11308: Remove unused import.
[arvados.git] / sdk / python / tests / test_arvfile.py
1 from __future__ import absolute_import
2 from builtins import hex
3 from builtins import str
4 from builtins import range
5 from builtins import object
6 import bz2
7 import datetime
8 import gzip
9 import io
10 import mock
11 import os
12 import unittest
13 import time
14
15 import arvados
16 from arvados._ranges import Range
17 from arvados.keep import KeepLocator
18 from arvados.collection import Collection, CollectionReader
19 from arvados.arvfile import ArvadosFile, ArvadosFileReader
20
21 from . import arvados_testutil as tutil
22 from .test_stream import StreamFileReaderTestCase, StreamRetryTestMixin
23
24 class ArvadosFileWriterTestCase(unittest.TestCase):
25     class MockKeep(object):
26         def __init__(self, blocks):
27             self.blocks = blocks
28             self.requests = []
29         def get(self, locator, num_retries=0):
30             self.requests.append(locator)
31             return self.blocks.get(locator)
32         def get_from_cache(self, locator):
33             self.requests.append(locator)
34             return self.blocks.get(locator)
35         def put(self, data, num_retries=None, copies=None):
36             pdh = tutil.str_keep_locator(data)
37             self.blocks[pdh] = bytes(data)
38             return pdh
39
40     class MockApi(object):
41         def __init__(self, b, r):
42             self.body = b
43             self.response = r
44             self._schema = ArvadosFileWriterTestCase.MockApi.MockSchema()
45             self._rootDesc = {}
46         class MockSchema(object):
47             def __init__(self):
48                 self.schemas = {'Collection': {'properties': {'replication_desired': {'type':'integer'}}}}
49         class MockCollections(object):
50             def __init__(self, b, r):
51                 self.body = b
52                 self.response = r
53             class Execute(object):
54                 def __init__(self, r):
55                     self.response = r
56                 def execute(self, num_retries=None):
57                     return self.response
58             def create(self, ensure_unique_name=False, body=None):
59                 if body != self.body:
60                     raise Exception("Body %s does not match expectation %s" % (body, self.body))
61                 return ArvadosFileWriterTestCase.MockApi.MockCollections.Execute(self.response)
62             def update(self, uuid=None, body=None):
63                 return ArvadosFileWriterTestCase.MockApi.MockCollections.Execute(self.response)
64         def collections(self):
65             return ArvadosFileWriterTestCase.MockApi.MockCollections(self.body, self.response)
66
67
68     def test_truncate(self):
69         keep = ArvadosFileWriterTestCase.MockKeep({
70             "781e5e245d69b566979b86e28d23f2c7+10": b"0123456789",
71         })
72         api = ArvadosFileWriterTestCase.MockApi({
73             "name": "test_truncate",
74             "manifest_text": ". 781e5e245d69b566979b86e28d23f2c7+10 0:8:count.txt\n",
75             "replication_desired": None,
76         }, {
77             "uuid": "zzzzz-4zz18-mockcollection0",
78             "manifest_text": ". 781e5e245d69b566979b86e28d23f2c7+10 0:8:count.txt\n",
79             "portable_data_hash":"7fcd0eaac3aad4c31a6a0e756475da92+52",
80         })
81         with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n',
82                              api_client=api, keep_client=keep) as c:
83             writer = c.open("count.txt", "rb+")
84             self.assertEqual(writer.size(), 10)
85             self.assertEqual(b"0123456789", writer.read(12))
86
87             writer.truncate(8)
88
89             # Make sure reading off the end doesn't break
90             self.assertEqual(b"", writer.read(12))
91
92             self.assertEqual(writer.size(), 8)
93             writer.seek(0, os.SEEK_SET)
94             self.assertEqual(b"01234567", writer.read(12))
95
96             self.assertIsNone(c.manifest_locator())
97             self.assertTrue(c.modified())
98             c.save_new("test_truncate")
99             self.assertEqual("zzzzz-4zz18-mockcollection0", c.manifest_locator())
100             self.assertFalse(c.modified())
101
102     def test_write_to_end(self):
103         keep = ArvadosFileWriterTestCase.MockKeep({
104             "781e5e245d69b566979b86e28d23f2c7+10": b"0123456789",
105         })
106         api = ArvadosFileWriterTestCase.MockApi({
107             "name": "test_append",
108             "manifest_text": ". 781e5e245d69b566979b86e28d23f2c7+10 acbd18db4cc2f85cedef654fccc4a4d8+3 0:13:count.txt\n",
109             "replication_desired": None,
110         }, {
111             "uuid": "zzzzz-4zz18-mockcollection0",
112             "manifest_text": ". 781e5e245d69b566979b86e28d23f2c7+10 acbd18db4cc2f85cedef654fccc4a4d8+3 0:13:count.txt\n",
113             "portable_data_hash": "c5c3af76565c8efb6a806546bcf073f3+88",
114         })
115         with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n',
116                              api_client=api, keep_client=keep) as c:
117             writer = c.open("count.txt", "rb+")
118             self.assertEqual(writer.size(), 10)
119
120             writer.seek(5, os.SEEK_SET)
121             self.assertEqual(b"56789", writer.read(8))
122
123             writer.seek(10, os.SEEK_SET)
124             writer.write("foo")
125             self.assertEqual(writer.size(), 13)
126
127             writer.seek(5, os.SEEK_SET)
128             self.assertEqual(b"56789foo", writer.read(8))
129
130             self.assertIsNone(c.manifest_locator())
131             self.assertTrue(c.modified())
132             self.assertIsNone(keep.get("acbd18db4cc2f85cedef654fccc4a4d8+3"))
133
134             c.save_new("test_append")
135             self.assertEqual("zzzzz-4zz18-mockcollection0", c.manifest_locator())
136             self.assertFalse(c.modified())
137             self.assertEqual(b"foo", keep.get("acbd18db4cc2f85cedef654fccc4a4d8+3"))
138
139
140     def test_append(self):
141         keep = ArvadosFileWriterTestCase.MockKeep({
142             "781e5e245d69b566979b86e28d23f2c7+10": b"0123456789",
143         })
144         c = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n', keep_client=keep)
145         writer = c.open("count.txt", "ab+")
146         self.assertEqual(writer.read(20), b"0123456789")
147         writer.seek(0, os.SEEK_SET)
148
149         writer.write("hello")
150         self.assertEqual(writer.read(20), b"0123456789hello")
151         writer.seek(0, os.SEEK_SET)
152
153         writer.write("world")
154         self.assertEqual(writer.read(20), b"0123456789helloworld")
155
156         self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 fc5e038d38a57032085441e7fe7010b0+10 0:20:count.txt\n", c.portable_manifest_text())
157
158     def test_write_at_beginning(self):
159         keep = ArvadosFileWriterTestCase.MockKeep({
160             "781e5e245d69b566979b86e28d23f2c7+10": b"0123456789",
161         })
162         with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n',
163                              keep_client=keep) as c:
164             writer = c.open("count.txt", "rb+")
165             self.assertEqual(b"0123456789", writer.readfrom(0, 13))
166             writer.seek(0, os.SEEK_SET)
167             writer.write("foo")
168             self.assertEqual(writer.size(), 10)
169             self.assertEqual(b"foo3456789", writer.readfrom(0, 13))
170             self.assertEqual(". acbd18db4cc2f85cedef654fccc4a4d8+3 781e5e245d69b566979b86e28d23f2c7+10 0:3:count.txt 6:7:count.txt\n", c.portable_manifest_text())
171
172     def test_write_empty(self):
173         keep = ArvadosFileWriterTestCase.MockKeep({})
174         with Collection(keep_client=keep) as c:
175             writer = c.open("count.txt", "wb")
176             self.assertEqual(writer.size(), 0)
177             self.assertEqual(". d41d8cd98f00b204e9800998ecf8427e+0 0:0:count.txt\n", c.portable_manifest_text())
178
179     def test_save_manifest_text(self):
180         keep = ArvadosFileWriterTestCase.MockKeep({})
181         with Collection(keep_client=keep) as c:
182             writer = c.open("count.txt", "wb")
183             writer.write(b"0123456789")
184             self.assertEqual('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n', c.portable_manifest_text())
185             self.assertNotIn('781e5e245d69b566979b86e28d23f2c7+10', keep.blocks)
186
187             self.assertEqual('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n', c.save_new(create_collection_record=False))
188             self.assertIn('781e5e245d69b566979b86e28d23f2c7+10', keep.blocks)
189
190     def test_get_manifest_text_commits(self):
191          keep = ArvadosFileWriterTestCase.MockKeep({})
192          with Collection(keep_client=keep) as c:
193              writer = c.open("count.txt", "wb")
194              writer.write("0123456789")
195              self.assertEqual('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n', c.portable_manifest_text())
196              self.assertNotIn('781e5e245d69b566979b86e28d23f2c7+10', keep.blocks)
197              self.assertEqual('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n', c.manifest_text())
198              self.assertIn('781e5e245d69b566979b86e28d23f2c7+10', keep.blocks)
199
200
201     def test_write_in_middle(self):
202         keep = ArvadosFileWriterTestCase.MockKeep({"781e5e245d69b566979b86e28d23f2c7+10": b"0123456789"})
203         with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n',
204                              keep_client=keep) as c:
205             writer = c.open("count.txt", "rb+")
206             self.assertEqual(b"0123456789", writer.readfrom(0, 13))
207             writer.seek(3, os.SEEK_SET)
208             writer.write("foo")
209             self.assertEqual(writer.size(), 10)
210             self.assertEqual(b"012foo6789", writer.readfrom(0, 13))
211             self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 acbd18db4cc2f85cedef654fccc4a4d8+3 0:3:count.txt 10:3:count.txt 6:4:count.txt\n", c.portable_manifest_text())
212
213     def test_write_at_end(self):
214         keep = ArvadosFileWriterTestCase.MockKeep({"781e5e245d69b566979b86e28d23f2c7+10": b"0123456789"})
215         with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n',
216                              keep_client=keep) as c:
217             writer = c.open("count.txt", "rb+")
218             self.assertEqual(b"0123456789", writer.readfrom(0, 13))
219             writer.seek(7, os.SEEK_SET)
220             writer.write("foo")
221             self.assertEqual(writer.size(), 10)
222             self.assertEqual(b"0123456foo", writer.readfrom(0, 13))
223             self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 acbd18db4cc2f85cedef654fccc4a4d8+3 0:7:count.txt 10:3:count.txt\n", c.portable_manifest_text())
224
225     def test_write_across_segment_boundary(self):
226         keep = ArvadosFileWriterTestCase.MockKeep({"781e5e245d69b566979b86e28d23f2c7+10": b"0123456789"})
227         with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt 0:10:count.txt\n',
228                              keep_client=keep) as c:
229             writer = c.open("count.txt", "rb+")
230             self.assertEqual(b"012345678901234", writer.readfrom(0, 15))
231             writer.seek(7, os.SEEK_SET)
232             writer.write("foobar")
233             self.assertEqual(writer.size(), 20)
234             self.assertEqual(b"0123456foobar34", writer.readfrom(0, 15))
235             self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 3858f62230ac3c915f300c664312c63f+6 0:7:count.txt 10:6:count.txt 3:7:count.txt\n", c.portable_manifest_text())
236
237     def test_write_across_several_segments(self):
238         keep = ArvadosFileWriterTestCase.MockKeep({"781e5e245d69b566979b86e28d23f2c7+10": b"0123456789"})
239         with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:4:count.txt 0:4:count.txt 0:4:count.txt',
240                              keep_client=keep) as c:
241             writer = c.open("count.txt", "rb+")
242             self.assertEqual(b"012301230123", writer.readfrom(0, 15))
243             writer.seek(2, os.SEEK_SET)
244             writer.write("abcdefg")
245             self.assertEqual(writer.size(), 12)
246             self.assertEqual(b"01abcdefg123", writer.readfrom(0, 15))
247             self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 7ac66c0f148de9519b8bd264312c4d64+7 0:2:count.txt 10:7:count.txt 1:3:count.txt\n", c.portable_manifest_text())
248
249     def test_write_large(self):
250         keep = ArvadosFileWriterTestCase.MockKeep({})
251         api = ArvadosFileWriterTestCase.MockApi({"name":"test_write_large",
252                                                  "manifest_text": ". a5de24f4417cfba9d5825eadc2f4ca49+67108000 598cc1a4ccaef8ab6e4724d87e675d78+32892000 0:100000000:count.txt\n",
253                                                  "replication_desired":None},
254                                                 {"uuid":"zzzzz-4zz18-mockcollection0",
255                                                  "manifest_text": ". a5de24f4417cfba9d5825eadc2f4ca49+67108000 598cc1a4ccaef8ab6e4724d87e675d78+32892000 0:100000000:count.txt\n",
256                                                  "portable_data_hash":"9132ca8e3f671c76103a38f5bc24328c+108"})
257         with Collection('. ' + arvados.config.EMPTY_BLOCK_LOCATOR + ' 0:0:count.txt',
258                              api_client=api, keep_client=keep) as c:
259             writer = c.open("count.txt", "rb+")
260             text = "0123456789" * 100
261             for b in range(0, 100000):
262                 writer.write(text)
263             self.assertEqual(writer.size(), 100000000)
264
265             self.assertIsNone(c.manifest_locator())
266             self.assertTrue(c.modified())
267             c.save_new("test_write_large")
268             self.assertEqual("zzzzz-4zz18-mockcollection0", c.manifest_locator())
269             self.assertFalse(c.modified())
270
271
272     def test_large_write(self):
273         keep = ArvadosFileWriterTestCase.MockKeep({})
274         api = ArvadosFileWriterTestCase.MockApi({}, {})
275         with Collection('. ' + arvados.config.EMPTY_BLOCK_LOCATOR + ' 0:0:count.txt',
276                              api_client=api, keep_client=keep) as c:
277             writer = c.open("count.txt", "rb+")
278             self.assertEqual(writer.size(), 0)
279
280             text = "0123456789"
281             writer.write(text)
282             text = "0123456789" * 9999999
283             writer.write(text)
284             self.assertEqual(writer.size(), 100000000)
285
286             self.assertEqual(c.manifest_text(), ". 781e5e245d69b566979b86e28d23f2c7+10 48dd23ea1645fd47d789804d71b5bb8e+67108864 77c57dc6ac5a10bb2205caaa73187994+32891126 0:100000000:count.txt\n")
287
288     def test_rewrite_on_empty_file(self):
289         keep = ArvadosFileWriterTestCase.MockKeep({})
290         with Collection('. ' + arvados.config.EMPTY_BLOCK_LOCATOR + ' 0:0:count.txt',
291                              keep_client=keep) as c:
292             writer = c.open("count.txt", "rb+")
293             for b in range(0, 10):
294                 writer.seek(0, os.SEEK_SET)
295                 writer.write("0123456789")
296
297             self.assertEqual(writer.size(), 10)
298             self.assertEqual(b"0123456789", writer.readfrom(0, 20))
299             self.assertEqual(". 7a08b07e84641703e5f2c836aa59a170+100 90:10:count.txt\n", c.portable_manifest_text())
300             writer.flush()
301             self.assertEqual(writer.size(), 10)
302             self.assertEqual(b"0123456789", writer.readfrom(0, 20))
303             self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n", c.portable_manifest_text())
304
305     def test_rewrite_append_existing_file(self):
306         keep = ArvadosFileWriterTestCase.MockKeep({
307             "781e5e245d69b566979b86e28d23f2c7+10": b"0123456789",
308         })
309         with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt',
310                              keep_client=keep) as c:
311             writer = c.open("count.txt", "rb+")
312             for b in range(0, 10):
313                 writer.seek(10, os.SEEK_SET)
314                 writer.write("abcdefghij")
315
316             self.assertEqual(writer.size(), 20)
317             self.assertEqual(b"0123456789abcdefghij", writer.readfrom(0, 20))
318             self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 ae5f43bab79cf0be33f025fa97ae7398+100 0:10:count.txt 100:10:count.txt\n", c.portable_manifest_text())
319
320             writer.arvadosfile.flush()
321             self.assertEqual(writer.size(), 20)
322             self.assertEqual(b"0123456789abcdefghij", writer.readfrom(0, 20))
323             self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 a925576942e94b2ef57a066101b48876+10 0:20:count.txt\n", c.portable_manifest_text())
324
325     def test_rewrite_over_existing_file(self):
326         keep = ArvadosFileWriterTestCase.MockKeep({
327             "781e5e245d69b566979b86e28d23f2c7+10": b"0123456789",
328         })
329         with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt',
330                              keep_client=keep) as c:
331             writer = c.open("count.txt", "rb+")
332             for b in range(0, 10):
333                 writer.seek(5, os.SEEK_SET)
334                 writer.write("abcdefghij")
335
336             self.assertEqual(writer.size(), 15)
337             self.assertEqual(b"01234abcdefghij", writer.readfrom(0, 20))
338             self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 ae5f43bab79cf0be33f025fa97ae7398+100 0:5:count.txt 100:10:count.txt\n", c.portable_manifest_text())
339
340             writer.arvadosfile.flush()
341
342             self.assertEqual(writer.size(), 15)
343             self.assertEqual(b"01234abcdefghij", writer.readfrom(0, 20))
344             self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 a925576942e94b2ef57a066101b48876+10 0:5:count.txt 10:10:count.txt\n", c.portable_manifest_text())
345
346     def test_write_large_rewrite(self):
347         keep = ArvadosFileWriterTestCase.MockKeep({})
348         api = ArvadosFileWriterTestCase.MockApi({"name":"test_write_large",
349                                                  "manifest_text": ". 37400a68af9abdd76ca5bf13e819e42a+32892003 a5de24f4417cfba9d5825eadc2f4ca49+67108000 32892000:3:count.txt 32892006:67107997:count.txt 0:32892000:count.txt\n",
350                                                  "replication_desired":None},
351                                                 {"uuid":"zzzzz-4zz18-mockcollection0",
352                                                  "manifest_text": ". 37400a68af9abdd76ca5bf13e819e42a+32892003 a5de24f4417cfba9d5825eadc2f4ca49+67108000 32892000:3:count.txt 32892006:67107997:count.txt 0:32892000:count.txt\n",
353                                                  "portable_data_hash":"217665c6b713e1b78dfba7ebd42344db+156"})
354         with Collection('. ' + arvados.config.EMPTY_BLOCK_LOCATOR + ' 0:0:count.txt',
355                              api_client=api, keep_client=keep) as c:
356             writer = c.open("count.txt", "rb+")
357             text = b''.join([b"0123456789" for a in range(0, 100)])
358             for b in range(0, 100000):
359                 writer.write(text)
360             writer.seek(0, os.SEEK_SET)
361             writer.write("foo")
362             self.assertEqual(writer.size(), 100000000)
363
364             self.assertIsNone(c.manifest_locator())
365             self.assertTrue(c.modified())
366             c.save_new("test_write_large")
367             self.assertEqual("zzzzz-4zz18-mockcollection0", c.manifest_locator())
368             self.assertFalse(c.modified())
369
370     def test_create(self):
371         keep = ArvadosFileWriterTestCase.MockKeep({})
372         api = ArvadosFileWriterTestCase.MockApi({
373             "name":"test_create",
374             "manifest_text":". 2e9ec317e197819358fbc43afca7d837+8 0:8:count.txt\n",
375             "replication_desired":None,
376         }, {
377             "uuid":"zzzzz-4zz18-mockcollection0",
378             "manifest_text":". 2e9ec317e197819358fbc43afca7d837+8 0:8:count.txt\n",
379             "portable_data_hash":"7a461a8c58601798f690f8b368ac4423+51",
380         })
381         with Collection(api_client=api, keep_client=keep) as c:
382             writer = c.open("count.txt", "wb+")
383             self.assertEqual(writer.size(), 0)
384             writer.write("01234567")
385             self.assertEqual(writer.size(), 8)
386
387             self.assertIsNone(c.manifest_locator())
388             self.assertTrue(c.modified())
389             self.assertIsNone(keep.get("2e9ec317e197819358fbc43afca7d837+8"))
390             c.save_new("test_create")
391             self.assertEqual("zzzzz-4zz18-mockcollection0", c.manifest_locator())
392             self.assertFalse(c.modified())
393             self.assertEqual(b"01234567", keep.get("2e9ec317e197819358fbc43afca7d837+8"))
394
395
396     def test_create_subdir(self):
397         keep = ArvadosFileWriterTestCase.MockKeep({})
398         api = ArvadosFileWriterTestCase.MockApi({"name":"test_create",
399                                                  "manifest_text":"./foo/bar 2e9ec317e197819358fbc43afca7d837+8 0:8:count.txt\n",
400                                                  "replication_desired":None},
401                                                 {"uuid":"zzzzz-4zz18-mockcollection0",
402                                                  "manifest_text":"./foo/bar 2e9ec317e197819358fbc43afca7d837+8 0:8:count.txt\n",
403                                                  "portable_data_hash":"1b02aaa62528d28a5be41651cbb9d7c7+59"})
404         with Collection(api_client=api, keep_client=keep) as c:
405             self.assertIsNone(c.api_response())
406             writer = c.open("foo/bar/count.txt", "wb+")
407             writer.write("01234567")
408             self.assertFalse(c.committed())
409             c.save_new("test_create")
410             self.assertTrue(c.committed())
411             self.assertEqual(c.api_response(), api.response)
412
413     def test_overwrite(self):
414         keep = ArvadosFileWriterTestCase.MockKeep({"781e5e245d69b566979b86e28d23f2c7+10": "0123456789"})
415         api = ArvadosFileWriterTestCase.MockApi({"name":"test_overwrite",
416                                                  "manifest_text":". 2e9ec317e197819358fbc43afca7d837+8 0:8:count.txt\n",
417                                                  "replication_desired":None},
418                                                 {"uuid":"zzzzz-4zz18-mockcollection0",
419                                                  "manifest_text":". 2e9ec317e197819358fbc43afca7d837+8 0:8:count.txt\n",
420                                                  "portable_data_hash":"7a461a8c58601798f690f8b368ac4423+51"})
421         with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n',
422                              api_client=api, keep_client=keep) as c:
423             writer = c.open("count.txt", "wb+")
424             self.assertEqual(writer.size(), 0)
425             writer.write("01234567")
426             self.assertEqual(writer.size(), 8)
427
428             self.assertIsNone(c.manifest_locator())
429             self.assertTrue(c.modified())
430             c.save_new("test_overwrite")
431             self.assertEqual("zzzzz-4zz18-mockcollection0", c.manifest_locator())
432             self.assertFalse(c.modified())
433
434     def test_file_not_found(self):
435         with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n') as c:
436             with self.assertRaises(IOError):
437                 writer = c.open("nocount.txt", "rb")
438
439     def test_cannot_open_directory(self):
440         with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n') as c:
441             with self.assertRaises(IOError):
442                 writer = c.open(".", "rb")
443
444     def test_create_multiple(self):
445         keep = ArvadosFileWriterTestCase.MockKeep({})
446         api = ArvadosFileWriterTestCase.MockApi({"name":"test_create_multiple",
447                                                  "manifest_text":". 2e9ec317e197819358fbc43afca7d837+8 e8dc4081b13434b45189a720b77b6818+8 0:8:count1.txt 8:8:count2.txt\n",
448                                                  "replication_desired":None},
449                                                 {"uuid":"zzzzz-4zz18-mockcollection0",
450                                                  "manifest_text":". 2e9ec317e197819358fbc43afca7d837+8 e8dc4081b13434b45189a720b77b6818+8 0:8:count1.txt 8:8:count2.txt\n",
451                                                  "portable_data_hash":"71e7bb6c00d31fc2b4364199fd97be08+102"})
452         with Collection(api_client=api, keep_client=keep) as c:
453             w1 = c.open("count1.txt", "wb")
454             w2 = c.open("count2.txt", "wb")
455             w1.write("01234567")
456             w2.write("abcdefgh")
457             self.assertEqual(w1.size(), 8)
458             self.assertEqual(w2.size(), 8)
459
460             self.assertIsNone(c.manifest_locator())
461             self.assertTrue(c.modified())
462             self.assertIsNone(keep.get("2e9ec317e197819358fbc43afca7d837+8"))
463             c.save_new("test_create_multiple")
464             self.assertEqual("zzzzz-4zz18-mockcollection0", c.manifest_locator())
465             self.assertFalse(c.modified())
466             self.assertEqual(b"01234567", keep.get("2e9ec317e197819358fbc43afca7d837+8"))
467
468
469 class ArvadosFileReaderTestCase(StreamFileReaderTestCase):
470     class MockParent(object):
471         class MockBlockMgr(object):
472             def __init__(self, blocks, nocache):
473                 self.blocks = blocks
474                 self.nocache = nocache
475
476             def block_prefetch(self, loc):
477                 pass
478
479             def get_block_contents(self, loc, num_retries=0, cache_only=False):
480                 if self.nocache and cache_only:
481                     return None
482                 return self.blocks[loc]
483
484         def __init__(self, blocks, nocache):
485             self.blocks = blocks
486             self.nocache = nocache
487             self.lock = arvados.arvfile.NoopLock()
488
489         def root_collection(self):
490             return self
491
492         def _my_block_manager(self):
493             return ArvadosFileReaderTestCase.MockParent.MockBlockMgr(self.blocks, self.nocache)
494
495
496     def make_count_reader(self, nocache=False):
497         stream = []
498         n = 0
499         blocks = {}
500         for d in [b'01234', b'34567', b'67890']:
501             loc = tutil.str_keep_locator(d)
502             blocks[loc] = d
503             stream.append(Range(loc, n, len(d)))
504             n += len(d)
505         af = ArvadosFile(ArvadosFileReaderTestCase.MockParent(blocks, nocache), "count.txt", stream=stream, segments=[Range(1, 0, 3), Range(6, 3, 3), Range(11, 6, 3)])
506         return ArvadosFileReader(af, mode="rb")
507
508     def test_read_block_crossing_behavior(self):
509         # read() needs to return all the data requested if possible, even if it
510         # crosses uncached blocks: https://arvados.org/issues/5856
511         sfile = self.make_count_reader(nocache=True)
512         self.assertEqual(b'12345678', sfile.read(8))
513
514     def test_successive_reads(self):
515         # Override StreamFileReaderTestCase.test_successive_reads
516         sfile = self.make_count_reader(nocache=True)
517         self.assertEqual(b'1234', sfile.read(4))
518         self.assertEqual(b'5678', sfile.read(4))
519         self.assertEqual(b'9', sfile.read(4))
520         self.assertEqual(b'', sfile.read(4))
521
522     def test_tell_after_block_read(self):
523         # Override StreamFileReaderTestCase.test_tell_after_block_read
524         sfile = self.make_count_reader(nocache=True)
525         self.assertEqual(b'12345678', sfile.read(8))
526         self.assertEqual(8, sfile.tell())
527
528     def test_prefetch(self):
529         keep = ArvadosFileWriterTestCase.MockKeep({
530             "2e9ec317e197819358fbc43afca7d837+8": b"01234567",
531             "e8dc4081b13434b45189a720b77b6818+8": b"abcdefgh",
532         })
533         with Collection(". 2e9ec317e197819358fbc43afca7d837+8 e8dc4081b13434b45189a720b77b6818+8 0:16:count.txt\n", keep_client=keep) as c:
534             r = c.open("count.txt", "rb")
535             self.assertEqual(b"0123", r.read(4))
536         self.assertIn("2e9ec317e197819358fbc43afca7d837+8", keep.requests)
537         self.assertIn("e8dc4081b13434b45189a720b77b6818+8", keep.requests)
538
539     def test__eq__from_manifest(self):
540         with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt') as c1:
541             with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt') as c2:
542                 self.assertTrue(c1["count1.txt"] == c2["count1.txt"])
543                 self.assertFalse(c1["count1.txt"] != c2["count1.txt"])
544
545     def test__eq__from_writes(self):
546         with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt') as c1:
547             with Collection() as c2:
548                 f = c2.open("count1.txt", "wb")
549                 f.write("0123456789")
550
551                 self.assertTrue(c1["count1.txt"] == c2["count1.txt"])
552                 self.assertFalse(c1["count1.txt"] != c2["count1.txt"])
553
554     def test__ne__(self):
555         with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt') as c1:
556             with Collection() as c2:
557                 f = c2.open("count1.txt", "wb")
558                 f.write("1234567890")
559
560                 self.assertTrue(c1["count1.txt"] != c2["count1.txt"])
561                 self.assertFalse(c1["count1.txt"] == c2["count1.txt"])
562
563
564 class ArvadosFileReadTestCase(unittest.TestCase, StreamRetryTestMixin):
565     def reader_for(self, coll_name, **kwargs):
566         stream = []
567         segments = []
568         n = 0
569         for d in self.manifest_for(coll_name).split():
570             try:
571                 k = KeepLocator(d)
572                 segments.append(Range(n, n, k.size))
573                 stream.append(Range(d, n, k.size))
574                 n += k.size
575             except ValueError:
576                 pass
577
578         blockmanager = arvados.arvfile._BlockManager(self.keep_client())
579         blockmanager.prefetch_enabled = False
580         col = Collection(keep_client=self.keep_client(), block_manager=blockmanager)
581         af = ArvadosFile(col, "test",
582                          stream=stream,
583                          segments=segments)
584         kwargs.setdefault('mode', 'rb')
585         return ArvadosFileReader(af, **kwargs)
586
587     def read_for_test(self, reader, byte_count, **kwargs):
588         return reader.read(byte_count, **kwargs)
589
590
591 class ArvadosFileReadFromTestCase(ArvadosFileReadTestCase):
592     def read_for_test(self, reader, byte_count, **kwargs):
593         return reader.readfrom(0, byte_count, **kwargs)
594
595
596 class ArvadosFileReadAllTestCase(ArvadosFileReadTestCase):
597     def read_for_test(self, reader, byte_count, **kwargs):
598         return b''.join(reader.readall(**kwargs))
599
600
601 class ArvadosFileReadAllDecompressedTestCase(ArvadosFileReadTestCase):
602     def read_for_test(self, reader, byte_count, **kwargs):
603         return b''.join(reader.readall_decompressed(**kwargs))
604
605
606 class ArvadosFileReadlinesTestCase(ArvadosFileReadTestCase):
607     def read_for_test(self, reader, byte_count, **kwargs):
608         return ''.join(reader.readlines(**kwargs)).encode()
609
610
611 class ArvadosFileTestCase(unittest.TestCase):
612     def datetime_to_hex(self, dt):
613         return hex(int(time.mktime(dt.timetuple())))[2:]
614
615     def test_permission_expired(self):
616         base_manifest = ". 781e5e245d69b566979b86e28d23f2c7+10+A715fd31f8111894f717eb1003c1b0216799dd9ec@{} 0:10:count.txt\n"
617         now = datetime.datetime.now()
618         a_week_ago = now - datetime.timedelta(days=7)
619         a_month_ago = now - datetime.timedelta(days=30)
620         a_week_from_now = now + datetime.timedelta(days=7)
621         with Collection(base_manifest.format(self.datetime_to_hex(a_week_from_now))) as c:
622             self.assertFalse(c.find('count.txt').permission_expired())
623         with Collection(base_manifest.format(self.datetime_to_hex(a_week_ago))) as c:
624             f = c.find('count.txt')
625             self.assertTrue(f.permission_expired())
626             self.assertTrue(f.permission_expired(a_week_from_now))
627             self.assertFalse(f.permission_expired(a_month_ago))
628
629
630 class BlockManagerTest(unittest.TestCase):
631     def test_bufferblock_append(self):
632         keep = ArvadosFileWriterTestCase.MockKeep({})
633         with arvados.arvfile._BlockManager(keep) as blockmanager:
634             bufferblock = blockmanager.alloc_bufferblock()
635             bufferblock.append("foo")
636
637             self.assertEqual(bufferblock.size(), 3)
638             self.assertEqual(bufferblock.buffer_view[0:3], b"foo")
639             self.assertEqual(bufferblock.locator(), "acbd18db4cc2f85cedef654fccc4a4d8+3")
640
641             bufferblock.append("bar")
642
643             self.assertEqual(bufferblock.size(), 6)
644             self.assertEqual(bufferblock.buffer_view[0:6], b"foobar")
645             self.assertEqual(bufferblock.locator(), "3858f62230ac3c915f300c664312c63f+6")
646
647             bufferblock.set_state(arvados.arvfile._BufferBlock.PENDING)
648             with self.assertRaises(arvados.errors.AssertionError):
649                 bufferblock.append("bar")
650
651     def test_bufferblock_dup(self):
652         keep = ArvadosFileWriterTestCase.MockKeep({})
653         with arvados.arvfile._BlockManager(keep) as blockmanager:
654             bufferblock = blockmanager.alloc_bufferblock()
655             bufferblock.append("foo")
656
657             self.assertEqual(bufferblock.size(), 3)
658             self.assertEqual(bufferblock.buffer_view[0:3], b"foo")
659             self.assertEqual(bufferblock.locator(), "acbd18db4cc2f85cedef654fccc4a4d8+3")
660             bufferblock.set_state(arvados.arvfile._BufferBlock.PENDING)
661
662             bufferblock2 = blockmanager.dup_block(bufferblock, None)
663             self.assertNotEqual(bufferblock.blockid, bufferblock2.blockid)
664
665             bufferblock2.append("bar")
666
667             self.assertEqual(bufferblock2.size(), 6)
668             self.assertEqual(bufferblock2.buffer_view[0:6], b"foobar")
669             self.assertEqual(bufferblock2.locator(), "3858f62230ac3c915f300c664312c63f+6")
670
671             self.assertEqual(bufferblock.size(), 3)
672             self.assertEqual(bufferblock.buffer_view[0:3], b"foo")
673             self.assertEqual(bufferblock.locator(), "acbd18db4cc2f85cedef654fccc4a4d8+3")
674
675     def test_bufferblock_get(self):
676         keep = ArvadosFileWriterTestCase.MockKeep({
677             "781e5e245d69b566979b86e28d23f2c7+10": b"0123456789",
678         })
679         with arvados.arvfile._BlockManager(keep) as blockmanager:
680             bufferblock = blockmanager.alloc_bufferblock()
681             bufferblock.append("foo")
682
683             self.assertEqual(blockmanager.get_block_contents("781e5e245d69b566979b86e28d23f2c7+10", 1), b"0123456789")
684             self.assertEqual(blockmanager.get_block_contents(bufferblock.blockid, 1), b"foo")
685
686     def test_bufferblock_commit(self):
687         mockkeep = mock.MagicMock()
688         with arvados.arvfile._BlockManager(mockkeep) as blockmanager:
689             bufferblock = blockmanager.alloc_bufferblock()
690             bufferblock.owner = mock.MagicMock()
691             def flush(sync=None):
692                 blockmanager.commit_bufferblock(bufferblock, sync)
693             bufferblock.owner.flush.side_effect = flush
694             bufferblock.append("foo")
695             blockmanager.commit_all()
696             self.assertTrue(bufferblock.owner.flush.called)
697             self.assertTrue(mockkeep.put.called)
698             self.assertEqual(bufferblock.state(), arvados.arvfile._BufferBlock.COMMITTED)
699             self.assertIsNone(bufferblock.buffer_view)
700
701     def test_bufferblock_commit_pending(self):
702         # Test for bug #7225
703         mockkeep = mock.MagicMock()
704         mockkeep.put.side_effect = lambda x: time.sleep(1)
705         with arvados.arvfile._BlockManager(mockkeep) as blockmanager:
706             bufferblock = blockmanager.alloc_bufferblock()
707             bufferblock.append("foo")
708
709             blockmanager.commit_bufferblock(bufferblock, False)
710             self.assertEqual(bufferblock.state(), arvados.arvfile._BufferBlock.PENDING)
711
712             blockmanager.commit_bufferblock(bufferblock, True)
713             self.assertEqual(bufferblock.state(), arvados.arvfile._BufferBlock.COMMITTED)
714
715
716     def test_bufferblock_commit_with_error(self):
717         mockkeep = mock.MagicMock()
718         mockkeep.put.side_effect = arvados.errors.KeepWriteError("fail")
719         with arvados.arvfile._BlockManager(mockkeep) as blockmanager:
720             bufferblock = blockmanager.alloc_bufferblock()
721             bufferblock.owner = mock.MagicMock()
722             def flush(sync=None):
723                 blockmanager.commit_bufferblock(bufferblock, sync)
724             bufferblock.owner.flush.side_effect = flush
725             bufferblock.append("foo")
726             with self.assertRaises(arvados.errors.KeepWriteError) as err:
727                 blockmanager.commit_all()
728             self.assertTrue(bufferblock.owner.flush.called)
729             self.assertEqual(str(err.exception), "Error writing some blocks: block acbd18db4cc2f85cedef654fccc4a4d8+3 raised KeepWriteError (fail)")
730             self.assertEqual(bufferblock.state(), arvados.arvfile._BufferBlock.ERROR)