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 Log
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.revision_manager import NullRevisionRecorder
91 from cvs2svn_lib.revision_manager import NullRevisionExcluder
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 InternalRevisionRecorder
95 from cvs2svn_lib.checkout_internal import InternalRevisionExcluder
96 from cvs2svn_lib.checkout_internal import InternalRevisionReader
97 from cvs2svn_lib.symbol_strategy import AllBranchRule
98 from cvs2svn_lib.symbol_strategy import AllTagRule
99 from cvs2svn_lib.symbol_strategy import BranchIfCommitsRule
100 from cvs2svn_lib.symbol_strategy import ExcludeRegexpStrategyRule
101 from cvs2svn_lib.symbol_strategy import ForceBranchRegexpStrategyRule
102 from cvs2svn_lib.symbol_strategy import ForceTagRegexpStrategyRule
103 from cvs2svn_lib.symbol_strategy import ExcludeTrivialImportBranchRule
104 from cvs2svn_lib.symbol_strategy import ExcludeVendorBranchRule
105 from cvs2svn_lib.symbol_strategy import HeuristicStrategyRule
106 from cvs2svn_lib.symbol_strategy import UnambiguousUsageRule
107 from cvs2svn_lib.symbol_strategy import HeuristicPreferredParentRule
108 from cvs2svn_lib.symbol_strategy import SymbolHintsFileRule
109 from cvs2svn_lib.symbol_transform import ReplaceSubstringsSymbolTransform
110 from cvs2svn_lib.symbol_transform import RegexpSymbolTransform
111 from cvs2svn_lib.symbol_transform import IgnoreSymbolTransform
112 from cvs2svn_lib.symbol_transform import NormalizePathsSymbolTransform
113 from cvs2svn_lib.property_setters import AutoPropsPropertySetter
114 from cvs2svn_lib.property_setters import CVSBinaryFileDefaultMimeTypeSetter
115 from cvs2svn_lib.property_setters import CVSBinaryFileEOLStyleSetter
116 from cvs2svn_lib.property_setters import DefaultEOLStyleSetter
117 from cvs2svn_lib.property_setters import EOLStyleFromMimeTypeSetter
118 from cvs2svn_lib.property_setters import ExecutablePropertySetter
119 from cvs2svn_lib.property_setters import KeywordsPropertySetter
120 from cvs2svn_lib.property_setters import MimeMapper
121 from cvs2svn_lib.property_setters import SVNBinaryFileKeywordsPropertySetter
123 # To choose the level of logging output, uncomment one of the
125 #Log().log_level = Log.WARN
126 #Log().log_level = Log.QUIET
127 Log().log_level = Log.NORMAL
128 #Log().log_level = Log.VERBOSE
129 #Log().log_level = Log.DEBUG
132 # The directory to use for temporary files:
133 ctx.tmpdir = r'cvs2svn-tmp'
135 # cvs2hg reads the contents of file revisions during OutputPass, so no
136 # revision recorder is needed:
137 ctx.revision_recorder = NullRevisionRecorder()
139 # cvs2hg does not need to keep track of what revisions will be
140 # excluded, so leave this option unchanged:
141 ctx.revision_excluder = NullRevisionExcluder()
143 # cvs2hg's revision reader is set via the GitOutputOption constructor,
144 # so leave this option set to None.
145 ctx.revision_reader = None
147 # Change the following line to True if the conversion should only
148 # include the trunk of the repository (i.e., all branches and tags
149 # should be omitted from the conversion):
150 ctx.trunk_only = False
152 # How to convert CVS author names, log messages, and filenames to
153 # Unicode. The first argument to CVSTextDecoder is a list of encoders
154 # that are tried in order in 'strict' mode until one of them succeeds.
155 # If none of those succeeds, then fallback_encoder (if it is
156 # specified) is used in lossy 'replace' mode. Setting a fallback
157 # encoder ensures that the encoder always succeeds, but it can cause
159 ctx.cvs_author_decoder = CVSTextDecoder(
165 #fallback_encoding='ascii'
167 ctx.cvs_log_decoder = CVSTextDecoder(
173 #fallback_encoding='ascii'
175 # You might want to be especially strict when converting filenames to
176 # Unicode (e.g., maybe not specify a fallback_encoding).
177 ctx.cvs_filename_decoder = CVSTextDecoder(
183 #fallback_encoding='ascii'
186 # Template for the commit message to be used for initial project
188 ctx.initial_project_commit_message = (
189 'Standard project directories initialized by cvs2svn.'
192 # Template for the commit message to be used for post commits, in
193 # which modifications to a vendor branch are copied back to trunk.
194 # This message can use '%(revnum)d' to include the SVN revision number
195 # of the revision that included the change to the vendor branch
196 # (admittedly rather pointless in a cvs2hg conversion).
197 ctx.post_commit_message = (
198 'This commit was generated by cvs2svn to track changes on a CVS '
202 # Template for the commit message to be used for commits in which
203 # symbols are created. This message can use '%(symbol_type)s' to
204 # include the type of the symbol ('branch' or 'tag') or
205 # '%(symbol_name)s' to include the name of the symbol.
206 ctx.symbol_commit_message = (
207 "This commit was manufactured by cvs2svn to create %(symbol_type)s "
211 # Some CVS clients for MacOS store resource fork data into CVS along
212 # with the file contents itself by wrapping it all up in a container
213 # format called "AppleSingle". Subversion currently does not support
214 # MacOS resource forks. Nevertheless, sometimes the resource fork
215 # information is not necessary and can be discarded. Set the
216 # following option to True if you would like cvs2svn to identify files
217 # whose contents are encoded in AppleSingle format, and discard all
218 # but the data fork for such files before committing them to
219 # Subversion. (Please note that AppleSingle contents are identified
220 # by the AppleSingle magic number as the first four bytes of the file.
221 # This check is not failproof, so only set this option if you think
223 ctx.decode_apple_single = False
225 # This option can be set to the name of a filename to which are stored
226 # statistics and conversion decisions about the CVS symbols.
227 ctx.symbol_info_filename = None
228 #ctx.symbol_info_filename = 'symbol-info.txt'
230 # cvs2svn uses "symbol strategy rules" to help decide how to handle
231 # CVS symbols. The rules in a project's symbol_strategy_rules are
232 # applied in order, and each rule is allowed to modify the symbol.
233 # The result (after each of the rules has been applied) is used for
236 # 1. A CVS symbol might be used as a tag in one file and as a branch
237 # in another file. cvs2svn has to decide whether to convert such a
238 # symbol as a tag or as a branch. cvs2svn uses a series of
239 # heuristic rules to decide how to convert a symbol. The user can
240 # override the default rules for specific symbols or symbols
241 # matching regular expressions.
243 # 2. cvs2svn is also capable of excluding symbols from the conversion
244 # (provided no other symbols depend on them.
246 # 3. CVS does not record unambiguously the line of development from
247 # which a symbol sprouted. cvs2svn uses a heuristic to choose a
248 # symbol's "preferred parents".
250 # The standard branch/tag/exclude StrategyRules do not change a symbol
251 # that has already been processed by an earlier rule, so in effect the
252 # first matching rule is the one that is used.
254 global_symbol_strategy_rules = [
255 # It is possible to specify manually exactly how symbols should be
256 # converted and what line of development should be used as the
257 # preferred parent. To do so, create a file containing the symbol
258 # hints and enable the following option.
260 # The format of the hints file is described in the documentation
261 # for the --symbol-hints command-line option. The file output by
262 # the --write-symbol-info (i.e., ctx.symbol_info_filename) option
263 # is in the same format. The simplest way to use this option is
264 # to run the conversion through CollateSymbolsPass with
265 # --write-symbol-info option, copy the symbol info and edit it to
266 # create a hints file, then re-start the conversion at
267 # CollateSymbolsPass with this option enabled.
268 #SymbolHintsFileRule('symbol-hints.txt'),
270 # To force all symbols matching a regular expression to be
271 # converted as branches, add rules like the following:
272 #ForceBranchRegexpStrategyRule(r'branch.*'),
274 # To force all symbols matching a regular expression to be
275 # converted as tags, add rules like the following:
276 #ForceTagRegexpStrategyRule(r'tag.*'),
278 # To force all symbols matching a regular expression to be
279 # excluded from the conversion, add rules like the following:
280 #ExcludeRegexpStrategyRule(r'unknown-.*'),
282 # Sometimes people use "cvs import" to get their own source code
283 # into CVS. This practice creates a vendor branch 1.1.1 and
284 # imports the code onto the vendor branch as 1.1.1.1, then copies
285 # the same content to the trunk as version 1.1. Normally, such
286 # vendor branches are useless and they complicate the SVN history
287 # unnecessarily. The following rule excludes any branches that
288 # only existed as a vendor branch with a single import (leaving
289 # only the 1.1 revision). If you want to retain such branches,
290 # comment out the following line. (Please note that this rule
291 # does not exclude vendor *tags*, as they are not so easy to
293 ExcludeTrivialImportBranchRule(),
295 # To exclude all vendor branches (branches that had "cvs import"s
296 # on them bug no other kinds of commits), uncomment the following
298 #ExcludeVendorBranchRule(),
300 # Usually you want this rule, to convert unambiguous symbols
301 # (symbols that were only ever used as tags or only ever used as
302 # branches in CVS) the same way they were used in CVS:
303 UnambiguousUsageRule(),
305 # If there was ever a commit on a symbol, then it cannot be
306 # converted as a tag. This rule causes all such symbols to be
307 # converted as branches. If you would like to resolve such
308 # ambiguities manually, comment out the following line:
309 BranchIfCommitsRule(),
311 # Last in the list can be a catch-all rule that is used for
312 # symbols that were not matched by any of the more specific rules
313 # above. (Assuming that BranchIfCommitsRule() was included above,
314 # then the symbols that are still indeterminate at this point can
315 # sensibly be converted as branches or tags.) Include at most one
316 # of these lines. If none of these catch-all rules are included,
317 # then the presence of any ambiguous symbols (that haven't been
318 # disambiguated above) is an error:
320 # Convert ambiguous symbols based on whether they were used more
321 # often as branches or as tags:
322 HeuristicStrategyRule(),
323 # Convert all ambiguous symbols as branches:
325 # Convert all ambiguous symbols as tags:
328 # The last rule is here to choose the preferred parent of branches
329 # and tags, that is, the line of development from which the symbol
331 HeuristicPreferredParentRule(),
334 # Specify a username to be used for commits for which CVS doesn't
335 # record the original author (for example, the creation of a branch).
336 # This should be a simple (unix-style) username, but it can be
337 # translated into a hg-style name by the author_transforms map.
338 ctx.username = 'cvs2svn'
340 # ctx.svn_property_setters contains a list of rules used to set the
341 # svn properties on files in the converted archive. For each file,
342 # the rules are tried one by one. Any rule can add or suppress one or
343 # more svn properties. Typically the rules will not overwrite
344 # properties set by a previous rule (though they are free to do so).
346 # Obviously, SVN properties per se are not interesting for a cvs2hg
347 # conversion, but some of these properties have side-effects that do
348 # affect the hg output. FIXME: Document this in more detail.
349 ctx.svn_property_setters.extend([
350 # To read auto-props rules from a file, uncomment the following line
351 # and specify a filename. The boolean argument specifies whether
352 # case should be ignored when matching filenames to the filename
353 # patterns found in the auto-props file:
354 #AutoPropsPropertySetter(
355 # r'/home/username/.subversion/config',
359 # To read mime types from a file, uncomment the following line and
360 # specify a filename:
361 #MimeMapper(r'/etc/mime.types'),
363 # Omit the svn:eol-style property from any files that are listed
364 # as binary (i.e., mode '-kb') in CVS:
365 CVSBinaryFileEOLStyleSetter(),
367 # If the file is binary and its svn:mime-type property is not yet
368 # set, set svn:mime-type to 'application/octet-stream'.
369 CVSBinaryFileDefaultMimeTypeSetter(),
371 # To try to determine the eol-style from the mime type, uncomment
372 # the following line:
373 #EOLStyleFromMimeTypeSetter(),
375 # Choose one of the following lines to set the default
376 # svn:eol-style if none of the above rules applied. The argument
377 # is the svn:eol-style that should be applied, or None if no
378 # svn:eol-style should be set (i.e., the file should be treated as
381 # The default is to treat all files as binary unless one of the
382 # previous rules has determined otherwise, because this is the
383 # safest approach. However, if you have been diligent about
384 # marking binary files with -kb in CVS and/or you have used the
385 # above rules to definitely mark binary files as binary, then you
386 # might prefer to use 'native' as the default, as it is usually
387 # the most convenient setting for text files. Other possible
388 # options: 'CRLF', 'CR', 'LF'.
389 DefaultEOLStyleSetter(None),
390 #DefaultEOLStyleSetter('native'),
392 # Prevent svn:keywords from being set on files that have
393 # svn:eol-style unset.
394 SVNBinaryFileKeywordsPropertySetter(),
396 # If svn:keywords has not been set yet, set it based on the file's
398 KeywordsPropertySetter(config.SVN_KEYWORDS_VALUE),
400 # Set the svn:executable flag on any files that are marked in CVS as
402 ExecutablePropertySetter(),
406 # To skip the cleanup of temporary files, uncomment the following
408 #ctx.skip_cleanup = True
411 # In CVS, it is perfectly possible to make a single commit that
412 # affects more than one project or more than one branch of a single
413 # project. Subversion also allows such commits. Therefore, by
414 # default, when cvs2svn sees what looks like a cross-project or
415 # cross-branch CVS commit, it converts it into a
416 # cross-project/cross-branch Subversion commit.
418 # However, other tools and SCMs have trouble representing
419 # cross-project or cross-branch commits. (For example, Trac's Revtree
420 # plugin, http://www.trac-hacks.org/wiki/RevtreePlugin is confused by
421 # such commits.) Therefore, we provide the following two options to
422 # allow cross-project/cross-branch commits to be suppressed.
424 # cvs2hg only supports single-project conversions (multiple-project
425 # conversions wouldn't really make sense for hg anyway). So this
426 # option must be set to False:
427 ctx.cross_project_commits = False
429 # Mercurial itself doesn't allow commits that affect more than one
430 # branch, so this option must be set to False:
431 ctx.cross_branch_commits = False
433 # cvs2hg does not yet handle translating .cvsignore files into
434 # .hgignore content, so by default, the .cvsignore files are included
435 # inthe conversion output. If you would like to omit the .cvsignore
436 # files from the output, set this option to False:
437 ctx.keep_cvsignore = True
439 # By default, it is a fatal error for a CVS ",v" file to appear both
440 # inside and outside of an "Attic" subdirectory (this should never
441 # happen, but frequently occurs due to botched repository
442 # administration). If you would like to retain both versions of such
443 # files, change the following option to True, and the attic version of
444 # the file will be written to a subdirectory called "Attic" in the
446 ctx.retain_conflicting_attic_files = False
448 # CVS uses unix login names as author names whereas "hg fastimport"
449 # format requires author names to be of the form "foo <bar>". The
450 # default is to set the author to "cvsauthor <cvsauthor>".
451 # author_transforms can be used to map cvsauthor names (e.g.,
452 # "jrandom") to a true name and email address (e.g., "J. Random
453 # <jrandom@example.com>" for the example shown). All values should be
454 # either Unicode strings (i.e., with "u" as a prefix) or 8-bit strings
455 # in the utf-8 encoding. Please substitute your own project's
456 # usernames here to use with the author_transforms option of
457 # GitOutputOption below.
459 'jrandom' : ('J. Random', 'jrandom@example.com'),
460 'mhagger' : ('Michael Haggerty', 'mhagger@alum.mit.edu'),
461 'brane' : (u'Branko Čibej', 'brane@xbc.nu'),
462 'ringstrom' : ('Tobias Ringström', 'tobias@ringstrom.mine.nu'),
463 'dionisos' : (u'Erik Hülsmann', 'e.huelsmann@gmx.net'),
465 # This one will be used for commits for which CVS doesn't record
466 # the original author, as explained above.
467 'cvs2svn' : ('cvs2svn', 'admin@example.com'),
470 # This is the main option that causes cvs2svn to output to an "hg
471 # fastimport"-format dumpfile rather than to Subversion:
472 ctx.output_option = GitOutputOption(
473 # The file in which to write the "hg fastimport" stream that
474 # contains the changesets and branch/tag information:
475 os.path.join(ctx.tmpdir, 'hg-dump.dat'),
477 # Write the file contents inline in the "hg fastimport" stream,
478 # rather than using a separate blobs file (which "hg fastimport"
480 revision_writer=GitRevisionInlineWriter(
481 # cvs2hg uses either RCS's "co" command or CVS's "cvs co -p" to
482 # extract the content of file revisions. Here you can choose
483 # whether to use RCS (faster, but fails in some rare
484 # circumstances) or CVS (much slower, but more reliable).
485 #RCSRevisionReader(co_executable=r'co')
486 CVSRevisionReader(cvs_executable=r'cvs')
489 # Optional map from CVS author names to hg author names:
490 author_transforms=author_transforms,
493 # Change this option to True to turn on profiling of cvs2svn (for
494 # debugging purposes):
495 run_options.profiling = False
498 # Should CVSItem -> Changeset database files be memory mapped? In
499 # some tests, using memory mapping speeded up the overall conversion
500 # by about 5%. But this option can cause the conversion to fail with
501 # an out of memory error if the conversion computer runs out of
502 # virtual address space (e.g., when running a very large conversion on
503 # a 32-bit operating system). Therefore it is disabled by default.
504 # Uncomment the following line to allow these database files to be
506 #changeset_database.use_mmap_for_cvs_item_to_changeset_table = True
508 # Now set the project to be converted to hg. cvs2hg only supports
509 # single-project conversions, so this method must only be called once:
510 run_options.set_project(
511 # The filesystem path to the part of the CVS repository (*not* a
512 # CVS working copy) that should be converted. This may be a
513 # subdirectory (i.e., a module) within a larger CVS repository.
514 r'test-data/main-cvsrepos',
516 # A list of symbol transformations that can be used to rename
517 # symbols in this project.
519 # Use IgnoreSymbolTransforms like the following to completely
520 # ignore symbols matching a regular expression when parsing
521 # the CVS repository, for example to avoid warnings about
522 # branches with two names and to choose the preferred name.
523 # It is *not* recommended to use this instead of
524 # ExcludeRegexpStrategyRule; though more efficient,
525 # IgnoreSymbolTransforms are less flexible and don't exclude
526 # branches correctly. The argument is a Python-style regular
527 # expression that has to match the *whole* CVS symbol name:
528 #IgnoreSymbolTransform(r'nightly-build-tag-.*')
530 # RegexpSymbolTransforms transform symbols textually using a
531 # regular expression. The first argument is a Python regular
532 # expression pattern and the second is a replacement pattern.
533 # The pattern is matched against each symbol name. If it
534 # matches the whole symbol name, then the symbol name is
535 # replaced with the corresponding replacement text. The
536 # replacement can include substitution patterns (e.g., r'\1'
537 # or r'\g<name>'). Typically you will want to use raw strings
538 # (strings with a preceding 'r', like shown in the examples)
539 # for the regexp and its replacement to avoid backslash
540 # substitution within those strings.
541 #RegexpSymbolTransform(r'release-(\d+)_(\d+)',
543 #RegexpSymbolTransform(r'release-(\d+)_(\d+)_(\d+)',
544 # r'release-\1.\2.\3'),
546 # Simple 1:1 character replacements can also be done. The
547 # following transform, which converts backslashes into forward
548 # slashes, should usually be included:
549 ReplaceSubstringsSymbolTransform('\\','/'),
551 # This last rule eliminates leading, trailing, and repeated
552 # slashes within the output symbol names:
553 NormalizePathsSymbolTransform(),
556 # See the definition of global_symbol_strategy_rules above for a
557 # description of this option:
558 symbol_strategy_rules=global_symbol_strategy_rules,