sdk/python/tests/test_collections.py

   1 from __future__ import absolute_import
   2 # usage example:
   3 #
   4 # ARVADOS_API_TOKEN=abc ARVADOS_API_HOST=arvados.local python -m unittest discover
   5
   6 import arvados
   7 import copy
   8 import mock
   9 import os
  10 import pprint
  11 import re
  12 import tempfile
  13 import unittest
  14
  15 from . import run_test_server
  16 from arvados._ranges import Range, LocatorAndRange
  17 from arvados.collection import Collection, CollectionReader
  18 from . import arvados_testutil as tutil
  19
  20 class TestResumableWriter(arvados.ResumableCollectionWriter):
  21     KEEP_BLOCK_SIZE = 1024  # PUT to Keep every 1K.
  22
  23     def current_state(self):
  24         return self.dump_state(copy.deepcopy)
  25
  26
  27 class ArvadosCollectionsTest(run_test_server.TestCaseWithServers,
  28                              tutil.ArvadosBaseTestCase):
  29     MAIN_SERVER = {}
  30
  31     @classmethod
  32     def setUpClass(cls):
  33         super(ArvadosCollectionsTest, cls).setUpClass()
  34         run_test_server.authorize_with('active')
  35         cls.api_client = arvados.api('v1')
  36         cls.keep_client = arvados.KeepClient(api_client=cls.api_client,
  37                                              local_store=cls.local_store)
  38
  39     def write_foo_bar_baz(self):
  40         cw = arvados.CollectionWriter(self.api_client)
  41         self.assertEqual(cw.current_stream_name(), '.',
  42                          'current_stream_name() should be "." now')
  43         cw.set_current_file_name('foo.txt')
  44         cw.write('foo')
  45         self.assertEqual(cw.current_file_name(), 'foo.txt',
  46                          'current_file_name() should be foo.txt now')
  47         cw.start_new_file('bar.txt')
  48         cw.write('bar')
  49         cw.start_new_stream('baz')
  50         cw.write('baz')
  51         cw.set_current_file_name('baz.txt')
  52         self.assertEqual(cw.manifest_text(),
  53                          ". 3858f62230ac3c915f300c664312c63f+6 0:3:foo.txt 3:3:bar.txt\n" +
  54                          "./baz 73feffa4b7f6bb68e44cf984c85f6e88+3 0:3:baz.txt\n",
  55                          "wrong manifest: got {}".format(cw.manifest_text()))
  56         cw.finish()
  57         return cw.portable_data_hash()
  58
  59     def test_keep_local_store(self):
  60         self.assertEqual(self.keep_client.put('foo'), 'acbd18db4cc2f85cedef654fccc4a4d8+3', 'wrong md5 hash from Keep.put')
  61         self.assertEqual(self.keep_client.get('acbd18db4cc2f85cedef654fccc4a4d8+3'), 'foo', 'wrong data from Keep.get')
  62
  63     def test_local_collection_writer(self):
  64         self.assertEqual(self.write_foo_bar_baz(),
  65                          '23ca013983d6239e98931cc779e68426+114',
  66                          'wrong locator hash: ' + self.write_foo_bar_baz())
  67
  68     def test_local_collection_reader(self):
  69         foobarbaz = self.write_foo_bar_baz()
  70         cr = arvados.CollectionReader(
  71             foobarbaz + '+Xzizzle', self.api_client)
  72         got = []
  73         for s in cr.all_streams():
  74             for f in s.all_files():
  75                 got += [[f.size(), f.stream_name(), f.name(), f.read(2**26)]]
  76         expected = [[3, '.', 'foo.txt', 'foo'],
  77                     [3, '.', 'bar.txt', 'bar'],
  78                     [3, './baz', 'baz.txt', 'baz']]
  79         self.assertEqual(got,
  80                          expected)
  81         stream0 = cr.all_streams()[0]
  82         self.assertEqual(stream0.readfrom(0, 0),
  83                          '',
  84                          'reading zero bytes should have returned empty string')
  85         self.assertEqual(stream0.readfrom(0, 2**26),
  86                          'foobar',
  87                          'reading entire stream failed')
  88         self.assertEqual(stream0.readfrom(2**26, 0),
  89                          '',
  90                          'reading zero bytes should have returned empty string')
  91
  92     def _test_subset(self, collection, expected):
  93         cr = arvados.CollectionReader(collection, self.api_client)
  94         for s in cr.all_streams():
  95             for ex in expected:
  96                 if ex[0] == s:
  97                     f = s.files()[ex[2]]
  98                     got = [f.size(), f.stream_name(), f.name(), "".join(f.readall(2**26))]
  99                     self.assertEqual(got,
 100                                      ex,
 101                                      'all_files|as_manifest did not preserve manifest contents: got %s expected %s' % (got, ex))
 102
 103     def test_collection_manifest_subset(self):
 104         foobarbaz = self.write_foo_bar_baz()
 105         self._test_subset(foobarbaz,
 106                           [[3, '.',     'bar.txt', 'bar'],
 107                            [3, '.',     'foo.txt', 'foo'],
 108                            [3, './baz', 'baz.txt', 'baz']])
 109         self._test_subset((". %s %s 0:3:foo.txt 3:3:bar.txt\n" %
 110                            (self.keep_client.put("foo"),
 111                             self.keep_client.put("bar"))),
 112                           [[3, '.', 'bar.txt', 'bar'],
 113                            [3, '.', 'foo.txt', 'foo']])
 114         self._test_subset((". %s %s 0:2:fo.txt 2:4:obar.txt\n" %
 115                            (self.keep_client.put("foo"),
 116                             self.keep_client.put("bar"))),
 117                           [[2, '.', 'fo.txt', 'fo'],
 118                            [4, '.', 'obar.txt', 'obar']])
 119         self._test_subset((". %s %s 0:2:fo.txt 2:0:zero.txt 2:2:ob.txt 4:2:ar.txt\n" %
 120                            (self.keep_client.put("foo"),
 121                             self.keep_client.put("bar"))),
 122                           [[2, '.', 'ar.txt', 'ar'],
 123                            [2, '.', 'fo.txt', 'fo'],
 124                            [2, '.', 'ob.txt', 'ob'],
 125                            [0, '.', 'zero.txt', '']])
 126
 127     def test_collection_empty_file(self):
 128         cw = arvados.CollectionWriter(self.api_client)
 129         cw.start_new_file('zero.txt')
 130         cw.write('')
 131
 132         self.assertEqual(cw.manifest_text(), ". d41d8cd98f00b204e9800998ecf8427e+0 0:0:zero.txt\n")
 133         self.check_manifest_file_sizes(cw.manifest_text(), [0])
 134         cw = arvados.CollectionWriter(self.api_client)
 135         cw.start_new_file('zero.txt')
 136         cw.write('')
 137         cw.start_new_file('one.txt')
 138         cw.write('1')
 139         cw.start_new_stream('foo')
 140         cw.start_new_file('zero.txt')
 141         cw.write('')
 142         self.check_manifest_file_sizes(cw.manifest_text(), [0,1,0])
 143
 144     def test_no_implicit_normalize(self):
 145         cw = arvados.CollectionWriter(self.api_client)
 146         cw.start_new_file('b')
 147         cw.write('b')
 148         cw.start_new_file('a')
 149         cw.write('')
 150         self.check_manifest_file_sizes(cw.manifest_text(), [1,0])
 151         self.check_manifest_file_sizes(
 152             arvados.CollectionReader(
 153                 cw.manifest_text()).manifest_text(normalize=True),
 154             [0,1])
 155
 156     def check_manifest_file_sizes(self, manifest_text, expect_sizes):
 157         cr = arvados.CollectionReader(manifest_text, self.api_client)
 158         got_sizes = []
 159         for f in cr.all_files():
 160             got_sizes += [f.size()]
 161         self.assertEqual(got_sizes, expect_sizes, "got wrong file sizes %s, expected %s" % (got_sizes, expect_sizes))
 162
 163     def test_normalized_collection(self):
 164         m1 = """. 5348b82a029fd9e971a811ce1f71360b+43 0:43:md5sum.txt
 165 . 085c37f02916da1cad16f93c54d899b7+41 0:41:md5sum.txt
 166 . 8b22da26f9f433dea0a10e5ec66d73ba+43 0:43:md5sum.txt
 167 """
 168         self.assertEqual(arvados.CollectionReader(m1, self.api_client).manifest_text(normalize=True),
 169                          """. 5348b82a029fd9e971a811ce1f71360b+43 085c37f02916da1cad16f93c54d899b7+41 8b22da26f9f433dea0a10e5ec66d73ba+43 0:127:md5sum.txt
 170 """)
 171
 172         m2 = """. 204e43b8a1185621ca55a94839582e6f+67108864 b9677abbac956bd3e86b1deb28dfac03+67108864 fc15aff2a762b13f521baf042140acec+67108864 323d2a3ce20370c4ca1d3462a344f8fd+25885655 0:227212247:var-GS000016015-ASM.tsv.bz2
 173 """
 174         self.assertEqual(arvados.CollectionReader(m2, self.api_client).manifest_text(normalize=True), m2)
 175
 176         m3 = """. 5348b82a029fd9e971a811ce1f71360b+43 3:40:md5sum.txt
 177 . 085c37f02916da1cad16f93c54d899b7+41 0:41:md5sum.txt
 178 . 8b22da26f9f433dea0a10e5ec66d73ba+43 0:43:md5sum.txt
 179 """
 180         self.assertEqual(arvados.CollectionReader(m3, self.api_client).manifest_text(normalize=True),
 181                          """. 5348b82a029fd9e971a811ce1f71360b+43 085c37f02916da1cad16f93c54d899b7+41 8b22da26f9f433dea0a10e5ec66d73ba+43 3:124:md5sum.txt
 182 """)
 183
 184         m4 = """. 204e43b8a1185621ca55a94839582e6f+67108864 0:3:foo/bar
 185 ./zzz 204e43b8a1185621ca55a94839582e6f+67108864 0:999:zzz
 186 ./foo 323d2a3ce20370c4ca1d3462a344f8fd+25885655 0:3:bar
 187 """
 188         self.assertEqual(arvados.CollectionReader(m4, self.api_client).manifest_text(normalize=True),
 189                          """./foo 204e43b8a1185621ca55a94839582e6f+67108864 323d2a3ce20370c4ca1d3462a344f8fd+25885655 0:3:bar 67108864:3:bar
 190 ./zzz 204e43b8a1185621ca55a94839582e6f+67108864 0:999:zzz
 191 """)
 192
 193         m5 = """. 204e43b8a1185621ca55a94839582e6f+67108864 0:3:foo/bar
 194 ./zzz 204e43b8a1185621ca55a94839582e6f+67108864 0:999:zzz
 195 ./foo 204e43b8a1185621ca55a94839582e6f+67108864 3:3:bar
 196 """
 197         self.assertEqual(arvados.CollectionReader(m5, self.api_client).manifest_text(normalize=True),
 198                          """./foo 204e43b8a1185621ca55a94839582e6f+67108864 0:6:bar
 199 ./zzz 204e43b8a1185621ca55a94839582e6f+67108864 0:999:zzz
 200 """)
 201
 202         with self.data_file('1000G_ref_manifest') as f6:
 203             m6 = f6.read()
 204             self.assertEqual(arvados.CollectionReader(m6, self.api_client).manifest_text(normalize=True), m6)
 205
 206         with self.data_file('jlake_manifest') as f7:
 207             m7 = f7.read()
 208             self.assertEqual(arvados.CollectionReader(m7, self.api_client).manifest_text(normalize=True), m7)
 209
 210         m8 = """./a\\040b\\040c 59ca0efa9f5633cb0371bbc0355478d8+13 0:13:hello\\040world.txt
 211 """
 212         self.assertEqual(arvados.CollectionReader(m8, self.api_client).manifest_text(normalize=True), m8)
 213
 214     def test_locators_and_ranges(self):
 215         blocks2 = [Range('a', 0, 10),
 216                    Range('b', 10, 10),
 217                    Range('c', 20, 10),
 218                    Range('d', 30, 10),
 219                    Range('e', 40, 10),
 220                    Range('f', 50, 10)]
 221
 222         self.assertEqual(arvados.locators_and_ranges(blocks2,  2,  2), [LocatorAndRange('a', 10, 2, 2)])
 223         self.assertEqual(arvados.locators_and_ranges(blocks2, 12, 2), [LocatorAndRange('b', 10, 2, 2)])
 224         self.assertEqual(arvados.locators_and_ranges(blocks2, 22, 2), [LocatorAndRange('c', 10, 2, 2)])
 225         self.assertEqual(arvados.locators_and_ranges(blocks2, 32, 2), [LocatorAndRange('d', 10, 2, 2)])
 226         self.assertEqual(arvados.locators_and_ranges(blocks2, 42, 2), [LocatorAndRange('e', 10, 2, 2)])
 227         self.assertEqual(arvados.locators_and_ranges(blocks2, 52, 2), [LocatorAndRange('f', 10, 2, 2)])
 228         self.assertEqual(arvados.locators_and_ranges(blocks2, 62, 2), [])
 229         self.assertEqual(arvados.locators_and_ranges(blocks2, -2, 2), [])
 230
 231         self.assertEqual(arvados.locators_and_ranges(blocks2,  0,  2), [LocatorAndRange('a', 10, 0, 2)])
 232         self.assertEqual(arvados.locators_and_ranges(blocks2, 10, 2), [LocatorAndRange('b', 10, 0, 2)])
 233         self.assertEqual(arvados.locators_and_ranges(blocks2, 20, 2), [LocatorAndRange('c', 10, 0, 2)])
 234         self.assertEqual(arvados.locators_and_ranges(blocks2, 30, 2), [LocatorAndRange('d', 10, 0, 2)])
 235         self.assertEqual(arvados.locators_and_ranges(blocks2, 40, 2), [LocatorAndRange('e', 10, 0, 2)])
 236         self.assertEqual(arvados.locators_and_ranges(blocks2, 50, 2), [LocatorAndRange('f', 10, 0, 2)])
 237         self.assertEqual(arvados.locators_and_ranges(blocks2, 60, 2), [])
 238         self.assertEqual(arvados.locators_and_ranges(blocks2, -2, 2), [])
 239
 240         self.assertEqual(arvados.locators_and_ranges(blocks2,  9,  2), [LocatorAndRange('a', 10, 9, 1), LocatorAndRange('b', 10, 0, 1)])
 241         self.assertEqual(arvados.locators_and_ranges(blocks2, 19, 2), [LocatorAndRange('b', 10, 9, 1), LocatorAndRange('c', 10, 0, 1)])
 242         self.assertEqual(arvados.locators_and_ranges(blocks2, 29, 2), [LocatorAndRange('c', 10, 9, 1), LocatorAndRange('d', 10, 0, 1)])
 243         self.assertEqual(arvados.locators_and_ranges(blocks2, 39, 2), [LocatorAndRange('d', 10, 9, 1), LocatorAndRange('e', 10, 0, 1)])
 244         self.assertEqual(arvados.locators_and_ranges(blocks2, 49, 2), [LocatorAndRange('e', 10, 9, 1), LocatorAndRange('f', 10, 0, 1)])
 245         self.assertEqual(arvados.locators_and_ranges(blocks2, 59, 2), [LocatorAndRange('f', 10, 9, 1)])
 246
 247
 248         blocks3 = [Range('a', 0, 10),
 249                   Range('b', 10, 10),
 250                   Range('c', 20, 10),
 251                   Range('d', 30, 10),
 252                   Range('e', 40, 10),
 253                   Range('f', 50, 10),
 254                    Range('g', 60, 10)]
 255
 256         self.assertEqual(arvados.locators_and_ranges(blocks3,  2,  2), [LocatorAndRange('a', 10, 2, 2)])
 257         self.assertEqual(arvados.locators_and_ranges(blocks3, 12, 2), [LocatorAndRange('b', 10, 2, 2)])
 258         self.assertEqual(arvados.locators_and_ranges(blocks3, 22, 2), [LocatorAndRange('c', 10, 2, 2)])
 259         self.assertEqual(arvados.locators_and_ranges(blocks3, 32, 2), [LocatorAndRange('d', 10, 2, 2)])
 260         self.assertEqual(arvados.locators_and_ranges(blocks3, 42, 2), [LocatorAndRange('e', 10, 2, 2)])
 261         self.assertEqual(arvados.locators_and_ranges(blocks3, 52, 2), [LocatorAndRange('f', 10, 2, 2)])
 262         self.assertEqual(arvados.locators_and_ranges(blocks3, 62, 2), [LocatorAndRange('g', 10, 2, 2)])
 263
 264
 265         blocks = [Range('a', 0, 10),
 266                   Range('b', 10, 15),
 267                   Range('c', 25, 5)]
 268         self.assertEqual(arvados.locators_and_ranges(blocks, 1, 0), [])
 269         self.assertEqual(arvados.locators_and_ranges(blocks, 0, 5), [LocatorAndRange('a', 10, 0, 5)])
 270         self.assertEqual(arvados.locators_and_ranges(blocks, 3, 5), [LocatorAndRange('a', 10, 3, 5)])
 271         self.assertEqual(arvados.locators_and_ranges(blocks, 0, 10), [LocatorAndRange('a', 10, 0, 10)])
 272
 273         self.assertEqual(arvados.locators_and_ranges(blocks, 0, 11), [LocatorAndRange('a', 10, 0, 10),
 274                                                                       LocatorAndRange('b', 15, 0, 1)])
 275         self.assertEqual(arvados.locators_and_ranges(blocks, 1, 11), [LocatorAndRange('a', 10, 1, 9),
 276                                                                       LocatorAndRange('b', 15, 0, 2)])
 277         self.assertEqual(arvados.locators_and_ranges(blocks, 0, 25), [LocatorAndRange('a', 10, 0, 10),
 278                                                                       LocatorAndRange('b', 15, 0, 15)])
 279
 280         self.assertEqual(arvados.locators_and_ranges(blocks, 0, 30), [LocatorAndRange('a', 10, 0, 10),
 281                                                                       LocatorAndRange('b', 15, 0, 15),
 282                                                                       LocatorAndRange('c', 5, 0, 5)])
 283         self.assertEqual(arvados.locators_and_ranges(blocks, 1, 30), [LocatorAndRange('a', 10, 1, 9),
 284                                                                       LocatorAndRange('b', 15, 0, 15),
 285                                                                       LocatorAndRange('c', 5, 0, 5)])
 286         self.assertEqual(arvados.locators_and_ranges(blocks, 0, 31), [LocatorAndRange('a', 10, 0, 10),
 287                                                                       LocatorAndRange('b', 15, 0, 15),
 288                                                                       LocatorAndRange('c', 5, 0, 5)])
 289
 290         self.assertEqual(arvados.locators_and_ranges(blocks, 15, 5), [LocatorAndRange('b', 15, 5, 5)])
 291
 292         self.assertEqual(arvados.locators_and_ranges(blocks, 8, 17), [LocatorAndRange('a', 10, 8, 2),
 293                                                                       LocatorAndRange('b', 15, 0, 15)])
 294
 295         self.assertEqual(arvados.locators_and_ranges(blocks, 8, 20), [LocatorAndRange('a', 10, 8, 2),
 296                                                                       LocatorAndRange('b', 15, 0, 15),
 297                                                                       LocatorAndRange('c', 5, 0, 3)])
 298
 299         self.assertEqual(arvados.locators_and_ranges(blocks, 26, 2), [LocatorAndRange('c', 5, 1, 2)])
 300
 301         self.assertEqual(arvados.locators_and_ranges(blocks, 9, 15), [LocatorAndRange('a', 10, 9, 1),
 302                                                                       LocatorAndRange('b', 15, 0, 14)])
 303         self.assertEqual(arvados.locators_and_ranges(blocks, 10, 15), [LocatorAndRange('b', 15, 0, 15)])
 304         self.assertEqual(arvados.locators_and_ranges(blocks, 11, 15), [LocatorAndRange('b', 15, 1, 14),
 305                                                                        LocatorAndRange('c', 5, 0, 1)])
 306
 307     class MockKeep(object):
 308         def __init__(self, content, num_retries=0):
 309             self.content = content
 310
 311         def get(self, locator, num_retries=0):
 312             return self.content[locator]
 313
 314     def test_stream_reader(self):
 315         keepblocks = {'aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa+10': 'abcdefghij',
 316                       'bbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbb+15': 'klmnopqrstuvwxy',
 317                       'cccccccccccccccccccccccccccccccc+5': 'z0123'}
 318         mk = self.MockKeep(keepblocks)
 319
 320         sr = arvados.StreamReader([".", "aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa+10", "bbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbb+15", "cccccccccccccccccccccccccccccccc+5", "0:30:foo"], mk)
 321
 322         content = 'abcdefghijklmnopqrstuvwxyz0123456789'
 323
 324         self.assertEqual(sr.readfrom(0, 30), content[0:30])
 325         self.assertEqual(sr.readfrom(2, 30), content[2:30])
 326
 327         self.assertEqual(sr.readfrom(2, 8), content[2:10])
 328         self.assertEqual(sr.readfrom(0, 10), content[0:10])
 329
 330         self.assertEqual(sr.readfrom(0, 5), content[0:5])
 331         self.assertEqual(sr.readfrom(5, 5), content[5:10])
 332         self.assertEqual(sr.readfrom(10, 5), content[10:15])
 333         self.assertEqual(sr.readfrom(15, 5), content[15:20])
 334         self.assertEqual(sr.readfrom(20, 5), content[20:25])
 335         self.assertEqual(sr.readfrom(25, 5), content[25:30])
 336         self.assertEqual(sr.readfrom(30, 5), '')
 337
 338     def test_extract_file(self):
 339         m1 = """. 5348b82a029fd9e971a811ce1f71360b+43 0:43:md5sum.txt
 340 . 085c37f02916da1cad16f93c54d899b7+41 0:41:md6sum.txt
 341 . 8b22da26f9f433dea0a10e5ec66d73ba+43 0:43:md7sum.txt
 342 . 085c37f02916da1cad16f93c54d899b7+41 5348b82a029fd9e971a811ce1f71360b+43 8b22da26f9f433dea0a10e5ec66d73ba+43 47:80:md8sum.txt
 343 . 085c37f02916da1cad16f93c54d899b7+41 5348b82a029fd9e971a811ce1f71360b+43 8b22da26f9f433dea0a10e5ec66d73ba+43 40:80:md9sum.txt
 344 """
 345
 346         m2 = arvados.CollectionReader(m1, self.api_client).manifest_text(normalize=True)
 347
 348         self.assertEqual(m2,
 349                          ". 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")
 350         files = arvados.CollectionReader(
 351             m2, self.api_client).all_streams()[0].files()
 352
 353         self.assertEqual(files['md5sum.txt'].as_manifest(),
 354                          ". 5348b82a029fd9e971a811ce1f71360b+43 0:43:md5sum.txt\n")
 355         self.assertEqual(files['md6sum.txt'].as_manifest(),
 356                          ". 085c37f02916da1cad16f93c54d899b7+41 0:41:md6sum.txt\n")
 357         self.assertEqual(files['md7sum.txt'].as_manifest(),
 358                          ". 8b22da26f9f433dea0a10e5ec66d73ba+43 0:43:md7sum.txt\n")
 359         self.assertEqual(files['md9sum.txt'].as_manifest(),
 360                          ". 085c37f02916da1cad16f93c54d899b7+41 5348b82a029fd9e971a811ce1f71360b+43 8b22da26f9f433dea0a10e5ec66d73ba+43 40:80:md9sum.txt\n")
 361
 362     def test_write_directory_tree(self):
 363         cwriter = arvados.CollectionWriter(self.api_client)
 364         cwriter.write_directory_tree(self.build_directory_tree(
 365                 ['basefile', 'subdir/subfile']))
 366         self.assertEqual(cwriter.manifest_text(),
 367                          """. c5110c5ac93202d8e0f9e381f22bac0f+8 0:8:basefile
 368 ./subdir 1ca4dec89403084bf282ad31e6cf7972+14 0:14:subfile\n""")
 369
 370     def test_write_named_directory_tree(self):
 371         cwriter = arvados.CollectionWriter(self.api_client)
 372         cwriter.write_directory_tree(self.build_directory_tree(
 373                 ['basefile', 'subdir/subfile']), 'root')
 374         self.assertEqual(
 375             cwriter.manifest_text(),
 376             """./root c5110c5ac93202d8e0f9e381f22bac0f+8 0:8:basefile
 377 ./root/subdir 1ca4dec89403084bf282ad31e6cf7972+14 0:14:subfile\n""")
 378
 379     def test_write_directory_tree_in_one_stream(self):
 380         cwriter = arvados.CollectionWriter(self.api_client)
 381         cwriter.write_directory_tree(self.build_directory_tree(
 382                 ['basefile', 'subdir/subfile']), max_manifest_depth=0)
 383         self.assertEqual(cwriter.manifest_text(),
 384                          """. 4ace875ffdc6824a04950f06858f4465+22 0:8:basefile 8:14:subdir/subfile\n""")
 385
 386     def test_write_directory_tree_with_limited_recursion(self):
 387         cwriter = arvados.CollectionWriter(self.api_client)
 388         cwriter.write_directory_tree(
 389             self.build_directory_tree(['f1', 'd1/f2', 'd1/d2/f3']),
 390             max_manifest_depth=1)
 391         self.assertEqual(cwriter.manifest_text(),
 392                          """. bd19836ddb62c11c55ab251ccaca5645+2 0:2:f1
 393 ./d1 50170217e5b04312024aa5cd42934494+13 0:8:d2/f3 8:5:f2\n""")
 394
 395     def test_write_directory_tree_with_zero_recursion(self):
 396         cwriter = arvados.CollectionWriter(self.api_client)
 397         content = 'd1/d2/f3d1/f2f1'
 398         blockhash = tutil.str_keep_locator(content)
 399         cwriter.write_directory_tree(
 400             self.build_directory_tree(['f1', 'd1/f2', 'd1/d2/f3']),
 401             max_manifest_depth=0)
 402         self.assertEqual(
 403             cwriter.manifest_text(),
 404             ". {} 0:8:d1/d2/f3 8:5:d1/f2 13:2:f1\n".format(blockhash))
 405
 406     def test_write_one_file(self):
 407         cwriter = arvados.CollectionWriter(self.api_client)
 408         with self.make_test_file() as testfile:
 409             cwriter.write_file(testfile.name)
 410             self.assertEqual(
 411                 cwriter.manifest_text(),
 412                 ". 098f6bcd4621d373cade4e832627b4f6+4 0:4:{}\n".format(
 413                     os.path.basename(testfile.name)))
 414
 415     def test_write_named_file(self):
 416         cwriter = arvados.CollectionWriter(self.api_client)
 417         with self.make_test_file() as testfile:
 418             cwriter.write_file(testfile.name, 'foo')
 419             self.assertEqual(cwriter.manifest_text(),
 420                              ". 098f6bcd4621d373cade4e832627b4f6+4 0:4:foo\n")
 421
 422     def test_write_multiple_files(self):
 423         cwriter = arvados.CollectionWriter(self.api_client)
 424         for letter in 'ABC':
 425             with self.make_test_file(letter) as testfile:
 426                 cwriter.write_file(testfile.name, letter)
 427         self.assertEqual(
 428             cwriter.manifest_text(),
 429             ". 902fbdd2b1df0c4f70b4a5d23525e932+3 0:1:A 1:1:B 2:1:C\n")
 430
 431     def test_basic_resume(self):
 432         cwriter = TestResumableWriter()
 433         with self.make_test_file() as testfile:
 434             cwriter.write_file(testfile.name, 'test')
 435             resumed = TestResumableWriter.from_state(cwriter.current_state())
 436         self.assertEqual(cwriter.manifest_text(), resumed.manifest_text(),
 437                           "resumed CollectionWriter had different manifest")
 438
 439     def test_resume_fails_when_missing_dependency(self):
 440         cwriter = TestResumableWriter()
 441         with self.make_test_file() as testfile:
 442             cwriter.write_file(testfile.name, 'test')
 443         self.assertRaises(arvados.errors.StaleWriterStateError,
 444                           TestResumableWriter.from_state,
 445                           cwriter.current_state())
 446
 447     def test_resume_fails_when_dependency_mtime_changed(self):
 448         cwriter = TestResumableWriter()
 449         with self.make_test_file() as testfile:
 450             cwriter.write_file(testfile.name, 'test')
 451             os.utime(testfile.name, (0, 0))
 452             self.assertRaises(arvados.errors.StaleWriterStateError,
 453                               TestResumableWriter.from_state,
 454                               cwriter.current_state())
 455
 456     def test_resume_fails_when_dependency_is_nonfile(self):
 457         cwriter = TestResumableWriter()
 458         cwriter.write_file('/dev/null', 'empty')
 459         self.assertRaises(arvados.errors.StaleWriterStateError,
 460                           TestResumableWriter.from_state,
 461                           cwriter.current_state())
 462
 463     def test_resume_fails_when_dependency_size_changed(self):
 464         cwriter = TestResumableWriter()
 465         with self.make_test_file() as testfile:
 466             cwriter.write_file(testfile.name, 'test')
 467             orig_mtime = os.fstat(testfile.fileno()).st_mtime
 468             testfile.write('extra')
 469             testfile.flush()
 470             os.utime(testfile.name, (orig_mtime, orig_mtime))
 471             self.assertRaises(arvados.errors.StaleWriterStateError,
 472                               TestResumableWriter.from_state,
 473                               cwriter.current_state())
 474
 475     def test_resume_fails_with_expired_locator(self):
 476         cwriter = TestResumableWriter()
 477         state = cwriter.current_state()
 478         # Add an expired locator to the state.
 479         state['_current_stream_locators'].append(''.join([
 480                     'a' * 32, '+1+A', 'b' * 40, '@', '10000000']))
 481         self.assertRaises(arvados.errors.StaleWriterStateError,
 482                           TestResumableWriter.from_state, state)
 483
 484     def test_arbitrary_objects_not_resumable(self):
 485         cwriter = TestResumableWriter()
 486         with open('/dev/null') as badfile:
 487             self.assertRaises(arvados.errors.AssertionError,
 488                               cwriter.write_file, badfile)
 489
 490     def test_arbitrary_writes_not_resumable(self):
 491         cwriter = TestResumableWriter()
 492         self.assertRaises(arvados.errors.AssertionError,
 493                           cwriter.write, "badtext")
 494
 495     def test_read_arbitrary_data_with_collection_reader(self):
 496         # arv-get relies on this to do "arv-get {keep-locator} -".
 497         self.write_foo_bar_baz()
 498         self.assertEqual(
 499             'foobar',
 500             arvados.CollectionReader(
 501                 '3858f62230ac3c915f300c664312c63f+6'
 502                 ).manifest_text())
 503
 504
 505 class CollectionTestMixin(tutil.ApiClientMock):
 506     API_COLLECTIONS = run_test_server.fixture('collections')
 507     DEFAULT_COLLECTION = API_COLLECTIONS['foo_file']
 508     DEFAULT_DATA_HASH = DEFAULT_COLLECTION['portable_data_hash']
 509     DEFAULT_MANIFEST = DEFAULT_COLLECTION['manifest_text']
 510     DEFAULT_UUID = DEFAULT_COLLECTION['uuid']
 511     ALT_COLLECTION = API_COLLECTIONS['bar_file']
 512     ALT_DATA_HASH = ALT_COLLECTION['portable_data_hash']
 513     ALT_MANIFEST = ALT_COLLECTION['manifest_text']
 514
 515     def api_client_mock(self, status=200):
 516         client = super(CollectionTestMixin, self).api_client_mock()
 517         self.mock_keep_services(client, status=status, service_type='proxy', count=1)
 518         return client
 519
 520
 521 @tutil.skip_sleep
 522 class CollectionReaderTestCase(unittest.TestCase, CollectionTestMixin):
 523     def mock_get_collection(self, api_mock, code, body):
 524         body = self.API_COLLECTIONS.get(body)
 525         self._mock_api_call(api_mock.collections().get, code, body)
 526
 527     def api_client_mock(self, status=200):
 528         client = super(CollectionReaderTestCase, self).api_client_mock()
 529         self.mock_get_collection(client, status, 'foo_file')
 530         return client
 531
 532     def test_init_no_default_retries(self):
 533         client = self.api_client_mock(200)
 534         reader = arvados.CollectionReader(self.DEFAULT_UUID, api_client=client)
 535         reader.manifest_text()
 536         client.collections().get().execute.assert_called_with(num_retries=0)
 537
 538     def test_uuid_init_success(self):
 539         client = self.api_client_mock(200)
 540         reader = arvados.CollectionReader(self.DEFAULT_UUID, api_client=client,
 541                                           num_retries=3)
 542         self.assertEqual(self.DEFAULT_COLLECTION['manifest_text'],
 543                          reader.manifest_text())
 544         client.collections().get().execute.assert_called_with(num_retries=3)
 545
 546     def test_uuid_init_failure_raises_api_error(self):
 547         client = self.api_client_mock(500)
 548         with self.assertRaises(arvados.errors.ApiError):
 549             reader = arvados.CollectionReader(self.DEFAULT_UUID, api_client=client)
 550
 551     def test_locator_init(self):
 552         client = self.api_client_mock(200)
 553         # Ensure Keep will not return anything if asked.
 554         with tutil.mock_keep_responses(None, 404):
 555             reader = arvados.CollectionReader(self.DEFAULT_DATA_HASH,
 556                                               api_client=client)
 557             self.assertEqual(self.DEFAULT_MANIFEST, reader.manifest_text())
 558
 559     def test_locator_init_fallback_to_keep(self):
 560         # crunch-job needs this to read manifests that have only ever
 561         # been written to Keep.
 562         client = self.api_client_mock(200)
 563         self.mock_get_collection(client, 404, None)
 564         with tutil.mock_keep_responses(self.DEFAULT_MANIFEST, 200):
 565             reader = arvados.CollectionReader(self.DEFAULT_DATA_HASH,
 566                                               api_client=client)
 567             self.assertEqual(self.DEFAULT_MANIFEST, reader.manifest_text())
 568
 569     def test_uuid_init_no_fallback_to_keep(self):
 570         # Do not look up a collection UUID in Keep.
 571         client = self.api_client_mock(404)
 572         with tutil.mock_keep_responses(self.DEFAULT_MANIFEST, 200):
 573             with self.assertRaises(arvados.errors.ApiError):
 574                 reader = arvados.CollectionReader(self.DEFAULT_UUID,
 575                                                   api_client=client)
 576
 577     def test_try_keep_first_if_permission_hint(self):
 578         # To verify that CollectionReader tries Keep first here, we
 579         # mock API server to return the wrong data.
 580         client = self.api_client_mock(200)
 581         with tutil.mock_keep_responses(self.ALT_MANIFEST, 200):
 582             self.assertEqual(
 583                 self.ALT_MANIFEST,
 584                 arvados.CollectionReader(
 585                     self.ALT_DATA_HASH + '+Affffffffffffffffffffffffffffffffffffffff@fedcba98',
 586                     api_client=client).manifest_text())
 587
 588     def test_init_num_retries_propagated(self):
 589         # More of an integration test...
 590         client = self.api_client_mock(200)
 591         reader = arvados.CollectionReader(self.DEFAULT_UUID, api_client=client,
 592                                           num_retries=3)
 593         with tutil.mock_keep_responses('foo', 500, 500, 200):
 594             self.assertEqual('foo',
 595                              ''.join(f.read(9) for f in reader.all_files()))
 596
 597     def test_read_nonnormalized_manifest_with_collection_reader(self):
 598         # client should be able to use CollectionReader on a manifest without normalizing it
 599         client = self.api_client_mock(500)
 600         nonnormal = ". acbd18db4cc2f85cedef654fccc4a4d8+3+Aabadbadbee@abeebdee 0:3:foo.txt 1:0:bar.txt 0:3:foo.txt\n"
 601         reader = arvados.CollectionReader(
 602             nonnormal,
 603             api_client=client, num_retries=0)
 604         # Ensure stripped_manifest() doesn't mangle our manifest in
 605         # any way other than stripping hints.
 606         self.assertEqual(
 607             re.sub('\+[^\d\s\+]+', '', nonnormal),
 608             reader.stripped_manifest())
 609         # Ensure stripped_manifest() didn't mutate our reader.
 610         self.assertEqual(nonnormal, reader.manifest_text())
 611         # Ensure the files appear in the order given in the manifest.
 612         self.assertEqual(
 613             [[6, '.', 'foo.txt'],
 614              [0, '.', 'bar.txt']],
 615             [[f.size(), f.stream_name(), f.name()]
 616              for f in reader.all_streams()[0].all_files()])
 617
 618     def test_read_empty_collection(self):
 619         client = self.api_client_mock(200)
 620         self.mock_get_collection(client, 200, 'empty')
 621         reader = arvados.CollectionReader('d41d8cd98f00b204e9800998ecf8427e+0',
 622                                           api_client=client)
 623         self.assertEqual('', reader.manifest_text())
 624
 625     def test_api_response(self):
 626         client = self.api_client_mock()
 627         reader = arvados.CollectionReader(self.DEFAULT_UUID, api_client=client)
 628         self.assertEqual(self.DEFAULT_COLLECTION, reader.api_response())
 629
 630     def test_api_response_with_collection_from_keep(self):
 631         client = self.api_client_mock()
 632         self.mock_get_collection(client, 404, 'foo')
 633         with tutil.mock_keep_responses(self.DEFAULT_MANIFEST, 200):
 634             reader = arvados.CollectionReader(self.DEFAULT_DATA_HASH,
 635                                               api_client=client)
 636             api_response = reader.api_response()
 637         self.assertIsNone(api_response)
 638
 639     def check_open_file(self, coll_file, stream_name, file_name, file_size):
 640         self.assertFalse(coll_file.closed, "returned file is not open")
 641         self.assertEqual(stream_name, coll_file.stream_name())
 642         self.assertEqual(file_name, coll_file.name)
 643         self.assertEqual(file_size, coll_file.size())
 644
 645     def test_open_collection_file_one_argument(self):
 646         client = self.api_client_mock(200)
 647         reader = arvados.CollectionReader(self.DEFAULT_UUID, api_client=client)
 648         cfile = reader.open('./foo')
 649         self.check_open_file(cfile, '.', 'foo', 3)
 650
 651     def test_open_deep_file(self):
 652         coll_name = 'collection_with_files_in_subdir'
 653         client = self.api_client_mock(200)
 654         self.mock_get_collection(client, 200, coll_name)
 655         reader = arvados.CollectionReader(
 656             self.API_COLLECTIONS[coll_name]['uuid'], api_client=client)
 657         cfile = reader.open('./subdir2/subdir3/file2_in_subdir3.txt')
 658         self.check_open_file(cfile, './subdir2/subdir3', 'file2_in_subdir3.txt',
 659                              32)
 660
 661     def test_open_nonexistent_stream(self):
 662         client = self.api_client_mock(200)
 663         reader = arvados.CollectionReader(self.DEFAULT_UUID, api_client=client)
 664         self.assertRaises(IOError, reader.open, './nonexistent/foo')
 665
 666     def test_open_nonexistent_file(self):
 667         client = self.api_client_mock(200)
 668         reader = arvados.CollectionReader(self.DEFAULT_UUID, api_client=client)
 669         self.assertRaises(IOError, reader.open, 'nonexistent')
 670
 671
 672 @tutil.skip_sleep
 673 class CollectionWriterTestCase(unittest.TestCase, CollectionTestMixin):
 674     def mock_keep(self, body, *codes, **headers):
 675         headers.setdefault('x-keep-replicas-stored', 2)
 676         return tutil.mock_keep_responses(body, *codes, **headers)
 677
 678     def foo_writer(self, **kwargs):
 679         kwargs.setdefault('api_client', self.api_client_mock())
 680         writer = arvados.CollectionWriter(**kwargs)
 681         writer.start_new_file('foo')
 682         writer.write('foo')
 683         return writer
 684
 685     def test_write_whole_collection(self):
 686         writer = self.foo_writer()
 687         with self.mock_keep(self.DEFAULT_DATA_HASH, 200, 200):
 688             self.assertEqual(self.DEFAULT_DATA_HASH, writer.finish())
 689
 690     def test_write_no_default(self):
 691         writer = self.foo_writer()
 692         with self.mock_keep(None, 500):
 693             with self.assertRaises(arvados.errors.KeepWriteError):
 694                 writer.finish()
 695
 696     def test_write_insufficient_replicas_via_proxy(self):
 697         writer = self.foo_writer(replication=3)
 698         with self.mock_keep(None, 200, **{'x-keep-replicas-stored': 2}):
 699             with self.assertRaises(arvados.errors.KeepWriteError):
 700                 writer.manifest_text()
 701
 702     def test_write_insufficient_replicas_via_disks(self):
 703         client = mock.MagicMock(name='api_client')
 704         with self.mock_keep(
 705                 None, 200, 200,
 706                 **{'x-keep-replicas-stored': 1}) as keepmock:
 707             self.mock_keep_services(client, status=200, service_type='disk', count=2)
 708             writer = self.foo_writer(api_client=client, replication=3)
 709             with self.assertRaises(arvados.errors.KeepWriteError):
 710                 writer.manifest_text()
 711
 712     def test_write_three_replicas(self):
 713         client = mock.MagicMock(name='api_client')
 714         with self.mock_keep(
 715                 "", 500, 500, 500, 200, 200, 200,
 716                 **{'x-keep-replicas-stored': 1}) as keepmock:
 717             self.mock_keep_services(client, status=200, service_type='disk', count=6)
 718             writer = self.foo_writer(api_client=client, replication=3)
 719             writer.manifest_text()
 720             self.assertEqual(6, keepmock.call_count)
 721
 722     def test_write_whole_collection_through_retries(self):
 723         writer = self.foo_writer(num_retries=2)
 724         with self.mock_keep(self.DEFAULT_DATA_HASH,
 725                             500, 500, 200, 500, 500, 200):
 726             self.assertEqual(self.DEFAULT_DATA_HASH, writer.finish())
 727
 728     def test_flush_data_retries(self):
 729         writer = self.foo_writer(num_retries=2)
 730         foo_hash = self.DEFAULT_MANIFEST.split()[1]
 731         with self.mock_keep(foo_hash, 500, 200):
 732             writer.flush_data()
 733         self.assertEqual(self.DEFAULT_MANIFEST, writer.manifest_text())
 734
 735     def test_one_open(self):
 736         client = self.api_client_mock()
 737         writer = arvados.CollectionWriter(client)
 738         with writer.open('out') as out_file:
 739             self.assertEqual('.', writer.current_stream_name())
 740             self.assertEqual('out', writer.current_file_name())
 741             out_file.write('test data')
 742             data_loc = tutil.str_keep_locator('test data')
 743         self.assertTrue(out_file.closed, "writer file not closed after context")
 744         self.assertRaises(ValueError, out_file.write, 'extra text')
 745         with self.mock_keep(data_loc, 200) as keep_mock:
 746             self.assertEqual(". {} 0:9:out\n".format(data_loc),
 747                              writer.manifest_text())
 748
 749     def test_open_writelines(self):
 750         client = self.api_client_mock()
 751         writer = arvados.CollectionWriter(client)
 752         with writer.open('six') as out_file:
 753             out_file.writelines(['12', '34', '56'])
 754             data_loc = tutil.str_keep_locator('123456')
 755         with self.mock_keep(data_loc, 200) as keep_mock:
 756             self.assertEqual(". {} 0:6:six\n".format(data_loc),
 757                              writer.manifest_text())
 758
 759     def test_open_flush(self):
 760         client = self.api_client_mock()
 761         data_loc1 = tutil.str_keep_locator('flush1')
 762         data_loc2 = tutil.str_keep_locator('flush2')
 763         with self.mock_keep((data_loc1, 200), (data_loc2, 200)) as keep_mock:
 764             writer = arvados.CollectionWriter(client)
 765             with writer.open('flush_test') as out_file:
 766                 out_file.write('flush1')
 767                 out_file.flush()
 768                 out_file.write('flush2')
 769             self.assertEqual(". {} {} 0:12:flush_test\n".format(data_loc1,
 770                                                                 data_loc2),
 771                              writer.manifest_text())
 772
 773     def test_two_opens_same_stream(self):
 774         client = self.api_client_mock()
 775         writer = arvados.CollectionWriter(client)
 776         with writer.open('.', '1') as out_file:
 777             out_file.write('1st')
 778         with writer.open('.', '2') as out_file:
 779             out_file.write('2nd')
 780         data_loc = tutil.str_keep_locator('1st2nd')
 781         with self.mock_keep(data_loc, 200) as keep_mock:
 782             self.assertEqual(". {} 0:3:1 3:3:2\n".format(data_loc),
 783                              writer.manifest_text())
 784
 785     def test_two_opens_two_streams(self):
 786         client = self.api_client_mock()
 787         data_loc1 = tutil.str_keep_locator('file')
 788         data_loc2 = tutil.str_keep_locator('indir')
 789         with self.mock_keep((data_loc1, 200), (data_loc2, 200)) as keep_mock:
 790             writer = arvados.CollectionWriter(client)
 791             with writer.open('file') as out_file:
 792                 out_file.write('file')
 793             with writer.open('./dir', 'indir') as out_file:
 794                 out_file.write('indir')
 795             expected = ". {} 0:4:file\n./dir {} 0:5:indir\n".format(
 796                 data_loc1, data_loc2)
 797             self.assertEqual(expected, writer.manifest_text())
 798
 799     def test_dup_open_fails(self):
 800         client = self.api_client_mock()
 801         writer = arvados.CollectionWriter(client)
 802         file1 = writer.open('one')
 803         self.assertRaises(arvados.errors.AssertionError, writer.open, 'two')
 804
 805
 806 class NewCollectionTestCase(unittest.TestCase, CollectionTestMixin):
 807
 808     def test_replication_desired_kept_on_load(self):
 809         m = '. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt 0:10:count2.txt\n'
 810         c1 = Collection(m, replication_desired=1)
 811         c1.save_new()
 812         loc = c1.manifest_locator()
 813         c2 = Collection(loc)
 814         self.assertEqual(c1.manifest_text, c2.manifest_text)
 815         self.assertEqual(c1.replication_desired, c2.replication_desired)
 816
 817     def test_replication_desired_not_loaded_if_provided(self):
 818         m = '. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt 0:10:count2.txt\n'
 819         c1 = Collection(m, replication_desired=1)
 820         c1.save_new()
 821         loc = c1.manifest_locator()
 822         c2 = Collection(loc, replication_desired=2)
 823         self.assertEqual(c1.manifest_text, c2.manifest_text)
 824         self.assertNotEqual(c1.replication_desired, c2.replication_desired)
 825
 826     def test_init_manifest(self):
 827         m1 = """. 5348b82a029fd9e971a811ce1f71360b+43 0:43:md5sum.txt
 828 . 085c37f02916da1cad16f93c54d899b7+41 0:41:md5sum.txt
 829 . 8b22da26f9f433dea0a10e5ec66d73ba+43 0:43:md5sum.txt
 830 """
 831         self.assertEqual(m1, CollectionReader(m1).manifest_text(normalize=False))
 832         self.assertEqual(". 5348b82a029fd9e971a811ce1f71360b+43 085c37f02916da1cad16f93c54d899b7+41 8b22da26f9f433dea0a10e5ec66d73ba+43 0:127:md5sum.txt\n", CollectionReader(m1).manifest_text(normalize=True))
 833
 834     def test_init_manifest_with_collision(self):
 835         m1 = """. 5348b82a029fd9e971a811ce1f71360b+43 0:43:md5sum.txt
 836 ./md5sum.txt 085c37f02916da1cad16f93c54d899b7+41 0:41:md5sum.txt
 837 """
 838         with self.assertRaises(arvados.errors.ArgumentError):
 839             self.assertEqual(m1, CollectionReader(m1))
 840
 841     def test_init_manifest_with_error(self):
 842         m1 = """. 0:43:md5sum.txt"""
 843         with self.assertRaises(arvados.errors.ArgumentError):
 844             self.assertEqual(m1, CollectionReader(m1))
 845
 846     def test_remove(self):
 847         c = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt 0:10:count2.txt\n')
 848         self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt 0:10:count2.txt\n", c.portable_manifest_text())
 849         self.assertIn("count1.txt", c)
 850         c.remove("count1.txt")
 851         self.assertNotIn("count1.txt", c)
 852         self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count2.txt\n", c.portable_manifest_text())
 853         with self.assertRaises(arvados.errors.ArgumentError):
 854             c.remove("")
 855
 856     def test_find(self):
 857         c = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt 0:10:count2.txt\n')
 858         self.assertIs(c.find("."), c)
 859         self.assertIs(c.find("./count1.txt"), c["count1.txt"])
 860         self.assertIs(c.find("count1.txt"), c["count1.txt"])
 861         with self.assertRaises(IOError):
 862             c.find("/.")
 863         with self.assertRaises(arvados.errors.ArgumentError):
 864             c.find("")
 865         self.assertIs(c.find("./nonexistant.txt"), None)
 866         self.assertIs(c.find("./nonexistantsubdir/nonexistant.txt"), None)
 867
 868     def test_remove_in_subdir(self):
 869         c = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n./foo 781e5e245d69b566979b86e28d23f2c7+10 0:10:count2.txt\n')
 870         c.remove("foo/count2.txt")
 871         self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n", c.portable_manifest_text())
 872
 873     def test_remove_empty_subdir(self):
 874         c = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n./foo 781e5e245d69b566979b86e28d23f2c7+10 0:10:count2.txt\n')
 875         c.remove("foo/count2.txt")
 876         c.remove("foo")
 877         self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n", c.portable_manifest_text())
 878
 879     def test_remove_nonempty_subdir(self):
 880         c = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n./foo 781e5e245d69b566979b86e28d23f2c7+10 0:10:count2.txt\n')
 881         with self.assertRaises(IOError):
 882             c.remove("foo")
 883         c.remove("foo", recursive=True)
 884         self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n", c.portable_manifest_text())
 885
 886     def test_copy_to_file_in_dir(self):
 887         c = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n')
 888         c.copy("count1.txt", "foo/count2.txt")
 889         self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n./foo 781e5e245d69b566979b86e28d23f2c7+10 0:10:count2.txt\n", c.portable_manifest_text())
 890
 891     def test_copy_file(self):
 892         c = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n')
 893         c.copy("count1.txt", "count2.txt")
 894         self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt 0:10:count2.txt\n", c.portable_manifest_text())
 895
 896     def test_copy_to_existing_dir(self):
 897         c = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n./foo 781e5e245d69b566979b86e28d23f2c7+10 0:10:count2.txt\n')
 898         c.copy("count1.txt", "foo")
 899         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())
 900
 901     def test_copy_to_new_dir(self):
 902         c = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n')
 903         c.copy("count1.txt", "foo/")
 904         self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n./foo 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n", c.portable_manifest_text())
 905
 906     def test_rename_file(self):
 907         c = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n')
 908         c.rename("count1.txt", "count2.txt")
 909         self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count2.txt\n", c.manifest_text())
 910
 911     def test_move_file_to_dir(self):
 912         c = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n')
 913         c.mkdirs("foo")
 914         c.rename("count1.txt", "foo/count2.txt")
 915         self.assertEqual("./foo 781e5e245d69b566979b86e28d23f2c7+10 0:10:count2.txt\n", c.manifest_text())
 916
 917     def test_move_file_to_other(self):
 918         c1 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n')
 919         c2 = Collection()
 920         c2.rename("count1.txt", "count2.txt", source_collection=c1)
 921         self.assertEqual("", c1.manifest_text())
 922         self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count2.txt\n", c2.manifest_text())
 923
 924     def test_clone(self):
 925         c = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n./foo 781e5e245d69b566979b86e28d23f2c7+10 0:10:count2.txt\n')
 926         cl = c.clone()
 927         self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n./foo 781e5e245d69b566979b86e28d23f2c7+10 0:10:count2.txt\n", cl.portable_manifest_text())
 928
 929     def test_diff_del_add(self):
 930         c1 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n')
 931         c2 = Collection('. 5348b82a029fd9e971a811ce1f71360b+43 0:10:count2.txt\n')
 932         d = c2.diff(c1)
 933         self.assertEqual(d, [('del', './count2.txt', c2["count2.txt"]),
 934                              ('add', './count1.txt', c1["count1.txt"])])
 935         d = c1.diff(c2)
 936         self.assertEqual(d, [('del', './count1.txt', c1["count1.txt"]),
 937                              ('add', './count2.txt', c2["count2.txt"])])
 938         self.assertNotEqual(c1.portable_manifest_text(), c2.portable_manifest_text())
 939         c1.apply(d)
 940         self.assertEqual(c1.portable_manifest_text(), c2.portable_manifest_text())
 941
 942     def test_diff_same(self):
 943         c1 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n')
 944         c2 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n')
 945         d = c2.diff(c1)
 946         self.assertEqual(d, [('tok', './count1.txt', c2["count1.txt"], c1["count1.txt"])])
 947         d = c1.diff(c2)
 948         self.assertEqual(d, [('tok', './count1.txt', c2["count1.txt"], c1["count1.txt"])])
 949
 950         self.assertEqual(c1.portable_manifest_text(), c2.portable_manifest_text())
 951         c1.apply(d)
 952         self.assertEqual(c1.portable_manifest_text(), c2.portable_manifest_text())
 953
 954     def test_diff_mod(self):
 955         c1 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n')
 956         c2 = Collection('. 5348b82a029fd9e971a811ce1f71360b+43 0:10:count1.txt\n')
 957         d = c2.diff(c1)
 958         self.assertEqual(d, [('mod', './count1.txt', c2["count1.txt"], c1["count1.txt"])])
 959         d = c1.diff(c2)
 960         self.assertEqual(d, [('mod', './count1.txt', c1["count1.txt"], c2["count1.txt"])])
 961
 962         self.assertNotEqual(c1.portable_manifest_text(), c2.portable_manifest_text())
 963         c1.apply(d)
 964         self.assertEqual(c1.portable_manifest_text(), c2.portable_manifest_text())
 965
 966     def test_diff_add(self):
 967         c1 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n')
 968         c2 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 5348b82a029fd9e971a811ce1f71360b+43 0:10:count1.txt 10:20:count2.txt\n')
 969         d = c2.diff(c1)
 970         self.assertEqual(d, [('del', './count2.txt', c2["count2.txt"]),
 971                              ('tok', './count1.txt', c2["count1.txt"], c1["count1.txt"])])
 972         d = c1.diff(c2)
 973         self.assertEqual(d, [('add', './count2.txt', c2["count2.txt"]),
 974                              ('tok', './count1.txt', c2["count1.txt"], c1["count1.txt"])])
 975
 976         self.assertNotEqual(c1.portable_manifest_text(), c2.portable_manifest_text())
 977         c1.apply(d)
 978         self.assertEqual(c1.portable_manifest_text(), c2.portable_manifest_text())
 979
 980     def test_diff_add_in_subcollection(self):
 981         c1 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n')
 982         c2 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n./foo 5348b82a029fd9e971a811ce1f71360b+43 0:10:count2.txt\n')
 983         d = c2.diff(c1)
 984         self.assertEqual(d, [('del', './foo', c2["foo"]),
 985                              ('tok', './count1.txt', c2["count1.txt"], c1["count1.txt"])])
 986         d = c1.diff(c2)
 987         self.assertEqual(d, [('add', './foo', c2["foo"]),
 988                              ('tok', './count1.txt', c2["count1.txt"], c1["count1.txt"])])
 989
 990         self.assertNotEqual(c1.portable_manifest_text(), c2.portable_manifest_text())
 991         c1.apply(d)
 992         self.assertEqual(c1.portable_manifest_text(), c2.portable_manifest_text())
 993
 994     def test_diff_del_add_in_subcollection(self):
 995         c1 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n./foo 5348b82a029fd9e971a811ce1f71360b+43 0:10:count2.txt\n')
 996         c2 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n./foo 5348b82a029fd9e971a811ce1f71360b+43 0:3:count3.txt\n')
 997
 998         d = c2.diff(c1)
 999         self.assertEqual(d, [('del', './foo/count3.txt', c2.find("foo/count3.txt")),
1000                              ('add', './foo/count2.txt', c1.find("foo/count2.txt")),
1001                              ('tok', './count1.txt', c2["count1.txt"], c1["count1.txt"])])
1002         d = c1.diff(c2)
1003         self.assertEqual(d, [('del', './foo/count2.txt', c1.find("foo/count2.txt")),
1004                              ('add', './foo/count3.txt', c2.find("foo/count3.txt")),
1005                              ('tok', './count1.txt', c2["count1.txt"], c1["count1.txt"])])
1006
1007         self.assertNotEqual(c1.portable_manifest_text(), c2.portable_manifest_text())
1008         c1.apply(d)
1009         self.assertEqual(c1.portable_manifest_text(), c2.portable_manifest_text())
1010
1011     def test_diff_mod_in_subcollection(self):
1012         c1 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n./foo 5348b82a029fd9e971a811ce1f71360b+43 0:10:count2.txt\n')
1013         c2 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt 0:3:foo\n')
1014         d = c2.diff(c1)
1015         self.assertEqual(d, [('mod', './foo', c2["foo"], c1["foo"]),
1016                              ('tok', './count1.txt', c2["count1.txt"], c1["count1.txt"])])
1017         d = c1.diff(c2)
1018         self.assertEqual(d, [('mod', './foo', c1["foo"], c2["foo"]),
1019                              ('tok', './count1.txt', c2["count1.txt"], c1["count1.txt"])])
1020
1021         self.assertNotEqual(c1.portable_manifest_text(), c2.portable_manifest_text())
1022         c1.apply(d)
1023         self.assertEqual(c1.portable_manifest_text(), c2.portable_manifest_text())
1024
1025     def test_conflict_keep_local_change(self):
1026         c1 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n')
1027         c2 = Collection('. 5348b82a029fd9e971a811ce1f71360b+43 0:10:count2.txt\n')
1028         d = c1.diff(c2)
1029         self.assertEqual(d, [('del', './count1.txt', c1["count1.txt"]),
1030                              ('add', './count2.txt', c2["count2.txt"])])
1031         f = c1.open("count1.txt", "w")
1032         f.write("zzzzz")
1033
1034         # c1 changed, so it should not be deleted.
1035         c1.apply(d)
1036         self.assertEqual(c1.portable_manifest_text(), ". 95ebc3c7b3b9f1d2c40fec14415d3cb8+5 5348b82a029fd9e971a811ce1f71360b+43 0:5:count1.txt 5:10:count2.txt\n")
1037
1038     def test_conflict_mod(self):
1039         c1 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt')
1040         c2 = Collection('. 5348b82a029fd9e971a811ce1f71360b+43 0:10:count1.txt')
1041         d = c1.diff(c2)
1042         self.assertEqual(d, [('mod', './count1.txt', c1["count1.txt"], c2["count1.txt"])])
1043         f = c1.open("count1.txt", "w")
1044         f.write("zzzzz")
1045
1046         # c1 changed, so c2 mod will go to a conflict file
1047         c1.apply(d)
1048         self.assertRegexpMatches(c1.portable_manifest_text(), r"\. 95ebc3c7b3b9f1d2c40fec14415d3cb8\+5 5348b82a029fd9e971a811ce1f71360b\+43 0:5:count1\.txt 5:10:count1\.txt~\d\d\d\d\d\d\d\d-\d\d\d\d\d\d~conflict~$")
1049
1050     def test_conflict_add(self):
1051         c1 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count2.txt\n')
1052         c2 = Collection('. 5348b82a029fd9e971a811ce1f71360b+43 0:10:count1.txt\n')
1053         d = c1.diff(c2)
1054         self.assertEqual(d, [('del', './count2.txt', c1["count2.txt"]),
1055                              ('add', './count1.txt', c2["count1.txt"])])
1056         f = c1.open("count1.txt", "w")
1057         f.write("zzzzz")
1058
1059         # c1 added count1.txt, so c2 add will go to a conflict file
1060         c1.apply(d)
1061         self.assertRegexpMatches(c1.portable_manifest_text(), r"\. 95ebc3c7b3b9f1d2c40fec14415d3cb8\+5 5348b82a029fd9e971a811ce1f71360b\+43 0:5:count1\.txt 5:10:count1\.txt~\d\d\d\d\d\d\d\d-\d\d\d\d\d\d~conflict~$")
1062
1063     def test_conflict_del(self):
1064         c1 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt')
1065         c2 = Collection('. 5348b82a029fd9e971a811ce1f71360b+43 0:10:count1.txt')
1066         d = c1.diff(c2)
1067         self.assertEqual(d, [('mod', './count1.txt', c1["count1.txt"], c2["count1.txt"])])
1068         c1.remove("count1.txt")
1069
1070         # c1 deleted, so c2 mod will go to a conflict file
1071         c1.apply(d)
1072         self.assertRegexpMatches(c1.portable_manifest_text(), r"\. 5348b82a029fd9e971a811ce1f71360b\+43 0:10:count1\.txt~\d\d\d\d\d\d\d\d-\d\d\d\d\d\d~conflict~$")
1073
1074     def test_notify(self):
1075         c1 = Collection()
1076         events = []
1077         c1.subscribe(lambda event, collection, name, item: events.append((event, collection, name, item)))
1078         f = c1.open("foo.txt", "w")
1079         self.assertEqual(events[0], (arvados.collection.ADD, c1, "foo.txt", f.arvadosfile))
1080
1081     def test_open_w(self):
1082         c1 = Collection(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n")
1083         self.assertEqual(c1["count1.txt"].size(), 10)
1084         c1.open("count1.txt", "w").close()
1085         self.assertEqual(c1["count1.txt"].size(), 0)
1086
1087
1088 class NewCollectionTestCaseWithServers(run_test_server.TestCaseWithServers):
1089     def test_get_manifest_text_only_committed(self):
1090         c = Collection()
1091         with c.open("count.txt", "w") as f:
1092             # One file committed
1093             with c.open("foo.txt", "w") as foo:
1094                 foo.write("foo")
1095                 foo.flush() # Force block commit
1096             f.write("0123456789")
1097             # Other file not committed. Block not written to keep yet.
1098             self.assertEqual(
1099                 c._get_manifest_text(".",
1100                                      strip=False,
1101                                      normalize=False,
1102                                      only_committed=True),
1103                 '. acbd18db4cc2f85cedef654fccc4a4d8+3 0:0:count.txt 0:3:foo.txt\n')
1104             # And now with the file closed...
1105             f.flush() # Force block commit
1106         self.assertEqual(
1107             c._get_manifest_text(".",
1108                                  strip=False,
1109                                  normalize=False,
1110                                  only_committed=True),
1111             ". 781e5e245d69b566979b86e28d23f2c7+10 acbd18db4cc2f85cedef654fccc4a4d8+3 0:10:count.txt 10:3:foo.txt\n")
1112
1113     def test_only_small_blocks_are_packed_together(self):
1114         c = Collection()
1115         # Write a couple of small files,
1116         f = c.open("count.txt", "w")
1117         f.write("0123456789")
1118         f.close(flush=False)
1119         foo = c.open("foo.txt", "w")
1120         foo.write("foo")
1121         foo.close(flush=False)
1122         # Then, write a big file, it shouldn't be packed with the ones above
1123         big = c.open("bigfile.txt", "w")
1124         big.write("x" * 1024 * 1024 * 33) # 33 MB > KEEP_BLOCK_SIZE/2
1125         big.close(flush=False)
1126         self.assertEqual(
1127             c.manifest_text("."),
1128             '. 2d303c138c118af809f39319e5d507e9+34603008 a8430a058b8fbf408e1931b794dbd6fb+13 0:34603008:bigfile.txt 34603008:10:count.txt 34603018:3:foo.txt\n')
1129
1130
1131 class CollectionCreateUpdateTest(run_test_server.TestCaseWithServers):
1132     MAIN_SERVER = {}
1133     KEEP_SERVER = {}
1134
1135     def create_count_txt(self):
1136         # Create an empty collection, save it to the API server, then write a
1137         # file, but don't save it.
1138
1139         c = Collection()
1140         c.save_new("CollectionCreateUpdateTest", ensure_unique_name=True)
1141         self.assertEqual(c.portable_data_hash(), "d41d8cd98f00b204e9800998ecf8427e+0")
1142         self.assertEqual(c.api_response()["portable_data_hash"], "d41d8cd98f00b204e9800998ecf8427e+0" )
1143
1144         with c.open("count.txt", "w") as f:
1145             f.write("0123456789")
1146
1147         self.assertEqual(c.portable_manifest_text(), ". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n")
1148
1149         return c
1150
1151     def test_create_and_save(self):
1152         c = self.create_count_txt()
1153         c.save()
1154         self.assertRegexpMatches(c.manifest_text(), r"^\. 781e5e245d69b566979b86e28d23f2c7\+10\+A[a-f0-9]{40}@[a-f0-9]{8} 0:10:count\.txt$",)
1155
1156     def test_create_and_save_new(self):
1157         c = self.create_count_txt()
1158         c.save_new()
1159         self.assertRegexpMatches(c.manifest_text(), r"^\. 781e5e245d69b566979b86e28d23f2c7\+10\+A[a-f0-9]{40}@[a-f0-9]{8} 0:10:count\.txt$",)
1160
1161     def test_create_diff_apply(self):
1162         c1 = self.create_count_txt()
1163         c1.save()
1164
1165         c2 = Collection(c1.manifest_locator())
1166         with c2.open("count.txt", "w") as f:
1167             f.write("abcdefg")
1168
1169         diff = c1.diff(c2)
1170
1171         self.assertEqual(diff[0], (arvados.collection.MOD, u'./count.txt', c1["count.txt"], c2["count.txt"]))
1172
1173         c1.apply(diff)
1174         self.assertEqual(c1.portable_data_hash(), c2.portable_data_hash())
1175
1176     def test_diff_apply_with_token(self):
1177         baseline = CollectionReader(". 781e5e245d69b566979b86e28d23f2c7+10+A715fd31f8111894f717eb1003c1b0216799dd9ec@54f5dd1a 0:10:count.txt\n")
1178         c = Collection(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n")
1179         other = CollectionReader(". 7ac66c0f148de9519b8bd264312c4d64+7+A715fd31f8111894f717eb1003c1b0216799dd9ec@54f5dd1a 0:7:count.txt\n")
1180
1181         diff = baseline.diff(other)
1182         self.assertEqual(diff, [('mod', u'./count.txt', c["count.txt"], other["count.txt"])])
1183
1184         c.apply(diff)
1185
1186         self.assertEqual(c.manifest_text(), ". 7ac66c0f148de9519b8bd264312c4d64+7+A715fd31f8111894f717eb1003c1b0216799dd9ec@54f5dd1a 0:7:count.txt\n")
1187
1188
1189     def test_create_and_update(self):
1190         c1 = self.create_count_txt()
1191         c1.save()
1192
1193         c2 = arvados.collection.Collection(c1.manifest_locator())
1194         with c2.open("count.txt", "w") as f:
1195             f.write("abcdefg")
1196
1197         c2.save()
1198
1199         self.assertNotEqual(c1.portable_data_hash(), c2.portable_data_hash())
1200         c1.update()
1201         self.assertEqual(c1.portable_data_hash(), c2.portable_data_hash())
1202
1203
1204     def test_create_and_update_with_conflict(self):
1205         c1 = self.create_count_txt()
1206         c1.save()
1207
1208         with c1.open("count.txt", "w") as f:
1209             f.write("XYZ")
1210
1211         c2 = arvados.collection.Collection(c1.manifest_locator())
1212         with c2.open("count.txt", "w") as f:
1213             f.write("abcdefg")
1214
1215         c2.save()
1216
1217         c1.update()
1218         self.assertRegexpMatches(c1.manifest_text(), r"\. e65075d550f9b5bf9992fa1d71a131be\+3\S* 7ac66c0f148de9519b8bd264312c4d64\+7\S* 0:3:count\.txt 3:7:count\.txt~\d\d\d\d\d\d\d\d-\d\d\d\d\d\d~conflict~$")
1219
1220
1221 if __name__ == '__main__':
1222     unittest.main()