Chromium Code Reviews
chromiumcodereview-hr@appspot.gserviceaccount.com (chromiumcodereview-hr) | Please choose your nickname with Settings | Help | Chromium Project | Gerrit Changes | Sign out
(357)

Side by Side Diff: tests/gstools_unittest.py

Issue 12042069: Scripts to download files from google storage based on sha1 sums (Closed) Base URL: https://chromium.googlesource.com/chromium/tools/depot_tools.git@master
Patch Set: Review fixes Created 7 years, 9 months ago
Use n/p to move between diff chunks; N/P to move between comments. Draft comments are only viewable by you.
Jump to:
View unified diff | Download patch
OLDNEW
(Empty)
1 #!/usr/bin/env python
2 # Copyright (c) 2012 The Chromium Authors. All rights reserved.
3 # Use of this source code is governed by a BSD-style license that can be
4 # found in the LICENSE file.
5
6 """Unit tests for gstools.py and download_to/upload_from_google_storage.py."""
7
8 import os
9 import sys
10 import unittest
11 import threading
12 import StringIO
13 import Queue
14 import optparse
15
16 sys.path.insert(0, os.path.dirname(os.path.dirname(os.path.abspath(__file__))))
17
18 import gstools
19 import upload_to_google_storage
20 import download_from_google_storage
21
22 # ../third_party/gsutil/gsutil
23 GSUTIL_DEFAULT_PATH = os.path.join(
24 os.path.dirname(os.path.dirname(os.path.abspath(__file__))),
25 'third_party', 'gsutil', 'gsutil')
26
27
28 class GsutilMock(object):
29 def __init__(self, path, boto_path=None, timeout=None):
30 self.path = path
31 self.timeout = timeout
32 self.boto_path = boto_path
33 self.expected = []
34 self.history = []
35 self.lock = threading.Lock()
36
37 def add_expected(self, return_code, out, err):
38 self.expected.append((return_code, out, err))
39
40 def append_history(self, method, args):
41 with self.lock:
42 self.history.append((method, args))
43
44 def call(self, *args):
45 self.append_history('call', args)
46 if self.expected:
47 return self.expected.pop(0)[0]
48 else:
49 return 0
50
51 def check_call(self, *args):
52 self.append_history('check_call', args)
53 if self.expected:
54 return self.expected.pop(0)
55 else:
56 return (0, '', '')
57
58 def clone(self):
59 return self
60
61
62 class GstoolsUnitTests(unittest.TestCase):
63 def setUp(self):
64 self.base_path = os.path.join(
65 os.path.dirname(os.path.abspath(__file__)), 'gstools')
66
67 def test_gsutil(self):
68 gsutil = gstools.Gsutil(GSUTIL_DEFAULT_PATH)
69 self.assertEquals(gsutil.path, GSUTIL_DEFAULT_PATH)
70 code, _, err = gsutil.check_call()
71 self.assertEquals(code, 0)
72 self.assertEquals(err, '')
73
74 def test_gsutil_version(self):
75 gsutil = gstools.Gsutil(GSUTIL_DEFAULT_PATH)
76 _, _, err = gsutil.check_call('version')
77 err_lines = err.splitlines()
78 self.assertEquals(err_lines[0], 'gsutil version 3.25')
79 self.assertEquals(
80 err_lines[1],
81 'checksum 6ceb2cabffe2492167043c76fb067fd4 (OK)')
82
83 def test_get_sha1(self):
84 lorem_ipsum = os.path.join(self.base_path, 'lorem_ipsum.txt')
85 self.assertEquals(
86 gstools.GetSHA1(lorem_ipsum),
87 '7871c8e24da15bad8b0be2c36edc9dc77e37727f')
88
89 def test_get_md5(self):
90 lorem_ipsum = os.path.join(self.base_path, 'lorem_ipsum.txt')
91 lock = threading.Lock()
92 self.assertEquals(
93 gstools.GetMD5(lorem_ipsum, lock),
94 '634d7c1ed3545383837428f031840a1e')
95
96 def test_get_md5_cached_read(self):
97 lorem_ipsum = os.path.join(self.base_path, 'lorem_ipsum.txt')
98 lock = threading.Lock()
99 # Use a fake 'stale' MD5 sum. Expected behavior is to return stale sum.
100 self.assertEquals(
101 gstools.GetMD5Cached(lorem_ipsum, lock),
102 '734d7c1ed3545383837428f031840a1e')
103
104 def test_get_md5_cached_write(self):
105 lorem_ipsum2 = os.path.join(self.base_path, 'lorem_ipsum2.txt')
106 lorem_ipsum2_md5 = os.path.join(self.base_path, 'lorem_ipsum2.txt.md5')
107 if os.path.exists(lorem_ipsum2_md5):
108 os.remove(lorem_ipsum2_md5)
109 lock = threading.Lock()
110 # Use a fake 'stale' MD5 sum. Expected behavior is to return stale sum.
111 self.assertEquals(
112 gstools.GetMD5Cached(lorem_ipsum2, lock),
113 '4c02d1eb455a0f22c575265d17b84b6d')
114 self.assertTrue(os.path.exists(lorem_ipsum2_md5))
115 self.assertEquals(
116 open(lorem_ipsum2_md5, 'rb').read(),
117 '4c02d1eb455a0f22c575265d17b84b6d')
118 os.remove(lorem_ipsum2_md5) # Clean up.
119 self.assertFalse(os.path.exists(lorem_ipsum2_md5))
120
121
122 class UploadTests(unittest.TestCase):
123 def setUp(self):
124 self.gsutil = GsutilMock(GSUTIL_DEFAULT_PATH)
125 self.base_path = os.path.join(
126 os.path.dirname(os.path.abspath(__file__)), 'gstools')
127 self.base_url = 'gs://sometesturl'
128 self.parser = optparse.OptionParser()
129 self.ret_codes = Queue.Queue()
130 self.lorem_ipsum = os.path.join(self.base_path, 'lorem_ipsum.txt')
131 self.lorem_ipsum_sha1 = '7871c8e24da15bad8b0be2c36edc9dc77e37727f'
132
133 def test_upload_single_file(self):
134 filenames = [self.lorem_ipsum]
135 output_filename = '%s.sha1' % self.lorem_ipsum
136 if os.path.exists(output_filename):
137 os.remove(output_filename)
138 upload_to_google_storage.upload_to_google_storage(
139 filenames, self.base_url, self.gsutil, True, False, 1, False)
140 self.assertEquals(
141 self.gsutil.history,
142 [('check_call',
143 ('ls', '%s/%s' % (self.base_url, self.lorem_ipsum_sha1))),
144 ('check_call',
145 ('cp', '-q', filenames[0], '%s/%s' % (self.base_url,
146 self.lorem_ipsum_sha1)))])
147 self.assertTrue(os.path.exists(output_filename))
148 self.assertEquals(
149 open(output_filename, 'rb').read(),
150 '7871c8e24da15bad8b0be2c36edc9dc77e37727f')
151 os.remove(output_filename)
152
153 def test_upload_single_file_remote_exists(self):
154 filenames = [self.lorem_ipsum]
155 output_filename = '%s.sha1' % self.lorem_ipsum
156 etag_string = 'ETag: 634d7c1ed3545383837428f031840a1e'
157 if os.path.exists(output_filename):
158 os.remove(output_filename)
159 self.gsutil.add_expected(0, '', '')
160 self.gsutil.add_expected(0, etag_string, '')
161 upload_to_google_storage.upload_to_google_storage(
162 filenames, self.base_url, self.gsutil, False, False, 1, False)
163 self.assertEquals(
164 self.gsutil.history,
165 [('check_call',
166 ('ls', '%s/%s' % (self.base_url, self.lorem_ipsum_sha1))),
167 ('check_call',
168 ('ls', '-L', '%s/%s' % (self.base_url, self.lorem_ipsum_sha1)))])
169 self.assertTrue(os.path.exists(output_filename))
170 self.assertEquals(
171 open(output_filename, 'rb').read(),
172 '7871c8e24da15bad8b0be2c36edc9dc77e37727f')
173 os.remove(output_filename)
174
175 def test_upload_worker_errors(self):
176 work_queue = Queue.Queue()
177 work_queue.put((self.lorem_ipsum, self.lorem_ipsum_sha1))
178 work_queue.put((None, None))
179 self.gsutil.add_expected(1, '', '') # For the first ls call.
180 self.gsutil.add_expected(20, '', 'Expected error message')
181 # pylint: disable=W0212
182 upload_to_google_storage._upload_worker(
183 0,
184 work_queue,
185 self.base_url,
186 self.gsutil,
187 threading.Lock(),
188 False,
189 False,
190 self.ret_codes)
191 expected_ret_codes = [
192 (20,
193 'Encountered error on uploading %s to %s/%s\nExpected error message' %
194 (self.lorem_ipsum, self.base_url, self.lorem_ipsum_sha1))]
195 self.assertEquals(list(self.ret_codes.queue), expected_ret_codes)
196
197
198 def test_skip_hashing(self):
199 filenames = [self.lorem_ipsum]
200 output_filename = '%s.sha1' % self.lorem_ipsum
201 fake_hash = '6871c8e24da15bad8b0be2c36edc9dc77e37727f'
202 with open(output_filename, 'wb') as f:
203 f.write(fake_hash) # Fake hash.
204 upload_to_google_storage.upload_to_google_storage(
205 filenames, self.base_url, self.gsutil, False, False, 1, True)
206 self.assertEquals(
207 self.gsutil.history,
208 [('check_call',
209 ('ls', '%s/%s' % (self.base_url, fake_hash))),
210 ('check_call',
211 ('ls', '-L', '%s/%s' % (self.base_url, fake_hash))),
212 ('check_call',
213 ('cp', '-q', filenames[0], '%s/%s' % (self.base_url, fake_hash)))])
214 self.assertEquals(
215 open(output_filename, 'rb').read(), fake_hash)
216 os.remove(output_filename)
217
218 def test_get_targets_no_args(self):
219 try:
220 upload_to_google_storage.get_targets([], self.parser, False)
221 except SystemExit, e:
222 self.assertEquals(type(e), type(SystemExit()))
223 self.assertEquals(e.code, 2)
224 except Exception, e:
225 self.fail('unexpected exception: %s' % e)
226 else:
227 self.fail('SystemExit exception expected')
228
229 def test_get_targets_passthrough(self):
230 result = upload_to_google_storage.get_targets(
231 ['a', 'b', 'c', 'd', 'e'],
232 self.parser,
233 False)
234 self.assertEquals(result, ['a', 'b', 'c', 'd', 'e'])
235
236 def test_get_targets_multiple_stdin(self):
237 inputs = ['a', 'b', 'c', 'd', 'e']
238 sys.stdin = StringIO.StringIO(os.linesep.join(inputs))
239 result = upload_to_google_storage.get_targets(
240 ['-'],
241 self.parser,
242 False)
243 self.assertEquals(result, inputs)
244
245 def test_get_targets_multiple_stdin_null(self):
246 inputs = ['a', 'b', 'c', 'd', 'e']
247 sys.stdin = StringIO.StringIO('\0'.join(inputs))
248 result = upload_to_google_storage.get_targets(
249 ['-'],
250 self.parser,
251 True)
252 self.assertEquals(result, inputs)
253
254
255 class DownloadTests(unittest.TestCase):
256 def setUp(self):
257 self.gsutil = GsutilMock(GSUTIL_DEFAULT_PATH)
258 self.base_path = os.path.join(
259 os.path.dirname(os.path.abspath(__file__)),
260 'gstools',
261 'download_test_data')
262 self.base_url = 'gs://sometesturl'
263 self.parser = optparse.OptionParser()
264 self.queue = Queue.Queue()
265 self.lorem_ipsum = os.path.join(self.base_path, 'lorem_ipsum.txt')
266 self.lorem_ipsum_sha1 = '7871c8e24da15bad8b0be2c36edc9dc77e37727f'
267 self.maxDiff = None
268
269 def test_enumerate_files_non_recursive(self):
270 queue_size = download_from_google_storage.enumerate_work_queue(
271 self.base_path, self.queue, True, False, False, None, False)
272 result = list(self.queue.queue)
273 expected_queue = [
274 ('e6c4fbd4fe7607f3e6ebf68b2ea4ef694da7b4fe',
275 os.path.join(self.base_path, 'rootfolder_text.txt')),
276 ('7871c8e24da15bad8b0be2c36edc9dc77e37727f',
277 os.path.join(self.base_path, 'uploaded_lorem_ipsum.txt'))]
278 for item in result:
279 self.assertTrue(item in expected_queue)
280 self.assertEquals(queue_size, 2)
281
282 def test_enumerate_files_recursive(self):
283 queue_size = download_from_google_storage.enumerate_work_queue(
284 self.base_path, self.queue, True, True, False, None, False)
285 expected_queue = [
286 ('e6c4fbd4fe7607f3e6ebf68b2ea4ef694da7b4fe',
287 os.path.join(self.base_path, 'rootfolder_text.txt')),
288 ('7871c8e24da15bad8b0be2c36edc9dc77e37727f',
289 os.path.join(self.base_path, 'uploaded_lorem_ipsum.txt')),
290 ('b5415aa0b64006a95c0c409182e628881d6d6463',
291 os.path.join(self.base_path, 'subfolder', 'subfolder_text.txt'))]
292 result = list(self.queue.queue)
293 for item in result:
294 self.assertTrue(item in expected_queue)
295 self.assertEquals(queue_size, 3)
296
297 def test_download_worker_single_file(self):
298 sha1_hash = '7871c8e24da15bad8b0be2c36edc9dc77e37727f'
299 input_filename = '%s/%s' % (self.base_url, sha1_hash)
300 output_filename = os.path.join(self.base_path, 'uploaded_lorem_ipsum.txt')
301 self.queue.put((sha1_hash, output_filename))
302 self.queue.put((None, None))
303 stdout_queue = Queue.Queue()
304 # pylint: disable=W0212
305 download_from_google_storage._downloader_worker_thread(
306 0, self.queue, False, self.base_url, self.gsutil, stdout_queue)
307 expected_calls = [
308 ('check_call',
309 ('ls', input_filename)),
310 ('check_call',
311 ('cp', '-q', input_filename, output_filename))]
312 expected_output = [
313 'Downloading %s to %s...' % (input_filename, output_filename),
314 'Thread 0 is done']
315 self.assertEquals(list(stdout_queue.queue), expected_output)
316 self.assertEquals(self.gsutil.history, expected_calls)
317
318 def test_download_worker_skips_file(self):
319 sha1_hash = 'e6c4fbd4fe7607f3e6ebf68b2ea4ef694da7b4fe'
320 output_filename = os.path.join(self.base_path, 'rootfolder_text.txt')
321 self.queue.put((sha1_hash, output_filename))
322 self.queue.put((None, None))
323 stdout_queue = Queue.Queue()
324 # pylint: disable=W0212
325 download_from_google_storage._downloader_worker_thread(
326 0, self.queue, False, self.base_url, self.gsutil, stdout_queue)
327 expected_output = [
328 'File %s exists and SHA1 sum (%s) matches. Skipping.' %
329 (output_filename, sha1_hash),
330 'Thread 0 is done'
331 ]
332 self.assertEquals(list(stdout_queue.queue), expected_output)
333 self.assertEquals(self.gsutil.history, [])
334
335 def test_download_worker_skips_not_found_file(self):
336 sha1_hash = '7871c8e24da15bad8b0be2c36edc9dc77e37727f'
337 input_filename = '%s/%s' % (self.base_url, sha1_hash)
338 output_filename = os.path.join(self.base_path, 'uploaded_lorem_ipsum.txt')
339 self.queue.put((sha1_hash, output_filename))
340 self.queue.put((None, None))
341 stdout_queue = Queue.Queue()
342 self.gsutil.add_expected(1, '', '') # Return error when 'ls' is called.
343 # pylint: disable=W0212
344 download_from_google_storage._downloader_worker_thread(
345 0, self.queue, False, self.base_url, self.gsutil, stdout_queue)
346 expected_output = [
347 'File %s for %s does not exist, skipping.' % (
348 input_filename, output_filename),
349 'Thread 0 is done'
350 ]
351 expected_calls = [
352 ('check_call',
353 ('ls', input_filename))
354 ]
355 self.assertEquals(list(stdout_queue.queue), expected_output)
356 self.assertEquals(self.gsutil.history, expected_calls)
357
358 def test_download_directory_no_recursive_non_force(self):
359 sha1_hash = '7871c8e24da15bad8b0be2c36edc9dc77e37727f'
360 input_filename = '%s/%s' % (self.base_url, sha1_hash)
361 output_filename = os.path.join(self.base_path, 'uploaded_lorem_ipsum.txt')
362 download_from_google_storage.download_from_google_storage(
363 input_filename=self.base_path,
364 base_url=self.base_url,
365 gsutil=self.gsutil,
366 num_threads=1,
367 directory=True,
368 recursive=False,
369 force=False,
370 output=None,
371 ignore_errors=False,
372 sha1_file=False)
373 expected_calls = [
374 ('check_call',
375 ('ls', input_filename)),
376 ('check_call',
377 ('cp', '-q', input_filename, output_filename))]
378 self.assertEquals(self.gsutil.history, expected_calls)
379
380
381 if __name__ == '__main__':
382 unittest.main()
OLDNEW
« gstools.py ('K') | « tests/gstools/lorem_ipsum2.txt ('k') | upload_to_google_storage.py » ('j') | no next file with comments »

Powered by Google App Engine
This is Rietveld 408576698