5692: Backed out commit_all() from Collection.manifest_text().
[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
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_empty(self):
146         keep = ArvadosFileWriterTestCase.MockKeep({})
147         with Collection(keep_client=keep) as c:
148             writer = c.open("count.txt", "w")
149             self.assertEqual(writer.size(), 0)
150             self.assertEqual(". d41d8cd98f00b204e9800998ecf8427e+0 0:0:count.txt\n", c.manifest_text())
151
152     def test_save_manifest_text(self):
153         keep = ArvadosFileWriterTestCase.MockKeep({})
154         with Collection(keep_client=keep) as c:
155             writer = c.open("count.txt", "w")
156             writer.write("0123456789")
157             self.assertEqual('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n', c.manifest_text())
158             self.assertNotIn('781e5e245d69b566979b86e28d23f2c7+10', keep.blocks)
159
160             self.assertEqual('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n', c.save_new(create_collection_record=False))
161             self.assertIn('781e5e245d69b566979b86e28d23f2c7+10', keep.blocks)
162
163     def test_write_in_middle(self):
164         keep = ArvadosFileWriterTestCase.MockKeep({"781e5e245d69b566979b86e28d23f2c7+10": "0123456789"})
165         with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n',
166                              keep_client=keep) as c:
167             writer = c.open("count.txt", "r+")
168             self.assertEqual("0123456789", writer.readfrom(0, 13))
169             writer.seek(3, os.SEEK_SET)
170             writer.write("foo")
171             self.assertEqual(writer.size(), 10)
172             self.assertEqual("012foo6789", writer.readfrom(0, 13))
173             self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 acbd18db4cc2f85cedef654fccc4a4d8+3 0:3:count.txt 10:3:count.txt 6:4:count.txt\n", c.manifest_text())
174
175     def test_write_at_end(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(7, os.SEEK_SET)
182             writer.write("foo")
183             self.assertEqual(writer.size(), 10)
184             self.assertEqual("0123456foo", writer.readfrom(0, 13))
185             self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 acbd18db4cc2f85cedef654fccc4a4d8+3 0:7:count.txt 10:3:count.txt\n", c.manifest_text())
186
187     def test_write_across_segment_boundary(self):
188         keep = ArvadosFileWriterTestCase.MockKeep({"781e5e245d69b566979b86e28d23f2c7+10": "0123456789"})
189         with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt 0:10:count.txt\n',
190                              keep_client=keep) as c:
191             writer = c.open("count.txt", "r+")
192             self.assertEqual("012345678901234", writer.readfrom(0, 15))
193             writer.seek(7, os.SEEK_SET)
194             writer.write("foobar")
195             self.assertEqual(writer.size(), 20)
196             self.assertEqual("0123456foobar34", writer.readfrom(0, 15))
197             self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 3858f62230ac3c915f300c664312c63f+6 0:7:count.txt 10:6:count.txt 3:7:count.txt\n", c.manifest_text())
198
199     def test_write_across_several_segments(self):
200         keep = ArvadosFileWriterTestCase.MockKeep({"781e5e245d69b566979b86e28d23f2c7+10": "0123456789"})
201         with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:4:count.txt 0:4:count.txt 0:4:count.txt',
202                              keep_client=keep) as c:
203             writer = c.open("count.txt", "r+")
204             self.assertEqual("012301230123", writer.readfrom(0, 15))
205             writer.seek(2, os.SEEK_SET)
206             writer.write("abcdefg")
207             self.assertEqual(writer.size(), 12)
208             self.assertEqual("01abcdefg123", writer.readfrom(0, 15))
209             self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 7ac66c0f148de9519b8bd264312c4d64+7 0:2:count.txt 10:7:count.txt 1:3:count.txt\n", c.manifest_text())
210
211     def test_write_large(self):
212         keep = ArvadosFileWriterTestCase.MockKeep({})
213         api = ArvadosFileWriterTestCase.MockApi({"name":"test_write_large",
214                                                  "manifest_text": ". a5de24f4417cfba9d5825eadc2f4ca49+67108000 598cc1a4ccaef8ab6e4724d87e675d78+32892000 0:100000000:count.txt\n"},
215                                                 {"uuid":"zzzzz-4zz18-mockcollection0",
216                                                  "manifest_text": ". a5de24f4417cfba9d5825eadc2f4ca49+67108000 598cc1a4ccaef8ab6e4724d87e675d78+32892000 0:100000000:count.txt\n"})
217         with Collection('. ' + arvados.config.EMPTY_BLOCK_LOCATOR + ' 0:0:count.txt',
218                              api_client=api, keep_client=keep) as c:
219             writer = c.open("count.txt", "r+")
220             text = ''.join(["0123456789" for a in xrange(0, 100)])
221             for b in xrange(0, 100000):
222                 writer.write(text)
223             self.assertEqual(writer.size(), 100000000)
224
225             self.assertEqual(None, c.manifest_locator())
226             self.assertEqual(True, c.modified())
227             c.save_new("test_write_large")
228             self.assertEqual("zzzzz-4zz18-mockcollection0", c.manifest_locator())
229             self.assertEqual(False, c.modified())
230
231     def test_rewrite_on_empty_file(self):
232         keep = ArvadosFileWriterTestCase.MockKeep({})
233         with Collection('. ' + arvados.config.EMPTY_BLOCK_LOCATOR + ' 0:0:count.txt',
234                              keep_client=keep) as c:
235             writer = c.open("count.txt", "r+")
236             for b in xrange(0, 10):
237                 writer.seek(0, os.SEEK_SET)
238                 writer.write("0123456789")
239
240             self.assertEqual(writer.size(), 10)
241             self.assertEqual("0123456789", writer.readfrom(0, 20))
242             self.assertEqual(". 7a08b07e84641703e5f2c836aa59a170+100 90:10:count.txt\n", c.manifest_text(strip=True))
243             writer.flush()
244             self.assertEqual(writer.size(), 10)
245             self.assertEqual("0123456789", writer.readfrom(0, 20))
246             self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n", c.manifest_text(strip=True))
247
248     def test_rewrite_append_existing_file(self):
249         keep = ArvadosFileWriterTestCase.MockKeep({"781e5e245d69b566979b86e28d23f2c7+10": "0123456789"})
250         with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt',
251                              keep_client=keep) as c:
252             writer = c.open("count.txt", "r+")
253             for b in xrange(0, 10):
254                 writer.seek(10, os.SEEK_SET)
255                 writer.write("abcdefghij")
256
257             self.assertEqual(writer.size(), 20)
258             self.assertEqual("0123456789abcdefghij", writer.readfrom(0, 20))
259             self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 ae5f43bab79cf0be33f025fa97ae7398+100 0:10:count.txt 100:10:count.txt\n", c.manifest_text(strip=True))
260
261             writer.arvadosfile.flush()
262             self.assertEqual(writer.size(), 20)
263             self.assertEqual("0123456789abcdefghij", writer.readfrom(0, 20))
264             self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 a925576942e94b2ef57a066101b48876+10 0:20:count.txt\n", c.manifest_text(strip=True))
265
266     def test_rewrite_over_existing_file(self):
267         keep = ArvadosFileWriterTestCase.MockKeep({"781e5e245d69b566979b86e28d23f2c7+10": "0123456789"})
268         with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt',
269                              keep_client=keep) as c:
270             writer = c.open("count.txt", "r+")
271             for b in xrange(0, 10):
272                 writer.seek(5, os.SEEK_SET)
273                 writer.write("abcdefghij")
274
275             self.assertEqual(writer.size(), 15)
276             self.assertEqual("01234abcdefghij", writer.readfrom(0, 20))
277             self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 ae5f43bab79cf0be33f025fa97ae7398+100 0:5:count.txt 100:10:count.txt\n", c.manifest_text(strip=True))
278
279             writer.arvadosfile.flush()
280
281             self.assertEqual(writer.size(), 15)
282             self.assertEqual("01234abcdefghij", writer.readfrom(0, 20))
283             self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 a925576942e94b2ef57a066101b48876+10 0:5:count.txt 10:10:count.txt\n", c.manifest_text(strip=True))
284
285     def test_write_large_rewrite(self):
286         keep = ArvadosFileWriterTestCase.MockKeep({})
287         api = ArvadosFileWriterTestCase.MockApi({"name":"test_write_large",
288                                                  "manifest_text": ". 37400a68af9abdd76ca5bf13e819e42a+32892003 a5de24f4417cfba9d5825eadc2f4ca49+67108000 32892000:3:count.txt 32892006:67107997:count.txt 0:32892000:count.txt\n"},
289                                                 {"uuid":"zzzzz-4zz18-mockcollection0",
290                                                  "manifest_text": ". 37400a68af9abdd76ca5bf13e819e42a+32892003 a5de24f4417cfba9d5825eadc2f4ca49+67108000 32892000:3:count.txt 32892006:67107997:count.txt 0:32892000:count.txt\n"})
291         with Collection('. ' + arvados.config.EMPTY_BLOCK_LOCATOR + ' 0:0:count.txt',
292                              api_client=api, keep_client=keep) as c:
293             writer = c.open("count.txt", "r+")
294             text = ''.join(["0123456789" for a in xrange(0, 100)])
295             for b in xrange(0, 100000):
296                 writer.write(text)
297             writer.seek(0, os.SEEK_SET)
298             writer.write("foo")
299             self.assertEqual(writer.size(), 100000000)
300
301             self.assertEqual(None, c.manifest_locator())
302             self.assertEqual(True, c.modified())
303             c.save_new("test_write_large")
304             self.assertEqual("zzzzz-4zz18-mockcollection0", c.manifest_locator())
305             self.assertEqual(False, c.modified())
306
307     def test_create(self):
308         keep = ArvadosFileWriterTestCase.MockKeep({})
309         api = ArvadosFileWriterTestCase.MockApi({"name":"test_create",
310                                                  "manifest_text":". 2e9ec317e197819358fbc43afca7d837+8 0:8:count.txt\n"},
311                                                 {"uuid":"zzzzz-4zz18-mockcollection0",
312                                                  "manifest_text":". 2e9ec317e197819358fbc43afca7d837+8 0:8:count.txt\n"})
313         with Collection(api_client=api, keep_client=keep) as c:
314             writer = c.open("count.txt", "w+")
315             self.assertEqual(writer.size(), 0)
316             writer.write("01234567")
317             self.assertEqual(writer.size(), 8)
318
319             self.assertEqual(None, c.manifest_locator())
320             self.assertEqual(True, c.modified())
321             self.assertEqual(None, keep.get("2e9ec317e197819358fbc43afca7d837+8"))
322             c.save_new("test_create")
323             self.assertEqual("zzzzz-4zz18-mockcollection0", c.manifest_locator())
324             self.assertEqual(False, c.modified())
325             self.assertEqual("01234567", keep.get("2e9ec317e197819358fbc43afca7d837+8"))
326
327
328     def test_create_subdir(self):
329         keep = ArvadosFileWriterTestCase.MockKeep({})
330         api = ArvadosFileWriterTestCase.MockApi({"name":"test_create",
331                                                  "manifest_text":"./foo/bar 2e9ec317e197819358fbc43afca7d837+8 0:8:count.txt\n"},
332                                                 {"uuid":"zzzzz-4zz18-mockcollection0",
333                                                  "manifest_text":"./foo/bar 2e9ec317e197819358fbc43afca7d837+8 0:8:count.txt\n"})
334         with Collection(api_client=api, keep_client=keep) as c:
335             self.assertIsNone(c.api_response())
336             writer = c.open("foo/bar/count.txt", "w+")
337             writer.write("01234567")
338             c.save_new("test_create")
339             self.assertEqual(c.api_response(), api.response)
340
341     def test_overwrite(self):
342         keep = ArvadosFileWriterTestCase.MockKeep({"781e5e245d69b566979b86e28d23f2c7+10": "0123456789"})
343         api = ArvadosFileWriterTestCase.MockApi({"name":"test_overwrite",
344                                                  "manifest_text":". 2e9ec317e197819358fbc43afca7d837+8 0:8:count.txt\n"},
345                                                 {"uuid":"zzzzz-4zz18-mockcollection0",
346                                                  "manifest_text":". 2e9ec317e197819358fbc43afca7d837+8 0:8:count.txt\n"})
347         with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n',
348                              api_client=api, keep_client=keep) as c:
349             writer = c.open("count.txt", "w+")
350             self.assertEqual(writer.size(), 0)
351             writer.write("01234567")
352             self.assertEqual(writer.size(), 8)
353
354             self.assertEqual(None, c.manifest_locator())
355             self.assertEqual(True, c.modified())
356             c.save_new("test_overwrite")
357             self.assertEqual("zzzzz-4zz18-mockcollection0", c.manifest_locator())
358             self.assertEqual(False, c.modified())
359
360     def test_file_not_found(self):
361         with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n') as c:
362             with self.assertRaises(IOError):
363                 writer = c.open("nocount.txt", "r")
364
365     def test_cannot_open_directory(self):
366         with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n') as c:
367             with self.assertRaises(IOError):
368                 writer = c.open(".", "r")
369
370     def test_create_multiple(self):
371         keep = ArvadosFileWriterTestCase.MockKeep({})
372         api = ArvadosFileWriterTestCase.MockApi({"name":"test_create_multiple",
373                                                  "manifest_text":". 2e9ec317e197819358fbc43afca7d837+8 e8dc4081b13434b45189a720b77b6818+8 0:8:count1.txt 8:8:count2.txt\n"},
374                                                 {"uuid":"zzzzz-4zz18-mockcollection0",
375                                                  "manifest_text":". 2e9ec317e197819358fbc43afca7d837+8 e8dc4081b13434b45189a720b77b6818+8 0:8:count1.txt 8:8:count2.txt\n"})
376         with Collection(api_client=api, keep_client=keep) as c:
377             w1 = c.open("count1.txt", "w")
378             w2 = c.open("count2.txt", "w")
379             w1.write("01234567")
380             w2.write("abcdefgh")
381             self.assertEqual(w1.size(), 8)
382             self.assertEqual(w2.size(), 8)
383
384             self.assertEqual(None, c.manifest_locator())
385             self.assertEqual(True, c.modified())
386             self.assertEqual(None, keep.get("2e9ec317e197819358fbc43afca7d837+8"))
387             c.save_new("test_create_multiple")
388             self.assertEqual("zzzzz-4zz18-mockcollection0", c.manifest_locator())
389             self.assertEqual(False, c.modified())
390             self.assertEqual("01234567", keep.get("2e9ec317e197819358fbc43afca7d837+8"))
391
392
393 class ArvadosFileReaderTestCase(StreamFileReaderTestCase):
394     class MockParent(object):
395         class MockBlockMgr(object):
396             def __init__(self, blocks, nocache):
397                 self.blocks = blocks
398                 self.nocache = nocache
399
400             def block_prefetch(self, loc):
401                 pass
402
403             def get_block_contents(self, loc, num_retries=0, cache_only=False):
404                 if self.nocache and cache_only:
405                     return None
406                 return self.blocks[loc]
407
408         def __init__(self, blocks, nocache):
409             self.blocks = blocks
410             self.nocache = nocache
411             self.lock = arvados.arvfile.NoopLock()
412
413         def root_collection(self):
414             return self
415
416         def _my_block_manager(self):
417             return ArvadosFileReaderTestCase.MockParent.MockBlockMgr(self.blocks, self.nocache)
418
419
420     def make_count_reader(self, nocache=False):
421         stream = []
422         n = 0
423         blocks = {}
424         for d in ['01234', '34567', '67890']:
425             loc = '{}+{}'.format(hashlib.md5(d).hexdigest(), len(d))
426             blocks[loc] = d
427             stream.append(Range(loc, n, len(d)))
428             n += len(d)
429         af = ArvadosFile(ArvadosFileReaderTestCase.MockParent(blocks, nocache), stream=stream, segments=[Range(1, 0, 3), Range(6, 3, 3), Range(11, 6, 3)])
430         return ArvadosFileReader(af, "count.txt")
431
432     def test_read_returns_first_block(self):
433         # read() calls will be aligned on block boundaries - see #3663.
434         sfile = self.make_count_reader(nocache=True)
435         self.assertEqual('123', sfile.read(10))
436
437     def test_successive_reads(self):
438         sfile = self.make_count_reader(nocache=True)
439         for expect in ['123', '456', '789', '']:
440             self.assertEqual(expect, sfile.read(10))
441
442     def test_tell_after_block_read(self):
443         sfile = self.make_count_reader(nocache=True)
444         sfile.read(5)
445         self.assertEqual(3, sfile.tell())
446
447     def test_prefetch(self):
448         keep = ArvadosFileWriterTestCase.MockKeep({"2e9ec317e197819358fbc43afca7d837+8": "01234567", "e8dc4081b13434b45189a720b77b6818+8": "abcdefgh"})
449         with Collection(". 2e9ec317e197819358fbc43afca7d837+8 e8dc4081b13434b45189a720b77b6818+8 0:16:count.txt\n", keep_client=keep) as c:
450             r = c.open("count.txt", "r")
451             self.assertEqual("0123", r.read(4))
452         self.assertIn("2e9ec317e197819358fbc43afca7d837+8", keep.requests)
453         self.assertIn("e8dc4081b13434b45189a720b77b6818+8", keep.requests)
454
455     def test__eq__from_manifest(self):
456         with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt') as c1:
457             with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt') as c2:
458                 self.assertTrue(c1["count1.txt"] == c2["count1.txt"])
459                 self.assertFalse(c1["count1.txt"] != c2["count1.txt"])
460
461     def test__eq__from_writes(self):
462         with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt') as c1:
463             with Collection() as c2:
464                 with c2.open("count1.txt", "w") as f:
465                     f.write("0123456789")
466
467                 self.assertTrue(c1["count1.txt"] == c2["count1.txt"])
468                 self.assertFalse(c1["count1.txt"] != c2["count1.txt"])
469
470     def test__ne__(self):
471         with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt') as c1:
472             with Collection() as c2:
473                 with c2.open("count1.txt", "w") as f:
474                     f.write("1234567890")
475
476                 self.assertTrue(c1["count1.txt"] != c2["count1.txt"])
477                 self.assertFalse(c1["count1.txt"] == c2["count1.txt"])
478
479
480 class ArvadosFileReadTestCase(unittest.TestCase, StreamRetryTestMixin):
481     def reader_for(self, coll_name, **kwargs):
482         stream = []
483         segments = []
484         n = 0
485         for d in self.manifest_for(coll_name).split():
486             try:
487                 k = KeepLocator(d)
488                 segments.append(Range(n, n, k.size))
489                 stream.append(Range(d, n, k.size))
490                 n += k.size
491             except ValueError:
492                 pass
493
494         blockmanager = arvados.arvfile._BlockManager(self.keep_client())
495         blockmanager.prefetch_enabled = False
496         col = Collection(keep_client=self.keep_client(), block_manager=blockmanager)
497         af = ArvadosFile(col,
498                          stream=stream,
499                          segments=segments)
500         return ArvadosFileReader(af, "test", **kwargs)
501
502     def read_for_test(self, reader, byte_count, **kwargs):
503         return reader.read(byte_count, **kwargs)
504
505
506 class ArvadosFileReadFromTestCase(ArvadosFileReadTestCase):
507     def read_for_test(self, reader, byte_count, **kwargs):
508         return reader.readfrom(0, byte_count, **kwargs)
509
510
511 class ArvadosFileReadAllTestCase(ArvadosFileReadTestCase):
512     def read_for_test(self, reader, byte_count, **kwargs):
513         return ''.join(reader.readall(**kwargs))
514
515
516 class ArvadosFileReadAllDecompressedTestCase(ArvadosFileReadTestCase):
517     def read_for_test(self, reader, byte_count, **kwargs):
518         return ''.join(reader.readall_decompressed(**kwargs))
519
520
521 class ArvadosFileReadlinesTestCase(ArvadosFileReadTestCase):
522     def read_for_test(self, reader, byte_count, **kwargs):
523         return ''.join(reader.readlines(**kwargs))
524
525 class BlockManagerTest(unittest.TestCase):
526     def test_bufferblock_append(self):
527         keep = ArvadosFileWriterTestCase.MockKeep({})
528         blockmanager = arvados.arvfile._BlockManager(keep)
529         bufferblock = blockmanager.alloc_bufferblock()
530         bufferblock.append("foo")
531
532         self.assertEqual(bufferblock.size(), 3)
533         self.assertEqual(bufferblock.buffer_view[0:3], "foo")
534         self.assertEqual(bufferblock.locator(), "acbd18db4cc2f85cedef654fccc4a4d8+3")
535
536         bufferblock.append("bar")
537
538         self.assertEqual(bufferblock.size(), 6)
539         self.assertEqual(bufferblock.buffer_view[0:6], "foobar")
540         self.assertEqual(bufferblock.locator(), "3858f62230ac3c915f300c664312c63f+6")
541
542         bufferblock.set_state(arvados.arvfile._BufferBlock.PENDING)
543         with self.assertRaises(arvados.errors.AssertionError):
544             bufferblock.append("bar")
545
546     def test_bufferblock_dup(self):
547         keep = ArvadosFileWriterTestCase.MockKeep({})
548         blockmanager = arvados.arvfile._BlockManager(keep)
549         bufferblock = blockmanager.alloc_bufferblock()
550         bufferblock.append("foo")
551
552         self.assertEqual(bufferblock.size(), 3)
553         self.assertEqual(bufferblock.buffer_view[0:3], "foo")
554         self.assertEqual(bufferblock.locator(), "acbd18db4cc2f85cedef654fccc4a4d8+3")
555         bufferblock.set_state(arvados.arvfile._BufferBlock.PENDING)
556
557         bufferblock2 = blockmanager.dup_block(bufferblock, None)
558         self.assertNotEqual(bufferblock.blockid, bufferblock2.blockid)
559
560         bufferblock2.append("bar")
561
562         self.assertEqual(bufferblock2.size(), 6)
563         self.assertEqual(bufferblock2.buffer_view[0:6], "foobar")
564         self.assertEqual(bufferblock2.locator(), "3858f62230ac3c915f300c664312c63f+6")
565
566         self.assertEqual(bufferblock.size(), 3)
567         self.assertEqual(bufferblock.buffer_view[0:3], "foo")
568         self.assertEqual(bufferblock.locator(), "acbd18db4cc2f85cedef654fccc4a4d8+3")
569
570     def test_bufferblock_get(self):
571         keep = ArvadosFileWriterTestCase.MockKeep({"781e5e245d69b566979b86e28d23f2c7+10": "0123456789"})
572         blockmanager = arvados.arvfile._BlockManager(keep)
573         bufferblock = blockmanager.alloc_bufferblock()
574         bufferblock.append("foo")
575
576         self.assertEqual(blockmanager.get_block_contents("781e5e245d69b566979b86e28d23f2c7+10", 1), "0123456789")
577         self.assertEqual(blockmanager.get_block_contents(bufferblock.blockid, 1), "foo")
578
579     def test_bufferblock_commit(self):
580         mockkeep = mock.MagicMock()
581         blockmanager = arvados.arvfile._BlockManager(mockkeep)
582         bufferblock = blockmanager.alloc_bufferblock()
583         bufferblock.owner = ArvadosFile(Collection(block_manager=blockmanager),
584                                         [Range(bufferblock.blockid, 0, 3)], [Range(0, 0, 3)])
585         bufferblock.owner._current_bblock = bufferblock
586         bufferblock.append("foo")
587         blockmanager.commit_all()
588         self.assertTrue(mockkeep.put.called)
589         self.assertEqual(bufferblock.state(), arvados.arvfile._BufferBlock.COMMITTED)
590         self.assertIsNone(bufferblock.buffer_view)
591
592
593     def test_bufferblock_commit_with_error(self):
594         mockkeep = mock.MagicMock()
595         mockkeep.put.side_effect = arvados.errors.KeepWriteError("fail")
596         blockmanager = arvados.arvfile._BlockManager(mockkeep)
597         bufferblock = blockmanager.alloc_bufferblock()
598         bufferblock.owner = ArvadosFile(Collection(block_manager=blockmanager),
599                                         [Range(bufferblock.blockid, 0, 3)], [Range(0, 0, 3)])
600         bufferblock.owner._current_bblock = bufferblock
601         bufferblock.append("foo")
602         with self.assertRaises(arvados.errors.KeepWriteError) as err:
603             blockmanager.commit_all()
604         self.assertEqual(str(err.exception), "Error writing some blocks: block acbd18db4cc2f85cedef654fccc4a4d8+3 raised KeepWriteError (fail)")
605         self.assertEqual(bufferblock.state(), arvados.arvfile._BufferBlock.PENDING)