12 from arvados._ranges import Range
13 from arvados.keep import KeepLocator
14 from arvados.collection import Collection, CollectionReader
15 from arvados.arvfile import ArvadosFile, ArvadosFileReader
17 import arvados_testutil as tutil
18 from test_stream import StreamFileReaderTestCase, StreamRetryTestMixin
20 class ArvadosFileWriterTestCase(unittest.TestCase):
21 class MockKeep(object):
22 def __init__(self, blocks):
25 def get(self, locator, num_retries=0):
26 self.requests.append(locator)
27 return self.blocks.get(locator)
28 def get_from_cache(self, locator):
29 self.requests.append(locator)
30 return self.blocks.get(locator)
31 def put(self, data, num_retries=None):
32 pdh = "%s+%i" % (hashlib.md5(data).hexdigest(), len(data))
33 self.blocks[pdh] = str(data)
36 class MockApi(object):
37 def __init__(self, b, r):
40 class MockCollections(object):
41 def __init__(self, b, r):
44 class Execute(object):
45 def __init__(self, r):
47 def execute(self, num_retries=None):
49 def create(self, ensure_unique_name=False, body=None):
51 raise Exception("Body %s does not match expectation %s" % (body, self.body))
52 return ArvadosFileWriterTestCase.MockApi.MockCollections.Execute(self.response)
53 def update(self, uuid=None, body=None):
54 return ArvadosFileWriterTestCase.MockApi.MockCollections.Execute(self.response)
55 def collections(self):
56 return ArvadosFileWriterTestCase.MockApi.MockCollections(self.body, self.response)
59 def test_truncate(self):
60 keep = ArvadosFileWriterTestCase.MockKeep({"781e5e245d69b566979b86e28d23f2c7+10": "0123456789"})
61 api = ArvadosFileWriterTestCase.MockApi({"name":"test_truncate",
62 "manifest_text":". 781e5e245d69b566979b86e28d23f2c7+10 0:8:count.txt\n"},
63 {"uuid":"zzzzz-4zz18-mockcollection0",
64 "manifest_text":". 781e5e245d69b566979b86e28d23f2c7+10 0:8:count.txt\n"})
65 with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n',
66 api_client=api, keep_client=keep) as c:
67 writer = c.open("count.txt", "r+")
68 self.assertEqual(writer.size(), 10)
69 self.assertEqual("0123456789", writer.read(12))
73 # Make sure reading off the end doesn't break
74 self.assertEqual("", writer.read(12))
76 self.assertEqual(writer.size(), 8)
77 writer.seek(0, os.SEEK_SET)
78 self.assertEqual("01234567", writer.read(12))
80 self.assertEqual(None, c.manifest_locator())
81 self.assertEqual(True, c.modified())
82 c.save_new("test_truncate")
83 self.assertEqual("zzzzz-4zz18-mockcollection0", c.manifest_locator())
84 self.assertEqual(False, c.modified())
86 def test_write_to_end(self):
87 keep = ArvadosFileWriterTestCase.MockKeep({"781e5e245d69b566979b86e28d23f2c7+10": "0123456789"})
88 api = ArvadosFileWriterTestCase.MockApi({"name":"test_append",
89 "manifest_text": ". 781e5e245d69b566979b86e28d23f2c7+10 acbd18db4cc2f85cedef654fccc4a4d8+3 0:13:count.txt\n"},
90 {"uuid":"zzzzz-4zz18-mockcollection0",
91 "manifest_text": ". 781e5e245d69b566979b86e28d23f2c7+10 acbd18db4cc2f85cedef654fccc4a4d8+3 0:13:count.txt\n"})
92 with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n',
93 api_client=api, keep_client=keep) as c:
94 writer = c.open("count.txt", "r+")
95 self.assertEqual(writer.size(), 10)
97 writer.seek(5, os.SEEK_SET)
98 self.assertEqual("56789", writer.read(8))
100 writer.seek(10, os.SEEK_SET)
102 self.assertEqual(writer.size(), 13)
104 writer.seek(5, os.SEEK_SET)
105 self.assertEqual("56789foo", writer.read(8))
107 self.assertEqual(None, c.manifest_locator())
108 self.assertEqual(True, c.modified())
109 self.assertEqual(None, keep.get("acbd18db4cc2f85cedef654fccc4a4d8+3"))
111 c.save_new("test_append")
112 self.assertEqual("zzzzz-4zz18-mockcollection0", c.manifest_locator())
113 self.assertEqual(False, c.modified())
114 self.assertEqual("foo", keep.get("acbd18db4cc2f85cedef654fccc4a4d8+3"))
117 def test_append(self):
118 keep = ArvadosFileWriterTestCase.MockKeep({"781e5e245d69b566979b86e28d23f2c7+10": "0123456789"})
119 c = Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n', keep_client=keep)
120 writer = c.open("count.txt", "a+")
121 self.assertEqual(writer.read(20), "0123456789")
122 writer.seek(0, os.SEEK_SET)
124 writer.write("hello")
125 self.assertEqual(writer.read(20), "0123456789hello")
126 writer.seek(0, os.SEEK_SET)
128 writer.write("world")
129 self.assertEqual(writer.read(20), "0123456789helloworld")
131 self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 fc5e038d38a57032085441e7fe7010b0+10 0:20:count.txt\n", c.portable_manifest_text())
133 def test_write_at_beginning(self):
134 keep = ArvadosFileWriterTestCase.MockKeep({"781e5e245d69b566979b86e28d23f2c7+10": "0123456789"})
135 with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n',
136 keep_client=keep) as c:
137 writer = c.open("count.txt", "r+")
138 self.assertEqual("0123456789", writer.readfrom(0, 13))
139 writer.seek(0, os.SEEK_SET)
141 self.assertEqual(writer.size(), 10)
142 self.assertEqual("foo3456789", writer.readfrom(0, 13))
143 self.assertEqual(". acbd18db4cc2f85cedef654fccc4a4d8+3 781e5e245d69b566979b86e28d23f2c7+10 0:3:count.txt 6:7:count.txt\n", c.portable_manifest_text())
145 def test_write_empty(self):
146 keep = ArvadosFileWriterTestCase.MockKeep({})
147 with Collection(keep_client=keep) as c:
148 writer = c.open("count.txt", "w")
149 self.assertEqual(writer.size(), 0)
150 self.assertEqual(". d41d8cd98f00b204e9800998ecf8427e+0 0:0:count.txt\n", c.portable_manifest_text())
152 def test_save_manifest_text(self):
153 keep = ArvadosFileWriterTestCase.MockKeep({})
154 with Collection(keep_client=keep) as c:
155 writer = c.open("count.txt", "w")
156 writer.write("0123456789")
157 self.assertEqual('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n', c.portable_manifest_text())
158 self.assertNotIn('781e5e245d69b566979b86e28d23f2c7+10', keep.blocks)
160 self.assertEqual('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n', c.save_new(create_collection_record=False))
161 self.assertIn('781e5e245d69b566979b86e28d23f2c7+10', keep.blocks)
163 def test_get_manifest_text_commits(self):
164 keep = ArvadosFileWriterTestCase.MockKeep({})
165 with Collection(keep_client=keep) as c:
166 writer = c.open("count.txt", "w")
167 writer.write("0123456789")
168 self.assertEqual('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n', c.portable_manifest_text())
169 self.assertNotIn('781e5e245d69b566979b86e28d23f2c7+10', keep.blocks)
170 self.assertEqual('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n', c.manifest_text())
171 self.assertIn('781e5e245d69b566979b86e28d23f2c7+10', keep.blocks)
174 def test_write_in_middle(self):
175 keep = ArvadosFileWriterTestCase.MockKeep({"781e5e245d69b566979b86e28d23f2c7+10": "0123456789"})
176 with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n',
177 keep_client=keep) as c:
178 writer = c.open("count.txt", "r+")
179 self.assertEqual("0123456789", writer.readfrom(0, 13))
180 writer.seek(3, os.SEEK_SET)
182 self.assertEqual(writer.size(), 10)
183 self.assertEqual("012foo6789", writer.readfrom(0, 13))
184 self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 acbd18db4cc2f85cedef654fccc4a4d8+3 0:3:count.txt 10:3:count.txt 6:4:count.txt\n", c.portable_manifest_text())
186 def test_write_at_end(self):
187 keep = ArvadosFileWriterTestCase.MockKeep({"781e5e245d69b566979b86e28d23f2c7+10": "0123456789"})
188 with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n',
189 keep_client=keep) as c:
190 writer = c.open("count.txt", "r+")
191 self.assertEqual("0123456789", writer.readfrom(0, 13))
192 writer.seek(7, os.SEEK_SET)
194 self.assertEqual(writer.size(), 10)
195 self.assertEqual("0123456foo", writer.readfrom(0, 13))
196 self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 acbd18db4cc2f85cedef654fccc4a4d8+3 0:7:count.txt 10:3:count.txt\n", c.portable_manifest_text())
198 def test_write_across_segment_boundary(self):
199 keep = ArvadosFileWriterTestCase.MockKeep({"781e5e245d69b566979b86e28d23f2c7+10": "0123456789"})
200 with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt 0:10:count.txt\n',
201 keep_client=keep) as c:
202 writer = c.open("count.txt", "r+")
203 self.assertEqual("012345678901234", writer.readfrom(0, 15))
204 writer.seek(7, os.SEEK_SET)
205 writer.write("foobar")
206 self.assertEqual(writer.size(), 20)
207 self.assertEqual("0123456foobar34", writer.readfrom(0, 15))
208 self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 3858f62230ac3c915f300c664312c63f+6 0:7:count.txt 10:6:count.txt 3:7:count.txt\n", c.portable_manifest_text())
210 def test_write_across_several_segments(self):
211 keep = ArvadosFileWriterTestCase.MockKeep({"781e5e245d69b566979b86e28d23f2c7+10": "0123456789"})
212 with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:4:count.txt 0:4:count.txt 0:4:count.txt',
213 keep_client=keep) as c:
214 writer = c.open("count.txt", "r+")
215 self.assertEqual("012301230123", writer.readfrom(0, 15))
216 writer.seek(2, os.SEEK_SET)
217 writer.write("abcdefg")
218 self.assertEqual(writer.size(), 12)
219 self.assertEqual("01abcdefg123", writer.readfrom(0, 15))
220 self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 7ac66c0f148de9519b8bd264312c4d64+7 0:2:count.txt 10:7:count.txt 1:3:count.txt\n", c.portable_manifest_text())
222 def test_write_large(self):
223 keep = ArvadosFileWriterTestCase.MockKeep({})
224 api = ArvadosFileWriterTestCase.MockApi({"name":"test_write_large",
225 "manifest_text": ". a5de24f4417cfba9d5825eadc2f4ca49+67108000 598cc1a4ccaef8ab6e4724d87e675d78+32892000 0:100000000:count.txt\n"},
226 {"uuid":"zzzzz-4zz18-mockcollection0",
227 "manifest_text": ". a5de24f4417cfba9d5825eadc2f4ca49+67108000 598cc1a4ccaef8ab6e4724d87e675d78+32892000 0:100000000:count.txt\n"})
228 with Collection('. ' + arvados.config.EMPTY_BLOCK_LOCATOR + ' 0:0:count.txt',
229 api_client=api, keep_client=keep) as c:
230 writer = c.open("count.txt", "r+")
231 text = "0123456789" * 100
232 for b in xrange(0, 100000):
234 self.assertEqual(writer.size(), 100000000)
236 self.assertEqual(None, c.manifest_locator())
237 self.assertEqual(True, c.modified())
238 c.save_new("test_write_large")
239 self.assertEqual("zzzzz-4zz18-mockcollection0", c.manifest_locator())
240 self.assertEqual(False, c.modified())
243 def test_large_write(self):
244 keep = ArvadosFileWriterTestCase.MockKeep({})
245 api = ArvadosFileWriterTestCase.MockApi({}, {})
246 with Collection('. ' + arvados.config.EMPTY_BLOCK_LOCATOR + ' 0:0:count.txt',
247 api_client=api, keep_client=keep) as c:
248 writer = c.open("count.txt", "r+")
249 self.assertEqual(writer.size(), 0)
253 text = "0123456789" * 9999999
255 self.assertEqual(writer.size(), 100000000)
257 self.assertEqual(c.manifest_text(), ". 781e5e245d69b566979b86e28d23f2c7+10 48dd23ea1645fd47d789804d71b5bb8e+67108864 77c57dc6ac5a10bb2205caaa73187994+32891126 0:100000000:count.txt\n")
259 def test_rewrite_on_empty_file(self):
260 keep = ArvadosFileWriterTestCase.MockKeep({})
261 with Collection('. ' + arvados.config.EMPTY_BLOCK_LOCATOR + ' 0:0:count.txt',
262 keep_client=keep) as c:
263 writer = c.open("count.txt", "r+")
264 for b in xrange(0, 10):
265 writer.seek(0, os.SEEK_SET)
266 writer.write("0123456789")
268 self.assertEqual(writer.size(), 10)
269 self.assertEqual("0123456789", writer.readfrom(0, 20))
270 self.assertEqual(". 7a08b07e84641703e5f2c836aa59a170+100 90:10:count.txt\n", c.portable_manifest_text())
272 self.assertEqual(writer.size(), 10)
273 self.assertEqual("0123456789", writer.readfrom(0, 20))
274 self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n", c.portable_manifest_text())
276 def test_rewrite_append_existing_file(self):
277 keep = ArvadosFileWriterTestCase.MockKeep({"781e5e245d69b566979b86e28d23f2c7+10": "0123456789"})
278 with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt',
279 keep_client=keep) as c:
280 writer = c.open("count.txt", "r+")
281 for b in xrange(0, 10):
282 writer.seek(10, os.SEEK_SET)
283 writer.write("abcdefghij")
285 self.assertEqual(writer.size(), 20)
286 self.assertEqual("0123456789abcdefghij", writer.readfrom(0, 20))
287 self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 ae5f43bab79cf0be33f025fa97ae7398+100 0:10:count.txt 100:10:count.txt\n", c.portable_manifest_text())
289 writer.arvadosfile.flush()
290 self.assertEqual(writer.size(), 20)
291 self.assertEqual("0123456789abcdefghij", writer.readfrom(0, 20))
292 self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 a925576942e94b2ef57a066101b48876+10 0:20:count.txt\n", c.portable_manifest_text())
294 def test_rewrite_over_existing_file(self):
295 keep = ArvadosFileWriterTestCase.MockKeep({"781e5e245d69b566979b86e28d23f2c7+10": "0123456789"})
296 with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt',
297 keep_client=keep) as c:
298 writer = c.open("count.txt", "r+")
299 for b in xrange(0, 10):
300 writer.seek(5, os.SEEK_SET)
301 writer.write("abcdefghij")
303 self.assertEqual(writer.size(), 15)
304 self.assertEqual("01234abcdefghij", writer.readfrom(0, 20))
305 self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 ae5f43bab79cf0be33f025fa97ae7398+100 0:5:count.txt 100:10:count.txt\n", c.portable_manifest_text())
307 writer.arvadosfile.flush()
309 self.assertEqual(writer.size(), 15)
310 self.assertEqual("01234abcdefghij", writer.readfrom(0, 20))
311 self.assertEqual(". 781e5e245d69b566979b86e28d23f2c7+10 a925576942e94b2ef57a066101b48876+10 0:5:count.txt 10:10:count.txt\n", c.portable_manifest_text())
313 def test_write_large_rewrite(self):
314 keep = ArvadosFileWriterTestCase.MockKeep({})
315 api = ArvadosFileWriterTestCase.MockApi({"name":"test_write_large",
316 "manifest_text": ". 37400a68af9abdd76ca5bf13e819e42a+32892003 a5de24f4417cfba9d5825eadc2f4ca49+67108000 32892000:3:count.txt 32892006:67107997:count.txt 0:32892000:count.txt\n"},
317 {"uuid":"zzzzz-4zz18-mockcollection0",
318 "manifest_text": ". 37400a68af9abdd76ca5bf13e819e42a+32892003 a5de24f4417cfba9d5825eadc2f4ca49+67108000 32892000:3:count.txt 32892006:67107997:count.txt 0:32892000:count.txt\n"})
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", "r+")
322 text = ''.join(["0123456789" for a in xrange(0, 100)])
323 for b in xrange(0, 100000):
325 writer.seek(0, os.SEEK_SET)
327 self.assertEqual(writer.size(), 100000000)
329 self.assertEqual(None, c.manifest_locator())
330 self.assertEqual(True, c.modified())
331 c.save_new("test_write_large")
332 self.assertEqual("zzzzz-4zz18-mockcollection0", c.manifest_locator())
333 self.assertEqual(False, c.modified())
335 def test_create(self):
336 keep = ArvadosFileWriterTestCase.MockKeep({})
337 api = ArvadosFileWriterTestCase.MockApi({"name":"test_create",
338 "manifest_text":". 2e9ec317e197819358fbc43afca7d837+8 0:8:count.txt\n"},
339 {"uuid":"zzzzz-4zz18-mockcollection0",
340 "manifest_text":". 2e9ec317e197819358fbc43afca7d837+8 0:8:count.txt\n"})
341 with Collection(api_client=api, keep_client=keep) as c:
342 writer = c.open("count.txt", "w+")
343 self.assertEqual(writer.size(), 0)
344 writer.write("01234567")
345 self.assertEqual(writer.size(), 8)
347 self.assertEqual(None, c.manifest_locator())
348 self.assertEqual(True, c.modified())
349 self.assertEqual(None, keep.get("2e9ec317e197819358fbc43afca7d837+8"))
350 c.save_new("test_create")
351 self.assertEqual("zzzzz-4zz18-mockcollection0", c.manifest_locator())
352 self.assertEqual(False, c.modified())
353 self.assertEqual("01234567", keep.get("2e9ec317e197819358fbc43afca7d837+8"))
356 def test_create_subdir(self):
357 keep = ArvadosFileWriterTestCase.MockKeep({})
358 api = ArvadosFileWriterTestCase.MockApi({"name":"test_create",
359 "manifest_text":"./foo/bar 2e9ec317e197819358fbc43afca7d837+8 0:8:count.txt\n"},
360 {"uuid":"zzzzz-4zz18-mockcollection0",
361 "manifest_text":"./foo/bar 2e9ec317e197819358fbc43afca7d837+8 0:8:count.txt\n"})
362 with Collection(api_client=api, keep_client=keep) as c:
363 self.assertIsNone(c.api_response())
364 writer = c.open("foo/bar/count.txt", "w+")
365 writer.write("01234567")
366 c.save_new("test_create")
367 self.assertEqual(c.api_response(), api.response)
369 def test_overwrite(self):
370 keep = ArvadosFileWriterTestCase.MockKeep({"781e5e245d69b566979b86e28d23f2c7+10": "0123456789"})
371 api = ArvadosFileWriterTestCase.MockApi({"name":"test_overwrite",
372 "manifest_text":". 2e9ec317e197819358fbc43afca7d837+8 0:8:count.txt\n"},
373 {"uuid":"zzzzz-4zz18-mockcollection0",
374 "manifest_text":". 2e9ec317e197819358fbc43afca7d837+8 0:8:count.txt\n"})
375 with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n',
376 api_client=api, keep_client=keep) as c:
377 writer = c.open("count.txt", "w+")
378 self.assertEqual(writer.size(), 0)
379 writer.write("01234567")
380 self.assertEqual(writer.size(), 8)
382 self.assertEqual(None, c.manifest_locator())
383 self.assertEqual(True, c.modified())
384 c.save_new("test_overwrite")
385 self.assertEqual("zzzzz-4zz18-mockcollection0", c.manifest_locator())
386 self.assertEqual(False, c.modified())
388 def test_file_not_found(self):
389 with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n') as c:
390 with self.assertRaises(IOError):
391 writer = c.open("nocount.txt", "r")
393 def test_cannot_open_directory(self):
394 with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count.txt\n') as c:
395 with self.assertRaises(IOError):
396 writer = c.open(".", "r")
398 def test_create_multiple(self):
399 keep = ArvadosFileWriterTestCase.MockKeep({})
400 api = ArvadosFileWriterTestCase.MockApi({"name":"test_create_multiple",
401 "manifest_text":". 2e9ec317e197819358fbc43afca7d837+8 e8dc4081b13434b45189a720b77b6818+8 0:8:count1.txt 8:8:count2.txt\n"},
402 {"uuid":"zzzzz-4zz18-mockcollection0",
403 "manifest_text":". 2e9ec317e197819358fbc43afca7d837+8 e8dc4081b13434b45189a720b77b6818+8 0:8:count1.txt 8:8:count2.txt\n"})
404 with Collection(api_client=api, keep_client=keep) as c:
405 w1 = c.open("count1.txt", "w")
406 w2 = c.open("count2.txt", "w")
409 self.assertEqual(w1.size(), 8)
410 self.assertEqual(w2.size(), 8)
412 self.assertEqual(None, c.manifest_locator())
413 self.assertEqual(True, c.modified())
414 self.assertEqual(None, keep.get("2e9ec317e197819358fbc43afca7d837+8"))
415 c.save_new("test_create_multiple")
416 self.assertEqual("zzzzz-4zz18-mockcollection0", c.manifest_locator())
417 self.assertEqual(False, c.modified())
418 self.assertEqual("01234567", keep.get("2e9ec317e197819358fbc43afca7d837+8"))
421 class ArvadosFileReaderTestCase(StreamFileReaderTestCase):
422 class MockParent(object):
423 class MockBlockMgr(object):
424 def __init__(self, blocks, nocache):
426 self.nocache = nocache
428 def block_prefetch(self, loc):
431 def get_block_contents(self, loc, num_retries=0, cache_only=False):
432 if self.nocache and cache_only:
434 return self.blocks[loc]
436 def __init__(self, blocks, nocache):
438 self.nocache = nocache
439 self.lock = arvados.arvfile.NoopLock()
441 def root_collection(self):
444 def _my_block_manager(self):
445 return ArvadosFileReaderTestCase.MockParent.MockBlockMgr(self.blocks, self.nocache)
448 def make_count_reader(self, nocache=False):
452 for d in ['01234', '34567', '67890']:
453 loc = '{}+{}'.format(hashlib.md5(d).hexdigest(), len(d))
455 stream.append(Range(loc, n, len(d)))
457 af = ArvadosFile(ArvadosFileReaderTestCase.MockParent(blocks, nocache), "count.txt", stream=stream, segments=[Range(1, 0, 3), Range(6, 3, 3), Range(11, 6, 3)])
458 return ArvadosFileReader(af)
460 def test_read_block_crossing_behavior(self):
461 # read() needs to return all the data requested if possible, even if it
462 # crosses uncached blocks: https://arvados.org/issues/5856
463 sfile = self.make_count_reader(nocache=True)
464 self.assertEqual('12345678', sfile.read(8))
466 def test_successive_reads(self):
467 # Override StreamFileReaderTestCase.test_successive_reads
468 sfile = self.make_count_reader(nocache=True)
469 self.assertEqual('1234', sfile.read(4))
470 self.assertEqual('5678', sfile.read(4))
471 self.assertEqual('9', sfile.read(4))
472 self.assertEqual('', sfile.read(4))
474 def test_tell_after_block_read(self):
475 # Override StreamFileReaderTestCase.test_tell_after_block_read
476 sfile = self.make_count_reader(nocache=True)
477 self.assertEqual('12345678', sfile.read(8))
478 self.assertEqual(8, sfile.tell())
480 def test_prefetch(self):
481 keep = ArvadosFileWriterTestCase.MockKeep({"2e9ec317e197819358fbc43afca7d837+8": "01234567", "e8dc4081b13434b45189a720b77b6818+8": "abcdefgh"})
482 with Collection(". 2e9ec317e197819358fbc43afca7d837+8 e8dc4081b13434b45189a720b77b6818+8 0:16:count.txt\n", keep_client=keep) as c:
483 r = c.open("count.txt", "r")
484 self.assertEqual("0123", r.read(4))
485 self.assertIn("2e9ec317e197819358fbc43afca7d837+8", keep.requests)
486 self.assertIn("e8dc4081b13434b45189a720b77b6818+8", keep.requests)
488 def test__eq__from_manifest(self):
489 with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt') as c1:
490 with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt') as c2:
491 self.assertTrue(c1["count1.txt"] == c2["count1.txt"])
492 self.assertFalse(c1["count1.txt"] != c2["count1.txt"])
494 def test__eq__from_writes(self):
495 with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt') as c1:
496 with Collection() as c2:
497 f = c2.open("count1.txt", "w")
498 f.write("0123456789")
500 self.assertTrue(c1["count1.txt"] == c2["count1.txt"])
501 self.assertFalse(c1["count1.txt"] != c2["count1.txt"])
503 def test__ne__(self):
504 with Collection('. 781e5e245d69b566979b86e28d23f2c7+10 0:10:count1.txt') as c1:
505 with Collection() as c2:
506 f = c2.open("count1.txt", "w")
507 f.write("1234567890")
509 self.assertTrue(c1["count1.txt"] != c2["count1.txt"])
510 self.assertFalse(c1["count1.txt"] == c2["count1.txt"])
513 class ArvadosFileReadTestCase(unittest.TestCase, StreamRetryTestMixin):
514 def reader_for(self, coll_name, **kwargs):
518 for d in self.manifest_for(coll_name).split():
521 segments.append(Range(n, n, k.size))
522 stream.append(Range(d, n, k.size))
527 blockmanager = arvados.arvfile._BlockManager(self.keep_client())
528 blockmanager.prefetch_enabled = False
529 col = Collection(keep_client=self.keep_client(), block_manager=blockmanager)
530 af = ArvadosFile(col, "test",
533 return ArvadosFileReader(af, **kwargs)
535 def read_for_test(self, reader, byte_count, **kwargs):
536 return reader.read(byte_count, **kwargs)
539 class ArvadosFileReadFromTestCase(ArvadosFileReadTestCase):
540 def read_for_test(self, reader, byte_count, **kwargs):
541 return reader.readfrom(0, byte_count, **kwargs)
544 class ArvadosFileReadAllTestCase(ArvadosFileReadTestCase):
545 def read_for_test(self, reader, byte_count, **kwargs):
546 return ''.join(reader.readall(**kwargs))
549 class ArvadosFileReadAllDecompressedTestCase(ArvadosFileReadTestCase):
550 def read_for_test(self, reader, byte_count, **kwargs):
551 return ''.join(reader.readall_decompressed(**kwargs))
554 class ArvadosFileReadlinesTestCase(ArvadosFileReadTestCase):
555 def read_for_test(self, reader, byte_count, **kwargs):
556 return ''.join(reader.readlines(**kwargs))
558 class BlockManagerTest(unittest.TestCase):
559 def test_bufferblock_append(self):
560 keep = ArvadosFileWriterTestCase.MockKeep({})
561 blockmanager = arvados.arvfile._BlockManager(keep)
562 bufferblock = blockmanager.alloc_bufferblock()
563 bufferblock.append("foo")
565 self.assertEqual(bufferblock.size(), 3)
566 self.assertEqual(bufferblock.buffer_view[0:3], "foo")
567 self.assertEqual(bufferblock.locator(), "acbd18db4cc2f85cedef654fccc4a4d8+3")
569 bufferblock.append("bar")
571 self.assertEqual(bufferblock.size(), 6)
572 self.assertEqual(bufferblock.buffer_view[0:6], "foobar")
573 self.assertEqual(bufferblock.locator(), "3858f62230ac3c915f300c664312c63f+6")
575 bufferblock.set_state(arvados.arvfile._BufferBlock.PENDING)
576 with self.assertRaises(arvados.errors.AssertionError):
577 bufferblock.append("bar")
579 def test_bufferblock_dup(self):
580 keep = ArvadosFileWriterTestCase.MockKeep({})
581 blockmanager = arvados.arvfile._BlockManager(keep)
582 bufferblock = blockmanager.alloc_bufferblock()
583 bufferblock.append("foo")
585 self.assertEqual(bufferblock.size(), 3)
586 self.assertEqual(bufferblock.buffer_view[0:3], "foo")
587 self.assertEqual(bufferblock.locator(), "acbd18db4cc2f85cedef654fccc4a4d8+3")
588 bufferblock.set_state(arvados.arvfile._BufferBlock.PENDING)
590 bufferblock2 = blockmanager.dup_block(bufferblock, None)
591 self.assertNotEqual(bufferblock.blockid, bufferblock2.blockid)
593 bufferblock2.append("bar")
595 self.assertEqual(bufferblock2.size(), 6)
596 self.assertEqual(bufferblock2.buffer_view[0:6], "foobar")
597 self.assertEqual(bufferblock2.locator(), "3858f62230ac3c915f300c664312c63f+6")
599 self.assertEqual(bufferblock.size(), 3)
600 self.assertEqual(bufferblock.buffer_view[0:3], "foo")
601 self.assertEqual(bufferblock.locator(), "acbd18db4cc2f85cedef654fccc4a4d8+3")
603 def test_bufferblock_get(self):
604 keep = ArvadosFileWriterTestCase.MockKeep({"781e5e245d69b566979b86e28d23f2c7+10": "0123456789"})
605 blockmanager = arvados.arvfile._BlockManager(keep)
606 bufferblock = blockmanager.alloc_bufferblock()
607 bufferblock.append("foo")
609 self.assertEqual(blockmanager.get_block_contents("781e5e245d69b566979b86e28d23f2c7+10", 1), "0123456789")
610 self.assertEqual(blockmanager.get_block_contents(bufferblock.blockid, 1), "foo")
612 def test_bufferblock_commit(self):
613 mockkeep = mock.MagicMock()
614 blockmanager = arvados.arvfile._BlockManager(mockkeep)
615 bufferblock = blockmanager.alloc_bufferblock()
616 bufferblock.owner = mock.MagicMock()
617 bufferblock.owner.flush.side_effect = lambda x: blockmanager.commit_bufferblock(bufferblock, False)
618 bufferblock.append("foo")
619 blockmanager.commit_all()
620 self.assertTrue(bufferblock.owner.flush.called)
621 self.assertTrue(mockkeep.put.called)
622 self.assertEqual(bufferblock.state(), arvados.arvfile._BufferBlock.COMMITTED)
623 self.assertIsNone(bufferblock.buffer_view)
626 def test_bufferblock_commit_with_error(self):
627 mockkeep = mock.MagicMock()
628 mockkeep.put.side_effect = arvados.errors.KeepWriteError("fail")
629 blockmanager = arvados.arvfile._BlockManager(mockkeep)
630 bufferblock = blockmanager.alloc_bufferblock()
631 bufferblock.owner = mock.MagicMock()
632 bufferblock.owner.flush.side_effect = lambda x: blockmanager.commit_bufferblock(bufferblock, False)
633 bufferblock.append("foo")
634 with self.assertRaises(arvados.errors.KeepWriteError) as err:
635 blockmanager.commit_all()
636 self.assertTrue(bufferblock.owner.flush.called)
637 self.assertEqual(str(err.exception), "Error writing some blocks: block acbd18db4cc2f85cedef654fccc4a4d8+3 raised KeepWriteError (fail)")
638 self.assertEqual(bufferblock.state(), arvados.arvfile._BufferBlock.PENDING)