6194: Make splitting loop simpler since [n:n+KEEP_BLOCK_SIZE] returns a short
[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.portable_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.portable_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.portable_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.portable_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_get_manifest_text_commits(self):
164          keep = ArvadosFileWriterTestCase.MockKeep({})
165          with Collection(keep_client=keep) as c:
166              writer = c.open("count.txt", "w")
167              writer.write("0123456789")
168              self.assertEqual('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n', c.portable_manifest_text())
169              self.assertNotIn('781e5e245d69b566979b86e28d23f2c7+10', keep.blocks)
170              self.assertEqual('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n', c.manifest_text())
171              self.assertIn('781e5e245d69b566979b86e28d23f2c7+10', keep.blocks)
172
173
174     def test_write_in_middle(self):
175         keep = ArvadosFileWriterTestCase.MockKeep({"781e5e245d69b566979b86e28d23f2c7+10": "0123456789"})
176         with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n',
177                              keep_client=keep) as c:
178             writer = c.open("count.txt", "r+")
179             self.assertEqual("0123456789", writer.readfrom(0, 13))
180             writer.seek(3, os.SEEK_SET)
181             writer.write("foo")
182             self.assertEqual(writer.size(), 10)
183             self.assertEqual("012foo6789", writer.readfrom(0, 13))
184             self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 acbd18db4cc2f85cedef654fccc4a4d8+3 0:3:count.txt 10:3:count.txt 6:4:count.txt\n", c.portable_manifest_text())
185
186     def test_write_at_end(self):
187         keep = ArvadosFileWriterTestCase.MockKeep({"781e5e245d69b566979b86e28d23f2c7+10": "0123456789"})
188         with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n',
189                              keep_client=keep) as c:
190             writer = c.open("count.txt", "r+")
191             self.assertEqual("0123456789", writer.readfrom(0, 13))
192             writer.seek(7, os.SEEK_SET)
193             writer.write("foo")
194             self.assertEqual(writer.size(), 10)
195             self.assertEqual("0123456foo", writer.readfrom(0, 13))
196             self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 acbd18db4cc2f85cedef654fccc4a4d8+3 0:7:count.txt 10:3:count.txt\n", c.portable_manifest_text())
197
198     def test_write_across_segment_boundary(self):
199         keep = ArvadosFileWriterTestCase.MockKeep({"781e5e245d69b566979b86e28d23f2c7+10": "0123456789"})
200         with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt 0:10:count.txt\n',
201                              keep_client=keep) as c:
202             writer = c.open("count.txt", "r+")
203             self.assertEqual("012345678901234", writer.readfrom(0, 15))
204             writer.seek(7, os.SEEK_SET)
205             writer.write("foobar")
206             self.assertEqual(writer.size(), 20)
207             self.assertEqual("0123456foobar34", writer.readfrom(0, 15))
208             self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 3858f62230ac3c915f300c664312c63f+6 0:7:count.txt 10:6:count.txt 3:7:count.txt\n", c.portable_manifest_text())
209
210     def test_write_across_several_segments(self):
211         keep = ArvadosFileWriterTestCase.MockKeep({"781e5e245d69b566979b86e28d23f2c7+10": "0123456789"})
212         with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:4:count.txt 0:4:count.txt 0:4:count.txt',
213                              keep_client=keep) as c:
214             writer = c.open("count.txt", "r+")
215             self.assertEqual("012301230123", writer.readfrom(0, 15))
216             writer.seek(2, os.SEEK_SET)
217             writer.write("abcdefg")
218             self.assertEqual(writer.size(), 12)
219             self.assertEqual("01abcdefg123", writer.readfrom(0, 15))
220             self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 7ac66c0f148de9519b8bd264312c4d64+7 0:2:count.txt 10:7:count.txt 1:3:count.txt\n", c.portable_manifest_text())
221
222     def test_write_large(self):
223         keep = ArvadosFileWriterTestCase.MockKeep({})
224         api = ArvadosFileWriterTestCase.MockApi({"name":"test_write_large",
225                                                  "manifest_text": ". a5de24f4417cfba9d5825eadc2f4ca49+67108000 598cc1a4ccaef8ab6e4724d87e675d78+32892000 0:100000000:count.txt\n"},
226                                                 {"uuid":"zzzzz-4zz18-mockcollection0",
227                                                  "manifest_text": ". a5de24f4417cfba9d5825eadc2f4ca49+67108000 598cc1a4ccaef8ab6e4724d87e675d78+32892000 0:100000000:count.txt\n"})
228         with Collection('. ' + arvados.config.EMPTY_BLOCK_LOCATOR + ' 0:0:count.txt',
229                              api_client=api, keep_client=keep) as c:
230             writer = c.open("count.txt", "r+")
231             text = "0123456789" * 100
232             for b in xrange(0, 100000):
233                 writer.write(text)
234             self.assertEqual(writer.size(), 100000000)
235
236             self.assertEqual(None, c.manifest_locator())
237             self.assertEqual(True, c.modified())
238             c.save_new("test_write_large")
239             self.assertEqual("zzzzz-4zz18-mockcollection0", c.manifest_locator())
240             self.assertEqual(False, c.modified())
241
242
243     def test_large_write(self):
244         keep = ArvadosFileWriterTestCase.MockKeep({})
245         api = ArvadosFileWriterTestCase.MockApi({"name":"test_write_large",
246                                                  "manifest_text": ". 781e5e245d69b566979b86e28d23f2c7+10 48dd23ea1645fd47d789804d71b5bb8e+67108864 77c57dc6ac5a10bb2205caaa73187994+32891126 0:100000000:count.txt\n"},
247                                                 {"uuid":"zzzzz-4zz18-mockcollection0",
248                                                  "manifest_text": ". 781e5e245d69b566979b86e28d23f2c7+10 48dd23ea1645fd47d789804d71b5bb8e+67108864 77c57dc6ac5a10bb2205caaa73187994+32891126 0:100000000:count.txt\n"})
249         with Collection('. ' + arvados.config.EMPTY_BLOCK_LOCATOR + ' 0:0:count.txt',
250                              api_client=api, keep_client=keep) as c:
251             writer = c.open("count.txt", "r+")
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(None, c.manifest_locator())
259             self.assertEqual(True, c.modified())
260             c.save_new("test_write_large")
261             self.assertEqual("zzzzz-4zz18-mockcollection0", c.manifest_locator())
262             self.assertEqual(False, c.modified())
263
264     def test_rewrite_on_empty_file(self):
265         keep = ArvadosFileWriterTestCase.MockKeep({})
266         with Collection('. ' + arvados.config.EMPTY_BLOCK_LOCATOR + ' 0:0:count.txt',
267                              keep_client=keep) as c:
268             writer = c.open("count.txt", "r+")
269             for b in xrange(0, 10):
270                 writer.seek(0, os.SEEK_SET)
271                 writer.write("0123456789")
272
273             self.assertEqual(writer.size(), 10)
274             self.assertEqual("0123456789", writer.readfrom(0, 20))
275             self.assertEqual(". 7a08b07e84641703e5f2c836aa59a170+100 90:10:count.txt\n", c.portable_manifest_text())
276             writer.flush()
277             self.assertEqual(writer.size(), 10)
278             self.assertEqual("0123456789", writer.readfrom(0, 20))
279             self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n", c.portable_manifest_text())
280
281     def test_rewrite_append_existing_file(self):
282         keep = ArvadosFileWriterTestCase.MockKeep({"781e5e245d69b566979b86e28d23f2c7+10": "0123456789"})
283         with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt',
284                              keep_client=keep) as c:
285             writer = c.open("count.txt", "r+")
286             for b in xrange(0, 10):
287                 writer.seek(10, os.SEEK_SET)
288                 writer.write("abcdefghij")
289
290             self.assertEqual(writer.size(), 20)
291             self.assertEqual("0123456789abcdefghij", writer.readfrom(0, 20))
292             self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 ae5f43bab79cf0be33f025fa97ae7398+100 0:10:count.txt 100:10:count.txt\n", c.portable_manifest_text())
293
294             writer.arvadosfile.flush()
295             self.assertEqual(writer.size(), 20)
296             self.assertEqual("0123456789abcdefghij", writer.readfrom(0, 20))
297             self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 a925576942e94b2ef57a066101b48876+10 0:20:count.txt\n", c.portable_manifest_text())
298
299     def test_rewrite_over_existing_file(self):
300         keep = ArvadosFileWriterTestCase.MockKeep({"781e5e245d69b566979b86e28d23f2c7+10": "0123456789"})
301         with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt',
302                              keep_client=keep) as c:
303             writer = c.open("count.txt", "r+")
304             for b in xrange(0, 10):
305                 writer.seek(5, os.SEEK_SET)
306                 writer.write("abcdefghij")
307
308             self.assertEqual(writer.size(), 15)
309             self.assertEqual("01234abcdefghij", writer.readfrom(0, 20))
310             self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 ae5f43bab79cf0be33f025fa97ae7398+100 0:5:count.txt 100:10:count.txt\n", c.portable_manifest_text())
311
312             writer.arvadosfile.flush()
313
314             self.assertEqual(writer.size(), 15)
315             self.assertEqual("01234abcdefghij", writer.readfrom(0, 20))
316             self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 a925576942e94b2ef57a066101b48876+10 0:5:count.txt 10:10:count.txt\n", c.portable_manifest_text())
317
318     def test_write_large_rewrite(self):
319         keep = ArvadosFileWriterTestCase.MockKeep({})
320         api = ArvadosFileWriterTestCase.MockApi({"name":"test_write_large",
321                                                  "manifest_text": ". 37400a68af9abdd76ca5bf13e819e42a+32892003 a5de24f4417cfba9d5825eadc2f4ca49+67108000 32892000:3:count.txt 32892006:67107997:count.txt 0:32892000:count.txt\n"},
322                                                 {"uuid":"zzzzz-4zz18-mockcollection0",
323                                                  "manifest_text": ". 37400a68af9abdd76ca5bf13e819e42a+32892003 a5de24f4417cfba9d5825eadc2f4ca49+67108000 32892000:3:count.txt 32892006:67107997:count.txt 0:32892000:count.txt\n"})
324         with Collection('. ' + arvados.config.EMPTY_BLOCK_LOCATOR + ' 0:0:count.txt',
325                              api_client=api, keep_client=keep) as c:
326             writer = c.open("count.txt", "r+")
327             text = ''.join(["0123456789" for a in xrange(0, 100)])
328             for b in xrange(0, 100000):
329                 writer.write(text)
330             writer.seek(0, os.SEEK_SET)
331             writer.write("foo")
332             self.assertEqual(writer.size(), 100000000)
333
334             self.assertEqual(None, c.manifest_locator())
335             self.assertEqual(True, c.modified())
336             c.save_new("test_write_large")
337             self.assertEqual("zzzzz-4zz18-mockcollection0", c.manifest_locator())
338             self.assertEqual(False, c.modified())
339
340     def test_create(self):
341         keep = ArvadosFileWriterTestCase.MockKeep({})
342         api = ArvadosFileWriterTestCase.MockApi({"name":"test_create",
343                                                  "manifest_text":". 2e9ec317e197819358fbc43afca7d837+8 0:8:count.txt\n"},
344                                                 {"uuid":"zzzzz-4zz18-mockcollection0",
345                                                  "manifest_text":". 2e9ec317e197819358fbc43afca7d837+8 0:8:count.txt\n"})
346         with Collection(api_client=api, keep_client=keep) as c:
347             writer = c.open("count.txt", "w+")
348             self.assertEqual(writer.size(), 0)
349             writer.write("01234567")
350             self.assertEqual(writer.size(), 8)
351
352             self.assertEqual(None, c.manifest_locator())
353             self.assertEqual(True, c.modified())
354             self.assertEqual(None, keep.get("2e9ec317e197819358fbc43afca7d837+8"))
355             c.save_new("test_create")
356             self.assertEqual("zzzzz-4zz18-mockcollection0", c.manifest_locator())
357             self.assertEqual(False, c.modified())
358             self.assertEqual("01234567", keep.get("2e9ec317e197819358fbc43afca7d837+8"))
359
360
361     def test_create_subdir(self):
362         keep = ArvadosFileWriterTestCase.MockKeep({})
363         api = ArvadosFileWriterTestCase.MockApi({"name":"test_create",
364                                                  "manifest_text":"./foo/bar 2e9ec317e197819358fbc43afca7d837+8 0:8:count.txt\n"},
365                                                 {"uuid":"zzzzz-4zz18-mockcollection0",
366                                                  "manifest_text":"./foo/bar 2e9ec317e197819358fbc43afca7d837+8 0:8:count.txt\n"})
367         with Collection(api_client=api, keep_client=keep) as c:
368             self.assertIsNone(c.api_response())
369             writer = c.open("foo/bar/count.txt", "w+")
370             writer.write("01234567")
371             c.save_new("test_create")
372             self.assertEqual(c.api_response(), api.response)
373
374     def test_overwrite(self):
375         keep = ArvadosFileWriterTestCase.MockKeep({"781e5e245d69b566979b86e28d23f2c7+10": "0123456789"})
376         api = ArvadosFileWriterTestCase.MockApi({"name":"test_overwrite",
377                                                  "manifest_text":". 2e9ec317e197819358fbc43afca7d837+8 0:8:count.txt\n"},
378                                                 {"uuid":"zzzzz-4zz18-mockcollection0",
379                                                  "manifest_text":". 2e9ec317e197819358fbc43afca7d837+8 0:8:count.txt\n"})
380         with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n',
381                              api_client=api, keep_client=keep) as c:
382             writer = c.open("count.txt", "w+")
383             self.assertEqual(writer.size(), 0)
384             writer.write("01234567")
385             self.assertEqual(writer.size(), 8)
386
387             self.assertEqual(None, c.manifest_locator())
388             self.assertEqual(True, c.modified())
389             c.save_new("test_overwrite")
390             self.assertEqual("zzzzz-4zz18-mockcollection0", c.manifest_locator())
391             self.assertEqual(False, c.modified())
392
393     def test_file_not_found(self):
394         with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n') as c:
395             with self.assertRaises(IOError):
396                 writer = c.open("nocount.txt", "r")
397
398     def test_cannot_open_directory(self):
399         with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n') as c:
400             with self.assertRaises(IOError):
401                 writer = c.open(".", "r")
402
403     def test_create_multiple(self):
404         keep = ArvadosFileWriterTestCase.MockKeep({})
405         api = ArvadosFileWriterTestCase.MockApi({"name":"test_create_multiple",
406                                                  "manifest_text":". 2e9ec317e197819358fbc43afca7d837+8 e8dc4081b13434b45189a720b77b6818+8 0:8:count1.txt 8:8:count2.txt\n"},
407                                                 {"uuid":"zzzzz-4zz18-mockcollection0",
408                                                  "manifest_text":". 2e9ec317e197819358fbc43afca7d837+8 e8dc4081b13434b45189a720b77b6818+8 0:8:count1.txt 8:8:count2.txt\n"})
409         with Collection(api_client=api, keep_client=keep) as c:
410             w1 = c.open("count1.txt", "w")
411             w2 = c.open("count2.txt", "w")
412             w1.write("01234567")
413             w2.write("abcdefgh")
414             self.assertEqual(w1.size(), 8)
415             self.assertEqual(w2.size(), 8)
416
417             self.assertEqual(None, c.manifest_locator())
418             self.assertEqual(True, c.modified())
419             self.assertEqual(None, keep.get("2e9ec317e197819358fbc43afca7d837+8"))
420             c.save_new("test_create_multiple")
421             self.assertEqual("zzzzz-4zz18-mockcollection0", c.manifest_locator())
422             self.assertEqual(False, c.modified())
423             self.assertEqual("01234567", keep.get("2e9ec317e197819358fbc43afca7d837+8"))
424
425
426 class ArvadosFileReaderTestCase(StreamFileReaderTestCase):
427     class MockParent(object):
428         class MockBlockMgr(object):
429             def __init__(self, blocks, nocache):
430                 self.blocks = blocks
431                 self.nocache = nocache
432
433             def block_prefetch(self, loc):
434                 pass
435
436             def get_block_contents(self, loc, num_retries=0, cache_only=False):
437                 if self.nocache and cache_only:
438                     return None
439                 return self.blocks[loc]
440
441         def __init__(self, blocks, nocache):
442             self.blocks = blocks
443             self.nocache = nocache
444             self.lock = arvados.arvfile.NoopLock()
445
446         def root_collection(self):
447             return self
448
449         def _my_block_manager(self):
450             return ArvadosFileReaderTestCase.MockParent.MockBlockMgr(self.blocks, self.nocache)
451
452
453     def make_count_reader(self, nocache=False):
454         stream = []
455         n = 0
456         blocks = {}
457         for d in ['01234', '34567', '67890']:
458             loc = '{}+{}'.format(hashlib.md5(d).hexdigest(), len(d))
459             blocks[loc] = d
460             stream.append(Range(loc, n, len(d)))
461             n += len(d)
462         af = ArvadosFile(ArvadosFileReaderTestCase.MockParent(blocks, nocache), stream=stream, segments=[Range(1, 0, 3), Range(6, 3, 3), Range(11, 6, 3)])
463         return ArvadosFileReader(af, "count.txt")
464
465     def test_read_block_crossing_behavior(self):
466         # read() needs to return all the data requested if possible, even if it
467         # crosses uncached blocks: https://arvados.org/issues/5856
468         sfile = self.make_count_reader(nocache=True)
469         self.assertEqual('12345678', sfile.read(8))
470
471     def test_successive_reads(self):
472         # Override StreamFileReaderTestCase.test_successive_reads
473         sfile = self.make_count_reader(nocache=True)
474         self.assertEqual('1234', sfile.read(4))
475         self.assertEqual('5678', sfile.read(4))
476         self.assertEqual('9', sfile.read(4))
477         self.assertEqual('', sfile.read(4))
478
479     def test_tell_after_block_read(self):
480         # Override StreamFileReaderTestCase.test_tell_after_block_read
481         sfile = self.make_count_reader(nocache=True)
482         self.assertEqual('12345678', sfile.read(8))
483         self.assertEqual(8, sfile.tell())
484
485     def test_prefetch(self):
486         keep = ArvadosFileWriterTestCase.MockKeep({"2e9ec317e197819358fbc43afca7d837+8": "01234567", "e8dc4081b13434b45189a720b77b6818+8": "abcdefgh"})
487         with Collection(". 2e9ec317e197819358fbc43afca7d837+8 e8dc4081b13434b45189a720b77b6818+8 0:16:count.txt\n", keep_client=keep) as c:
488             r = c.open("count.txt", "r")
489             self.assertEqual("0123", r.read(4))
490         self.assertIn("2e9ec317e197819358fbc43afca7d837+8", keep.requests)
491         self.assertIn("e8dc4081b13434b45189a720b77b6818+8", keep.requests)
492
493     def test__eq__from_manifest(self):
494         with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt') as c1:
495             with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt') as c2:
496                 self.assertTrue(c1["count1.txt"] == c2["count1.txt"])
497                 self.assertFalse(c1["count1.txt"] != c2["count1.txt"])
498
499     def test__eq__from_writes(self):
500         with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt') as c1:
501             with Collection() as c2:
502                 with c2.open("count1.txt", "w") as f:
503                     f.write("0123456789")
504
505                 self.assertTrue(c1["count1.txt"] == c2["count1.txt"])
506                 self.assertFalse(c1["count1.txt"] != c2["count1.txt"])
507
508     def test__ne__(self):
509         with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt') as c1:
510             with Collection() as c2:
511                 with c2.open("count1.txt", "w") as f:
512                     f.write("1234567890")
513
514                 self.assertTrue(c1["count1.txt"] != c2["count1.txt"])
515                 self.assertFalse(c1["count1.txt"] == c2["count1.txt"])
516
517
518 class ArvadosFileReadTestCase(unittest.TestCase, StreamRetryTestMixin):
519     def reader_for(self, coll_name, **kwargs):
520         stream = []
521         segments = []
522         n = 0
523         for d in self.manifest_for(coll_name).split():
524             try:
525                 k = KeepLocator(d)
526                 segments.append(Range(n, n, k.size))
527                 stream.append(Range(d, n, k.size))
528                 n += k.size
529             except ValueError:
530                 pass
531
532         blockmanager = arvados.arvfile._BlockManager(self.keep_client())
533         blockmanager.prefetch_enabled = False
534         col = Collection(keep_client=self.keep_client(), block_manager=blockmanager)
535         af = ArvadosFile(col,
536                          stream=stream,
537                          segments=segments)
538         return ArvadosFileReader(af, "test", **kwargs)
539
540     def read_for_test(self, reader, byte_count, **kwargs):
541         return reader.read(byte_count, **kwargs)
542
543
544 class ArvadosFileReadFromTestCase(ArvadosFileReadTestCase):
545     def read_for_test(self, reader, byte_count, **kwargs):
546         return reader.readfrom(0, byte_count, **kwargs)
547
548
549 class ArvadosFileReadAllTestCase(ArvadosFileReadTestCase):
550     def read_for_test(self, reader, byte_count, **kwargs):
551         return ''.join(reader.readall(**kwargs))
552
553
554 class ArvadosFileReadAllDecompressedTestCase(ArvadosFileReadTestCase):
555     def read_for_test(self, reader, byte_count, **kwargs):
556         return ''.join(reader.readall_decompressed(**kwargs))
557
558
559 class ArvadosFileReadlinesTestCase(ArvadosFileReadTestCase):
560     def read_for_test(self, reader, byte_count, **kwargs):
561         return ''.join(reader.readlines(**kwargs))
562
563 class BlockManagerTest(unittest.TestCase):
564     def test_bufferblock_append(self):
565         keep = ArvadosFileWriterTestCase.MockKeep({})
566         blockmanager = arvados.arvfile._BlockManager(keep)
567         bufferblock = blockmanager.alloc_bufferblock()
568         bufferblock.append("foo")
569
570         self.assertEqual(bufferblock.size(), 3)
571         self.assertEqual(bufferblock.buffer_view[0:3], "foo")
572         self.assertEqual(bufferblock.locator(), "acbd18db4cc2f85cedef654fccc4a4d8+3")
573
574         bufferblock.append("bar")
575
576         self.assertEqual(bufferblock.size(), 6)
577         self.assertEqual(bufferblock.buffer_view[0:6], "foobar")
578         self.assertEqual(bufferblock.locator(), "3858f62230ac3c915f300c664312c63f+6")
579
580         bufferblock.set_state(arvados.arvfile._BufferBlock.PENDING)
581         with self.assertRaises(arvados.errors.AssertionError):
582             bufferblock.append("bar")
583
584     def test_bufferblock_dup(self):
585         keep = ArvadosFileWriterTestCase.MockKeep({})
586         blockmanager = arvados.arvfile._BlockManager(keep)
587         bufferblock = blockmanager.alloc_bufferblock()
588         bufferblock.append("foo")
589
590         self.assertEqual(bufferblock.size(), 3)
591         self.assertEqual(bufferblock.buffer_view[0:3], "foo")
592         self.assertEqual(bufferblock.locator(), "acbd18db4cc2f85cedef654fccc4a4d8+3")
593         bufferblock.set_state(arvados.arvfile._BufferBlock.PENDING)
594
595         bufferblock2 = blockmanager.dup_block(bufferblock, None)
596         self.assertNotEqual(bufferblock.blockid, bufferblock2.blockid)
597
598         bufferblock2.append("bar")
599
600         self.assertEqual(bufferblock2.size(), 6)
601         self.assertEqual(bufferblock2.buffer_view[0:6], "foobar")
602         self.assertEqual(bufferblock2.locator(), "3858f62230ac3c915f300c664312c63f+6")
603
604         self.assertEqual(bufferblock.size(), 3)
605         self.assertEqual(bufferblock.buffer_view[0:3], "foo")
606         self.assertEqual(bufferblock.locator(), "acbd18db4cc2f85cedef654fccc4a4d8+3")
607
608     def test_bufferblock_get(self):
609         keep = ArvadosFileWriterTestCase.MockKeep({"781e5e245d69b566979b86e28d23f2c7+10": "0123456789"})
610         blockmanager = arvados.arvfile._BlockManager(keep)
611         bufferblock = blockmanager.alloc_bufferblock()
612         bufferblock.append("foo")
613
614         self.assertEqual(blockmanager.get_block_contents("781e5e245d69b566979b86e28d23f2c7+10", 1), "0123456789")
615         self.assertEqual(blockmanager.get_block_contents(bufferblock.blockid, 1), "foo")
616
617     def test_bufferblock_commit(self):
618         mockkeep = mock.MagicMock()
619         blockmanager = arvados.arvfile._BlockManager(mockkeep)
620         bufferblock = blockmanager.alloc_bufferblock()
621         bufferblock.owner = mock.MagicMock()
622         bufferblock.owner.flush.side_effect = lambda: blockmanager.commit_bufferblock(bufferblock)
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
631     def test_bufferblock_commit_with_error(self):
632         mockkeep = mock.MagicMock()
633         mockkeep.put.side_effect = arvados.errors.KeepWriteError("fail")
634         blockmanager = arvados.arvfile._BlockManager(mockkeep)
635         bufferblock = blockmanager.alloc_bufferblock()
636         bufferblock.owner = mock.MagicMock()
637         bufferblock.owner.flush.side_effect = lambda: blockmanager.commit_bufferblock(bufferblock)
638         bufferblock.append("foo")
639         with self.assertRaises(arvados.errors.KeepWriteError) as err:
640             blockmanager.commit_all()
641         self.assertTrue(bufferblock.owner.flush.called)
642         self.assertEqual(str(err.exception), "Error writing some blocks: block acbd18db4cc2f85cedef654fccc4a4d8+3 raised KeepWriteError (fail)")
643         self.assertEqual(bufferblock.state(), arvados.arvfile._BufferBlock.PENDING)