_locator_re, c.manifest_text())), 0)
1343         self.assertEqual(
1344             len(re.findall(self.local_locator_re, c.manifest_text())), 1)
1345
1346     @mock.patch('arvados.keep.KeepClient.refresh_signature')
1347     def test_copy_remote_blocks_on_save(self, rs_mock):
1348         remote_block_loc = "acbd18db4cc2f85cedef654fccc4a4d8+3+Remote-" + "a" * 40 + "@abcdef01"
1349         local_block_loc = "acbd18db4cc2f85cedef654fccc4a4d8+3+A" + "b" * 40 + "@abcdef01"
1350         rs_mock.return_value = local_block_loc
1351         # Remote collection
1352         remote_c = Collection(". " + remote_block_loc + " 0:3:foofile.txt\n")
1353         self.assertEqual(
1354             len(re.findall(self.remote_locator_re, remote_c.manifest_text())), 1)
1355         # Local collection
1356         local_c = Collection()
1357         with local_c.open('barfile.txt', 'wb') as f:
1358             f.write('bar')
1359         local_c.save_new()
1360         self.assertEqual(
1361             len(re.findall(self.local_locator_re, local_c.manifest_text())), 1)
1362         self.assertEqual(
1363             len(re.findall(self.remote_locator_re, local_c.manifest_text())), 0)
1364         # Copy remote file to local collection
1365         local_c.copy('./foofile.txt', './copied/foofile.txt', remote_c)
1366         self.assertEqual(
1367             len(re.findall(self.local_locator_re, local_c.manifest_text())), 1)
1368         self.assertEqual(
1369             len(re.findall(self.remote_locator_re, local_c.manifest_text())), 1)
1370         # Save local collection: remote block should be copied
1371         local_c.save()
1372         rs_mock.assert_called()
1373         self.assertEqual(
1374             len(re.findall(self.local_locator_re, local_c.manifest_text())), 2)
1375         self.assertEqual(
1376             len(re.findall(self.remote_locator_re, local_c.manifest_text())), 0)
1377
1378
1379 class NewCollectionTestCaseWithServers(run_test_server.TestCaseWithServers):
1380     def test_preserve_version_on_save(self):
1381         c = Collection()
1382         c.save_new(preserve_version=True)
1383         coll_record = arvados.api().collections().get(uuid=c.manifest_locator()).execute()
1384         self.assertEqual(coll_record['version'], 1)
1385         self.assertEqual(coll_record['preserve_version'], True)
1386         with c.open("foo.txt", "wb") as foo:
1387             foo.write(b"foo")
1388         c.save(preserve_version=True)
1389         coll_record = arvados.api().collections().get(uuid=c.manifest_locator()).execute()
1390         self.assertEqual(coll_record['version'], 2)
1391         self.assertEqual(coll_record['preserve_version'], True)
1392         with c.open("bar.txt", "wb") as foo:
1393             foo.write(b"bar")
1394         c.save(preserve_version=False)
1395         coll_record = arvados.api().collections().get(uuid=c.manifest_locator()).execute()
1396         self.assertEqual(coll_record['version'], 3)
1397         self.assertEqual(coll_record['preserve_version'], False)
1398
1399     def test_get_manifest_text_only_committed(self):
1400         c = Collection()
1401         with c.open("count.txt", "wb") as f:
1402             # One file committed
1403             with c.open("foo.txt", "wb") as foo:
1404                 foo.write(b"foo")
1405                 foo.flush() # Force block commit
1406             f.write(b"0123456789")
1407             # Other file not committed. Block not written to keep yet.
1408             self.assertEqual(
1409                 c._get_manifest_text(".",
1410                                      strip=False,
1411                                      normalize=False,
1412                                      only_committed=True),
1413                 '. acbd18db4cc2f85cedef654fccc4a4d8+3 0:0:count.txt 0:3:foo.txt\n')
1414             # And now with the file closed...
1415             f.flush() # Force block commit
1416         self.assertEqual(
1417             c._get_manifest_text(".",
1418                                  strip=False,
1419                                  normalize=False,
1420                                  only_committed=True),
1421             ". 781e5e245d69b566979b86e28d23f2c7+10 acbd18db4cc2f85cedef654fccc4a4d8+3 0:10:count.txt 10:3:foo.txt\n")
1422
1423     def test_only_small_blocks_are_packed_together(self):
1424         c = Collection()
1425         # Write a couple of small files,
1426         f = c.open("count.txt", "wb")
1427         f.write(b"0123456789")
1428         f.close(flush=False)
1429         foo = c.open("foo.txt", "wb")
1430         foo.write(b"foo")
1431         foo.close(flush=False)
1432         # Then, write a big file, it shouldn't be packed with the ones above
1433         big = c.open("bigfile.txt", "wb")
1434         big.write(b"x" * 1024 * 1024 * 33) # 33 MB > KEEP_BLOCK_SIZE/2
1435         big.close(flush=False)
1436         self.assertEqual(
1437             c.manifest_text("."),
1438             '. 2d303c138c118af809f39319e5d507e9+34603008 a8430a058b8fbf408e1931b794dbd6fb+13 0:34603008:bigfile.txt 34603008:10:count.txt 34603018:3:foo.txt\n')
1439
1440     def test_flush_after_small_block_packing(self):
1441         c = Collection()
1442         # Write a couple of small files,
1443         f = c.open("count.txt", "wb")
1444         f.write(b"0123456789")
1445         f.close(flush=False)
1446         foo = c.open("foo.txt", "wb")
1447         foo.write(b"foo")
1448         foo.close(flush=False)
1449
1450         self.assertEqual(
1451             c.manifest_text(),
1452             '. a8430a058b8fbf408e1931b794dbd6fb+13 0:10:count.txt 10:3:foo.txt\n')
1453
1454         f = c.open("count.txt", "rb+")
1455         f.close(flush=True)
1456
1457         self.assertEqual(
1458             c.manifest_text(),
1459             '. a8430a058b8fbf408e1931b794dbd6fb+13 0:10:count.txt 10:3:foo.txt\n')
1460
1461     def test_write_after_small_block_packing2(self):
1462         c = Collection()
1463         # Write a couple of small files,
1464         f = c.open("count.txt", "wb")
1465         f.write(b"0123456789")
1466         f.close(flush=False)
1467         foo = c.open("foo.txt", "wb")
1468         foo.write(b"foo")
1469         foo.close(flush=False)
1470
1471         self.assertEqual(
1472             c.manifest_text(),
1473             '. a8430a058b8fbf408e1931b794dbd6fb+13 0:10:count.txt 10:3:foo.txt\n')
1474
1475         f = c.open("count.txt", "rb+")
1476         f.write(b"abc")
1477         f.close(flush=False)
1478
1479         self.assertEqual(
1480             c.manifest_text(),
1481             '. 900150983cd24fb0d6963f7d28e17f72+3 a8430a058b8fbf408e1931b794dbd6fb+13 0:3:count.txt 6:7:count.txt 13:3:foo.txt\n')
1482
1483
1484     def test_small_block_packing_with_overwrite(self):
1485         c = Collection()
1486         c.open("b1", "wb").close()
1487         c["b1"].writeto(0, b"b1", 0)
1488
1489         c.open("b2", "wb").close()
1490         c["b2"].writeto(0, b"b2", 0)
1491
1492         c["b1"].writeto(0, b"1b", 0)
1493
1494         self.assertEqual(c.manifest_text(), ". ed4f3f67c70b02b29c50ce1ea26666bd+4 0:2:b1 2:2:b2\n")
1495         self.assertEqual(c["b1"].manifest_text(), ". ed4f3f67c70b02b29c50ce1ea26666bd+4 0:2:b1\n")
1496         self.assertEqual(c["b2"].manifest_text(), ". ed4f3f67c70b02b29c50ce1ea26666bd+4 2:2:b2\n")
1497
1498
1499 class CollectionCreateUpdateTest(run_test_server.TestCaseWithServers):
1500     MAIN_SERVER = {}
1501     KEEP_SERVER = {}
1502
1503     def create_count_txt(self):
1504         # Create an empty collection, save it to the API server, then write a
1505         # file, but don't save it.
1506
1507         c = Collection()
1508         c.save_new("CollectionCreateUpdateTest", ensure_unique_name=True)
1509         self.assertEqual(c.portable_data_hash(), "d41d8cd98f00b204e9800998ecf8427e+0")
1510         self.assertEqual(c.api_response()["portable_data_hash"], "d41d8cd98f00b204e9800998ecf8427e+0" )
1511
1512         with c.open("count.txt", "wb") as f:
1513             f.write(b"0123456789")
1514
1515         self.assertEqual(c.portable_manifest_text(), ". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n")
1516
1517         return c
1518
1519     def test_create_and_save(self):
1520         c = self.create_count_txt()
1521         c.save(properties={'type' : 'Intermediate'},
1522                storage_classes=['archive'],
1523                trash_at=datetime.datetime(2111, 1, 1, 11, 11, 11, 111111))
1524
1525         self.assertRegex(
1526             c.manifest_text(),
1527             r"^\. 781e5e245d69b566979b86e28d23f2c7\+10\+A[a-f0-9]{40}@[a-f0-9]{8} 0:10:count\.txt$",)
1528         self.assertEqual(c.api_response()["storage_classes_desired"], ['archive'])
1529         self.assertEqual(c.api_response()["properties"], {'type' : 'Intermediate'})
1530         self.assertEqual(c.api_response()["trash_at"], '2111-01-01T11:11:11.111111000Z')
1531
1532
1533     def test_create_and_save_new(self):
1534         c = self.create_count_txt()
1535         c.save_new(properties={'type' : 'Intermediate'},
1536                    storage_classes=['archive'],
1537                    trash_at=datetime.datetime(2111, 1, 1, 11, 11, 11, 111111))
1538
1539         self.assertRegex(
1540             c.manifest_text(),
1541             r"^\. 781e5e245d69b566979b86e28d23f2c7\+10\+A[a-f0-9]{40}@[a-f0-9]{8} 0:10:count\.txt$",)
1542         self.assertEqual(c.api_response()["storage_classes_desired"], ['archive'])
1543         self.assertEqual(c.api_response()["properties"], {'type' : 'Intermediate'})
1544         self.assertEqual(c.api_response()["trash_at"], '2111-01-01T11:11:11.111111000Z')
1545
1546     def test_create_and_save_after_commiting(self):
1547         c = self.create_count_txt()
1548         c.save(properties={'type' : 'Intermediate'},
1549                storage_classes=['hot'],
1550                trash_at=datetime.datetime(2111, 1, 1, 11, 11, 11, 111111))
1551         c.save(properties={'type' : 'Output'},
1552                storage_classes=['cold'],
1553                trash_at=datetime.datetime(2222, 2, 2, 22, 22, 22, 222222))
1554
1555         self.assertEqual(c.api_response()["storage_classes_desired"], ['cold'])
1556         self.assertEqual(c.api_response()["properties"], {'type' : 'Output'})
1557         self.assertEqual(c.api_response()["trash_at"], '2222-02-02T22:22:22.222222000Z')
1558
1559     def test_create_diff_apply(self):
1560         c1 = self.create_count_txt()
1561         c1.save()
1562
1563         c2 = Collection(c1.manifest_locator())
1564         with c2.open("count.txt", "wb") as f:
1565             f.write(b"abcdefg")
1566
1567         diff = c1.diff(c2)
1568
1569         self.assertEqual(diff[0], (arvados.collection.MOD, u'./count.txt', c1["count.txt"], c2["count.txt"]))
1570
1571         c1.apply(diff)
1572         self.assertEqual(c1.portable_data_hash(), c2.portable_data_hash())
1573
1574     def test_diff_apply_with_token(self):
1575         baseline = CollectionReader(". 781e5e245d69b566979b86e28d23f2c7+10+A715fd31f8111894f717eb1003c1b0216799dd9ec@54f5dd1a 0:10:count.txt\n")
1576         c = Collection(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n")
1577         other = CollectionReader(". 7ac66c0f148de9519b8bd264312c4d64+7+A715fd31f8111894f717eb1003c1b0216799dd9ec@54f5dd1a 0:7:count.txt\n")
1578
1579         diff = baseline.diff(other)
1580         self.assertEqual(diff, [('mod', u'./count.txt', c["count.txt"], other["count.txt"])])
1581
1582         c.apply(diff)
1583
1584         self.assertEqual(c.manifest_text(), ". 7ac66c0f148de9519b8bd264312c4d64+7+A715fd31f8111894f717eb1003c1b0216799dd9ec@54f5dd1a 0:7:count.txt\n")
1585
1586
1587     def test_create_and_update(self):
1588         c1 = self.create_count_txt()
1589         c1.save()
1590
1591         c2 = arvados.collection.Collection(c1.manifest_locator())
1592         with c2.open("count.txt", "wb") as f:
1593             f.write(b"abcdefg")
1594
1595         c2.save()
1596
1597         self.assertNotEqual(c1.portable_data_hash(), c2.portable_data_hash())
1598         c1.update()
1599         self.assertEqual(c1.portable_data_hash(), c2.portable_data_hash())
1600
1601
1602     def test_create_and_update_with_conflict(self):
1603         c1 = self.create_count_txt()
1604         c1.save()
1605
1606         with c1.open("count.txt", "wb") as f:
1607             f.write(b"XYZ")
1608
1609         c2 = arvados.collection.Collection(c1.manifest_locator())
1610         with c2.open("count.txt", "wb") as f:
1611             f.write(b"abcdefg")
1612
1613         c2.save()
1614
1615         c1.update()
1616         self.assertRegex(
1617             c1.manifest_text(),
1618             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~$")
1619
1620     def test_pdh_is_native_str(self):
1621         c1 = self.create_count_txt()
1622         pdh = c1.portable_data_hash()
1623         self.assertEqual(type(''), type(pdh))
1624
1625
1626 if __name__ == '__main__':
1627     unittest.main()