sdk/python/tests/test_collections.py

   1 # usage example:
   2 #
   3 # ARVADOS_API_TOKEN=abc ARVADOS_API_HOST=arvados.local python -m unittest discover
   4
   5 import arvados
   6 import bz2
   7 import copy
   8 import mock
   9 import os
  10 import pprint
  11 import subprocess
  12 import tempfile
  13 import unittest
  14
  15 import run_test_server
  16 import arvados_testutil as tutil
  17
  18 class TestResumableWriter(arvados.ResumableCollectionWriter):
  19     KEEP_BLOCK_SIZE = 1024  # PUT to Keep every 1K.
  20
  21     def current_state(self):
  22         return self.dump_state(copy.deepcopy)
  23
  24
  25 class ArvadosCollectionsTest(run_test_server.TestCaseWithServers,
  26                              tutil.ArvadosBaseTestCase):
  27     MAIN_SERVER = {}
  28
  29     @classmethod
  30     def setUpClass(cls):
  31         super(ArvadosCollectionsTest, cls).setUpClass()
  32         run_test_server.authorize_with('active')
  33         cls.api_client = arvados.api('v1')
  34         cls.keep_client = arvados.KeepClient(api_client=cls.api_client,
  35                                              local_store=cls.local_store)
  36
  37     def write_foo_bar_baz(self):
  38         cw = arvados.CollectionWriter(self.api_client)
  39         self.assertEqual(cw.current_stream_name(), '.',
  40                          'current_stream_name() should be "." now')
  41         cw.set_current_file_name('foo.txt')
  42         cw.write('foo')
  43         self.assertEqual(cw.current_file_name(), 'foo.txt',
  44                          'current_file_name() should be foo.txt now')
  45         cw.start_new_file('bar.txt')
  46         cw.write('bar')
  47         cw.start_new_stream('baz')
  48         cw.write('baz')
  49         cw.set_current_file_name('baz.txt')
  50         self.assertEqual(cw.manifest_text(),
  51                          ". 3858f62230ac3c915f300c664312c63f+6 0:3:foo.txt 3:3:bar.txt\n" +
  52                          "./baz 73feffa4b7f6bb68e44cf984c85f6e88+3 0:3:baz.txt\n",
  53                          "wrong manifest: got {}".format(cw.manifest_text()))
  54         cw.finish()
  55         return cw.portable_data_hash()
  56
  57     def test_keep_local_store(self):
  58         self.assertEqual(self.keep_client.put('foo'), 'acbd18db4cc2f85cedef654fccc4a4d8+3', 'wrong md5 hash from Keep.put')
  59         self.assertEqual(self.keep_client.get('acbd18db4cc2f85cedef654fccc4a4d8+3'), 'foo', 'wrong data from Keep.get')
  60
  61     def test_local_collection_writer(self):
  62         self.assertEqual(self.write_foo_bar_baz(),
  63                          '23ca013983d6239e98931cc779e68426+114',
  64                          'wrong locator hash: ' + self.write_foo_bar_baz())
  65
  66     def test_local_collection_reader(self):
  67         foobarbaz = self.write_foo_bar_baz()
  68         cr = arvados.CollectionReader(
  69             foobarbaz + '+Xzizzle', self.api_client)
  70         got = []
  71         for s in cr.all_streams():
  72             for f in s.all_files():
  73                 got += [[f.size(), f.stream_name(), f.name(), f.read(2**26)]]
  74         expected = [[3, '.', 'foo.txt', 'foo'],
  75                     [3, '.', 'bar.txt', 'bar'],
  76                     [3, './baz', 'baz.txt', 'baz']]
  77         self.assertEqual(got,
  78                          expected)
  79         stream0 = cr.all_streams()[0]
  80         self.assertEqual(stream0.readfrom(0, 0),
  81                          '',
  82                          'reading zero bytes should have returned empty string')
  83         self.assertEqual(stream0.readfrom(0, 2**26),
  84                          'foobar',
  85                          'reading entire stream failed')
  86         self.assertEqual(stream0.readfrom(2**26, 0),
  87                          '',
  88                          'reading zero bytes should have returned empty string')
  89
  90     def _test_subset(self, collection, expected):
  91         cr = arvados.CollectionReader(collection, self.api_client)
  92         for s in cr.all_streams():
  93             for ex in expected:
  94                 if ex[0] == s:
  95                     f = s.files()[ex[2]]
  96                     got = [f.size(), f.stream_name(), f.name(), "".join(f.readall(2**26))]
  97                     self.assertEqual(got,
  98                                      ex,
  99                                      'all_files|as_manifest did not preserve manifest contents: got %s expected %s' % (got, ex))
 100
 101     def test_collection_manifest_subset(self):
 102         foobarbaz = self.write_foo_bar_baz()
 103         self._test_subset(foobarbaz,
 104                           [[3, '.',     'bar.txt', 'bar'],
 105                            [3, '.',     'foo.txt', 'foo'],
 106                            [3, './baz', 'baz.txt', 'baz']])
 107         self._test_subset((". %s %s 0:3:foo.txt 3:3:bar.txt\n" %
 108                            (self.keep_client.put("foo"),
 109                             self.keep_client.put("bar"))),
 110                           [[3, '.', 'bar.txt', 'bar'],
 111                            [3, '.', 'foo.txt', 'foo']])
 112         self._test_subset((". %s %s 0:2:fo.txt 2:4:obar.txt\n" %
 113                            (self.keep_client.put("foo"),
 114                             self.keep_client.put("bar"))),
 115                           [[2, '.', 'fo.txt', 'fo'],
 116                            [4, '.', 'obar.txt', 'obar']])
 117         self._test_subset((". %s %s 0:2:fo.txt 2:0:zero.txt 2:2:ob.txt 4:2:ar.txt\n" %
 118                            (self.keep_client.put("foo"),
 119                             self.keep_client.put("bar"))),
 120                           [[2, '.', 'ar.txt', 'ar'],
 121                            [2, '.', 'fo.txt', 'fo'],
 122                            [2, '.', 'ob.txt', 'ob'],
 123                            [0, '.', 'zero.txt', '']])
 124
 125     def _test_readline(self, what_in, what_out):
 126         cw = arvados.CollectionWriter(self.api_client)
 127         cw.start_new_file('test.txt')
 128         cw.write(what_in)
 129         test1 = cw.finish()
 130         cr = arvados.CollectionReader(test1, self.api_client)
 131         got = []
 132         for x in list(cr.all_files())[0].readlines():
 133             got += [x]
 134         self.assertEqual(got,
 135                          what_out,
 136                          "readlines did not split lines correctly: %s" % got)
 137
 138     def test_collection_readline(self):
 139         self._test_readline("\na\nbcd\n\nefg\nz",
 140                             ["\n", "a\n", "bcd\n", "\n", "efg\n", "z"])
 141         self._test_readline("ab\ncd\n",
 142                             ["ab\n", "cd\n"])
 143
 144     def test_collection_empty_file(self):
 145         cw = arvados.CollectionWriter(self.api_client)
 146         cw.start_new_file('zero.txt')
 147         cw.write('')
 148
 149         self.assertEqual(cw.manifest_text(), ". d41d8cd98f00b204e9800998ecf8427e+0 0:0:zero.txt\n")
 150         self.check_manifest_file_sizes(cw.manifest_text(), [0])
 151         cw = arvados.CollectionWriter(self.api_client)
 152         cw.start_new_file('zero.txt')
 153         cw.write('')
 154         cw.start_new_file('one.txt')
 155         cw.write('1')
 156         cw.start_new_stream('foo')
 157         cw.start_new_file('zero.txt')
 158         cw.write('')
 159         self.check_manifest_file_sizes(cw.manifest_text(), [0,1,0])
 160
 161     def test_no_implicit_normalize(self):
 162         cw = arvados.CollectionWriter(self.api_client)
 163         cw.start_new_file('b')
 164         cw.write('b')
 165         cw.start_new_file('a')
 166         cw.write('')
 167         self.check_manifest_file_sizes(cw.manifest_text(), [1,0])
 168         self.check_manifest_file_sizes(
 169             arvados.CollectionReader(
 170                 cw.manifest_text()).manifest_text(normalize=True),
 171             [0,1])
 172
 173     def check_manifest_file_sizes(self, manifest_text, expect_sizes):
 174         cr = arvados.CollectionReader(manifest_text, self.api_client)
 175         got_sizes = []
 176         for f in cr.all_files():
 177             got_sizes += [f.size()]
 178         self.assertEqual(got_sizes, expect_sizes, "got wrong file sizes %s, expected %s" % (got_sizes, expect_sizes))
 179
 180     def test_collection_bz2_decompression(self):
 181         n_lines_in = 2**18
 182         data_in = "abc\n"
 183         for x in xrange(0, 18):
 184             data_in += data_in
 185         compressed_data_in = bz2.compress(data_in)
 186         cw = arvados.CollectionWriter(self.api_client)
 187         cw.start_new_file('test.bz2')
 188         cw.write(compressed_data_in)
 189         bz2_manifest = cw.manifest_text()
 190
 191         cr = arvados.CollectionReader(bz2_manifest, self.api_client)
 192
 193         got = 0
 194         for x in list(cr.all_files())[0].readlines():
 195             self.assertEqual(x, "abc\n", "decompression returned wrong data: %s" % x)
 196             got += 1
 197         self.assertEqual(got,
 198                          n_lines_in,
 199                          "decompression returned %d lines instead of %d" % (got, n_lines_in))
 200
 201     def test_collection_gzip_decompression(self):
 202         n_lines_in = 2**18
 203         data_in = "abc\n"
 204         for x in xrange(0, 18):
 205             data_in += data_in
 206         p = subprocess.Popen(["gzip", "-1cn"],
 207                              stdout=subprocess.PIPE,
 208                              stdin=subprocess.PIPE,
 209                              stderr=subprocess.PIPE,
 210                              shell=False, close_fds=True)
 211         compressed_data_in, stderrdata = p.communicate(data_in)
 212
 213         cw = arvados.CollectionWriter(self.api_client)
 214         cw.start_new_file('test.gz')
 215         cw.write(compressed_data_in)
 216         gzip_manifest = cw.manifest_text()
 217
 218         cr = arvados.CollectionReader(gzip_manifest, self.api_client)
 219         got = 0
 220         for x in list(cr.all_files())[0].readlines():
 221             self.assertEqual(x, "abc\n", "decompression returned wrong data: %s" % x)
 222             got += 1
 223         self.assertEqual(got,
 224                          n_lines_in,
 225                          "decompression returned %d lines instead of %d" % (got, n_lines_in))
 226
 227     def test_normalized_collection(self):
 228         m1 = """. 5348b82a029fd9e971a811ce1f71360b+43 0:43:md5sum.txt
 229 . 085c37f02916da1cad16f93c54d899b7+41 0:41:md5sum.txt
 230 . 8b22da26f9f433dea0a10e5ec66d73ba+43 0:43:md5sum.txt
 231 """
 232         self.assertEqual(arvados.CollectionReader(m1, self.api_client).manifest_text(normalize=True),
 233                          """. 5348b82a029fd9e971a811ce1f71360b+43 085c37f02916da1cad16f93c54d899b7+41 8b22da26f9f433dea0a10e5ec66d73ba+43 0:127:md5sum.txt
 234 """)
 235
 236         m2 = """. 204e43b8a1185621ca55a94839582e6f+67108864 b9677abbac956bd3e86b1deb28dfac03+67108864 fc15aff2a762b13f521baf042140acec+67108864 323d2a3ce20370c4ca1d3462a344f8fd+25885655 0:227212247:var-GS000016015-ASM.tsv.bz2
 237 """
 238         self.assertEqual(arvados.CollectionReader(m2, self.api_client).manifest_text(normalize=True), m2)
 239
 240         m3 = """. 5348b82a029fd9e971a811ce1f71360b+43 3:40:md5sum.txt
 241 . 085c37f02916da1cad16f93c54d899b7+41 0:41:md5sum.txt
 242 . 8b22da26f9f433dea0a10e5ec66d73ba+43 0:43:md5sum.txt
 243 """
 244         self.assertEqual(arvados.CollectionReader(m3, self.api_client).manifest_text(normalize=True),
 245                          """. 5348b82a029fd9e971a811ce1f71360b+43 085c37f02916da1cad16f93c54d899b7+41 8b22da26f9f433dea0a10e5ec66d73ba+43 3:124:md5sum.txt
 246 """)
 247
 248         m4 = """. 204e43b8a1185621ca55a94839582e6f+67108864 0:3:foo/bar
 249 ./zzz 204e43b8a1185621ca55a94839582e6f+67108864 0:999:zzz
 250 ./foo 323d2a3ce20370c4ca1d3462a344f8fd+25885655 0:3:bar
 251 """
 252         self.assertEqual(arvados.CollectionReader(m4, self.api_client).manifest_text(normalize=True),
 253                          """./foo 204e43b8a1185621ca55a94839582e6f+67108864 323d2a3ce20370c4ca1d3462a344f8fd+25885655 0:3:bar 67108864:3:bar
 254 ./zzz 204e43b8a1185621ca55a94839582e6f+67108864 0:999:zzz
 255 """)
 256
 257         m5 = """. 204e43b8a1185621ca55a94839582e6f+67108864 0:3:foo/bar
 258 ./zzz 204e43b8a1185621ca55a94839582e6f+67108864 0:999:zzz
 259 ./foo 204e43b8a1185621ca55a94839582e6f+67108864 3:3:bar
 260 """
 261         self.assertEqual(arvados.CollectionReader(m5, self.api_client).manifest_text(normalize=True),
 262                          """./foo 204e43b8a1185621ca55a94839582e6f+67108864 0:6:bar
 263 ./zzz 204e43b8a1185621ca55a94839582e6f+67108864 0:999:zzz
 264 """)
 265
 266         with self.data_file('1000G_ref_manifest') as f6:
 267             m6 = f6.read()
 268             self.assertEqual(arvados.CollectionReader(m6, self.api_client).manifest_text(normalize=True), m6)
 269
 270         with self.data_file('jlake_manifest') as f7:
 271             m7 = f7.read()
 272             self.assertEqual(arvados.CollectionReader(m7, self.api_client).manifest_text(normalize=True), m7)
 273
 274         m8 = """./a\\040b\\040c 59ca0efa9f5633cb0371bbc0355478d8+13 0:13:hello\\040world.txt
 275 """
 276         self.assertEqual(arvados.CollectionReader(m8, self.api_client).manifest_text(normalize=True), m8)
 277
 278     def test_locators_and_ranges(self):
 279         blocks2 = [['a', 10, 0],
 280                   ['b', 10, 10],
 281                   ['c', 10, 20],
 282                   ['d', 10, 30],
 283                   ['e', 10, 40],
 284                   ['f', 10, 50]]
 285
 286         self.assertEqual(arvados.locators_and_ranges(blocks2,  2,  2), [['a', 10, 2, 2]])
 287         self.assertEqual(arvados.locators_and_ranges(blocks2, 12, 2), [['b', 10, 2, 2]])
 288         self.assertEqual(arvados.locators_and_ranges(blocks2, 22, 2), [['c', 10, 2, 2]])
 289         self.assertEqual(arvados.locators_and_ranges(blocks2, 32, 2), [['d', 10, 2, 2]])
 290         self.assertEqual(arvados.locators_and_ranges(blocks2, 42, 2), [['e', 10, 2, 2]])
 291         self.assertEqual(arvados.locators_and_ranges(blocks2, 52, 2), [['f', 10, 2, 2]])
 292         self.assertEqual(arvados.locators_and_ranges(blocks2, 62, 2), [])
 293         self.assertEqual(arvados.locators_and_ranges(blocks2, -2, 2), [])
 294
 295         self.assertEqual(arvados.locators_and_ranges(blocks2,  0,  2), [['a', 10, 0, 2]])
 296         self.assertEqual(arvados.locators_and_ranges(blocks2, 10, 2), [['b', 10, 0, 2]])
 297         self.assertEqual(arvados.locators_and_ranges(blocks2, 20, 2), [['c', 10, 0, 2]])
 298         self.assertEqual(arvados.locators_and_ranges(blocks2, 30, 2), [['d', 10, 0, 2]])
 299         self.assertEqual(arvados.locators_and_ranges(blocks2, 40, 2), [['e', 10, 0, 2]])
 300         self.assertEqual(arvados.locators_and_ranges(blocks2, 50, 2), [['f', 10, 0, 2]])
 301         self.assertEqual(arvados.locators_and_ranges(blocks2, 60, 2), [])
 302         self.assertEqual(arvados.locators_and_ranges(blocks2, -2, 2), [])
 303
 304         self.assertEqual(arvados.locators_and_ranges(blocks2,  9,  2), [['a', 10, 9, 1], ['b', 10, 0, 1]])
 305         self.assertEqual(arvados.locators_and_ranges(blocks2, 19, 2), [['b', 10, 9, 1], ['c', 10, 0, 1]])
 306         self.assertEqual(arvados.locators_and_ranges(blocks2, 29, 2), [['c', 10, 9, 1], ['d', 10, 0, 1]])
 307         self.assertEqual(arvados.locators_and_ranges(blocks2, 39, 2), [['d', 10, 9, 1], ['e', 10, 0, 1]])
 308         self.assertEqual(arvados.locators_and_ranges(blocks2, 49, 2), [['e', 10, 9, 1], ['f', 10, 0, 1]])
 309         self.assertEqual(arvados.locators_and_ranges(blocks2, 59, 2), [['f', 10, 9, 1]])
 310
 311
 312         blocks3 = [['a', 10, 0],
 313                   ['b', 10, 10],
 314                   ['c', 10, 20],
 315                   ['d', 10, 30],
 316                   ['e', 10, 40],
 317                   ['f', 10, 50],
 318                   ['g', 10, 60]]
 319
 320         self.assertEqual(arvados.locators_and_ranges(blocks3,  2,  2), [['a', 10, 2, 2]])
 321         self.assertEqual(arvados.locators_and_ranges(blocks3, 12, 2), [['b', 10, 2, 2]])
 322         self.assertEqual(arvados.locators_and_ranges(blocks3, 22, 2), [['c', 10, 2, 2]])
 323         self.assertEqual(arvados.locators_and_ranges(blocks3, 32, 2), [['d', 10, 2, 2]])
 324         self.assertEqual(arvados.locators_and_ranges(blocks3, 42, 2), [['e', 10, 2, 2]])
 325         self.assertEqual(arvados.locators_and_ranges(blocks3, 52, 2), [['f', 10, 2, 2]])
 326         self.assertEqual(arvados.locators_and_ranges(blocks3, 62, 2), [['g', 10, 2, 2]])
 327
 328
 329         blocks = [['a', 10, 0],
 330                   ['b', 15, 10],
 331                   ['c', 5, 25]]
 332         self.assertEqual(arvados.locators_and_ranges(blocks, 1, 0), [])
 333         self.assertEqual(arvados.locators_and_ranges(blocks, 0, 5), [['a', 10, 0, 5]])
 334         self.assertEqual(arvados.locators_and_ranges(blocks, 3, 5), [['a', 10, 3, 5]])
 335         self.assertEqual(arvados.locators_and_ranges(blocks, 0, 10), [['a', 10, 0, 10]])
 336
 337         self.assertEqual(arvados.locators_and_ranges(blocks, 0, 11), [['a', 10, 0, 10],
 338                                                                       ['b', 15, 0, 1]])
 339         self.assertEqual(arvados.locators_and_ranges(blocks, 1, 11), [['a', 10, 1, 9],
 340                                                                       ['b', 15, 0, 2]])
 341         self.assertEqual(arvados.locators_and_ranges(blocks, 0, 25), [['a', 10, 0, 10],
 342                                                                       ['b', 15, 0, 15]])
 343
 344         self.assertEqual(arvados.locators_and_ranges(blocks, 0, 30), [['a', 10, 0, 10],
 345                                                                       ['b', 15, 0, 15],
 346                                                                       ['c', 5, 0, 5]])
 347         self.assertEqual(arvados.locators_and_ranges(blocks, 1, 30), [['a', 10, 1, 9],
 348                                                                       ['b', 15, 0, 15],
 349                                                                       ['c', 5, 0, 5]])
 350         self.assertEqual(arvados.locators_and_ranges(blocks, 0, 31), [['a', 10, 0, 10],
 351                                                                       ['b', 15, 0, 15],
 352                                                                       ['c', 5, 0, 5]])
 353
 354         self.assertEqual(arvados.locators_and_ranges(blocks, 15, 5), [['b', 15, 5, 5]])
 355
 356         self.assertEqual(arvados.locators_and_ranges(blocks, 8, 17), [['a', 10, 8, 2],
 357                                                                       ['b', 15, 0, 15]])
 358
 359         self.assertEqual(arvados.locators_and_ranges(blocks, 8, 20), [['a', 10, 8, 2],
 360                                                                       ['b', 15, 0, 15],
 361                                                                       ['c', 5, 0, 3]])
 362
 363         self.assertEqual(arvados.locators_and_ranges(blocks, 26, 2), [['c', 5, 1, 2]])
 364
 365         self.assertEqual(arvados.locators_and_ranges(blocks, 9, 15), [['a', 10, 9, 1],
 366                                                                       ['b', 15, 0, 14]])
 367         self.assertEqual(arvados.locators_and_ranges(blocks, 10, 15), [['b', 15, 0, 15]])
 368         self.assertEqual(arvados.locators_and_ranges(blocks, 11, 15), [['b', 15, 1, 14],
 369                                                                        ['c', 5, 0, 1]])
 370
 371     class MockStreamReader(object):
 372         def __init__(self, content):
 373             self.content = content
 374             self.num_retries = 0
 375
 376         def readfrom(self, start, size, num_retries=0):
 377             return self.content[start:start+size]
 378
 379     def test_file_stream(self):
 380         content = 'abcdefghijklmnopqrstuvwxyz0123456789'
 381         msr = self.MockStreamReader(content)
 382         segments = [[0, 10, 0],
 383                     [10, 15, 10],
 384                     [25, 5, 25]]
 385
 386         sfr = arvados.StreamFileReader(msr, segments, "test")
 387
 388         self.assertEqual(sfr.name(), "test")
 389         self.assertEqual(sfr.size(), 30)
 390
 391         self.assertEqual(sfr.readfrom(0, 30), content[0:30])
 392         self.assertEqual(sfr.readfrom(2, 30), content[2:30])
 393
 394         self.assertEqual(sfr.readfrom(2, 8), content[2:10])
 395         self.assertEqual(sfr.readfrom(0, 10), content[0:10])
 396
 397         self.assertEqual(sfr.tell(), 0)
 398         self.assertEqual(sfr.read(5), content[0:5])
 399         self.assertEqual(sfr.tell(), 5)
 400         self.assertEqual(sfr.read(5), content[5:10])
 401         self.assertEqual(sfr.tell(), 10)
 402         self.assertEqual(sfr.read(5), content[10:15])
 403         self.assertEqual(sfr.tell(), 15)
 404         self.assertEqual(sfr.read(5), content[15:20])
 405         self.assertEqual(sfr.tell(), 20)
 406         self.assertEqual(sfr.read(5), content[20:25])
 407         self.assertEqual(sfr.tell(), 25)
 408         self.assertEqual(sfr.read(5), content[25:30])
 409         self.assertEqual(sfr.tell(), 30)
 410         self.assertEqual(sfr.read(5), '')
 411         self.assertEqual(sfr.tell(), 30)
 412
 413         segments = [[26, 10, 0],
 414                     [0, 15, 10],
 415                     [15, 5, 25]]
 416
 417         sfr = arvados.StreamFileReader(msr, segments, "test")
 418
 419         self.assertEqual(sfr.size(), 30)
 420
 421         self.assertEqual(sfr.readfrom(0, 30), content[26:36] + content[0:20])
 422         self.assertEqual(sfr.readfrom(2, 30), content[28:36] + content[0:20])
 423
 424         self.assertEqual(sfr.readfrom(2, 8), content[28:36])
 425         self.assertEqual(sfr.readfrom(0, 10), content[26:36])
 426
 427         self.assertEqual(sfr.tell(), 0)
 428         self.assertEqual(sfr.read(5), content[26:31])
 429         self.assertEqual(sfr.tell(), 5)
 430         self.assertEqual(sfr.read(5), content[31:36])
 431         self.assertEqual(sfr.tell(), 10)
 432         self.assertEqual(sfr.read(5), content[0:5])
 433         self.assertEqual(sfr.tell(), 15)
 434         self.assertEqual(sfr.read(5), content[5:10])
 435         self.assertEqual(sfr.tell(), 20)
 436         self.assertEqual(sfr.read(5), content[10:15])
 437         self.assertEqual(sfr.tell(), 25)
 438         self.assertEqual(sfr.read(5), content[15:20])
 439         self.assertEqual(sfr.tell(), 30)
 440         self.assertEqual(sfr.read(5), '')
 441         self.assertEqual(sfr.tell(), 30)
 442
 443
 444     class MockKeep(object):
 445         def __init__(self, content, num_retries=0):
 446             self.content = content
 447
 448         def get(self, locator, num_retries=0):
 449             return self.content[locator]
 450
 451     def test_stream_reader(self):
 452         keepblocks = {'aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa+10': 'abcdefghij',
 453                       'bbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbb+15': 'klmnopqrstuvwxy',
 454                       'cccccccccccccccccccccccccccccccc+5': 'z0123'}
 455         mk = self.MockKeep(keepblocks)
 456
 457         sr = arvados.StreamReader([".", "aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa+10", "bbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbb+15", "cccccccccccccccccccccccccccccccc+5", "0:30:foo"], mk)
 458
 459         content = 'abcdefghijklmnopqrstuvwxyz0123456789'
 460
 461         self.assertEqual(sr.readfrom(0, 30), content[0:30])
 462         self.assertEqual(sr.readfrom(2, 30), content[2:30])
 463
 464         self.assertEqual(sr.readfrom(2, 8), content[2:10])
 465         self.assertEqual(sr.readfrom(0, 10), content[0:10])
 466
 467         self.assertEqual(sr.readfrom(0, 5), content[0:5])
 468         self.assertEqual(sr.readfrom(5, 5), content[5:10])
 469         self.assertEqual(sr.readfrom(10, 5), content[10:15])
 470         self.assertEqual(sr.readfrom(15, 5), content[15:20])
 471         self.assertEqual(sr.readfrom(20, 5), content[20:25])
 472         self.assertEqual(sr.readfrom(25, 5), content[25:30])
 473         self.assertEqual(sr.readfrom(30, 5), '')
 474
 475     def test_file_reader(self):
 476         keepblocks = {'aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa+10': 'abcdefghij',
 477                       'bbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbb+15': 'klmnopqrstuvwxy',
 478                       'cccccccccccccccccccccccccccccccc+5': 'z0123'}
 479         mk = self.MockKeep(keepblocks)
 480
 481         sr = arvados.StreamReader([".", "aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa+10", "bbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbb+15", "cccccccccccccccccccccccccccccccc+5", "0:10:foo", "15:10:foo"], mk)
 482
 483         content = 'abcdefghijpqrstuvwxy'
 484
 485         f = sr.files()["foo"]
 486
 487         # f.read() calls will be aligned on block boundaries (as a
 488         # result of ticket #3663).
 489
 490         f.seek(0)
 491         self.assertEqual(f.read(20), content[0:10])
 492
 493         f.seek(0)
 494         self.assertEqual(f.read(6), content[0:6])
 495         self.assertEqual(f.read(6), content[6:10])
 496         self.assertEqual(f.read(6), content[10:16])
 497         self.assertEqual(f.read(6), content[16:20])
 498
 499     def test_extract_file(self):
 500         m1 = """. 5348b82a029fd9e971a811ce1f71360b+43 0:43:md5sum.txt
 501 . 085c37f02916da1cad16f93c54d899b7+41 0:41:md6sum.txt
 502 . 8b22da26f9f433dea0a10e5ec66d73ba+43 0:43:md7sum.txt
 503 . 085c37f02916da1cad16f93c54d899b7+41 5348b82a029fd9e971a811ce1f71360b+43 8b22da26f9f433dea0a10e5ec66d73ba+43 47:80:md8sum.txt
 504 . 085c37f02916da1cad16f93c54d899b7+41 5348b82a029fd9e971a811ce1f71360b+43 8b22da26f9f433dea0a10e5ec66d73ba+43 40:80:md9sum.txt
 505 """
 506
 507         m2 = arvados.CollectionReader(m1, self.api_client).manifest_text(normalize=True)
 508
 509         self.assertEqual(m2,
 510                          ". 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")
 511         files = arvados.CollectionReader(
 512             m2, self.api_client).all_streams()[0].files()
 513
 514         self.assertEqual(files['md5sum.txt'].as_manifest(),
 515                          ". 5348b82a029fd9e971a811ce1f71360b+43 0:43:md5sum.txt\n")
 516         self.assertEqual(files['md6sum.txt'].as_manifest(),
 517                          ". 085c37f02916da1cad16f93c54d899b7+41 0:41:md6sum.txt\n")
 518         self.assertEqual(files['md7sum.txt'].as_manifest(),
 519                          ". 8b22da26f9f433dea0a10e5ec66d73ba+43 0:43:md7sum.txt\n")
 520         self.assertEqual(files['md9sum.txt'].as_manifest(),
 521                          ". 085c37f02916da1cad16f93c54d899b7+41 5348b82a029fd9e971a811ce1f71360b+43 8b22da26f9f433dea0a10e5ec66d73ba+43 40:80:md9sum.txt\n")
 522
 523     def test_write_directory_tree(self):
 524         cwriter = arvados.CollectionWriter(self.api_client)
 525         cwriter.write_directory_tree(self.build_directory_tree(
 526                 ['basefile', 'subdir/subfile']))
 527         self.assertEqual(cwriter.manifest_text(),
 528                          """. c5110c5ac93202d8e0f9e381f22bac0f+8 0:8:basefile
 529 ./subdir 1ca4dec89403084bf282ad31e6cf7972+14 0:14:subfile\n""")
 530
 531     def test_write_named_directory_tree(self):
 532         cwriter = arvados.CollectionWriter(self.api_client)
 533         cwriter.write_directory_tree(self.build_directory_tree(
 534                 ['basefile', 'subdir/subfile']), 'root')
 535         self.assertEqual(
 536             cwriter.manifest_text(),
 537             """./root c5110c5ac93202d8e0f9e381f22bac0f+8 0:8:basefile
 538 ./root/subdir 1ca4dec89403084bf282ad31e6cf7972+14 0:14:subfile\n""")
 539
 540     def test_write_directory_tree_in_one_stream(self):
 541         cwriter = arvados.CollectionWriter(self.api_client)
 542         cwriter.write_directory_tree(self.build_directory_tree(
 543                 ['basefile', 'subdir/subfile']), max_manifest_depth=0)
 544         self.assertEqual(cwriter.manifest_text(),
 545                          """. 4ace875ffdc6824a04950f06858f4465+22 0:8:basefile 8:14:subdir/subfile\n""")
 546
 547     def test_write_directory_tree_with_limited_recursion(self):
 548         cwriter = arvados.CollectionWriter(self.api_client)
 549         cwriter.write_directory_tree(
 550             self.build_directory_tree(['f1', 'd1/f2', 'd1/d2/f3']),
 551             max_manifest_depth=1)
 552         self.assertEqual(cwriter.manifest_text(),
 553                          """. bd19836ddb62c11c55ab251ccaca5645+2 0:2:f1
 554 ./d1 50170217e5b04312024aa5cd42934494+13 0:8:d2/f3 8:5:f2\n""")
 555
 556     def test_write_one_file(self):
 557         cwriter = arvados.CollectionWriter(self.api_client)
 558         with self.make_test_file() as testfile:
 559             cwriter.write_file(testfile.name)
 560             self.assertEqual(
 561                 cwriter.manifest_text(),
 562                 ". 098f6bcd4621d373cade4e832627b4f6+4 0:4:{}\n".format(
 563                     os.path.basename(testfile.name)))
 564
 565     def test_write_named_file(self):
 566         cwriter = arvados.CollectionWriter(self.api_client)
 567         with self.make_test_file() as testfile:
 568             cwriter.write_file(testfile.name, 'foo')
 569             self.assertEqual(cwriter.manifest_text(),
 570                              ". 098f6bcd4621d373cade4e832627b4f6+4 0:4:foo\n")
 571
 572     def test_write_multiple_files(self):
 573         cwriter = arvados.CollectionWriter(self.api_client)
 574         for letter in 'ABC':
 575             with self.make_test_file(letter) as testfile:
 576                 cwriter.write_file(testfile.name, letter)
 577         self.assertEqual(
 578             cwriter.manifest_text(),
 579             ". 902fbdd2b1df0c4f70b4a5d23525e932+3 0:1:A 1:1:B 2:1:C\n")
 580
 581     def test_basic_resume(self):
 582         cwriter = TestResumableWriter()
 583         with self.make_test_file() as testfile:
 584             cwriter.write_file(testfile.name, 'test')
 585             resumed = TestResumableWriter.from_state(cwriter.current_state())
 586         self.assertEquals(cwriter.manifest_text(), resumed.manifest_text(),
 587                           "resumed CollectionWriter had different manifest")
 588
 589     def test_resume_fails_when_missing_dependency(self):
 590         cwriter = TestResumableWriter()
 591         with self.make_test_file() as testfile:
 592             cwriter.write_file(testfile.name, 'test')
 593         self.assertRaises(arvados.errors.StaleWriterStateError,
 594                           TestResumableWriter.from_state,
 595                           cwriter.current_state())
 596
 597     def test_resume_fails_when_dependency_mtime_changed(self):
 598         cwriter = TestResumableWriter()
 599         with self.make_test_file() as testfile:
 600             cwriter.write_file(testfile.name, 'test')
 601             os.utime(testfile.name, (0, 0))
 602             self.assertRaises(arvados.errors.StaleWriterStateError,
 603                               TestResumableWriter.from_state,
 604                               cwriter.current_state())
 605
 606     def test_resume_fails_when_dependency_is_nonfile(self):
 607         cwriter = TestResumableWriter()
 608         cwriter.write_file('/dev/null', 'empty')
 609         self.assertRaises(arvados.errors.StaleWriterStateError,
 610                           TestResumableWriter.from_state,
 611                           cwriter.current_state())
 612
 613     def test_resume_fails_when_dependency_size_changed(self):
 614         cwriter = TestResumableWriter()
 615         with self.make_test_file() as testfile:
 616             cwriter.write_file(testfile.name, 'test')
 617             orig_mtime = os.fstat(testfile.fileno()).st_mtime
 618             testfile.write('extra')
 619             testfile.flush()
 620             os.utime(testfile.name, (orig_mtime, orig_mtime))
 621             self.assertRaises(arvados.errors.StaleWriterStateError,
 622                               TestResumableWriter.from_state,
 623                               cwriter.current_state())
 624
 625     def test_resume_fails_with_expired_locator(self):
 626         cwriter = TestResumableWriter()
 627         state = cwriter.current_state()
 628         # Add an expired locator to the state.
 629         state['_current_stream_locators'].append(''.join([
 630                     'a' * 32, '+1+A', 'b' * 40, '@', '10000000']))
 631         self.assertRaises(arvados.errors.StaleWriterStateError,
 632                           TestResumableWriter.from_state, state)
 633
 634     def test_arbitrary_objects_not_resumable(self):
 635         cwriter = TestResumableWriter()
 636         with open('/dev/null') as badfile:
 637             self.assertRaises(arvados.errors.AssertionError,
 638                               cwriter.write_file, badfile)
 639
 640     def test_arbitrary_writes_not_resumable(self):
 641         cwriter = TestResumableWriter()
 642         self.assertRaises(arvados.errors.AssertionError,
 643                           cwriter.write, "badtext")
 644
 645     def test_read_arbitrary_data_with_collection_reader(self):
 646         # arv-get relies on this to do "arv-get {keep-locator} -".
 647         self.write_foo_bar_baz()
 648         self.assertEqual(
 649             'foobar',
 650             arvados.CollectionReader(
 651                 '3858f62230ac3c915f300c664312c63f+6'
 652                 ).manifest_text())
 653
 654
 655 class CollectionTestMixin(object):
 656     PROXY_RESPONSE = {
 657         'items_available': 1,
 658         'items': [{
 659                 'uuid': 'zzzzz-bi6l4-mockproxy012345',
 660                 'owner_uuid': 'zzzzz-tpzed-mockowner012345',
 661                 'service_host': tutil.TEST_HOST,
 662                 'service_port': 65535,
 663                 'service_ssl_flag': True,
 664                 'service_type': 'proxy',
 665                 }]}
 666     API_COLLECTIONS = run_test_server.fixture('collections')
 667     DEFAULT_COLLECTION = API_COLLECTIONS['foo_file']
 668     DEFAULT_DATA_HASH = DEFAULT_COLLECTION['portable_data_hash']
 669     DEFAULT_MANIFEST = DEFAULT_COLLECTION['manifest_text']
 670     DEFAULT_UUID = DEFAULT_COLLECTION['uuid']
 671     ALT_COLLECTION = API_COLLECTIONS['bar_file']
 672     ALT_DATA_HASH = ALT_COLLECTION['portable_data_hash']
 673     ALT_MANIFEST = ALT_COLLECTION['manifest_text']
 674
 675     def _mock_api_call(self, mock_method, code, body):
 676         mock_method = mock_method().execute
 677         if code == 200:
 678             mock_method.return_value = body
 679         else:
 680             mock_method.side_effect = arvados.errors.ApiError(
 681                 tutil.fake_httplib2_response(code), "{}")
 682
 683     def mock_keep_services(self, api_mock, code, body):
 684         self._mock_api_call(api_mock.keep_services().accessible, code, body)
 685
 686     def api_client_mock(self, code=200):
 687         client = mock.MagicMock(name='api_client')
 688         self.mock_keep_services(client, code, self.PROXY_RESPONSE)
 689         return client
 690
 691
 692 @tutil.skip_sleep
 693 class CollectionReaderTestCase(unittest.TestCase, CollectionTestMixin):
 694     def mock_get_collection(self, api_mock, code, body):
 695         body = self.API_COLLECTIONS.get(body)
 696         self._mock_api_call(api_mock.collections().get, code, body)
 697
 698     def api_client_mock(self, code=200):
 699         client = super(CollectionReaderTestCase, self).api_client_mock(code)
 700         self.mock_get_collection(client, code, 'foo_file')
 701         return client
 702
 703     def test_init_no_default_retries(self):
 704         client = self.api_client_mock(200)
 705         reader = arvados.CollectionReader(self.DEFAULT_UUID, api_client=client)
 706         reader.manifest_text()
 707         client.collections().get().execute.assert_called_with(num_retries=0)
 708
 709     def test_uuid_init_success(self):
 710         client = self.api_client_mock(200)
 711         reader = arvados.CollectionReader(self.DEFAULT_UUID, api_client=client,
 712                                           num_retries=3)
 713         self.assertEqual(self.DEFAULT_COLLECTION['manifest_text'],
 714                          reader.manifest_text())
 715         client.collections().get().execute.assert_called_with(num_retries=3)
 716
 717     def test_uuid_init_failure_raises_api_error(self):
 718         client = self.api_client_mock(500)
 719         reader = arvados.CollectionReader(self.DEFAULT_UUID, api_client=client)
 720         with self.assertRaises(arvados.errors.ApiError):
 721             reader.manifest_text()
 722
 723     def test_locator_init(self):
 724         client = self.api_client_mock(200)
 725         # Ensure Keep will not return anything if asked.
 726         with tutil.mock_responses(None, 404):
 727             reader = arvados.CollectionReader(self.DEFAULT_DATA_HASH,
 728                                               api_client=client)
 729             self.assertEqual(self.DEFAULT_MANIFEST, reader.manifest_text())
 730
 731     def test_locator_init_fallback_to_keep(self):
 732         # crunch-job needs this to read manifests that have only ever
 733         # been written to Keep.
 734         client = self.api_client_mock(200)
 735         with tutil.mock_responses(self.DEFAULT_MANIFEST, 404, 200):
 736             reader = arvados.CollectionReader(self.DEFAULT_DATA_HASH,
 737                                               api_client=client)
 738             self.assertEqual(self.DEFAULT_MANIFEST, reader.manifest_text())
 739
 740     def test_uuid_init_no_fallback_to_keep(self):
 741         # Do not look up a collection UUID in Keep.
 742         client = self.api_client_mock(404)
 743         reader = arvados.CollectionReader(self.DEFAULT_UUID,
 744                                           api_client=client)
 745         with tutil.mock_responses(self.DEFAULT_MANIFEST, 200):
 746             with self.assertRaises(arvados.errors.ApiError):
 747                 reader.manifest_text()
 748
 749     def test_try_keep_first_if_permission_hint(self):
 750         # To verify that CollectionReader tries Keep first here, we
 751         # mock API server to return the wrong data.
 752         client = self.api_client_mock(200)
 753         with tutil.mock_responses(self.ALT_MANIFEST, 200):
 754             self.assertEqual(
 755                 self.ALT_MANIFEST,
 756                 arvados.CollectionReader(
 757                     self.ALT_DATA_HASH + '+Affffffffffffffffffffffffffffffffffffffff@fedcba98',
 758                     api_client=client).manifest_text())
 759
 760     def test_init_num_retries_propagated(self):
 761         # More of an integration test...
 762         client = self.api_client_mock(200)
 763         reader = arvados.CollectionReader(self.DEFAULT_UUID, api_client=client,
 764                                           num_retries=3)
 765         with tutil.mock_responses('foo', 500, 500, 200):
 766             self.assertEqual('foo',
 767                              ''.join(f.read(9) for f in reader.all_files()))
 768
 769     def test_read_nonnormalized_manifest_with_collection_reader(self):
 770         # client should be able to use CollectionReader on a manifest without normalizing it
 771         client = self.api_client_mock(500)
 772         nonnormal = ". acbd18db4cc2f85cedef654fccc4a4d8+3+Aabadbadbee@abeebdee 0:3:foo.txt 1:0:bar.txt 0:3:foo.txt\n"
 773         self.assertEqual(
 774             ". acbd18db4cc2f85cedef654fccc4a4d8+3 0:3:foo.txt 1:0:bar.txt 0:3:foo.txt\n",
 775             arvados.CollectionReader(
 776                 nonnormal,
 777                 api_client=client, num_retries=0).stripped_manifest())
 778         self.assertEqual(
 779             [[6, '.', 'foo.txt'],
 780              [0, '.', 'bar.txt']],
 781             [[f.size(), f.stream_name(), f.name()]
 782              for f in
 783              arvados.CollectionReader(
 784                     nonnormal,
 785                     api_client=client, num_retries=0).all_streams()[0].all_files()])
 786
 787
 788 @tutil.skip_sleep
 789 class CollectionWriterTestCase(unittest.TestCase, CollectionTestMixin):
 790     def mock_keep(self, body, *codes, **headers):
 791         headers.setdefault('x-keep-replicas-stored', 2)
 792         return tutil.mock_responses(body, *codes, **headers)
 793
 794     def foo_writer(self, **kwargs):
 795         api_client = self.api_client_mock()
 796         writer = arvados.CollectionWriter(api_client, **kwargs)
 797         writer.start_new_file('foo')
 798         writer.write('foo')
 799         return writer
 800
 801     def test_write_whole_collection(self):
 802         writer = self.foo_writer()
 803         with self.mock_keep(self.DEFAULT_DATA_HASH, 200, 200):
 804             self.assertEqual(self.DEFAULT_DATA_HASH, writer.finish())
 805
 806     def test_write_no_default(self):
 807         writer = self.foo_writer()
 808         with self.mock_keep(None, 500):
 809             with self.assertRaises(arvados.errors.KeepWriteError):
 810                 writer.finish()
 811
 812     def test_write_whole_collection_through_retries(self):
 813         writer = self.foo_writer(num_retries=2)
 814         with self.mock_keep(self.DEFAULT_DATA_HASH,
 815                             500, 500, 200, 500, 500, 200):
 816             self.assertEqual(self.DEFAULT_DATA_HASH, writer.finish())
 817
 818     def test_flush_data_retries(self):
 819         writer = self.foo_writer(num_retries=2)
 820         foo_hash = self.DEFAULT_MANIFEST.split()[1]
 821         with self.mock_keep(foo_hash, 500, 200):
 822             writer.flush_data()
 823         self.assertEqual(self.DEFAULT_MANIFEST, writer.manifest_text())
 824
 825
 826 if __name__ == '__main__':
 827     unittest.main()