1 from __future__ import absolute_import
3 from builtins import object
14 from . import run_test_server
15 from arvados._ranges import Range, LocatorAndRange
16 from arvados.collection import Collection, CollectionReader
17 from . import arvados_testutil as tutil
19 class TestResumableWriter(arvados.ResumableCollectionWriter):
20 KEEP_BLOCK_SIZE = 1024 # PUT to Keep every 1K.
22 def current_state(self):
23 return self.dump_state(copy.deepcopy)
26 class ArvadosCollectionsTest(run_test_server.TestCaseWithServers,
27 tutil.ArvadosBaseTestCase):
32 super(ArvadosCollectionsTest, cls).setUpClass()
33 run_test_server.authorize_with('active')
34 cls.api_client = arvados.api('v1')
35 cls.keep_client = arvados.KeepClient(api_client=cls.api_client,
36 local_store=cls.local_store)
38 def write_foo_bar_baz(self):
39 cw = arvados.CollectionWriter(self.api_client)
40 self.assertEqual(cw.current_stream_name(), '.',
41 'current_stream_name() should be "." now')
42 cw.set_current_file_name('foo.txt')
44 self.assertEqual(cw.current_file_name(), 'foo.txt',
45 'current_file_name() should be foo.txt now')
46 cw.start_new_file('bar.txt')
48 cw.start_new_stream('baz')
50 cw.set_current_file_name('baz.txt')
51 self.assertEqual(cw.manifest_text(),
52 ". 3858f62230ac3c915f300c664312c63f+6 0:3:foo.txt 3:3:bar.txt\n" +
53 "./baz 73feffa4b7f6bb68e44cf984c85f6e88+3 0:3:baz.txt\n",
54 "wrong manifest: got {}".format(cw.manifest_text()))
56 return cw.portable_data_hash()
58 def test_keep_local_store(self):
59 self.assertEqual(self.keep_client.put(b'foo'), 'acbd18db4cc2f85cedef654fccc4a4d8+3', 'wrong md5 hash from Keep.put')
60 self.assertEqual(self.keep_client.get('acbd18db4cc2f85cedef654fccc4a4d8+3'), b'foo', 'wrong data from Keep.get')
62 def test_local_collection_writer(self):
63 self.assertEqual(self.write_foo_bar_baz(),
64 '23ca013983d6239e98931cc779e68426+114',
65 'wrong locator hash: ' + self.write_foo_bar_baz())
67 def test_local_collection_reader(self):
68 foobarbaz = self.write_foo_bar_baz()
69 cr = arvados.CollectionReader(
70 foobarbaz + '+Xzizzle', self.api_client)
72 for s in cr.all_streams():
73 for f in s.all_files():
74 got += [[f.size(), f.stream_name(), f.name(), f.read(2**26)]]
75 expected = [[3, '.', 'foo.txt', b'foo'],
76 [3, '.', 'bar.txt', b'bar'],
77 [3, './baz', 'baz.txt', b'baz']]
80 stream0 = cr.all_streams()[0]
81 self.assertEqual(stream0.readfrom(0, 0),
83 'reading zero bytes should have returned empty string')
84 self.assertEqual(stream0.readfrom(0, 2**26),
86 'reading entire stream failed')
87 self.assertEqual(stream0.readfrom(2**26, 0),
89 'reading zero bytes should have returned empty string')
91 def _test_subset(self, collection, expected):
92 cr = arvados.CollectionReader(collection, self.api_client)
93 for s in cr.all_streams():
97 got = [f.size(), f.stream_name(), f.name(), "".join(f.readall(2**26))]
100 'all_files|as_manifest did not preserve manifest contents: got %s expected %s' % (got, ex))
102 def test_collection_manifest_subset(self):
103 foobarbaz = self.write_foo_bar_baz()
104 self._test_subset(foobarbaz,
105 [[3, '.', 'bar.txt', b'bar'],
106 [3, '.', 'foo.txt', b'foo'],
107 [3, './baz', 'baz.txt', b'baz']])
108 self._test_subset((". %s %s 0:3:foo.txt 3:3:bar.txt\n" %
109 (self.keep_client.put(b"foo"),
110 self.keep_client.put(b"bar"))),
111 [[3, '.', 'bar.txt', b'bar'],
112 [3, '.', 'foo.txt', b'foo']])
113 self._test_subset((". %s %s 0:2:fo.txt 2:4:obar.txt\n" %
114 (self.keep_client.put(b"foo"),
115 self.keep_client.put(b"bar"))),
116 [[2, '.', 'fo.txt', b'fo'],
117 [4, '.', 'obar.txt', b'obar']])
118 self._test_subset((". %s %s 0:2:fo.txt 2:0:zero.txt 2:2:ob.txt 4:2:ar.txt\n" %
119 (self.keep_client.put(b"foo"),
120 self.keep_client.put(b"bar"))),
121 [[2, '.', 'ar.txt', b'ar'],
122 [2, '.', 'fo.txt', b'fo'],
123 [2, '.', 'ob.txt', b'ob'],
124 [0, '.', 'zero.txt', b'']])
126 def test_collection_empty_file(self):
127 cw = arvados.CollectionWriter(self.api_client)
128 cw.start_new_file('zero.txt')
131 self.assertEqual(cw.manifest_text(), ". d41d8cd98f00b204e9800998ecf8427e+0 0:0:zero.txt\n")
132 self.check_manifest_file_sizes(cw.manifest_text(), [0])
133 cw = arvados.CollectionWriter(self.api_client)
134 cw.start_new_file('zero.txt')
136 cw.start_new_file('one.txt')
138 cw.start_new_stream('foo')
139 cw.start_new_file('zero.txt')
141 self.check_manifest_file_sizes(cw.manifest_text(), [0,1,0])
143 def test_no_implicit_normalize(self):
144 cw = arvados.CollectionWriter(self.api_client)
145 cw.start_new_file('b')
147 cw.start_new_file('a')
149 self.check_manifest_file_sizes(cw.manifest_text(), [1,0])
150 self.check_manifest_file_sizes(
151 arvados.CollectionReader(
152 cw.manifest_text()).manifest_text(normalize=True),
155 def check_manifest_file_sizes(self, manifest_text, expect_sizes):
156 cr = arvados.CollectionReader(manifest_text, self.api_client)
158 for f in cr.all_files():
159 got_sizes += [f.size()]
160 self.assertEqual(got_sizes, expect_sizes, "got wrong file sizes %s, expected %s" % (got_sizes, expect_sizes))
162 def test_normalized_collection(self):
163 m1 = """. 5348b82a029fd9e971a811ce1f71360b+43 0:43:md5sum.txt
164 . 085c37f02916da1cad16f93c54d899b7+41 0:41:md5sum.txt
165 . 8b22da26f9f433dea0a10e5ec66d73ba+43 0:43:md5sum.txt
167 self.assertEqual(arvados.CollectionReader(m1, self.api_client).manifest_text(normalize=True),
168 """. 5348b82a029fd9e971a811ce1f71360b+43 085c37f02916da1cad16f93c54d899b7+41 8b22da26f9f433dea0a10e5ec66d73ba+43 0:127:md5sum.txt
171 m2 = """. 204e43b8a1185621ca55a94839582e6f+67108864 b9677abbac956bd3e86b1deb28dfac03+67108864 fc15aff2a762b13f521baf042140acec+67108864 323d2a3ce20370c4ca1d3462a344f8fd+25885655 0:227212247:var-GS000016015-ASM.tsv.bz2
173 self.assertEqual(arvados.CollectionReader(m2, self.api_client).manifest_text(normalize=True), m2)
175 m3 = """. 5348b82a029fd9e971a811ce1f71360b+43 3:40:md5sum.txt
176 . 085c37f02916da1cad16f93c54d899b7+41 0:41:md5sum.txt
177 . 8b22da26f9f433dea0a10e5ec66d73ba+43 0:43:md5sum.txt
179 self.assertEqual(arvados.CollectionReader(m3, self.api_client).manifest_text(normalize=True),
180 """. 5348b82a029fd9e971a811ce1f71360b+43 085c37f02916da1cad16f93c54d899b7+41 8b22da26f9f433dea0a10e5ec66d73ba+43 3:124:md5sum.txt
183 m4 = """. 204e43b8a1185621ca55a94839582e6f+67108864 0:3:foo/bar
184 ./zzz 204e43b8a1185621ca55a94839582e6f+67108864 0:999:zzz
185 ./foo 323d2a3ce20370c4ca1d3462a344f8fd+25885655 0:3:bar
187 self.assertEqual(arvados.CollectionReader(m4, self.api_client).manifest_text(normalize=True),
188 """./foo 204e43b8a1185621ca55a94839582e6f+67108864 323d2a3ce20370c4ca1d3462a344f8fd+25885655 0:3:bar 67108864:3:bar
189 ./zzz 204e43b8a1185621ca55a94839582e6f+67108864 0:999:zzz
192 m5 = """. 204e43b8a1185621ca55a94839582e6f+67108864 0:3:foo/bar
193 ./zzz 204e43b8a1185621ca55a94839582e6f+67108864 0:999:zzz
194 ./foo 204e43b8a1185621ca55a94839582e6f+67108864 3:3:bar
196 self.assertEqual(arvados.CollectionReader(m5, self.api_client).manifest_text(normalize=True),
197 """./foo 204e43b8a1185621ca55a94839582e6f+67108864 0:6:bar
198 ./zzz 204e43b8a1185621ca55a94839582e6f+67108864 0:999:zzz
201 with self.data_file('1000G_ref_manifest') as f6:
203 self.assertEqual(arvados.CollectionReader(m6, self.api_client).manifest_text(normalize=True), m6)
205 with self.data_file('jlake_manifest') as f7:
207 self.assertEqual(arvados.CollectionReader(m7, self.api_client).manifest_text(normalize=True), m7)
209 m8 = """./a\\040b\\040c 59ca0efa9f5633cb0371bbc0355478d8+13 0:13:hello\\040world.txt
211 self.assertEqual(arvados.CollectionReader(m8, self.api_client).manifest_text(normalize=True), m8)
213 def test_locators_and_ranges(self):
214 blocks2 = [Range('a', 0, 10),
221 self.assertEqual(arvados.locators_and_ranges(blocks2, 2, 2), [LocatorAndRange('a', 10, 2, 2)])
222 self.assertEqual(arvados.locators_and_ranges(blocks2, 12, 2), [LocatorAndRange('b', 10, 2, 2)])
223 self.assertEqual(arvados.locators_and_ranges(blocks2, 22, 2), [LocatorAndRange('c', 10, 2, 2)])
224 self.assertEqual(arvados.locators_and_ranges(blocks2, 32, 2), [LocatorAndRange('d', 10, 2, 2)])
225 self.assertEqual(arvados.locators_and_ranges(blocks2, 42, 2), [LocatorAndRange('e', 10, 2, 2)])
226 self.assertEqual(arvados.locators_and_ranges(blocks2, 52, 2), [LocatorAndRange('f', 10, 2, 2)])
227 self.assertEqual(arvados.locators_and_ranges(blocks2, 62, 2), [])
228 self.assertEqual(arvados.locators_and_ranges(blocks2, -2, 2), [])
230 self.assertEqual(arvados.locators_and_ranges(blocks2, 0, 2), [LocatorAndRange('a', 10, 0, 2)])
231 self.assertEqual(arvados.locators_and_ranges(blocks2, 10, 2), [LocatorAndRange('b', 10, 0, 2)])
232 self.assertEqual(arvados.locators_and_ranges(blocks2, 20, 2), [LocatorAndRange('c', 10, 0, 2)])
233 self.assertEqual(arvados.locators_and_ranges(blocks2, 30, 2), [LocatorAndRange('d', 10, 0, 2)])
234 self.assertEqual(arvados.locators_and_ranges(blocks2, 40, 2), [LocatorAndRange('e', 10, 0, 2)])
235 self.assertEqual(arvados.locators_and_ranges(blocks2, 50, 2), [LocatorAndRange('f', 10, 0, 2)])
236 self.assertEqual(arvados.locators_and_ranges(blocks2, 60, 2), [])
237 self.assertEqual(arvados.locators_and_ranges(blocks2, -2, 2), [])
239 self.assertEqual(arvados.locators_and_ranges(blocks2, 9, 2), [LocatorAndRange('a', 10, 9, 1), LocatorAndRange('b', 10, 0, 1)])
240 self.assertEqual(arvados.locators_and_ranges(blocks2, 19, 2), [LocatorAndRange('b', 10, 9, 1), LocatorAndRange('c', 10, 0, 1)])
241 self.assertEqual(arvados.locators_and_ranges(blocks2, 29, 2), [LocatorAndRange('c', 10, 9, 1), LocatorAndRange('d', 10, 0, 1)])
242 self.assertEqual(arvados.locators_and_ranges(blocks2, 39, 2), [LocatorAndRange('d', 10, 9, 1), LocatorAndRange('e', 10, 0, 1)])
243 self.assertEqual(arvados.locators_and_ranges(blocks2, 49, 2), [LocatorAndRange('e', 10, 9, 1), LocatorAndRange('f', 10, 0, 1)])
244 self.assertEqual(arvados.locators_and_ranges(blocks2, 59, 2), [LocatorAndRange('f', 10, 9, 1)])
247 blocks3 = [Range('a', 0, 10),
255 self.assertEqual(arvados.locators_and_ranges(blocks3, 2, 2), [LocatorAndRange('a', 10, 2, 2)])
256 self.assertEqual(arvados.locators_and_ranges(blocks3, 12, 2), [LocatorAndRange('b', 10, 2, 2)])
257 self.assertEqual(arvados.locators_and_ranges(blocks3, 22, 2), [LocatorAndRange('c', 10, 2, 2)])
258 self.assertEqual(arvados.locators_and_ranges(blocks3, 32, 2), [LocatorAndRange('d', 10, 2, 2)])
259 self.assertEqual(arvados.locators_and_ranges(blocks3, 42, 2), [LocatorAndRange('e', 10, 2, 2)])
260 self.assertEqual(arvados.locators_and_ranges(blocks3, 52, 2), [LocatorAndRange('f', 10, 2, 2)])
261 self.assertEqual(arvados.locators_and_ranges(blocks3, 62, 2), [LocatorAndRange('g', 10, 2, 2)])
264 blocks = [Range('a', 0, 10),
267 self.assertEqual(arvados.locators_and_ranges(blocks, 1, 0), [])
268 self.assertEqual(arvados.locators_and_ranges(blocks, 0, 5), [LocatorAndRange('a', 10, 0, 5)])
269 self.assertEqual(arvados.locators_and_ranges(blocks, 3, 5), [LocatorAndRange('a', 10, 3, 5)])
270 self.assertEqual(arvados.locators_and_ranges(blocks, 0, 10), [LocatorAndRange('a', 10, 0, 10)])
272 self.assertEqual(arvados.locators_and_ranges(blocks, 0, 11), [LocatorAndRange('a', 10, 0, 10),
273 LocatorAndRange('b', 15, 0, 1)])
274 self.assertEqual(arvados.locators_and_ranges(blocks, 1, 11), [LocatorAndRange('a', 10, 1, 9),
275 LocatorAndRange('b', 15, 0, 2)])
276 self.assertEqual(arvados.locators_and_ranges(blocks, 0, 25), [LocatorAndRange('a', 10, 0, 10),
277 LocatorAndRange('b', 15, 0, 15)])
279 self.assertEqual(arvados.locators_and_ranges(blocks, 0, 30), [LocatorAndRange('a', 10, 0, 10),
280 LocatorAndRange('b', 15, 0, 15),
281 LocatorAndRange('c', 5, 0, 5)])
282 self.assertEqual(arvados.locators_and_ranges(blocks, 1, 30), [LocatorAndRange('a', 10, 1, 9),
283 LocatorAndRange('b', 15, 0, 15),
284 LocatorAndRange('c', 5, 0, 5)])
285 self.assertEqual(arvados.locators_and_ranges(blocks, 0, 31), [LocatorAndRange('a', 10, 0, 10),
286 LocatorAndRange('b', 15, 0, 15),
287 LocatorAndRange('c', 5, 0, 5)])
289 self.assertEqual(arvados.locators_and_ranges(blocks, 15, 5), [LocatorAndRange('b', 15, 5, 5)])
291 self.assertEqual(arvados.locators_and_ranges(blocks, 8, 17), [LocatorAndRange('a', 10, 8, 2),
292 LocatorAndRange('b', 15, 0, 15)])
294 self.assertEqual(arvados.locators_and_ranges(blocks, 8, 20), [LocatorAndRange('a', 10, 8, 2),
295 LocatorAndRange('b', 15, 0, 15),
296 LocatorAndRange('c', 5, 0, 3)])
298 self.assertEqual(arvados.locators_and_ranges(blocks, 26, 2), [LocatorAndRange('c', 5, 1, 2)])
300 self.assertEqual(arvados.locators_and_ranges(blocks, 9, 15), [LocatorAndRange('a', 10, 9, 1),
301 LocatorAndRange('b', 15, 0, 14)])
302 self.assertEqual(arvados.locators_and_ranges(blocks, 10, 15), [LocatorAndRange('b', 15, 0, 15)])
303 self.assertEqual(arvados.locators_and_ranges(blocks, 11, 15), [LocatorAndRange('b', 15, 1, 14),
304 LocatorAndRange('c', 5, 0, 1)])
306 class MockKeep(object):
307 def __init__(self, content, num_retries=0):
308 self.content = content
310 def get(self, locator, num_retries=0):
311 return self.content[locator]
313 def test_stream_reader(self):
315 'aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa+10': b'abcdefghij',
316 'bbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbb+15': b'klmnopqrstuvwxy',
317 'cccccccccccccccccccccccccccccccc+5': b'z0123',
319 mk = self.MockKeep(keepblocks)
321 sr = arvados.StreamReader([".", "aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa+10", "bbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbb+15", "cccccccccccccccccccccccccccccccc+5", "0:30:foo"], mk)
323 content = b'abcdefghijklmnopqrstuvwxyz0123456789'
325 self.assertEqual(sr.readfrom(0, 30), content[0:30])
326 self.assertEqual(sr.readfrom(2, 30), content[2:30])
328 self.assertEqual(sr.readfrom(2, 8), content[2:10])
329 self.assertEqual(sr.readfrom(0, 10), content[0:10])
331 self.assertEqual(sr.readfrom(0, 5), content[0:5])
332 self.assertEqual(sr.readfrom(5, 5), content[5:10])
333 self.assertEqual(sr.readfrom(10, 5), content[10:15])
334 self.assertEqual(sr.readfrom(15, 5), content[15:20])
335 self.assertEqual(sr.readfrom(20, 5), content[20:25])
336 self.assertEqual(sr.readfrom(25, 5), content[25:30])
337 self.assertEqual(sr.readfrom(30, 5), b'')
339 def test_extract_file(self):
340 m1 = """. 5348b82a029fd9e971a811ce1f71360b+43 0:43:md5sum.txt
341 . 085c37f02916da1cad16f93c54d899b7+41 0:41:md6sum.txt
342 . 8b22da26f9f433dea0a10e5ec66d73ba+43 0:43:md7sum.txt
343 . 085c37f02916da1cad16f93c54d899b7+41 5348b82a029fd9e971a811ce1f71360b+43 8b22da26f9f433dea0a10e5ec66d73ba+43 47:80:md8sum.txt
344 . 085c37f02916da1cad16f93c54d899b7+41 5348b82a029fd9e971a811ce1f71360b+43 8b22da26f9f433dea0a10e5ec66d73ba+43 40:80:md9sum.txt
347 m2 = arvados.CollectionReader(m1, self.api_client).manifest_text(normalize=True)
350 ". 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")
351 files = arvados.CollectionReader(
352 m2, self.api_client).all_streams()[0].files()
354 self.assertEqual(files['md5sum.txt'].as_manifest(),
355 ". 5348b82a029fd9e971a811ce1f71360b+43 0:43:md5sum.txt\n")
356 self.assertEqual(files['md6sum.txt'].as_manifest(),
357 ". 085c37f02916da1cad16f93c54d899b7+41 0:41:md6sum.txt\n")
358 self.assertEqual(files['md7sum.txt'].as_manifest(),
359 ". 8b22da26f9f433dea0a10e5ec66d73ba+43 0:43:md7sum.txt\n")
360 self.assertEqual(files['md9sum.txt'].as_manifest(),
361 ". 085c37f02916da1cad16f93c54d899b7+41 5348b82a029fd9e971a811ce1f71360b+43 8b22da26f9f433dea0a10e5ec66d73ba+43 40:80:md9sum.txt\n")
363 def test_write_directory_tree(self):
364 cwriter = arvados.CollectionWriter(self.api_client)
365 cwriter.write_directory_tree(self.build_directory_tree(
366 ['basefile', 'subdir/subfile']))
367 self.assertEqual(cwriter.manifest_text(),
368 """. c5110c5ac93202d8e0f9e381f22bac0f+8 0:8:basefile
369 ./subdir 1ca4dec89403084bf282ad31e6cf7972+14 0:14:subfile\n""")
371 def test_write_named_directory_tree(self):
372 cwriter = arvados.CollectionWriter(self.api_client)
373 cwriter.write_directory_tree(self.build_directory_tree(
374 ['basefile', 'subdir/subfile']), 'root')
376 cwriter.manifest_text(),
377 """./root c5110c5ac93202d8e0f9e381f22bac0f+8 0:8:basefile
378 ./root/subdir 1ca4dec89403084bf282ad31e6cf7972+14 0:14:subfile\n""")
380 def test_write_directory_tree_in_one_stream(self):
381 cwriter = arvados.CollectionWriter(self.api_client)
382 cwriter.write_directory_tree(self.build_directory_tree(
383 ['basefile', 'subdir/subfile']), max_manifest_depth=0)
384 self.assertEqual(cwriter.manifest_text(),
385 """. 4ace875ffdc6824a04950f06858f4465+22 0:8:basefile 8:14:subdir/subfile\n""")
387 def test_write_directory_tree_with_limited_recursion(self):
388 cwriter = arvados.CollectionWriter(self.api_client)
389 cwriter.write_directory_tree(
390 self.build_directory_tree(['f1', 'd1/f2', 'd1/d2/f3']),
391 max_manifest_depth=1)
392 self.assertEqual(cwriter.manifest_text(),
393 """. bd19836ddb62c11c55ab251ccaca5645+2 0:2:f1
394 ./d1 50170217e5b04312024aa5cd42934494+13 0:8:d2/f3 8:5:f2\n""")
396 def test_write_directory_tree_with_zero_recursion(self):
397 cwriter = arvados.CollectionWriter(self.api_client)
398 content = 'd1/d2/f3d1/f2f1'
399 blockhash = tutil.str_keep_locator(content)
400 cwriter.write_directory_tree(
401 self.build_directory_tree(['f1', 'd1/f2', 'd1/d2/f3']),
402 max_manifest_depth=0)
404 cwriter.manifest_text(),
405 ". {} 0:8:d1/d2/f3 8:5:d1/f2 13:2:f1\n".format(blockhash))
407 def test_write_one_file(self):
408 cwriter = arvados.CollectionWriter(self.api_client)
409 with self.make_test_file() as testfile:
410 cwriter.write_file(testfile.name)
412 cwriter.manifest_text(),
413 ". 098f6bcd4621d373cade4e832627b4f6+4 0:4:{}\n".format(
414 os.path.basename(testfile.name)))
416 def test_write_named_file(self):
417 cwriter = arvados.CollectionWriter(self.api_client)
418 with self.make_test_file() as testfile:
419 cwriter.write_file(testfile.name, 'foo')
420 self.assertEqual(cwriter.manifest_text(),
421 ". 098f6bcd4621d373cade4e832627b4f6+4 0:4:foo\n")
423 def test_write_multiple_files(self):
424 cwriter = arvados.CollectionWriter(self.api_client)
426 with self.make_test_file(letter.encode()) as testfile:
427 cwriter.write_file(testfile.name, letter)
429 cwriter.manifest_text(),
430 ". 902fbdd2b1df0c4f70b4a5d23525e932+3 0:1:A 1:1:B 2:1:C\n")
432 def test_basic_resume(self):
433 cwriter = TestResumableWriter()
434 with self.make_test_file() as testfile:
435 cwriter.write_file(testfile.name, 'test')
436 resumed = TestResumableWriter.from_state(cwriter.current_state())
437 self.assertEqual(cwriter.manifest_text(), resumed.manifest_text(),
438 "resumed CollectionWriter had different manifest")
440 def test_resume_fails_when_missing_dependency(self):
441 cwriter = TestResumableWriter()
442 with self.make_test_file() as testfile:
443 cwriter.write_file(testfile.name, 'test')
444 self.assertRaises(arvados.errors.StaleWriterStateError,
445 TestResumableWriter.from_state,
446 cwriter.current_state())
448 def test_resume_fails_when_dependency_mtime_changed(self):
449 cwriter = TestResumableWriter()
450 with self.make_test_file() as testfile:
451 cwriter.write_file(testfile.name, 'test')
452 os.utime(testfile.name, (0, 0))
453 self.assertRaises(arvados.errors.StaleWriterStateError,
454 TestResumableWriter.from_state,
455 cwriter.current_state())
457 def test_resume_fails_when_dependency_is_nonfile(self):
458 cwriter = TestResumableWriter()
459 cwriter.write_file('/dev/null', 'empty')
460 self.assertRaises(arvados.errors.StaleWriterStateError,
461 TestResumableWriter.from_state,
462 cwriter.current_state())
464 def test_resume_fails_when_dependency_size_changed(self):
465 cwriter = TestResumableWriter()
466 with self.make_test_file() as testfile:
467 cwriter.write_file(testfile.name, 'test')
468 orig_mtime = os.fstat(testfile.fileno()).st_mtime
469 testfile.write(b'extra')
471 os.utime(testfile.name, (orig_mtime, orig_mtime))
472 self.assertRaises(arvados.errors.StaleWriterStateError,
473 TestResumableWriter.from_state,
474 cwriter.current_state())
476 def test_resume_fails_with_expired_locator(self):
477 cwriter = TestResumableWriter()
478 state = cwriter.current_state()
479 # Add an expired locator to the state.
480 state['_current_stream_locators'].append(''.join([
481 'a' * 32, '+1+A', 'b' * 40, '@', '10000000']))
482 self.assertRaises(arvados.errors.StaleWriterStateError,
483 TestResumableWriter.from_state, state)
485 def test_arbitrary_objects_not_resumable(self):
486 cwriter = TestResumableWriter()
487 with open('/dev/null') as badfile:
488 self.assertRaises(arvados.errors.AssertionError,
489 cwriter.write_file, badfile)
491 def test_arbitrary_writes_not_resumable(self):
492 cwriter = TestResumableWriter()
493 self.assertRaises(arvados.errors.AssertionError,
494 cwriter.write, "badtext")
496 def test_read_arbitrary_data_with_collection_reader(self):
497 # arv-get relies on this to do "arv-get {keep-locator} -".
498 self.write_foo_bar_baz()
501 arvados.CollectionReader(
502 '3858f62230ac3c915f300c664312c63f+6'
506 class CollectionTestMixin(tutil.ApiClientMock):
507 API_COLLECTIONS = run_test_server.fixture('collections')
508 DEFAULT_COLLECTION = API_COLLECTIONS['foo_file']
509 DEFAULT_DATA_HASH = DEFAULT_COLLECTION['portable_data_hash']
510 DEFAULT_MANIFEST = DEFAULT_COLLECTION['manifest_text']
511 DEFAULT_UUID = DEFAULT_COLLECTION['uuid']
512 ALT_COLLECTION = API_COLLECTIONS['bar_file']
513 ALT_DATA_HASH = ALT_COLLECTION['portable_data_hash']
514 ALT_MANIFEST = ALT_COLLECTION['manifest_text']
516 def api_client_mock(self, status=200):
517 client = super(CollectionTestMixin, self).api_client_mock()
518 self.mock_keep_services(client, status=status, service_type='proxy', count=1)
523 class CollectionReaderTestCase(unittest.TestCase, CollectionTestMixin):
524 def mock_get_collection(self, api_mock, code, fixturename):
525 body = self.API_COLLECTIONS.get(fixturename)
526 self._mock_api_call(api_mock.collections().get, code, body)
528 def api_client_mock(self, status=200):
529 client = super(CollectionReaderTestCase, self).api_client_mock()
530 self.mock_get_collection(client, status, 'foo_file')
533 def test_init_no_default_retries(self):
534 client = self.api_client_mock(200)
535 reader = arvados.CollectionReader(self.DEFAULT_UUID, api_client=client)
536 reader.manifest_text()
537 client.collections().get().execute.assert_called_with(num_retries=0)
539 def test_uuid_init_success(self):
540 client = self.api_client_mock(200)
541 reader = arvados.CollectionReader(self.DEFAULT_UUID, api_client=client,
543 self.assertEqual(self.DEFAULT_COLLECTION['manifest_text'],
544 reader.manifest_text())
545 client.collections().get().execute.assert_called_with(num_retries=3)
547 def test_uuid_init_failure_raises_api_error(self):
548 client = self.api_client_mock(500)
549 with self.assertRaises(arvados.errors.ApiError):
550 reader = arvados.CollectionReader(self.DEFAULT_UUID, api_client=client)
552 def test_locator_init(self):
553 client = self.api_client_mock(200)
554 # Ensure Keep will not return anything if asked.
555 with tutil.mock_keep_responses(None, 404):
556 reader = arvados.CollectionReader(self.DEFAULT_DATA_HASH,
558 self.assertEqual(self.DEFAULT_MANIFEST, reader.manifest_text())
560 def test_locator_init_fallback_to_keep(self):
561 # crunch-job needs this to read manifests that have only ever
562 # been written to Keep.
563 client = self.api_client_mock(200)
564 self.mock_get_collection(client, 404, None)
565 with tutil.mock_keep_responses(self.DEFAULT_MANIFEST, 200):
566 reader = arvados.CollectionReader(self.DEFAULT_DATA_HASH,
568 self.assertEqual(self.DEFAULT_MANIFEST, reader.manifest_text())
570 def test_uuid_init_no_fallback_to_keep(self):
571 # Do not look up a collection UUID in Keep.
572 client = self.api_client_mock(404)
573 with tutil.mock_keep_responses(self.DEFAULT_MANIFEST, 200):
574 with self.assertRaises(arvados.errors.ApiError):
575 reader = arvados.CollectionReader(self.DEFAULT_UUID,
578 def test_try_keep_first_if_permission_hint(self):
579 # To verify that CollectionReader tries Keep first here, we
580 # mock API server to return the wrong data.
581 client = self.api_client_mock(200)
582 with tutil.mock_keep_responses(self.ALT_MANIFEST, 200):
585 arvados.CollectionReader(
586 self.ALT_DATA_HASH + '+Affffffffffffffffffffffffffffffffffffffff@fedcba98',
587 api_client=client).manifest_text())
589 def test_init_num_retries_propagated(self):
590 # More of an integration test...
591 client = self.api_client_mock(200)
592 reader = arvados.CollectionReader(self.DEFAULT_UUID, api_client=client,
594 with tutil.mock_keep_responses('foo', 500, 500, 200):
595 self.assertEqual(b'foo',
596 b''.join(f.read(9) for f in reader.all_files()))
598 def test_read_nonnormalized_manifest_with_collection_reader(self):
599 # client should be able to use CollectionReader on a manifest without normalizing it
600 client = self.api_client_mock(500)
601 nonnormal = ". acbd18db4cc2f85cedef654fccc4a4d8+3+Aabadbadbee@abeebdee 0:3:foo.txt 1:0:bar.txt 0:3:foo.txt\n"
602 reader = arvados.CollectionReader(
604 api_client=client, num_retries=0)
605 # Ensure stripped_manifest() doesn't mangle our manifest in
606 # any way other than stripping hints.
608 re.sub('\+[^\d\s\+]+', '', nonnormal),
609 reader.stripped_manifest())
610 # Ensure stripped_manifest() didn't mutate our reader.
611 self.assertEqual(nonnormal, reader.manifest_text())
612 # Ensure the files appear in the order given in the manifest.
614 [[6, '.', 'foo.txt'],
615 [0, '.', 'bar.txt']],
616 [[f.size(), f.stream_name(), f.name()]
617 for f in reader.all_streams()[0].all_files()])
619 def test_read_empty_collection(self):
620 client = self.api_client_mock(200)
621 self.mock_get_collection(client, 200, 'empty')
622 reader = arvados.CollectionReader('d41d8cd98f00b204e9800998ecf8427e+0',
624 self.assertEqual('', reader.manifest_text())
626 def test_api_response(self):
627 client = self.api_client_mock()
628 reader = arvados.CollectionReader(self.DEFAULT_UUID, api_client=client)
629 self.assertEqual(self.DEFAULT_COLLECTION, reader.api_response())
631 def test_api_response_with_collection_from_keep(self):
632 client = self.api_client_mock()
633 self.mock_get_collection(client, 404, 'foo')
634 with tutil.mock_keep_responses(self.DEFAULT_MANIFEST, 200):
635 reader = arvados.CollectionReader(self.DEFAULT_DATA_HASH,
637 api_response = reader.api_response()
638 self.assertIsNone(api_response)
640 def check_open_file(self, coll_file, stream_name, file_name, file_size):
641 self.assertFalse(coll_file.closed, "returned file is not open")
642 self.assertEqual(stream_name, coll_file.stream_name())
643 self.assertEqual(file_name, coll_file.name)
644 self.assertEqual(file_size, coll_file.size())
646 def test_open_collection_file_one_argument(self):
647 client = self.api_client_mock(200)
648 reader = arvados.CollectionReader(self.DEFAULT_UUID, api_client=client)
649 cfile = reader.open('./foo', 'rb')
650 self.check_open_file(cfile, '.', 'foo', 3)
652 def test_open_deep_file(self):
653 coll_name = 'collection_with_files_in_subdir'
654 client = self.api_client_mock(200)
655 self.mock_get_collection(client, 200, coll_name)
656 reader = arvados.CollectionReader(
657 self.API_COLLECTIONS[coll_name]['uuid'], api_client=client)
658 cfile = reader.open('./subdir2/subdir3/file2_in_subdir3.txt', 'rb')
659 self.check_open_file(cfile, './subdir2/subdir3', 'file2_in_subdir3.txt',
662 def test_open_nonexistent_stream(self):
663 client = self.api_client_mock(200)
664 reader = arvados.CollectionReader(self.DEFAULT_UUID, api_client=client)
665 self.assertRaises(IOError, reader.open, './nonexistent/foo')
667 def test_open_nonexistent_file(self):
668 client = self.api_client_mock(200)
669 reader = arvados.CollectionReader(self.DEFAULT_UUID, api_client=client)
670 self.assertRaises(IOError, reader.open, 'nonexistent')
674 class CollectionWriterTestCase(unittest.TestCase, CollectionTestMixin):
675 def mock_keep(self, body, *codes, **headers):
676 headers.setdefault('x-keep-replicas-stored', 2)
677 return tutil.mock_keep_responses(body, *codes, **headers)
679 def foo_writer(self, **kwargs):
680 kwargs.setdefault('api_client', self.api_client_mock())
681 writer = arvados.CollectionWriter(**kwargs)
682 writer.start_new_file('foo')
686 def test_write_whole_collection(self):
687 writer = self.foo_writer()
688 with self.mock_keep(self.DEFAULT_DATA_HASH, 200, 200):
689 self.assertEqual(self.DEFAULT_DATA_HASH, writer.finish())
691 def test_write_no_default(self):
692 writer = self.foo_writer()
693 with self.mock_keep(None, 500):
694 with self.assertRaises(arvados.errors.KeepWriteError):
697 def test_write_insufficient_replicas_via_proxy(self):
698 writer = self.foo_writer(replication=3)
699 with self.mock_keep(None, 200, **{'x-keep-replicas-stored': 2}):
700 with self.assertRaises(arvados.errors.KeepWriteError):
701 writer.manifest_text()
703 def test_write_insufficient_replicas_via_disks(self):
704 client = mock.MagicMock(name='api_client')
707 **{'x-keep-replicas-stored': 1}) as keepmock:
708 self.mock_keep_services(client, status=200, service_type='disk', count=2)
709 writer = self.foo_writer(api_client=client, replication=3)
710 with self.assertRaises(arvados.errors.KeepWriteError):
711 writer.manifest_text()
713 def test_write_three_replicas(self):
714 client = mock.MagicMock(name='api_client')
716 "", 500, 500, 500, 200, 200, 200,
717 **{'x-keep-replicas-stored': 1}) as keepmock:
718 self.mock_keep_services(client, status=200, service_type='disk', count=6)
719 writer = self.foo_writer(api_client=client, replication=3)
720 writer.manifest_text()
721 self.assertEqual(6, keepmock.call_count)
723 def test_write_whole_collection_through_retries(self):
724 writer = self.foo_writer(num_retries=2)
725 with self.mock_keep(self.DEFAULT_DATA_HASH,
726 500, 500, 200, 500, 500, 200):
727 self.assertEqual(self.DEFAULT_DATA_HASH, writer.finish())
729 def test_flush_data_retries(self):
730 writer = self.foo_writer(num_retries=2)
731 foo_hash = self.DEFAULT_MANIFEST.split()[1]
732 with self.mock_keep(foo_hash, 500, 200):
734 self.assertEqual(self.DEFAULT_MANIFEST, writer.manifest_text())
736 def test_one_open(self):
737 client = self.api_client_mock()
738 writer = arvados.CollectionWriter(client)
739 with writer.open('out') as out_file:
740 self.assertEqual('.', writer.current_stream_name())
741 self.assertEqual('out', writer.current_file_name())
742 out_file.write(b'test data')
743 data_loc = tutil.str_keep_locator('test data')
744 self.assertTrue(out_file.closed, "writer file not closed after context")
745 self.assertRaises(ValueError, out_file.write, 'extra text')
746 with self.mock_keep(data_loc, 200) as keep_mock:
747 self.assertEqual(". {} 0:9:out\n".format(data_loc),
748 writer.manifest_text())
750 def test_open_writelines(self):
751 client = self.api_client_mock()
752 writer = arvados.CollectionWriter(client)
753 with writer.open('six') as out_file:
754 out_file.writelines(['12', '34', '56'])
755 data_loc = tutil.str_keep_locator('123456')
756 with self.mock_keep(data_loc, 200) as keep_mock:
757 self.assertEqual(". {} 0:6:six\n".format(data_loc),
758 writer.manifest_text())
760 def test_open_flush(self):
761 client = self.api_client_mock()
762 data_loc1 = tutil.str_keep_locator('flush1')
763 data_loc2 = tutil.str_keep_locator('flush2')
764 with self.mock_keep((data_loc1, 200), (data_loc2, 200)) as keep_mock:
765 writer = arvados.CollectionWriter(client)
766 with writer.open('flush_test') as out_file:
767 out_file.write(b'flush1')
769 out_file.write(b'flush2')
770 self.assertEqual(". {} {} 0:12:flush_test\n".format(data_loc1,
772 writer.manifest_text())
774 def test_two_opens_same_stream(self):
775 client = self.api_client_mock()
776 writer = arvados.CollectionWriter(client)
777 with writer.open('.', '1') as out_file:
778 out_file.write(b'1st')
779 with writer.open('.', '2') as out_file:
780 out_file.write(b'2nd')
781 data_loc = tutil.str_keep_locator('1st2nd')
782 with self.mock_keep(data_loc, 200) as keep_mock:
783 self.assertEqual(". {} 0:3:1 3:3:2\n".format(data_loc),
784 writer.manifest_text())
786 def test_two_opens_two_streams(self):
787 client = self.api_client_mock()
788 data_loc1 = tutil.str_keep_locator('file')
789 data_loc2 = tutil.str_keep_locator('indir')
790 with self.mock_keep((data_loc1, 200), (data_loc2, 200)) as keep_mock:
791 writer = arvados.CollectionWriter(client)
792 with writer.open('file') as out_file:
793 out_file.write(b'file')
794 with writer.open('./dir', 'indir') as out_file:
795 out_file.write(b'indir')
796 expected = ". {} 0:4:file\n./dir {} 0:5:indir\n".format(
797 data_loc1, data_loc2)
798 self.assertEqual(expected, writer.manifest_text())
800 def test_dup_open_fails(self):
801 client = self.api_client_mock()
802 writer = arvados.CollectionWriter(client)
803 file1 = writer.open('one')
804 self.assertRaises(arvados.errors.AssertionError, writer.open, 'two')
807 class CollectionOpenModes(run_test_server.TestCaseWithServers):
809 def test_open_binary_modes(self):
811 for mode in ['wb', 'wb+', 'ab', 'ab+']:
812 with c.open('foo', 'wb') as f:
815 def test_open_invalid_modes(self):
817 for mode in ['+r', 'aa', '++', 'r+b', 'beer', '', None]:
818 with self.assertRaises(Exception):
821 def test_open_text_modes(self):
823 with c.open('foo', 'wb') as f:
825 for mode in ['r', 'rt', 'r+', 'rt+', 'w', 'wt', 'a', 'at']:
826 if sys.version_info >= (3, 0):
827 with self.assertRaises(NotImplementedError):
830 with c.open('foo', mode) as f:
831 if mode[0] == 'r' and '+' not in mode:
832 self.assertEqual('foo', f.read(3))
835 f.seek(-3, os.SEEK_CUR)
836 self.assertEqual('bar', f.read(3))
839 class NewCollectionTestCase(unittest.TestCase, CollectionTestMixin):
841 def test_replication_desired_kept_on_load(self):
842 m = '. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt 0:10:count2.txt\n'
843 c1 = Collection(m, replication_desired=1)
845 loc = c1.manifest_locator()
847 self.assertEqual(c1.manifest_text, c2.manifest_text)
848 self.assertEqual(c1.replication_desired, c2.replication_desired)
850 def test_replication_desired_not_loaded_if_provided(self):
851 m = '. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt 0:10:count2.txt\n'
852 c1 = Collection(m, replication_desired=1)
854 loc = c1.manifest_locator()
855 c2 = Collection(loc, replication_desired=2)
856 self.assertEqual(c1.manifest_text, c2.manifest_text)
857 self.assertNotEqual(c1.replication_desired, c2.replication_desired)
859 def test_init_manifest(self):
860 m1 = """. 5348b82a029fd9e971a811ce1f71360b+43 0:43:md5sum.txt
861 . 085c37f02916da1cad16f93c54d899b7+41 0:41:md5sum.txt
862 . 8b22da26f9f433dea0a10e5ec66d73ba+43 0:43:md5sum.txt
864 self.assertEqual(m1, CollectionReader(m1).manifest_text(normalize=False))
865 self.assertEqual(". 5348b82a029fd9e971a811ce1f71360b+43 085c37f02916da1cad16f93c54d899b7+41 8b22da26f9f433dea0a10e5ec66d73ba+43 0:127:md5sum.txt\n", CollectionReader(m1).manifest_text(normalize=True))
867 def test_init_manifest_with_collision(self):
868 m1 = """. 5348b82a029fd9e971a811ce1f71360b+43 0:43:md5sum.txt
869 ./md5sum.txt 085c37f02916da1cad16f93c54d899b7+41 0:41:md5sum.txt
871 with self.assertRaises(arvados.errors.ArgumentError):
872 self.assertEqual(m1, CollectionReader(m1))
874 def test_init_manifest_with_error(self):
875 m1 = """. 0:43:md5sum.txt"""
876 with self.assertRaises(arvados.errors.ArgumentError):
877 self.assertEqual(m1, CollectionReader(m1))
879 def test_remove(self):
880 c = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt 0:10:count2.txt\n')
881 self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt 0:10:count2.txt\n", c.portable_manifest_text())
882 self.assertIn("count1.txt", c)
883 c.remove("count1.txt")
884 self.assertNotIn("count1.txt", c)
885 self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count2.txt\n", c.portable_manifest_text())
886 with self.assertRaises(arvados.errors.ArgumentError):
890 c = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt 0:10:count2.txt\n')
891 self.assertIs(c.find("."), c)
892 self.assertIs(c.find("./count1.txt"), c["count1.txt"])
893 self.assertIs(c.find("count1.txt"), c["count1.txt"])
894 with self.assertRaises(IOError):
896 with self.assertRaises(arvados.errors.ArgumentError):
898 self.assertIs(c.find("./nonexistant.txt"), None)
899 self.assertIs(c.find("./nonexistantsubdir/nonexistant.txt"), None)
901 def test_remove_in_subdir(self):
902 c = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n./foo 781e5e245d69b566979b86e28d23f2c7+10 0:10:count2.txt\n')
903 c.remove("foo/count2.txt")
904 self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n", c.portable_manifest_text())
906 def test_remove_empty_subdir(self):
907 c = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n./foo 781e5e245d69b566979b86e28d23f2c7+10 0:10:count2.txt\n')
908 c.remove("foo/count2.txt")
910 self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n", c.portable_manifest_text())
912 def test_remove_nonempty_subdir(self):
913 c = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n./foo 781e5e245d69b566979b86e28d23f2c7+10 0:10:count2.txt\n')
914 with self.assertRaises(IOError):
916 c.remove("foo", recursive=True)
917 self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n", c.portable_manifest_text())
919 def test_copy_to_file_in_dir(self):
920 c = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n')
921 c.copy("count1.txt", "foo/count2.txt")
922 self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n./foo 781e5e245d69b566979b86e28d23f2c7+10 0:10:count2.txt\n", c.portable_manifest_text())
924 def test_copy_file(self):
925 c = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n')
926 c.copy("count1.txt", "count2.txt")
927 self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt 0:10:count2.txt\n", c.portable_manifest_text())
929 def test_copy_to_existing_dir(self):
930 c = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n./foo 781e5e245d69b566979b86e28d23f2c7+10 0:10:count2.txt\n')
931 c.copy("count1.txt", "foo")
932 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())
934 def test_copy_to_new_dir(self):
935 c = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n')
936 c.copy("count1.txt", "foo/")
937 self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n./foo 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n", c.portable_manifest_text())
939 def test_rename_file(self):
940 c = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n')
941 c.rename("count1.txt", "count2.txt")
942 self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count2.txt\n", c.manifest_text())
944 def test_move_file_to_dir(self):
945 c = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n')
947 c.rename("count1.txt", "foo/count2.txt")
948 self.assertEqual("./foo 781e5e245d69b566979b86e28d23f2c7+10 0:10:count2.txt\n", c.manifest_text())
950 def test_move_file_to_other(self):
951 c1 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n')
953 c2.rename("count1.txt", "count2.txt", source_collection=c1)
954 self.assertEqual("", c1.manifest_text())
955 self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count2.txt\n", c2.manifest_text())
957 def test_clone(self):
958 c = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n./foo 781e5e245d69b566979b86e28d23f2c7+10 0:10:count2.txt\n')
960 self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n./foo 781e5e245d69b566979b86e28d23f2c7+10 0:10:count2.txt\n", cl.portable_manifest_text())
962 def test_diff_del_add(self):
963 c1 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n')
964 c2 = Collection('. 5348b82a029fd9e971a811ce1f71360b+43 0:10:count2.txt\n')
966 self.assertEqual(sorted(d), [
967 ('add', './count1.txt', c1["count1.txt"]),
968 ('del', './count2.txt', c2["count2.txt"]),
971 self.assertEqual(sorted(d), [
972 ('add', './count2.txt', c2["count2.txt"]),
973 ('del', './count1.txt', c1["count1.txt"]),
975 self.assertNotEqual(c1.portable_manifest_text(), c2.portable_manifest_text())
977 self.assertEqual(c1.portable_manifest_text(), c2.portable_manifest_text())
979 def test_diff_same(self):
980 c1 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n')
981 c2 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n')
983 self.assertEqual(d, [('tok', './count1.txt', c2["count1.txt"], c1["count1.txt"])])
985 self.assertEqual(d, [('tok', './count1.txt', c2["count1.txt"], c1["count1.txt"])])
987 self.assertEqual(c1.portable_manifest_text(), c2.portable_manifest_text())
989 self.assertEqual(c1.portable_manifest_text(), c2.portable_manifest_text())
991 def test_diff_mod(self):
992 c1 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n')
993 c2 = Collection('. 5348b82a029fd9e971a811ce1f71360b+43 0:10:count1.txt\n')
995 self.assertEqual(d, [('mod', './count1.txt', c2["count1.txt"], c1["count1.txt"])])
997 self.assertEqual(d, [('mod', './count1.txt', c1["count1.txt"], c2["count1.txt"])])
999 self.assertNotEqual(c1.portable_manifest_text(), c2.portable_manifest_text())
1001 self.assertEqual(c1.portable_manifest_text(), c2.portable_manifest_text())
1003 def test_diff_add(self):
1004 c1 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n')
1005 c2 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 5348b82a029fd9e971a811ce1f71360b+43 0:10:count1.txt 10:20:count2.txt\n')
1007 self.assertEqual(sorted(d), [
1008 ('del', './count2.txt', c2["count2.txt"]),
1009 ('tok', './count1.txt', c2["count1.txt"], c1["count1.txt"]),
1012 self.assertEqual(sorted(d), [
1013 ('add', './count2.txt', c2["count2.txt"]),
1014 ('tok', './count1.txt', c2["count1.txt"], c1["count1.txt"]),
1017 self.assertNotEqual(c1.portable_manifest_text(), c2.portable_manifest_text())
1019 self.assertEqual(c1.portable_manifest_text(), c2.portable_manifest_text())
1021 def test_diff_add_in_subcollection(self):
1022 c1 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n')
1023 c2 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n./foo 5348b82a029fd9e971a811ce1f71360b+43 0:10:count2.txt\n')
1025 self.assertEqual(sorted(d), [
1026 ('del', './foo', c2["foo"]),
1027 ('tok', './count1.txt', c2["count1.txt"], c1["count1.txt"]),
1030 self.assertEqual(sorted(d), [
1031 ('add', './foo', c2["foo"]),
1032 ('tok', './count1.txt', c2["count1.txt"], c1["count1.txt"]),
1034 self.assertNotEqual(c1.portable_manifest_text(), c2.portable_manifest_text())
1036 self.assertEqual(c1.portable_manifest_text(), c2.portable_manifest_text())
1038 def test_diff_del_add_in_subcollection(self):
1039 c1 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n./foo 5348b82a029fd9e971a811ce1f71360b+43 0:10:count2.txt\n')
1040 c2 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n./foo 5348b82a029fd9e971a811ce1f71360b+43 0:3:count3.txt\n')
1042 self.assertEqual(sorted(d), [
1043 ('add', './foo/count2.txt', c1.find("foo/count2.txt")),
1044 ('del', './foo/count3.txt', c2.find("foo/count3.txt")),
1045 ('tok', './count1.txt', c2["count1.txt"], c1["count1.txt"]),
1048 self.assertEqual(sorted(d), [
1049 ('add', './foo/count3.txt', c2.find("foo/count3.txt")),
1050 ('del', './foo/count2.txt', c1.find("foo/count2.txt")),
1051 ('tok', './count1.txt', c2["count1.txt"], c1["count1.txt"]),
1054 self.assertNotEqual(c1.portable_manifest_text(), c2.portable_manifest_text())
1056 self.assertEqual(c1.portable_manifest_text(), c2.portable_manifest_text())
1058 def test_diff_mod_in_subcollection(self):
1059 c1 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n./foo 5348b82a029fd9e971a811ce1f71360b+43 0:10:count2.txt\n')
1060 c2 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt 0:3:foo\n')
1062 self.assertEqual(sorted(d), [
1063 ('mod', './foo', c2["foo"], c1["foo"]),
1064 ('tok', './count1.txt', c2["count1.txt"], c1["count1.txt"]),
1067 self.assertEqual(sorted(d), [
1068 ('mod', './foo', c1["foo"], c2["foo"]),
1069 ('tok', './count1.txt', c2["count1.txt"], c1["count1.txt"]),
1072 self.assertNotEqual(c1.portable_manifest_text(), c2.portable_manifest_text())
1074 self.assertEqual(c1.portable_manifest_text(), c2.portable_manifest_text())
1076 def test_conflict_keep_local_change(self):
1077 c1 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n')
1078 c2 = Collection('. 5348b82a029fd9e971a811ce1f71360b+43 0:10:count2.txt\n')
1080 self.assertEqual(sorted(d), [
1081 ('add', './count2.txt', c2["count2.txt"]),
1082 ('del', './count1.txt', c1["count1.txt"]),
1084 f = c1.open("count1.txt", "wb")
1087 # c1 changed, so it should not be deleted.
1089 self.assertEqual(c1.portable_manifest_text(), ". 95ebc3c7b3b9f1d2c40fec14415d3cb8+5 5348b82a029fd9e971a811ce1f71360b+43 0:5:count1.txt 5:10:count2.txt\n")
1091 def test_conflict_mod(self):
1092 c1 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt')
1093 c2 = Collection('. 5348b82a029fd9e971a811ce1f71360b+43 0:10:count1.txt')
1095 self.assertEqual(d, [('mod', './count1.txt', c1["count1.txt"], c2["count1.txt"])])
1096 f = c1.open("count1.txt", "wb")
1099 # c1 changed, so c2 mod will go to a conflict file
1102 c1.portable_manifest_text(),
1103 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~$")
1105 def test_conflict_add(self):
1106 c1 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count2.txt\n')
1107 c2 = Collection('. 5348b82a029fd9e971a811ce1f71360b+43 0:10:count1.txt\n')
1109 self.assertEqual(sorted(d), [
1110 ('add', './count1.txt', c2["count1.txt"]),
1111 ('del', './count2.txt', c1["count2.txt"]),
1113 f = c1.open("count1.txt", "wb")
1116 # c1 added count1.txt, so c2 add will go to a conflict file
1119 c1.portable_manifest_text(),
1120 r"\. 95ebc3c7b3b9f1d2c40fec14415d3cb8\+5 5348b82a029fd9e971a811ce1f71360b\+43 0:5:count1\.txt 5:10:count1\.txt~\d\d\d\d\d\d\d\d-\d\d\d\d\d\d~conflict~$")
1122 def test_conflict_del(self):
1123 c1 = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt')
1124 c2 = Collection('. 5348b82a029fd9e971a811ce1f71360b+43 0:10:count1.txt')
1126 self.assertEqual(d, [('mod', './count1.txt', c1["count1.txt"], c2["count1.txt"])])
1127 c1.remove("count1.txt")
1129 # c1 deleted, so c2 mod will go to a conflict file
1132 c1.portable_manifest_text(),
1133 r"\. 5348b82a029fd9e971a811ce1f71360b\+43 0:10:count1\.txt~\d\d\d\d\d\d\d\d-\d\d\d\d\d\d~conflict~$")
1135 def test_notify(self):
1138 c1.subscribe(lambda event, collection, name, item: events.append((event, collection, name, item)))
1139 f = c1.open("foo.txt", "wb")
1140 self.assertEqual(events[0], (arvados.collection.ADD, c1, "foo.txt", f.arvadosfile))
1142 def test_open_w(self):
1143 c1 = Collection(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n")
1144 self.assertEqual(c1["count1.txt"].size(), 10)
1145 c1.open("count1.txt", "wb").close()
1146 self.assertEqual(c1["count1.txt"].size(), 0)
1149 class NewCollectionTestCaseWithServers(run_test_server.TestCaseWithServers):
1150 def test_get_manifest_text_only_committed(self):
1152 with c.open("count.txt", "wb") as f:
1153 # One file committed
1154 with c.open("foo.txt", "wb") as foo:
1156 foo.flush() # Force block commit
1157 f.write(b"0123456789")
1158 # Other file not committed. Block not written to keep yet.
1160 c._get_manifest_text(".",
1163 only_committed=True),
1164 '. acbd18db4cc2f85cedef654fccc4a4d8+3 0:0:count.txt 0:3:foo.txt\n')
1165 # And now with the file closed...
1166 f.flush() # Force block commit
1168 c._get_manifest_text(".",
1171 only_committed=True),
1172 ". 781e5e245d69b566979b86e28d23f2c7+10 acbd18db4cc2f85cedef654fccc4a4d8+3 0:10:count.txt 10:3:foo.txt\n")
1174 def test_only_small_blocks_are_packed_together(self):
1176 # Write a couple of small files,
1177 f = c.open("count.txt", "wb")
1178 f.write(b"0123456789")
1179 f.close(flush=False)
1180 foo = c.open("foo.txt", "wb")
1182 foo.close(flush=False)
1183 # Then, write a big file, it shouldn't be packed with the ones above
1184 big = c.open("bigfile.txt", "wb")
1185 big.write(b"x" * 1024 * 1024 * 33) # 33 MB > KEEP_BLOCK_SIZE/2
1186 big.close(flush=False)
1188 c.manifest_text("."),
1189 '. 2d303c138c118af809f39319e5d507e9+34603008 a8430a058b8fbf408e1931b794dbd6fb+13 0:34603008:bigfile.txt 34603008:10:count.txt 34603018:3:foo.txt\n')
1192 class CollectionCreateUpdateTest(run_test_server.TestCaseWithServers):
1196 def create_count_txt(self):
1197 # Create an empty collection, save it to the API server, then write a
1198 # file, but don't save it.
1201 c.save_new("CollectionCreateUpdateTest", ensure_unique_name=True)
1202 self.assertEqual(c.portable_data_hash(), "d41d8cd98f00b204e9800998ecf8427e+0")
1203 self.assertEqual(c.api_response()["portable_data_hash"], "d41d8cd98f00b204e9800998ecf8427e+0" )
1205 with c.open("count.txt", "wb") as f:
1206 f.write(b"0123456789")
1208 self.assertEqual(c.portable_manifest_text(), ". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n")
1212 def test_create_and_save(self):
1213 c = self.create_count_txt()
1217 r"^\. 781e5e245d69b566979b86e28d23f2c7\+10\+A[a-f0-9]{40}@[a-f0-9]{8} 0:10:count\.txt$",)
1219 def test_create_and_save_new(self):
1220 c = self.create_count_txt()
1224 r"^\. 781e5e245d69b566979b86e28d23f2c7\+10\+A[a-f0-9]{40}@[a-f0-9]{8} 0:10:count\.txt$",)
1226 def test_create_diff_apply(self):
1227 c1 = self.create_count_txt()
1230 c2 = Collection(c1.manifest_locator())
1231 with c2.open("count.txt", "wb") as f:
1236 self.assertEqual(diff[0], (arvados.collection.MOD, u'./count.txt', c1["count.txt"], c2["count.txt"]))
1239 self.assertEqual(c1.portable_data_hash(), c2.portable_data_hash())
1241 def test_diff_apply_with_token(self):
1242 baseline = CollectionReader(". 781e5e245d69b566979b86e28d23f2c7+10+A715fd31f8111894f717eb1003c1b0216799dd9ec@54f5dd1a 0:10:count.txt\n")
1243 c = Collection(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n")
1244 other = CollectionReader(". 7ac66c0f148de9519b8bd264312c4d64+7+A715fd31f8111894f717eb1003c1b0216799dd9ec@54f5dd1a 0:7:count.txt\n")
1246 diff = baseline.diff(other)
1247 self.assertEqual(diff, [('mod', u'./count.txt', c["count.txt"], other["count.txt"])])
1251 self.assertEqual(c.manifest_text(), ". 7ac66c0f148de9519b8bd264312c4d64+7+A715fd31f8111894f717eb1003c1b0216799dd9ec@54f5dd1a 0:7:count.txt\n")
1254 def test_create_and_update(self):
1255 c1 = self.create_count_txt()
1258 c2 = arvados.collection.Collection(c1.manifest_locator())
1259 with c2.open("count.txt", "wb") as f:
1264 self.assertNotEqual(c1.portable_data_hash(), c2.portable_data_hash())
1266 self.assertEqual(c1.portable_data_hash(), c2.portable_data_hash())
1269 def test_create_and_update_with_conflict(self):
1270 c1 = self.create_count_txt()
1273 with c1.open("count.txt", "wb") as f:
1276 c2 = arvados.collection.Collection(c1.manifest_locator())
1277 with c2.open("count.txt", "wb") as f:
1285 r"\. e65075d550f9b5bf9992fa1d71a131be\+3\S* 7ac66c0f148de9519b8bd264312c4d64\+7\S* 0:3:count\.txt 3:7:count\.txt~\d\d\d\d\d\d\d\d-\d\d\d\d\d\d~conflict~$")
1288 if __name__ == '__main__':