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 Mercurial rather than to Subversion. See
23 # www/cvs2git.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 # "cvs2hg" is shorthand for "cvs2git in the mode where it is
28 # outputting to Mercurial instead of git". But the program that needs
29 # to be run is still called "cvs2git". Run it with the --options
30 # option, passing it this file as argument:
32 # cvs2git --options=cvs2hg-example.options
34 # Mercurial can (experimentally at this time) read git-fast-import
35 # format via its "hg fastimport" extension, with exceptions:
37 # 1. "hg fastimport" does not support blobs, so the contents of the
38 # revisions are output inline rather than in a separate blobs file.
39 # This increases the size of the output, because file contents that
40 # appear identically on multiple branches have to be output
43 # Many options do not have defaults, so it is easier to copy this file
44 # and modify what you need rather than creating a new options file
45 # from scratch. This file is in Python syntax, but you don't need to
46 # know Python to modify it. But if you *do* know Python, then you
47 # will be happy to know that you can use arbitary Python constructs to
48 # do fancy configuration tricks.
50 # But please be aware of the following:
52 # * In many places, leading whitespace is significant in Python (it is
53 # used instead of curly braces to group statements together).
54 # Therefore, if you don't know what you are doing, it is best to
55 # leave the whitespace as it is.
57 # * In normal strings, Python treats a backslash ("\") as an escape
58 # character. Therefore, if you want to specify a string that
59 # contains a backslash, you need either to escape the backslash with
60 # another backslash ("\\"), or use a "raw string", as in one if the
61 # following equivalent examples:
63 # cvs_executable = 'c:\\windows\\system32\\cvs.exe'
64 # cvs_executable = r'c:\windows\system32\cvs.exe'
66 # See http://docs.python.org/tutorial/introduction.html#strings for
69 # Two identifiers will have been defined before this file is executed,
70 # and can be used freely within this file:
72 # ctx -- a Ctx object (see cvs2svn_lib/context.py), which holds
73 # many configuration options
75 # run_options -- an instance of the GitRunOptions class (see
76 # cvs2svn_lib/git_run_options.py), which holds some variables
77 # governing how cvs2git is run
80 # Import some modules that are used in setting the options:
83 from cvs2svn_lib import config
84 from cvs2svn_lib import changeset_database
85 from cvs2svn_lib.common import CVSTextDecoder
86 from cvs2svn_lib.log import logger
87 from cvs2svn_lib.project import Project
88 from cvs2svn_lib.git_output_option import GitRevisionInlineWriter
89 from cvs2svn_lib.git_output_option import GitOutputOption
90 from cvs2svn_lib.dvcs_common import KeywordHandlingPropertySetter
91 from cvs2svn_lib.revision_manager import NullRevisionCollector
92 from cvs2svn_lib.rcs_revision_manager import RCSRevisionReader
93 from cvs2svn_lib.cvs_revision_manager import CVSRevisionReader
94 from cvs2svn_lib.checkout_internal import InternalRevisionCollector
95 from cvs2svn_lib.checkout_internal import InternalRevisionReader
96 from cvs2svn_lib.symbol_strategy import AllBranchRule
97 from cvs2svn_lib.symbol_strategy import AllTagRule
98 from cvs2svn_lib.symbol_strategy import BranchIfCommitsRule
99 from cvs2svn_lib.symbol_strategy import ExcludeRegexpStrategyRule
100 from cvs2svn_lib.symbol_strategy import ForceBranchRegexpStrategyRule
101 from cvs2svn_lib.symbol_strategy import ForceTagRegexpStrategyRule
102 from cvs2svn_lib.symbol_strategy import ExcludeTrivialImportBranchRule
103 from cvs2svn_lib.symbol_strategy import ExcludeVendorBranchRule
104 from cvs2svn_lib.symbol_strategy import HeuristicStrategyRule
105 from cvs2svn_lib.symbol_strategy import UnambiguousUsageRule
106 from cvs2svn_lib.symbol_strategy import HeuristicPreferredParentRule
107 from cvs2svn_lib.symbol_strategy import SymbolHintsFileRule
108 from cvs2svn_lib.symbol_transform import ReplaceSubstringsSymbolTransform
109 from cvs2svn_lib.symbol_transform import RegexpSymbolTransform
110 from cvs2svn_lib.symbol_transform import IgnoreSymbolTransform
111 from cvs2svn_lib.symbol_transform import NormalizePathsSymbolTransform
112 from cvs2svn_lib.property_setters import AutoPropsPropertySetter
113 from cvs2svn_lib.property_setters import ConditionalPropertySetter
114 from cvs2svn_lib.property_setters import cvs_file_is_binary
115 from cvs2svn_lib.property_setters import CVSBinaryFileDefaultMimeTypeSetter
116 from cvs2svn_lib.property_setters import CVSBinaryFileEOLStyleSetter
117 from cvs2svn_lib.property_setters import DefaultEOLStyleSetter
118 from cvs2svn_lib.property_setters import EOLStyleFromMimeTypeSetter
119 from cvs2svn_lib.property_setters import ExecutablePropertySetter
120 from cvs2svn_lib.property_setters import KeywordsPropertySetter
121 from cvs2svn_lib.property_setters import MimeMapper
122 from cvs2svn_lib.property_setters import SVNBinaryFileKeywordsPropertySetter
124 # To choose the level of logging output, uncomment one of the
126 #logger.log_level = logger.WARN
127 #logger.log_level = logger.QUIET
128 logger.log_level = logger.NORMAL
129 #logger.log_level = logger.VERBOSE
130 #logger.log_level = logger.DEBUG
133 # The directory to use for temporary files:
134 ctx.tmpdir = r'cvs2svn-tmp'
136 # cvs2hg does not need to keep track of what revisions will be
137 # excluded, so leave this option unchanged:
138 ctx.revision_collector = NullRevisionCollector()
140 # cvs2hg's revision reader is set via the GitOutputOption constructor,
141 # so leave this option set to None.
142 ctx.revision_reader = None
144 # Change the following line to True if the conversion should only
145 # include the trunk of the repository (i.e., all branches and tags
146 # should be omitted from the conversion):
147 ctx.trunk_only = False
149 # How to convert CVS author names, log messages, and filenames to
150 # Unicode. The first argument to CVSTextDecoder is a list of encoders
151 # that are tried in order in 'strict' mode until one of them succeeds.
152 # If none of those succeeds, then fallback_encoder (if it is
153 # specified) is used in lossy 'replace' mode. Setting a fallback
154 # encoder ensures that the encoder always succeeds, but it can cause
156 ctx.cvs_author_decoder = CVSTextDecoder(
162 #fallback_encoding='ascii'
164 ctx.cvs_log_decoder = CVSTextDecoder(
170 #fallback_encoding='ascii',
173 # You might want to be especially strict when converting filenames to
174 # Unicode (e.g., maybe not specify a fallback_encoding).
175 ctx.cvs_filename_decoder = CVSTextDecoder(
181 #fallback_encoding='ascii'
184 # Template for the commit message to be used for initial project
186 ctx.initial_project_commit_message = (
187 'Standard project directories initialized by cvs2svn.'
190 # Template for the commit message to be used for post commits, in
191 # which modifications to a vendor branch are copied back to trunk.
192 # This message can use '%(revnum)d' to include the SVN revision number
193 # of the revision that included the change to the vendor branch
194 # (admittedly rather pointless in a cvs2hg conversion).
195 ctx.post_commit_message = (
196 'This commit was generated by cvs2svn to track changes on a CVS '
200 # Template for the commit message to be used for commits in which
201 # symbols are created. This message can use '%(symbol_type)s' to
202 # include the type of the symbol ('branch' or 'tag') or
203 # '%(symbol_name)s' to include the name of the symbol.
204 ctx.symbol_commit_message = (
205 "This commit was manufactured by cvs2svn to create %(symbol_type)s "
209 # Some CVS clients for MacOS store resource fork data into CVS along
210 # with the file contents itself by wrapping it all up in a container
211 # format called "AppleSingle". Subversion currently does not support
212 # MacOS resource forks. Nevertheless, sometimes the resource fork
213 # information is not necessary and can be discarded. Set the
214 # following option to True if you would like cvs2svn to identify files
215 # whose contents are encoded in AppleSingle format, and discard all
216 # but the data fork for such files before committing them to
217 # Subversion. (Please note that AppleSingle contents are identified
218 # by the AppleSingle magic number as the first four bytes of the file.
219 # This check is not failproof, so only set this option if you think
221 ctx.decode_apple_single = False
223 # This option can be set to the name of a filename to which are stored
224 # statistics and conversion decisions about the CVS symbols.
225 ctx.symbol_info_filename = None
226 #ctx.symbol_info_filename = 'symbol-info.txt'
228 # cvs2svn uses "symbol strategy rules" to help decide how to handle
229 # CVS symbols. The rules in a project's symbol_strategy_rules are
230 # applied in order, and each rule is allowed to modify the symbol.
231 # The result (after each of the rules has been applied) is used for
234 # 1. A CVS symbol might be used as a tag in one file and as a branch
235 # in another file. cvs2svn has to decide whether to convert such a
236 # symbol as a tag or as a branch. cvs2svn uses a series of
237 # heuristic rules to decide how to convert a symbol. The user can
238 # override the default rules for specific symbols or symbols
239 # matching regular expressions.
241 # 2. cvs2svn is also capable of excluding symbols from the conversion
242 # (provided no other symbols depend on them.
244 # 3. CVS does not record unambiguously the line of development from
245 # which a symbol sprouted. cvs2svn uses a heuristic to choose a
246 # symbol's "preferred parents".
248 # The standard branch/tag/exclude StrategyRules do not change a symbol
249 # that has already been processed by an earlier rule, so in effect the
250 # first matching rule is the one that is used.
252 global_symbol_strategy_rules = [
253 # It is possible to specify manually exactly how symbols should be
254 # converted and what line of development should be used as the
255 # preferred parent. To do so, create a file containing the symbol
256 # hints and enable the following option.
258 # The format of the hints file is described in the documentation
259 # for the --symbol-hints command-line option. The file output by
260 # the --write-symbol-info (i.e., ctx.symbol_info_filename) option
261 # is in the same format. The simplest way to use this option is
262 # to run the conversion through CollateSymbolsPass with
263 # --write-symbol-info option, copy the symbol info and edit it to
264 # create a hints file, then re-start the conversion at
265 # CollateSymbolsPass with this option enabled.
266 #SymbolHintsFileRule('symbol-hints.txt'),
268 # To force all symbols matching a regular expression to be
269 # converted as branches, add rules like the following:
270 #ForceBranchRegexpStrategyRule(r'branch.*'),
272 # To force all symbols matching a regular expression to be
273 # converted as tags, add rules like the following:
274 #ForceTagRegexpStrategyRule(r'tag.*'),
276 # To force all symbols matching a regular expression to be
277 # excluded from the conversion, add rules like the following:
278 #ExcludeRegexpStrategyRule(r'unknown-.*'),
280 # Sometimes people use "cvs import" to get their own source code
281 # into CVS. This practice creates a vendor branch 1.1.1 and
282 # imports the code onto the vendor branch as 1.1.1.1, then copies
283 # the same content to the trunk as version 1.1. Normally, such
284 # vendor branches are useless and they complicate the SVN history
285 # unnecessarily. The following rule excludes any branches that
286 # only existed as a vendor branch with a single import (leaving
287 # only the 1.1 revision). If you want to retain such branches,
288 # comment out the following line. (Please note that this rule
289 # does not exclude vendor *tags*, as they are not so easy to
291 ExcludeTrivialImportBranchRule(),
293 # To exclude all vendor branches (branches that had "cvs import"s
294 # on them but no other kinds of commits), uncomment the following
296 #ExcludeVendorBranchRule(),
298 # Usually you want this rule, to convert unambiguous symbols
299 # (symbols that were only ever used as tags or only ever used as
300 # branches in CVS) the same way they were used in CVS:
301 UnambiguousUsageRule(),
303 # If there was ever a commit on a symbol, then it cannot be
304 # converted as a tag. This rule causes all such symbols to be
305 # converted as branches. If you would like to resolve such
306 # ambiguities manually, comment out the following line:
307 BranchIfCommitsRule(),
309 # Last in the list can be a catch-all rule that is used for
310 # symbols that were not matched by any of the more specific rules
311 # above. (Assuming that BranchIfCommitsRule() was included above,
312 # then the symbols that are still indeterminate at this point can
313 # sensibly be converted as branches or tags.) Include at most one
314 # of these lines. If none of these catch-all rules are included,
315 # then the presence of any ambiguous symbols (that haven't been
316 # disambiguated above) is an error:
318 # Convert ambiguous symbols based on whether they were used more
319 # often as branches or as tags:
320 HeuristicStrategyRule(),
321 # Convert all ambiguous symbols as branches:
323 # Convert all ambiguous symbols as tags:
326 # The last rule is here to choose the preferred parent of branches
327 # and tags, that is, the line of development from which the symbol
329 HeuristicPreferredParentRule(),
332 # Specify a username to be used for commits for which CVS doesn't
333 # record the original author (for example, the creation of a branch).
334 # This should be a simple (unix-style) username, but it can be
335 # translated into a hg-style name by the author_transforms map.
336 ctx.username = 'cvs2svn'
338 # ctx.file_property_setters and ctx.revision_property_setters contain
339 # rules used to set the svn properties on files in the converted
340 # archive. For each file, the rules are tried one by one. Any rule
341 # can add or suppress one or more svn properties. Typically the rules
342 # will not overwrite properties set by a previous rule (though they
343 # are free to do so). ctx.file_property_setters should be used for
344 # properties that remain the same for the life of the file; these
345 # should implement FilePropertySetter. ctx.revision_property_setters
346 # should be used for properties that are allowed to vary from revision
347 # to revision; these should implement RevisionPropertySetter.
349 # Obviously, SVN properties per se are not interesting for a cvs2hg
350 # conversion, but some of these properties have side-effects that do
351 # affect the hg output. FIXME: Document this in more detail.
352 ctx.file_property_setters.extend([
353 # To read auto-props rules from a file, uncomment the following line
354 # and specify a filename. The boolean argument specifies whether
355 # case should be ignored when matching filenames to the filename
356 # patterns found in the auto-props file:
357 #AutoPropsPropertySetter(
358 # r'/home/username/.subversion/config',
362 # To read mime types from a file and use them to set svn:mime-type
363 # based on the filename extensions, uncomment the following line
364 # and specify a filename (see
365 # http://en.wikipedia.org/wiki/Mime.types for information about
367 #MimeMapper(r'/etc/mime.types', ignore_case=False),
369 # Omit the svn:eol-style property from any files that are listed
370 # as binary (i.e., mode '-kb') in CVS:
371 CVSBinaryFileEOLStyleSetter(),
373 # If the file is binary and its svn:mime-type property is not yet
374 # set, set svn:mime-type to 'application/octet-stream'.
375 CVSBinaryFileDefaultMimeTypeSetter(),
377 # To try to determine the eol-style from the mime type, uncomment
378 # the following line:
379 #EOLStyleFromMimeTypeSetter(),
381 # Choose one of the following lines to set the default
382 # svn:eol-style if none of the above rules applied. The argument
383 # is the svn:eol-style that should be applied, or None if no
384 # svn:eol-style should be set (i.e., the file should be treated as
387 # The default is to treat all files as binary unless one of the
388 # previous rules has determined otherwise, because this is the
389 # safest approach. However, if you have been diligent about
390 # marking binary files with -kb in CVS and/or you have used the
391 # above rules to definitely mark binary files as binary, then you
392 # might prefer to use 'native' as the default, as it is usually
393 # the most convenient setting for text files. Other possible
394 # options: 'CRLF', 'CR', 'LF'.
395 DefaultEOLStyleSetter(None),
396 #DefaultEOLStyleSetter('native'),
398 # Prevent svn:keywords from being set on files that have
399 # svn:eol-style unset.
400 SVNBinaryFileKeywordsPropertySetter(),
402 # If svn:keywords has not been set yet, set it based on the file's
404 KeywordsPropertySetter(config.SVN_KEYWORDS_VALUE),
406 # Set the svn:executable flag on any files that are marked in CVS as
408 ExecutablePropertySetter(),
410 # The following causes keywords to be untouched in binary files and
411 # collapsed in all text to be committed:
412 ConditionalPropertySetter(
413 cvs_file_is_binary, KeywordHandlingPropertySetter('untouched'),
415 KeywordHandlingPropertySetter('collapsed'),
418 ctx.revision_property_setters.extend([
421 # To skip the cleanup of temporary files, uncomment the following
423 #ctx.skip_cleanup = True
426 # In CVS, it is perfectly possible to make a single commit that
427 # affects more than one project or more than one branch of a single
428 # project. Subversion also allows such commits. Therefore, by
429 # default, when cvs2svn sees what looks like a cross-project or
430 # cross-branch CVS commit, it converts it into a
431 # cross-project/cross-branch Subversion commit.
433 # However, other tools and SCMs have trouble representing
434 # cross-project or cross-branch commits. (For example, Trac's Revtree
435 # plugin, http://www.trac-hacks.org/wiki/RevtreePlugin is confused by
436 # such commits.) Therefore, we provide the following two options to
437 # allow cross-project/cross-branch commits to be suppressed.
439 # cvs2hg only supports single-project conversions (multiple-project
440 # conversions wouldn't really make sense for hg anyway). So this
441 # option must be set to False:
442 ctx.cross_project_commits = False
444 # Mercurial itself doesn't allow commits that affect more than one
445 # branch, so this option must be set to False:
446 ctx.cross_branch_commits = False
448 # cvs2hg does not yet handle translating .cvsignore files into
449 # .hgignore content, so by default, the .cvsignore files are included
450 # inthe conversion output. If you would like to omit the .cvsignore
451 # files from the output, set this option to False:
452 ctx.keep_cvsignore = True
454 # By default, it is a fatal error for a CVS ",v" file to appear both
455 # inside and outside of an "Attic" subdirectory (this should never
456 # happen, but frequently occurs due to botched repository
457 # administration). If you would like to retain both versions of such
458 # files, change the following option to True, and the attic version of
459 # the file will be written to a subdirectory called "Attic" in the
461 ctx.retain_conflicting_attic_files = False
463 # CVS uses unix login names as author names whereas "hg fastimport"
464 # format requires author names to be of the form "foo <bar>". The
465 # default is to set the author to "cvsauthor <cvsauthor>".
466 # author_transforms can be used to map cvsauthor names (e.g.,
467 # "jrandom") to a true name and email address (e.g., "J. Random
468 # <jrandom@example.com>" for the example shown). All strings should
469 # be either Unicode strings (i.e., with "u" as a prefix) or 8-bit
470 # strings in the utf-8 encoding. The values can either be strings in
471 # the form "name <email>" or tuples (name, email). Please substitute
472 # your own project's usernames here to use with the author_transforms
473 # option of GitOutputOption below.
475 'jrandom' : ('J. Random', 'jrandom@example.com'),
476 'mhagger' : 'Michael Haggerty <mhagger@alum.mit.edu>',
477 'brane' : (u'Branko Čibej', 'brane@xbc.nu'),
478 'ringstrom' : 'Tobias Ringström <tobias@ringstrom.mine.nu>',
479 'dionisos' : (u'Erik Hülsmann', 'e.huelsmann@gmx.net'),
481 # This one will be used for commits for which CVS doesn't record
482 # the original author, as explained above.
483 'cvs2svn' : 'cvs2svn <admin@example.com>',
486 # This is the main option that causes cvs2svn to output to an "hg
487 # fastimport"-format dumpfile rather than to Subversion:
488 ctx.output_option = GitOutputOption(
489 # The file in which to write the "hg fastimport" stream that
490 # contains the changesets and branch/tag information:
491 os.path.join(ctx.tmpdir, 'hg-dump.dat'),
493 # Write the file contents inline in the "hg fastimport" stream,
494 # rather than using a separate blobs file (which "hg fastimport"
496 revision_writer=GitRevisionInlineWriter(
497 # cvs2hg uses either RCS's "co" command or CVS's "cvs co -p" to
498 # extract the content of file revisions. Here you can choose
499 # whether to use RCS (faster, but fails in some rare
500 # circumstances) or CVS (much slower, but more reliable).
501 #RCSRevisionReader(co_executable=r'co')
502 CVSRevisionReader(cvs_executable=r'cvs')
505 # Optional map from CVS author names to hg author names:
506 author_transforms=author_transforms,
509 # Change this option to True to turn on profiling of cvs2svn (for
510 # debugging purposes):
511 run_options.profiling = False
514 # Should CVSItem -> Changeset database files be memory mapped? In
515 # some tests, using memory mapping speeded up the overall conversion
516 # by about 5%. But this option can cause the conversion to fail with
517 # an out of memory error if the conversion computer runs out of
518 # virtual address space (e.g., when running a very large conversion on
519 # a 32-bit operating system). Therefore it is disabled by default.
520 # Uncomment the following line to allow these database files to be
522 #changeset_database.use_mmap_for_cvs_item_to_changeset_table = True
524 # Now set the project to be converted to hg. cvs2hg only supports
525 # single-project conversions, so this method must only be called once:
526 run_options.set_project(
527 # The filesystem path to the part of the CVS repository (*not* a
528 # CVS working copy) that should be converted. This may be a
529 # subdirectory (i.e., a module) within a larger CVS repository.
530 r'test-data/main-cvsrepos',
532 # A list of symbol transformations that can be used to rename
533 # symbols in this project.
535 # Use IgnoreSymbolTransforms like the following to completely
536 # ignore symbols matching a regular expression when parsing
537 # the CVS repository, for example to avoid warnings about
538 # branches with two names and to choose the preferred name.
539 # It is *not* recommended to use this instead of
540 # ExcludeRegexpStrategyRule; though more efficient,
541 # IgnoreSymbolTransforms are less flexible and don't exclude
542 # branches correctly. The argument is a Python-style regular
543 # expression that has to match the *whole* CVS symbol name:
544 #IgnoreSymbolTransform(r'nightly-build-tag-.*')
546 # RegexpSymbolTransforms transform symbols textually using a
547 # regular expression. The first argument is a Python regular
548 # expression pattern and the second is a replacement pattern.
549 # The pattern is matched against each symbol name. If it
550 # matches the whole symbol name, then the symbol name is
551 # replaced with the corresponding replacement text. The
552 # replacement can include substitution patterns (e.g., r'\1'
553 # or r'\g<name>'). Typically you will want to use raw strings
554 # (strings with a preceding 'r', like shown in the examples)
555 # for the regexp and its replacement to avoid backslash
556 # substitution within those strings.
557 #RegexpSymbolTransform(r'release-(\d+)_(\d+)',
559 #RegexpSymbolTransform(r'release-(\d+)_(\d+)_(\d+)',
560 # r'release-\1.\2.\3'),
562 # Simple 1:1 character replacements can also be done. The
563 # following transform, which converts backslashes into forward
564 # slashes, should usually be included:
565 ReplaceSubstringsSymbolTransform('\\','/'),
567 # This last rule eliminates leading, trailing, and repeated
568 # slashes within the output symbol names:
569 NormalizePathsSymbolTransform(),
572 # See the definition of global_symbol_strategy_rules above for a
573 # description of this option:
574 symbol_strategy_rules=global_symbol_strategy_rules,
576 # Exclude paths from the conversion. Should be relative to
577 # repository path and use forward slashes:
578 #exclude_paths=['file-to-exclude.txt,v', 'dir/to/exclude'],