Merge branch 'master' into 4823-python-sdk-writable-collection-api
[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
11 import arvados
12 from arvados._ranges import Range
13 from arvados.keep import KeepLocator
14 from arvados.collection import Collection, CollectionReader
15 from arvados.arvfile import ArvadosFile, ArvadosFileReader, SYNC_READONLY, SYNC_EXPLICIT
16
17 import arvados_testutil as tutil
18 from test_stream import StreamFileReaderTestCase, StreamRetryTestMixin
19
20 class ArvadosFileWriterTestCase(unittest.TestCase):
21     class MockKeep(object):
22         def __init__(self, blocks):
23             self.blocks = blocks
24             self.requests = []
25         def get(self, locator, num_retries=0):
26             self.requests.append(locator)
27             return self.blocks.get(locator)
28         def get_from_cache(self, locator):
29             self.requests.append(locator)
30             return self.blocks.get(locator)
31         def put(self, data, num_retries=None):
32             pdh = "%s+%i" % (hashlib.md5(data).hexdigest(), len(data))
33             self.blocks[pdh] = str(data)
34             return pdh
35
36     class MockApi(object):
37         def __init__(self, b, r):
38             self.body = b
39             self.response = r
40         class MockCollections(object):
41             def __init__(self, b, r):
42                 self.body = b
43                 self.response = r
44             class Execute(object):
45                 def __init__(self, r):
46                     self.response = r
47                 def execute(self, num_retries=None):
48                     return self.response
49             def create(self, ensure_unique_name=False, body=None):
50                 if body != self.body:
51                     raise Exception("Body %s does not match expectation %s" % (body, self.body))
52                 return ArvadosFileWriterTestCase.MockApi.MockCollections.Execute(self.response)
53             def update(self, uuid=None, body=None):
54                 return ArvadosFileWriterTestCase.MockApi.MockCollections.Execute(self.response)
55         def collections(self):
56             return ArvadosFileWriterTestCase.MockApi.MockCollections(self.body, self.response)
57
58
59     def test_truncate(self):
60         keep = ArvadosFileWriterTestCase.MockKeep({"781e5e245d69b566979b86e28d23f2c7+10": "0123456789"})
61         api = ArvadosFileWriterTestCase.MockApi({"name":"test_truncate",
62                                                  "manifest_text":". 781e5e245d69b566979b86e28d23f2c7+10 0:8:count.txt\n"},
63                                                 {"uuid":"zzzzz-4zz18-mockcollection0",
64                                                  "manifest_text":". 781e5e245d69b566979b86e28d23f2c7+10 0:8:count.txt\n"})
65         with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n',
66                              api_client=api, keep_client=keep) as c:
67             writer = c.open("count.txt", "r+")
68             self.assertEqual(writer.size(), 10)
69             self.assertEqual("0123456789", writer.read(12))
70
71             writer.truncate(8)
72
73             # Make sure reading off the end doesn't break
74             self.assertEqual("", writer.read(12))
75
76             self.assertEqual(writer.size(), 8)
77             writer.seek(0, os.SEEK_SET)
78             self.assertEqual("01234567", writer.read(12))
79
80             self.assertEqual(None, c._manifest_locator)
81             self.assertEqual(True, c.modified())
82             c.save_new("test_truncate")
83             self.assertEqual("zzzzz-4zz18-mockcollection0", c._manifest_locator)
84             self.assertEqual(False, c.modified())
85
86     def test_write_to_end(self):
87         keep = ArvadosFileWriterTestCase.MockKeep({"781e5e245d69b566979b86e28d23f2c7+10": "0123456789"})
88         api = ArvadosFileWriterTestCase.MockApi({"name":"test_append",
89                                                  "manifest_text": ". 781e5e245d69b566979b86e28d23f2c7+10 acbd18db4cc2f85cedef654fccc4a4d8+3 0:13:count.txt\n"},
90                                                 {"uuid":"zzzzz-4zz18-mockcollection0",
91                                                  "manifest_text": ". 781e5e245d69b566979b86e28d23f2c7+10 acbd18db4cc2f85cedef654fccc4a4d8+3 0:13:count.txt\n"})
92         with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n',
93                              api_client=api, keep_client=keep) as c:
94             writer = c.open("count.txt", "r+")
95             self.assertEqual(writer.size(), 10)
96
97             writer.seek(5, os.SEEK_SET)
98             self.assertEqual("56789", writer.read(8))
99
100             writer.seek(10, os.SEEK_SET)
101             writer.write("foo")
102             self.assertEqual(writer.size(), 13)
103
104             writer.seek(5, os.SEEK_SET)
105             self.assertEqual("56789foo", writer.read(8))
106
107             self.assertEqual(None, c._manifest_locator)
108             self.assertEqual(True, c.modified())
109             self.assertEqual(None, keep.get("acbd18db4cc2f85cedef654fccc4a4d8+3"))
110
111             c.save_new("test_append")
112             self.assertEqual("zzzzz-4zz18-mockcollection0", c._manifest_locator)
113             self.assertEqual(False, c.modified())
114             self.assertEqual("foo", keep.get("acbd18db4cc2f85cedef654fccc4a4d8+3"))
115
116
117     def test_append(self):
118         keep = ArvadosFileWriterTestCase.MockKeep({"781e5e245d69b566979b86e28d23f2c7+10": "0123456789"})
119         c = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n', keep_client=keep)
120         writer = c.open("count.txt", "a+")
121         self.assertEqual(writer.read(20), "0123456789")
122         writer.seek(0, os.SEEK_SET)
123
124         writer.write("hello")
125         self.assertEqual(writer.read(20), "0123456789hello")
126         writer.seek(0, os.SEEK_SET)
127
128         writer.write("world")
129         self.assertEqual(writer.read(20), "0123456789helloworld")
130
131         self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 fc5e038d38a57032085441e7fe7010b0+10 0:20:count.txt\n", c.manifest_text())
132
133     def test_write_at_beginning(self):
134         keep = ArvadosFileWriterTestCase.MockKeep({"781e5e245d69b566979b86e28d23f2c7+10": "0123456789"})
135         with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n',
136                              keep_client=keep) as c:
137             writer = c.open("count.txt", "r+")
138             self.assertEqual("0123456789", writer.readfrom(0, 13))
139             writer.seek(0, os.SEEK_SET)
140             writer.write("foo")
141             self.assertEqual(writer.size(), 10)
142             self.assertEqual("foo3456789", writer.readfrom(0, 13))
143             self.assertEqual(". acbd18db4cc2f85cedef654fccc4a4d8+3 781e5e245d69b566979b86e28d23f2c7+10 0:3:count.txt 6:7:count.txt\n", c.manifest_text())
144
145     def test_write_in_middle(self):
146         keep = ArvadosFileWriterTestCase.MockKeep({"781e5e245d69b566979b86e28d23f2c7+10": "0123456789"})
147         with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n',
148                              keep_client=keep) as c:
149             writer = c.open("count.txt", "r+")
150             self.assertEqual("0123456789", writer.readfrom(0, 13))
151             writer.seek(3, os.SEEK_SET)
152             writer.write("foo")
153             self.assertEqual(writer.size(), 10)
154             self.assertEqual("012foo6789", writer.readfrom(0, 13))
155             self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 acbd18db4cc2f85cedef654fccc4a4d8+3 0:3:count.txt 10:3:count.txt 6:4:count.txt\n", c.manifest_text())
156
157     def test_write_at_end(self):
158         keep = ArvadosFileWriterTestCase.MockKeep({"781e5e245d69b566979b86e28d23f2c7+10": "0123456789"})
159         with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n',
160                              keep_client=keep) as c:
161             writer = c.open("count.txt", "r+")
162             self.assertEqual("0123456789", writer.readfrom(0, 13))
163             writer.seek(7, os.SEEK_SET)
164             writer.write("foo")
165             self.assertEqual(writer.size(), 10)
166             self.assertEqual("0123456foo", writer.readfrom(0, 13))
167             self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 acbd18db4cc2f85cedef654fccc4a4d8+3 0:7:count.txt 10:3:count.txt\n", c.manifest_text())
168
169     def test_write_across_segment_boundary(self):
170         keep = ArvadosFileWriterTestCase.MockKeep({"781e5e245d69b566979b86e28d23f2c7+10": "0123456789"})
171         with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt 0:10:count.txt\n',
172                              keep_client=keep) as c:
173             writer = c.open("count.txt", "r+")
174             self.assertEqual("012345678901234", writer.readfrom(0, 15))
175             writer.seek(7, os.SEEK_SET)
176             writer.write("foobar")
177             self.assertEqual(writer.size(), 20)
178             self.assertEqual("0123456foobar34", writer.readfrom(0, 15))
179             self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 3858f62230ac3c915f300c664312c63f+6 0:7:count.txt 10:6:count.txt 3:7:count.txt\n", c.manifest_text())
180
181     def test_write_across_several_segments(self):
182         keep = ArvadosFileWriterTestCase.MockKeep({"781e5e245d69b566979b86e28d23f2c7+10": "0123456789"})
183         with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:4:count.txt 0:4:count.txt 0:4:count.txt',
184                              keep_client=keep) as c:
185             writer = c.open("count.txt", "r+")
186             self.assertEqual("012301230123", writer.readfrom(0, 15))
187             writer.seek(2, os.SEEK_SET)
188             writer.write("abcdefg")
189             self.assertEqual(writer.size(), 12)
190             self.assertEqual("01abcdefg123", writer.readfrom(0, 15))
191             self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 7ac66c0f148de9519b8bd264312c4d64+7 0:2:count.txt 10:7:count.txt 1:3:count.txt\n", c.manifest_text())
192
193     def test_write_large(self):
194         keep = ArvadosFileWriterTestCase.MockKeep({})
195         api = ArvadosFileWriterTestCase.MockApi({"name":"test_write_large",
196                                                  "manifest_text": ". a5de24f4417cfba9d5825eadc2f4ca49+67108000 598cc1a4ccaef8ab6e4724d87e675d78+32892000 0:100000000:count.txt\n"},
197                                                 {"uuid":"zzzzz-4zz18-mockcollection0",
198                                                  "manifest_text": ". a5de24f4417cfba9d5825eadc2f4ca49+67108000 598cc1a4ccaef8ab6e4724d87e675d78+32892000 0:100000000:count.txt\n"})
199         with Collection('. ' + arvados.config.EMPTY_BLOCK_LOCATOR + ' 0:0:count.txt',
200                              api_client=api, keep_client=keep) as c:
201             writer = c.open("count.txt", "r+")
202             text = ''.join(["0123456789" for a in xrange(0, 100)])
203             for b in xrange(0, 100000):
204                 writer.write(text)
205             self.assertEqual(writer.size(), 100000000)
206
207             self.assertEqual(None, c._manifest_locator)
208             self.assertEqual(True, c.modified())
209             c.save_new("test_write_large")
210             self.assertEqual("zzzzz-4zz18-mockcollection0", c._manifest_locator)
211             self.assertEqual(False, c.modified())
212
213     def test_rewrite_on_empty_file(self):
214         keep = ArvadosFileWriterTestCase.MockKeep({})
215         with Collection('. ' + arvados.config.EMPTY_BLOCK_LOCATOR + ' 0:0:count.txt',
216                              keep_client=keep) as c:
217             writer = c.open("count.txt", "r+")
218             for b in xrange(0, 10):
219                 writer.seek(0, os.SEEK_SET)
220                 writer.write("0123456789")
221             writer.arvadosfile._repack_writes()
222             self.assertEqual(writer.size(), 10)
223             self.assertEqual("0123456789", writer.readfrom(0, 20))
224             self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n", c.manifest_text())
225
226     def test_rewrite_append_existing_file(self):
227         keep = ArvadosFileWriterTestCase.MockKeep({"781e5e245d69b566979b86e28d23f2c7+10": "0123456789"})
228         with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt',
229                              keep_client=keep) as c:
230             writer = c.open("count.txt", "r+")
231             for b in xrange(0, 10):
232                 writer.seek(10, os.SEEK_SET)
233                 writer.write("abcdefghij")
234             writer.arvadosfile._repack_writes()
235             self.assertEqual(writer.size(), 20)
236             self.assertEqual("0123456789abcdefghij", writer.readfrom(0, 20))
237             self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 a925576942e94b2ef57a066101b48876+10 0:20:count.txt\n", c.manifest_text())
238
239     def test_rewrite_over_existing_file(self):
240         keep = ArvadosFileWriterTestCase.MockKeep({"781e5e245d69b566979b86e28d23f2c7+10": "0123456789"})
241         with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt',
242                              keep_client=keep) as c:
243             writer = c.open("count.txt", "r+")
244             for b in xrange(0, 10):
245                 writer.seek(5, os.SEEK_SET)
246                 writer.write("abcdefghij")
247             writer.arvadosfile._repack_writes()
248             self.assertEqual(writer.size(), 15)
249             self.assertEqual("01234abcdefghij", writer.readfrom(0, 20))
250             self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 a925576942e94b2ef57a066101b48876+10 0:5:count.txt 10:10:count.txt\n", c.manifest_text())
251
252     def test_write_large_rewrite(self):
253         keep = ArvadosFileWriterTestCase.MockKeep({})
254         api = ArvadosFileWriterTestCase.MockApi({"name":"test_write_large",
255                                                  "manifest_text": ". 37400a68af9abdd76ca5bf13e819e42a+32892003 a5de24f4417cfba9d5825eadc2f4ca49+67108000 32892000:3:count.txt 32892006:67107997:count.txt 0:32892000:count.txt\n"},
256                                                 {"uuid":"zzzzz-4zz18-mockcollection0",
257                                                  "manifest_text": ". 37400a68af9abdd76ca5bf13e819e42a+32892003 a5de24f4417cfba9d5825eadc2f4ca49+67108000 32892000:3:count.txt 32892006:67107997:count.txt 0:32892000:count.txt\n"})
258         with Collection('. ' + arvados.config.EMPTY_BLOCK_LOCATOR + ' 0:0:count.txt',
259                              api_client=api, keep_client=keep) as c:
260             writer = c.open("count.txt", "r+")
261             text = ''.join(["0123456789" for a in xrange(0, 100)])
262             for b in xrange(0, 100000):
263                 writer.write(text)
264             writer.seek(0, os.SEEK_SET)
265             writer.write("foo")
266             self.assertEqual(writer.size(), 100000000)
267
268             self.assertEqual(None, c._manifest_locator)
269             self.assertEqual(True, c.modified())
270             c.save_new("test_write_large")
271             self.assertEqual("zzzzz-4zz18-mockcollection0", c._manifest_locator)
272             self.assertEqual(False, c.modified())
273
274     def test_create(self):
275         keep = ArvadosFileWriterTestCase.MockKeep({})
276         api = ArvadosFileWriterTestCase.MockApi({"name":"test_create",
277                                                  "manifest_text":". 2e9ec317e197819358fbc43afca7d837+8 0:8:count.txt\n"},
278                                                 {"uuid":"zzzzz-4zz18-mockcollection0",
279                                                  "manifest_text":". 2e9ec317e197819358fbc43afca7d837+8 0:8:count.txt\n"})
280         with Collection(api_client=api, keep_client=keep) as c:
281             writer = c.open("count.txt", "w+")
282             self.assertEqual(writer.size(), 0)
283             writer.write("01234567")
284             self.assertEqual(writer.size(), 8)
285
286             self.assertEqual(None, c._manifest_locator)
287             self.assertEqual(True, c.modified())
288             self.assertEqual(None, keep.get("2e9ec317e197819358fbc43afca7d837+8"))
289             c.save_new("test_create")
290             self.assertEqual("zzzzz-4zz18-mockcollection0", c._manifest_locator)
291             self.assertEqual(False, c.modified())
292             self.assertEqual("01234567", keep.get("2e9ec317e197819358fbc43afca7d837+8"))
293
294
295     def test_create_subdir(self):
296         keep = ArvadosFileWriterTestCase.MockKeep({})
297         api = ArvadosFileWriterTestCase.MockApi({"name":"test_create",
298                                                  "manifest_text":"./foo/bar 2e9ec317e197819358fbc43afca7d837+8 0:8:count.txt\n"},
299                                                 {"uuid":"zzzzz-4zz18-mockcollection0",
300                                                  "manifest_text":"./foo/bar 2e9ec317e197819358fbc43afca7d837+8 0:8:count.txt\n"})
301         with Collection(api_client=api, keep_client=keep) as c:
302             self.assertIsNone(c.api_response())
303             writer = c.open("foo/bar/count.txt", "w+")
304             writer.write("01234567")
305             c.save_new("test_create")
306             self.assertEqual(c.api_response(), api.response)
307
308     def test_overwrite(self):
309         keep = ArvadosFileWriterTestCase.MockKeep({"781e5e245d69b566979b86e28d23f2c7+10": "0123456789"})
310         api = ArvadosFileWriterTestCase.MockApi({"name":"test_overwrite",
311                                                  "manifest_text":". 2e9ec317e197819358fbc43afca7d837+8 0:8:count.txt\n"},
312                                                 {"uuid":"zzzzz-4zz18-mockcollection0",
313                                                  "manifest_text":". 2e9ec317e197819358fbc43afca7d837+8 0:8:count.txt\n"})
314         with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n',
315                              api_client=api, keep_client=keep) as c:
316             writer = c.open("count.txt", "w+")
317             self.assertEqual(writer.size(), 0)
318             writer.write("01234567")
319             self.assertEqual(writer.size(), 8)
320
321             self.assertEqual(None, c._manifest_locator)
322             self.assertEqual(True, c.modified())
323             c.save_new("test_overwrite")
324             self.assertEqual("zzzzz-4zz18-mockcollection0", c._manifest_locator)
325             self.assertEqual(False, c.modified())
326
327     def test_file_not_found(self):
328         with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n') as c:
329             with self.assertRaises(IOError):
330                 writer = c.open("nocount.txt", "r")
331
332     def test_cannot_open_directory(self):
333         with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n') as c:
334             with self.assertRaises(IOError):
335                 writer = c.open(".", "r")
336
337     def test_create_multiple(self):
338         keep = ArvadosFileWriterTestCase.MockKeep({})
339         api = ArvadosFileWriterTestCase.MockApi({"name":"test_create_multiple",
340                                                  "manifest_text":". 2e9ec317e197819358fbc43afca7d837+8 e8dc4081b13434b45189a720b77b6818+8 0:8:count1.txt 8:8:count2.txt\n"},
341                                                 {"uuid":"zzzzz-4zz18-mockcollection0",
342                                                  "manifest_text":". 2e9ec317e197819358fbc43afca7d837+8 e8dc4081b13434b45189a720b77b6818+8 0:8:count1.txt 8:8:count2.txt\n"})
343         with Collection(api_client=api, keep_client=keep) as c:
344             w1 = c.open("count1.txt", "w")
345             w2 = c.open("count2.txt", "w")
346             w1.write("01234567")
347             w2.write("abcdefgh")
348             self.assertEqual(w1.size(), 8)
349             self.assertEqual(w2.size(), 8)
350
351             self.assertEqual(None, c._manifest_locator)
352             self.assertEqual(True, c.modified())
353             self.assertEqual(None, keep.get("2e9ec317e197819358fbc43afca7d837+8"))
354             c.save_new("test_create_multiple")
355             self.assertEqual("zzzzz-4zz18-mockcollection0", c._manifest_locator)
356             self.assertEqual(False, c.modified())
357             self.assertEqual("01234567", keep.get("2e9ec317e197819358fbc43afca7d837+8"))
358
359
360 class ArvadosFileReaderTestCase(StreamFileReaderTestCase):
361     class MockParent(object):
362         class MockBlockMgr(object):
363             def __init__(self, blocks, nocache):
364                 self.blocks = blocks
365                 self.nocache = nocache
366
367             def block_prefetch(self, loc):
368                 pass
369
370             def get_block_contents(self, loc, num_retries=0, cache_only=False):
371                 if self.nocache and cache_only:
372                     return None
373                 return self.blocks[loc]
374
375         def __init__(self, blocks, nocache):
376             self.blocks = blocks
377             self.nocache = nocache
378             self.lock = arvados.arvfile.NoopLock()
379
380         def root_collection(self):
381             return self
382
383         def _my_block_manager(self):
384             return ArvadosFileReaderTestCase.MockParent.MockBlockMgr(self.blocks, self.nocache)
385
386         def sync_mode(self):
387             return SYNC_READONLY
388
389
390     def make_count_reader(self, nocache=False):
391         stream = []
392         n = 0
393         blocks = {}
394         for d in ['01234', '34567', '67890']:
395             loc = '{}+{}'.format(hashlib.md5(d).hexdigest(), len(d))
396             blocks[loc] = d
397             stream.append(Range(loc, n, len(d)))
398             n += len(d)
399         af = ArvadosFile(ArvadosFileReaderTestCase.MockParent(blocks, nocache), stream=stream, segments=[Range(1, 0, 3), Range(6, 3, 3), Range(11, 6, 3)])
400         return ArvadosFileReader(af, "count.txt")
401
402     def test_read_returns_first_block(self):
403         # read() calls will be aligned on block boundaries - see #3663.
404         sfile = self.make_count_reader(nocache=True)
405         self.assertEqual('123', sfile.read(10))
406
407     def test_successive_reads(self):
408         sfile = self.make_count_reader(nocache=True)
409         for expect in ['123', '456', '789', '']:
410             self.assertEqual(expect, sfile.read(10))
411
412     def test_tell_after_block_read(self):
413         sfile = self.make_count_reader(nocache=True)
414         sfile.read(5)
415         self.assertEqual(3, sfile.tell())
416
417     def test_prefetch(self):
418         keep = ArvadosFileWriterTestCase.MockKeep({"2e9ec317e197819358fbc43afca7d837+8": "01234567", "e8dc4081b13434b45189a720b77b6818+8": "abcdefgh"})
419         with Collection(". 2e9ec317e197819358fbc43afca7d837+8 e8dc4081b13434b45189a720b77b6818+8 0:16:count.txt\n", keep_client=keep) as c:
420             r = c.open("count.txt", "r")
421             self.assertEqual("0123", r.read(4))
422         self.assertIn("2e9ec317e197819358fbc43afca7d837+8", keep.requests)
423         self.assertIn("e8dc4081b13434b45189a720b77b6818+8", keep.requests)
424
425     def test__eq__from_manifest(self):
426         with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt') as c1:
427             with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt') as c2:
428                 self.assertTrue(c1["count1.txt"] == c2["count1.txt"])
429                 self.assertFalse(c1["count1.txt"] != c2["count1.txt"])
430
431     def test__eq__from_writes(self):
432         with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt') as c1:
433             with Collection() as c2:
434                 with c2.open("count1.txt", "w") as f:
435                     f.write("0123456789")
436
437                 self.assertTrue(c1["count1.txt"] == c2["count1.txt"])
438                 self.assertFalse(c1["count1.txt"] != c2["count1.txt"])
439
440     def test__ne__(self):
441         with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt') as c1:
442             with Collection() as c2:
443                 with c2.open("count1.txt", "w") as f:
444                     f.write("1234567890")
445
446                 self.assertTrue(c1["count1.txt"] != c2["count1.txt"])
447                 self.assertFalse(c1["count1.txt"] == c2["count1.txt"])
448
449
450 class ArvadosFileReadTestCase(unittest.TestCase, StreamRetryTestMixin):
451     def reader_for(self, coll_name, **kwargs):
452         stream = []
453         segments = []
454         n = 0
455         for d in self.manifest_for(coll_name).split():
456             try:
457                 k = KeepLocator(d)
458                 segments.append(Range(n, n, k.size))
459                 stream.append(Range(d, n, k.size))
460                 n += k.size
461             except ValueError:
462                 pass
463         col = Collection(keep_client=self.keep_client())
464         col._my_block_manager().prefetch_enabled = False
465         af = ArvadosFile(col,
466                          stream=stream,
467                          segments=segments)
468         return ArvadosFileReader(af, "test", **kwargs)
469
470     def read_for_test(self, reader, byte_count, **kwargs):
471         return reader.read(byte_count, **kwargs)
472
473
474 class ArvadosFileReadFromTestCase(ArvadosFileReadTestCase):
475     def read_for_test(self, reader, byte_count, **kwargs):
476         return reader.readfrom(0, byte_count, **kwargs)
477
478
479 class ArvadosFileReadAllTestCase(ArvadosFileReadTestCase):
480     def read_for_test(self, reader, byte_count, **kwargs):
481         return ''.join(reader.readall(**kwargs))
482
483
484 class ArvadosFileReadAllDecompressedTestCase(ArvadosFileReadTestCase):
485     def read_for_test(self, reader, byte_count, **kwargs):
486         return ''.join(reader.readall_decompressed(**kwargs))
487
488
489 class ArvadosFileReadlinesTestCase(ArvadosFileReadTestCase):
490     def read_for_test(self, reader, byte_count, **kwargs):
491         return ''.join(reader.readlines(**kwargs))
492
493 class BlockManagerTest(unittest.TestCase):
494     def test_bufferblock_append(self):
495         keep = ArvadosFileWriterTestCase.MockKeep({})
496         blockmanager = arvados.arvfile._BlockManager(keep)
497         bufferblock = blockmanager.alloc_bufferblock()
498         bufferblock.append("foo")
499
500         self.assertEqual(bufferblock.size(), 3)
501         self.assertEqual(bufferblock.buffer_view[0:3], "foo")
502         self.assertEqual(bufferblock.locator(), "acbd18db4cc2f85cedef654fccc4a4d8+3")
503
504         bufferblock.append("bar")
505
506         self.assertEqual(bufferblock.size(), 6)
507         self.assertEqual(bufferblock.buffer_view[0:6], "foobar")
508         self.assertEqual(bufferblock.locator(), "3858f62230ac3c915f300c664312c63f+6")
509
510         bufferblock.set_state(arvados.arvfile._BufferBlock.PENDING)
511         with self.assertRaises(arvados.errors.AssertionError):
512             bufferblock.append("bar")
513
514     def test_bufferblock_dup(self):
515         keep = ArvadosFileWriterTestCase.MockKeep({})
516         blockmanager = arvados.arvfile._BlockManager(keep)
517         bufferblock = blockmanager.alloc_bufferblock()
518         bufferblock.append("foo")
519
520         self.assertEqual(bufferblock.size(), 3)
521         self.assertEqual(bufferblock.buffer_view[0:3], "foo")
522         self.assertEqual(bufferblock.locator(), "acbd18db4cc2f85cedef654fccc4a4d8+3")
523         bufferblock.set_state(arvados.arvfile._BufferBlock.PENDING)
524
525         bufferblock2 = blockmanager.dup_block(bufferblock, None)
526         self.assertNotEqual(bufferblock.blockid, bufferblock2.blockid)
527
528         bufferblock2.append("bar")
529
530         self.assertEqual(bufferblock2.size(), 6)
531         self.assertEqual(bufferblock2.buffer_view[0:6], "foobar")
532         self.assertEqual(bufferblock2.locator(), "3858f62230ac3c915f300c664312c63f+6")
533
534         self.assertEqual(bufferblock.size(), 3)
535         self.assertEqual(bufferblock.buffer_view[0:3], "foo")
536         self.assertEqual(bufferblock.locator(), "acbd18db4cc2f85cedef654fccc4a4d8+3")
537
538     def test_bufferblock_get(self):
539         keep = ArvadosFileWriterTestCase.MockKeep({"781e5e245d69b566979b86e28d23f2c7+10": "0123456789"})
540         blockmanager = arvados.arvfile._BlockManager(keep)
541         bufferblock = blockmanager.alloc_bufferblock()
542         bufferblock.append("foo")
543
544         self.assertEqual(blockmanager.get_block_contents("781e5e245d69b566979b86e28d23f2c7+10", 1), "0123456789")
545         self.assertEqual(blockmanager.get_block_contents(bufferblock.blockid, 1), "foo")
546
547     def test_bufferblock_commit(self):
548         mockkeep = mock.MagicMock()
549         blockmanager = arvados.arvfile._BlockManager(mockkeep)
550         bufferblock = blockmanager.alloc_bufferblock()
551         bufferblock.append("foo")
552         blockmanager.commit_all()
553         self.assertTrue(mockkeep.put.called)
554         self.assertEqual(bufferblock.state(), arvados.arvfile._BufferBlock.COMMITTED)
555         self.assertIsNone(bufferblock.buffer_view)
556
557
558     def test_bufferblock_commit_with_error(self):
559         mockkeep = mock.MagicMock()
560         mockkeep.put.side_effect = arvados.errors.KeepWriteError("fail")
561         blockmanager = arvados.arvfile._BlockManager(mockkeep)
562         bufferblock = blockmanager.alloc_bufferblock()
563         bufferblock.append("foo")
564         with self.assertRaises(arvados.errors.KeepWriteError) as err:
565             blockmanager.commit_all()
566         self.assertEquals(str(err.exception), "Error writing some blocks: acbd18db4cc2f85cedef654fccc4a4d8+3 raised KeepWriteError (fail)")
567         self.assertEqual(bufferblock.state(), arvados.arvfile._BufferBlock.PENDING)