5824: Fix Keep server shutdown, check errors, simplify stderr redirection.
[arvados.git] / sdk / python / tests / test_arvfile.py
1 #!/usr/bin/env python
2
3 import bz2
4 import gzip
5 import io
6 import mock
7 import os
8 import unittest
9 import hashlib
10 import time
11
12 import arvados
13 from arvados._ranges import Range
14 from arvados.keep import KeepLocator
15 from arvados.collection import Collection, CollectionReader
16 from arvados.arvfile import ArvadosFile, ArvadosFileReader
17
18 import arvados_testutil as tutil
19 from test_stream import StreamFileReaderTestCase, StreamRetryTestMixin
20
21 class ArvadosFileWriterTestCase(unittest.TestCase):
22     class MockKeep(object):
23         def __init__(self, blocks):
24             self.blocks = blocks
25             self.requests = []
26         def get(self, locator, num_retries=0):
27             self.requests.append(locator)
28             return self.blocks.get(locator)
29         def get_from_cache(self, locator):
30             self.requests.append(locator)
31             return self.blocks.get(locator)
32         def put(self, data, num_retries=None):
33             pdh = "%s+%i" % (hashlib.md5(data).hexdigest(), len(data))
34             self.blocks[pdh] = str(data)
35             return pdh
36
37     class MockApi(object):
38         def __init__(self, b, r):
39             self.body = b
40             self.response = r
41         class MockCollections(object):
42             def __init__(self, b, r):
43                 self.body = b
44                 self.response = r
45             class Execute(object):
46                 def __init__(self, r):
47                     self.response = r
48                 def execute(self, num_retries=None):
49                     return self.response
50             def create(self, ensure_unique_name=False, body=None):
51                 if body != self.body:
52                     raise Exception("Body %s does not match expectation %s" % (body, self.body))
53                 return ArvadosFileWriterTestCase.MockApi.MockCollections.Execute(self.response)
54             def update(self, uuid=None, body=None):
55                 return ArvadosFileWriterTestCase.MockApi.MockCollections.Execute(self.response)
56         def collections(self):
57             return ArvadosFileWriterTestCase.MockApi.MockCollections(self.body, self.response)
58
59
60     def test_truncate(self):
61         keep = ArvadosFileWriterTestCase.MockKeep({"781e5e245d69b566979b86e28d23f2c7+10": "0123456789"})
62         api = ArvadosFileWriterTestCase.MockApi({"name":"test_truncate",
63                                                  "manifest_text":". 781e5e245d69b566979b86e28d23f2c7+10 0:8:count.txt\n"},
64                                                 {"uuid":"zzzzz-4zz18-mockcollection0",
65                                                  "manifest_text":". 781e5e245d69b566979b86e28d23f2c7+10 0:8:count.txt\n"})
66         with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n',
67                              api_client=api, keep_client=keep) as c:
68             writer = c.open("count.txt", "r+")
69             self.assertEqual(writer.size(), 10)
70             self.assertEqual("0123456789", writer.read(12))
71
72             writer.truncate(8)
73
74             # Make sure reading off the end doesn't break
75             self.assertEqual("", writer.read(12))
76
77             self.assertEqual(writer.size(), 8)
78             writer.seek(0, os.SEEK_SET)
79             self.assertEqual("01234567", writer.read(12))
80
81             self.assertIsNone(c.manifest_locator())
82             self.assertTrue(c.modified())
83             c.save_new("test_truncate")
84             self.assertEqual("zzzzz-4zz18-mockcollection0", c.manifest_locator())
85             self.assertFalse(c.modified())
86
87     def test_write_to_end(self):
88         keep = ArvadosFileWriterTestCase.MockKeep({"781e5e245d69b566979b86e28d23f2c7+10": "0123456789"})
89         api = ArvadosFileWriterTestCase.MockApi({"name":"test_append",
90                                                  "manifest_text": ". 781e5e245d69b566979b86e28d23f2c7+10 acbd18db4cc2f85cedef654fccc4a4d8+3 0:13:count.txt\n"},
91                                                 {"uuid":"zzzzz-4zz18-mockcollection0",
92                                                  "manifest_text": ". 781e5e245d69b566979b86e28d23f2c7+10 acbd18db4cc2f85cedef654fccc4a4d8+3 0:13:count.txt\n"})
93         with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n',
94                              api_client=api, keep_client=keep) as c:
95             writer = c.open("count.txt", "r+")
96             self.assertEqual(writer.size(), 10)
97
98             writer.seek(5, os.SEEK_SET)
99             self.assertEqual("56789", writer.read(8))
100
101             writer.seek(10, os.SEEK_SET)
102             writer.write("foo")
103             self.assertEqual(writer.size(), 13)
104
105             writer.seek(5, os.SEEK_SET)
106             self.assertEqual("56789foo", writer.read(8))
107
108             self.assertIsNone(c.manifest_locator())
109             self.assertTrue(c.modified())
110             self.assertIsNone(keep.get("acbd18db4cc2f85cedef654fccc4a4d8+3"))
111
112             c.save_new("test_append")
113             self.assertEqual("zzzzz-4zz18-mockcollection0", c.manifest_locator())
114             self.assertFalse(c.modified())
115             self.assertEqual("foo", keep.get("acbd18db4cc2f85cedef654fccc4a4d8+3"))
116
117
118     def test_append(self):
119         keep = ArvadosFileWriterTestCase.MockKeep({"781e5e245d69b566979b86e28d23f2c7+10": "0123456789"})
120         c = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n', keep_client=keep)
121         writer = c.open("count.txt", "a+")
122         self.assertEqual(writer.read(20), "0123456789")
123         writer.seek(0, os.SEEK_SET)
124
125         writer.write("hello")
126         self.assertEqual(writer.read(20), "0123456789hello")
127         writer.seek(0, os.SEEK_SET)
128
129         writer.write("world")
130         self.assertEqual(writer.read(20), "0123456789helloworld")
131
132         self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 fc5e038d38a57032085441e7fe7010b0+10 0:20:count.txt\n", c.portable_manifest_text())
133
134     def test_write_at_beginning(self):
135         keep = ArvadosFileWriterTestCase.MockKeep({"781e5e245d69b566979b86e28d23f2c7+10": "0123456789"})
136         with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n',
137                              keep_client=keep) as c:
138             writer = c.open("count.txt", "r+")
139             self.assertEqual("0123456789", writer.readfrom(0, 13))
140             writer.seek(0, os.SEEK_SET)
141             writer.write("foo")
142             self.assertEqual(writer.size(), 10)
143             self.assertEqual("foo3456789", writer.readfrom(0, 13))
144             self.assertEqual(". acbd18db4cc2f85cedef654fccc4a4d8+3 781e5e245d69b566979b86e28d23f2c7+10 0:3:count.txt 6:7:count.txt\n", c.portable_manifest_text())
145
146     def test_write_empty(self):
147         keep = ArvadosFileWriterTestCase.MockKeep({})
148         with Collection(keep_client=keep) as c:
149             writer = c.open("count.txt", "w")
150             self.assertEqual(writer.size(), 0)
151             self.assertEqual(". d41d8cd98f00b204e9800998ecf8427e+0 0:0:count.txt\n", c.portable_manifest_text())
152
153     def test_save_manifest_text(self):
154         keep = ArvadosFileWriterTestCase.MockKeep({})
155         with Collection(keep_client=keep) as c:
156             writer = c.open("count.txt", "w")
157             writer.write("0123456789")
158             self.assertEqual('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n', c.portable_manifest_text())
159             self.assertNotIn('781e5e245d69b566979b86e28d23f2c7+10', keep.blocks)
160
161             self.assertEqual('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n', c.save_new(create_collection_record=False))
162             self.assertIn('781e5e245d69b566979b86e28d23f2c7+10', keep.blocks)
163
164     def test_get_manifest_text_commits(self):
165          keep = ArvadosFileWriterTestCase.MockKeep({})
166          with Collection(keep_client=keep) as c:
167              writer = c.open("count.txt", "w")
168              writer.write("0123456789")
169              self.assertEqual('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n', c.portable_manifest_text())
170              self.assertNotIn('781e5e245d69b566979b86e28d23f2c7+10', keep.blocks)
171              self.assertEqual('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n', c.manifest_text())
172              self.assertIn('781e5e245d69b566979b86e28d23f2c7+10', keep.blocks)
173
174
175     def test_write_in_middle(self):
176         keep = ArvadosFileWriterTestCase.MockKeep({"781e5e245d69b566979b86e28d23f2c7+10": "0123456789"})
177         with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n',
178                              keep_client=keep) as c:
179             writer = c.open("count.txt", "r+")
180             self.assertEqual("0123456789", writer.readfrom(0, 13))
181             writer.seek(3, os.SEEK_SET)
182             writer.write("foo")
183             self.assertEqual(writer.size(), 10)
184             self.assertEqual("012foo6789", writer.readfrom(0, 13))
185             self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 acbd18db4cc2f85cedef654fccc4a4d8+3 0:3:count.txt 10:3:count.txt 6:4:count.txt\n", c.portable_manifest_text())
186
187     def test_write_at_end(self):
188         keep = ArvadosFileWriterTestCase.MockKeep({"781e5e245d69b566979b86e28d23f2c7+10": "0123456789"})
189         with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n',
190                              keep_client=keep) as c:
191             writer = c.open("count.txt", "r+")
192             self.assertEqual("0123456789", writer.readfrom(0, 13))
193             writer.seek(7, os.SEEK_SET)
194             writer.write("foo")
195             self.assertEqual(writer.size(), 10)
196             self.assertEqual("0123456foo", writer.readfrom(0, 13))
197             self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 acbd18db4cc2f85cedef654fccc4a4d8+3 0:7:count.txt 10:3:count.txt\n", c.portable_manifest_text())
198
199     def test_write_across_segment_boundary(self):
200         keep = ArvadosFileWriterTestCase.MockKeep({"781e5e245d69b566979b86e28d23f2c7+10": "0123456789"})
201         with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt 0:10:count.txt\n',
202                              keep_client=keep) as c:
203             writer = c.open("count.txt", "r+")
204             self.assertEqual("012345678901234", writer.readfrom(0, 15))
205             writer.seek(7, os.SEEK_SET)
206             writer.write("foobar")
207             self.assertEqual(writer.size(), 20)
208             self.assertEqual("0123456foobar34", writer.readfrom(0, 15))
209             self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 3858f62230ac3c915f300c664312c63f+6 0:7:count.txt 10:6:count.txt 3:7:count.txt\n", c.portable_manifest_text())
210
211     def test_write_across_several_segments(self):
212         keep = ArvadosFileWriterTestCase.MockKeep({"781e5e245d69b566979b86e28d23f2c7+10": "0123456789"})
213         with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:4:count.txt 0:4:count.txt 0:4:count.txt',
214                              keep_client=keep) as c:
215             writer = c.open("count.txt", "r+")
216             self.assertEqual("012301230123", writer.readfrom(0, 15))
217             writer.seek(2, os.SEEK_SET)
218             writer.write("abcdefg")
219             self.assertEqual(writer.size(), 12)
220             self.assertEqual("01abcdefg123", writer.readfrom(0, 15))
221             self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 7ac66c0f148de9519b8bd264312c4d64+7 0:2:count.txt 10:7:count.txt 1:3:count.txt\n", c.portable_manifest_text())
222
223     def test_write_large(self):
224         keep = ArvadosFileWriterTestCase.MockKeep({})
225         api = ArvadosFileWriterTestCase.MockApi({"name":"test_write_large",
226                                                  "manifest_text": ". a5de24f4417cfba9d5825eadc2f4ca49+67108000 598cc1a4ccaef8ab6e4724d87e675d78+32892000 0:100000000:count.txt\n"},
227                                                 {"uuid":"zzzzz-4zz18-mockcollection0",
228                                                  "manifest_text": ". a5de24f4417cfba9d5825eadc2f4ca49+67108000 598cc1a4ccaef8ab6e4724d87e675d78+32892000 0:100000000:count.txt\n"})
229         with Collection('. ' + arvados.config.EMPTY_BLOCK_LOCATOR + ' 0:0:count.txt',
230                              api_client=api, keep_client=keep) as c:
231             writer = c.open("count.txt", "r+")
232             text = "0123456789" * 100
233             for b in xrange(0, 100000):
234                 writer.write(text)
235             self.assertEqual(writer.size(), 100000000)
236
237             self.assertIsNone(c.manifest_locator())
238             self.assertTrue(c.modified())
239             c.save_new("test_write_large")
240             self.assertEqual("zzzzz-4zz18-mockcollection0", c.manifest_locator())
241             self.assertFalse(c.modified())
242
243
244     def test_large_write(self):
245         keep = ArvadosFileWriterTestCase.MockKeep({})
246         api = ArvadosFileWriterTestCase.MockApi({}, {})
247         with Collection('. ' + arvados.config.EMPTY_BLOCK_LOCATOR + ' 0:0:count.txt',
248                              api_client=api, keep_client=keep) as c:
249             writer = c.open("count.txt", "r+")
250             self.assertEqual(writer.size(), 0)
251
252             text = "0123456789"
253             writer.write(text)
254             text = "0123456789" * 9999999
255             writer.write(text)
256             self.assertEqual(writer.size(), 100000000)
257
258             self.assertEqual(c.manifest_text(), ". 781e5e245d69b566979b86e28d23f2c7+10 48dd23ea1645fd47d789804d71b5bb8e+67108864 77c57dc6ac5a10bb2205caaa73187994+32891126 0:100000000:count.txt\n")
259
260     def test_rewrite_on_empty_file(self):
261         keep = ArvadosFileWriterTestCase.MockKeep({})
262         with Collection('. ' + arvados.config.EMPTY_BLOCK_LOCATOR + ' 0:0:count.txt',
263                              keep_client=keep) as c:
264             writer = c.open("count.txt", "r+")
265             for b in xrange(0, 10):
266                 writer.seek(0, os.SEEK_SET)
267                 writer.write("0123456789")
268
269             self.assertEqual(writer.size(), 10)
270             self.assertEqual("0123456789", writer.readfrom(0, 20))
271             self.assertEqual(". 7a08b07e84641703e5f2c836aa59a170+100 90:10:count.txt\n", c.portable_manifest_text())
272             writer.flush()
273             self.assertEqual(writer.size(), 10)
274             self.assertEqual("0123456789", writer.readfrom(0, 20))
275             self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n", c.portable_manifest_text())
276
277     def test_rewrite_append_existing_file(self):
278         keep = ArvadosFileWriterTestCase.MockKeep({"781e5e245d69b566979b86e28d23f2c7+10": "0123456789"})
279         with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt',
280                              keep_client=keep) as c:
281             writer = c.open("count.txt", "r+")
282             for b in xrange(0, 10):
283                 writer.seek(10, os.SEEK_SET)
284                 writer.write("abcdefghij")
285
286             self.assertEqual(writer.size(), 20)
287             self.assertEqual("0123456789abcdefghij", writer.readfrom(0, 20))
288             self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 ae5f43bab79cf0be33f025fa97ae7398+100 0:10:count.txt 100:10:count.txt\n", c.portable_manifest_text())
289
290             writer.arvadosfile.flush()
291             self.assertEqual(writer.size(), 20)
292             self.assertEqual("0123456789abcdefghij", writer.readfrom(0, 20))
293             self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 a925576942e94b2ef57a066101b48876+10 0:20:count.txt\n", c.portable_manifest_text())
294
295     def test_rewrite_over_existing_file(self):
296         keep = ArvadosFileWriterTestCase.MockKeep({"781e5e245d69b566979b86e28d23f2c7+10": "0123456789"})
297         with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt',
298                              keep_client=keep) as c:
299             writer = c.open("count.txt", "r+")
300             for b in xrange(0, 10):
301                 writer.seek(5, os.SEEK_SET)
302                 writer.write("abcdefghij")
303
304             self.assertEqual(writer.size(), 15)
305             self.assertEqual("01234abcdefghij", writer.readfrom(0, 20))
306             self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 ae5f43bab79cf0be33f025fa97ae7398+100 0:5:count.txt 100:10:count.txt\n", c.portable_manifest_text())
307
308             writer.arvadosfile.flush()
309
310             self.assertEqual(writer.size(), 15)
311             self.assertEqual("01234abcdefghij", writer.readfrom(0, 20))
312             self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 a925576942e94b2ef57a066101b48876+10 0:5:count.txt 10:10:count.txt\n", c.portable_manifest_text())
313
314     def test_write_large_rewrite(self):
315         keep = ArvadosFileWriterTestCase.MockKeep({})
316         api = ArvadosFileWriterTestCase.MockApi({"name":"test_write_large",
317                                                  "manifest_text": ". 37400a68af9abdd76ca5bf13e819e42a+32892003 a5de24f4417cfba9d5825eadc2f4ca49+67108000 32892000:3:count.txt 32892006:67107997:count.txt 0:32892000:count.txt\n"},
318                                                 {"uuid":"zzzzz-4zz18-mockcollection0",
319                                                  "manifest_text": ". 37400a68af9abdd76ca5bf13e819e42a+32892003 a5de24f4417cfba9d5825eadc2f4ca49+67108000 32892000:3:count.txt 32892006:67107997:count.txt 0:32892000:count.txt\n"})
320         with Collection('. ' + arvados.config.EMPTY_BLOCK_LOCATOR + ' 0:0:count.txt',
321                              api_client=api, keep_client=keep) as c:
322             writer = c.open("count.txt", "r+")
323             text = ''.join(["0123456789" for a in xrange(0, 100)])
324             for b in xrange(0, 100000):
325                 writer.write(text)
326             writer.seek(0, os.SEEK_SET)
327             writer.write("foo")
328             self.assertEqual(writer.size(), 100000000)
329
330             self.assertIsNone(c.manifest_locator())
331             self.assertTrue(c.modified())
332             c.save_new("test_write_large")
333             self.assertEqual("zzzzz-4zz18-mockcollection0", c.manifest_locator())
334             self.assertFalse(c.modified())
335
336     def test_create(self):
337         keep = ArvadosFileWriterTestCase.MockKeep({})
338         api = ArvadosFileWriterTestCase.MockApi({"name":"test_create",
339                                                  "manifest_text":". 2e9ec317e197819358fbc43afca7d837+8 0:8:count.txt\n"},
340                                                 {"uuid":"zzzzz-4zz18-mockcollection0",
341                                                  "manifest_text":". 2e9ec317e197819358fbc43afca7d837+8 0:8:count.txt\n"})
342         with Collection(api_client=api, keep_client=keep) as c:
343             writer = c.open("count.txt", "w+")
344             self.assertEqual(writer.size(), 0)
345             writer.write("01234567")
346             self.assertEqual(writer.size(), 8)
347
348             self.assertIsNone(c.manifest_locator())
349             self.assertTrue(c.modified())
350             self.assertIsNone(keep.get("2e9ec317e197819358fbc43afca7d837+8"))
351             c.save_new("test_create")
352             self.assertEqual("zzzzz-4zz18-mockcollection0", c.manifest_locator())
353             self.assertFalse(c.modified())
354             self.assertEqual("01234567", keep.get("2e9ec317e197819358fbc43afca7d837+8"))
355
356
357     def test_create_subdir(self):
358         keep = ArvadosFileWriterTestCase.MockKeep({})
359         api = ArvadosFileWriterTestCase.MockApi({"name":"test_create",
360                                                  "manifest_text":"./foo/bar 2e9ec317e197819358fbc43afca7d837+8 0:8:count.txt\n"},
361                                                 {"uuid":"zzzzz-4zz18-mockcollection0",
362                                                  "manifest_text":"./foo/bar 2e9ec317e197819358fbc43afca7d837+8 0:8:count.txt\n"})
363         with Collection(api_client=api, keep_client=keep) as c:
364             self.assertIsNone(c.api_response())
365             writer = c.open("foo/bar/count.txt", "w+")
366             writer.write("01234567")
367             self.assertFalse(c.committed())
368             c.save_new("test_create")
369             self.assertTrue(c.committed())
370             self.assertEqual(c.api_response(), api.response)
371
372     def test_overwrite(self):
373         keep = ArvadosFileWriterTestCase.MockKeep({"781e5e245d69b566979b86e28d23f2c7+10": "0123456789"})
374         api = ArvadosFileWriterTestCase.MockApi({"name":"test_overwrite",
375                                                  "manifest_text":". 2e9ec317e197819358fbc43afca7d837+8 0:8:count.txt\n"},
376                                                 {"uuid":"zzzzz-4zz18-mockcollection0",
377                                                  "manifest_text":". 2e9ec317e197819358fbc43afca7d837+8 0:8:count.txt\n"})
378         with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n',
379                              api_client=api, keep_client=keep) as c:
380             writer = c.open("count.txt", "w+")
381             self.assertEqual(writer.size(), 0)
382             writer.write("01234567")
383             self.assertEqual(writer.size(), 8)
384
385             self.assertIsNone(c.manifest_locator())
386             self.assertTrue(c.modified())
387             c.save_new("test_overwrite")
388             self.assertEqual("zzzzz-4zz18-mockcollection0", c.manifest_locator())
389             self.assertFalse(c.modified())
390
391     def test_file_not_found(self):
392         with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n') as c:
393             with self.assertRaises(IOError):
394                 writer = c.open("nocount.txt", "r")
395
396     def test_cannot_open_directory(self):
397         with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n') as c:
398             with self.assertRaises(IOError):
399                 writer = c.open(".", "r")
400
401     def test_create_multiple(self):
402         keep = ArvadosFileWriterTestCase.MockKeep({})
403         api = ArvadosFileWriterTestCase.MockApi({"name":"test_create_multiple",
404                                                  "manifest_text":". 2e9ec317e197819358fbc43afca7d837+8 e8dc4081b13434b45189a720b77b6818+8 0:8:count1.txt 8:8:count2.txt\n"},
405                                                 {"uuid":"zzzzz-4zz18-mockcollection0",
406                                                  "manifest_text":". 2e9ec317e197819358fbc43afca7d837+8 e8dc4081b13434b45189a720b77b6818+8 0:8:count1.txt 8:8:count2.txt\n"})
407         with Collection(api_client=api, keep_client=keep) as c:
408             w1 = c.open("count1.txt", "w")
409             w2 = c.open("count2.txt", "w")
410             w1.write("01234567")
411             w2.write("abcdefgh")
412             self.assertEqual(w1.size(), 8)
413             self.assertEqual(w2.size(), 8)
414
415             self.assertIsNone(c.manifest_locator())
416             self.assertTrue(c.modified())
417             self.assertIsNone(keep.get("2e9ec317e197819358fbc43afca7d837+8"))
418             c.save_new("test_create_multiple")
419             self.assertEqual("zzzzz-4zz18-mockcollection0", c.manifest_locator())
420             self.assertFalse(c.modified())
421             self.assertEqual("01234567", keep.get("2e9ec317e197819358fbc43afca7d837+8"))
422
423
424 class ArvadosFileReaderTestCase(StreamFileReaderTestCase):
425     class MockParent(object):
426         class MockBlockMgr(object):
427             def __init__(self, blocks, nocache):
428                 self.blocks = blocks
429                 self.nocache = nocache
430
431             def block_prefetch(self, loc):
432                 pass
433
434             def get_block_contents(self, loc, num_retries=0, cache_only=False):
435                 if self.nocache and cache_only:
436                     return None
437                 return self.blocks[loc]
438
439         def __init__(self, blocks, nocache):
440             self.blocks = blocks
441             self.nocache = nocache
442             self.lock = arvados.arvfile.NoopLock()
443
444         def root_collection(self):
445             return self
446
447         def _my_block_manager(self):
448             return ArvadosFileReaderTestCase.MockParent.MockBlockMgr(self.blocks, self.nocache)
449
450
451     def make_count_reader(self, nocache=False):
452         stream = []
453         n = 0
454         blocks = {}
455         for d in ['01234', '34567', '67890']:
456             loc = '{}+{}'.format(hashlib.md5(d).hexdigest(), len(d))
457             blocks[loc] = d
458             stream.append(Range(loc, n, len(d)))
459             n += len(d)
460         af = ArvadosFile(ArvadosFileReaderTestCase.MockParent(blocks, nocache), "count.txt", stream=stream, segments=[Range(1, 0, 3), Range(6, 3, 3), Range(11, 6, 3)])
461         return ArvadosFileReader(af)
462
463     def test_read_block_crossing_behavior(self):
464         # read() needs to return all the data requested if possible, even if it
465         # crosses uncached blocks: https://arvados.org/issues/5856
466         sfile = self.make_count_reader(nocache=True)
467         self.assertEqual('12345678', sfile.read(8))
468
469     def test_successive_reads(self):
470         # Override StreamFileReaderTestCase.test_successive_reads
471         sfile = self.make_count_reader(nocache=True)
472         self.assertEqual('1234', sfile.read(4))
473         self.assertEqual('5678', sfile.read(4))
474         self.assertEqual('9', sfile.read(4))
475         self.assertEqual('', sfile.read(4))
476
477     def test_tell_after_block_read(self):
478         # Override StreamFileReaderTestCase.test_tell_after_block_read
479         sfile = self.make_count_reader(nocache=True)
480         self.assertEqual('12345678', sfile.read(8))
481         self.assertEqual(8, sfile.tell())
482
483     def test_prefetch(self):
484         keep = ArvadosFileWriterTestCase.MockKeep({"2e9ec317e197819358fbc43afca7d837+8": "01234567", "e8dc4081b13434b45189a720b77b6818+8": "abcdefgh"})
485         with Collection(". 2e9ec317e197819358fbc43afca7d837+8 e8dc4081b13434b45189a720b77b6818+8 0:16:count.txt\n", keep_client=keep) as c:
486             r = c.open("count.txt", "r")
487             self.assertEqual("0123", r.read(4))
488         self.assertIn("2e9ec317e197819358fbc43afca7d837+8", keep.requests)
489         self.assertIn("e8dc4081b13434b45189a720b77b6818+8", keep.requests)
490
491     def test__eq__from_manifest(self):
492         with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt') as c1:
493             with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt') as c2:
494                 self.assertTrue(c1["count1.txt"] == c2["count1.txt"])
495                 self.assertFalse(c1["count1.txt"] != c2["count1.txt"])
496
497     def test__eq__from_writes(self):
498         with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt') as c1:
499             with Collection() as c2:
500                 f = c2.open("count1.txt", "w")
501                 f.write("0123456789")
502
503                 self.assertTrue(c1["count1.txt"] == c2["count1.txt"])
504                 self.assertFalse(c1["count1.txt"] != c2["count1.txt"])
505
506     def test__ne__(self):
507         with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt') as c1:
508             with Collection() as c2:
509                 f = c2.open("count1.txt", "w")
510                 f.write("1234567890")
511
512                 self.assertTrue(c1["count1.txt"] != c2["count1.txt"])
513                 self.assertFalse(c1["count1.txt"] == c2["count1.txt"])
514
515
516 class ArvadosFileReadTestCase(unittest.TestCase, StreamRetryTestMixin):
517     def reader_for(self, coll_name, **kwargs):
518         stream = []
519         segments = []
520         n = 0
521         for d in self.manifest_for(coll_name).split():
522             try:
523                 k = KeepLocator(d)
524                 segments.append(Range(n, n, k.size))
525                 stream.append(Range(d, n, k.size))
526                 n += k.size
527             except ValueError:
528                 pass
529
530         blockmanager = arvados.arvfile._BlockManager(self.keep_client())
531         blockmanager.prefetch_enabled = False
532         col = Collection(keep_client=self.keep_client(), block_manager=blockmanager)
533         af = ArvadosFile(col, "test",
534                          stream=stream,
535                          segments=segments)
536         return ArvadosFileReader(af, **kwargs)
537
538     def read_for_test(self, reader, byte_count, **kwargs):
539         return reader.read(byte_count, **kwargs)
540
541
542 class ArvadosFileReadFromTestCase(ArvadosFileReadTestCase):
543     def read_for_test(self, reader, byte_count, **kwargs):
544         return reader.readfrom(0, byte_count, **kwargs)
545
546
547 class ArvadosFileReadAllTestCase(ArvadosFileReadTestCase):
548     def read_for_test(self, reader, byte_count, **kwargs):
549         return ''.join(reader.readall(**kwargs))
550
551
552 class ArvadosFileReadAllDecompressedTestCase(ArvadosFileReadTestCase):
553     def read_for_test(self, reader, byte_count, **kwargs):
554         return ''.join(reader.readall_decompressed(**kwargs))
555
556
557 class ArvadosFileReadlinesTestCase(ArvadosFileReadTestCase):
558     def read_for_test(self, reader, byte_count, **kwargs):
559         return ''.join(reader.readlines(**kwargs))
560
561 class BlockManagerTest(unittest.TestCase):
562     def test_bufferblock_append(self):
563         keep = ArvadosFileWriterTestCase.MockKeep({})
564         with arvados.arvfile._BlockManager(keep) as blockmanager:
565             bufferblock = blockmanager.alloc_bufferblock()
566             bufferblock.append("foo")
567
568             self.assertEqual(bufferblock.size(), 3)
569             self.assertEqual(bufferblock.buffer_view[0:3], "foo")
570             self.assertEqual(bufferblock.locator(), "acbd18db4cc2f85cedef654fccc4a4d8+3")
571
572             bufferblock.append("bar")
573
574             self.assertEqual(bufferblock.size(), 6)
575             self.assertEqual(bufferblock.buffer_view[0:6], "foobar")
576             self.assertEqual(bufferblock.locator(), "3858f62230ac3c915f300c664312c63f+6")
577
578             bufferblock.set_state(arvados.arvfile._BufferBlock.PENDING)
579             with self.assertRaises(arvados.errors.AssertionError):
580                 bufferblock.append("bar")
581
582     def test_bufferblock_dup(self):
583         keep = ArvadosFileWriterTestCase.MockKeep({})
584         with arvados.arvfile._BlockManager(keep) as blockmanager:
585             bufferblock = blockmanager.alloc_bufferblock()
586             bufferblock.append("foo")
587
588             self.assertEqual(bufferblock.size(), 3)
589             self.assertEqual(bufferblock.buffer_view[0:3], "foo")
590             self.assertEqual(bufferblock.locator(), "acbd18db4cc2f85cedef654fccc4a4d8+3")
591             bufferblock.set_state(arvados.arvfile._BufferBlock.PENDING)
592
593             bufferblock2 = blockmanager.dup_block(bufferblock, None)
594             self.assertNotEqual(bufferblock.blockid, bufferblock2.blockid)
595
596             bufferblock2.append("bar")
597
598             self.assertEqual(bufferblock2.size(), 6)
599             self.assertEqual(bufferblock2.buffer_view[0:6], "foobar")
600             self.assertEqual(bufferblock2.locator(), "3858f62230ac3c915f300c664312c63f+6")
601
602             self.assertEqual(bufferblock.size(), 3)
603             self.assertEqual(bufferblock.buffer_view[0:3], "foo")
604             self.assertEqual(bufferblock.locator(), "acbd18db4cc2f85cedef654fccc4a4d8+3")
605
606     def test_bufferblock_get(self):
607         keep = ArvadosFileWriterTestCase.MockKeep({"781e5e245d69b566979b86e28d23f2c7+10": "0123456789"})
608         with arvados.arvfile._BlockManager(keep) as blockmanager:
609             bufferblock = blockmanager.alloc_bufferblock()
610             bufferblock.append("foo")
611
612             self.assertEqual(blockmanager.get_block_contents("781e5e245d69b566979b86e28d23f2c7+10", 1), "0123456789")
613             self.assertEqual(blockmanager.get_block_contents(bufferblock.blockid, 1), "foo")
614
615     def test_bufferblock_commit(self):
616         mockkeep = mock.MagicMock()
617         with arvados.arvfile._BlockManager(mockkeep) as blockmanager:
618             bufferblock = blockmanager.alloc_bufferblock()
619             bufferblock.owner = mock.MagicMock()
620             def flush(sync=None):
621                 blockmanager.commit_bufferblock(bufferblock, sync)
622             bufferblock.owner.flush.side_effect = flush
623             bufferblock.append("foo")
624             blockmanager.commit_all()
625             self.assertTrue(bufferblock.owner.flush.called)
626             self.assertTrue(mockkeep.put.called)
627             self.assertEqual(bufferblock.state(), arvados.arvfile._BufferBlock.COMMITTED)
628             self.assertIsNone(bufferblock.buffer_view)
629
630     def test_bufferblock_commit_pending(self):
631         # Test for bug #7225
632         mockkeep = mock.MagicMock()
633         mockkeep.put.side_effect = lambda x: time.sleep(1)
634         with arvados.arvfile._BlockManager(mockkeep) as blockmanager:
635             bufferblock = blockmanager.alloc_bufferblock()
636             bufferblock.append("foo")
637
638             blockmanager.commit_bufferblock(bufferblock, False)
639             self.assertEqual(bufferblock.state(), arvados.arvfile._BufferBlock.PENDING)
640
641             blockmanager.commit_bufferblock(bufferblock, True)
642             self.assertEqual(bufferblock.state(), arvados.arvfile._BufferBlock.COMMITTED)
643
644
645     def test_bufferblock_commit_with_error(self):
646         mockkeep = mock.MagicMock()
647         mockkeep.put.side_effect = arvados.errors.KeepWriteError("fail")
648         with arvados.arvfile._BlockManager(mockkeep) as blockmanager:
649             bufferblock = blockmanager.alloc_bufferblock()
650             bufferblock.owner = mock.MagicMock()
651             def flush(sync=None):
652                 blockmanager.commit_bufferblock(bufferblock, sync)
653             bufferblock.owner.flush.side_effect = flush
654             bufferblock.append("foo")
655             with self.assertRaises(arvados.errors.KeepWriteError) as err:
656                 blockmanager.commit_all()
657             self.assertTrue(bufferblock.owner.flush.called)
658             self.assertEqual(str(err.exception), "Error writing some blocks: block acbd18db4cc2f85cedef654fccc4a4d8+3 raised KeepWriteError (fail)")
659             self.assertEqual(bufferblock.state(), arvados.arvfile._BufferBlock.ERROR)