2 # (Be in -*- python -*- mode.)
4 # ====================================================================
5 # Copyright (c) 2000-2009 CollabNet. All rights reserved.
7 # This software is licensed as described in the file COPYING, which
8 # you should have received as part of this distribution. The terms
9 # are also available at http://subversion.tigris.org/license-1.html.
10 # If newer versions of this license are posted there, you may use a
11 # newer version instead, at your option.
13 # This software consists of voluntary contributions made by many
14 # individuals. For exact contribution history, see the revision
15 # history and logs, available at http://cvs2svn.tigris.org/.
16 # ====================================================================
18 # The purpose of verify-cvs2svn is to verify the result of a cvs2svn
19 # repository conversion. The following tests are performed:
21 # 1. Content checking of the HEAD revision of trunk, all tags and all
22 # branches. Only the tags and branches in the Subversion
23 # repository are checked, i.e. there are no checks to verify that
24 # all tags and branches in the CVS repository are present.
26 # This program only works if you converted a subdirectory of a CVS
27 # repository, and not the whole repository. If you really did convert
28 # a whole repository and need to check it, you must create a CVSROOT
29 # directory above the current root using cvs init.
31 # ====================================================================
41 # CVS and Subversion command line client commands
48 """Run cmd as a pipe. Return (output, status)."""
49 child
= subprocess
.Popen(cmd
, stdout
=subprocess
.PIPE
)
50 output
= child
.stdout
.read()
52 return (output
, status
)
54 def cmd_failed(cmd
, output
, status
):
55 print 'CMD FAILED:', ' '.join(cmd
)
57 sys
.stdout
.write(output
)
58 raise RuntimeError('%s command failed!' % cmd
[0])
61 def __init__(self
, path
):
62 """Open the CVS repository at PATH."""
63 path
= os
.path
.abspath(path
)
64 if not os
.path
.isdir(path
):
65 raise RuntimeError('CVS path is not a directory')
67 if os
.path
.exists(os
.path
.join(path
, 'CVSROOT')):
68 # The whole repository
72 self
.cvsroot
= os
.path
.dirname(path
)
73 self
.module
= os
.path
.basename(path
)
74 while not os
.path
.exists(os
.path
.join(self
.cvsroot
, 'CVSROOT')):
75 parent
= os
.path
.dirname(self
.cvsroot
)
76 if parent
== self
.cvsroot
:
77 raise RuntimeError('Cannot find the CVSROOT')
78 self
.module
= os
.path
.join(os
.path
.basename(self
.cvsroot
), self
.module
)
82 return os
.path
.basename(self
.cvsroot
)
84 def export(self
, dest_path
, rev
=None, keyword_opt
=None):
85 """Export revision REV to DEST_PATH where REV can be None to export
86 the HEAD revision, or any valid CVS revision string to export that
89 cmd
= [ CVS_CMD
, '-Q', '-d', ':local:' + self
.cvsroot
, 'export' ]
91 cmd
.extend([ '-r', rev
])
93 cmd
.extend([ '-D', 'now' ])
95 cmd
.append(keyword_opt
)
96 cmd
.extend([ '-d', dest_path
, self
.module
])
97 (output
, status
) = pipe(cmd
)
99 cmd_failed(cmd
, output
, status
)
105 def __init__(self
, url
):
106 """Open the Subversion repository at URL."""
107 # Check if the user supplied an URL or a path
108 if url
.find('://') == -1:
109 abspath
= os
.path
.abspath(url
)
110 url
= 'file://' + (abspath
[0] != '/' and '/' or '') + abspath
112 url
= url
.replace(os
.sep
, '/')
116 # Cache a list of all tags and branches
119 self
.tag_list
= self
.list('tags')
122 if 'branches' in list:
123 self
.branch_list
= self
.list('branches')
125 self
.branch_list
= []
128 return self
.url
.split('/')[-1]
130 def export(self
, path
, dest_path
):
131 """Export PATH to DEST_PATH."""
132 url
= '/'.join([self
.url
, path
])
133 cmd
= [ SVN_CMD
, 'export', '-q', url
, dest_path
]
134 (output
, status
) = pipe(cmd
)
136 cmd_failed(cmd
, output
, status
)
138 def export_trunk(self
, dest_path
):
139 """Export trunk to DEST_PATH."""
140 self
.export('trunk', dest_path
)
142 def export_tag(self
, dest_path
, tag
):
143 """Export the tag TAG to DEST_PATH."""
144 self
.export('tags/' + tag
, dest_path
)
146 def export_branch(self
, dest_path
, branch
):
147 """Export the branch BRANCH to DEST_PATH."""
148 self
.export('branches/' + branch
, dest_path
)
150 def list(self
, path
):
151 """Return a list of all files and directories in PATH."""
152 cmd
= [ SVN_CMD
, 'ls', self
.url
+ '/' + path
]
153 (output
, status
) = pipe(cmd
)
155 cmd_failed(cmd
, output
, status
)
157 for line
in output
.split("\n"):
159 entries
.append(line
[:-1])
163 """Return a list of all tags in the repository."""
167 """Return a list of all branches in the repository."""
168 return self
.branch_list
173 def __init__(self
, path
):
175 self
.base_cmd
= [HG_CMD
, '-R', self
.path
]
177 self
._branches
= None # cache result of branches()
178 self
._have
_default
= None # so export_trunk() doesn't blow up
181 return os
.path
.basename(self
.path
)
183 def _export(self
, dest_path
, rev
):
184 cmd
= self
.base_cmd
+ ['archive',
187 '--exclude', 're:^\.hg',
189 (output
, status
) = pipe(cmd
)
191 cmd_failed(cmd
, output
, status
)
193 # If Mercurial has nothing to export, then it doesn't create
194 # dest_path. This breaks tree_compare(), so just check that the
195 # manifest for the chosen revision really is empty, and if so create
197 if not os
.path
.exists(dest_path
):
198 cmd
= self
.base_cmd
+ ['manifest', '--rev', rev
]
200 (output
, status
) = pipe(cmd
)
202 cmd_failed(cmd
, output
, status
)
203 manifest
= [fn
for fn
in output
.split("\n")[:-1]
204 if not fn
.startswith('.hg')]
208 def export_trunk(self
, dest_path
):
209 self
.branches() # ensure _have_default is set
210 if self
._have
_default
:
211 self
._export
(dest_path
, 'default')
213 # same as CVS does when exporting empty trunk
216 def export_tag(self
, dest_path
, tag
):
217 self
._export
(dest_path
, tag
)
219 def export_branch(self
, dest_path
, branch
):
220 self
._export
(dest_path
, branch
)
223 cmd
= self
.base_cmd
+ ['tags', '-q']
224 tags
= self
._split
_output
(cmd
)
229 if self
._branches
is None:
230 cmd
= self
.base_cmd
+ ['branches', '-q']
231 self
._branches
= branches
= self
._split
_output
(cmd
)
233 branches
.remove('default')
234 self
._have
_default
= True
236 self
._have
_default
= False
238 return self
._branches
240 def _split_output(self
, cmd
):
241 (output
, status
) = pipe(cmd
)
243 cmd_failed(cmd
, output
, status
)
244 return output
.split("\n")[:-1]
249 def __init__(self
, path
):
250 raise NotImplementedError()
252 def transform_symbol(ctx
, name
):
253 """Transform the symbol NAME using the renaming rules specified
254 with --symbol-transform. Return the transformed symbol name."""
256 for (pattern
, replacement
) in ctx
.symbol_transforms
:
257 newname
= pattern
.sub(replacement
, name
)
259 print " symbol '%s' transformed to '%s'" % (name
, newname
)
265 class Failures(object):
267 self
.count
= 0 # number of failures seen
269 def report(self
, summary
, details
=None):
271 sys
.stdout
.write(' FAIL: %s\n' % summary
)
274 sys
.stdout
.write(' %s\n' % line
)
276 def __nonzero__(self
):
277 return self
.count
> 0
279 def file_compare(failures
, base1
, base2
, run_diff
, rel_path
):
280 """Compare the mode and contents of two files. The paths are
281 specified as two base paths BASE1 and BASE2, and a path REL_PATH that
282 is relative to the two base paths. Return 1 if the file mode and
283 contents are identical, else 0."""
285 path1
= os
.path
.join(base1
, rel_path
)
286 path2
= os
.path
.join(base2
, rel_path
)
287 mode1
= os
.stat(path1
).st_mode
& 0700 # only look at owner bits
288 mode2
= os
.stat(path2
).st_mode
& 0700
290 failures
.report('File modes differ for %s' % rel_path
,
291 details
=['%s: %o' % (path1
, mode1
),
292 '%s: %o' % (path2
, mode2
)])
295 file1
= open(path1
, 'rb')
296 file2
= open(path2
, 'rb')
298 data1
= file1
.read(8192)
299 data2
= file2
.read(8192)
302 cmd
= ['diff', '-u', path1
, path2
]
303 (output
, status
) = pipe(cmd
)
304 diff
= output
.split('\n')
307 failures
.report('File contents differ for %s' % rel_path
,
317 def tree_compare(failures
, base1
, base2
, run_diff
, rel_path
=''):
318 """Compare the contents of two directory trees, including the contents
319 of all files. The paths are specified as two base paths BASE1 and BASE2,
320 and a path REL_PATH that is relative to the two base paths. Return 1
321 if the trees are identical, else 0."""
326 path1
= os
.path
.join(base1
, rel_path
)
327 path2
= os
.path
.join(base2
, rel_path
)
328 if not os
.path
.exists(path1
):
329 failures
.report('%s does not exist' % path1
)
331 if not os
.path
.exists(path2
):
332 failures
.report('%s does not exist' % path2
)
334 if os
.path
.isfile(path1
) and os
.path
.isfile(path2
):
335 return file_compare(failures
, base1
, base2
, run_diff
, rel_path
)
336 if not (os
.path
.isdir(path1
) and os
.path
.isdir(path2
)):
337 failures
.report('Path types differ for %r' % rel_path
)
339 entries1
= os
.listdir(path1
)
341 entries2
= os
.listdir(path2
)
343 missing
= filter(lambda x
: x
not in entries2
, entries1
)
344 extra
= filter(lambda x
: x
not in entries1
, entries2
)
346 failures
.report('Directory /%s is missing entries: %s' %
347 (rel_path
, ', '.join(missing
)))
349 failures
.report('Directory /%s has extra entries: %s' %
350 (rel_path
, ', '.join(extra
)))
352 for entry
in entries1
:
353 new_rel_path
= os
.path
.join(rel_path
, entry
)
354 if not tree_compare(failures
, base1
, base2
, run_diff
, new_rel_path
):
359 def verify_contents_single(failures
, cvsrepos
, verifyrepos
, kind
, label
, ctx
):
360 """Verify that the contents of the HEAD revision of all directories
361 and files in the conversion repository VERIFYREPOS matches the ones in
362 the CVS repository CVSREPOS. KIND can be either 'trunk', 'tag' or
363 'branch'. If KIND is either 'tag' or 'branch', LABEL is used to
364 specify the name of the tag or branch. CTX has the attributes:
365 CTX.tmpdir: specifying the directory for all temporary files.
366 CTX.skip_cleanup: if true, the temporary files are not deleted.
367 CTX.run_diff: if true, run diff on differing files.
369 itemname
= kind
+ (kind
!= 'trunk' and '-' + label
or '')
370 cvs_export_dir
= os
.path
.join(
371 ctx
.tmpdir
, 'cvs-export-%s' % itemname
)
372 vrf_export_dir
= os
.path
.join(
373 ctx
.tmpdir
, '%s-export-%s' % (verifyrepos
.name
, itemname
))
376 cvslabel
= transform_symbol(ctx
, label
)
381 cvsrepos
.export(cvs_export_dir
, cvslabel
, ctx
.keyword_opt
)
383 verifyrepos
.export_trunk(vrf_export_dir
)
385 verifyrepos
.export_tag(vrf_export_dir
, label
)
387 verifyrepos
.export_branch(vrf_export_dir
, label
)
390 failures
, cvs_export_dir
, vrf_export_dir
, ctx
.run_diff
394 if not ctx
.skip_cleanup
:
395 if os
.path
.exists(cvs_export_dir
):
396 shutil
.rmtree(cvs_export_dir
)
397 if os
.path
.exists(vrf_export_dir
):
398 shutil
.rmtree(vrf_export_dir
)
402 def verify_contents(failures
, cvsrepos
, verifyrepos
, ctx
):
403 """Verify that the contents of the HEAD revision of all directories
404 and files in the trunk, all tags and all branches in the conversion
405 repository VERIFYREPOS matches the ones in the CVS repository CVSREPOS.
406 CTX is passed through to verify_contents_single()."""
408 # branches/tags that failed:
411 # Verify contents of trunk
412 print 'Verifying trunk'
413 if not verify_contents_single(
414 failures
, cvsrepos
, verifyrepos
, 'trunk', None, ctx
416 locations
.append('trunk')
418 # Verify contents of all tags
419 for tag
in verifyrepos
.tags():
420 print 'Verifying tag', tag
421 if not verify_contents_single(
422 failures
, cvsrepos
, verifyrepos
, 'tag', tag
, ctx
424 locations
.append('tag:' + tag
)
426 # Verify contents of all branches
427 for branch
in verifyrepos
.branches():
428 if branch
[:10] == 'unlabeled-':
429 print 'Skipped branch', branch
431 print 'Verifying branch', branch
432 if not verify_contents_single(
433 failures
, cvsrepos
, verifyrepos
, 'branch', branch
, ctx
435 locations
.append('branch:' + branch
)
437 assert bool(failures
) == bool(locations
)
441 sys
.stdout
.write('FAIL: %s != %s: %d failure(s) in:\n'
442 % (cvsrepos
, verifyrepos
, failures
.count
))
443 for location
in locations
:
444 sys
.stdout
.write(' %s\n' % location
)
446 sys
.stdout
.write('PASS: %s == %s\n' % (cvsrepos
, verifyrepos
))
453 parser
= optparse
.OptionParser(
454 usage
='%prog [options] cvs-repos verify-repos')
455 parser
.add_option('--branch',
456 help='verify contents of the branch BRANCH only')
457 parser
.add_option('--diff', action
='store_true', dest
='run_diff',
458 help='run diff on differing files')
459 parser
.add_option('--tag',
460 help='verify contents of the tag TAG only')
461 parser
.add_option('--tmpdir',
463 help='path to store temporary files')
464 parser
.add_option('--trunk', action
='store_true',
465 help='verify contents of trunk only')
466 parser
.add_option('--symbol-transform', action
='append',
468 help='transform symbol names from P to S like cvs2svn, '
469 'except transforms SVN symbol to CVS symbol')
470 parser
.add_option('--svn',
471 action
='store_const', dest
='repos_type', const
='svn',
472 help='assume verify-repos is svn [default]')
473 parser
.add_option('--hg',
474 action
='store_const', dest
='repos_type', const
='hg',
475 help='assume verify-repos is hg')
476 parser
.add_option('--git',
477 action
='store_const', dest
='repos_type', const
='git',
478 help='assume verify-repos is git (not implemented!)')
479 parser
.add_option('--suppress-keywords',
480 action
='store_const', dest
='keyword_opt', const
='-kk',
481 help='suppress CVS keyword expansion '
482 '(equivalent to --keyword-opt=-kk)')
483 parser
.add_option('--keyword-opt',
485 help='control CVS keyword expansion by adding OPT to '
486 'cvs export command line')
488 parser
.set_defaults(run_diff
=False,
491 symbol_transforms
=[],
493 (options
, args
) = parser
.parse_args()
495 symbol_transforms
= []
496 for value
in options
.symbol_transforms
:
498 [pattern
, replacement
] = value
.split(":")
500 symbol_transforms
.append(
501 RegexpSymbolTransform(pattern
, replacement
))
503 parser
.error("'%s' is not a valid regexp." % (pattern
,))
506 """Print an error to sys.stderr."""
507 sys
.stderr
.write('Error: ' + str(msg
) + '\n')
509 verify_branch
= options
.branch
510 verify_tag
= options
.tag
511 verify_trunk
= options
.trunk
513 # Consistency check for options and arguments.
515 parser
.error("wrong number of arguments")
518 verify_path
= args
[1]
519 verify_klass
= {'svn': SvnRepos
,
521 'git': GitRepos
}[options
.repos_type
]
523 failures
= Failures()
525 # Open the repositories
526 cvsrepos
= CvsRepos(cvs_path
)
527 verifyrepos
= verify_klass(verify_path
)
531 print 'Verifying branch', verify_branch
532 verify_contents_single(
533 failures
, cvsrepos
, verifyrepos
, 'branch', verify_branch
, options
536 print 'Verifying tag', verify_tag
537 verify_contents_single(
538 failures
, cvsrepos
, verifyrepos
, 'tag', verify_tag
, options
541 print 'Verifying trunk'
542 verify_contents_single(
543 failures
, cvsrepos
, verifyrepos
, 'trunk', None, options
546 # Verify trunk, tags and branches
547 verify_contents(failures
, cvsrepos
, verifyrepos
, options
)
548 except RuntimeError, e
:
550 except KeyboardInterrupt:
553 sys
.exit(failures
and 1 or 0)
555 if __name__
== '__main__':