closes #10903
[arvados.git] / sdk / python / tests / test_arvfile.py
1 #!/usr/bin/env python
2
3 import bz2
4 import datetime
5 import gzip
6 import io
7 import mock
8 import os
9 import unittest
10 import time
11
12 import arvados
13 from arvados._ranges import Range
14 from arvados.keep import KeepLocator
15 from arvados.collection import Collection, CollectionReader
16 from arvados.arvfile import ArvadosFile, ArvadosFileReader
17
18 import arvados_testutil as tutil
19 from test_stream import StreamFileReaderTestCase, StreamRetryTestMixin
20
21 class ArvadosFileWriterTestCase(unittest.TestCase):
22     class MockKeep(object):
23         def __init__(self, blocks):
24             self.blocks = blocks
25             self.requests = []
26         def get(self, locator, num_retries=0):
27             self.requests.append(locator)
28             return self.blocks.get(locator)
29         def get_from_cache(self, locator):
30             self.requests.append(locator)
31             return self.blocks.get(locator)
32         def put(self, data, num_retries=None, copies=None):
33             pdh = tutil.str_keep_locator(data)
34             self.blocks[pdh] = str(data)
35             return pdh
36
37     class MockApi(object):
38         def __init__(self, b, r):
39             self.body = b
40             self.response = r
41             self._schema = ArvadosFileWriterTestCase.MockApi.MockSchema()
42             self._rootDesc = {}
43         class MockSchema(object):
44             def __init__(self):
45                 self.schemas = {'Collection': {'properties': {'replication_desired': {'type':'integer'}}}}
46         class MockCollections(object):
47             def __init__(self, b, r):
48                 self.body = b
49                 self.response = r
50             class Execute(object):
51                 def __init__(self, r):
52                     self.response = r
53                 def execute(self, num_retries=None):
54                     return self.response
55             def create(self, ensure_unique_name=False, body=None):
56                 if body != self.body:
57                     raise Exception("Body %s does not match expectation %s" % (body, self.body))
58                 return ArvadosFileWriterTestCase.MockApi.MockCollections.Execute(self.response)
59             def update(self, uuid=None, body=None):
60                 return ArvadosFileWriterTestCase.MockApi.MockCollections.Execute(self.response)
61         def collections(self):
62             return ArvadosFileWriterTestCase.MockApi.MockCollections(self.body, self.response)
63
64
65     def test_truncate(self):
66         keep = ArvadosFileWriterTestCase.MockKeep({"781e5e245d69b566979b86e28d23f2c7+10": "0123456789"})
67         api = ArvadosFileWriterTestCase.MockApi({"name":"test_truncate",
68                                                  "manifest_text":". 781e5e245d69b566979b86e28d23f2c7+10 0:8:count.txt\n",
69                                                  "replication_desired":None},
70                                                 {"uuid":"zzzzz-4zz18-mockcollection0",
71                                                  "manifest_text":". 781e5e245d69b566979b86e28d23f2c7+10 0:8:count.txt\n",
72                                                  "portable_data_hash":"7fcd0eaac3aad4c31a6a0e756475da92+52"})
73         with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n',
74                              api_client=api, keep_client=keep) as c:
75             writer = c.open("count.txt", "r+")
76             self.assertEqual(writer.size(), 10)
77             self.assertEqual("0123456789", writer.read(12))
78
79             writer.truncate(8)
80
81             # Make sure reading off the end doesn't break
82             self.assertEqual("", writer.read(12))
83
84             self.assertEqual(writer.size(), 8)
85             writer.seek(0, os.SEEK_SET)
86             self.assertEqual("01234567", writer.read(12))
87
88             self.assertIsNone(c.manifest_locator())
89             self.assertTrue(c.modified())
90             c.save_new("test_truncate")
91             self.assertEqual("zzzzz-4zz18-mockcollection0", c.manifest_locator())
92             self.assertFalse(c.modified())
93
94     def test_write_to_end(self):
95         keep = ArvadosFileWriterTestCase.MockKeep({"781e5e245d69b566979b86e28d23f2c7+10": "0123456789"})
96         api = ArvadosFileWriterTestCase.MockApi({"name":"test_append",
97                                                  "manifest_text": ". 781e5e245d69b566979b86e28d23f2c7+10 acbd18db4cc2f85cedef654fccc4a4d8+3 0:13:count.txt\n",
98                                                  "replication_desired":None},
99                                                 {"uuid":"zzzzz-4zz18-mockcollection0",
100                                                  "manifest_text": ". 781e5e245d69b566979b86e28d23f2c7+10 acbd18db4cc2f85cedef654fccc4a4d8+3 0:13:count.txt\n",
101                                                  "portable_data_hash":"c5c3af76565c8efb6a806546bcf073f3+88"})
102         with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n',
103                              api_client=api, keep_client=keep) as c:
104             writer = c.open("count.txt", "r+")
105             self.assertEqual(writer.size(), 10)
106
107             writer.seek(5, os.SEEK_SET)
108             self.assertEqual("56789", writer.read(8))
109
110             writer.seek(10, os.SEEK_SET)
111             writer.write("foo")
112             self.assertEqual(writer.size(), 13)
113
114             writer.seek(5, os.SEEK_SET)
115             self.assertEqual("56789foo", writer.read(8))
116
117             self.assertIsNone(c.manifest_locator())
118             self.assertTrue(c.modified())
119             self.assertIsNone(keep.get("acbd18db4cc2f85cedef654fccc4a4d8+3"))
120
121             c.save_new("test_append")
122             self.assertEqual("zzzzz-4zz18-mockcollection0", c.manifest_locator())
123             self.assertFalse(c.modified())
124             self.assertEqual("foo", keep.get("acbd18db4cc2f85cedef654fccc4a4d8+3"))
125
126
127     def test_append(self):
128         keep = ArvadosFileWriterTestCase.MockKeep({"781e5e245d69b566979b86e28d23f2c7+10": "0123456789"})
129         c = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n', keep_client=keep)
130         writer = c.open("count.txt", "a+")
131         self.assertEqual(writer.read(20), "0123456789")
132         writer.seek(0, os.SEEK_SET)
133
134         writer.write("hello")
135         self.assertEqual(writer.read(20), "0123456789hello")
136         writer.seek(0, os.SEEK_SET)
137
138         writer.write("world")
139         self.assertEqual(writer.read(20), "0123456789helloworld")
140
141         self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 fc5e038d38a57032085441e7fe7010b0+10 0:20:count.txt\n", c.portable_manifest_text())
142
143     def test_write_at_beginning(self):
144         keep = ArvadosFileWriterTestCase.MockKeep({"781e5e245d69b566979b86e28d23f2c7+10": "0123456789"})
145         with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n',
146                              keep_client=keep) as c:
147             writer = c.open("count.txt", "r+")
148             self.assertEqual("0123456789", writer.readfrom(0, 13))
149             writer.seek(0, os.SEEK_SET)
150             writer.write("foo")
151             self.assertEqual(writer.size(), 10)
152             self.assertEqual("foo3456789", writer.readfrom(0, 13))
153             self.assertEqual(". acbd18db4cc2f85cedef654fccc4a4d8+3 781e5e245d69b566979b86e28d23f2c7+10 0:3:count.txt 6:7:count.txt\n", c.portable_manifest_text())
154
155     def test_write_empty(self):
156         keep = ArvadosFileWriterTestCase.MockKeep({})
157         with Collection(keep_client=keep) as c:
158             writer = c.open("count.txt", "w")
159             self.assertEqual(writer.size(), 0)
160             self.assertEqual(". d41d8cd98f00b204e9800998ecf8427e+0 0:0:count.txt\n", c.portable_manifest_text())
161
162     def test_save_manifest_text(self):
163         keep = ArvadosFileWriterTestCase.MockKeep({})
164         with Collection(keep_client=keep) as c:
165             writer = c.open("count.txt", "w")
166             writer.write("0123456789")
167             self.assertEqual('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n', c.portable_manifest_text())
168             self.assertNotIn('781e5e245d69b566979b86e28d23f2c7+10', keep.blocks)
169
170             self.assertEqual('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n', c.save_new(create_collection_record=False))
171             self.assertIn('781e5e245d69b566979b86e28d23f2c7+10', keep.blocks)
172
173     def test_get_manifest_text_commits(self):
174          keep = ArvadosFileWriterTestCase.MockKeep({})
175          with Collection(keep_client=keep) as c:
176              writer = c.open("count.txt", "w")
177              writer.write("0123456789")
178              self.assertEqual('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n', c.portable_manifest_text())
179              self.assertNotIn('781e5e245d69b566979b86e28d23f2c7+10', keep.blocks)
180              self.assertEqual('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n', c.manifest_text())
181              self.assertIn('781e5e245d69b566979b86e28d23f2c7+10', keep.blocks)
182
183
184     def test_write_in_middle(self):
185         keep = ArvadosFileWriterTestCase.MockKeep({"781e5e245d69b566979b86e28d23f2c7+10": "0123456789"})
186         with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n',
187                              keep_client=keep) as c:
188             writer = c.open("count.txt", "r+")
189             self.assertEqual("0123456789", writer.readfrom(0, 13))
190             writer.seek(3, os.SEEK_SET)
191             writer.write("foo")
192             self.assertEqual(writer.size(), 10)
193             self.assertEqual("012foo6789", writer.readfrom(0, 13))
194             self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 acbd18db4cc2f85cedef654fccc4a4d8+3 0:3:count.txt 10:3:count.txt 6:4:count.txt\n", c.portable_manifest_text())
195
196     def test_write_at_end(self):
197         keep = ArvadosFileWriterTestCase.MockKeep({"781e5e245d69b566979b86e28d23f2c7+10": "0123456789"})
198         with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n',
199                              keep_client=keep) as c:
200             writer = c.open("count.txt", "r+")
201             self.assertEqual("0123456789", writer.readfrom(0, 13))
202             writer.seek(7, os.SEEK_SET)
203             writer.write("foo")
204             self.assertEqual(writer.size(), 10)
205             self.assertEqual("0123456foo", writer.readfrom(0, 13))
206             self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 acbd18db4cc2f85cedef654fccc4a4d8+3 0:7:count.txt 10:3:count.txt\n", c.portable_manifest_text())
207
208     def test_write_across_segment_boundary(self):
209         keep = ArvadosFileWriterTestCase.MockKeep({"781e5e245d69b566979b86e28d23f2c7+10": "0123456789"})
210         with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt 0:10:count.txt\n',
211                              keep_client=keep) as c:
212             writer = c.open("count.txt", "r+")
213             self.assertEqual("012345678901234", writer.readfrom(0, 15))
214             writer.seek(7, os.SEEK_SET)
215             writer.write("foobar")
216             self.assertEqual(writer.size(), 20)
217             self.assertEqual("0123456foobar34", writer.readfrom(0, 15))
218             self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 3858f62230ac3c915f300c664312c63f+6 0:7:count.txt 10:6:count.txt 3:7:count.txt\n", c.portable_manifest_text())
219
220     def test_write_across_several_segments(self):
221         keep = ArvadosFileWriterTestCase.MockKeep({"781e5e245d69b566979b86e28d23f2c7+10": "0123456789"})
222         with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:4:count.txt 0:4:count.txt 0:4:count.txt',
223                              keep_client=keep) as c:
224             writer = c.open("count.txt", "r+")
225             self.assertEqual("012301230123", writer.readfrom(0, 15))
226             writer.seek(2, os.SEEK_SET)
227             writer.write("abcdefg")
228             self.assertEqual(writer.size(), 12)
229             self.assertEqual("01abcdefg123", writer.readfrom(0, 15))
230             self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 7ac66c0f148de9519b8bd264312c4d64+7 0:2:count.txt 10:7:count.txt 1:3:count.txt\n", c.portable_manifest_text())
231
232     def test_write_large(self):
233         keep = ArvadosFileWriterTestCase.MockKeep({})
234         api = ArvadosFileWriterTestCase.MockApi({"name":"test_write_large",
235                                                  "manifest_text": ". a5de24f4417cfba9d5825eadc2f4ca49+67108000 598cc1a4ccaef8ab6e4724d87e675d78+32892000 0:100000000:count.txt\n",
236                                                  "replication_desired":None},
237                                                 {"uuid":"zzzzz-4zz18-mockcollection0",
238                                                  "manifest_text": ". a5de24f4417cfba9d5825eadc2f4ca49+67108000 598cc1a4ccaef8ab6e4724d87e675d78+32892000 0:100000000:count.txt\n",
239                                                  "portable_data_hash":"9132ca8e3f671c76103a38f5bc24328c+108"})
240         with Collection('. ' + arvados.config.EMPTY_BLOCK_LOCATOR + ' 0:0:count.txt',
241                              api_client=api, keep_client=keep) as c:
242             writer = c.open("count.txt", "r+")
243             text = "0123456789" * 100
244             for b in xrange(0, 100000):
245                 writer.write(text)
246             self.assertEqual(writer.size(), 100000000)
247
248             self.assertIsNone(c.manifest_locator())
249             self.assertTrue(c.modified())
250             c.save_new("test_write_large")
251             self.assertEqual("zzzzz-4zz18-mockcollection0", c.manifest_locator())
252             self.assertFalse(c.modified())
253
254
255     def test_large_write(self):
256         keep = ArvadosFileWriterTestCase.MockKeep({})
257         api = ArvadosFileWriterTestCase.MockApi({}, {})
258         with Collection('. ' + arvados.config.EMPTY_BLOCK_LOCATOR + ' 0:0:count.txt',
259                              api_client=api, keep_client=keep) as c:
260             writer = c.open("count.txt", "r+")
261             self.assertEqual(writer.size(), 0)
262
263             text = "0123456789"
264             writer.write(text)
265             text = "0123456789" * 9999999
266             writer.write(text)
267             self.assertEqual(writer.size(), 100000000)
268
269             self.assertEqual(c.manifest_text(), ". 781e5e245d69b566979b86e28d23f2c7+10 48dd23ea1645fd47d789804d71b5bb8e+67108864 77c57dc6ac5a10bb2205caaa73187994+32891126 0:100000000:count.txt\n")
270
271     def test_rewrite_on_empty_file(self):
272         keep = ArvadosFileWriterTestCase.MockKeep({})
273         with Collection('. ' + arvados.config.EMPTY_BLOCK_LOCATOR + ' 0:0:count.txt',
274                              keep_client=keep) as c:
275             writer = c.open("count.txt", "r+")
276             for b in xrange(0, 10):
277                 writer.seek(0, os.SEEK_SET)
278                 writer.write("0123456789")
279
280             self.assertEqual(writer.size(), 10)
281             self.assertEqual("0123456789", writer.readfrom(0, 20))
282             self.assertEqual(". 7a08b07e84641703e5f2c836aa59a170+100 90:10:count.txt\n", c.portable_manifest_text())
283             writer.flush()
284             self.assertEqual(writer.size(), 10)
285             self.assertEqual("0123456789", writer.readfrom(0, 20))
286             self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n", c.portable_manifest_text())
287
288     def test_rewrite_append_existing_file(self):
289         keep = ArvadosFileWriterTestCase.MockKeep({"781e5e245d69b566979b86e28d23f2c7+10": "0123456789"})
290         with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt',
291                              keep_client=keep) as c:
292             writer = c.open("count.txt", "r+")
293             for b in xrange(0, 10):
294                 writer.seek(10, os.SEEK_SET)
295                 writer.write("abcdefghij")
296
297             self.assertEqual(writer.size(), 20)
298             self.assertEqual("0123456789abcdefghij", writer.readfrom(0, 20))
299             self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 ae5f43bab79cf0be33f025fa97ae7398+100 0:10:count.txt 100:10:count.txt\n", c.portable_manifest_text())
300
301             writer.arvadosfile.flush()
302             self.assertEqual(writer.size(), 20)
303             self.assertEqual("0123456789abcdefghij", writer.readfrom(0, 20))
304             self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 a925576942e94b2ef57a066101b48876+10 0:20:count.txt\n", c.portable_manifest_text())
305
306     def test_rewrite_over_existing_file(self):
307         keep = ArvadosFileWriterTestCase.MockKeep({"781e5e245d69b566979b86e28d23f2c7+10": "0123456789"})
308         with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt',
309                              keep_client=keep) as c:
310             writer = c.open("count.txt", "r+")
311             for b in xrange(0, 10):
312                 writer.seek(5, os.SEEK_SET)
313                 writer.write("abcdefghij")
314
315             self.assertEqual(writer.size(), 15)
316             self.assertEqual("01234abcdefghij", writer.readfrom(0, 20))
317             self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 ae5f43bab79cf0be33f025fa97ae7398+100 0:5:count.txt 100:10:count.txt\n", c.portable_manifest_text())
318
319             writer.arvadosfile.flush()
320
321             self.assertEqual(writer.size(), 15)
322             self.assertEqual("01234abcdefghij", writer.readfrom(0, 20))
323             self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 a925576942e94b2ef57a066101b48876+10 0:5:count.txt 10:10:count.txt\n", c.portable_manifest_text())
324
325     def test_write_large_rewrite(self):
326         keep = ArvadosFileWriterTestCase.MockKeep({})
327         api = ArvadosFileWriterTestCase.MockApi({"name":"test_write_large",
328                                                  "manifest_text": ". 37400a68af9abdd76ca5bf13e819e42a+32892003 a5de24f4417cfba9d5825eadc2f4ca49+67108000 32892000:3:count.txt 32892006:67107997:count.txt 0:32892000:count.txt\n",
329                                                  "replication_desired":None},
330                                                 {"uuid":"zzzzz-4zz18-mockcollection0",
331                                                  "manifest_text": ". 37400a68af9abdd76ca5bf13e819e42a+32892003 a5de24f4417cfba9d5825eadc2f4ca49+67108000 32892000:3:count.txt 32892006:67107997:count.txt 0:32892000:count.txt\n",
332                                                  "portable_data_hash":"217665c6b713e1b78dfba7ebd42344db+156"})
333         with Collection('. ' + arvados.config.EMPTY_BLOCK_LOCATOR + ' 0:0:count.txt',
334                              api_client=api, keep_client=keep) as c:
335             writer = c.open("count.txt", "r+")
336             text = ''.join(["0123456789" for a in xrange(0, 100)])
337             for b in xrange(0, 100000):
338                 writer.write(text)
339             writer.seek(0, os.SEEK_SET)
340             writer.write("foo")
341             self.assertEqual(writer.size(), 100000000)
342
343             self.assertIsNone(c.manifest_locator())
344             self.assertTrue(c.modified())
345             c.save_new("test_write_large")
346             self.assertEqual("zzzzz-4zz18-mockcollection0", c.manifest_locator())
347             self.assertFalse(c.modified())
348
349     def test_create(self):
350         keep = ArvadosFileWriterTestCase.MockKeep({})
351         api = ArvadosFileWriterTestCase.MockApi({"name":"test_create",
352                                                  "manifest_text":". 2e9ec317e197819358fbc43afca7d837+8 0:8:count.txt\n",
353                                                  "replication_desired":None},
354                                                 {"uuid":"zzzzz-4zz18-mockcollection0",
355                                                  "manifest_text":". 2e9ec317e197819358fbc43afca7d837+8 0:8:count.txt\n",
356                                                  "portable_data_hash":"7a461a8c58601798f690f8b368ac4423+51"})
357         with Collection(api_client=api, keep_client=keep) as c:
358             writer = c.open("count.txt", "w+")
359             self.assertEqual(writer.size(), 0)
360             writer.write("01234567")
361             self.assertEqual(writer.size(), 8)
362
363             self.assertIsNone(c.manifest_locator())
364             self.assertTrue(c.modified())
365             self.assertIsNone(keep.get("2e9ec317e197819358fbc43afca7d837+8"))
366             c.save_new("test_create")
367             self.assertEqual("zzzzz-4zz18-mockcollection0", c.manifest_locator())
368             self.assertFalse(c.modified())
369             self.assertEqual("01234567", keep.get("2e9ec317e197819358fbc43afca7d837+8"))
370
371
372     def test_create_subdir(self):
373         keep = ArvadosFileWriterTestCase.MockKeep({})
374         api = ArvadosFileWriterTestCase.MockApi({"name":"test_create",
375                                                  "manifest_text":"./foo/bar 2e9ec317e197819358fbc43afca7d837+8 0:8:count.txt\n",
376                                                  "replication_desired":None},
377                                                 {"uuid":"zzzzz-4zz18-mockcollection0",
378                                                  "manifest_text":"./foo/bar 2e9ec317e197819358fbc43afca7d837+8 0:8:count.txt\n",
379                                                  "portable_data_hash":"1b02aaa62528d28a5be41651cbb9d7c7+59"})
380         with Collection(api_client=api, keep_client=keep) as c:
381             self.assertIsNone(c.api_response())
382             writer = c.open("foo/bar/count.txt", "w+")
383             writer.write("01234567")
384             self.assertFalse(c.committed())
385             c.save_new("test_create")
386             self.assertTrue(c.committed())
387             self.assertEqual(c.api_response(), api.response)
388
389     def test_overwrite(self):
390         keep = ArvadosFileWriterTestCase.MockKeep({"781e5e245d69b566979b86e28d23f2c7+10": "0123456789"})
391         api = ArvadosFileWriterTestCase.MockApi({"name":"test_overwrite",
392                                                  "manifest_text":". 2e9ec317e197819358fbc43afca7d837+8 0:8:count.txt\n",
393                                                  "replication_desired":None},
394                                                 {"uuid":"zzzzz-4zz18-mockcollection0",
395                                                  "manifest_text":". 2e9ec317e197819358fbc43afca7d837+8 0:8:count.txt\n",
396                                                  "portable_data_hash":"7a461a8c58601798f690f8b368ac4423+51"})
397         with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n',
398                              api_client=api, keep_client=keep) as c:
399             writer = c.open("count.txt", "w+")
400             self.assertEqual(writer.size(), 0)
401             writer.write("01234567")
402             self.assertEqual(writer.size(), 8)
403
404             self.assertIsNone(c.manifest_locator())
405             self.assertTrue(c.modified())
406             c.save_new("test_overwrite")
407             self.assertEqual("zzzzz-4zz18-mockcollection0", c.manifest_locator())
408             self.assertFalse(c.modified())
409
410     def test_file_not_found(self):
411         with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n') as c:
412             with self.assertRaises(IOError):
413                 writer = c.open("nocount.txt", "r")
414
415     def test_cannot_open_directory(self):
416         with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n') as c:
417             with self.assertRaises(IOError):
418                 writer = c.open(".", "r")
419
420     def test_create_multiple(self):
421         keep = ArvadosFileWriterTestCase.MockKeep({})
422         api = ArvadosFileWriterTestCase.MockApi({"name":"test_create_multiple",
423                                                  "manifest_text":". 2e9ec317e197819358fbc43afca7d837+8 e8dc4081b13434b45189a720b77b6818+8 0:8:count1.txt 8:8:count2.txt\n",
424                                                  "replication_desired":None},
425                                                 {"uuid":"zzzzz-4zz18-mockcollection0",
426                                                  "manifest_text":". 2e9ec317e197819358fbc43afca7d837+8 e8dc4081b13434b45189a720b77b6818+8 0:8:count1.txt 8:8:count2.txt\n",
427                                                  "portable_data_hash":"71e7bb6c00d31fc2b4364199fd97be08+102"})
428         with Collection(api_client=api, keep_client=keep) as c:
429             w1 = c.open("count1.txt", "w")
430             w2 = c.open("count2.txt", "w")
431             w1.write("01234567")
432             w2.write("abcdefgh")
433             self.assertEqual(w1.size(), 8)
434             self.assertEqual(w2.size(), 8)
435
436             self.assertIsNone(c.manifest_locator())
437             self.assertTrue(c.modified())
438             self.assertIsNone(keep.get("2e9ec317e197819358fbc43afca7d837+8"))
439             c.save_new("test_create_multiple")
440             self.assertEqual("zzzzz-4zz18-mockcollection0", c.manifest_locator())
441             self.assertFalse(c.modified())
442             self.assertEqual("01234567", keep.get("2e9ec317e197819358fbc43afca7d837+8"))
443
444
445 class ArvadosFileReaderTestCase(StreamFileReaderTestCase):
446     class MockParent(object):
447         class MockBlockMgr(object):
448             def __init__(self, blocks, nocache):
449                 self.blocks = blocks
450                 self.nocache = nocache
451
452             def block_prefetch(self, loc):
453                 pass
454
455             def get_block_contents(self, loc, num_retries=0, cache_only=False):
456                 if self.nocache and cache_only:
457                     return None
458                 return self.blocks[loc]
459
460         def __init__(self, blocks, nocache):
461             self.blocks = blocks
462             self.nocache = nocache
463             self.lock = arvados.arvfile.NoopLock()
464
465         def root_collection(self):
466             return self
467
468         def _my_block_manager(self):
469             return ArvadosFileReaderTestCase.MockParent.MockBlockMgr(self.blocks, self.nocache)
470
471
472     def make_count_reader(self, nocache=False):
473         stream = []
474         n = 0
475         blocks = {}
476         for d in ['01234', '34567', '67890']:
477             loc = tutil.str_keep_locator(d)
478             blocks[loc] = d
479             stream.append(Range(loc, n, len(d)))
480             n += len(d)
481         af = ArvadosFile(ArvadosFileReaderTestCase.MockParent(blocks, nocache), "count.txt", stream=stream, segments=[Range(1, 0, 3), Range(6, 3, 3), Range(11, 6, 3)])
482         return ArvadosFileReader(af)
483
484     def test_read_block_crossing_behavior(self):
485         # read() needs to return all the data requested if possible, even if it
486         # crosses uncached blocks: https://arvados.org/issues/5856
487         sfile = self.make_count_reader(nocache=True)
488         self.assertEqual('12345678', sfile.read(8))
489
490     def test_successive_reads(self):
491         # Override StreamFileReaderTestCase.test_successive_reads
492         sfile = self.make_count_reader(nocache=True)
493         self.assertEqual('1234', sfile.read(4))
494         self.assertEqual('5678', sfile.read(4))
495         self.assertEqual('9', sfile.read(4))
496         self.assertEqual('', sfile.read(4))
497
498     def test_tell_after_block_read(self):
499         # Override StreamFileReaderTestCase.test_tell_after_block_read
500         sfile = self.make_count_reader(nocache=True)
501         self.assertEqual('12345678', sfile.read(8))
502         self.assertEqual(8, sfile.tell())
503
504     def test_prefetch(self):
505         keep = ArvadosFileWriterTestCase.MockKeep({"2e9ec317e197819358fbc43afca7d837+8": "01234567", "e8dc4081b13434b45189a720b77b6818+8": "abcdefgh"})
506         with Collection(". 2e9ec317e197819358fbc43afca7d837+8 e8dc4081b13434b45189a720b77b6818+8 0:16:count.txt\n", keep_client=keep) as c:
507             r = c.open("count.txt", "r")
508             self.assertEqual("0123", r.read(4))
509         self.assertIn("2e9ec317e197819358fbc43afca7d837+8", keep.requests)
510         self.assertIn("e8dc4081b13434b45189a720b77b6818+8", keep.requests)
511
512     def test__eq__from_manifest(self):
513         with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt') as c1:
514             with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt') as c2:
515                 self.assertTrue(c1["count1.txt"] == c2["count1.txt"])
516                 self.assertFalse(c1["count1.txt"] != c2["count1.txt"])
517
518     def test__eq__from_writes(self):
519         with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt') as c1:
520             with Collection() as c2:
521                 f = c2.open("count1.txt", "w")
522                 f.write("0123456789")
523
524                 self.assertTrue(c1["count1.txt"] == c2["count1.txt"])
525                 self.assertFalse(c1["count1.txt"] != c2["count1.txt"])
526
527     def test__ne__(self):
528         with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt') as c1:
529             with Collection() as c2:
530                 f = c2.open("count1.txt", "w")
531                 f.write("1234567890")
532
533                 self.assertTrue(c1["count1.txt"] != c2["count1.txt"])
534                 self.assertFalse(c1["count1.txt"] == c2["count1.txt"])
535
536
537 class ArvadosFileReadTestCase(unittest.TestCase, StreamRetryTestMixin):
538     def reader_for(self, coll_name, **kwargs):
539         stream = []
540         segments = []
541         n = 0
542         for d in self.manifest_for(coll_name).split():
543             try:
544                 k = KeepLocator(d)
545                 segments.append(Range(n, n, k.size))
546                 stream.append(Range(d, n, k.size))
547                 n += k.size
548             except ValueError:
549                 pass
550
551         blockmanager = arvados.arvfile._BlockManager(self.keep_client())
552         blockmanager.prefetch_enabled = False
553         col = Collection(keep_client=self.keep_client(), block_manager=blockmanager)
554         af = ArvadosFile(col, "test",
555                          stream=stream,
556                          segments=segments)
557         return ArvadosFileReader(af, **kwargs)
558
559     def read_for_test(self, reader, byte_count, **kwargs):
560         return reader.read(byte_count, **kwargs)
561
562
563 class ArvadosFileReadFromTestCase(ArvadosFileReadTestCase):
564     def read_for_test(self, reader, byte_count, **kwargs):
565         return reader.readfrom(0, byte_count, **kwargs)
566
567
568 class ArvadosFileReadAllTestCase(ArvadosFileReadTestCase):
569     def read_for_test(self, reader, byte_count, **kwargs):
570         return ''.join(reader.readall(**kwargs))
571
572
573 class ArvadosFileReadAllDecompressedTestCase(ArvadosFileReadTestCase):
574     def read_for_test(self, reader, byte_count, **kwargs):
575         return ''.join(reader.readall_decompressed(**kwargs))
576
577
578 class ArvadosFileReadlinesTestCase(ArvadosFileReadTestCase):
579     def read_for_test(self, reader, byte_count, **kwargs):
580         return ''.join(reader.readlines(**kwargs))
581
582
583 class ArvadosFileTestCase(unittest.TestCase):
584     def datetime_to_hex(self, dt):
585         return hex(int(time.mktime(dt.timetuple())))[2:]
586
587     def test_permission_expired(self):
588         base_manifest = ". 781e5e245d69b566979b86e28d23f2c7+10+A715fd31f8111894f717eb1003c1b0216799dd9ec@{} 0:10:count.txt\n"
589         now = datetime.datetime.now()
590         a_week_ago = now - datetime.timedelta(days=7)
591         a_month_ago = now - datetime.timedelta(days=30)
592         a_week_from_now = now + datetime.timedelta(days=7)
593         with Collection(base_manifest.format(self.datetime_to_hex(a_week_from_now))) as c:
594             self.assertFalse(c.find('count.txt').permission_expired())
595         with Collection(base_manifest.format(self.datetime_to_hex(a_week_ago))) as c:
596             f = c.find('count.txt')
597             self.assertTrue(f.permission_expired())
598             self.assertTrue(f.permission_expired(a_week_from_now))
599             self.assertFalse(f.permission_expired(a_month_ago))
600
601
602 class BlockManagerTest(unittest.TestCase):
603     def test_bufferblock_append(self):
604         keep = ArvadosFileWriterTestCase.MockKeep({})
605         with arvados.arvfile._BlockManager(keep) as blockmanager:
606             bufferblock = blockmanager.alloc_bufferblock()
607             bufferblock.append("foo")
608
609             self.assertEqual(bufferblock.size(), 3)
610             self.assertEqual(bufferblock.buffer_view[0:3], "foo")
611             self.assertEqual(bufferblock.locator(), "acbd18db4cc2f85cedef654fccc4a4d8+3")
612
613             bufferblock.append("bar")
614
615             self.assertEqual(bufferblock.size(), 6)
616             self.assertEqual(bufferblock.buffer_view[0:6], "foobar")
617             self.assertEqual(bufferblock.locator(), "3858f62230ac3c915f300c664312c63f+6")
618
619             bufferblock.set_state(arvados.arvfile._BufferBlock.PENDING)
620             with self.assertRaises(arvados.errors.AssertionError):
621                 bufferblock.append("bar")
622
623     def test_bufferblock_dup(self):
624         keep = ArvadosFileWriterTestCase.MockKeep({})
625         with arvados.arvfile._BlockManager(keep) as blockmanager:
626             bufferblock = blockmanager.alloc_bufferblock()
627             bufferblock.append("foo")
628
629             self.assertEqual(bufferblock.size(), 3)
630             self.assertEqual(bufferblock.buffer_view[0:3], "foo")
631             self.assertEqual(bufferblock.locator(), "acbd18db4cc2f85cedef654fccc4a4d8+3")
632             bufferblock.set_state(arvados.arvfile._BufferBlock.PENDING)
633
634             bufferblock2 = blockmanager.dup_block(bufferblock, None)
635             self.assertNotEqual(bufferblock.blockid, bufferblock2.blockid)
636
637             bufferblock2.append("bar")
638
639             self.assertEqual(bufferblock2.size(), 6)
640             self.assertEqual(bufferblock2.buffer_view[0:6], "foobar")
641             self.assertEqual(bufferblock2.locator(), "3858f62230ac3c915f300c664312c63f+6")
642
643             self.assertEqual(bufferblock.size(), 3)
644             self.assertEqual(bufferblock.buffer_view[0:3], "foo")
645             self.assertEqual(bufferblock.locator(), "acbd18db4cc2f85cedef654fccc4a4d8+3")
646
647     def test_bufferblock_get(self):
648         keep = ArvadosFileWriterTestCase.MockKeep({"781e5e245d69b566979b86e28d23f2c7+10": "0123456789"})
649         with arvados.arvfile._BlockManager(keep) as blockmanager:
650             bufferblock = blockmanager.alloc_bufferblock()
651             bufferblock.append("foo")
652
653             self.assertEqual(blockmanager.get_block_contents("781e5e245d69b566979b86e28d23f2c7+10", 1), "0123456789")
654             self.assertEqual(blockmanager.get_block_contents(bufferblock.blockid, 1), "foo")
655
656     def test_bufferblock_commit(self):
657         mockkeep = mock.MagicMock()
658         with arvados.arvfile._BlockManager(mockkeep) as blockmanager:
659             bufferblock = blockmanager.alloc_bufferblock()
660             bufferblock.owner = mock.MagicMock()
661             def flush(sync=None):
662                 blockmanager.commit_bufferblock(bufferblock, sync)
663             bufferblock.owner.flush.side_effect = flush
664             bufferblock.append("foo")
665             blockmanager.commit_all()
666             self.assertTrue(bufferblock.owner.flush.called)
667             self.assertTrue(mockkeep.put.called)
668             self.assertEqual(bufferblock.state(), arvados.arvfile._BufferBlock.COMMITTED)
669             self.assertIsNone(bufferblock.buffer_view)
670
671     def test_bufferblock_commit_pending(self):
672         # Test for bug #7225
673         mockkeep = mock.MagicMock()
674         mockkeep.put.side_effect = lambda x: time.sleep(1)
675         with arvados.arvfile._BlockManager(mockkeep) as blockmanager:
676             bufferblock = blockmanager.alloc_bufferblock()
677             bufferblock.append("foo")
678
679             blockmanager.commit_bufferblock(bufferblock, False)
680             self.assertEqual(bufferblock.state(), arvados.arvfile._BufferBlock.PENDING)
681
682             blockmanager.commit_bufferblock(bufferblock, True)
683             self.assertEqual(bufferblock.state(), arvados.arvfile._BufferBlock.COMMITTED)
684
685
686     def test_bufferblock_commit_with_error(self):
687         mockkeep = mock.MagicMock()
688         mockkeep.put.side_effect = arvados.errors.KeepWriteError("fail")
689         with arvados.arvfile._BlockManager(mockkeep) as blockmanager:
690             bufferblock = blockmanager.alloc_bufferblock()
691             bufferblock.owner = mock.MagicMock()
692             def flush(sync=None):
693                 blockmanager.commit_bufferblock(bufferblock, sync)
694             bufferblock.owner.flush.side_effect = flush
695             bufferblock.append("foo")
696             with self.assertRaises(arvados.errors.KeepWriteError) as err:
697                 blockmanager.commit_all()
698             self.assertTrue(bufferblock.owner.flush.called)
699             self.assertEqual(str(err.exception), "Error writing some blocks: block acbd18db4cc2f85cedef654fccc4a4d8+3 raised KeepWriteError (fail)")
700             self.assertEqual(bufferblock.state(), arvados.arvfile._BufferBlock.ERROR)