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
270 return str(self
.count
)
273 return "<%s at 0x%x: %s>" % (self
.__class
__.__name
__, id(self
), self
.count
)
275 def report(self
, summary
, details
=None):
277 sys
.stdout
.write(' FAIL: %s\n' % summary
)
280 sys
.stdout
.write(' %s\n' % line
)
282 def __nonzero__(self
):
283 return self
.count
> 0
286 def file_compare(failures
, base1
, base2
, run_diff
, rel_path
):
287 """Compare the mode and contents of two files.
289 The paths are specified as two base paths BASE1 and BASE2, and a
290 path REL_PATH that is relative to the two base paths. Return True
291 iff the file mode and contents are identical."""
294 path1
= os
.path
.join(base1
, rel_path
)
295 path2
= os
.path
.join(base2
, rel_path
)
296 mode1
= os
.stat(path1
).st_mode
& 0700 # only look at owner bits
297 mode2
= os
.stat(path2
).st_mode
& 0700
299 failures
.report('File modes differ for %s' % rel_path
,
300 details
=['%s: %o' % (path1
, mode1
),
301 '%s: %o' % (path2
, mode2
)])
304 file1
= open(path1
, 'rb')
305 file2
= open(path2
, 'rb')
307 data1
= file1
.read(8192)
308 data2
= file2
.read(8192)
311 cmd
= ['diff', '-u', path1
, path2
]
312 (output
, status
) = pipe(cmd
)
313 diff
= output
.split('\n')
316 failures
.report('File contents differ for %s' % rel_path
,
327 def tree_compare(failures
, base1
, base2
, run_diff
, rel_path
=''):
328 """Compare the contents of two directory trees, including file contents.
330 The paths are specified as two base paths BASE1 and BASE2, and a
331 path REL_PATH that is relative to the two base paths. Return True
332 iff the trees are identical."""
338 path1
= os
.path
.join(base1
, rel_path
)
339 path2
= os
.path
.join(base2
, rel_path
)
340 if not os
.path
.exists(path1
):
341 failures
.report('%s does not exist' % path1
)
343 if not os
.path
.exists(path2
):
344 failures
.report('%s does not exist' % path2
)
346 if os
.path
.isfile(path1
) and os
.path
.isfile(path2
):
347 return file_compare(failures
, base1
, base2
, run_diff
, rel_path
)
348 if not (os
.path
.isdir(path1
) and os
.path
.isdir(path2
)):
349 failures
.report('Path types differ for %r' % rel_path
)
351 entries1
= os
.listdir(path1
)
353 entries2
= os
.listdir(path2
)
358 missing
= filter(lambda x
: x
not in entries2
, entries1
)
359 extra
= filter(lambda x
: x
not in entries1
, entries2
)
361 failures
.report('Directory /%s is missing entries: %s' %
362 (rel_path
, ', '.join(missing
)))
365 failures
.report('Directory /%s has extra entries: %s' %
366 (rel_path
, ', '.join(extra
)))
369 for entry
in entries1
:
370 new_rel_path
= os
.path
.join(rel_path
, entry
)
371 if not tree_compare(failures
, base1
, base2
, run_diff
, new_rel_path
):
376 def verify_contents_single(failures
, cvsrepos
, verifyrepos
, kind
, label
, ctx
):
377 """Verify the HEAD revision of a trunk, tag, or branch.
379 Verify that the contents of the HEAD revision of all directories and
380 files in the conversion repository VERIFYREPOS match the ones in the
381 CVS repository CVSREPOS. KIND can be either 'trunk', 'tag' or
382 'branch'. If KIND is either 'tag' or 'branch', LABEL is used to
383 specify the name of the tag or branch. CTX has the attributes:
384 CTX.tmpdir: specifying the directory for all temporary files.
385 CTX.skip_cleanup: if true, the temporary files are not deleted.
386 CTX.run_diff: if true, run diff on differing files."""
388 itemname
= kind
+ (kind
!= 'trunk' and '-' + label
or '')
389 cvs_export_dir
= os
.path
.join(
390 ctx
.tmpdir
, 'cvs-export-%s' % itemname
)
391 vrf_export_dir
= os
.path
.join(
392 ctx
.tmpdir
, '%s-export-%s' % (verifyrepos
.name
, itemname
))
395 cvslabel
= transform_symbol(ctx
, label
)
400 cvsrepos
.export(cvs_export_dir
, cvslabel
, ctx
.keyword_opt
)
402 verifyrepos
.export_trunk(vrf_export_dir
)
404 verifyrepos
.export_tag(vrf_export_dir
, label
)
406 verifyrepos
.export_branch(vrf_export_dir
, label
)
409 failures
, cvs_export_dir
, vrf_export_dir
, ctx
.run_diff
413 if not ctx
.skip_cleanup
:
414 if os
.path
.exists(cvs_export_dir
):
415 shutil
.rmtree(cvs_export_dir
)
416 if os
.path
.exists(vrf_export_dir
):
417 shutil
.rmtree(vrf_export_dir
)
421 def verify_contents(failures
, cvsrepos
, verifyrepos
, ctx
):
422 """Verify that the contents of the HEAD revision of all directories
423 and files in the trunk, all tags and all branches in the conversion
424 repository VERIFYREPOS matches the ones in the CVS repository CVSREPOS.
425 CTX is passed through to verify_contents_single()."""
427 # branches/tags that failed:
430 # Verify contents of trunk
431 print 'Verifying trunk'
432 if not verify_contents_single(
433 failures
, cvsrepos
, verifyrepos
, 'trunk', None, ctx
435 locations
.append('trunk')
437 # Verify contents of all tags
438 for tag
in verifyrepos
.tags():
439 print 'Verifying tag', tag
440 if not verify_contents_single(
441 failures
, cvsrepos
, verifyrepos
, 'tag', tag
, ctx
443 locations
.append('tag:' + tag
)
445 # Verify contents of all branches
446 for branch
in verifyrepos
.branches():
447 if branch
[:10] == 'unlabeled-':
448 print 'Skipped branch', branch
450 print 'Verifying branch', branch
451 if not verify_contents_single(
452 failures
, cvsrepos
, verifyrepos
, 'branch', branch
, ctx
454 locations
.append('branch:' + branch
)
456 assert bool(failures
) == bool(locations
), \
457 "failures = %r\nlocations = %r" % (failures
, locations
)
461 sys
.stdout
.write('FAIL: %s != %s: %d failure(s) in:\n'
462 % (cvsrepos
, verifyrepos
, failures
.count
))
463 for location
in locations
:
464 sys
.stdout
.write(' %s\n' % location
)
466 sys
.stdout
.write('PASS: %s == %s\n' % (cvsrepos
, verifyrepos
))
473 parser
= optparse
.OptionParser(
474 usage
='%prog [options] cvs-repos verify-repos')
475 parser
.add_option('--branch',
476 help='verify contents of the branch BRANCH only')
477 parser
.add_option('--diff', action
='store_true', dest
='run_diff',
478 help='run diff on differing files')
479 parser
.add_option('--tag',
480 help='verify contents of the tag TAG only')
481 parser
.add_option('--tmpdir',
483 help='path to store temporary files')
484 parser
.add_option('--trunk', action
='store_true',
485 help='verify contents of trunk only')
486 parser
.add_option('--symbol-transform', action
='append',
488 help='transform symbol names from P to S like cvs2svn, '
489 'except transforms SVN symbol to CVS symbol')
490 parser
.add_option('--svn',
491 action
='store_const', dest
='repos_type', const
='svn',
492 help='assume verify-repos is svn [default]')
493 parser
.add_option('--hg',
494 action
='store_const', dest
='repos_type', const
='hg',
495 help='assume verify-repos is hg')
496 parser
.add_option('--git',
497 action
='store_const', dest
='repos_type', const
='git',
498 help='assume verify-repos is git (not implemented!)')
499 parser
.add_option('--suppress-keywords',
500 action
='store_const', dest
='keyword_opt', const
='-kk',
501 help='suppress CVS keyword expansion '
502 '(equivalent to --keyword-opt=-kk)')
503 parser
.add_option('--keyword-opt',
505 help='control CVS keyword expansion by adding OPT to '
506 'cvs export command line')
508 parser
.set_defaults(run_diff
=False,
511 symbol_transforms
=[],
513 (options
, args
) = parser
.parse_args()
515 symbol_transforms
= []
516 for value
in options
.symbol_transforms
:
518 [pattern
, replacement
] = value
.split(":")
520 symbol_transforms
.append(
521 RegexpSymbolTransform(pattern
, replacement
))
523 parser
.error("'%s' is not a valid regexp." % (pattern
,))
526 """Print an error to sys.stderr."""
527 sys
.stderr
.write('Error: ' + str(msg
) + '\n')
529 verify_branch
= options
.branch
530 verify_tag
= options
.tag
531 verify_trunk
= options
.trunk
533 # Consistency check for options and arguments.
535 parser
.error("wrong number of arguments")
538 verify_path
= args
[1]
539 verify_klass
= {'svn': SvnRepos
,
541 'git': GitRepos
}[options
.repos_type
]
543 failures
= Failures()
545 # Open the repositories
546 cvsrepos
= CvsRepos(cvs_path
)
547 verifyrepos
= verify_klass(verify_path
)
551 print 'Verifying branch', verify_branch
552 verify_contents_single(
553 failures
, cvsrepos
, verifyrepos
, 'branch', verify_branch
, options
556 print 'Verifying tag', verify_tag
557 verify_contents_single(
558 failures
, cvsrepos
, verifyrepos
, 'tag', verify_tag
, options
561 print 'Verifying trunk'
562 verify_contents_single(
563 failures
, cvsrepos
, verifyrepos
, 'trunk', None, options
566 # Verify trunk, tags and branches
567 verify_contents(failures
, cvsrepos
, verifyrepos
, options
)
568 except RuntimeError, e
:
570 except KeyboardInterrupt:
573 sys
.exit(failures
and 1 or 0)
575 if __name__
== '__main__':