Chromium Code Reviews
chromiumcodereview-hr@appspot.gserviceaccount.com (chromiumcodereview-hr) | Please choose your nickname with Settings | Help | Chromium Project | Gerrit Changes | Sign out
(781)

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 changes, fixed race condition 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 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 upload_to_google_storage
19 import download_from_google_storage
20
21 # ../third_party/gsutil/gsutil
22 GSUTIL_DEFAULT_PATH = os.path.join(
23 os.path.dirname(os.path.dirname(os.path.abspath(__file__))),
24 'third_party', 'gsutil', 'gsutil')
25
26
27 class GsutilMock(object):
28 def __init__(self, path, boto_path=None, timeout=None):
29 self.path = path
30 self.timeout = timeout
31 self.boto_path = boto_path
32 self.expected = []
33 self.history = []
34 self.lock = threading.Lock()
35
36 def add_expected(self, return_code, out, err):
37 self.expected.append((return_code, out, err))
38
39 def append_history(self, method, args):
40 with self.lock:
41 self.history.append((method, args))
42
43 def call(self, *args):
44 self.append_history('call', args)
45 if self.expected:
46 return self.expected.pop(0)[0]
47 else:
48 return 0
49
50 def check_call(self, *args):
51 self.append_history('check_call', args)
52 if self.expected:
53 return self.expected.pop(0)
54 else:
55 return (0, '', '')
56
57 def clone(self):
58 return self
59
60
61 class GstoolsUnitTests(unittest.TestCase):
62 def setUp(self):
63 self.base_path = os.path.join(
64 os.path.dirname(os.path.abspath(__file__)), 'gstools')
65
66 def test_gsutil(self):
67 gsutil = download_from_google_storage.Gsutil(GSUTIL_DEFAULT_PATH)
68 self.assertEquals(gsutil.path, GSUTIL_DEFAULT_PATH)
69 code, _, err = gsutil.check_call()
70 self.assertEquals(code, 0)
71 self.assertEquals(err, '')
72
73 def test_gsutil_version(self):
74 gsutil = download_from_google_storage.Gsutil(GSUTIL_DEFAULT_PATH)
75 _, _, err = gsutil.check_call('version')
76 err_lines = err.splitlines()
77 self.assertEquals(err_lines[0], 'gsutil version 3.25')
78 self.assertEquals(
79 err_lines[1],
80 'checksum 79d281187a4cf35d3625357aaf4fbdae (OK)')
81
82 def test_get_sha1(self):
83 lorem_ipsum = os.path.join(self.base_path, 'lorem_ipsum.txt')
84 self.assertEquals(
85 download_from_google_storage.GetSHA1(lorem_ipsum),
86 '7871c8e24da15bad8b0be2c36edc9dc77e37727f')
87
88 def test_get_md5(self):
89 lorem_ipsum = os.path.join(self.base_path, 'lorem_ipsum.txt')
90 lock = threading.Lock()
91 self.assertEquals(
92 upload_to_google_storage.GetMD5(lorem_ipsum, lock),
93 '634d7c1ed3545383837428f031840a1e')
94
95 def test_get_md5_cached_read(self):
96 lorem_ipsum = os.path.join(self.base_path, 'lorem_ipsum.txt')
97 lock = threading.Lock()
98 # Use a fake 'stale' MD5 sum. Expected behavior is to return stale sum.
99 self.assertEquals(
100 upload_to_google_storage.GetMD5Cached(lorem_ipsum, lock),
101 '734d7c1ed3545383837428f031840a1e')
102
103 def test_get_md5_cached_write(self):
104 lorem_ipsum2 = os.path.join(self.base_path, 'lorem_ipsum2.txt')
105 lorem_ipsum2_md5 = os.path.join(self.base_path, 'lorem_ipsum2.txt.md5')
106 if os.path.exists(lorem_ipsum2_md5):
107 os.remove(lorem_ipsum2_md5)
108 lock = threading.Lock()
109 # Use a fake 'stale' MD5 sum. Expected behavior is to return stale sum.
110 self.assertEquals(
111 upload_to_google_storage.GetMD5Cached(lorem_ipsum2, lock),
112 '4c02d1eb455a0f22c575265d17b84b6d')
113 self.assertTrue(os.path.exists(lorem_ipsum2_md5))
114 self.assertEquals(
115 open(lorem_ipsum2_md5, 'rb').read(),
116 '4c02d1eb455a0f22c575265d17b84b6d')
117 os.remove(lorem_ipsum2_md5) # Clean up.
118 self.assertFalse(os.path.exists(lorem_ipsum2_md5))
119
120
121 class UploadTests(unittest.TestCase):
122 def setUp(self):
123 self.gsutil = GsutilMock(GSUTIL_DEFAULT_PATH)
124 self.base_path = os.path.join(
125 os.path.dirname(os.path.abspath(__file__)), 'gstools')
126 self.base_url = 'gs://sometesturl'
127 self.parser = optparse.OptionParser()
128 self.ret_codes = Queue.Queue()
129 self.lorem_ipsum = os.path.join(self.base_path, 'lorem_ipsum.txt')
130 self.lorem_ipsum_sha1 = '7871c8e24da15bad8b0be2c36edc9dc77e37727f'
131
132 def test_upload_single_file(self):
133 filenames = [self.lorem_ipsum]
134 output_filename = '%s.sha1' % self.lorem_ipsum
135 if os.path.exists(output_filename):
136 os.remove(output_filename)
137 upload_to_google_storage.upload_to_google_storage(
138 filenames, self.base_url, self.gsutil, True, False, 1, False)
139 self.assertEquals(
140 self.gsutil.history,
141 [('check_call',
142 ('ls', '%s/%s' % (self.base_url, self.lorem_ipsum_sha1))),
143 ('check_call',
144 ('cp', '-q', filenames[0], '%s/%s' % (self.base_url,
145 self.lorem_ipsum_sha1)))])
146 self.assertTrue(os.path.exists(output_filename))
147 self.assertEquals(
148 open(output_filename, 'rb').read(),
149 '7871c8e24da15bad8b0be2c36edc9dc77e37727f')
150 os.remove(output_filename)
151
152 def test_upload_single_file_remote_exists(self):
153 filenames = [self.lorem_ipsum]
154 output_filename = '%s.sha1' % self.lorem_ipsum
155 etag_string = 'ETag: 634d7c1ed3545383837428f031840a1e'
156 if os.path.exists(output_filename):
157 os.remove(output_filename)
158 self.gsutil.add_expected(0, '', '')
159 self.gsutil.add_expected(0, etag_string, '')
160 upload_to_google_storage.upload_to_google_storage(
161 filenames, self.base_url, self.gsutil, False, False, 1, False)
162 self.assertEquals(
163 self.gsutil.history,
164 [('check_call',
165 ('ls', '%s/%s' % (self.base_url, self.lorem_ipsum_sha1))),
166 ('check_call',
167 ('ls', '-L', '%s/%s' % (self.base_url, self.lorem_ipsum_sha1)))])
168 self.assertTrue(os.path.exists(output_filename))
169 self.assertEquals(
170 open(output_filename, 'rb').read(),
171 '7871c8e24da15bad8b0be2c36edc9dc77e37727f')
172 os.remove(output_filename)
173
174 def test_upload_worker_errors(self):
175 work_queue = Queue.Queue()
176 work_queue.put((self.lorem_ipsum, self.lorem_ipsum_sha1))
177 work_queue.put((None, None))
178 self.gsutil.add_expected(1, '', '') # For the first ls call.
179 self.gsutil.add_expected(20, '', 'Expected error message')
180 # pylint: disable=W0212
181 upload_to_google_storage._upload_worker(
182 0,
183 work_queue,
184 self.base_url,
185 self.gsutil,
186 threading.Lock(),
187 False,
188 False,
189 self.ret_codes)
190 expected_ret_codes = [
191 (20,
192 'Encountered error on uploading %s to %s/%s\nExpected error message' %
193 (self.lorem_ipsum, self.base_url, self.lorem_ipsum_sha1))]
194 self.assertEquals(list(self.ret_codes.queue), expected_ret_codes)
195
196
197 def test_skip_hashing(self):
198 filenames = [self.lorem_ipsum]
199 output_filename = '%s.sha1' % self.lorem_ipsum
200 fake_hash = '6871c8e24da15bad8b0be2c36edc9dc77e37727f'
201 with open(output_filename, 'wb') as f:
202 f.write(fake_hash) # Fake hash.
203 upload_to_google_storage.upload_to_google_storage(
204 filenames, self.base_url, self.gsutil, False, False, 1, True)
205 self.assertEquals(
206 self.gsutil.history,
207 [('check_call',
208 ('ls', '%s/%s' % (self.base_url, fake_hash))),
209 ('check_call',
210 ('ls', '-L', '%s/%s' % (self.base_url, fake_hash))),
211 ('check_call',
212 ('cp', '-q', filenames[0], '%s/%s' % (self.base_url, fake_hash)))])
213 self.assertEquals(
214 open(output_filename, 'rb').read(), fake_hash)
215 os.remove(output_filename)
216
217 def test_get_targets_no_args(self):
218 try:
219 upload_to_google_storage.get_targets([], self.parser, False)
220 except SystemExit, e:
221 self.assertEquals(type(e), type(SystemExit()))
222 self.assertEquals(e.code, 2)
223 except Exception, e:
224 self.fail('unexpected exception: %s' % e)
225 else:
226 self.fail('SystemExit exception expected')
227
228 def test_get_targets_passthrough(self):
229 result = upload_to_google_storage.get_targets(
230 ['a', 'b', 'c', 'd', 'e'],
231 self.parser,
232 False)
233 self.assertEquals(result, ['a', 'b', 'c', 'd', 'e'])
234
235 def test_get_targets_multiple_stdin(self):
236 inputs = ['a', 'b', 'c', 'd', 'e']
237 sys.stdin = StringIO.StringIO(os.linesep.join(inputs))
238 result = upload_to_google_storage.get_targets(
239 ['-'],
240 self.parser,
241 False)
242 self.assertEquals(result, inputs)
243
244 def test_get_targets_multiple_stdin_null(self):
245 inputs = ['a', 'b', 'c', 'd', 'e']
246 sys.stdin = StringIO.StringIO('\0'.join(inputs))
247 result = upload_to_google_storage.get_targets(
248 ['-'],
249 self.parser,
250 True)
251 self.assertEquals(result, inputs)
252
253
254 class DownloadTests(unittest.TestCase):
255 def setUp(self):
256 self.gsutil = GsutilMock(GSUTIL_DEFAULT_PATH)
257 self.base_path = os.path.join(
258 os.path.dirname(os.path.abspath(__file__)),
259 'gstools',
260 'download_test_data')
261 self.base_url = 'gs://sometesturl'
262 self.parser = optparse.OptionParser()
263 self.queue = Queue.Queue()
264 self.lorem_ipsum = os.path.join(self.base_path, 'lorem_ipsum.txt')
265 self.lorem_ipsum_sha1 = '7871c8e24da15bad8b0be2c36edc9dc77e37727f'
266 self.maxDiff = None
267
268 def test_enumerate_files_non_recursive(self):
269 queue_size = download_from_google_storage.enumerate_work_queue(
270 self.base_path, self.queue, True, False, False, None, False)
271 result = list(self.queue.queue)
272 expected_queue = [
273 ('e6c4fbd4fe7607f3e6ebf68b2ea4ef694da7b4fe',
274 os.path.join(self.base_path, 'rootfolder_text.txt')),
275 ('7871c8e24da15bad8b0be2c36edc9dc77e37727f',
276 os.path.join(self.base_path, 'uploaded_lorem_ipsum.txt'))]
277 for item in result:
278 self.assertTrue(item in expected_queue)
279 self.assertEquals(queue_size, 2)
280
281 def test_enumerate_files_recursive(self):
282 queue_size = download_from_google_storage.enumerate_work_queue(
283 self.base_path, self.queue, True, True, False, None, False)
284 expected_queue = [
285 ('e6c4fbd4fe7607f3e6ebf68b2ea4ef694da7b4fe',
286 os.path.join(self.base_path, 'rootfolder_text.txt')),
287 ('7871c8e24da15bad8b0be2c36edc9dc77e37727f',
288 os.path.join(self.base_path, 'uploaded_lorem_ipsum.txt')),
289 ('b5415aa0b64006a95c0c409182e628881d6d6463',
290 os.path.join(self.base_path, 'subfolder', 'subfolder_text.txt'))]
291 result = list(self.queue.queue)
292 for item in result:
293 self.assertTrue(item in expected_queue)
294 self.assertEquals(queue_size, 3)
295
296 def test_download_worker_single_file(self):
297 sha1_hash = '7871c8e24da15bad8b0be2c36edc9dc77e37727f'
298 input_filename = '%s/%s' % (self.base_url, sha1_hash)
299 output_filename = os.path.join(self.base_path, 'uploaded_lorem_ipsum.txt')
300 self.queue.put((sha1_hash, output_filename))
301 self.queue.put((None, None))
302 stdout_queue = Queue.Queue()
303 # pylint: disable=W0212
304 download_from_google_storage._downloader_worker_thread(
305 0, self.queue, False, self.base_url, self.gsutil, stdout_queue)
306 expected_calls = [
307 ('check_call',
308 ('ls', input_filename)),
309 ('check_call',
310 ('cp', '-q', input_filename, output_filename))]
311 expected_output = [
312 'Downloading %s to %s...' % (input_filename, output_filename),
313 'Thread 0 is done']
314 self.assertEquals(list(stdout_queue.queue), expected_output)
315 self.assertEquals(self.gsutil.history, expected_calls)
316
317 def test_download_worker_skips_file(self):
318 sha1_hash = 'e6c4fbd4fe7607f3e6ebf68b2ea4ef694da7b4fe'
319 output_filename = os.path.join(self.base_path, 'rootfolder_text.txt')
320 self.queue.put((sha1_hash, output_filename))
321 self.queue.put((None, None))
322 stdout_queue = Queue.Queue()
323 # pylint: disable=W0212
324 download_from_google_storage._downloader_worker_thread(
325 0, self.queue, False, self.base_url, self.gsutil, stdout_queue)
326 expected_output = [
327 'File %s exists and SHA1 sum (%s) matches. Skipping.' %
328 (output_filename, sha1_hash),
329 'Thread 0 is done'
330 ]
331 self.assertEquals(list(stdout_queue.queue), expected_output)
332 self.assertEquals(self.gsutil.history, [])
333
334 def test_download_worker_skips_not_found_file(self):
335 sha1_hash = '7871c8e24da15bad8b0be2c36edc9dc77e37727f'
336 input_filename = '%s/%s' % (self.base_url, sha1_hash)
337 output_filename = os.path.join(self.base_path, 'uploaded_lorem_ipsum.txt')
338 self.queue.put((sha1_hash, output_filename))
339 self.queue.put((None, None))
340 stdout_queue = Queue.Queue()
341 self.gsutil.add_expected(1, '', '') # Return error when 'ls' is called.
342 # pylint: disable=W0212
343 download_from_google_storage._downloader_worker_thread(
344 0, self.queue, False, self.base_url, self.gsutil, stdout_queue)
345 expected_output = [
346 'File %s for %s does not exist, skipping.' % (
347 input_filename, output_filename),
348 'Thread 0 is done'
349 ]
350 expected_calls = [
351 ('check_call',
352 ('ls', input_filename))
353 ]
354 self.assertEquals(list(stdout_queue.queue), expected_output)
355 self.assertEquals(self.gsutil.history, expected_calls)
356
357 def test_download_directory_no_recursive_non_force(self):
358 sha1_hash = '7871c8e24da15bad8b0be2c36edc9dc77e37727f'
359 input_filename = '%s/%s' % (self.base_url, sha1_hash)
360 output_filename = os.path.join(self.base_path, 'uploaded_lorem_ipsum.txt')
361 download_from_google_storage.download_from_google_storage(
362 input_filename=self.base_path,
363 base_url=self.base_url,
364 gsutil=self.gsutil,
365 num_threads=1,
366 directory=True,
367 recursive=False,
368 force=False,
369 output=None,
370 ignore_errors=False,
371 sha1_file=False)
372 expected_calls = [
373 ('check_call',
374 ('ls', input_filename)),
375 ('check_call',
376 ('cp', '-q', input_filename, output_filename))]
377 self.assertEquals(self.gsutil.history, expected_calls)
378
379
380 if __name__ == '__main__':
381 unittest.main()
OLDNEW

Powered by Google App Engine
This is Rietveld 408576698