sdk/python/tests/test_collections.py

   1 # Copyright (C) The Arvados Authors. All rights reserved.
   2 #
   3 # SPDX-License-Identifier: Apache-2.0
   4
   5 from __future__ import absolute_import
   6
   7 from builtins import object
   8 import arvados
   9 import copy
  10 import mock
  11 import os
  12 import pprint
  13 import random
  14 import re
  15 import sys
  16 import tempfile
  17 import datetime
  18 import ciso8601
  19 import time
  20 import unittest
  21
  22 from . import run_test_server
  23 from arvados._ranges import Range, LocatorAndRange
  24 from arvados.collection import Collection, CollectionReader
  25 from . import arvados_testutil as tutil
  26
  27 class TestResumableWriter(arvados.ResumableCollectionWriter):
  28     KEEP_BLOCK_SIZE = 1024  # PUT to Keep every 1K.
  29
  30     def current_state(self):
  31         return self.dump_state(copy.deepcopy)
  32
  33
  34 class ArvadosCollectionsTest(run_test_server.TestCaseWithServers,
  35                              tutil.ArvadosBaseTestCase):
  36     MAIN_SERVER = {}
  37
  38     @classmethod
  39     def setUpClass(cls):
  40         super(ArvadosCollectionsTest, cls).setUpClass()
  41         # need admin privileges to make collections with unsigned blocks
  42         run_test_server.authorize_with('admin')
  43         cls.api_client = arvados.api('v1')
  44         cls.keep_client = arvados.KeepClient(api_client=cls.api_client,
  45                                              local_store=cls.local_store)
  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
 325         def get(self, locator, num_retries=0):
 326             return self.content[locator]
 327
 328     def test_stream_reader(self):
 329         keepblocks = {
 330             'aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa+10': b'abcdefghij',
 331             'bbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbb+15': b'klmnopqrstuvwxy',
 332             'cccccccccccccccccccccccccccccccc+5': b'z0123',
 333         }
 334         mk = self.MockKeep(keepblocks)
 335
 336         sr = arvados.StreamReader([".", "aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa+10", "bbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbb+15", "cccccccccccccccccccccccccccccccc+5", "0:30:foo"], mk)
 337
 338         content = b'abcdefghijklmnopqrstuvwxyz0123456789'
 339
 340         self.assertEqual(sr.readfrom(0, 30), content[0:30])
 341         self.assertEqual(sr.readfrom(2, 30), content[2:30])
 342
 343         self.assertEqual(sr.readfrom(2, 8), content[2:10])
 344         self.assertEqual(sr.readfrom(0, 10), content[0:10])
 345
 346         self.assertEqual(sr.readfrom(0, 5), content[0:5])
 347         self.assertEqual(sr.readfrom(5, 5), content[5:10])
 348         self.assertEqual(sr.readfrom(10, 5), content[10:15])
 349         self.assertEqual(sr.readfrom(15, 5), content[15:20])
 350         self.assertEqual(sr.readfrom(20, 5), content[20:25])
 351         self.assertEqual(sr.readfrom(25, 5), content[25:30])
 352         self.assertEqual(sr.readfrom(30, 5), b'')
 353
 354     def test_extract_file(self):
 355         m1 = """. 5348b82a029fd9e971a811ce1f71360b+43 0:43:md5sum.txt
 356 . 085c37f02916da1cad16f93c54d899b7+41 0:41:md6sum.txt
 357 . 8b22da26f9f433dea0a10e5ec66d73ba+43 0:43:md7sum.txt
 358 . 085c37f02916da1cad16f93c54d899b7+41 5348b82a029fd9e971a811ce1f71360b+43 8b22da26f9f433dea0a10e5ec66d73ba+43 47:80:md8sum.txt
 359 . 085c37f02916da1cad16f93c54d899b7+41 5348b82a029fd9e971a811ce1f71360b+43 8b22da26f9f433dea0a10e5ec66d73ba+43 40:80:md9sum.txt
 360 """
 361
 362         m2 = arvados.CollectionReader(m1, self.api_client).manifest_text(normalize=True)
 363
 364         self.assertEqual(m2,
 365                          ". 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")
 366         files = arvados.CollectionReader(
 367             m2, self.api_client).all_streams()[0].files()
 368
 369         self.assertEqual(files['md5sum.txt'].as_manifest(),
 370                          ". 5348b82a029fd9e971a811ce1f71360b+43 0:43:md5sum.txt\n")
 371         self.assertEqual(files['md6sum.txt'].as_manifest(),
 372                          ". 085c37f02916da1cad16f93c54d899b7+41 0:41:md6sum.txt\n")
 373         self.assertEqual(files['md7sum.txt'].as_manifest(),
 374                          ". 8b22da26f9f433dea0a10e5ec66d73ba+43 0:43:md7sum.txt\n")
 375         self.assertEqual(files['md9sum.txt'].as_manifest(),
 376                          ". 085c37f02916da1cad16f93c54d899b7+41 5348b82a029fd9e971a811ce1f71360b+43 8b22da26f9f433dea0a10e5ec66d73ba+43 40:80:md9sum.txt\n")
 377
 378     def test_write_directory_tree(self):
 379         cwriter = arvados.CollectionWriter(self.api_client)
 380         cwriter.write_directory_tree(self.build_directory_tree(
 381                 ['basefile', 'subdir/subfile']))
 382         self.assertEqual(cwriter.manifest_text(),
 383                          """. c5110c5ac93202d8e0f9e381f22bac0f+8 0:8:basefile
 384 ./subdir 1ca4dec89403084bf282ad31e6cf7972+14 0:14:subfile\n""")
 385
 386     def test_write_named_directory_tree(self):
 387         cwriter = arvados.CollectionWriter(self.api_client)
 388         cwriter.write_directory_tree(self.build_directory_tree(
 389                 ['basefile', 'subdir/subfile']), 'root')
 390         self.assertEqual(
 391             cwriter.manifest_text(),
 392             """./root c5110c5ac93202d8e0f9e381f22bac0f+8 0:8:basefile
 393 ./root/subdir 1ca4dec89403084bf282ad31e6cf7972+14 0:14:subfile\n""")
 394
 395     def test_write_directory_tree_in_one_stream(self):
 396         cwriter = arvados.CollectionWriter(self.api_client)
 397         cwriter.write_directory_tree(self.build_directory_tree(
 398                 ['basefile', 'subdir/subfile']), max_manifest_depth=0)
 399         self.assertEqual(cwriter.manifest_text(),
 400                          """. 4ace875ffdc6824a04950f06858f4465+22 0:8:basefile 8:14:subdir/subfile\n""")
 401
 402     def test_write_directory_tree_with_limited_recursion(self):
 403         cwriter = arvados.CollectionWriter(self.api_client)
 404         cwriter.write_directory_tree(
 405             self.build_directory_tree(['f1', 'd1/f2', 'd1/d2/f3']),
 406             max_manifest_depth=1)
 407         self.assertEqual(cwriter.manifest_text(),
 408                          """. bd19836ddb62c11c55ab251ccaca5645+2 0:2:f1
 409 ./d1 50170217e5b04312024aa5cd42934494+13 0:8:d2/f3 8:5:f2\n""")
 410
 411     def test_write_directory_tree_with_zero_recursion(self):
 412         cwriter = arvados.CollectionWriter(self.api_client)
 413         content = 'd1/d2/f3d1/f2f1'
 414         blockhash = tutil.str_keep_locator(content)
 415         cwriter.write_directory_tree(
 416             self.build_directory_tree(['f1', 'd1/f2', 'd1/d2/f3']),
 417             max_manifest_depth=0)
 418         self.assertEqual(
 419             cwriter.manifest_text(),
 420             ". {} 0:8:d1/d2/f3 8:5:d1/f2 13:2:f1\n".format(blockhash))
 421
 422     def test_write_one_file(self):
 423         cwriter = arvados.CollectionWriter(self.api_client)
 424         with self.make_test_file() as testfile:
 425             cwriter.write_file(testfile.name)
 426             self.assertEqual(
 427                 cwriter.manifest_text(),
 428                 ". 098f6bcd4621d373cade4e832627b4f6+4 0:4:{}\n".format(
 429                     os.path.basename(testfile.name)))
 430
 431     def test_write_named_file(self):
 432         cwriter = arvados.CollectionWriter(self.api_client)
 433         with self.make_test_file() as testfile:
 434             cwriter.write_file(testfile.name, 'foo')
 435             self.assertEqual(cwriter.manifest_text(),
 436                              ". 098f6bcd4621d373cade4e832627b4f6+4 0:4:foo\n")
 437
 438     def test_write_multiple_files(self):
 439         cwriter = arvados.CollectionWriter(self.api_client)
 440         for letter in 'ABC':
 441             with self.make_test_file(letter.encode()) as testfile:
 442                 cwriter.write_file(testfile.name, letter)
 443         self.assertEqual(
 444             cwriter.manifest_text(),
 445             ". 902fbdd2b1df0c4f70b4a5d23525e932+3 0:1:A 1:1:B 2:1:C\n")
 446
 447     def test_basic_resume(self):
 448         cwriter = TestResumableWriter()
 449         with self.make_test_file() as testfile:
 450             cwriter.write_file(testfile.name, 'test')
 451             resumed = TestResumableWriter.from_state(cwriter.current_state())
 452         self.assertEqual(cwriter.manifest_text(), resumed.manifest_text(),
 453                           "resumed CollectionWriter had different manifest")
 454
 455     def test_resume_fails_when_missing_dependency(self):
 456         cwriter = TestResumableWriter()
 457         with self.make_test_file() as testfile:
 458             cwriter.write_file(testfile.name, 'test')
 459         self.assertRaises(arvados.errors.StaleWriterStateError,
 460                           TestResumableWriter.from_state,
 461                           cwriter.current_state())
 462
 463     def test_resume_fails_when_dependency_mtime_changed(self):
 464         cwriter = TestResumableWriter()
 465         with self.make_test_file() as testfile:
 466             cwriter.write_file(testfile.name, 'test')
 467             os.utime(testfile.name, (0, 0))
 468             self.assertRaises(arvados.errors.StaleWriterStateError,
 469                               TestResumableWriter.from_state,
 470                               cwriter.current_state())
 471
 472     def test_resume_fails_when_dependency_is_nonfile(self):
 473         cwriter = TestResumableWriter()
 474         cwriter.write_file('/dev/null', 'empty')
 475         self.assertRaises(arvados.errors.StaleWriterStateError,
 476                           TestResumableWriter.from_state,
 477                           cwriter.current_state())
 478
 479     def test_resume_fails_when_dependency_size_changed(self):
 480         cwriter = TestResumableWriter()
 481         with self.make_test_file() as testfile:
 482             cwriter.write_file(testfile.name, 'test')
 483             orig_mtime = os.fstat(testfile.fileno()).st_mtime
 484             testfile.write(b'extra')
 485             testfile.flush()
 486             os.utime(testfile.name, (orig_mtime, orig_mtime))
 487             self.assertRaises(arvados.errors.StaleWriterStateError,
 488                               TestResumableWriter.from_state,
 489                               cwriter.current_state())
 490
 491     def test_resume_fails_with_expired_locator(self):
 492         cwriter = TestResumableWriter()
 493         state = cwriter.current_state()
 494         # Add an expired locator to the state.
 495         state['_current_stream_locators'].append(''.join([
 496                     'a' * 32, '+1+A', 'b' * 40, '@', '10000000']))
 497         self.assertRaises(arvados.errors.StaleWriterStateError,
 498                           TestResumableWriter.from_state, state)
 499
 500     def test_arbitrary_objects_not_resumable(self):
 501         cwriter = TestResumableWriter()
 502         with open('/dev/null') as badfile:
 503             self.assertRaises(arvados.errors.AssertionError,
 504                               cwriter.write_file, badfile)
 505
 506     def test_arbitrary_writes_not_resumable(self):
 507         cwriter = TestResumableWriter()
 508         self.assertRaises(arvados.errors.AssertionError,
 509                           cwriter.write, "badtext")
 510
 511
 512 class CollectionTestMixin(tutil.ApiClientMock):
 513     API_COLLECTIONS = run_test_server.fixture('collections')
 514     DEFAULT_COLLECTION = API_COLLECTIONS['foo_file']
 515     DEFAULT_DATA_HASH = DEFAULT_COLLECTION['portable_data_hash']
 516     DEFAULT_MANIFEST = DEFAULT_COLLECTION['manifest_text']
 517     DEFAULT_UUID = DEFAULT_COLLECTION['uuid']
 518     ALT_COLLECTION = API_COLLECTIONS['bar_file']
 519     ALT_DATA_HASH = ALT_COLLECTION['portable_data_hash']
 520     ALT_MANIFEST = ALT_COLLECTION['manifest_text']
 521
 522     def api_client_mock(self, status=200):
 523         client = super(CollectionTestMixin, self).api_client_mock()
 524         self.mock_keep_services(client, status=status, service_type='proxy', count=1)
 525         return client
 526
 527
 528 @tutil.skip_sleep
 529 class CollectionReaderTestCase(unittest.TestCase, CollectionTestMixin):
 530     def mock_get_collection(self, api_mock, code, fixturename):
 531         body = self.API_COLLECTIONS.get(fixturename)
 532         self._mock_api_call(api_mock.collections().get, code, body)
 533
 534     def api_client_mock(self, status=200):
 535         client = super(CollectionReaderTestCase, self).api_client_mock()
 536         self.mock_get_collection(client, status, 'foo_file')
 537         return client
 538
 539     def test_init_no_default_retries(self):
 540         client = self.api_client_mock(200)
 541         reader = arvados.CollectionReader(self.DEFAULT_UUID, api_client=client)
 542         reader.manifest_text()
 543         client.collections().get().execute.assert_called_with(num_retries=0)
 544
 545     def test_uuid_init_success(self):
 546         client = self.api_client_mock(200)
 547         reader = arvados.CollectionReader(self.DEFAULT_UUID, api_client=client,
 548                                           num_retries=3)
 549         self.assertEqual(self.DEFAULT_COLLECTION['manifest_text'],
 550                          reader.manifest_text())
 551         client.collections().get().execute.assert_called_with(num_retries=3)
 552
 553     def test_uuid_init_failure_raises_api_error(self):
 554         client = self.api_client_mock(500)
 555         with self.assertRaises(arvados.errors.ApiError):
 556             reader = arvados.CollectionReader(self.DEFAULT_UUID, api_client=client)
 557
 558     def test_locator_init(self):
 559         client = self.api_client_mock(200)
 560         # Ensure Keep will not return anything if asked.
 561         with tutil.mock_keep_responses(None, 404):
 562             reader = arvados.CollectionReader(self.DEFAULT_DATA_HASH,
 563                                               api_client=client)
 564             self.assertEqual(self.DEFAULT_MANIFEST, reader.manifest_text())
 565
 566     def test_init_no_fallback_to_keep(self):
 567         # Do not look up a collection UUID or PDH in Keep.
 568         for key in [self.DEFAULT_UUID, self.DEFAULT_DATA_HASH]:
 569             client = self.api_client_mock(404)
 570             with tutil.mock_keep_responses(self.DEFAULT_MANIFEST, 200):
 571                 with self.assertRaises(arvados.errors.ApiError):
 572                     reader = arvados.CollectionReader(key, api_client=client)
 573
 574     def test_init_num_retries_propagated(self):
 575         # More of an integration test...
 576         client = self.api_client_mock(200)
 577         reader = arvados.CollectionReader(self.DEFAULT_UUID, api_client=client,
 578                                           num_retries=3)
 579         with tutil.mock_keep_responses('foo', 500, 500, 200):
 580             self.assertEqual(b'foo',
 581                              b''.join(f.read(9) for f in reader.all_files()))
 582
 583     def test_read_nonnormalized_manifest_with_collection_reader(self):
 584         # client should be able to use CollectionReader on a manifest without normalizing it
 585         client = self.api_client_mock(500)
 586         nonnormal = ". acbd18db4cc2f85cedef654fccc4a4d8+3+Aabadbadbee@abeebdee 0:3:foo.txt 1:0:bar.txt 0:3:foo.txt\n"
 587         reader = arvados.CollectionReader(
 588             nonnormal,
 589             api_client=client, num_retries=0)
 590         # Ensure stripped_manifest() doesn't mangle our manifest in
 591         # any way other than stripping hints.
 592         self.assertEqual(
 593             re.sub('\+[^\d\s\+]+', '', nonnormal),
 594             reader.stripped_manifest())
 595         # Ensure stripped_manifest() didn't mutate our reader.
 596         self.assertEqual(nonnormal, reader.manifest_text())
 597         # Ensure the files appear in the order given in the manifest.
 598         self.assertEqual(
 599             [[6, '.', 'foo.txt'],
 600              [0, '.', 'bar.txt']],
 601             [[f.size(), f.stream_name(), f.name()]
 602              for f in reader.all_streams()[0].all_files()])
 603
 604     def test_read_empty_collection(self):
 605         client = self.api_client_mock(200)
 606         self.mock_get_collection(client, 200, 'empty')
 607         reader = arvados.CollectionReader('d41d8cd98f00b204e9800998ecf8427e+0',
 608                                           api_client=client)
 609         self.assertEqual('', reader.manifest_text())
 610         self.assertEqual(0, len(reader))
 611         self.assertFalse(reader)
 612
 613     def test_api_response(self):
 614         client = self.api_client_mock()
 615         reader = arvados.CollectionReader(self.DEFAULT_UUID, api_client=client)
 616         self.assertEqual(self.DEFAULT_COLLECTION, reader.api_response())
 617
 618     def check_open_file(self, coll_file, stream_name, file_name, file_size):
 619         self.assertFalse(coll_file.closed, "returned file is not open")
 620         self.assertEqual(stream_name, coll_file.stream_name())
 621         self.assertEqual(file_name, coll_file.name)
 622         self.assertEqual(file_size, coll_file.size())
 623
 624     def test_open_collection_file_one_argument(self):
 625         client = self.api_client_mock(200)
 626         reader = arvados.CollectionReader(self.DEFAULT_UUID, api_client=client)
 627         cfile = reader.open('./foo', 'rb')
 628         self.check_open_file(cfile, '.', 'foo', 3)
 629
 630     def test_open_deep_file(self):
 631         coll_name = 'collection_with_files_in_subdir'
 632         client = self.api_client_mock(200)
 633         self.mock_get_collection(client, 200, coll_name)
 634         reader = arvados.CollectionReader(
 635             self.API_COLLECTIONS[coll_name]['uuid'], api_client=client)
 636         cfile = reader.open('./subdir2/subdir3/file2_in_subdir3.txt', 'rb')
 637         self.check_open_file(cfile, './subdir2/subdir3', 'file2_in_subdir3.txt',
 638                              32)
 639
 640     def test_open_nonexistent_stream(self):
 641         client = self.api_client_mock(200)
 642         reader = arvados.CollectionReader(self.DEFAULT_UUID, api_client=client)
 643         self.assertRaises(IOError, reader.open, './nonexistent/foo')
 644
 645     def test_open_nonexistent_file(self):
 646         client = self.api_client_mock(200)
 647         reader = arvados.CollectionReader(self.DEFAULT_UUID, api_client=client)
 648         self.assertRaises(IOError, reader.open, 'nonexistent')
 649
 650
 651 @tutil.skip_sleep
 652 class CollectionWriterTestCase(unittest.TestCase, CollectionTestMixin):
 653     def mock_keep(self, body, *codes, **headers):
 654         headers.setdefault('x-keep-replicas-stored', 2)
 655         return tutil.mock_keep_responses(body, *codes, **headers)
 656
 657     def foo_writer(self, **kwargs):
 658         kwargs.setdefault('api_client', self.api_client_mock())
 659         writer = arvados.CollectionWriter(**kwargs)
 660         writer.start_new_file('foo')
 661         writer.write(b'foo')
 662         return writer
 663
 664     def test_write_whole_collection(self):
 665         writer = self.foo_writer()
 666         with self.mock_keep(self.DEFAULT_DATA_HASH, 200, 200):
 667             self.assertEqual(self.DEFAULT_DATA_HASH, writer.finish())
 668
 669     def test_write_no_default(self):
 670         writer = self.foo_writer()
 671         with self.mock_keep(None, 500):
 672             with self.assertRaises(arvados.errors.KeepWriteError):
 673                 writer.finish()
 674
 675     def test_write_insufficient_replicas_via_proxy(self):
 676         writer = self.foo_writer(replication=3)
 677         with self.mock_keep(None, 200, **{'x-keep-replicas-stored': 2}):
 678             with self.assertRaises(arvados.errors.KeepWriteError):
 679                 writer.manifest_text()
 680
 681     def test_write_insufficient_replicas_via_disks(self):
 682         client = mock.MagicMock(name='api_client')
 683         with self.mock_keep(
 684                 None, 200, 200,
 685                 **{'x-keep-replicas-stored': 1}) as keepmock:
 686             self.mock_keep_services(client, status=200, service_type='disk', count=2)
 687             writer = self.foo_writer(api_client=client, replication=3)
 688             with self.assertRaises(arvados.errors.KeepWriteError):
 689                 writer.manifest_text()
 690
 691     def test_write_three_replicas(self):
 692         client = mock.MagicMock(name='api_client')
 693         with self.mock_keep(
 694                 "", 500, 500, 500, 200, 200, 200,
 695                 **{'x-keep-replicas-stored': 1}) as keepmock:
 696             self.mock_keep_services(client, status=200, service_type='disk', count=6)
 697             writer = self.foo_writer(api_client=client, replication=3)
 698             writer.manifest_text()
 699             self.assertEqual(6, keepmock.call_count)
 700
 701     def test_write_whole_collection_through_retries(self):
 702         writer = self.foo_writer(num_retries=2)
 703         with self.mock_keep(self.DEFAULT_DATA_HASH,
 704                             500, 500, 200, 500, 500, 200):
 705             self.assertEqual(self.DEFAULT_DATA_HASH, writer.finish())
 706
 707     def test_flush_data_retries(self):
 708         writer = self.foo_writer(num_retries=2)
 709         foo_hash = self.DEFAULT_MANIFEST.split()[1]
 710         with self.mock_keep(foo_hash, 500, 200):
 711             writer.flush_data()
 712         self.assertEqual(self.DEFAULT_MANIFEST, writer.manifest_text())
 713
 714     def test_one_open(self):
 715         client = self.api_client_mock()
 716         writer = arvados.CollectionWriter(client)
 717         with writer.open('out') as out_file:
 718             self.assertEqual('.', writer.current_stream_name())
 719             self.assertEqual('out', writer.current_file_name())
 720             out_file.write(b'test data')
 721             data_loc = tutil.str_keep_locator('test data')
 722         self.assertTrue(out_file.closed, "writer file not closed after context")
 723         self.assertRaises(ValueError, out_file.write, 'extra text')
 724         with self.mock_keep(data_loc, 200) as keep_mock:
 725             self.assertEqual(". {} 0:9:out\n".format(data_loc),
 726                              writer.manifest_text())
 727
 728     def test_open_writelines(self):
 729         client = self.api_client_mock()
 730         writer = arvados.CollectionWriter(client)
 731         with writer.open('six') as out_file:
 732             out_file.writelines(['12', '34', '56'])
 733             data_loc = tutil.str_keep_locator('123456')
 734         with self.mock_keep(data_loc, 200) as keep_mock:
 735             self.assertEqual(". {} 0:6:six\n".format(data_loc),
 736                              writer.manifest_text())
 737
 738     def test_open_flush(self):
 739         client = self.api_client_mock()
 740         data_loc1 = tutil.str_keep_locator('flush1')
 741         data_loc2 = tutil.str_keep_locator('flush2')
 742         with self.mock_keep((data_loc1, 200), (data_loc2, 200)) as keep_mock:
 743             writer = arvados.CollectionWriter(client)
 744             with writer.open('flush_test') as out_file:
 745                 out_file.write(b'flush1')
 746                 out_file.flush()
 747                 out_file.write(b'flush2')
 748             self.assertEqual(". {} {} 0:12:flush_test\n".format(data_loc1,
 749                                                                 data_loc2),
 750                              writer.manifest_text())
 751
 752     def test_two_opens_same_stream(self):
 753         client = self.api_client_mock()
 754         writer = arvados.CollectionWriter(client)
 755         with writer.open('.', '1') as out_file:
 756             out_file.write(b'1st')
 757         with writer.open('.', '2') as out_file:
 758             out_file.write(b'2nd')
 759         data_loc = tutil.str_keep_locator('1st2nd')
 760         with self.mock_keep(data_loc, 200) as keep_mock:
 761             self.assertEqual(". {} 0:3:1 3:3:2\n".format(data_loc),
 762                              writer.manifest_text())
 763
 764     def test_two_opens_two_streams(self):
 765         client = self.api_client_mock()
 766         data_loc1 = tutil.str_keep_locator('file')
 767         data_loc2 = tutil.str_keep_locator('indir')
 768         with self.mock_keep((data_loc1, 200), (data_loc2, 200)) as keep_mock:
 769             writer = arvados.CollectionWriter(client)
 770             with writer.open('file') as out_file:
 771                 out_file.write(b'file')
 772             with writer.open('./dir', 'indir') as out_file:
 773                 out_file.write(b'indir')
 774             expected = ". {} 0:4:file\n./dir {} 0:5:indir\n".format(
 775                 data_loc1, data_loc2)
 776             self.assertEqual(expected, writer.manifest_text())
 777
 778     def test_dup_open_fails(self):
 779         client = self.api_client_mock()
 780         writer = arvados.CollectionWriter(client)
 781         file1 = writer.open('one')
 782         self.assertRaises(arvados.errors.AssertionError, writer.open, 'two')
 783
 784
 785 class CollectionMethods(run_test_server.TestCaseWithServers):
 786
 787     def test_keys_values_items_support_indexing(self):
 788         c = Collection()
 789         with c.open('foo', 'wb') as f:
 790             f.write(b'foo')
 791         with c.open('bar', 'wb') as f:
 792             f.write(b'bar')
 793         self.assertEqual(2, len(c.keys()))
 794         if sys.version_info < (3, 0):
 795             # keys() supports indexing only for python2 callers.
 796             fn0 = c.keys()[0]
 797             fn1 = c.keys()[1]
 798         else:
 799             fn0, fn1 = c.keys()
 800         self.assertEqual(2, len(c.values()))
 801         f0 = c.values()[0]
 802         f1 = c.values()[1]
 803         self.assertEqual(2, len(c.items()))
 804         self.assertEqual(fn0, c.items()[0][0])
 805         self.assertEqual(fn1, c.items()[1][0])
 806
 807     def test_get_properties(self):
 808         c = Collection()
 809         self.assertEqual(c.get_properties(), {})
 810         c.save_new(properties={"foo":"bar"})
 811         self.assertEqual(c.get_properties(), {"foo":"bar"})
 812
 813     def test_get_trash_at(self):
 814         c = Collection()
 815         self.assertEqual(c.get_trash_at(), None)
 816         c.save_new(trash_at=datetime.datetime(2111, 1, 1, 11, 11, 11, 111111))
 817         self.assertEqual(c.get_trash_at(), ciso8601.parse_datetime('2111-01-01T11:11:11.111111000Z'))
 818
 819
 820 class CollectionOpenModes(run_test_server.TestCaseWithServers):
 821
 822     def test_open_binary_modes(self):
 823         c = Collection()
 824         for mode in ['wb', 'wb+', 'ab', 'ab+']:
 825             with c.open('foo', mode) as f:
 826                 f.write(b'foo')
 827
 828     def test_open_invalid_modes(self):
 829         c = Collection()
 830         for mode in ['+r', 'aa', '++', 'r+b', 'beer', '', None]:
 831             with self.assertRaises(Exception):
 832                 c.open('foo', mode)
 833
 834     def test_open_text_modes(self):
 835         c = Collection()
 836         with c.open('foo', 'wb') as f:
 837             f.write('foo')
 838         for mode in ['r', 'rt', 'r+', 'rt+', 'w', 'wt', 'a', 'at']:
 839             if sys.version_info >= (3, 0):
 840                 with self.assertRaises(NotImplementedError):
 841                     c.open('foo', mode)
 842             else:
 843                 with c.open('foo', mode) as f:
 844                     if mode[0] == 'r' and '+' not in mode:
 845                         self.assertEqual('foo', f.read(3))
 846                     else:
 847                         f.write('bar')
 848                         f.seek(-3, os.SEEK_CUR)
 849                         self.assertEqual('bar', f.read(3))
 850
 851
 852 class NewCollectionTestCase(unittest.TestCase, CollectionTestMixin):
 853
 854     def test_replication_desired_kept_on_load(self):
 855         m = '. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt 0:10:count2.txt\n'
 856         c1 = Collection(m, replication_desired=1)
 857         c1.save_new()
 858         loc = c1.manifest_locator()
 859         c2 = Collection(loc)
 860         self.assertEqual(c1.manifest_text, c2.manifest_text)
 861         self.assertEqual(c1.replication_desired, c2.replication_desired)
 862
 863     def test_replication_desired_not_loaded_if_provided(self):
 864         m = '. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt 0:10:count2.txt\n'
 865         c1 = Collection(m, replication_desired=1)
 866         c1.save_new()
 867         loc = c1.manifest_locator()
 868         c2 = Collection(loc, replication_desired=2)
 869         self.assertEqual(c1.manifest_text, c2.manifest_text)
 870         self.assertNotEqual(c1.replication_desired, c2.replication_desired)
 871
 872     def test_init_manifest(self):
 873         m1 = """. 5348b82a029fd9e971a811ce1f71360b+43 0:43:md5sum.txt
 874 . 085c37f02916da1cad16f93c54d899b7+41 0:41:md5sum.txt
 875 . 8b22da26f9f433dea0a10e5ec66d73ba+43 0:43:md5sum.txt
 876 """
 877         self.assertEqual(m1, CollectionReader(m1).manifest_text(normalize=False))
 878         self.assertEqual(". 5348b82a029fd9e971a811ce1f71360b+43 085c37f02916da1cad16f93c54d899b7+41 8b22da26f9f433dea0a10e5ec66d73ba+43 0:127:md5sum.txt\n", CollectionReader(m1).manifest_text(normalize=True))
 879
 880     def test_init_manifest_with_collision(self):
 881         m1 = """. 5348b82a029fd9e971a811ce1f71360b+43 0:43:md5sum.txt
 882 ./md5sum.txt 085c37f02916da1cad16f93c54d899b7+41 0:41:md5sum.txt
 883 """
 884         with self.assertRaises(arvados.errors.ArgumentError):
 885             self.assertEqual(m1, CollectionReader(m1))
 886
 887     def test_init_manifest_with_error(self):
 888         m1 = """. 0:43:md5sum.txt"""
 889         with self.assertRaises(arvados.errors.ArgumentError):
 890             self.assertEqual(m1, CollectionReader(m1))
 891
 892     def test_remove(self):
 893         c = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt 0:10:count2.txt\n')
 894         self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt 0:10:count2.txt\n", c.portable_manifest_text())
 895         self.assertIn("count1.txt", c)
 896         c.remove("count1.txt")
 897         self.assertNotIn("count1.txt", c)
 898         self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count2.txt\n", c.portable_manifest_text())
 899         with self.assertRaises(arvados.errors.ArgumentError):
 900             c.remove("")
 901
 902     def test_find(self):
 903         c = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt 0:10:count2.txt\n')
 904         self.assertIs(c.find("."), c)
 905         self.assertIs(c.find("./count1.txt"), c["count1.txt"])
 906         self.assertIs(c.find("count1.txt"), c["count1.txt"])
 907         with self.assertRaises(IOError):
 908             c.find("/.")
 909         with self.assertRaises(arvados.errors.ArgumentError):
 910             c.find("")
 911         self.assertIs(c.find("./nonexistant.txt"), None)
 912         self.assertIs(c.find("./nonexistantsubdir/nonexistant.txt"), None)
 913
 914     def test_remove_in_subdir(self):
 915         c = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n./foo 781e5e245d69b566979b86e28d23f2c7+10 0:10:count2.txt\n')
 916         c.remove("foo/count2.txt")
 917         self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n", c.portable_manifest_text())
 918
 919     def test_remove_empty_subdir(self):
 920         c = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n./foo 781e5e245d69b566979b86e28d23f2c7+10 0:10:count2.txt\n')
 921         c.remove("foo/count2.txt")
 922         c.remove("foo")
 923         self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n", c.portable_manifest_text())
 924
 925     def test_remove_nonempty_subdir(self):
 926         c = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n./foo 781e5e245d69b566979b86e28d23f2c7+10 0:10:count2.txt\n')
 927         with self.assertRaises(IOError):
 928             c.remove("foo")
 929         c.remove("foo", recursive=True)
 930         self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n", c.portable_manifest_text())
 931
 932     def test_copy_to_file_in_dir(self):
 933         c = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n')
 934         c.copy("count1.txt", "foo/count2.txt")
 935         self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n./foo 781e5e245d69b566979b86e28d23f2c7+10 0:10:count2.txt\n", c.portable_manifest_text())
 936
 937     def test_copy_file(self):
 938         c = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n')
 939         c.copy("count1.txt", "count2.txt")
 940         self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt 0:10:count2.txt\n", c.portable_manifest_text())
 941
 942     def test_copy_to_existing_dir(self):
 943         c = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n./foo 781e5e245d69b566979b86e28d23f2c7+10 0:10:count2.txt\n')
 944         c.copy("count1.txt", "foo")
 945         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())
 946
 947     def test_copy_to_new_dir(self):
 948         c = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n')
 949         c.copy("count1.txt", "foo/")
 950         self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n./foo 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n", c.portable_manifest_text())
 951
 952     def test_rename_file(self):
 953         c = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n')
 954         c.rename("count1.txt", "count2.txt")
 955         self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count2.txt\n", c.manifest_text())
 956
 957     def test_move_file_to_dir(self):
 958         c = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n')
 959         c.mkdirs("foo")
 960         c.rename("count1.txt", "foo/count2.txt")
 961         self.assertEqual("./foo 781e5e245d69b566979b86e28d23f2c7+10 0:10:count2.txt\n", c.manifest_text())
 962
 963     def test_move_file_to_other(self):
 964         c1 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n')
 965         c2 = Collection()
 966         c2.rename("count1.txt", "count2.txt", source_collection=c1)
 967         self.assertEqual("", c1.manifest_text())
 968         self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count2.txt\n", c2.manifest_text())
 969
 970     def test_clone(self):
 971         c = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n./foo 781e5e245d69b566979b86e28d23f2c7+10 0:10:count2.txt\n')
 972         cl = c.clone()
 973         self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n./foo 781e5e245d69b566979b86e28d23f2c7+10 0:10:count2.txt\n", cl.portable_manifest_text())
 974
 975     def test_diff_del_add(self):
 976         c1 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n')
 977         c2 = Collection('. 5348b82a029fd9e971a811ce1f71360b+43 0:10:count2.txt\n')
 978         d = c2.diff(c1)
 979         self.assertEqual(sorted(d), [
 980             ('add', './count1.txt', c1["count1.txt"]),
 981             ('del', './count2.txt', c2["count2.txt"]),
 982         ])
 983         d = c1.diff(c2)
 984         self.assertEqual(sorted(d), [
 985             ('add', './count2.txt', c2["count2.txt"]),
 986             ('del', './count1.txt', c1["count1.txt"]),
 987         ])
 988         self.assertNotEqual(c1.portable_manifest_text(), c2.portable_manifest_text())
 989         c1.apply(d)
 990         self.assertEqual(c1.portable_manifest_text(), c2.portable_manifest_text())
 991
 992     def test_diff_same(self):
 993         c1 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n')
 994         c2 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n')
 995         d = c2.diff(c1)
 996         self.assertEqual(d, [('tok', './count1.txt', c2["count1.txt"], c1["count1.txt"])])
 997         d = c1.diff(c2)
 998         self.assertEqual(d, [('tok', './count1.txt', c2["count1.txt"], c1["count1.txt"])])
 999
1000         self.assertEqual(c1.portable_manifest_text(), c2.portable_manifest_text())
1001         c1.apply(d)
1002         self.assertEqual(c1.portable_manifest_text(), c2.portable_manifest_text())
1003
1004     def test_diff_mod(self):
1005         c1 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n')
1006         c2 = Collection('. 5348b82a029fd9e971a811ce1f71360b+43 0:10:count1.txt\n')
1007         d = c2.diff(c1)
1008         self.assertEqual(d, [('mod', './count1.txt', c2["count1.txt"], c1["count1.txt"])])
1009         d = c1.diff(c2)
1010         self.assertEqual(d, [('mod', './count1.txt', c1["count1.txt"], c2["count1.txt"])])
1011
1012         self.assertNotEqual(c1.portable_manifest_text(), c2.portable_manifest_text())
1013         c1.apply(d)
1014         self.assertEqual(c1.portable_manifest_text(), c2.portable_manifest_text())
1015
1016     def test_diff_add(self):
1017         c1 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n')
1018         c2 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 5348b82a029fd9e971a811ce1f71360b+43 0:10:count1.txt 10:20:count2.txt\n')
1019         d = c2.diff(c1)
1020         self.assertEqual(sorted(d), [
1021             ('del', './count2.txt', c2["count2.txt"]),
1022             ('tok', './count1.txt', c2["count1.txt"], c1["count1.txt"]),
1023         ])
1024         d = c1.diff(c2)
1025         self.assertEqual(sorted(d), [
1026             ('add', './count2.txt', c2["count2.txt"]),
1027             ('tok', './count1.txt', c2["count1.txt"], c1["count1.txt"]),
1028         ])
1029
1030         self.assertNotEqual(c1.portable_manifest_text(), c2.portable_manifest_text())
1031         c1.apply(d)
1032         self.assertEqual(c1.portable_manifest_text(), c2.portable_manifest_text())
1033
1034     def test_diff_add_in_subcollection(self):
1035         c1 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n')
1036         c2 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n./foo 5348b82a029fd9e971a811ce1f71360b+43 0:10:count2.txt\n')
1037         d = c2.diff(c1)
1038         self.assertEqual(sorted(d), [
1039             ('del', './foo', c2["foo"]),
1040             ('tok', './count1.txt', c2["count1.txt"], c1["count1.txt"]),
1041         ])
1042         d = c1.diff(c2)
1043         self.assertEqual(sorted(d), [
1044             ('add', './foo', c2["foo"]),
1045             ('tok', './count1.txt', c2["count1.txt"], c1["count1.txt"]),
1046         ])
1047         self.assertNotEqual(c1.portable_manifest_text(), c2.portable_manifest_text())
1048         c1.apply(d)
1049         self.assertEqual(c1.portable_manifest_text(), c2.portable_manifest_text())
1050
1051     def test_diff_del_add_in_subcollection(self):
1052         c1 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n./foo 5348b82a029fd9e971a811ce1f71360b+43 0:10:count2.txt\n')
1053         c2 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n./foo 5348b82a029fd9e971a811ce1f71360b+43 0:3:count3.txt\n')
1054         d = c2.diff(c1)
1055         self.assertEqual(sorted(d), [
1056             ('add', './foo/count2.txt', c1.find("foo/count2.txt")),
1057             ('del', './foo/count3.txt', c2.find("foo/count3.txt")),
1058             ('tok', './count1.txt', c2["count1.txt"], c1["count1.txt"]),
1059         ])
1060         d = c1.diff(c2)
1061         self.assertEqual(sorted(d), [
1062             ('add', './foo/count3.txt', c2.find("foo/count3.txt")),
1063             ('del', './foo/count2.txt', c1.find("foo/count2.txt")),
1064             ('tok', './count1.txt', c2["count1.txt"], c1["count1.txt"]),
1065         ])
1066
1067         self.assertNotEqual(c1.portable_manifest_text(), c2.portable_manifest_text())
1068         c1.apply(d)
1069         self.assertEqual(c1.portable_manifest_text(), c2.portable_manifest_text())
1070
1071     def test_diff_mod_in_subcollection(self):
1072         c1 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n./foo 5348b82a029fd9e971a811ce1f71360b+43 0:10:count2.txt\n')
1073         c2 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt 0:3:foo\n')
1074         d = c2.diff(c1)
1075         self.assertEqual(sorted(d), [
1076             ('mod', './foo', c2["foo"], c1["foo"]),
1077             ('tok', './count1.txt', c2["count1.txt"], c1["count1.txt"]),
1078         ])
1079         d = c1.diff(c2)
1080         self.assertEqual(sorted(d), [
1081             ('mod', './foo', c1["foo"], c2["foo"]),
1082             ('tok', './count1.txt', c2["count1.txt"], c1["count1.txt"]),
1083         ])
1084
1085         self.assertNotEqual(c1.portable_manifest_text(), c2.portable_manifest_text())
1086         c1.apply(d)
1087         self.assertEqual(c1.portable_manifest_text(), c2.portable_manifest_text())
1088
1089     def test_conflict_keep_local_change(self):
1090         c1 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n')
1091         c2 = Collection('. 5348b82a029fd9e971a811ce1f71360b+43 0:10:count2.txt\n')
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         f = c1.open("count1.txt", "wb")
1098         f.write(b"zzzzz")
1099
1100         # c1 changed, so it should not be deleted.
1101         c1.apply(d)
1102         self.assertEqual(c1.portable_manifest_text(), ". 95ebc3c7b3b9f1d2c40fec14415d3cb8+5 5348b82a029fd9e971a811ce1f71360b+43 0:5:count1.txt 5:10:count2.txt\n")
1103
1104     def test_conflict_mod(self):
1105         c1 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt')
1106         c2 = Collection('. 5348b82a029fd9e971a811ce1f71360b+43 0:10:count1.txt')
1107         d = c1.diff(c2)
1108         self.assertEqual(d, [('mod', './count1.txt', c1["count1.txt"], c2["count1.txt"])])
1109         f = c1.open("count1.txt", "wb")
1110         f.write(b"zzzzz")
1111
1112         # c1 changed, so c2 mod will go to a conflict file
1113         c1.apply(d)
1114         self.assertRegex(
1115             c1.portable_manifest_text(),
1116             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~$")
1117
1118     def test_conflict_add(self):
1119         c1 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count2.txt\n')
1120         c2 = Collection('. 5348b82a029fd9e971a811ce1f71360b+43 0:10:count1.txt\n')
1121         d = c1.diff(c2)
1122         self.assertEqual(sorted(d), [
1123             ('add', './count1.txt', c2["count1.txt"]),
1124             ('del', './count2.txt', c1["count2.txt"]),
1125         ])
1126         f = c1.open("count1.txt", "wb")
1127         f.write(b"zzzzz")
1128
1129         # c1 added count1.txt, so c2 add will go to a conflict file
1130         c1.apply(d)
1131         self.assertRegex(
1132             c1.portable_manifest_text(),
1133             r"\. 95ebc3c7b3b9f1d2c40fec14415d3cb8\+5 5348b82a029fd9e971a811ce1f71360b\+43 0:5:count1\.txt 5:10:count1\.txt~\d\d\d\d\d\d\d\d-\d\d\d\d\d\d~conflict~$")
1134
1135     def test_conflict_del(self):
1136         c1 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt')
1137         c2 = Collection('. 5348b82a029fd9e971a811ce1f71360b+43 0:10:count1.txt')
1138         d = c1.diff(c2)
1139         self.assertEqual(d, [('mod', './count1.txt', c1["count1.txt"], c2["count1.txt"])])
1140         c1.remove("count1.txt")
1141
1142         # c1 deleted, so c2 mod will go to a conflict file
1143         c1.apply(d)
1144         self.assertRegex(
1145             c1.portable_manifest_text(),
1146             r"\. 5348b82a029fd9e971a811ce1f71360b\+43 0:10:count1\.txt~\d\d\d\d\d\d\d\d-\d\d\d\d\d\d~conflict~$")
1147
1148     def test_notify(self):
1149         c1 = Collection()
1150         events = []
1151         c1.subscribe(lambda event, collection, name, item: events.append((event, collection, name, item)))
1152         f = c1.open("foo.txt", "wb")
1153         self.assertEqual(events[0], (arvados.collection.ADD, c1, "foo.txt", f.arvadosfile))
1154
1155     def test_open_w(self):
1156         c1 = Collection(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n")
1157         self.assertEqual(c1["count1.txt"].size(), 10)
1158         c1.open("count1.txt", "wb").close()
1159         self.assertEqual(c1["count1.txt"].size(), 0)
1160
1161
1162 class NewCollectionTestCaseWithServersAndTokens(run_test_server.TestCaseWithServers):
1163     MAIN_SERVER = {}
1164     KEEP_SERVER = {}
1165     local_locator_re = r"[0-9a-f]{32}\+\d+\+A[a-f0-9]{40}@[a-f0-9]{8}"
1166     remote_locator_re = r"[0-9a-f]{32}\+\d+\+R[a-z]{5}-[a-f0-9]{40}@[a-f0-9]{8}"
1167
1168     def setUp(self):
1169         self.keep_put = getattr(arvados.keep.KeepClient, 'put')
1170
1171     @mock.patch('arvados.keep.KeepClient.put', autospec=True)
1172     def test_repacked_block_submission_get_permission_token(self, mocked_put):
1173         '''
1174         Make sure that those blocks that are committed after repacking small ones,
1175         get their permission tokens assigned on the collection manifest.
1176         '''
1177         def wrapped_keep_put(*args, **kwargs):
1178             # Simulate slow put operations
1179             time.sleep(1)
1180             return self.keep_put(*args, **kwargs)
1181
1182         mocked_put.side_effect = wrapped_keep_put
1183         c = Collection()
1184         # Write 70 files ~1MiB each so we force to produce 1 big block by repacking
1185         # small ones before finishing the upload.
1186         for i in range(70):
1187             f = c.open("file_{}.txt".format(i), 'wb')
1188             f.write(random.choice('abcdefghijklmnopqrstuvwxyz') * (2**20+i))
1189             f.close(flush=False)
1190         # We should get 2 blocks with their tokens
1191         self.assertEqual(len(re.findall(self.local_locator_re, c.manifest_text())), 2)
1192
1193     @mock.patch('arvados.keep.KeepClient.refresh_signature')
1194     def test_copy_remote_blocks_on_save_new(self, rs_mock):
1195         remote_block_loc = "acbd18db4cc2f85cedef654fccc4a4d8+3+Remote-" + "a" * 40 + "@abcdef01"
1196         local_block_loc = "acbd18db4cc2f85cedef654fccc4a4d8+3+A" + "b" * 40 + "@abcdef01"
1197         rs_mock.return_value = local_block_loc
1198         c = Collection(". " + remote_block_loc + " 0:3:foofile.txt\n")
1199         self.assertEqual(
1200             len(re.findall(self.remote_locator_re, c.manifest_text())), 1)
1201         self.assertEqual(
1202             len(re.findall(self.local_locator_re, c.manifest_text())), 0)
1203         c.save_new()
1204         rs_mock.assert_called()
1205         self.assertEqual(
1206             len(re.findall(self.remote_locator_re, c.manifest_text())), 0)
1207         self.assertEqual(
1208             len(re.findall(self.local_locator_re, c.manifest_text())), 1)
1209
1210     @mock.patch('arvados.keep.KeepClient.refresh_signature')
1211     def test_copy_remote_blocks_on_save(self, rs_mock):
1212         remote_block_loc = "acbd18db4cc2f85cedef654fccc4a4d8+3+Remote-" + "a" * 40 + "@abcdef01"
1213         local_block_loc = "acbd18db4cc2f85cedef654fccc4a4d8+3+A" + "b" * 40 + "@abcdef01"
1214         rs_mock.return_value = local_block_loc
1215         # Remote collection
1216         remote_c = Collection(". " + remote_block_loc + " 0:3:foofile.txt\n")
1217         self.assertEqual(
1218             len(re.findall(self.remote_locator_re, remote_c.manifest_text())), 1)
1219         # Local collection
1220         local_c = Collection()
1221         with local_c.open('barfile.txt', 'wb') as f:
1222             f.write('bar')
1223         local_c.save_new()
1224         self.assertEqual(
1225             len(re.findall(self.local_locator_re, local_c.manifest_text())), 1)
1226         self.assertEqual(
1227             len(re.findall(self.remote_locator_re, local_c.manifest_text())), 0)
1228         # Copy remote file to local collection
1229         local_c.copy('./foofile.txt', './copied/foofile.txt', remote_c)
1230         self.assertEqual(
1231             len(re.findall(self.local_locator_re, local_c.manifest_text())), 1)
1232         self.assertEqual(
1233             len(re.findall(self.remote_locator_re, local_c.manifest_text())), 1)
1234         # Save local collection: remote block should be copied
1235         local_c.save()
1236         rs_mock.assert_called()
1237         self.assertEqual(
1238             len(re.findall(self.local_locator_re, local_c.manifest_text())), 2)
1239         self.assertEqual(
1240             len(re.findall(self.remote_locator_re, local_c.manifest_text())), 0)
1241
1242
1243 class NewCollectionTestCaseWithServers(run_test_server.TestCaseWithServers):
1244     def test_get_manifest_text_only_committed(self):
1245         c = Collection()
1246         with c.open("count.txt", "wb") as f:
1247             # One file committed
1248             with c.open("foo.txt", "wb") as foo:
1249                 foo.write(b"foo")
1250                 foo.flush() # Force block commit
1251             f.write(b"0123456789")
1252             # Other file not committed. Block not written to keep yet.
1253             self.assertEqual(
1254                 c._get_manifest_text(".",
1255                                      strip=False,
1256                                      normalize=False,
1257                                      only_committed=True),
1258                 '. acbd18db4cc2f85cedef654fccc4a4d8+3 0:0:count.txt 0:3:foo.txt\n')
1259             # And now with the file closed...
1260             f.flush() # Force block commit
1261         self.assertEqual(
1262             c._get_manifest_text(".",
1263                                  strip=False,
1264                                  normalize=False,
1265                                  only_committed=True),
1266             ". 781e5e245d69b566979b86e28d23f2c7+10 acbd18db4cc2f85cedef654fccc4a4d8+3 0:10:count.txt 10:3:foo.txt\n")
1267
1268     def test_only_small_blocks_are_packed_together(self):
1269         c = Collection()
1270         # Write a couple of small files,
1271         f = c.open("count.txt", "wb")
1272         f.write(b"0123456789")
1273         f.close(flush=False)
1274         foo = c.open("foo.txt", "wb")
1275         foo.write(b"foo")
1276         foo.close(flush=False)
1277         # Then, write a big file, it shouldn't be packed with the ones above
1278         big = c.open("bigfile.txt", "wb")
1279         big.write(b"x" * 1024 * 1024 * 33) # 33 MB > KEEP_BLOCK_SIZE/2
1280         big.close(flush=False)
1281         self.assertEqual(
1282             c.manifest_text("."),
1283             '. 2d303c138c118af809f39319e5d507e9+34603008 a8430a058b8fbf408e1931b794dbd6fb+13 0:34603008:bigfile.txt 34603008:10:count.txt 34603018:3:foo.txt\n')
1284
1285     def test_flush_after_small_block_packing(self):
1286         c = Collection()
1287         # Write a couple of small files,
1288         f = c.open("count.txt", "wb")
1289         f.write(b"0123456789")
1290         f.close(flush=False)
1291         foo = c.open("foo.txt", "wb")
1292         foo.write(b"foo")
1293         foo.close(flush=False)
1294
1295         self.assertEqual(
1296             c.manifest_text(),
1297             '. a8430a058b8fbf408e1931b794dbd6fb+13 0:10:count.txt 10:3:foo.txt\n')
1298
1299         f = c.open("count.txt", "rb+")
1300         f.close(flush=True)
1301
1302         self.assertEqual(
1303             c.manifest_text(),
1304             '. a8430a058b8fbf408e1931b794dbd6fb+13 0:10:count.txt 10:3:foo.txt\n')
1305
1306     def test_write_after_small_block_packing2(self):
1307         c = Collection()
1308         # Write a couple of small files,
1309         f = c.open("count.txt", "wb")
1310         f.write(b"0123456789")
1311         f.close(flush=False)
1312         foo = c.open("foo.txt", "wb")
1313         foo.write(b"foo")
1314         foo.close(flush=False)
1315
1316         self.assertEqual(
1317             c.manifest_text(),
1318             '. a8430a058b8fbf408e1931b794dbd6fb+13 0:10:count.txt 10:3:foo.txt\n')
1319
1320         f = c.open("count.txt", "rb+")
1321         f.write(b"abc")
1322         f.close(flush=False)
1323
1324         self.assertEqual(
1325             c.manifest_text(),
1326             '. 900150983cd24fb0d6963f7d28e17f72+3 a8430a058b8fbf408e1931b794dbd6fb+13 0:3:count.txt 6:7:count.txt 13:3:foo.txt\n')
1327
1328
1329     def test_small_block_packing_with_overwrite(self):
1330         c = Collection()
1331         c.open("b1", "wb").close()
1332         c["b1"].writeto(0, b"b1", 0)
1333
1334         c.open("b2", "wb").close()
1335         c["b2"].writeto(0, b"b2", 0)
1336
1337         c["b1"].writeto(0, b"1b", 0)
1338
1339         self.assertEqual(c.manifest_text(), ". ed4f3f67c70b02b29c50ce1ea26666bd+4 0:2:b1 2:2:b2\n")
1340         self.assertEqual(c["b1"].manifest_text(), ". ed4f3f67c70b02b29c50ce1ea26666bd+4 0:2:b1\n")
1341         self.assertEqual(c["b2"].manifest_text(), ". ed4f3f67c70b02b29c50ce1ea26666bd+4 2:2:b2\n")
1342
1343
1344 class CollectionCreateUpdateTest(run_test_server.TestCaseWithServers):
1345     MAIN_SERVER = {}
1346     KEEP_SERVER = {}
1347
1348     def create_count_txt(self):
1349         # Create an empty collection, save it to the API server, then write a
1350         # file, but don't save it.
1351
1352         c = Collection()
1353         c.save_new("CollectionCreateUpdateTest", ensure_unique_name=True)
1354         self.assertEqual(c.portable_data_hash(), "d41d8cd98f00b204e9800998ecf8427e+0")
1355         self.assertEqual(c.api_response()["portable_data_hash"], "d41d8cd98f00b204e9800998ecf8427e+0" )
1356
1357         with c.open("count.txt", "wb") as f:
1358             f.write(b"0123456789")
1359
1360         self.assertEqual(c.portable_manifest_text(), ". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n")
1361
1362         return c
1363
1364     def test_create_and_save(self):
1365         c = self.create_count_txt()
1366         c.save(properties={'type' : 'Intermediate'},
1367                storage_classes=['archive'],
1368                trash_at=datetime.datetime(2111, 1, 1, 11, 11, 11, 111111))
1369
1370         self.assertRegex(
1371             c.manifest_text(),
1372             r"^\. 781e5e245d69b566979b86e28d23f2c7\+10\+A[a-f0-9]{40}@[a-f0-9]{8} 0:10:count\.txt$",)
1373         self.assertEqual(c.api_response()["storage_classes_desired"], ['archive'])
1374         self.assertEqual(c.api_response()["properties"], {'type' : 'Intermediate'})
1375         self.assertEqual(c.api_response()["trash_at"], '2111-01-01T11:11:11.111111000Z')
1376
1377
1378     def test_create_and_save_new(self):
1379         c = self.create_count_txt()
1380         c.save_new(properties={'type' : 'Intermediate'},
1381                    storage_classes=['archive'],
1382                    trash_at=datetime.datetime(2111, 1, 1, 11, 11, 11, 111111))
1383
1384         self.assertRegex(
1385             c.manifest_text(),
1386             r"^\. 781e5e245d69b566979b86e28d23f2c7\+10\+A[a-f0-9]{40}@[a-f0-9]{8} 0:10:count\.txt$",)
1387         self.assertEqual(c.api_response()["storage_classes_desired"], ['archive'])
1388         self.assertEqual(c.api_response()["properties"], {'type' : 'Intermediate'})
1389         self.assertEqual(c.api_response()["trash_at"], '2111-01-01T11:11:11.111111000Z')
1390
1391     def test_create_and_save_after_commiting(self):
1392         c = self.create_count_txt()
1393         c.save(properties={'type' : 'Intermediate'},
1394                storage_classes=['hot'],
1395                trash_at=datetime.datetime(2111, 1, 1, 11, 11, 11, 111111))
1396         c.save(properties={'type' : 'Output'},
1397                storage_classes=['cold'],
1398                trash_at=datetime.datetime(2222, 2, 2, 22, 22, 22, 222222))
1399
1400         self.assertEqual(c.api_response()["storage_classes_desired"], ['cold'])
1401         self.assertEqual(c.api_response()["properties"], {'type' : 'Output'})
1402         self.assertEqual(c.api_response()["trash_at"], '2222-02-02T22:22:22.222222000Z')
1403
1404     def test_create_diff_apply(self):
1405         c1 = self.create_count_txt()
1406         c1.save()
1407
1408         c2 = Collection(c1.manifest_locator())
1409         with c2.open("count.txt", "wb") as f:
1410             f.write(b"abcdefg")
1411
1412         diff = c1.diff(c2)
1413
1414         self.assertEqual(diff[0], (arvados.collection.MOD, u'./count.txt', c1["count.txt"], c2["count.txt"]))
1415
1416         c1.apply(diff)
1417         self.assertEqual(c1.portable_data_hash(), c2.portable_data_hash())
1418
1419     def test_diff_apply_with_token(self):
1420         baseline = CollectionReader(". 781e5e245d69b566979b86e28d23f2c7+10+A715fd31f8111894f717eb1003c1b0216799dd9ec@54f5dd1a 0:10:count.txt\n")
1421         c = Collection(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n")
1422         other = CollectionReader(". 7ac66c0f148de9519b8bd264312c4d64+7+A715fd31f8111894f717eb1003c1b0216799dd9ec@54f5dd1a 0:7:count.txt\n")
1423
1424         diff = baseline.diff(other)
1425         self.assertEqual(diff, [('mod', u'./count.txt', c["count.txt"], other["count.txt"])])
1426
1427         c.apply(diff)
1428
1429         self.assertEqual(c.manifest_text(), ". 7ac66c0f148de9519b8bd264312c4d64+7+A715fd31f8111894f717eb1003c1b0216799dd9ec@54f5dd1a 0:7:count.txt\n")
1430
1431
1432     def test_create_and_update(self):
1433         c1 = self.create_count_txt()
1434         c1.save()
1435
1436         c2 = arvados.collection.Collection(c1.manifest_locator())
1437         with c2.open("count.txt", "wb") as f:
1438             f.write(b"abcdefg")
1439
1440         c2.save()
1441
1442         self.assertNotEqual(c1.portable_data_hash(), c2.portable_data_hash())
1443         c1.update()
1444         self.assertEqual(c1.portable_data_hash(), c2.portable_data_hash())
1445
1446
1447     def test_create_and_update_with_conflict(self):
1448         c1 = self.create_count_txt()
1449         c1.save()
1450
1451         with c1.open("count.txt", "wb") as f:
1452             f.write(b"XYZ")
1453
1454         c2 = arvados.collection.Collection(c1.manifest_locator())
1455         with c2.open("count.txt", "wb") as f:
1456             f.write(b"abcdefg")
1457
1458         c2.save()
1459
1460         c1.update()
1461         self.assertRegex(
1462             c1.manifest_text(),
1463             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~$")
1464
1465     def test_pdh_is_native_str(self):
1466         c1 = self.create_count_txt()
1467         pdh = c1.portable_data_hash()
1468         self.assertEqual(type(''), type(pdh))
1469
1470
1471 if __name__ == '__main__':
1472     unittest.main()