11308: Merge branch 'master' into 11308-python3
[arvados.git] / sdk / python / tests / test_arvfile.py
1 from __future__ import absolute_import
2 from builtins import hex
3 from builtins import str
4 from builtins import range
5 from builtins import object
6 import bz2
7 import datetime
8 import gzip
9 import io
10 import mock
11 import os
12 import unittest
13 import time
14
15 import arvados
16 from arvados._ranges import Range
17 from arvados.keep import KeepLocator
18 from arvados.collection import Collection, CollectionReader
19 from arvados.arvfile import ArvadosFile, ArvadosFileReader
20
21 from . import arvados_testutil as tutil
22 from .test_stream import StreamFileReaderTestCase, StreamRetryTestMixin
23
24 class ArvadosFileWriterTestCase(unittest.TestCase):
25     class MockKeep(object):
26         def __init__(self, blocks):
27             self.blocks = blocks
28             self.requests = []
29         def get(self, locator, num_retries=0):
30             self.requests.append(locator)
31             return self.blocks.get(locator)
32         def get_from_cache(self, locator):
33             self.requests.append(locator)
34             return self.blocks.get(locator)
35         def put(self, data, num_retries=None, copies=None):
36             pdh = tutil.str_keep_locator(data)
37             self.blocks[pdh] = bytes(data)
38             return pdh
39
40     class MockApi(object):
41         def __init__(self, b, r):
42             self.body = b
43             self.response = r
44             self._schema = ArvadosFileWriterTestCase.MockApi.MockSchema()
45             self._rootDesc = {}
46         class MockSchema(object):
47             def __init__(self):
48                 self.schemas = {'Collection': {'properties': {'replication_desired': {'type':'integer'}}}}
49         class MockCollections(object):
50             def __init__(self, b, r):
51                 self.body = b
52                 self.response = r
53             class Execute(object):
54                 def __init__(self, r):
55                     self.response = r
56                 def execute(self, num_retries=None):
57                     return self.response
58             def create(self, ensure_unique_name=False, body=None):
59                 if body != self.body:
60                     raise Exception("Body %s does not match expectation %s" % (body, self.body))
61                 return ArvadosFileWriterTestCase.MockApi.MockCollections.Execute(self.response)
62             def update(self, uuid=None, body=None):
63                 return ArvadosFileWriterTestCase.MockApi.MockCollections.Execute(self.response)
64         def collections(self):
65             return ArvadosFileWriterTestCase.MockApi.MockCollections(self.body, self.response)
66
67
68     def test_truncate(self):
69         keep = ArvadosFileWriterTestCase.MockKeep({
70             "781e5e245d69b566979b86e28d23f2c7+10": b"0123456789",
71         })
72         api = ArvadosFileWriterTestCase.MockApi({
73             "name": "test_truncate",
74             "manifest_text": ". 781e5e245d69b566979b86e28d23f2c7+10 0:8:count.txt\n",
75             "replication_desired": None,
76         }, {
77             "uuid": "zzzzz-4zz18-mockcollection0",
78             "manifest_text": ". 781e5e245d69b566979b86e28d23f2c7+10 0:8:count.txt\n",
79             "portable_data_hash":"7fcd0eaac3aad4c31a6a0e756475da92+52",
80         })
81         with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n',
82                              api_client=api, keep_client=keep) as c:
83             writer = c.open("count.txt", "rb+")
84             self.assertEqual(writer.size(), 10)
85             self.assertEqual(b"0123456789", writer.read(12))
86
87             writer.truncate(8)
88
89             # Make sure reading off the end doesn't break
90             self.assertEqual(b"", writer.read(12))
91
92             self.assertEqual(writer.size(), 8)
93             writer.seek(0, os.SEEK_SET)
94             self.assertEqual(b"01234567", writer.read(12))
95
96             self.assertIsNone(c.manifest_locator())
97             self.assertTrue(c.modified())
98             c.save_new("test_truncate")
99             self.assertEqual("zzzzz-4zz18-mockcollection0", c.manifest_locator())
100             self.assertFalse(c.modified())
101
102
103     def test_truncate2(self):
104         keep = ArvadosFileWriterTestCase.MockKeep({"781e5e245d69b566979b86e28d23f2c7+10": "0123456789"})
105         api = ArvadosFileWriterTestCase.MockApi({"name":"test_truncate2",
106                                                  "manifest_text":". 781e5e245d69b566979b86e28d23f2c7+10 7f614da9329cd3aebf59b91aadc30bf0+67108864 0:12:count.txt\n",
107                                                  "replication_desired":None},
108                                                 {"uuid":"zzzzz-4zz18-mockcollection0",
109                                                  "manifest_text":". 781e5e245d69b566979b86e28d23f2c7+10 7f614da9329cd3aebf59b91aadc30bf0+67108864 0:12:count.txt\n",
110                                                  "portable_data_hash":"272da898abdf86ddc71994835e3155f8+95"})
111         with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n',
112                              api_client=api, keep_client=keep) as c:
113             writer = c.open("count.txt", "r+")
114             self.assertEqual(writer.size(), 10)
115             self.assertEqual("0123456789", writer.read(12))
116
117             # extend file size
118             writer.truncate(12)
119
120             self.assertEqual(writer.size(), 12)
121             writer.seek(0, os.SEEK_SET)
122             self.assertEqual(b"0123456789\x00\x00", writer.read(12))
123
124             self.assertIsNone(c.manifest_locator())
125             self.assertTrue(c.modified())
126             c.save_new("test_truncate2")
127             self.assertEqual("zzzzz-4zz18-mockcollection0", c.manifest_locator())
128             self.assertFalse(c.modified())
129
130     def test_truncate3(self):
131         keep = ArvadosFileWriterTestCase.MockKeep({"781e5e245d69b566979b86e28d23f2c7+10": "0123456789",
132                                                    "a925576942e94b2ef57a066101b48876+10": "abcdefghij"})
133         api = ArvadosFileWriterTestCase.MockApi({"name":"test_truncate",
134                                                  "manifest_text":". 781e5e245d69b566979b86e28d23f2c7+10 0:8:count.txt\n",
135                                                  "replication_desired":None},
136                                                 {"uuid":"zzzzz-4zz18-mockcollection0",
137                                                  "manifest_text":". 781e5e245d69b566979b86e28d23f2c7+10 0:8:count.txt\n",
138                                                  "portable_data_hash":"7fcd0eaac3aad4c31a6a0e756475da92+52"})
139         with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 a925576942e94b2ef57a066101b48876+10 0:20:count.txt\n',
140                              api_client=api, keep_client=keep) as c:
141             writer = c.open("count.txt", "r+")
142             self.assertEqual(writer.size(), 20)
143             self.assertEqual("0123456789ab", writer.read(12))
144             self.assertEqual(12, writer.tell())
145
146             writer.truncate(8)
147
148             # Make sure reading off the end doesn't break
149             self.assertEqual(12, writer.tell())
150             self.assertEqual("", writer.read(12))
151
152             self.assertEqual(writer.size(), 8)
153             self.assertEqual(2, writer.seek(-10, os.SEEK_CUR))
154             self.assertEqual("234567", writer.read(12))
155
156             self.assertIsNone(c.manifest_locator())
157             self.assertTrue(c.modified())
158             c.save_new("test_truncate")
159             self.assertEqual("zzzzz-4zz18-mockcollection0", c.manifest_locator())
160             self.assertFalse(c.modified())
161
162
163
164     def test_write_to_end(self):
165         keep = ArvadosFileWriterTestCase.MockKeep({
166             "781e5e245d69b566979b86e28d23f2c7+10": b"0123456789",
167         })
168         api = ArvadosFileWriterTestCase.MockApi({
169             "name": "test_append",
170             "manifest_text": ". 781e5e245d69b566979b86e28d23f2c7+10 acbd18db4cc2f85cedef654fccc4a4d8+3 0:13:count.txt\n",
171             "replication_desired": None,
172         }, {
173             "uuid": "zzzzz-4zz18-mockcollection0",
174             "manifest_text": ". 781e5e245d69b566979b86e28d23f2c7+10 acbd18db4cc2f85cedef654fccc4a4d8+3 0:13:count.txt\n",
175             "portable_data_hash": "c5c3af76565c8efb6a806546bcf073f3+88",
176         })
177         with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n',
178                              api_client=api, keep_client=keep) as c:
179             writer = c.open("count.txt", "rb+")
180             self.assertEqual(writer.size(), 10)
181
182             self.assertEqual(5, writer.seek(5, os.SEEK_SET))
183             self.assertEqual(b"56789", writer.read(8))
184
185             writer.seek(10, os.SEEK_SET)
186             writer.write("foo")
187             self.assertEqual(writer.size(), 13)
188
189             writer.seek(5, os.SEEK_SET)
190             self.assertEqual(b"56789foo", writer.read(8))
191
192             self.assertIsNone(c.manifest_locator())
193             self.assertTrue(c.modified())
194             self.assertIsNone(keep.get("acbd18db4cc2f85cedef654fccc4a4d8+3"))
195
196             c.save_new("test_append")
197             self.assertEqual("zzzzz-4zz18-mockcollection0", c.manifest_locator())
198             self.assertFalse(c.modified())
199             self.assertEqual(b"foo", keep.get("acbd18db4cc2f85cedef654fccc4a4d8+3"))
200
201
202     def test_append(self):
203         keep = ArvadosFileWriterTestCase.MockKeep({
204             "781e5e245d69b566979b86e28d23f2c7+10": b"0123456789",
205         })
206         c = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n', keep_client=keep)
207         writer = c.open("count.txt", "ab+")
208         self.assertEqual(writer.read(20), b"0123456789")
209         writer.seek(0, os.SEEK_SET)
210
211         writer.write("hello")
212         self.assertEqual(writer.read(20), b"0123456789hello")
213         writer.seek(0, os.SEEK_SET)
214
215         writer.write("world")
216         self.assertEqual(writer.read(20), b"0123456789helloworld")
217
218         self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 fc5e038d38a57032085441e7fe7010b0+10 0:20:count.txt\n", c.portable_manifest_text())
219
220     def test_write_at_beginning(self):
221         keep = ArvadosFileWriterTestCase.MockKeep({
222             "781e5e245d69b566979b86e28d23f2c7+10": b"0123456789",
223         })
224         with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n',
225                              keep_client=keep) as c:
226             writer = c.open("count.txt", "rb+")
227             self.assertEqual(b"0123456789", writer.readfrom(0, 13))
228             writer.seek(0, os.SEEK_SET)
229             writer.write("foo")
230             self.assertEqual(writer.size(), 10)
231             self.assertEqual(b"foo3456789", writer.readfrom(0, 13))
232             self.assertEqual(". acbd18db4cc2f85cedef654fccc4a4d8+3 781e5e245d69b566979b86e28d23f2c7+10 0:3:count.txt 6:7:count.txt\n", c.portable_manifest_text())
233
234     def test_write_empty(self):
235         keep = ArvadosFileWriterTestCase.MockKeep({})
236         with Collection(keep_client=keep) as c:
237             writer = c.open("count.txt", "wb")
238             self.assertEqual(writer.size(), 0)
239             self.assertEqual(". d41d8cd98f00b204e9800998ecf8427e+0 0:0:count.txt\n", c.portable_manifest_text())
240
241     def test_save_manifest_text(self):
242         keep = ArvadosFileWriterTestCase.MockKeep({})
243         with Collection(keep_client=keep) as c:
244             writer = c.open("count.txt", "wb")
245             writer.write(b"0123456789")
246             self.assertEqual('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n', c.portable_manifest_text())
247             self.assertNotIn('781e5e245d69b566979b86e28d23f2c7+10', keep.blocks)
248
249             self.assertEqual('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n', c.save_new(create_collection_record=False))
250             self.assertIn('781e5e245d69b566979b86e28d23f2c7+10', keep.blocks)
251
252     def test_get_manifest_text_commits(self):
253          keep = ArvadosFileWriterTestCase.MockKeep({})
254          with Collection(keep_client=keep) as c:
255              writer = c.open("count.txt", "wb")
256              writer.write("0123456789")
257              self.assertEqual('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n', c.portable_manifest_text())
258              self.assertNotIn('781e5e245d69b566979b86e28d23f2c7+10', keep.blocks)
259              self.assertEqual('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n', c.manifest_text())
260              self.assertIn('781e5e245d69b566979b86e28d23f2c7+10', keep.blocks)
261
262
263     def test_write_in_middle(self):
264         keep = ArvadosFileWriterTestCase.MockKeep({"781e5e245d69b566979b86e28d23f2c7+10": b"0123456789"})
265         with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n',
266                              keep_client=keep) as c:
267             writer = c.open("count.txt", "rb+")
268             self.assertEqual(b"0123456789", writer.readfrom(0, 13))
269             writer.seek(3, os.SEEK_SET)
270             writer.write("foo")
271             self.assertEqual(writer.size(), 10)
272             self.assertEqual(b"012foo6789", writer.readfrom(0, 13))
273             self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 acbd18db4cc2f85cedef654fccc4a4d8+3 0:3:count.txt 10:3:count.txt 6:4:count.txt\n", c.portable_manifest_text())
274
275     def test_write_at_end(self):
276         keep = ArvadosFileWriterTestCase.MockKeep({"781e5e245d69b566979b86e28d23f2c7+10": b"0123456789"})
277         with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n',
278                              keep_client=keep) as c:
279             writer = c.open("count.txt", "rb+")
280             self.assertEqual(b"0123456789", writer.readfrom(0, 13))
281             writer.seek(7, os.SEEK_SET)
282             writer.write("foo")
283             self.assertEqual(writer.size(), 10)
284             self.assertEqual(b"0123456foo", writer.readfrom(0, 13))
285             self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 acbd18db4cc2f85cedef654fccc4a4d8+3 0:7:count.txt 10:3:count.txt\n", c.portable_manifest_text())
286
287     def test_write_across_segment_boundary(self):
288         keep = ArvadosFileWriterTestCase.MockKeep({"781e5e245d69b566979b86e28d23f2c7+10": b"0123456789"})
289         with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt 0:10:count.txt\n',
290                              keep_client=keep) as c:
291             writer = c.open("count.txt", "rb+")
292             self.assertEqual(b"012345678901234", writer.readfrom(0, 15))
293             writer.seek(7, os.SEEK_SET)
294             writer.write("foobar")
295             self.assertEqual(writer.size(), 20)
296             self.assertEqual(b"0123456foobar34", writer.readfrom(0, 15))
297             self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 3858f62230ac3c915f300c664312c63f+6 0:7:count.txt 10:6:count.txt 3:7:count.txt\n", c.portable_manifest_text())
298
299     def test_write_across_several_segments(self):
300         keep = ArvadosFileWriterTestCase.MockKeep({"781e5e245d69b566979b86e28d23f2c7+10": b"0123456789"})
301         with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:4:count.txt 0:4:count.txt 0:4:count.txt',
302                              keep_client=keep) as c:
303             writer = c.open("count.txt", "rb+")
304             self.assertEqual(b"012301230123", writer.readfrom(0, 15))
305             writer.seek(2, os.SEEK_SET)
306             writer.write("abcdefg")
307             self.assertEqual(writer.size(), 12)
308             self.assertEqual(b"01abcdefg123", writer.readfrom(0, 15))
309             self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 7ac66c0f148de9519b8bd264312c4d64+7 0:2:count.txt 10:7:count.txt 1:3:count.txt\n", c.portable_manifest_text())
310
311     def test_write_large(self):
312         keep = ArvadosFileWriterTestCase.MockKeep({})
313         api = ArvadosFileWriterTestCase.MockApi({"name":"test_write_large",
314                                                  "manifest_text": ". a5de24f4417cfba9d5825eadc2f4ca49+67108000 598cc1a4ccaef8ab6e4724d87e675d78+32892000 0:100000000:count.txt\n",
315                                                  "replication_desired":None},
316                                                 {"uuid":"zzzzz-4zz18-mockcollection0",
317                                                  "manifest_text": ". a5de24f4417cfba9d5825eadc2f4ca49+67108000 598cc1a4ccaef8ab6e4724d87e675d78+32892000 0:100000000:count.txt\n",
318                                                  "portable_data_hash":"9132ca8e3f671c76103a38f5bc24328c+108"})
319         with Collection('. ' + arvados.config.EMPTY_BLOCK_LOCATOR + ' 0:0:count.txt',
320                              api_client=api, keep_client=keep) as c:
321             writer = c.open("count.txt", "rb+")
322             text = "0123456789" * 100
323             for b in range(0, 100000):
324                 writer.write(text)
325             self.assertEqual(writer.size(), 100000000)
326
327             self.assertIsNone(c.manifest_locator())
328             self.assertTrue(c.modified())
329             c.save_new("test_write_large")
330             self.assertEqual("zzzzz-4zz18-mockcollection0", c.manifest_locator())
331             self.assertFalse(c.modified())
332
333
334     def test_large_write(self):
335         keep = ArvadosFileWriterTestCase.MockKeep({})
336         api = ArvadosFileWriterTestCase.MockApi({}, {})
337         with Collection('. ' + arvados.config.EMPTY_BLOCK_LOCATOR + ' 0:0:count.txt',
338                              api_client=api, keep_client=keep) as c:
339             writer = c.open("count.txt", "rb+")
340             self.assertEqual(writer.size(), 0)
341
342             text = "0123456789"
343             writer.write(text)
344             text = "0123456789" * 9999999
345             writer.write(text)
346             self.assertEqual(writer.size(), 100000000)
347
348             self.assertEqual(c.manifest_text(), ". 781e5e245d69b566979b86e28d23f2c7+10 48dd23ea1645fd47d789804d71b5bb8e+67108864 77c57dc6ac5a10bb2205caaa73187994+32891126 0:100000000:count.txt\n")
349
350     def test_sparse_write(self):
351         keep = ArvadosFileWriterTestCase.MockKeep({})
352         api = ArvadosFileWriterTestCase.MockApi({}, {})
353         with Collection('. ' + arvados.config.EMPTY_BLOCK_LOCATOR + ' 0:0:count.txt',
354                              api_client=api, keep_client=keep) as c:
355             writer = c.open("count.txt", "r+")
356             self.assertEqual(writer.size(), 0)
357
358             text = "0123456789"
359             writer.seek(2)
360             writer.write(text)
361             self.assertEqual(writer.size(), 12)
362             writer.seek(0, os.SEEK_SET)
363             self.assertEqual(writer.read(), b"\x00\x00"+text)
364
365             self.assertEqual(c.manifest_text(), ". 7f614da9329cd3aebf59b91aadc30bf0+67108864 781e5e245d69b566979b86e28d23f2c7+10 0:2:count.txt 67108864:10:count.txt\n")
366
367
368     def test_sparse_write2(self):
369         keep = ArvadosFileWriterTestCase.MockKeep({})
370         api = ArvadosFileWriterTestCase.MockApi({}, {})
371         with Collection('. ' + arvados.config.EMPTY_BLOCK_LOCATOR + ' 0:0:count.txt',
372                              api_client=api, keep_client=keep) as c:
373             writer = c.open("count.txt", "r+")
374             self.assertEqual(writer.size(), 0)
375
376             text = "0123456789"
377             writer.seek((arvados.config.KEEP_BLOCK_SIZE*2) + 2)
378             writer.write(text)
379             self.assertEqual(writer.size(), (arvados.config.KEEP_BLOCK_SIZE*2) + 12)
380             writer.seek(0, os.SEEK_SET)
381
382             self.assertEqual(c.manifest_text(), ". 7f614da9329cd3aebf59b91aadc30bf0+67108864 781e5e245d69b566979b86e28d23f2c7+10 0:67108864:count.txt 0:67108864:count.txt 0:2:count.txt 67108864:10:count.txt\n")
383
384
385     def test_sparse_write3(self):
386         keep = ArvadosFileWriterTestCase.MockKeep({})
387         api = ArvadosFileWriterTestCase.MockApi({}, {})
388         for r in [[0, 1, 2, 3, 4], [4, 3, 2, 1, 0], [3, 2, 0, 4, 1]]:
389             with Collection() as c:
390                 writer = c.open("count.txt", "r+")
391                 self.assertEqual(writer.size(), 0)
392
393                 for i in r:
394                     w = ("%s" % i) * 10
395                     writer.seek(i*10)
396                     writer.write(w)
397                 writer.seek(0)
398                 self.assertEqual(writer.read(), "00000000001111111111222222222233333333334444444444")
399
400     def test_sparse_write4(self):
401         keep = ArvadosFileWriterTestCase.MockKeep({})
402         api = ArvadosFileWriterTestCase.MockApi({}, {})
403         for r in [[0, 1, 2, 4], [4, 2, 1, 0], [2, 0, 4, 1]]:
404             with Collection() as c:
405                 writer = c.open("count.txt", "r+")
406                 self.assertEqual(writer.size(), 0)
407
408                 for i in r:
409                     w = ("%s" % i) * 10
410                     writer.seek(i*10)
411                     writer.write(w)
412                 writer.seek(0)
413                 self.assertEqual(writer.read(), b"000000000011111111112222222222\x00\x00\x00\x00\x00\x00\x00\x00\x00\x004444444444")
414
415
416     def test_rewrite_on_empty_file(self):
417         keep = ArvadosFileWriterTestCase.MockKeep({})
418         with Collection('. ' + arvados.config.EMPTY_BLOCK_LOCATOR + ' 0:0:count.txt',
419                              keep_client=keep) as c:
420             writer = c.open("count.txt", "rb+")
421             for b in range(0, 10):
422                 writer.seek(0, os.SEEK_SET)
423                 writer.write("0123456789")
424
425             self.assertEqual(writer.size(), 10)
426             self.assertEqual(b"0123456789", writer.readfrom(0, 20))
427             self.assertEqual(". 7a08b07e84641703e5f2c836aa59a170+100 90:10:count.txt\n", c.portable_manifest_text())
428             writer.flush()
429             self.assertEqual(writer.size(), 10)
430             self.assertEqual(b"0123456789", writer.readfrom(0, 20))
431             self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n", c.portable_manifest_text())
432
433     def test_rewrite_append_existing_file(self):
434         keep = ArvadosFileWriterTestCase.MockKeep({
435             "781e5e245d69b566979b86e28d23f2c7+10": b"0123456789",
436         })
437         with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt',
438                              keep_client=keep) as c:
439             writer = c.open("count.txt", "rb+")
440             for b in range(0, 10):
441                 writer.seek(10, os.SEEK_SET)
442                 writer.write("abcdefghij")
443
444             self.assertEqual(writer.size(), 20)
445             self.assertEqual(b"0123456789abcdefghij", writer.readfrom(0, 20))
446             self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 ae5f43bab79cf0be33f025fa97ae7398+100 0:10:count.txt 100:10:count.txt\n", c.portable_manifest_text())
447
448             writer.arvadosfile.flush()
449             self.assertEqual(writer.size(), 20)
450             self.assertEqual(b"0123456789abcdefghij", writer.readfrom(0, 20))
451             self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 a925576942e94b2ef57a066101b48876+10 0:20:count.txt\n", c.portable_manifest_text())
452
453     def test_rewrite_over_existing_file(self):
454         keep = ArvadosFileWriterTestCase.MockKeep({
455             "781e5e245d69b566979b86e28d23f2c7+10": b"0123456789",
456         })
457         with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt',
458                              keep_client=keep) as c:
459             writer = c.open("count.txt", "rb+")
460             for b in range(0, 10):
461                 writer.seek(5, os.SEEK_SET)
462                 writer.write("abcdefghij")
463
464             self.assertEqual(writer.size(), 15)
465             self.assertEqual(b"01234abcdefghij", writer.readfrom(0, 20))
466             self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 ae5f43bab79cf0be33f025fa97ae7398+100 0:5:count.txt 100:10:count.txt\n", c.portable_manifest_text())
467
468             writer.arvadosfile.flush()
469
470             self.assertEqual(writer.size(), 15)
471             self.assertEqual(b"01234abcdefghij", writer.readfrom(0, 20))
472             self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 a925576942e94b2ef57a066101b48876+10 0:5:count.txt 10:10:count.txt\n", c.portable_manifest_text())
473
474     def test_write_large_rewrite(self):
475         keep = ArvadosFileWriterTestCase.MockKeep({})
476         api = ArvadosFileWriterTestCase.MockApi({"name":"test_write_large",
477                                                  "manifest_text": ". 3dc0d4bc21f48060bedcb2c91af4f906+32892003 a5de24f4417cfba9d5825eadc2f4ca49+67108000 0:3:count.txt 32892006:67107997:count.txt 3:32892000:count.txt\n",
478                                                  "replication_desired":None},
479                                                 {"uuid":"zzzzz-4zz18-mockcollection0",
480                                                  "manifest_text": ". 3dc0d4bc21f48060bedcb2c91af4f906+32892003 a5de24f4417cfba9d5825eadc2f4ca49+67108000 0:3:count.txt 32892006:67107997:count.txt 3:32892000:count.txt\n",
481                                                  "portable_data_hash":"217665c6b713e1b78dfba7ebd42344db+156"})
482         with Collection('. ' + arvados.config.EMPTY_BLOCK_LOCATOR + ' 0:0:count.txt',
483                              api_client=api, keep_client=keep) as c:
484             writer = c.open("count.txt", "rb+")
485             text = b''.join([b"0123456789" for a in range(0, 100)])
486             for b in range(0, 100000):
487                 writer.write(text)
488             writer.seek(0, os.SEEK_SET)
489             writer.write("foo")
490             self.assertEqual(writer.size(), 100000000)
491
492             self.assertIsNone(c.manifest_locator())
493             self.assertTrue(c.modified())
494             c.save_new("test_write_large")
495             self.assertEqual("zzzzz-4zz18-mockcollection0", c.manifest_locator())
496             self.assertFalse(c.modified())
497
498     def test_create(self):
499         keep = ArvadosFileWriterTestCase.MockKeep({})
500         api = ArvadosFileWriterTestCase.MockApi({
501             "name":"test_create",
502             "manifest_text":". 2e9ec317e197819358fbc43afca7d837+8 0:8:count.txt\n",
503             "replication_desired":None,
504         }, {
505             "uuid":"zzzzz-4zz18-mockcollection0",
506             "manifest_text":". 2e9ec317e197819358fbc43afca7d837+8 0:8:count.txt\n",
507             "portable_data_hash":"7a461a8c58601798f690f8b368ac4423+51",
508         })
509         with Collection(api_client=api, keep_client=keep) as c:
510             writer = c.open("count.txt", "wb+")
511             self.assertEqual(writer.size(), 0)
512             writer.write("01234567")
513             self.assertEqual(writer.size(), 8)
514
515             self.assertIsNone(c.manifest_locator())
516             self.assertTrue(c.modified())
517             self.assertIsNone(keep.get("2e9ec317e197819358fbc43afca7d837+8"))
518             c.save_new("test_create")
519             self.assertEqual("zzzzz-4zz18-mockcollection0", c.manifest_locator())
520             self.assertFalse(c.modified())
521             self.assertEqual(b"01234567", keep.get("2e9ec317e197819358fbc43afca7d837+8"))
522
523
524     def test_create_subdir(self):
525         keep = ArvadosFileWriterTestCase.MockKeep({})
526         api = ArvadosFileWriterTestCase.MockApi({"name":"test_create",
527                                                  "manifest_text":"./foo/bar 2e9ec317e197819358fbc43afca7d837+8 0:8:count.txt\n",
528                                                  "replication_desired":None},
529                                                 {"uuid":"zzzzz-4zz18-mockcollection0",
530                                                  "manifest_text":"./foo/bar 2e9ec317e197819358fbc43afca7d837+8 0:8:count.txt\n",
531                                                  "portable_data_hash":"1b02aaa62528d28a5be41651cbb9d7c7+59"})
532         with Collection(api_client=api, keep_client=keep) as c:
533             self.assertIsNone(c.api_response())
534             writer = c.open("foo/bar/count.txt", "wb+")
535             writer.write("01234567")
536             self.assertFalse(c.committed())
537             c.save_new("test_create")
538             self.assertTrue(c.committed())
539             self.assertEqual(c.api_response(), api.response)
540
541     def test_overwrite(self):
542         keep = ArvadosFileWriterTestCase.MockKeep({"781e5e245d69b566979b86e28d23f2c7+10": "0123456789"})
543         api = ArvadosFileWriterTestCase.MockApi({"name":"test_overwrite",
544                                                  "manifest_text":". 2e9ec317e197819358fbc43afca7d837+8 0:8:count.txt\n",
545                                                  "replication_desired":None},
546                                                 {"uuid":"zzzzz-4zz18-mockcollection0",
547                                                  "manifest_text":". 2e9ec317e197819358fbc43afca7d837+8 0:8:count.txt\n",
548                                                  "portable_data_hash":"7a461a8c58601798f690f8b368ac4423+51"})
549         with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n',
550                              api_client=api, keep_client=keep) as c:
551             writer = c.open("count.txt", "wb+")
552             self.assertEqual(writer.size(), 0)
553             writer.write("01234567")
554             self.assertEqual(writer.size(), 8)
555
556             self.assertIsNone(c.manifest_locator())
557             self.assertTrue(c.modified())
558             c.save_new("test_overwrite")
559             self.assertEqual("zzzzz-4zz18-mockcollection0", c.manifest_locator())
560             self.assertFalse(c.modified())
561
562     def test_file_not_found(self):
563         with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n') as c:
564             with self.assertRaises(IOError):
565                 writer = c.open("nocount.txt", "rb")
566
567     def test_cannot_open_directory(self):
568         with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n') as c:
569             with self.assertRaises(IOError):
570                 writer = c.open(".", "rb")
571
572     def test_create_multiple(self):
573         keep = ArvadosFileWriterTestCase.MockKeep({})
574         api = ArvadosFileWriterTestCase.MockApi({"name":"test_create_multiple",
575                                                  "manifest_text":". 2e9ec317e197819358fbc43afca7d837+8 e8dc4081b13434b45189a720b77b6818+8 0:8:count1.txt 8:8:count2.txt\n",
576                                                  "replication_desired":None},
577                                                 {"uuid":"zzzzz-4zz18-mockcollection0",
578                                                  "manifest_text":". 2e9ec317e197819358fbc43afca7d837+8 e8dc4081b13434b45189a720b77b6818+8 0:8:count1.txt 8:8:count2.txt\n",
579                                                  "portable_data_hash":"71e7bb6c00d31fc2b4364199fd97be08+102"})
580         with Collection(api_client=api, keep_client=keep) as c:
581             w1 = c.open("count1.txt", "wb")
582             w2 = c.open("count2.txt", "wb")
583             w1.write("01234567")
584             w2.write("abcdefgh")
585             self.assertEqual(w1.size(), 8)
586             self.assertEqual(w2.size(), 8)
587
588             self.assertIsNone(c.manifest_locator())
589             self.assertTrue(c.modified())
590             self.assertIsNone(keep.get("2e9ec317e197819358fbc43afca7d837+8"))
591             c.save_new("test_create_multiple")
592             self.assertEqual("zzzzz-4zz18-mockcollection0", c.manifest_locator())
593             self.assertFalse(c.modified())
594             self.assertEqual(b"01234567", keep.get("2e9ec317e197819358fbc43afca7d837+8"))
595
596
597 class ArvadosFileReaderTestCase(StreamFileReaderTestCase):
598     class MockParent(object):
599         class MockBlockMgr(object):
600             def __init__(self, blocks, nocache):
601                 self.blocks = blocks
602                 self.nocache = nocache
603
604             def block_prefetch(self, loc):
605                 pass
606
607             def get_block_contents(self, loc, num_retries=0, cache_only=False):
608                 if self.nocache and cache_only:
609                     return None
610                 return self.blocks[loc]
611
612         def __init__(self, blocks, nocache):
613             self.blocks = blocks
614             self.nocache = nocache
615             self.lock = arvados.arvfile.NoopLock()
616
617         def root_collection(self):
618             return self
619
620         def _my_block_manager(self):
621             return ArvadosFileReaderTestCase.MockParent.MockBlockMgr(self.blocks, self.nocache)
622
623
624     def make_count_reader(self, nocache=False):
625         stream = []
626         n = 0
627         blocks = {}
628         for d in [b'01234', b'34567', b'67890']:
629             loc = tutil.str_keep_locator(d)
630             blocks[loc] = d
631             stream.append(Range(loc, n, len(d)))
632             n += len(d)
633         af = ArvadosFile(ArvadosFileReaderTestCase.MockParent(blocks, nocache), "count.txt", stream=stream, segments=[Range(1, 0, 3), Range(6, 3, 3), Range(11, 6, 3)])
634         return ArvadosFileReader(af, mode="rb")
635
636     def test_read_block_crossing_behavior(self):
637         # read() needs to return all the data requested if possible, even if it
638         # crosses uncached blocks: https://arvados.org/issues/5856
639         sfile = self.make_count_reader(nocache=True)
640         self.assertEqual(b'12345678', sfile.read(8))
641
642     def test_successive_reads(self):
643         # Override StreamFileReaderTestCase.test_successive_reads
644         sfile = self.make_count_reader(nocache=True)
645         self.assertEqual(b'1234', sfile.read(4))
646         self.assertEqual(b'5678', sfile.read(4))
647         self.assertEqual(b'9', sfile.read(4))
648         self.assertEqual(b'', sfile.read(4))
649
650     def test_tell_after_block_read(self):
651         # Override StreamFileReaderTestCase.test_tell_after_block_read
652         sfile = self.make_count_reader(nocache=True)
653         self.assertEqual(b'12345678', sfile.read(8))
654         self.assertEqual(8, sfile.tell())
655
656     def test_prefetch(self):
657         keep = ArvadosFileWriterTestCase.MockKeep({
658             "2e9ec317e197819358fbc43afca7d837+8": b"01234567",
659             "e8dc4081b13434b45189a720b77b6818+8": b"abcdefgh",
660         })
661         with Collection(". 2e9ec317e197819358fbc43afca7d837+8 e8dc4081b13434b45189a720b77b6818+8 0:16:count.txt\n", keep_client=keep) as c:
662             r = c.open("count.txt", "rb")
663             self.assertEqual(b"0123", r.read(4))
664         self.assertIn("2e9ec317e197819358fbc43afca7d837+8", keep.requests)
665         self.assertIn("e8dc4081b13434b45189a720b77b6818+8", keep.requests)
666
667     def test__eq__from_manifest(self):
668         with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt') as c1:
669             with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt') as c2:
670                 self.assertTrue(c1["count1.txt"] == c2["count1.txt"])
671                 self.assertFalse(c1["count1.txt"] != c2["count1.txt"])
672
673     def test__eq__from_writes(self):
674         with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt') as c1:
675             with Collection() as c2:
676                 f = c2.open("count1.txt", "wb")
677                 f.write("0123456789")
678
679                 self.assertTrue(c1["count1.txt"] == c2["count1.txt"])
680                 self.assertFalse(c1["count1.txt"] != c2["count1.txt"])
681
682     def test__ne__(self):
683         with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt') as c1:
684             with Collection() as c2:
685                 f = c2.open("count1.txt", "wb")
686                 f.write("1234567890")
687
688                 self.assertTrue(c1["count1.txt"] != c2["count1.txt"])
689                 self.assertFalse(c1["count1.txt"] == c2["count1.txt"])
690
691
692 class ArvadosFileReadTestCase(unittest.TestCase, StreamRetryTestMixin):
693     def reader_for(self, coll_name, **kwargs):
694         stream = []
695         segments = []
696         n = 0
697         for d in self.manifest_for(coll_name).split():
698             try:
699                 k = KeepLocator(d)
700                 segments.append(Range(n, n, k.size))
701                 stream.append(Range(d, n, k.size))
702                 n += k.size
703             except ValueError:
704                 pass
705
706         blockmanager = arvados.arvfile._BlockManager(self.keep_client())
707         blockmanager.prefetch_enabled = False
708         col = Collection(keep_client=self.keep_client(), block_manager=blockmanager)
709         af = ArvadosFile(col, "test",
710                          stream=stream,
711                          segments=segments)
712         kwargs.setdefault('mode', 'rb')
713         return ArvadosFileReader(af, **kwargs)
714
715     def read_for_test(self, reader, byte_count, **kwargs):
716         return reader.read(byte_count, **kwargs)
717
718
719 class ArvadosFileReadFromTestCase(ArvadosFileReadTestCase):
720     def read_for_test(self, reader, byte_count, **kwargs):
721         return reader.readfrom(0, byte_count, **kwargs)
722
723
724 class ArvadosFileReadAllTestCase(ArvadosFileReadTestCase):
725     def read_for_test(self, reader, byte_count, **kwargs):
726         return b''.join(reader.readall(**kwargs))
727
728
729 class ArvadosFileReadAllDecompressedTestCase(ArvadosFileReadTestCase):
730     def read_for_test(self, reader, byte_count, **kwargs):
731         return b''.join(reader.readall_decompressed(**kwargs))
732
733
734 class ArvadosFileReadlinesTestCase(ArvadosFileReadTestCase):
735     def read_for_test(self, reader, byte_count, **kwargs):
736         return ''.join(reader.readlines(**kwargs)).encode()
737
738
739 class ArvadosFileTestCase(unittest.TestCase):
740     def datetime_to_hex(self, dt):
741         return hex(int(time.mktime(dt.timetuple())))[2:]
742
743     def test_permission_expired(self):
744         base_manifest = ". 781e5e245d69b566979b86e28d23f2c7+10+A715fd31f8111894f717eb1003c1b0216799dd9ec@{} 0:10:count.txt\n"
745         now = datetime.datetime.now()
746         a_week_ago = now - datetime.timedelta(days=7)
747         a_month_ago = now - datetime.timedelta(days=30)
748         a_week_from_now = now + datetime.timedelta(days=7)
749         with Collection(base_manifest.format(self.datetime_to_hex(a_week_from_now))) as c:
750             self.assertFalse(c.find('count.txt').permission_expired())
751         with Collection(base_manifest.format(self.datetime_to_hex(a_week_ago))) as c:
752             f = c.find('count.txt')
753             self.assertTrue(f.permission_expired())
754             self.assertTrue(f.permission_expired(a_week_from_now))
755             self.assertFalse(f.permission_expired(a_month_ago))
756
757
758 class BlockManagerTest(unittest.TestCase):
759     def test_bufferblock_append(self):
760         keep = ArvadosFileWriterTestCase.MockKeep({})
761         with arvados.arvfile._BlockManager(keep) as blockmanager:
762             bufferblock = blockmanager.alloc_bufferblock()
763             bufferblock.append("foo")
764
765             self.assertEqual(bufferblock.size(), 3)
766             self.assertEqual(bufferblock.buffer_view[0:3], b"foo")
767             self.assertEqual(bufferblock.locator(), "acbd18db4cc2f85cedef654fccc4a4d8+3")
768
769             bufferblock.append("bar")
770
771             self.assertEqual(bufferblock.size(), 6)
772             self.assertEqual(bufferblock.buffer_view[0:6], b"foobar")
773             self.assertEqual(bufferblock.locator(), "3858f62230ac3c915f300c664312c63f+6")
774
775             bufferblock.set_state(arvados.arvfile._BufferBlock.PENDING)
776             with self.assertRaises(arvados.errors.AssertionError):
777                 bufferblock.append("bar")
778
779     def test_bufferblock_dup(self):
780         keep = ArvadosFileWriterTestCase.MockKeep({})
781         with arvados.arvfile._BlockManager(keep) as blockmanager:
782             bufferblock = blockmanager.alloc_bufferblock()
783             bufferblock.append("foo")
784
785             self.assertEqual(bufferblock.size(), 3)
786             self.assertEqual(bufferblock.buffer_view[0:3], b"foo")
787             self.assertEqual(bufferblock.locator(), "acbd18db4cc2f85cedef654fccc4a4d8+3")
788             bufferblock.set_state(arvados.arvfile._BufferBlock.PENDING)
789
790             bufferblock2 = blockmanager.dup_block(bufferblock, None)
791             self.assertNotEqual(bufferblock.blockid, bufferblock2.blockid)
792
793             bufferblock2.append("bar")
794
795             self.assertEqual(bufferblock2.size(), 6)
796             self.assertEqual(bufferblock2.buffer_view[0:6], b"foobar")
797             self.assertEqual(bufferblock2.locator(), "3858f62230ac3c915f300c664312c63f+6")
798
799             self.assertEqual(bufferblock.size(), 3)
800             self.assertEqual(bufferblock.buffer_view[0:3], b"foo")
801             self.assertEqual(bufferblock.locator(), "acbd18db4cc2f85cedef654fccc4a4d8+3")
802
803     def test_bufferblock_get(self):
804         keep = ArvadosFileWriterTestCase.MockKeep({
805             "781e5e245d69b566979b86e28d23f2c7+10": b"0123456789",
806         })
807         with arvados.arvfile._BlockManager(keep) as blockmanager:
808             bufferblock = blockmanager.alloc_bufferblock()
809             bufferblock.append("foo")
810
811             self.assertEqual(blockmanager.get_block_contents("781e5e245d69b566979b86e28d23f2c7+10", 1), b"0123456789")
812             self.assertEqual(blockmanager.get_block_contents(bufferblock.blockid, 1), b"foo")
813
814     def test_bufferblock_commit(self):
815         mockkeep = mock.MagicMock()
816         with arvados.arvfile._BlockManager(mockkeep) as blockmanager:
817             bufferblock = blockmanager.alloc_bufferblock()
818             bufferblock.owner = mock.MagicMock()
819             def flush(sync=None):
820                 blockmanager.commit_bufferblock(bufferblock, sync)
821             bufferblock.owner.flush.side_effect = flush
822             bufferblock.append("foo")
823             blockmanager.commit_all()
824             self.assertTrue(bufferblock.owner.flush.called)
825             self.assertTrue(mockkeep.put.called)
826             self.assertEqual(bufferblock.state(), arvados.arvfile._BufferBlock.COMMITTED)
827             self.assertIsNone(bufferblock.buffer_view)
828
829     def test_bufferblock_commit_pending(self):
830         # Test for bug #7225
831         mockkeep = mock.MagicMock()
832         mockkeep.put.side_effect = lambda x: time.sleep(1)
833         with arvados.arvfile._BlockManager(mockkeep) as blockmanager:
834             bufferblock = blockmanager.alloc_bufferblock()
835             bufferblock.append("foo")
836
837             blockmanager.commit_bufferblock(bufferblock, False)
838             self.assertEqual(bufferblock.state(), arvados.arvfile._BufferBlock.PENDING)
839
840             blockmanager.commit_bufferblock(bufferblock, True)
841             self.assertEqual(bufferblock.state(), arvados.arvfile._BufferBlock.COMMITTED)
842
843
844     def test_bufferblock_commit_with_error(self):
845         mockkeep = mock.MagicMock()
846         mockkeep.put.side_effect = arvados.errors.KeepWriteError("fail")
847         with arvados.arvfile._BlockManager(mockkeep) as blockmanager:
848             bufferblock = blockmanager.alloc_bufferblock()
849             bufferblock.owner = mock.MagicMock()
850             def flush(sync=None):
851                 blockmanager.commit_bufferblock(bufferblock, sync)
852             bufferblock.owner.flush.side_effect = flush
853             bufferblock.append("foo")
854             with self.assertRaises(arvados.errors.KeepWriteError) as err:
855                 blockmanager.commit_all()
856             self.assertTrue(bufferblock.owner.flush.called)
857             self.assertEqual(str(err.exception), "Error writing some blocks: block acbd18db4cc2f85cedef654fccc4a4d8+3 raised KeepWriteError (fail)")
858             self.assertEqual(bufferblock.state(), arvados.arvfile._BufferBlock.ERROR)