1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
|
#!/usr/bin/python
# Copyright (c) 2011 The Chromium Authors. All rights reserved.
# Use of this source code is governed by a BSD-style license that can be
# found in the LICENSE file.
"""Snapshot Build Bisect Tool
This script bisects a snapshot archive using binary search. It starts at
a bad revision (it will try to guess HEAD) and asks for a last known-good
revision. It will then binary search across this revision range by downloading,
unzipping, and opening Chromium for you. After testing the specific revision,
it will ask you whether it is good or bad before continuing the search.
"""
# The root URL for storage.
BASE_URL = 'http://commondatastorage.googleapis.com/chromium-browser-snapshots'
# URL to the ViewVC commit page.
BUILD_VIEWVC_URL = 'http://src.chromium.org/viewvc/chrome?view=rev&revision=%d'
# Changelogs URL.
CHANGELOG_URL = 'http://build.chromium.org/f/chromium/' \
'perf/dashboard/ui/changelog.html?url=/trunk/src&range=%d:%d'
###############################################################################
import math
import optparse
import os
import pipes
import re
import shutil
import sys
import tempfile
import urllib
from xml.etree import ElementTree
import zipfile
class PathContext(object):
"""A PathContext is used to carry the information used to construct URLs and
paths when dealing with the storage server and archives."""
def __init__(self, platform, good_revision, bad_revision):
super(PathContext, self).__init__()
# Store off the input parameters.
self.platform = platform # What's passed in to the '-a/--archive' option.
self.good_revision = good_revision
self.bad_revision = bad_revision
# The name of the ZIP file in a revision directory on the server.
self.archive_name = None
# Set some internal members:
# _listing_platform_dir = Directory that holds revisions. Ends with a '/'.
# _archive_extract_dir = Uncompressed directory in the archive_name file.
# _binary_name = The name of the executable to run.
if self.platform == 'linux' or self.platform == 'linux64':
self._listing_platform_dir = 'Linux/'
self.archive_name = 'chrome-linux.zip'
self._archive_extract_dir = 'chrome-linux'
self._binary_name = 'chrome'
# Linux and x64 share all the same path data except for the archive dir.
if self.platform == 'linux64':
self._listing_platform_dir = 'Linux_x64/'
elif self.platform == 'mac':
self._listing_platform_dir = 'Mac/'
self.archive_name = 'chrome-mac.zip'
self._archive_extract_dir = 'chrome-mac'
self._binary_name = 'Chromium.app/Contents/MacOS/Chromium'
elif self.platform == 'win':
self._listing_platform_dir = 'Win/'
self.archive_name = 'chrome-win32.zip'
self._archive_extract_dir = 'chrome-win32'
self._binary_name = 'chrome.exe'
else:
raise Exception("Invalid platform")
def GetListingURL(self, marker=None):
"""Returns the URL for a directory listing, with an optional marker."""
marker_param = ''
if marker:
marker_param = '&marker=' + str(marker)
return BASE_URL + '/?delimiter=/&prefix=' + self._listing_platform_dir + \
marker_param
def GetDownloadURL(self, revision):
"""Gets the download URL for a build archive of a specific revision."""
return "%s/%s%d/%s" % (
BASE_URL, self._listing_platform_dir, revision, self.archive_name)
def GetLastChangeURL(self):
"""Returns a URL to the LAST_CHANGE file."""
return BASE_URL + '/' + self._listing_platform_dir + 'LAST_CHANGE'
def GetLaunchPath(self):
"""Returns a relative path (presumably from the archive extraction location)
that is used to run the executable."""
return os.path.join(self._archive_extract_dir, self._binary_name)
def UnzipFilenameToDir(filename, dir):
"""Unzip |filename| to directory |dir|."""
zf = zipfile.ZipFile(filename)
# Make base.
pushd = os.getcwd()
try:
if not os.path.isdir(dir):
os.mkdir(dir)
os.chdir(dir)
# Extract files.
for info in zf.infolist():
name = info.filename
if name.endswith('/'): # dir
if not os.path.isdir(name):
os.makedirs(name)
else: # file
dir = os.path.dirname(name)
if not os.path.isdir(dir):
os.makedirs(dir)
out = open(name, 'wb')
out.write(zf.read(name))
out.close()
# Set permissions. Permission info in external_attr is shifted 16 bits.
os.chmod(name, info.external_attr >> 16L)
os.chdir(pushd)
except Exception, e:
print >>sys.stderr, e
sys.exit(1)
def ParseDirectoryIndex(context):
"""Parses the Google Storage directory listing into a list of revision
numbers. The range starts with context.good_revision and goes until the latest
revision."""
def _FetchAndParse(url):
"""Fetches a URL and returns a 2-Tuple of ([revisions], next-marker). If
next-marker is not None, then the listing is a partial listing and another
fetch should be performed with next-marker being the marker= GET
parameter."""
handle = urllib.urlopen(url)
document = ElementTree.parse(handle)
# All nodes in the tree are namespaced. Get the root's tag name to extract
# the namespace. Etree does namespaces as |{namespace}tag|.
root_tag = document.getroot().tag
end_ns_pos = root_tag.find('}')
if end_ns_pos == -1:
raise Exception("Could not locate end namespace for directory index")
namespace = root_tag[:end_ns_pos + 1]
# Find the prefix (_listing_platform_dir) and whether or not the list is
# truncated.
prefix_len = len(document.find(namespace + 'Prefix').text)
next_marker = None
is_truncated = document.find(namespace + 'IsTruncated')
if is_truncated is not None and is_truncated.text.lower() == 'true':
next_marker = document.find(namespace + 'NextMarker').text
# Get a list of all the revisions.
all_prefixes = document.findall(namespace + 'CommonPrefixes/' +
namespace + 'Prefix')
# The <Prefix> nodes have content of the form of
# |_listing_platform_dir/revision/|. Strip off the platform dir and the
# trailing slash to just have a number.
revisions = []
for prefix in all_prefixes:
revnum = prefix.text[prefix_len:-1]
try:
revnum = int(revnum)
revisions.append(revnum)
except ValueError:
pass
return (revisions, next_marker)
# Fetch the first list of revisions.
(revisions, next_marker) = _FetchAndParse(context.GetListingURL())
# If the result list was truncated, refetch with the next marker. Do this
# until an entire directory listing is done.
while next_marker:
(new_revisions, next_marker) = _FetchAndParse(
context.GetListingURL(next_marker))
revisions.extend(new_revisions)
return revisions
def GetRevList(context):
"""Gets the list of revision numbers between |good_revision| and
|bad_revision| of the |context|."""
# Download the revlist and filter for just the range between good and bad.
rev_range = range(context.good_revision, context.bad_revision)
revlist = map(int, ParseDirectoryIndex(context))
revlist = filter(lambda r: r in rev_range, revlist)
revlist.sort()
return revlist
def TryRevision(context, rev, profile, args):
"""Downloads revision |rev|, unzips it, and opens it for the user to test.
|profile| is the profile to use."""
# Do this in a temp dir so we don't collide with user files.
cwd = os.getcwd()
tempdir = tempfile.mkdtemp(prefix='bisect_tmp')
os.chdir(tempdir)
# Download the file.
download_url = context.GetDownloadURL(rev)
def _ReportHook(blocknum, blocksize, totalsize):
size = blocknum * blocksize
if totalsize == -1: # Total size not known.
progress = "Received %d bytes" % size
else:
size = min(totalsize, size)
progress = "Received %d of %d bytes, %.2f%%" % (
size, totalsize, 100.0 * size / totalsize)
# Send a \r to let all progress messages use just one line of output.
sys.stdout.write("\r" + progress)
sys.stdout.flush()
try:
print 'Fetching ' + download_url
urllib.urlretrieve(download_url, context.archive_name, _ReportHook)
print
except Exception, e:
print('Could not retrieve the download. Sorry.')
sys.exit(-1)
# Unzip the file.
print 'Unzipping ...'
UnzipFilenameToDir(context.archive_name, os.curdir)
# Tell the system to open the app.
args = ['--user-data-dir=%s' % profile] + args
flags = ' '.join(map(pipes.quote, args))
cmd = '%s %s' % (context.GetLaunchPath(), flags)
print 'Running %s' % cmd
os.system(cmd)
os.chdir(cwd)
print 'Cleaning temp dir ...'
try:
shutil.rmtree(tempdir, True)
except Exception, e:
pass
def AskIsGoodBuild(rev):
"""Ask the user whether build |rev| is good or bad."""
# Loop until we get a response that we can parse.
while True:
response = raw_input('\nBuild %d is [(g)ood/(b)ad]: ' % int(rev))
if response and response in ('g', 'b'):
return response == 'g'
def Bisect(good,
bad,
revlist,
context,
try_args=(),
profile='profile',
predicate=AskIsGoodBuild):
"""Tries to find the exact commit where a regression was introduced by
running a binary search on all archived builds in a given revision range.
@param good The index in revlist of the last known good revision.
@param bad The index in revlist of the first known bad revision.
@param revlist A list of chromium revision numbers to check.
@param context A PathContext object.
@param try_args A tuple of arguments to pass to the predicate function.
@param profile The user profile with which to run chromium.
@param predicate A predicate function which returns True iff the argument
chromium revision is good.
"""
last_known_good_rev = revlist[good]
first_known_bad_rev = revlist[bad]
# Binary search time!
while good < bad:
candidates = revlist[good:bad]
num_poss = len(candidates)
if num_poss > 10:
print('%d candidates. %d tries left.' %
(num_poss, round(math.log(num_poss, 2))))
else:
print('Candidates: %s' % revlist[good:bad])
# Cut the problem in half...
test = int((bad - good) / 2) + good
test_rev = revlist[test]
# Let the user give this rev a spin (in her own profile, if she wants).
TryRevision(context, test_rev, profile, try_args)
if predicate(test_rev):
last_known_good_rev = revlist[good]
good = test + 1
else:
bad = test
return (last_known_good_rev, first_known_bad_rev)
def main():
usage = ('%prog [options] [-- chromium-options]\n'
'Perform binary search on the snapshot builds.\n'
'\n'
'Tip: add "-- --no-first-run" to bypass the first run prompts.')
parser = optparse.OptionParser(usage=usage)
# Strangely, the default help output doesn't include the choice list.
choices = ['mac', 'win', 'linux', 'linux64']
# linux-chromiumos lacks a continuous archive http://crbug.com/78158
parser.add_option('-a', '--archive',
choices = choices,
help = 'The buildbot archive to bisect [%s].' %
'|'.join(choices))
parser.add_option('-b', '--bad', type = 'int',
help = 'The bad revision to bisect to.')
parser.add_option('-g', '--good', type = 'int',
help = 'The last known good revision to bisect from.')
parser.add_option('-p', '--profile', '--user-data-dir', type = 'str',
help = 'Profile to use; this will not reset every run. ' +
'Defaults to a clean profile.', default = 'profile')
(opts, args) = parser.parse_args()
if opts.archive is None:
print 'Error: missing required parameter: --archive'
print
parser.print_help()
return 1
if opts.bad and opts.good and (opts.good > opts.bad):
print ('The good revision (%d) must precede the bad revision (%d).\n' %
(opts.good, opts.bad))
parser.print_help()
return 1
# Create the context. Initialize 0 for the revisions as they are set below.
context = PathContext(opts.archive, 0, 0)
# Pick a starting point, try to get HEAD for this.
if opts.bad:
bad_rev = opts.bad
else:
bad_rev = 0
try:
# Location of the latest build revision number
nh = urllib.urlopen(context.GetLastChangeURL())
latest = int(nh.read())
nh.close()
bad_rev = raw_input('Bad revision [HEAD:%d]: ' % latest)
if (bad_rev == ''):
bad_rev = latest
bad_rev = int(bad_rev)
except Exception, e:
print('Could not determine latest revision. This could be bad...')
bad_rev = int(raw_input('Bad revision: '))
# Find out when we were good.
if opts.good:
good_rev = opts.good
else:
good_rev = 0
try:
good_rev = int(raw_input('Last known good [0]: '))
except Exception, e:
pass
# Set the input parameters now that they've been validated.
context.good_revision = good_rev
context.bad_revision = bad_rev
# Get a list of revisions to bisect across.
revlist = GetRevList(context)
if len(revlist) < 2: # Don't have enough builds to bisect
print 'We don\'t have enough builds to bisect. revlist: %s' % revlist
sys.exit(1)
# If we don't have a |good_rev|, set it to be the first revision possible.
if good_rev == 0:
good_rev = revlist[0]
# These are indexes of |revlist|.
good = 0
bad = len(revlist) - 1
(last_known_good_rev, first_known_bad_rev) = Bisect(
good, bad, revlist, context, args, opts.profile)
# We're done. Let the user know the results in an official manner.
print('You are probably looking for build %d.' % first_known_bad_rev)
print('CHANGELOG URL:')
print(CHANGELOG_URL % (last_known_good_rev, first_known_bad_rev))
print('Built at revision:')
print(BUILD_VIEWVC_URL % first_known_bad_rev)
if __name__ == '__main__':
sys.exit(main())
|