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