3198: Fixed up the rest of the tests to mock the session.
[arvados.git] / 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 copy
7 import hashlib
8 import mock
9 import os
10 import pprint
11 import re
12 import tempfile
13 import unittest
14
15 import run_test_server
16 import arvados_testutil as tutil
17 from arvados.ranges import Range, LocatorAndRange
18
19 class TestResumableWriter(arvados.ResumableCollectionWriter):
20     KEEP_BLOCK_SIZE = 1024  # PUT to Keep every 1K.
21
22     def current_state(self):
23         return self.dump_state(copy.deepcopy)
24
25
26 class ArvadosCollectionsTest(run_test_server.TestCaseWithServers,
27                              tutil.ArvadosBaseTestCase):
28     MAIN_SERVER = {}
29
30     @classmethod
31     def setUpClass(cls):
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)
37
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')
43         cw.write('foo')
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')
47         cw.write('bar')
48         cw.start_new_stream('baz')
49         cw.write('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()))
55         cw.finish()
56         return cw.portable_data_hash()
57
58     def test_keep_local_store(self):
59         self.assertEqual(self.keep_client.put('foo'), 'acbd18db4cc2f85cedef654fccc4a4d8+3', 'wrong md5 hash from Keep.put')
60         self.assertEqual(self.keep_client.get('acbd18db4cc2f85cedef654fccc4a4d8+3'), 'foo', 'wrong data from Keep.get')
61
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())
66
67     def test_local_collection_reader(self):
68         foobarbaz = self.write_foo_bar_baz()
69         cr = arvados.CollectionReader(
70             foobarbaz + '+Xzizzle', self.api_client)
71         got = []
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', 'foo'],
76                     [3, '.', 'bar.txt', 'bar'],
77                     [3, './baz', 'baz.txt', 'baz']]
78         self.assertEqual(got,
79                          expected)
80         stream0 = cr.all_streams()[0]
81         self.assertEqual(stream0.readfrom(0, 0),
82                          '',
83                          'reading zero bytes should have returned empty string')
84         self.assertEqual(stream0.readfrom(0, 2**26),
85                          'foobar',
86                          'reading entire stream failed')
87         self.assertEqual(stream0.readfrom(2**26, 0),
88                          '',
89                          'reading zero bytes should have returned empty string')
90
91     def _test_subset(self, collection, expected):
92         cr = arvados.CollectionReader(collection, self.api_client)
93         for s in cr.all_streams():
94             for ex in expected:
95                 if ex[0] == s:
96                     f = s.files()[ex[2]]
97                     got = [f.size(), f.stream_name(), f.name(), "".join(f.readall(2**26))]
98                     self.assertEqual(got,
99                                      ex,
100                                      'all_files|as_manifest did not preserve manifest contents: got %s expected %s' % (got, ex))
101
102     def test_collection_manifest_subset(self):
103         foobarbaz = self.write_foo_bar_baz()
104         self._test_subset(foobarbaz,
105                           [[3, '.',     'bar.txt', 'bar'],
106                            [3, '.',     'foo.txt', 'foo'],
107                            [3, './baz', 'baz.txt', 'baz']])
108         self._test_subset((". %s %s 0:3:foo.txt 3:3:bar.txt\n" %
109                            (self.keep_client.put("foo"),
110                             self.keep_client.put("bar"))),
111                           [[3, '.', 'bar.txt', 'bar'],
112                            [3, '.', 'foo.txt', 'foo']])
113         self._test_subset((". %s %s 0:2:fo.txt 2:4:obar.txt\n" %
114                            (self.keep_client.put("foo"),
115                             self.keep_client.put("bar"))),
116                           [[2, '.', 'fo.txt', 'fo'],
117                            [4, '.', 'obar.txt', '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("foo"),
120                             self.keep_client.put("bar"))),
121                           [[2, '.', 'ar.txt', 'ar'],
122                            [2, '.', 'fo.txt', 'fo'],
123                            [2, '.', 'ob.txt', 'ob'],
124                            [0, '.', 'zero.txt', '']])
125
126     def test_collection_empty_file(self):
127         cw = arvados.CollectionWriter(self.api_client)
128         cw.start_new_file('zero.txt')
129         cw.write('')
130
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')
135         cw.write('')
136         cw.start_new_file('one.txt')
137         cw.write('1')
138         cw.start_new_stream('foo')
139         cw.start_new_file('zero.txt')
140         cw.write('')
141         self.check_manifest_file_sizes(cw.manifest_text(), [0,1,0])
142
143     def test_no_implicit_normalize(self):
144         cw = arvados.CollectionWriter(self.api_client)
145         cw.start_new_file('b')
146         cw.write('b')
147         cw.start_new_file('a')
148         cw.write('')
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),
153             [0,1])
154
155     def check_manifest_file_sizes(self, manifest_text, expect_sizes):
156         cr = arvados.CollectionReader(manifest_text, self.api_client)
157         got_sizes = []
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))
161
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
166 """
167         self.assertEqual(arvados.CollectionReader(m1, self.api_client).manifest_text(normalize=True),
168                          """. 5348b82a029fd9e971a811ce1f71360b+43 085c37f02916da1cad16f93c54d899b7+41 8b22da26f9f433dea0a10e5ec66d73ba+43 0:127:md5sum.txt
169 """)
170
171         m2 = """. 204e43b8a1185621ca55a94839582e6f+67108864 b9677abbac956bd3e86b1deb28dfac03+67108864 fc15aff2a762b13f521baf042140acec+67108864 323d2a3ce20370c4ca1d3462a344f8fd+25885655 0:227212247:var-GS000016015-ASM.tsv.bz2
172 """
173         self.assertEqual(arvados.CollectionReader(m2, self.api_client).manifest_text(normalize=True), m2)
174
175         m3 = """. 5348b82a029fd9e971a811ce1f71360b+43 3:40:md5sum.txt
176 . 085c37f02916da1cad16f93c54d899b7+41 0:41:md5sum.txt
177 . 8b22da26f9f433dea0a10e5ec66d73ba+43 0:43:md5sum.txt
178 """
179         self.assertEqual(arvados.CollectionReader(m3, self.api_client).manifest_text(normalize=True),
180                          """. 5348b82a029fd9e971a811ce1f71360b+43 085c37f02916da1cad16f93c54d899b7+41 8b22da26f9f433dea0a10e5ec66d73ba+43 3:124:md5sum.txt
181 """)
182
183         m4 = """. 204e43b8a1185621ca55a94839582e6f+67108864 0:3:foo/bar
184 ./zzz 204e43b8a1185621ca55a94839582e6f+67108864 0:999:zzz
185 ./foo 323d2a3ce20370c4ca1d3462a344f8fd+25885655 0:3:bar
186 """
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
190 """)
191
192         m5 = """. 204e43b8a1185621ca55a94839582e6f+67108864 0:3:foo/bar
193 ./zzz 204e43b8a1185621ca55a94839582e6f+67108864 0:999:zzz
194 ./foo 204e43b8a1185621ca55a94839582e6f+67108864 3:3:bar
195 """
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
199 """)
200
201         with self.data_file('1000G_ref_manifest') as f6:
202             m6 = f6.read()
203             self.assertEqual(arvados.CollectionReader(m6, self.api_client).manifest_text(normalize=True), m6)
204
205         with self.data_file('jlake_manifest') as f7:
206             m7 = f7.read()
207             self.assertEqual(arvados.CollectionReader(m7, self.api_client).manifest_text(normalize=True), m7)
208
209         m8 = """./a\\040b\\040c 59ca0efa9f5633cb0371bbc0355478d8+13 0:13:hello\\040world.txt
210 """
211         self.assertEqual(arvados.CollectionReader(m8, self.api_client).manifest_text(normalize=True), m8)
212
213     def test_locators_and_ranges(self):
214         blocks2 = [Range('a', 0, 10),
215                    Range('b', 10, 10),
216                    Range('c', 20, 10),
217                    Range('d', 30, 10),
218                    Range('e', 40, 10),
219                    Range('f', 50, 10)]
220
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), [])
229
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), [])
238
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)])
245
246
247         blocks3 = [Range('a', 0, 10),
248                   Range('b', 10, 10),
249                   Range('c', 20, 10),
250                   Range('d', 30, 10),
251                   Range('e', 40, 10),
252                   Range('f', 50, 10),
253                    Range('g', 60, 10)]
254
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)])
262
263
264         blocks = [Range('a', 0, 10),
265                   Range('b', 10, 15),
266                   Range('c', 25, 5)]
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)])
271
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)])
278
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)])
288
289         self.assertEqual(arvados.locators_and_ranges(blocks, 15, 5), [LocatorAndRange('b', 15, 5, 5)])
290
291         self.assertEqual(arvados.locators_and_ranges(blocks, 8, 17), [LocatorAndRange('a', 10, 8, 2),
292                                                                       LocatorAndRange('b', 15, 0, 15)])
293
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)])
297
298         self.assertEqual(arvados.locators_and_ranges(blocks, 26, 2), [LocatorAndRange('c', 5, 1, 2)])
299
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)])
305
306     class MockKeep(object):
307         def __init__(self, content, num_retries=0):
308             self.content = content
309
310         def get(self, locator, num_retries=0):
311             return self.content[locator]
312
313     def test_stream_reader(self):
314         keepblocks = {'aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa+10': 'abcdefghij',
315                       'bbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbb+15': 'klmnopqrstuvwxy',
316                       'cccccccccccccccccccccccccccccccc+5': 'z0123'}
317         mk = self.MockKeep(keepblocks)
318
319         sr = arvados.StreamReader([".", "aaaaaaaaaaaaaaaaaaaaaaaaaaaaaaaa+10", "bbbbbbbbbbbbbbbbbbbbbbbbbbbbbbbb+15", "cccccccccccccccccccccccccccccccc+5", "0:30:foo"], mk)
320
321         content = 'abcdefghijklmnopqrstuvwxyz0123456789'
322
323         self.assertEqual(sr.readfrom(0, 30), content[0:30])
324         self.assertEqual(sr.readfrom(2, 30), content[2:30])
325
326         self.assertEqual(sr.readfrom(2, 8), content[2:10])
327         self.assertEqual(sr.readfrom(0, 10), content[0:10])
328
329         self.assertEqual(sr.readfrom(0, 5), content[0:5])
330         self.assertEqual(sr.readfrom(5, 5), content[5:10])
331         self.assertEqual(sr.readfrom(10, 5), content[10:15])
332         self.assertEqual(sr.readfrom(15, 5), content[15:20])
333         self.assertEqual(sr.readfrom(20, 5), content[20:25])
334         self.assertEqual(sr.readfrom(25, 5), content[25:30])
335         self.assertEqual(sr.readfrom(30, 5), '')
336
337     def test_extract_file(self):
338         m1 = """. 5348b82a029fd9e971a811ce1f71360b+43 0:43:md5sum.txt
339 . 085c37f02916da1cad16f93c54d899b7+41 0:41:md6sum.txt
340 . 8b22da26f9f433dea0a10e5ec66d73ba+43 0:43:md7sum.txt
341 . 085c37f02916da1cad16f93c54d899b7+41 5348b82a029fd9e971a811ce1f71360b+43 8b22da26f9f433dea0a10e5ec66d73ba+43 47:80:md8sum.txt
342 . 085c37f02916da1cad16f93c54d899b7+41 5348b82a029fd9e971a811ce1f71360b+43 8b22da26f9f433dea0a10e5ec66d73ba+43 40:80:md9sum.txt
343 """
344
345         m2 = arvados.CollectionReader(m1, self.api_client).manifest_text(normalize=True)
346
347         self.assertEqual(m2,
348                          ". 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")
349         files = arvados.CollectionReader(
350             m2, self.api_client).all_streams()[0].files()
351
352         self.assertEqual(files['md5sum.txt'].as_manifest(),
353                          ". 5348b82a029fd9e971a811ce1f71360b+43 0:43:md5sum.txt\n")
354         self.assertEqual(files['md6sum.txt'].as_manifest(),
355                          ". 085c37f02916da1cad16f93c54d899b7+41 0:41:md6sum.txt\n")
356         self.assertEqual(files['md7sum.txt'].as_manifest(),
357                          ". 8b22da26f9f433dea0a10e5ec66d73ba+43 0:43:md7sum.txt\n")
358         self.assertEqual(files['md9sum.txt'].as_manifest(),
359                          ". 085c37f02916da1cad16f93c54d899b7+41 5348b82a029fd9e971a811ce1f71360b+43 8b22da26f9f433dea0a10e5ec66d73ba+43 40:80:md9sum.txt\n")
360
361     def test_write_directory_tree(self):
362         cwriter = arvados.CollectionWriter(self.api_client)
363         cwriter.write_directory_tree(self.build_directory_tree(
364                 ['basefile', 'subdir/subfile']))
365         self.assertEqual(cwriter.manifest_text(),
366                          """. c5110c5ac93202d8e0f9e381f22bac0f+8 0:8:basefile
367 ./subdir 1ca4dec89403084bf282ad31e6cf7972+14 0:14:subfile\n""")
368
369     def test_write_named_directory_tree(self):
370         cwriter = arvados.CollectionWriter(self.api_client)
371         cwriter.write_directory_tree(self.build_directory_tree(
372                 ['basefile', 'subdir/subfile']), 'root')
373         self.assertEqual(
374             cwriter.manifest_text(),
375             """./root c5110c5ac93202d8e0f9e381f22bac0f+8 0:8:basefile
376 ./root/subdir 1ca4dec89403084bf282ad31e6cf7972+14 0:14:subfile\n""")
377
378     def test_write_directory_tree_in_one_stream(self):
379         cwriter = arvados.CollectionWriter(self.api_client)
380         cwriter.write_directory_tree(self.build_directory_tree(
381                 ['basefile', 'subdir/subfile']), max_manifest_depth=0)
382         self.assertEqual(cwriter.manifest_text(),
383                          """. 4ace875ffdc6824a04950f06858f4465+22 0:8:basefile 8:14:subdir/subfile\n""")
384
385     def test_write_directory_tree_with_limited_recursion(self):
386         cwriter = arvados.CollectionWriter(self.api_client)
387         cwriter.write_directory_tree(
388             self.build_directory_tree(['f1', 'd1/f2', 'd1/d2/f3']),
389             max_manifest_depth=1)
390         self.assertEqual(cwriter.manifest_text(),
391                          """. bd19836ddb62c11c55ab251ccaca5645+2 0:2:f1
392 ./d1 50170217e5b04312024aa5cd42934494+13 0:8:d2/f3 8:5:f2\n""")
393
394     def test_write_directory_tree_with_zero_recursion(self):
395         cwriter = arvados.CollectionWriter(self.api_client)
396         content = 'd1/d2/f3d1/f2f1'
397         blockhash = hashlib.md5(content).hexdigest() + '+' + str(len(content))
398         cwriter.write_directory_tree(
399             self.build_directory_tree(['f1', 'd1/f2', 'd1/d2/f3']),
400             max_manifest_depth=0)
401         self.assertEqual(
402             cwriter.manifest_text(),
403             ". {} 0:8:d1/d2/f3 8:5:d1/f2 13:2:f1\n".format(blockhash))
404
405     def test_write_one_file(self):
406         cwriter = arvados.CollectionWriter(self.api_client)
407         with self.make_test_file() as testfile:
408             cwriter.write_file(testfile.name)
409             self.assertEqual(
410                 cwriter.manifest_text(),
411                 ". 098f6bcd4621d373cade4e832627b4f6+4 0:4:{}\n".format(
412                     os.path.basename(testfile.name)))
413
414     def test_write_named_file(self):
415         cwriter = arvados.CollectionWriter(self.api_client)
416         with self.make_test_file() as testfile:
417             cwriter.write_file(testfile.name, 'foo')
418             self.assertEqual(cwriter.manifest_text(),
419                              ". 098f6bcd4621d373cade4e832627b4f6+4 0:4:foo\n")
420
421     def test_write_multiple_files(self):
422         cwriter = arvados.CollectionWriter(self.api_client)
423         for letter in 'ABC':
424             with self.make_test_file(letter) as testfile:
425                 cwriter.write_file(testfile.name, letter)
426         self.assertEqual(
427             cwriter.manifest_text(),
428             ". 902fbdd2b1df0c4f70b4a5d23525e932+3 0:1:A 1:1:B 2:1:C\n")
429
430     def test_basic_resume(self):
431         cwriter = TestResumableWriter()
432         with self.make_test_file() as testfile:
433             cwriter.write_file(testfile.name, 'test')
434             resumed = TestResumableWriter.from_state(cwriter.current_state())
435         self.assertEquals(cwriter.manifest_text(), resumed.manifest_text(),
436                           "resumed CollectionWriter had different manifest")
437
438     def test_resume_fails_when_missing_dependency(self):
439         cwriter = TestResumableWriter()
440         with self.make_test_file() as testfile:
441             cwriter.write_file(testfile.name, 'test')
442         self.assertRaises(arvados.errors.StaleWriterStateError,
443                           TestResumableWriter.from_state,
444                           cwriter.current_state())
445
446     def test_resume_fails_when_dependency_mtime_changed(self):
447         cwriter = TestResumableWriter()
448         with self.make_test_file() as testfile:
449             cwriter.write_file(testfile.name, 'test')
450             os.utime(testfile.name, (0, 0))
451             self.assertRaises(arvados.errors.StaleWriterStateError,
452                               TestResumableWriter.from_state,
453                               cwriter.current_state())
454
455     def test_resume_fails_when_dependency_is_nonfile(self):
456         cwriter = TestResumableWriter()
457         cwriter.write_file('/dev/null', 'empty')
458         self.assertRaises(arvados.errors.StaleWriterStateError,
459                           TestResumableWriter.from_state,
460                           cwriter.current_state())
461
462     def test_resume_fails_when_dependency_size_changed(self):
463         cwriter = TestResumableWriter()
464         with self.make_test_file() as testfile:
465             cwriter.write_file(testfile.name, 'test')
466             orig_mtime = os.fstat(testfile.fileno()).st_mtime
467             testfile.write('extra')
468             testfile.flush()
469             os.utime(testfile.name, (orig_mtime, orig_mtime))
470             self.assertRaises(arvados.errors.StaleWriterStateError,
471                               TestResumableWriter.from_state,
472                               cwriter.current_state())
473
474     def test_resume_fails_with_expired_locator(self):
475         cwriter = TestResumableWriter()
476         state = cwriter.current_state()
477         # Add an expired locator to the state.
478         state['_current_stream_locators'].append(''.join([
479                     'a' * 32, '+1+A', 'b' * 40, '@', '10000000']))
480         self.assertRaises(arvados.errors.StaleWriterStateError,
481                           TestResumableWriter.from_state, state)
482
483     def test_arbitrary_objects_not_resumable(self):
484         cwriter = TestResumableWriter()
485         with open('/dev/null') as badfile:
486             self.assertRaises(arvados.errors.AssertionError,
487                               cwriter.write_file, badfile)
488
489     def test_arbitrary_writes_not_resumable(self):
490         cwriter = TestResumableWriter()
491         self.assertRaises(arvados.errors.AssertionError,
492                           cwriter.write, "badtext")
493
494     def test_read_arbitrary_data_with_collection_reader(self):
495         # arv-get relies on this to do "arv-get {keep-locator} -".
496         self.write_foo_bar_baz()
497         self.assertEqual(
498             'foobar',
499             arvados.CollectionReader(
500                 '3858f62230ac3c915f300c664312c63f+6'
501                 ).manifest_text())
502
503
504 class CollectionTestMixin(object):
505     PROXY_RESPONSE = {
506         'items_available': 1,
507         'items': [{
508                 'uuid': 'zzzzz-bi6l4-mockproxy012345',
509                 'owner_uuid': 'zzzzz-tpzed-mockowner012345',
510                 'service_host': tutil.TEST_HOST,
511                 'service_port': 65535,
512                 'service_ssl_flag': True,
513                 'service_type': 'proxy',
514                 }]}
515     API_COLLECTIONS = run_test_server.fixture('collections')
516     DEFAULT_COLLECTION = API_COLLECTIONS['foo_file']
517     DEFAULT_DATA_HASH = DEFAULT_COLLECTION['portable_data_hash']
518     DEFAULT_MANIFEST = DEFAULT_COLLECTION['manifest_text']
519     DEFAULT_UUID = DEFAULT_COLLECTION['uuid']
520     ALT_COLLECTION = API_COLLECTIONS['bar_file']
521     ALT_DATA_HASH = ALT_COLLECTION['portable_data_hash']
522     ALT_MANIFEST = ALT_COLLECTION['manifest_text']
523
524     def _mock_api_call(self, mock_method, code, body):
525         mock_method = mock_method().execute
526         if code == 200:
527             mock_method.return_value = body
528         else:
529             mock_method.side_effect = arvados.errors.ApiError(
530                 tutil.fake_httplib2_response(code), "{}")
531
532     def mock_keep_services(self, api_mock, code, body):
533         self._mock_api_call(api_mock.keep_services().accessible, code, body)
534
535     def api_client_mock(self, code=200):
536         client = mock.MagicMock(name='api_client')
537         self.mock_keep_services(client, code, self.PROXY_RESPONSE)
538         return client
539
540
541 @tutil.skip_sleep
542 class CollectionReaderTestCase(unittest.TestCase, CollectionTestMixin):
543     def mock_get_collection(self, api_mock, code, body):
544         body = self.API_COLLECTIONS.get(body)
545         self._mock_api_call(api_mock.collections().get, code, body)
546
547     def api_client_mock(self, code=200):
548         client = super(CollectionReaderTestCase, self).api_client_mock(code)
549         self.mock_get_collection(client, code, 'foo_file')
550         return client
551
552     def test_init_no_default_retries(self):
553         client = self.api_client_mock(200)
554         reader = arvados.CollectionReader(self.DEFAULT_UUID, api_client=client)
555         reader.manifest_text()
556         client.collections().get().execute.assert_called_with(num_retries=0)
557
558     def test_uuid_init_success(self):
559         client = self.api_client_mock(200)
560         reader = arvados.CollectionReader(self.DEFAULT_UUID, api_client=client,
561                                           num_retries=3)
562         self.assertEqual(self.DEFAULT_COLLECTION['manifest_text'],
563                          reader.manifest_text())
564         client.collections().get().execute.assert_called_with(num_retries=3)
565
566     def test_uuid_init_failure_raises_api_error(self):
567         client = self.api_client_mock(500)
568         reader = arvados.CollectionReader(self.DEFAULT_UUID, api_client=client)
569         with self.assertRaises(arvados.errors.ApiError):
570             reader.manifest_text()
571
572     def test_locator_init(self):
573         client = self.api_client_mock(200)
574         # Ensure Keep will not return anything if asked.
575         with tutil.mock_get_responses(None, 404):
576             reader = arvados.CollectionReader(self.DEFAULT_DATA_HASH,
577                                               api_client=client)
578             self.assertEqual(self.DEFAULT_MANIFEST, reader.manifest_text())
579
580     def test_locator_init_fallback_to_keep(self):
581         # crunch-job needs this to read manifests that have only ever
582         # been written to Keep.
583         client = self.api_client_mock(200)
584         self.mock_get_collection(client, 404, None)
585         with tutil.mock_get_responses(self.DEFAULT_MANIFEST, 200):
586             reader = arvados.CollectionReader(self.DEFAULT_DATA_HASH,
587                                               api_client=client)
588             self.assertEqual(self.DEFAULT_MANIFEST, reader.manifest_text())
589
590     def test_uuid_init_no_fallback_to_keep(self):
591         # Do not look up a collection UUID in Keep.
592         client = self.api_client_mock(404)
593         reader = arvados.CollectionReader(self.DEFAULT_UUID,
594                                           api_client=client)
595         with tutil.mock_get_responses(self.DEFAULT_MANIFEST, 200):
596             with self.assertRaises(arvados.errors.ApiError):
597                 reader.manifest_text()
598
599     def test_try_keep_first_if_permission_hint(self):
600         # To verify that CollectionReader tries Keep first here, we
601         # mock API server to return the wrong data.
602         client = self.api_client_mock(200)
603         with tutil.mock_get_responses(self.ALT_MANIFEST, 200):
604             self.assertEqual(
605                 self.ALT_MANIFEST,
606                 arvados.CollectionReader(
607                     self.ALT_DATA_HASH + '+Affffffffffffffffffffffffffffffffffffffff@fedcba98',
608                     api_client=client).manifest_text())
609
610     def test_init_num_retries_propagated(self):
611         # More of an integration test...
612         client = self.api_client_mock(200)
613         reader = arvados.CollectionReader(self.DEFAULT_UUID, api_client=client,
614                                           num_retries=3)
615         with tutil.mock_get_responses('foo', 500, 500, 200):
616             self.assertEqual('foo',
617                              ''.join(f.read(9) for f in reader.all_files()))
618
619     def test_read_nonnormalized_manifest_with_collection_reader(self):
620         # client should be able to use CollectionReader on a manifest without normalizing it
621         client = self.api_client_mock(500)
622         nonnormal = ". acbd18db4cc2f85cedef654fccc4a4d8+3+Aabadbadbee@abeebdee 0:3:foo.txt 1:0:bar.txt 0:3:foo.txt\n"
623         reader = arvados.CollectionReader(
624             nonnormal,
625             api_client=client, num_retries=0)
626         # Ensure stripped_manifest() doesn't mangle our manifest in
627         # any way other than stripping hints.
628         self.assertEqual(
629             re.sub('\+[^\d\s\+]+', '', nonnormal),
630             reader.stripped_manifest())
631         # Ensure stripped_manifest() didn't mutate our reader.
632         self.assertEqual(nonnormal, reader.manifest_text())
633         # Ensure the files appear in the order given in the manifest.
634         self.assertEqual(
635             [[6, '.', 'foo.txt'],
636              [0, '.', 'bar.txt']],
637             [[f.size(), f.stream_name(), f.name()]
638              for f in reader.all_streams()[0].all_files()])
639
640     def test_read_empty_collection(self):
641         client = self.api_client_mock(200)
642         self.mock_get_collection(client, 200, 'empty')
643         reader = arvados.CollectionReader('d41d8cd98f00b204e9800998ecf8427e+0',
644                                           api_client=client)
645         self.assertEqual('', reader.manifest_text())
646
647     def test_api_response(self):
648         client = self.api_client_mock()
649         reader = arvados.CollectionReader(self.DEFAULT_UUID, api_client=client)
650         self.assertEqual(self.DEFAULT_COLLECTION, reader.api_response())
651
652     def test_api_response_with_collection_from_keep(self):
653         client = self.api_client_mock()
654         self.mock_get_collection(client, 404, 'foo')
655         with tutil.mock_get_responses(self.DEFAULT_MANIFEST, 200):
656             reader = arvados.CollectionReader(self.DEFAULT_DATA_HASH,
657                                               api_client=client)
658             api_response = reader.api_response()
659         self.assertIsNone(api_response)
660
661     def check_open_file(self, coll_file, stream_name, file_name, file_size):
662         self.assertFalse(coll_file.closed, "returned file is not open")
663         self.assertEqual(stream_name, coll_file.stream_name())
664         self.assertEqual(file_name, coll_file.name())
665         self.assertEqual(file_size, coll_file.size())
666
667     def test_open_collection_file_one_argument(self):
668         client = self.api_client_mock(200)
669         reader = arvados.CollectionReader(self.DEFAULT_UUID, api_client=client)
670         cfile = reader.open('./foo')
671         self.check_open_file(cfile, '.', 'foo', 3)
672
673     def test_open_collection_file_two_arguments(self):
674         client = self.api_client_mock(200)
675         reader = arvados.CollectionReader(self.DEFAULT_UUID, api_client=client)
676         cfile = reader.open('.', 'foo')
677         self.check_open_file(cfile, '.', 'foo', 3)
678
679     def test_open_deep_file(self):
680         coll_name = 'collection_with_files_in_subdir'
681         client = self.api_client_mock(200)
682         self.mock_get_collection(client, 200, coll_name)
683         reader = arvados.CollectionReader(
684             self.API_COLLECTIONS[coll_name]['uuid'], api_client=client)
685         cfile = reader.open('./subdir2/subdir3/file2_in_subdir3.txt')
686         self.check_open_file(cfile, './subdir2/subdir3', 'file2_in_subdir3.txt',
687                              32)
688
689     def test_open_nonexistent_stream(self):
690         client = self.api_client_mock(200)
691         reader = arvados.CollectionReader(self.DEFAULT_UUID, api_client=client)
692         self.assertRaises(ValueError, reader.open, './nonexistent', 'foo')
693
694     def test_open_nonexistent_file(self):
695         client = self.api_client_mock(200)
696         reader = arvados.CollectionReader(self.DEFAULT_UUID, api_client=client)
697         self.assertRaises(ValueError, reader.open, '.', 'nonexistent')
698
699
700 @tutil.skip_sleep
701 class CollectionWriterTestCase(unittest.TestCase, CollectionTestMixin):
702     def mock_keep(self, body, *codes, **headers):
703         headers.setdefault('x-keep-replicas-stored', 2)
704         return tutil.mock_put_responses(body, *codes, **headers)
705
706     def foo_writer(self, **kwargs):
707         api_client = self.api_client_mock()
708         writer = arvados.CollectionWriter(api_client, **kwargs)
709         writer.start_new_file('foo')
710         writer.write('foo')
711         return writer
712
713     def test_write_whole_collection(self):
714         writer = self.foo_writer()
715         with self.mock_keep(self.DEFAULT_DATA_HASH, 200, 200):
716             self.assertEqual(self.DEFAULT_DATA_HASH, writer.finish())
717
718     def test_write_no_default(self):
719         writer = self.foo_writer()
720         with self.mock_keep(None, 500):
721             with self.assertRaises(arvados.errors.KeepWriteError):
722                 writer.finish()
723
724     def test_write_whole_collection_through_retries(self):
725         writer = self.foo_writer(num_retries=2)
726         with self.mock_keep(self.DEFAULT_DATA_HASH,
727                             500, 500, 200, 500, 500, 200):
728             self.assertEqual(self.DEFAULT_DATA_HASH, writer.finish())
729
730     def test_flush_data_retries(self):
731         writer = self.foo_writer(num_retries=2)
732         foo_hash = self.DEFAULT_MANIFEST.split()[1]
733         with self.mock_keep(foo_hash, 500, 200):
734             writer.flush_data()
735         self.assertEqual(self.DEFAULT_MANIFEST, writer.manifest_text())
736
737     def test_one_open(self):
738         client = self.api_client_mock()
739         writer = arvados.CollectionWriter(client)
740         with writer.open('out') as out_file:
741             self.assertEqual('.', writer.current_stream_name())
742             self.assertEqual('out', writer.current_file_name())
743             out_file.write('test data')
744             data_loc = hashlib.md5('test data').hexdigest() + '+9'
745         self.assertTrue(out_file.closed, "writer file not closed after context")
746         self.assertRaises(ValueError, out_file.write, 'extra text')
747         with self.mock_keep(data_loc, 200) as keep_mock:
748             self.assertEqual(". {} 0:9:out\n".format(data_loc),
749                              writer.manifest_text())
750
751     def test_open_writelines(self):
752         client = self.api_client_mock()
753         writer = arvados.CollectionWriter(client)
754         with writer.open('six') as out_file:
755             out_file.writelines(['12', '34', '56'])
756             data_loc = hashlib.md5('123456').hexdigest() + '+6'
757         with self.mock_keep(data_loc, 200) as keep_mock:
758             self.assertEqual(". {} 0:6:six\n".format(data_loc),
759                              writer.manifest_text())
760
761     def test_open_flush(self):
762         client = self.api_client_mock()
763         data_loc1 = hashlib.md5('flush1').hexdigest() + '+6'
764         data_loc2 = hashlib.md5('flush2').hexdigest() + '+6'
765         with self.mock_keep((data_loc1, 200), (data_loc2, 200)) as keep_mock:
766             writer = arvados.CollectionWriter(client)
767             with writer.open('flush_test') as out_file:
768                 out_file.write('flush1')
769                 out_file.flush()
770                 out_file.write('flush2')
771             self.assertEqual(". {} {} 0:12:flush_test\n".format(data_loc1,
772                                                                 data_loc2),
773                              writer.manifest_text())
774
775     def test_two_opens_same_stream(self):
776         client = self.api_client_mock()
777         writer = arvados.CollectionWriter(client)
778         with writer.open('.', '1') as out_file:
779             out_file.write('1st')
780         with writer.open('.', '2') as out_file:
781             out_file.write('2nd')
782         data_loc = hashlib.md5('1st2nd').hexdigest() + '+6'
783         with self.mock_keep(data_loc, 200) as keep_mock:
784             self.assertEqual(". {} 0:3:1 3:3:2\n".format(data_loc),
785                              writer.manifest_text())
786
787     def test_two_opens_two_streams(self):
788         client = self.api_client_mock()
789         data_loc1 = hashlib.md5('file').hexdigest() + '+4'
790         data_loc2 = hashlib.md5('indir').hexdigest() + '+5'
791         with self.mock_keep((data_loc1, 200), (data_loc2, 200)) as keep_mock:
792             writer = arvados.CollectionWriter(client)
793             with writer.open('file') as out_file:
794                 out_file.write('file')
795             with writer.open('./dir', 'indir') as out_file:
796                 out_file.write('indir')
797             expected = ". {} 0:4:file\n./dir {} 0:5:indir\n".format(
798                 data_loc1, data_loc2)
799             self.assertEqual(expected, writer.manifest_text())
800
801     def test_dup_open_fails(self):
802         client = self.api_client_mock()
803         writer = arvados.CollectionWriter(client)
804         file1 = writer.open('one')
805         self.assertRaises(arvados.errors.AssertionError, writer.open, 'two')
806
807
808 class NewCollectionTestCase(unittest.TestCase, CollectionTestMixin):
809     def test_import_manifest(self):
810         m1 = """. 5348b82a029fd9e971a811ce1f71360b+43 0:43:md5sum.txt
811 . 085c37f02916da1cad16f93c54d899b7+41 0:41:md5sum.txt
812 . 8b22da26f9f433dea0a10e5ec66d73ba+43 0:43:md5sum.txt
813 """
814         self.assertEqual(". 5348b82a029fd9e971a811ce1f71360b+43 085c37f02916da1cad16f93c54d899b7+41 8b22da26f9f433dea0a10e5ec66d73ba+43 0:127:md5sum.txt\n", arvados.export_manifest(arvados.import_manifest(m1)))
815
816     def test_init_manifest(self):
817         m1 = """. 5348b82a029fd9e971a811ce1f71360b+43 0:43:md5sum.txt
818 . 085c37f02916da1cad16f93c54d899b7+41 0:41:md5sum.txt
819 . 8b22da26f9f433dea0a10e5ec66d73ba+43 0:43:md5sum.txt
820 """
821         self.assertEqual(". 5348b82a029fd9e971a811ce1f71360b+43 085c37f02916da1cad16f93c54d899b7+41 8b22da26f9f433dea0a10e5ec66d73ba+43 0:127:md5sum.txt\n", arvados.export_manifest(arvados.Collection(m1)))
822
823 if __name__ == '__main__':
824     unittest.main()