1 # (Be in -*- mode: python; coding: utf-8 -*- mode.)
3 # ====================================================================
4 # Copyright (c) 2006-2008 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 a couple of
38 # 1. "hg fastimport" does not support blobs, so the contents of the
39 # revisions are output inline rather than in a separate blobs file.
40 # This increases the size of the output, because file contents that
41 # appear identically on multiple branches have to be output
44 # 2. Mercurial only allows a revsion to have two parents (not an
45 # arbitrary number, as allowed by git). This affects the way that
46 # cvs2svn handles a symbol that includes content from multiple
47 # parent branches. For output to Mercurial, files are only copied
48 # from one parent branch in each commit (therefore potentially
49 # taking multiple commits to create the symbol).
51 # Many options do not have defaults, so it is easier to copy this file
52 # and modify what you need rather than creating a new options file
53 # from scratch. This file is in Python syntax, but you don't need to
54 # know Python to modify it. But if you *do* know Python, then you
55 # will be happy to know that you can use arbitary Python constructs to
56 # do fancy configuration tricks.
58 # But please be aware of the following:
60 # * In many places, leading whitespace is significant in Python (it is
61 # used instead of curly braces to group statements together).
62 # Therefore, if you don't know what you are doing, it is best to
63 # leave the whitespace as it is.
65 # * In normal strings, Python treats a backslash ("\") as an escape
66 # character. Therefore, if you want to specify a string that
67 # contains a backslash, you need either to escape the backslash with
68 # another backslash ("\\"), or use a "raw string", as in one if the
69 # following equivalent examples:
71 # ctx.sort_executable = 'c:\\windows\\system32\\sort.exe'
72 # ctx.sort_executable = r'c:\windows\system32\sort.exe'
74 # See http://docs.python.org/tutorial/introduction.html#strings for
77 # Two identifiers will have been defined before this file is executed,
78 # and can be used freely within this file:
80 # ctx -- a Ctx object (see cvs2svn_lib/context.py), which holds
81 # many configuration options
83 # run_options -- an instance of the GitRunOptions class (see
84 # cvs2svn_lib/git_run_options.py), which holds some variables
85 # governing how cvs2git is run
88 # Import some modules that are used in setting the options:
91 from cvs2svn_lib import config
92 from cvs2svn_lib import changeset_database
93 from cvs2svn_lib.common import CVSTextDecoder
94 from cvs2svn_lib.log import Log
95 from cvs2svn_lib.project import Project
96 from cvs2svn_lib.git_revision_recorder import GitRevisionRecorder
97 from cvs2svn_lib.git_output_option import GitRevisionInlineWriter
98 from cvs2svn_lib.git_output_option import GitOutputOption
99 from cvs2svn_lib.revision_manager import NullRevisionRecorder
100 from cvs2svn_lib.revision_manager import NullRevisionExcluder
101 from cvs2svn_lib.rcs_revision_manager import RCSRevisionReader
102 from cvs2svn_lib.cvs_revision_manager import CVSRevisionReader
103 from cvs2svn_lib.checkout_internal import InternalRevisionRecorder
104 from cvs2svn_lib.checkout_internal import InternalRevisionExcluder
105 from cvs2svn_lib.checkout_internal import InternalRevisionReader
106 from cvs2svn_lib.symbol_strategy import AllBranchRule
107 from cvs2svn_lib.symbol_strategy import AllTagRule
108 from cvs2svn_lib.symbol_strategy import BranchIfCommitsRule
109 from cvs2svn_lib.symbol_strategy import ExcludeRegexpStrategyRule
110 from cvs2svn_lib.symbol_strategy import ForceBranchRegexpStrategyRule
111 from cvs2svn_lib.symbol_strategy import ForceTagRegexpStrategyRule
112 from cvs2svn_lib.symbol_strategy import ExcludeTrivialImportBranchRule
113 from cvs2svn_lib.symbol_strategy import ExcludeVendorBranchRule
114 from cvs2svn_lib.symbol_strategy import HeuristicStrategyRule
115 from cvs2svn_lib.symbol_strategy import UnambiguousUsageRule
116 from cvs2svn_lib.symbol_strategy import HeuristicPreferredParentRule
117 from cvs2svn_lib.symbol_strategy import SymbolHintsFileRule
118 from cvs2svn_lib.symbol_transform import ReplaceSubstringsSymbolTransform
119 from cvs2svn_lib.symbol_transform import RegexpSymbolTransform
120 from cvs2svn_lib.symbol_transform import IgnoreSymbolTransform
121 from cvs2svn_lib.symbol_transform import NormalizePathsSymbolTransform
122 from cvs2svn_lib.property_setters import AutoPropsPropertySetter
123 from cvs2svn_lib.property_setters import CVSBinaryFileDefaultMimeTypeSetter
124 from cvs2svn_lib.property_setters import CVSBinaryFileEOLStyleSetter
125 from cvs2svn_lib.property_setters import CVSRevisionNumberSetter
126 from cvs2svn_lib.property_setters import DefaultEOLStyleSetter
127 from cvs2svn_lib.property_setters import EOLStyleFromMimeTypeSetter
128 from cvs2svn_lib.property_setters import ExecutablePropertySetter
129 from cvs2svn_lib.property_setters import KeywordsPropertySetter
130 from cvs2svn_lib.property_setters import MimeMapper
131 from cvs2svn_lib.property_setters import SVNBinaryFileKeywordsPropertySetter
133 # To choose the level of logging output, uncomment one of the
135 #Log().log_level = Log.WARN
136 #Log().log_level = Log.QUIET
137 Log().log_level = Log.NORMAL
138 #Log().log_level = Log.VERBOSE
139 #Log().log_level = Log.DEBUG
142 # cvs2hg reads the contents of file revisions during OutputPass, so no
143 # revision recorder is needed:
144 ctx.revision_recorder = NullRevisionRecorder()
146 # cvs2hg does not need to keep track of what revisions will be
147 # excluded, so leave this option unchanged:
148 ctx.revision_excluder = NullRevisionExcluder()
150 # cvs2hg's revision reader is set via the GitOutputOption constructor,
151 # so leave this option set to None.
152 ctx.revision_reader = None
154 # Set the name (and optionally the path) of some other executables
155 # required by cvs2svn:
156 ctx.sort_executable = r'sort'
158 # Change the following line to True if the conversion should only
159 # include the trunk of the repository (i.e., all branches and tags
160 # should be omitted from the conversion):
161 ctx.trunk_only = False
163 # How to convert CVS author names, log messages, and filenames to
164 # unicode. The first argument to CVSTextDecoder is a list of encoders
165 # that are tried in order in 'strict' mode until one of them succeeds.
166 # If none of those succeeds, then fallback_encoder (if it is
167 # specified) is used in lossy 'replace' mode. Setting a fallback
168 # encoder ensures that the encoder always succeeds, but it can cause
170 ctx.cvs_author_decoder = CVSTextDecoder(
176 #fallback_encoding='ascii'
178 ctx.cvs_log_decoder = CVSTextDecoder(
184 #fallback_encoding='ascii'
186 # You might want to be especially strict when converting filenames to
187 # unicode (e.g., maybe not specify a fallback_encoding).
188 ctx.cvs_filename_decoder = CVSTextDecoder(
194 #fallback_encoding='ascii'
197 # Template for the commit message to be used for initial project
199 ctx.initial_project_commit_message = (
200 'Standard project directories initialized by cvs2svn.'
203 # Template for the commit message to be used for post commits, in
204 # which modifications to a vendor branch are copied back to trunk.
205 # This message can use '%(revnum)d' to include the SVN revision number
206 # of the revision that included the change to the vendor branch
207 # (admittedly rather pointless in a cvs2hg conversion).
208 ctx.post_commit_message = (
209 'This commit was generated by cvs2svn to track changes on a CVS '
213 # Template for the commit message to be used for commits in which
214 # symbols are created. This message can use '%(symbol_type)d' to
215 # include the type of the symbol ('branch' or 'tag') or
216 # '%(symbol_name)' to include the name of the symbol.
217 ctx.symbol_commit_message = (
218 "This commit was manufactured by cvs2svn to create %(symbol_type)s "
222 # Some CVS clients for MacOS store resource fork data into CVS along
223 # with the file contents itself by wrapping it all up in a container
224 # format called "AppleSingle". Subversion currently does not support
225 # MacOS resource forks. Nevertheless, sometimes the resource fork
226 # information is not necessary and can be discarded. Set the
227 # following option to True if you would like cvs2svn to identify files
228 # whose contents are encoded in AppleSingle format, and discard all
229 # but the data fork for such files before committing them to
230 # Subversion. (Please note that AppleSingle contents are identified
231 # by the AppleSingle magic number as the first four bytes of the file.
232 # This check is not failproof, so only set this option if you think
234 ctx.decode_apple_single = False
236 # This option can be set to the name of a filename to which are stored
237 # statistics and conversion decisions about the CVS symbols.
238 ctx.symbol_info_filename = None
239 #ctx.symbol_info_filename = 'symbol-info.txt'
241 # cvs2svn uses "symbol strategy rules" to help decide how to handle
242 # CVS symbols. The rules in a project's symbol_strategy_rules are
243 # applied in order, and each rule is allowed to modify the symbol.
244 # The result (after each of the rules has been applied) is used for
247 # 1. A CVS symbol might be used as a tag in one file and as a branch
248 # in another file. cvs2svn has to decide whether to convert such a
249 # symbol as a tag or as a branch. cvs2svn uses a series of
250 # heuristic rules to decide how to convert a symbol. The user can
251 # override the default rules for specific symbols or symbols
252 # matching regular expressions.
254 # 2. cvs2svn is also capable of excluding symbols from the conversion
255 # (provided no other symbols depend on them.
257 # 3. CVS does not record unambiguously the line of development from
258 # which a symbol sprouted. cvs2svn uses a heuristic to choose a
259 # symbol's "preferred parents".
261 # The standard branch/tag/exclude StrategyRules do not change a symbol
262 # that has already been processed by an earlier rule, so in effect the
263 # first matching rule is the one that is used.
265 global_symbol_strategy_rules = [
266 # It is possible to specify manually exactly how symbols should be
267 # converted and what line of development should be used as the
268 # preferred parent. To do so, create a file containing the symbol
269 # hints and enable the following option.
271 # The format of the hints file is described in the documentation
272 # for the --symbol-hints command-line option. The file output by
273 # the --write-symbol-info (i.e., ctx.symbol_info_filename) option
274 # is in the same format. The simplest way to use this option is
275 # to run the conversion through CollateSymbolsPass with
276 # --write-symbol-info option, copy the symbol info and edit it to
277 # create a hints file, then re-start the conversion at
278 # CollateSymbolsPass with this option enabled.
279 #SymbolHintsFileRule('symbol-hints.txt'),
281 # To force all symbols matching a regular expression to be
282 # converted as branches, add rules like the following:
283 #ForceBranchRegexpStrategyRule(r'branch.*'),
285 # To force all symbols matching a regular expression to be
286 # converted as tags, add rules like the following:
287 #ForceTagRegexpStrategyRule(r'tag.*'),
289 # To force all symbols matching a regular expression to be
290 # excluded from the conversion, add rules like the following:
291 #ExcludeRegexpStrategyRule(r'unknown-.*'),
293 # Sometimes people use "cvs import" to get their own source code
294 # into CVS. This practice creates a vendor branch 1.1.1 and
295 # imports the code onto the vendor branch as 1.1.1.1, then copies
296 # the same content to the trunk as version 1.1. Normally, such
297 # vendor branches are useless and they complicate the SVN history
298 # unnecessarily. The following rule excludes any branches that
299 # only existed as a vendor branch with a single import (leaving
300 # only the 1.1 revision). If you want to retain such branches,
301 # comment out the following line. (Please note that this rule
302 # does not exclude vendor *tags*, as they are not so easy to
304 ExcludeTrivialImportBranchRule(),
306 # To exclude all vendor branches (branches that had "cvs import"s
307 # on them bug no other kinds of commits), uncomment the following
309 #ExcludeVendorBranchRule(),
311 # Usually you want this rule, to convert unambiguous symbols
312 # (symbols that were only ever used as tags or only ever used as
313 # branches in CVS) the same way they were used in CVS:
314 UnambiguousUsageRule(),
316 # If there was ever a commit on a symbol, then it cannot be
317 # converted as a tag. This rule causes all such symbols to be
318 # converted as branches. If you would like to resolve such
319 # ambiguities manually, comment out the following line:
320 BranchIfCommitsRule(),
322 # Last in the list can be a catch-all rule that is used for
323 # symbols that were not matched by any of the more specific rules
324 # above. (Assuming that BranchIfCommitsRule() was included above,
325 # then the symbols that are still indeterminate at this point can
326 # sensibly be converted as branches or tags.) Include at most one
327 # of these lines. If none of these catch-all rules are included,
328 # then the presence of any ambiguous symbols (that haven't been
329 # disambiguated above) is an error:
331 # Convert ambiguous symbols based on whether they were used more
332 # often as branches or as tags:
333 HeuristicStrategyRule(),
334 # Convert all ambiguous symbols as branches:
336 # Convert all ambiguous symbols as tags:
339 # The last rule is here to choose the preferred parent of branches
340 # and tags, that is, the line of development from which the symbol
342 HeuristicPreferredParentRule(),
345 # Specify a username to be used for commits for which CVS doesn't
346 # record the original author (for example, the creation of a branch).
347 # This should be a simple (unix-style) username, but it can be
348 # translated into a hg-style name by the author_transforms map.
349 ctx.username = 'cvs2svn'
351 # ctx.svn_property_setters contains a list of rules used to set the
352 # svn properties on files in the converted archive. For each file,
353 # the rules are tried one by one. Any rule can add or suppress one or
354 # more svn properties. Typically the rules will not overwrite
355 # properties set by a previous rule (though they are free to do so).
357 # Obviously, SVN properties per se are not interesting for a cvs2hg
358 # conversion, but some of these properties have side-effects that do
359 # affect the hg output. FIXME: Document this in more detail.
360 ctx.svn_property_setters.extend([
361 # To read auto-props rules from a file, uncomment the following line
362 # and specify a filename. The boolean argument specifies whether
363 # case should be ignored when matching filenames to the filename
364 # patterns found in the auto-props file:
365 #AutoPropsPropertySetter(
366 # r'/home/username/.subversion/config',
370 # To read mime types from a file, uncomment the following line and
371 # specify a filename:
372 #MimeMapper(r'/etc/mime.types'),
374 # Omit the svn:eol-style property from any files that are listed
375 # as binary (i.e., mode '-kb') in CVS:
376 CVSBinaryFileEOLStyleSetter(),
378 # If the file is binary and its svn:mime-type property is not yet
379 # set, set svn:mime-type to 'application/octet-stream'.
380 CVSBinaryFileDefaultMimeTypeSetter(),
382 # To try to determine the eol-style from the mime type, uncomment
383 # the following line:
384 #EOLStyleFromMimeTypeSetter(),
386 # Choose one of the following lines to set the default
387 # svn:eol-style if none of the above rules applied. The argument
388 # is the svn:eol-style that should be applied, or None if no
389 # svn:eol-style should be set (i.e., the file should be treated as
392 # The default is to treat all files as binary unless one of the
393 # previous rules has determined otherwise, because this is the
394 # safest approach. However, if you have been diligent about
395 # marking binary files with -kb in CVS and/or you have used the
396 # above rules to definitely mark binary files as binary, then you
397 # might prefer to use 'native' as the default, as it is usually
398 # the most convenient setting for text files. Other possible
399 # options: 'CRLF', 'CR', 'LF'.
400 DefaultEOLStyleSetter(None),
401 #DefaultEOLStyleSetter('native'),
403 # Prevent svn:keywords from being set on files that have
404 # svn:eol-style unset.
405 SVNBinaryFileKeywordsPropertySetter(),
407 # If svn:keywords has not been set yet, set it based on the file's
409 KeywordsPropertySetter(config.SVN_KEYWORDS_VALUE),
411 # Set the svn:executable flag on any files that are marked in CVS as
413 ExecutablePropertySetter(),
417 # The directory to use for temporary files:
418 ctx.tmpdir = r'cvs2svn-tmp'
420 # To skip the cleanup of temporary files, uncomment the following
422 #ctx.skip_cleanup = True
425 # In CVS, it is perfectly possible to make a single commit that
426 # affects more than one project or more than one branch of a single
427 # project. Subversion also allows such commits. Therefore, by
428 # default, when cvs2svn sees what looks like a cross-project or
429 # cross-branch CVS commit, it converts it into a
430 # cross-project/cross-branch Subversion commit.
432 # However, other tools and SCMs have trouble representing
433 # cross-project or cross-branch commits. (For example, Trac's Revtree
434 # plugin, http://www.trac-hacks.org/wiki/RevtreePlugin is confused by
435 # such commits.) Therefore, we provide the following two options to
436 # allow cross-project/cross-branch commits to be suppressed.
438 # cvs2hg only supports single-project conversions (multiple-project
439 # conversions wouldn't really make sense for hg anyway). So this
440 # option must be set to False:
441 ctx.cross_project_commits = False
443 # Mercurial itself doesn't allow commits that affect more than one
444 # branch, so this option must be set to False:
445 ctx.cross_branch_commits = False
447 # cvs2hg does not yet handle translating .cvsignore files into
448 # .hgignore content, so by default, the .cvsignore files are included
449 # inthe conversion output. If you would like to omit the .cvsignore
450 # files from the output, set this option to False:
451 ctx.keep_cvsignore = True
453 # By default, it is a fatal error for a CVS ",v" file to appear both
454 # inside and outside of an "Attic" subdirectory (this should never
455 # happen, but frequently occurs due to botched repository
456 # administration). If you would like to retain both versions of such
457 # files, change the following option to True, and the attic version of
458 # the file will be written to a subdirectory called "Attic" in the
460 ctx.retain_conflicting_attic_files = False
462 # CVS uses unix login names as author names whereas "hg fastimport"
463 # format requires author names to be of the form "foo <bar>". The
464 # default is to set the author to "cvsauthor <cvsauthor>".
465 # author_transforms can be used to map cvsauthor names (e.g.,
466 # "jrandom") to a true name and email address (e.g., "J. Random
467 # <jrandom@example.com>" for the example shown). All values should be
468 # either 16-bit strings (i.e., with "u" as a prefix) or 8-bit strings
469 # in the utf-8 encoding. Please substitute your own project's
470 # usernames here to use with the author_transforms option of
471 # GitOutputOption below.
473 'jrandom' : ('J. Random', 'jrandom@example.com'),
474 'mhagger' : ('Michael Haggerty', 'mhagger@alum.mit.edu'),
475 'brane' : (u'Branko Čibej', 'brane@xbc.nu'),
476 'ringstrom' : ('Tobias Ringström', 'tobias@ringstrom.mine.nu'),
477 'dionisos' : (u'Erik Hülsmann', 'e.huelsmann@gmx.net'),
479 # This one will be used for commits for which CVS doesn't record
480 # the original author, as explained above.
481 'cvs2svn' : ('cvs2svn', 'admin@example.com'),
484 # This is the main option that causes cvs2svn to output to an "hg
485 # fastimport"-format dumpfile rather than to Subversion:
486 ctx.output_option = GitOutputOption(
487 # The file in which to write the "hg fastimport" stream that
488 # contains the changesets and branch/tag information:
489 'cvs2svn-tmp/hg-dump.dat',
491 # Write the file contents inline in the "hg fastimport" stream,
492 # rather than using a separate blobs file (which "hg fastimport"
494 revision_writer=GitRevisionInlineWriter(
495 # cvs2hg uses either RCS's "co" command or CVS's "cvs co -p" to
496 # extract the content of file revisions. Here you can choose
497 # whether to use RCS (faster, but fails in some rare
498 # circumstances) or CVS (much slower, but more reliable).
499 #RCSRevisionReader(co_executable=r'co')
500 CVSRevisionReader(cvs_executable=r'cvs')
503 # This option limits the number of revisions that are merged with
504 # the main parent in any commit. For Mercurial output, this
505 # should be set to 1 (see the comment at the top of this file for
509 # Optional map from CVS author names to hg author names:
510 author_transforms=author_transforms,
513 # Change this option to True to turn on profiling of cvs2svn (for
514 # debugging purposes):
515 run_options.profiling = False
518 # Should CVSItem -> Changeset database files be memory mapped? In
519 # some tests, using memory mapping speeded up the overall conversion
520 # by about 5%. But this option can cause the conversion to fail with
521 # an out of memory error if the conversion computer runs out of
522 # virtual address space (e.g., when running a very large conversion on
523 # a 32-bit operating system). Therefore it is disabled by default.
524 # Uncomment the following line to allow these database files to be
526 #changeset_database.use_mmap_for_cvs_item_to_changeset_table = True
528 # Now set the project to be converted to hg. cvs2hg only supports
529 # single-project conversions, so this method must only be called once:
530 run_options.set_project(
531 # The filesystem path to the part of the CVS repository (*not* a
532 # CVS working copy) that should be converted. This may be a
533 # subdirectory (i.e., a module) within a larger CVS repository.
534 r'test-data/main-cvsrepos',
536 # A list of symbol transformations that can be used to rename
537 # symbols in this project.
539 # Use IgnoreSymbolTransforms like the following to completely
540 # ignore symbols matching a regular expression when parsing
541 # the CVS repository, for example to avoid warnings about
542 # branches with two names and to choose the preferred name.
543 # It is *not* recommended to use this instead of
544 # ExcludeRegexpStrategyRule; though more efficient,
545 # IgnoreSymbolTransforms are less flexible and don't exclude
546 # branches correctly. The argument is a Python-style regular
547 # expression that has to match the *whole* CVS symbol name:
548 #IgnoreSymbolTransform(r'nightly-build-tag-.*')
550 # RegexpSymbolTransforms transform symbols textually using a
551 # regular expression. The first argument is a Python regular
552 # expression pattern and the second is a replacement pattern.
553 # The pattern is matched against each symbol name. If it
554 # matches the whole symbol name, then the symbol name is
555 # replaced with the corresponding replacement text. The
556 # replacement can include substitution patterns (e.g., r'\1'
557 # or r'\g<name>'). Typically you will want to use raw strings
558 # (strings with a preceding 'r', like shown in the examples)
559 # for the regexp and its replacement to avoid backslash
560 # substitution within those strings.
561 #RegexpSymbolTransform(r'release-(\d+)_(\d+)',
563 #RegexpSymbolTransform(r'release-(\d+)_(\d+)_(\d+)',
564 # r'release-\1.\2.\3'),
566 # Simple 1:1 character replacements can also be done. The
567 # following transform, which converts backslashes into forward
568 # slashes, should usually be included:
569 ReplaceSubstringsSymbolTransform('\\','/'),
571 # This last rule eliminates leading, trailing, and repeated
572 # slashes within the output symbol names:
573 NormalizePathsSymbolTransform(),
576 # See the definition of global_symbol_strategy_rules above for a
577 # description of this option:
578 symbol_strategy_rules=global_symbol_strategy_rules,