Merge branch 'master' into 4823-python-sdk-writable-collection-api
[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 import ArvadosFile, ArvadosFileReader, Range, import_manifest, export_manifest
13
14 import arvados_testutil as tutil
15 from test_stream import StreamFileReaderTestCase
16
17 class ArvadosFileWriterTestCase(unittest.TestCase):
18     class MockKeep(object):
19         def __init__(self, blocks):
20             self.blocks = blocks
21             self.requests = []
22         def get(self, locator, num_retries=0, cache_only=False):
23             self.requests.append(locator)
24             return self.blocks.get(locator)
25         def put(self, data):
26             pdh = "%s+%i" % (hashlib.md5(data).hexdigest(), len(data))
27             self.blocks[pdh] = str(data)
28             return pdh
29
30     class MockApi(object):
31         def __init__(self, b, r):
32             self.b = b
33             self.r = r
34         class MockCollections(object):
35             def __init__(self, b, r):
36                 self.b = b
37                 self.r = r
38             class Execute(object):
39                 def __init__(self, r):
40                     self.r = r
41                 def execute(self, num_retries=None):
42                     return self.r
43             def create(self, ensure_unique_name=False, body=None):
44                 if body != self.b:
45                     raise Exception("Body %s does not match expectation %s" % (body, self.b))
46                 return ArvadosFileWriterTestCase.MockApi.MockCollections.Execute(self.r)
47             def update(self, uuid=None, body=None):
48                 return ArvadosFileWriterTestCase.MockApi.MockCollections.Execute(self.r)
49         def collections(self):
50             return ArvadosFileWriterTestCase.MockApi.MockCollections(self.b, self.r)
51
52
53     def test_truncate(self):
54         keep = ArvadosFileWriterTestCase.MockKeep({"781e5e245d69b566979b86e28d23f2c7+10": "0123456789"})
55         api = ArvadosFileWriterTestCase.MockApi({"name":"test_truncate",
56                                                  "manifest_text":". 781e5e245d69b566979b86e28d23f2c7+10 0:8:count.txt\n"},
57                                                 {"uuid":"zzzzz-4zz18-mockcollection0"})
58         with import_manifest('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n',
59                              api_client=api, keep=keep) as c:
60             writer = c.open("count.txt", "r+")
61             self.assertEqual(writer.size(), 10)
62             writer.seek(5)
63             self.assertEqual("56789", writer.read(8))
64             writer.truncate(8)
65             writer.seek(5, os.SEEK_SET)
66             self.assertEqual("567", writer.read(8))
67
68             self.assertEqual(None, c._manifest_locator)
69             self.assertEqual(True, c.modified())
70             c.save_as("test_truncate")
71             self.assertEqual("zzzzz-4zz18-mockcollection0", c._manifest_locator)
72             self.assertEqual(False, c.modified())
73
74     def test_append(self):
75         keep = ArvadosFileWriterTestCase.MockKeep({"781e5e245d69b566979b86e28d23f2c7+10": "0123456789"})
76         api = ArvadosFileWriterTestCase.MockApi({"name":"test_append",
77                                                  "manifest_text": ". 781e5e245d69b566979b86e28d23f2c7+10 acbd18db4cc2f85cedef654fccc4a4d8+3 0:13:count.txt\n"},
78                                                 {"uuid":"zzzzz-4zz18-mockcollection0"})
79         with import_manifest('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n',
80                              api_client=api, keep=keep) as c:
81             writer = c.open("count.txt", "r+")
82             writer.seek(5, os.SEEK_SET)
83             self.assertEqual("56789", writer.read(8))
84             writer.seek(10, os.SEEK_SET)
85             writer.write("foo")
86             self.assertEqual(writer.size(), 13)
87             writer.seek(5, os.SEEK_SET)
88             self.assertEqual("56789foo", writer.read(8))
89
90             self.assertEqual(None, c._manifest_locator)
91             self.assertEqual(True, c.modified())
92             self.assertEqual(None, keep.get("acbd18db4cc2f85cedef654fccc4a4d8+3"))
93             c.save_as("test_append")
94             self.assertEqual("zzzzz-4zz18-mockcollection0", c._manifest_locator)
95             self.assertEqual(False, c.modified())
96             self.assertEqual("foo", keep.get("acbd18db4cc2f85cedef654fccc4a4d8+3"))
97
98     def test_write0(self):
99         keep = ArvadosFileWriterTestCase.MockKeep({"781e5e245d69b566979b86e28d23f2c7+10": "0123456789"})
100         with import_manifest('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n',
101                              keep=keep) as c:
102             writer = c.open("count.txt", "r+")
103             self.assertEqual("0123456789", writer.readfrom(0, 13))
104             writer.seek(0, os.SEEK_SET)
105             writer.write("foo")
106             self.assertEqual(writer.size(), 10)
107             self.assertEqual("foo3456789", writer.readfrom(0, 13))
108             self.assertEqual(". acbd18db4cc2f85cedef654fccc4a4d8+3 781e5e245d69b566979b86e28d23f2c7+10 0:3:count.txt 6:7:count.txt\n", export_manifest(c))
109
110     def test_write1(self):
111         keep = ArvadosFileWriterTestCase.MockKeep({"781e5e245d69b566979b86e28d23f2c7+10": "0123456789"})
112         with import_manifest('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n',
113                              keep=keep) as c:
114             writer = c.open("count.txt", "r+")
115             self.assertEqual("0123456789", writer.readfrom(0, 13))
116             writer.seek(3, os.SEEK_SET)
117             writer.write("foo")
118             self.assertEqual(writer.size(), 10)
119             self.assertEqual("012foo6789", writer.readfrom(0, 13))
120             self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 acbd18db4cc2f85cedef654fccc4a4d8+3 0:3:count.txt 10:3:count.txt 6:4:count.txt\n", export_manifest(c))
121
122     def test_write2(self):
123         keep = ArvadosFileWriterTestCase.MockKeep({"781e5e245d69b566979b86e28d23f2c7+10": "0123456789"})
124         with import_manifest('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n',
125                              keep=keep) as c:
126             writer = c.open("count.txt", "r+")
127             self.assertEqual("0123456789", writer.readfrom(0, 13))
128             writer.seek(7, os.SEEK_SET)
129             writer.write("foo")
130             self.assertEqual(writer.size(), 10)
131             self.assertEqual("0123456foo", writer.readfrom(0, 13))
132             self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 acbd18db4cc2f85cedef654fccc4a4d8+3 0:7:count.txt 10:3:count.txt\n", export_manifest(c))
133
134     def test_write3(self):
135         keep = ArvadosFileWriterTestCase.MockKeep({"781e5e245d69b566979b86e28d23f2c7+10": "0123456789"})
136         with import_manifest('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt 0:10:count.txt\n',
137                              keep=keep) as c:
138             writer = c.open("count.txt", "r+")
139             self.assertEqual("012345678901234", writer.readfrom(0, 15))
140             writer.seek(7, os.SEEK_SET)
141             writer.write("foobar")
142             self.assertEqual(writer.size(), 20)
143             self.assertEqual("0123456foobar34", writer.readfrom(0, 15))
144             self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 3858f62230ac3c915f300c664312c63f+6 0:7:count.txt 10:6:count.txt 3:7:count.txt\n", export_manifest(c))
145
146     def test_write4(self):
147         keep = ArvadosFileWriterTestCase.MockKeep({"781e5e245d69b566979b86e28d23f2c7+10": "0123456789"})
148         with import_manifest('. 781e5e245d69b566979b86e28d23f2c7+10 0:4:count.txt 0:4:count.txt 0:4:count.txt',
149                              keep=keep) as c:
150             writer = c.open("count.txt", "r+")
151             self.assertEqual("012301230123", writer.readfrom(0, 15))
152             writer.seek(2, os.SEEK_SET)
153             writer.write("abcdefg")
154             self.assertEqual(writer.size(), 12)
155             self.assertEqual("01abcdefg123", writer.readfrom(0, 15))
156             self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 7ac66c0f148de9519b8bd264312c4d64+7 0:2:count.txt 10:7:count.txt 1:3:count.txt\n", export_manifest(c))
157
158     def test_write_large(self):
159         keep = ArvadosFileWriterTestCase.MockKeep({})
160         api = ArvadosFileWriterTestCase.MockApi({"name":"test_write_large",
161                                                  "manifest_text": ". a5de24f4417cfba9d5825eadc2f4ca49+67108000 598cc1a4ccaef8ab6e4724d87e675d78+32892000 0:100000000:count.txt\n"},
162                                                 {"uuid":"zzzzz-4zz18-mockcollection0"})
163         with import_manifest('. ' + arvados.config.EMPTY_BLOCK_LOCATOR + ' 0:0:count.txt',
164                              api_client=api, keep=keep) as c:
165             writer = c.open("count.txt", "r+")
166             text = ''.join(["0123456789" for a in xrange(0, 100)])
167             for b in xrange(0, 100000):
168                 writer.write(text)
169             self.assertEqual(writer.size(), 100000000)
170
171             self.assertEqual(None, c._manifest_locator)
172             self.assertEqual(True, c.modified())
173             c.save_as("test_write_large")
174             self.assertEqual("zzzzz-4zz18-mockcollection0", c._manifest_locator)
175             self.assertEqual(False, c.modified())
176
177     def test_write_rewrite0(self):
178         keep = ArvadosFileWriterTestCase.MockKeep({})
179         with import_manifest('. ' + arvados.config.EMPTY_BLOCK_LOCATOR + ' 0:0:count.txt',
180                              keep=keep) as c:
181             writer = c.open("count.txt", "r+")
182             for b in xrange(0, 10):
183                 writer.seek(0, os.SEEK_SET)
184                 writer.write("0123456789")
185             writer.arvadosfile._repack_writes()
186             self.assertEqual(writer.size(), 10)
187             self.assertEqual("0123456789", writer.readfrom(0, 20))
188             self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n", export_manifest(c))
189
190     def test_write_rewrite1(self):
191         keep = ArvadosFileWriterTestCase.MockKeep({"781e5e245d69b566979b86e28d23f2c7+10": "0123456789"})
192         with import_manifest('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt',
193                              keep=keep) as c:
194             writer = c.open("count.txt", "r+")
195             for b in xrange(0, 10):
196                 writer.seek(10, os.SEEK_SET)
197                 writer.write("abcdefghij")
198             writer.arvadosfile._repack_writes()
199             self.assertEqual(writer.size(), 20)
200             self.assertEqual("0123456789abcdefghij", writer.readfrom(0, 20))
201             self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 a925576942e94b2ef57a066101b48876+10 0:20:count.txt\n", export_manifest(c))
202
203     def test_write_rewrite2(self):
204         keep = ArvadosFileWriterTestCase.MockKeep({"781e5e245d69b566979b86e28d23f2c7+10": "0123456789"})
205         with import_manifest('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt',
206                              keep=keep) as c:
207             writer = c.open("count.txt", "r+")
208             for b in xrange(0, 10):
209                 writer.seek(5, os.SEEK_SET)
210                 writer.write("abcdefghij")
211             writer.arvadosfile._repack_writes()
212             self.assertEqual(writer.size(), 15)
213             self.assertEqual("01234abcdefghij", writer.readfrom(0, 20))
214             self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 a925576942e94b2ef57a066101b48876+10 0:5:count.txt 10:10:count.txt\n", export_manifest(c))
215
216     def test_write_large_rewrite0(self):
217         keep = ArvadosFileWriterTestCase.MockKeep({})
218         api = ArvadosFileWriterTestCase.MockApi({"name":"test_write_large",
219                                                  "manifest_text": ". 37400a68af9abdd76ca5bf13e819e42a+32892003 a5de24f4417cfba9d5825eadc2f4ca49+67108000 32892000:3:count.txt 32892006:67107997:count.txt 0:32892000:count.txt\n"},
220                                                 {"uuid":"zzzzz-4zz18-mockcollection0"})
221         with import_manifest('. ' + arvados.config.EMPTY_BLOCK_LOCATOR + ' 0:0:count.txt',
222                              api_client=api, keep=keep) as c:
223             writer = c.open("count.txt", "r+")
224             text = ''.join(["0123456789" for a in xrange(0, 100)])
225             for b in xrange(0, 100000):
226                 writer.write(text)
227             writer.seek(0, os.SEEK_SET)
228             writer.write("foo")
229             self.assertEqual(writer.size(), 100000000)
230
231             self.assertEqual(None, c._manifest_locator)
232             self.assertEqual(True, c.modified())
233             c.save_as("test_write_large")
234             self.assertEqual("zzzzz-4zz18-mockcollection0", c._manifest_locator)
235             self.assertEqual(False, c.modified())
236
237     def test_create(self):
238         keep = ArvadosFileWriterTestCase.MockKeep({})
239         api = ArvadosFileWriterTestCase.MockApi({"name":"test_create",
240                                                  "manifest_text":". 2e9ec317e197819358fbc43afca7d837+8 0:8:count.txt\n"},
241                                                 {"uuid":"zzzzz-4zz18-mockcollection0"})
242         with arvados.Collection(api_client=api, keep_client=keep) as c:
243             writer = c.open("count.txt", "w+")
244             self.assertEqual(writer.size(), 0)
245             writer.write("01234567")
246             self.assertEqual(writer.size(), 8)
247
248             self.assertEqual(None, c._manifest_locator)
249             self.assertEqual(True, c.modified())
250             self.assertEqual(None, keep.get("2e9ec317e197819358fbc43afca7d837+8"))
251             c.save_as("test_create")
252             self.assertEqual("zzzzz-4zz18-mockcollection0", c._manifest_locator)
253             self.assertEqual(False, c.modified())
254             self.assertEqual("01234567", keep.get("2e9ec317e197819358fbc43afca7d837+8"))
255
256
257     def test_create_subdir(self):
258         keep = ArvadosFileWriterTestCase.MockKeep({})
259         api = ArvadosFileWriterTestCase.MockApi({"name":"test_create",
260                                                  "manifest_text":"./foo/bar 2e9ec317e197819358fbc43afca7d837+8 0:8:count.txt\n"},
261                                                 {"uuid":"zzzzz-4zz18-mockcollection0"})
262         with arvados.Collection(api_client=api, keep_client=keep) as c:
263             writer = c.open("foo/bar/count.txt", "w+")
264             writer.write("01234567")
265             c.save_as("test_create")
266
267     def test_overwrite(self):
268         keep = ArvadosFileWriterTestCase.MockKeep({"781e5e245d69b566979b86e28d23f2c7+10": "0123456789"})
269         api = ArvadosFileWriterTestCase.MockApi({"name":"test_overwrite",
270                                                  "manifest_text":". 2e9ec317e197819358fbc43afca7d837+8 0:8:count.txt\n"},
271                                                 {"uuid":"zzzzz-4zz18-mockcollection0"})
272         with import_manifest('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n',
273                              api_client=api, keep=keep) as c:
274             writer = c.open("count.txt", "w+")
275             self.assertEqual(writer.size(), 0)
276             writer.write("01234567")
277             self.assertEqual(writer.size(), 8)
278
279             self.assertEqual(None, c._manifest_locator)
280             self.assertEqual(True, c.modified())
281             c.save_as("test_overwrite")
282             self.assertEqual("zzzzz-4zz18-mockcollection0", c._manifest_locator)
283             self.assertEqual(False, c.modified())
284
285     def test_file_not_found(self):
286         with import_manifest('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n') as c:
287             with self.assertRaises(IOError):
288                 writer = c.open("nocount.txt", "r")
289
290     def test_cannot_open_directory(self):
291         with import_manifest('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n') as c:
292             with self.assertRaises(IOError):
293                 writer = c.open(".", "r")
294
295     def test_create_multiple(self):
296         keep = ArvadosFileWriterTestCase.MockKeep({})
297         api = ArvadosFileWriterTestCase.MockApi({"name":"test_create_multiple",
298                                                  "manifest_text":". 2e9ec317e197819358fbc43afca7d837+8 e8dc4081b13434b45189a720b77b6818+8 0:8:count1.txt 8:8:count2.txt\n"},
299                                                 {"uuid":"zzzzz-4zz18-mockcollection0"})
300         with arvados.Collection(api_client=api, keep_client=keep) as c:
301             w1 = c.open("count1.txt", "w")
302             w2 = c.open("count2.txt", "w")
303             w1.write("01234567")
304             w2.write("abcdefgh")
305             self.assertEqual(w1.size(), 8)
306             self.assertEqual(w2.size(), 8)
307
308             self.assertEqual(None, c._manifest_locator)
309             self.assertEqual(True, c.modified())
310             self.assertEqual(None, keep.get("2e9ec317e197819358fbc43afca7d837+8"))
311             c.save_as("test_create_multiple")
312             self.assertEqual("zzzzz-4zz18-mockcollection0", c._manifest_locator)
313             self.assertEqual(False, c.modified())
314             self.assertEqual("01234567", keep.get("2e9ec317e197819358fbc43afca7d837+8"))
315
316     def test_remove(self):
317         with import_manifest('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt 0:10:count2.txt\n') as c:
318             self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt 0:10:count2.txt\n", export_manifest(c))
319             self.assertTrue("count1.txt" in c)
320             c.remove("count1.txt")
321             self.assertFalse("count1.txt" in c)
322             self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count2.txt\n", export_manifest(c))
323
324     def test_remove_in_subdir(self):
325         with import_manifest('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n./foo 781e5e245d69b566979b86e28d23f2c7+10 0:10:count2.txt\n') as c:
326             c.remove("foo/count2.txt")
327             self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n", export_manifest(c))
328
329     def test_remove_subdir(self):
330         with import_manifest('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n./foo 781e5e245d69b566979b86e28d23f2c7+10 0:10:count2.txt\n') as c:
331             c.remove("foo")
332             self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt\n", export_manifest(c))
333
334     def test_prefetch(self):
335         keep = ArvadosFileWriterTestCase.MockKeep({"2e9ec317e197819358fbc43afca7d837+8": "01234567", "e8dc4081b13434b45189a720b77b6818+8": "abcdefgh"})
336         with import_manifest(". 2e9ec317e197819358fbc43afca7d837+8 e8dc4081b13434b45189a720b77b6818+8 0:16:count.txt\n", keep=keep) as c:
337             r = c.open("count.txt", "r")
338             self.assertEqual("0123", r.read(4))
339         self.assertEqual(["2e9ec317e197819358fbc43afca7d837+8", "2e9ec317e197819358fbc43afca7d837+8", "e8dc4081b13434b45189a720b77b6818+8"], keep.requests)
340
341
342 class ArvadosFileReaderTestCase(StreamFileReaderTestCase):
343     class MockParent(object):
344         class MockBlockMgr(object):
345             def __init__(self, blocks, nocache):
346                 self.blocks = blocks
347                 self.nocache = nocache
348
349             def block_prefetch(self, loc):
350                 pass
351
352             def get_block(self, loc, num_retries=0, cache_only=False):
353                 if self.nocache and cache_only:
354                     return None
355                 return self.blocks[loc]
356
357         def __init__(self, blocks, nocache):
358             self.blocks = blocks
359             self.nocache = nocache
360
361         def _my_block_manager(self):
362             return ArvadosFileReaderTestCase.MockParent.MockBlockMgr(self.blocks, self.nocache)
363
364     def make_count_reader(self, nocache=False):
365         stream = []
366         n = 0
367         blocks = {}
368         for d in ['01234', '34567', '67890']:
369             loc = '{}+{}'.format(hashlib.md5(d).hexdigest(), len(d))
370             blocks[loc] = d
371             stream.append(Range(loc, n, len(d)))
372             n += len(d)
373         af = ArvadosFile(ArvadosFileReaderTestCase.MockParent(blocks, nocache), stream=stream, segments=[Range(1, 0, 3), Range(6, 3, 3), Range(11, 6, 3)])
374         return ArvadosFileReader(af, "count.txt")
375
376     def test_read_returns_first_block(self):
377         # read() calls will be aligned on block boundaries - see #3663.
378         sfile = self.make_count_reader(nocache=True)
379         self.assertEqual('123', sfile.read(10))
380
381     def test_successive_reads(self):
382         sfile = self.make_count_reader(nocache=True)
383         for expect in ['123', '456', '789', '']:
384             self.assertEqual(expect, sfile.read(10))
385
386     def test_tell_after_block_read(self):
387         sfile = self.make_count_reader(nocache=True)
388         sfile.read(5)
389         self.assertEqual(3, sfile.tell())
390
391 # class StreamReaderTestCase(unittest.TestCase, StreamRetryTestMixin):
392 #     def reader_for(self, coll_name, **kwargs):
393 #         return StreamReader(self.manifest_for(coll_name).split(),
394 #                             self.keep_client(), **kwargs)
395
396 #     def read_for_test(self, reader, byte_count, **kwargs):
397 #         return reader.readfrom(0, byte_count, **kwargs)
398
399 #     def test_manifest_text_without_keep_client(self):
400 #         mtext = self.manifest_for('multilevel_collection_1')
401 #         for line in mtext.rstrip('\n').split('\n'):
402 #             reader = StreamReader(line.split())
403 #             self.assertEqual(line + '\n', reader.manifest_text())
404
405
406 # class StreamFileReadTestCase(unittest.TestCase, StreamRetryTestMixin):
407 #     def reader_for(self, coll_name, **kwargs):
408 #         return StreamReader(self.manifest_for(coll_name).split(),
409 #                             self.keep_client(), **kwargs).all_files()[0]
410
411 #     def read_for_test(self, reader, byte_count, **kwargs):
412 #         return reader.read(byte_count, **kwargs)
413
414
415 # class StreamFileReadFromTestCase(StreamFileReadTestCase):
416 #     def read_for_test(self, reader, byte_count, **kwargs):
417 #         return reader.readfrom(0, byte_count, **kwargs)
418
419
420 # class StreamFileReadAllTestCase(StreamFileReadTestCase):
421 #     def read_for_test(self, reader, byte_count, **kwargs):
422 #         return ''.join(reader.readall(**kwargs))
423
424
425 # class StreamFileReadAllDecompressedTestCase(StreamFileReadTestCase):
426 #     def read_for_test(self, reader, byte_count, **kwargs):
427 #         return ''.join(reader.readall_decompressed(**kwargs))
428
429
430 # class StreamFileReadlinesTestCase(StreamFileReadTestCase):
431 #     def read_for_test(self, reader, byte_count, **kwargs):
432 #         return ''.join(reader.readlines(**kwargs))