1 # (Be in -*- mode: python; coding: utf-8 -*- mode.)
3 # ====================================================================
4 # Copyright (c) 2006-2009 CollabNet. All rights reserved.
6 # This software is licensed as described in the file COPYING, which
7 # you should have received as part of this distribution. The terms
8 # are also available at http://subversion.tigris.org/license-1.html.
9 # If newer versions of this license are posted there, you may use a
10 # newer version instead, at your option.
12 # This software consists of voluntary contributions made by many
13 # individuals. For exact contribution history, see the revision
14 # history and logs, available at http://cvs2svn.tigris.org/.
15 # ====================================================================
17 # #####################
18 # ## PLEASE READ ME! ##
19 # #####################
21 # This is a template for an options file that can be used to configure
22 # cvs2svn to convert to Bazaar rather than to Subversion. See
23 # www/cvs2bzr.html and www/cvs2svn.html for general information, and
24 # see the comments in this file for information about what options are
25 # available and how they can be set.
27 # The program that is run to convert from CVS to Bazaar is called
28 # cvs2bzr. Run it with the --options option, passing it this file
31 # cvs2bzr --options=cvs2bzr-example.options
33 # The output of cvs2bzr is a dump file that can be loaded into Bazaar
34 # using the "bzr fast-import" command. Please read www/cvs2bzr.html
35 # for more information.
37 # Many options do not have defaults, so it is easier to copy this file
38 # and modify what you need rather than creating a new options file
39 # from scratch. This file is in Python syntax, but you don't need to
40 # know Python to modify it. But if you *do* know Python, then you
41 # will be happy to know that you can use arbitary Python constructs to
42 # do fancy configuration tricks.
44 # But please be aware of the following:
46 # * In many places, leading whitespace is significant in Python (it is
47 # used instead of curly braces to group statements together).
48 # Therefore, if you don't know what you are doing, it is best to
49 # leave the whitespace as it is.
51 # * In normal strings, Python treats a backslash ("\") as an escape
52 # character. Therefore, if you want to specify a string that
53 # contains a backslash, you need either to escape the backslash with
54 # another backslash ("\\"), or use a "raw string", as in one if the
55 # following equivalent examples:
57 # cvs_executable = 'c:\\windows\\system32\\cvs.exe'
58 # cvs_executable = r'c:\windows\system32\cvs.exe'
60 # See http://docs.python.org/tutorial/introduction.html#strings for
63 # Two identifiers will have been defined before this file is executed,
64 # and can be used freely within this file:
66 # ctx -- a Ctx object (see cvs2svn_lib/context.py), which holds
67 # many configuration options
69 # run_options -- an instance of the BzrRunOptions class (see
70 # cvs2svn_lib/bzr_run_options.py), which holds some variables
71 # governing how cvs2bzr is run
74 # Import some modules that are used in setting the options:
77 from cvs2svn_lib import config
78 from cvs2svn_lib import changeset_database
79 from cvs2svn_lib.common import CVSTextDecoder
80 from cvs2svn_lib.log import logger
81 from cvs2svn_lib.project import Project
82 from cvs2svn_lib.git_output_option import GitRevisionInlineWriter
83 from cvs2svn_lib.bzr_output_option import BzrOutputOption
84 from cvs2svn_lib.dvcs_common import KeywordHandlingPropertySetter
85 from cvs2svn_lib.revision_manager import NullRevisionCollector
86 from cvs2svn_lib.rcs_revision_manager import RCSRevisionReader
87 from cvs2svn_lib.cvs_revision_manager import CVSRevisionReader
88 from cvs2svn_lib.checkout_internal import InternalRevisionCollector
89 from cvs2svn_lib.checkout_internal import InternalRevisionReader
90 from cvs2svn_lib.symbol_strategy import AllBranchRule
91 from cvs2svn_lib.symbol_strategy import AllTagRule
92 from cvs2svn_lib.symbol_strategy import BranchIfCommitsRule
93 from cvs2svn_lib.symbol_strategy import ExcludeRegexpStrategyRule
94 from cvs2svn_lib.symbol_strategy import ForceBranchRegexpStrategyRule
95 from cvs2svn_lib.symbol_strategy import ForceTagRegexpStrategyRule
96 from cvs2svn_lib.symbol_strategy import ExcludeTrivialImportBranchRule
97 from cvs2svn_lib.symbol_strategy import ExcludeVendorBranchRule
98 from cvs2svn_lib.symbol_strategy import HeuristicStrategyRule
99 from cvs2svn_lib.symbol_strategy import UnambiguousUsageRule
100 from cvs2svn_lib.symbol_strategy import HeuristicPreferredParentRule
101 from cvs2svn_lib.symbol_strategy import SymbolHintsFileRule
102 from cvs2svn_lib.symbol_transform import ReplaceSubstringsSymbolTransform
103 from cvs2svn_lib.symbol_transform import RegexpSymbolTransform
104 from cvs2svn_lib.symbol_transform import IgnoreSymbolTransform
105 from cvs2svn_lib.symbol_transform import NormalizePathsSymbolTransform
106 from cvs2svn_lib.property_setters import AutoPropsPropertySetter
107 from cvs2svn_lib.property_setters import ConditionalPropertySetter
108 from cvs2svn_lib.property_setters import cvs_file_is_binary
109 from cvs2svn_lib.property_setters import CVSBinaryFileDefaultMimeTypeSetter
110 from cvs2svn_lib.property_setters import CVSBinaryFileEOLStyleSetter
111 from cvs2svn_lib.property_setters import DefaultEOLStyleSetter
112 from cvs2svn_lib.property_setters import EOLStyleFromMimeTypeSetter
113 from cvs2svn_lib.property_setters import ExecutablePropertySetter
114 from cvs2svn_lib.property_setters import KeywordsPropertySetter
115 from cvs2svn_lib.property_setters import MimeMapper
116 from cvs2svn_lib.property_setters import SVNBinaryFileKeywordsPropertySetter
118 # To choose the level of logging output, uncomment one of the
120 #logger.log_level = logger.WARN
121 #logger.log_level = logger.QUIET
122 logger.log_level = logger.NORMAL
123 #logger.log_level = logger.VERBOSE
124 #logger.log_level = logger.DEBUG
127 # The directory to use for temporary files:
128 ctx.tmpdir = r'cvs2svn-tmp'
130 # cvs2bzr does not need to keep track of what revisions will be
131 # excluded, so leave this option unchanged:
132 ctx.revision_collector = NullRevisionCollector()
134 # cvs2bzr's revision reader is set via the BzrOutputOption constructor,
135 # so leave this option set to None.
136 ctx.revision_reader = None
138 # Change the following line to True if the conversion should only
139 # include the trunk of the repository (i.e., all branches and tags
140 # should be omitted from the conversion):
141 ctx.trunk_only = False
143 # How to convert CVS author names, log messages, and filenames to
144 # Unicode. The first argument to CVSTextDecoder is a list of encoders
145 # that are tried in order in 'strict' mode until one of them succeeds.
146 # If none of those succeeds, then fallback_encoder (if it is
147 # specified) is used in lossy 'replace' mode. Setting a fallback
148 # encoder ensures that the encoder always succeeds, but it can cause
150 ctx.cvs_author_decoder = CVSTextDecoder(
156 #fallback_encoding='ascii'
158 ctx.cvs_log_decoder = CVSTextDecoder(
164 #fallback_encoding='ascii',
167 # You might want to be especially strict when converting filenames to
168 # Unicode (e.g., maybe not specify a fallback_encoding).
169 ctx.cvs_filename_decoder = CVSTextDecoder(
175 #fallback_encoding='ascii'
178 # Template for the commit message to be used for initial project
180 ctx.initial_project_commit_message = (
181 'Standard project directories initialized by cvs2svn.'
184 # Template for the commit message to be used for post commits, in
185 # which modifications to a vendor branch are copied back to trunk.
186 # This message can use '%(revnum)d' to include the SVN revision number
187 # of the revision that included the change to the vendor branch
188 # (admittedly rather pointless in a cvs2bzr conversion).
189 ctx.post_commit_message = (
190 'This commit was generated by cvs2svn to track changes on a CVS '
194 # Template for the commit message to be used for commits in which
195 # symbols are created. This message can use '%(symbol_type)s' to
196 # include the type of the symbol ('branch' or 'tag') or
197 # '%(symbol_name)s' to include the name of the symbol.
198 ctx.symbol_commit_message = (
199 "This commit was manufactured by cvs2svn to create %(symbol_type)s "
203 # Template for the commit message to be used for commits in which
204 # tags are pseudo-merged back to their source branch. This message can
205 # use '%(symbol_name)s' to include the name of the symbol.
206 # (Not used by default unless you enable tie_tag_fixup_branches on
208 ctx.tie_tag_ancestry_message = (
209 "This commit was manufactured by cvs2svn to tie ancestry for "
210 "tag '%(symbol_name)s' back to the source branch."
213 # Some CVS clients for MacOS store resource fork data into CVS along
214 # with the file contents itself by wrapping it all up in a container
215 # format called "AppleSingle". Subversion currently does not support
216 # MacOS resource forks. Nevertheless, sometimes the resource fork
217 # information is not necessary and can be discarded. Set the
218 # following option to True if you would like cvs2svn to identify files
219 # whose contents are encoded in AppleSingle format, and discard all
220 # but the data fork for such files before committing them to
221 # Subversion. (Please note that AppleSingle contents are identified
222 # by the AppleSingle magic number as the first four bytes of the file.
223 # This check is not failproof, so only set this option if you think
225 ctx.decode_apple_single = False
227 # This option can be set to the name of a filename to which are stored
228 # statistics and conversion decisions about the CVS symbols.
229 ctx.symbol_info_filename = None
230 #ctx.symbol_info_filename = 'symbol-info.txt'
232 # cvs2svn uses "symbol strategy rules" to help decide how to handle
233 # CVS symbols. The rules in a project's symbol_strategy_rules are
234 # applied in order, and each rule is allowed to modify the symbol.
235 # The result (after each of the rules has been applied) is used for
238 # 1. A CVS symbol might be used as a tag in one file and as a branch
239 # in another file. cvs2svn has to decide whether to convert such a
240 # symbol as a tag or as a branch. cvs2svn uses a series of
241 # heuristic rules to decide how to convert a symbol. The user can
242 # override the default rules for specific symbols or symbols
243 # matching regular expressions.
245 # 2. cvs2svn is also capable of excluding symbols from the conversion
246 # (provided no other symbols depend on them.
248 # 3. CVS does not record unambiguously the line of development from
249 # which a symbol sprouted. cvs2svn uses a heuristic to choose a
250 # symbol's "preferred parents".
252 # The standard branch/tag/exclude StrategyRules do not change a symbol
253 # that has already been processed by an earlier rule, so in effect the
254 # first matching rule is the one that is used.
256 global_symbol_strategy_rules = [
257 # It is possible to specify manually exactly how symbols should be
258 # converted and what line of development should be used as the
259 # preferred parent. To do so, create a file containing the symbol
260 # hints and enable the following option.
262 # The format of the hints file is described in the documentation
263 # for the --symbol-hints command-line option. The file output by
264 # the --write-symbol-info (i.e., ctx.symbol_info_filename) option
265 # is in the same format. The simplest way to use this option is
266 # to run the conversion through CollateSymbolsPass with
267 # --write-symbol-info option, copy the symbol info and edit it to
268 # create a hints file, then re-start the conversion at
269 # CollateSymbolsPass with this option enabled.
270 #SymbolHintsFileRule('symbol-hints.txt'),
272 # To force all symbols matching a regular expression to be
273 # converted as branches, add rules like the following:
274 #ForceBranchRegexpStrategyRule(r'branch.*'),
276 # To force all symbols matching a regular expression to be
277 # converted as tags, add rules like the following:
278 #ForceTagRegexpStrategyRule(r'tag.*'),
280 # To force all symbols matching a regular expression to be
281 # excluded from the conversion, add rules like the following:
282 #ExcludeRegexpStrategyRule(r'unknown-.*'),
284 # Sometimes people use "cvs import" to get their own source code
285 # into CVS. This practice creates a vendor branch 1.1.1 and
286 # imports the code onto the vendor branch as 1.1.1.1, then copies
287 # the same content to the trunk as version 1.1. Normally, such
288 # vendor branches are useless and they complicate the SVN history
289 # unnecessarily. The following rule excludes any branches that
290 # only existed as a vendor branch with a single import (leaving
291 # only the 1.1 revision). If you want to retain such branches,
292 # comment out the following line. (Please note that this rule
293 # does not exclude vendor *tags*, as they are not so easy to
295 ExcludeTrivialImportBranchRule(),
297 # To exclude all vendor branches (branches that had "cvs import"s
298 # on them but no other kinds of commits), uncomment the following
300 #ExcludeVendorBranchRule(),
302 # Usually you want this rule, to convert unambiguous symbols
303 # (symbols that were only ever used as tags or only ever used as
304 # branches in CVS) the same way they were used in CVS:
305 UnambiguousUsageRule(),
307 # If there was ever a commit on a symbol, then it cannot be
308 # converted as a tag. This rule causes all such symbols to be
309 # converted as branches. If you would like to resolve such
310 # ambiguities manually, comment out the following line:
311 BranchIfCommitsRule(),
313 # Last in the list can be a catch-all rule that is used for
314 # symbols that were not matched by any of the more specific rules
315 # above. (Assuming that BranchIfCommitsRule() was included above,
316 # then the symbols that are still indeterminate at this point can
317 # sensibly be converted as branches or tags.) Include at most one
318 # of these lines. If none of these catch-all rules are included,
319 # then the presence of any ambiguous symbols (that haven't been
320 # disambiguated above) is an error:
322 # Convert ambiguous symbols based on whether they were used more
323 # often as branches or as tags:
324 HeuristicStrategyRule(),
325 # Convert all ambiguous symbols as branches:
327 # Convert all ambiguous symbols as tags:
330 # The last rule is here to choose the preferred parent of branches
331 # and tags, that is, the line of development from which the symbol
333 HeuristicPreferredParentRule(),
336 # Specify a username to be used for commits for which CVS doesn't
337 # record the original author (for example, the creation of a branch).
338 # This should be a simple (unix-style) username, but it can be
339 # translated into a Bazaar-style name by the author_transforms map.
340 ctx.username = 'cvs2svn'
342 # ctx.file_property_setters and ctx.revision_property_setters contain
343 # rules used to set the svn properties on files in the converted
344 # archive. For each file, the rules are tried one by one. Any rule
345 # can add or suppress one or more svn properties. Typically the rules
346 # will not overwrite properties set by a previous rule (though they
347 # are free to do so). ctx.file_property_setters should be used for
348 # properties that remain the same for the life of the file; these
349 # should implement FilePropertySetter. ctx.revision_property_setters
350 # should be used for properties that are allowed to vary from revision
351 # to revision; these should implement RevisionPropertySetter.
353 # Obviously, SVN properties per se are not interesting for a cvs2bzr
354 # conversion, but some of these properties have side-effects that do
355 # affect the Bazaar output. FIXME: Document this in more detail.
356 ctx.file_property_setters.extend([
357 # To read auto-props rules from a file, uncomment the following line
358 # and specify a filename. The boolean argument specifies whether
359 # case should be ignored when matching filenames to the filename
360 # patterns found in the auto-props file:
361 #AutoPropsPropertySetter(
362 # r'/home/username/.subversion/config',
366 # To read mime types from a file and use them to set svn:mime-type
367 # based on the filename extensions, uncomment the following line
368 # and specify a filename (see
369 # http://en.wikipedia.org/wiki/Mime.types for information about
371 #MimeMapper(r'/etc/mime.types', ignore_case=False),
373 # Omit the svn:eol-style property from any files that are listed
374 # as binary (i.e., mode '-kb') in CVS:
375 CVSBinaryFileEOLStyleSetter(),
377 # If the file is binary and its svn:mime-type property is not yet
378 # set, set svn:mime-type to 'application/octet-stream'.
379 CVSBinaryFileDefaultMimeTypeSetter(),
381 # To try to determine the eol-style from the mime type, uncomment
382 # the following line:
383 #EOLStyleFromMimeTypeSetter(),
385 # Choose one of the following lines to set the default
386 # svn:eol-style if none of the above rules applied. The argument
387 # is the svn:eol-style that should be applied, or None if no
388 # svn:eol-style should be set (i.e., the file should be treated as
391 # The default is to treat all files as binary unless one of the
392 # previous rules has determined otherwise, because this is the
393 # safest approach. However, if you have been diligent about
394 # marking binary files with -kb in CVS and/or you have used the
395 # above rules to definitely mark binary files as binary, then you
396 # might prefer to use 'native' as the default, as it is usually
397 # the most convenient setting for text files. Other possible
398 # options: 'CRLF', 'CR', 'LF'.
399 DefaultEOLStyleSetter(None),
400 #DefaultEOLStyleSetter('native'),
402 # Prevent svn:keywords from being set on files that have
403 # svn:eol-style unset.
404 SVNBinaryFileKeywordsPropertySetter(),
406 # If svn:keywords has not been set yet, set it based on the file's
408 KeywordsPropertySetter(config.SVN_KEYWORDS_VALUE),
410 # Set the svn:executable flag on any files that are marked in CVS as
412 ExecutablePropertySetter(),
414 # The following causes keywords to be untouched in binary files and
415 # collapsed in all text to be committed:
416 ConditionalPropertySetter(
417 cvs_file_is_binary, KeywordHandlingPropertySetter('untouched'),
419 KeywordHandlingPropertySetter('collapsed'),
422 ctx.revision_property_setters.extend([
425 # To skip the cleanup of temporary files, uncomment the following
427 #ctx.skip_cleanup = True
430 # In CVS, it is perfectly possible to make a single commit that
431 # affects more than one project or more than one branch of a single
432 # project. Subversion also allows such commits. Therefore, by
433 # default, when cvs2svn sees what looks like a cross-project or
434 # cross-branch CVS commit, it converts it into a
435 # cross-project/cross-branch Subversion commit.
437 # However, other tools and SCMs have trouble representing
438 # cross-project or cross-branch commits. (For example, Trac's Revtree
439 # plugin, http://www.trac-hacks.org/wiki/RevtreePlugin is confused by
440 # such commits.) Therefore, we provide the following two options to
441 # allow cross-project/cross-branch commits to be suppressed.
443 # cvs2bzr only supports single-project conversions (multiple-project
444 # conversions wouldn't really make sense for Bazaar anyway). So this
445 # option must be set to False:
446 ctx.cross_project_commits = False
448 # Bazaar itself doesn't allow commits that affect more than one branch,
449 # so this option must be set to False:
450 ctx.cross_branch_commits = False
452 # cvs2bzr does not yet handle translating .cvsignore files into
453 # .bzrignore content, so by default, the .cvsignore files are included
454 # in the conversion output. If you would like to omit the .cvsignore
455 # files from the output, set this option to False:
456 ctx.keep_cvsignore = True
458 # By default, it is a fatal error for a CVS ",v" file to appear both
459 # inside and outside of an "Attic" subdirectory (this should never
460 # happen, but frequently occurs due to botched repository
461 # administration). If you would like to retain both versions of such
462 # files, change the following option to True, and the attic version of
463 # the file will be written to a subdirectory called "Attic" in the
465 ctx.retain_conflicting_attic_files = False
467 # CVS uses unix login names as author names whereas Bazaar requires
468 # author names to be of the form "foo <bar>". The default is to set
469 # the Bazaar author to "cvsauthor <cvsauthor>". author_transforms can
470 # be used to map cvsauthor names (e.g., "jrandom") to a true name and
471 # email address (e.g., "J. Random <jrandom@example.com>" for the
472 # example shown). All strings should be either Unicode strings (i.e.,
473 # with "u" as a prefix) or 8-bit strings in the utf-8 encoding. The
474 # values can either be strings in the form "name <email>" or tuples
475 # (name, email). Please substitute your own project's usernames here
476 # to use with the author_transforms option of BzrOutputOption below.
478 'jrandom' : ('J. Random', 'jrandom@example.com'),
479 'mhagger' : 'Michael Haggerty <mhagger@alum.mit.edu>',
480 'brane' : (u'Branko Čibej', 'brane@xbc.nu'),
481 'ringstrom' : 'Tobias Ringström <tobias@ringstrom.mine.nu>',
482 'dionisos' : (u'Erik Hülsmann', 'e.huelsmann@gmx.net'),
484 # This one will be used for commits for which CVS doesn't record
485 # the original author, as explained above.
486 'cvs2svn' : 'cvs2svn <admin@example.com>',
489 # This is the main option that causes cvs2svn to output to a
490 # "fastimport"-format dumpfile rather than to Subversion:
491 ctx.output_option = BzrOutputOption(
492 # The file in which to write the "fastimport" stream:
493 os.path.join(ctx.tmpdir, 'dumpfile.fi'),
495 # Write the file contents inline in the "fastimport" stream,
496 # rather than using a separate blobs file (which "bzr fastimport"
497 # can't handle as easily).
498 revision_writer=GitRevisionInlineWriter(
499 # cvs2bzr uses either RCS's "co" command or CVS's "cvs co -p" to
500 # extract the content of file revisions. Here you can choose
501 # whether to use RCS (faster, but fails in some rare
502 # circumstances) or CVS (much slower, but more reliable).
503 #RCSRevisionReader(co_executable=r'co')
504 CVSRevisionReader(cvs_executable=r'cvs')
507 # Optional map from CVS author names to Bazaar author names:
508 author_transforms=author_transforms,
511 # Change this option to True to turn on profiling of cvs2svn (for
512 # debugging purposes):
513 run_options.profiling = False
516 # Should CVSItem -> Changeset database files be memory mapped? In
517 # some tests, using memory mapping speeded up the overall conversion
518 # by about 5%. But this option can cause the conversion to fail with
519 # an out of memory error if the conversion computer runs out of
520 # virtual address space (e.g., when running a very large conversion on
521 # a 32-bit operating system). Therefore it is disabled by default.
522 # Uncomment the following line to allow these database files to be
524 #changeset_database.use_mmap_for_cvs_item_to_changeset_table = True
526 # Now set the project to be converted to Bazaar. cvs2bzr only supports
527 # single-project conversions, so this method must only be called
529 run_options.set_project(
530 # The filesystem path to the part of the CVS repository (*not* a
531 # CVS working copy) that should be converted. This may be a
532 # subdirectory (i.e., a module) within a larger CVS repository.
533 r'test-data/main-cvsrepos',
535 # A list of symbol transformations that can be used to rename
536 # symbols in this project.
538 # Use IgnoreSymbolTransforms like the following to completely
539 # ignore symbols matching a regular expression when parsing
540 # the CVS repository, for example to avoid warnings about
541 # branches with two names and to choose the preferred name.
542 # It is *not* recommended to use this instead of
543 # ExcludeRegexpStrategyRule; though more efficient,
544 # IgnoreSymbolTransforms are less flexible and don't exclude
545 # branches correctly. The argument is a Python-style regular
546 # expression that has to match the *whole* CVS symbol name:
547 #IgnoreSymbolTransform(r'nightly-build-tag-.*')
549 # RegexpSymbolTransforms transform symbols textually using a
550 # regular expression. The first argument is a Python regular
551 # expression pattern and the second is a replacement pattern.
552 # The pattern is matched against each symbol name. If it
553 # matches the whole symbol name, then the symbol name is
554 # replaced with the corresponding replacement text. The
555 # replacement can include substitution patterns (e.g., r'\1'
556 # or r'\g<name>'). Typically you will want to use raw strings
557 # (strings with a preceding 'r', like shown in the examples)
558 # for the regexp and its replacement to avoid backslash
559 # substitution within those strings.
560 #RegexpSymbolTransform(r'release-(\d+)_(\d+)',
562 #RegexpSymbolTransform(r'release-(\d+)_(\d+)_(\d+)',
563 # r'release-\1.\2.\3'),
565 # Simple 1:1 character replacements can also be done. The
566 # following transform, which converts backslashes into forward
567 # slashes, should usually be included:
568 ReplaceSubstringsSymbolTransform('\\','/'),
570 # This last rule eliminates leading, trailing, and repeated
571 # slashes within the output symbol names:
572 NormalizePathsSymbolTransform(),
575 # See the definition of global_symbol_strategy_rules above for a
576 # description of this option:
577 symbol_strategy_rules=global_symbol_strategy_rules,
579 # Exclude paths from the conversion. Should be relative to
580 # repository path and use forward slashes:
581 #exclude_paths=['file-to-exclude.txt,v', 'dir/to/exclude'],