hgkw/keyword.py
author Christian Ebert <blacktrash@gmx.net>
Fri, 08 Feb 2008 10:14:47 +0100
branch0.9.2compat
changeset 387 ecb2935f9bb8
parent 380 0ed26effe190
child 388 05fbd226cb5c
permissions -rw-r--r--
(0.9.2compat) incorporate changes from default branch Testing reveals no name conflict with RuleDispatch's dispatch module. At least not with 0.9.* release versions.

# keyword.py - $Keyword$ expansion for Mercurial
#
# Copyright 2007, 2008 Christian Ebert <blacktrash@gmx.net>
#
# This software may be used and distributed according to the terms
# of the GNU General Public License, incorporated herein by reference.
#
# $Id$
#
# Keyword expansion hack against the grain of a DSCM
#
# There are many good reasons why this is not needed in a distributed
# SCM, still it may be useful in very small projects based on single
# files (like LaTeX packages), that are mostly addressed to an audience
# not running a version control system.
#
# For in-depth discussion refer to
# <http://www.selenic.com/mercurial/wiki/index.cgi/KeywordPlan>.
#
# Keyword expansion is based on Mercurial's changeset template mappings.
#
# Binary files are not touched.
#
# Setup in hgrc:
#
#   [extensions]
#   # enable extension
#   keyword = /full/path/to/hgkw/keyword.py
#   # or, if script in canonical hgext folder:
#   # hgext.keyword =
#
# Files to act upon/ignore are specified in the [keyword] section.
# Customized keyword template mappings in the [keywordmaps] section.
#
# Run "hg help keyword" and "hg kwdemo" to get info on configuration.

'''keyword expansion in local repositories

This extension expands RCS/CVS-like or self-customized $Keywords$
in tracked text files selected by your configuration.

Keywords are only expanded in local repositories and not stored in
the change history. The mechanism can be regarded as a convenience
for the current user or for archive distribution.

Configuration is done in the [keyword] and [keywordmaps] sections
of hgrc files.

Example:

    [keyword]
    # expand keywords in every python file except those matching "x*"
    **.py =
    x*    = ignore

Note: the more specific you are in your filename patterns
      the less you lose speed in huge repos.

For [keywordmaps] template mapping and expansion demonstration and
control run "hg kwdemo".

An additional date template filter {date|utcdate} is provided.

The default template mappings (view with "hg kwdemo -d") can be replaced
with customized keywords and templates.
Again, run "hg kwdemo" to control the results of your config changes.

Before changing/disabling active keywords, run "hg kwshrink" to avoid
the risk of inadvertedly storing expanded keywords in the change history.

To force expansion after enabling it, or a configuration change, run
"hg kwexpand".

Also, when committing with the record extension or using mq's qrecord, be aware
that keywords cannot be updated. Again, run "hg kwexpand" on the files in
question to update keyword expansions after all changes have been checked in.

Expansions spanning more than one line and incremental expansions,
like CVS' $Log$, are not supported. A keyword template map
"Log = {desc}" expands to the first line of the changeset description.

Caveat: With Mercurial versions prior to 4574925db5c0 "hg import" might
        cause rejects if the patch context contains an active keyword.
        In that case run "hg kwshrink", and then reimport.
        Or, better, use bundle/unbundle to share changes.
'''

from mercurial import commands, cmdutil, context, fancyopts
from mercurial import filelog, localrepo, revlog, templater, util
from mercurial.node import *
from mercurial.i18n import gettext as _
import getopt, os.path, re, shutil, tempfile, time

commands.optionalrepo += ' kwdemo'

# hg commands that do not act on keywords
nokwcommands = ('add addremove bundle copy export grep identify incoming init'
                ' log outgoing push remove rename rollback tip convert email')

# hg commands that trigger expansion only when writing to working dir,
# not when reading filelog, and unexpand when reading from working dir
restricted = 'diff1 record qfold qimport qnew qpush qrefresh qrecord'

def utcdate(date):
    '''Returns hgdate in cvs-like UTC format.'''
    return time.strftime('%Y/%m/%d %H:%M:%S', time.gmtime(date[0]))


_kwtemplater, _cmd, _cmdoptions = None, None, None

# backwards compatibility hacks

try:
    # cmdutil.parse moves to dispatch._parse in 18a9fbb5cd78
    from mercurial import dispatch
    _dispatch_parse = dispatch._parse
except ImportError:
    try:
        # commands.parse moves to cmdutil.parse in 0c61124ad877
        _dispatch_parse = cmdutil.parse
    except AttributeError:
        _dispatch_parse = commands.parse

def _kwdispatch_parse(ui, args):
    '''Monkeypatch dispatch._parse to obtain
    current command and command options (global _cmd, _cmdoptions).'''
    global _cmd, _cmdoptions
    _cmd, func, args, options, _cmdoptions = _dispatch_parse(ui, args)
    return _cmd, func, args, options, _cmdoptions

try:
    setattr(dispatch, '_parse', _kwdispatch_parse)
except (NameError, ImportError):
    # 0.9.4 needs ImportError
    if hasattr(cmdutil, 'parse'):
        cmdutil.parse = _kwdispatch_parse
    else:
        commands.parse = _kwdispatch_parse

try:
    # avoid spurious rejects if patchfile is available
    from mercurial.patch import patchfile
    _patchfile_init = patchfile.__init__

    def _kwpatchfile_init(self, ui, fname, missing=False):
        '''Monkeypatch/wrap patch.patchfile.__init__ to avoid
        rejects or conflicts due to expanded keywords in working dir.'''
        _patchfile_init(self, ui, fname, missing=missing)
        if _kwtemplater.matcher(self.fname):
            # shrink keywords read from working dir
            kwshrunk = _kwtemplater.shrink(''.join(self.lines))
            self.lines = kwshrunk.splitlines(True)
except ImportError:
    pass

try:
    # templatefilters module introduced in 9f1e6ab76069
    from mercurial import templatefilters
    template_filters = templatefilters.filters
    template_firstline = templatefilters.firstline
except ImportError:
    template_filters = templater.common_filters
    template_firstline = templater.firstline

def _wwrite(repo, f, data, mf):
    '''Makes repo.wwrite backwards compatible.'''
    # 656e06eebda7 removed file descriptor argument
    # 67982d3ee76c added flags argument
    try:
        repo.wwrite(f, data, mf.flags(f))
    except (AttributeError, TypeError):
        repo.wwrite(f, data)

def _normal(repo, files):
    '''Backwards compatible repo.dirstate.normal/update.'''
    # 6fd953d5faea introduced dirstate.normal()
    try:
        for f in files:
            repo.dirstate.normal(f)
    except AttributeError:
        repo.dirstate.update(files, 'n')

def _pathto(repo, f, cwd=None):
    '''kwfiles behaves similar to status, using pathto since 78b6add1f966.'''
    try:
        return repo.pathto(f, cwd)
    except AttributeError:
        return f

# commands.parse/cmdutil.parse returned nothing for
# "hg diff --rev" before 88803a69b24a due to bug in fancyopts
def _fancyopts(args, options, state):
    '''Fixed fancyopts from a9b7e425674f.'''
    namelist = []
    shortlist = ''
    argmap = {}
    defmap = {}

    for short, name, default, comment in options:
        # convert opts to getopt format
        oname = name
        name = name.replace('-', '_')

        argmap['-' + short] = argmap['--' + oname] = name
        defmap[name] = default

        # copy defaults to state
        if isinstance(default, list):
            state[name] = default[:]
        elif callable(default):
            print "whoa", name, default
            state[name] = None
        else:
            state[name] = default

        # does it take a parameter?
        if not (default is None or default is True or default is False):
            if short: short += ':'
            if oname: oname += '='
        if short:
            shortlist += short
        if name:
            namelist.append(oname)

    # parse arguments
    opts, args = getopt.getopt(args, shortlist, namelist)

    # transfer result to state
    for opt, val in opts:
        name = argmap[opt]
        t = type(defmap[name])
        if t is type(fancyopts):
            state[name] = defmap[name](val)
        elif t is type(1):
            state[name] = int(val)
        elif t is type(''):
            state[name] = val
        elif t is type([]):
            state[name].append(val)
        elif t is type(None) or t is type(False):
            state[name] = True

    # return unparsed args
    return args

fancyopts.fancyopts = _fancyopts


class kwtemplater(object):
    '''
    Sets up keyword templates, corresponding keyword regex, and
    provides keyword substitution functions.
    '''
    templates = {
        'Revision': '{node|short}',
        'Author': '{author|user}',
        'Date': '{date|utcdate}',
        'RCSFile': '{file|basename},v',
        'Source': '{root}/{file},v',
        'Id': '{file|basename},v {node|short} {date|utcdate} {author|user}',
        'Header': '{root}/{file},v {node|short} {date|utcdate} {author|user}',
    }

    def __init__(self, ui, repo, inc, exc, restrict):
        self.ui = ui
        self.repo = repo
        self.matcher = util.matcher(repo.root, inc=inc, exc=exc)[1]
        self.restrict = restrict
        self.commitnode = None
        self.path = ''

        kwmaps = self.ui.configitems('keywordmaps')
        if kwmaps: # override default templates
            kwmaps = [(k, templater.parsestring(v, quoted=False))
                      for (k, v) in kwmaps]
            self.templates = dict(kwmaps)
        escaped = map(re.escape, self.templates.keys())
        kwpat = r'\$(%s)(: [^$\n\r]*? )??\$' % '|'.join(escaped)
        self.re_kw = re.compile(kwpat)

        template_filters['utcdate'] = utcdate
        self.ct = self._changeset_templater()

    def _changeset_templater(self):
        '''Backwards compatible cmdutil.changeset_templater.'''
        # before 1e0b94cfba0e there was an extra "brinfo" argument
        try:
            return cmdutil.changeset_templater(self.ui, self.repo,
                                               False, '', False)
        except TypeError:
            return cmdutil.changeset_templater(self.ui, self.repo,
                                               False, None, '', False)

    def substitute(self, node, data, subfunc):
        '''Obtains file's changenode if commit node not given,
        and calls given substitution function.'''
        if self.commitnode:
            fnode = self.commitnode
        else:
            c = context.filectx(self.repo, self.path, fileid=node)
            fnode = c.node()

        def kwsub(mobj):
            '''Substitutes keyword using corresponding template.'''
            kw = mobj.group(1)
            self.ct.use_template(self.templates[kw])
            self.ui.pushbuffer()
            self.ct.show(changenode=fnode, root=self.repo.root, file=self.path)
            return '$%s: %s $' % (kw, template_firstline(self.ui.popbuffer()))

        return subfunc(kwsub, data)

    def expand(self, node, data):
        '''Returns data with keywords expanded.'''
        if self.restrict or util.binary(data):
            return data
        return self.substitute(node, data, self.re_kw.sub)

    def process(self, node, data, expand):
        '''Returns a tuple: data, count.
        Count is number of keywords/keyword substitutions,
        telling caller whether to act on file containing data.'''
        if util.binary(data):
            return data, None
        if expand:
            return self.substitute(node, data, self.re_kw.subn)
        return data, self.re_kw.search(data)

    def shrink(self, text):
        '''Returns text with all keyword substitutions removed.'''
        if util.binary(text):
            return text
        return self.re_kw.sub(r'$\1$', text)

class kwfilelog(filelog.filelog):
    '''
    Subclass of filelog to hook into its read, add, cmp methods.
    Keywords are "stored" unexpanded, and processed on reading.
    '''
    def __init__(self, opener, path):
        super(kwfilelog, self).__init__(opener, path)
        _kwtemplater.path = path

    def kwctread(self, node, expand):
        '''Reads expanding and counting keywords, called from _overwrite.'''
        data = super(kwfilelog, self).read(node)
        return _kwtemplater.process(node, data, expand)

    def read(self, node):
        '''Expands keywords when reading filelog.'''
        data = super(kwfilelog, self).read(node)
        return _kwtemplater.expand(node, data)

    def add(self, text, meta, tr, link, p1=None, p2=None):
        '''Removes keyword substitutions when adding to filelog.'''
        text = _kwtemplater.shrink(text)
        return super(kwfilelog, self).add(text, meta, tr, link, p1=p1, p2=p2)

    def cmp(self, node, text):
        '''Removes keyword substitutions for comparison.'''
        text = _kwtemplater.shrink(text)
        if self.renamed(node):
            t2 = super(kwfilelog, self).read(node)
            return t2 != text
        return revlog.revlog.cmp(self, node, text)

def _iskwfile(f, link):
    return not link(f) and _kwtemplater.matcher(f)

def _status(ui, repo, *pats, **opts):
    '''Bails out if [keyword] configuration is not active.
    Returns status of working directory.'''
    if _kwtemplater:
        files, match, anypats = cmdutil.matchpats(repo, pats, opts)
        return repo.status(files=files, match=match, list_clean=True)
    if ui.configitems('keyword'):
        raise util.Abort(_('[keyword] patterns cannot match'))
    raise util.Abort(_('no [keyword] patterns configured'))

def _overwrite(ui, repo, node=None, expand=True, files=None):
    '''Overwrites selected files expanding/shrinking keywords.'''
    ctx = repo.changectx(node)
    mf = ctx.manifest()
    if node is not None:   # commit
        _kwtemplater.commitnode = node
        files = [f for f in ctx.files() if f in mf]
        notify = ui.debug
    else:                  # kwexpand/kwshrink
        notify = ui.note
    candidates = [f for f in files if _iskwfile(f, mf.linkf)]
    if candidates:
        overwritten = []
        candidates.sort()
        action = expand and 'expanding' or 'shrinking'
        for f in candidates:
            fp = repo.file(f, kwmatch=True)
            data, kwfound = fp.kwctread(mf[f], expand)
            if kwfound:
                notify(_('overwriting %s %s keywords\n') % (f, action))
                _wwrite(repo, f, data, mf)
                overwritten.append(f)
        _normal(repo, overwritten)

def _kwfwrite(ui, repo, expand, *pats, **opts):
    '''Selects files and passes them to _overwrite.'''
    status = _status(ui, repo, *pats, **opts)
    modified, added, removed, deleted, unknown, ignored, clean = status
    if modified or added or removed or deleted:
        raise util.Abort(_('outstanding uncommitted changes in given files'))
    wlock = lock = None
    try:
        wlock = repo.wlock()
        lock = repo.lock()
        _overwrite(ui, repo, expand=expand, files=clean)
    finally:
        del wlock, lock


def demo(ui, repo, *args, **opts):
    '''print [keywordmaps] configuration and an expansion example

    Show current, custom, or default keyword template maps
    and their expansion.

    Extend current configuration by specifying maps as arguments
    and optionally by reading from an additional hgrc file.

    Override current keyword template maps with "default" option.
    '''
    def demostatus(stat):
        ui.status(_('\n\t%s\n') % stat)

    def demoitems(section, items):
        ui.write('[%s]\n' % section)
        for k, v in items:
            ui.write('%s = %s\n' % (k, v))

    msg = 'hg keyword config and expansion example'
    kwstatus = 'current'
    fn = 'demo.txt'
    branchname = 'demobranch'
    tmpdir = tempfile.mkdtemp('', 'kwdemo.')
    ui.note(_('creating temporary repo at %s\n') % tmpdir)
    repo = localrepo.localrepository(ui, path=tmpdir, create=True)
    ui.setconfig('keyword', fn, '')
    if args or opts.get('rcfile'):
        kwstatus = 'custom'
    if opts.get('rcfile'):
        ui.readconfig(opts.get('rcfile'))
    if opts.get('default'):
        kwstatus = 'default'
        kwmaps = kwtemplater.templates
        if ui.configitems('keywordmaps'):
            # override maps from optional rcfile
            for k, v in kwmaps.iteritems():
                ui.setconfig('keywordmaps', k, v)
    elif args:
        # simulate hgrc parsing
        rcmaps = ['[keywordmaps]\n'] + [a + '\n' for a in args]
        fp = repo.opener('hgrc', 'w')
        fp.writelines(rcmaps)
        fp.close()
        ui.readconfig(repo.join('hgrc'))
    if not opts.get('default'):
        kwmaps = dict(ui.configitems('keywordmaps')) or kwtemplater.templates
    reposetup(ui, repo)
    for k, v in ui.configitems('extensions'):
        if k.endswith('keyword'):
            extension = '%s = %s' % (k, v)
            break
    demostatus('config using %s keyword template maps' % kwstatus)
    ui.write('[extensions]\n%s\n' % extension)
    demoitems('keyword', ui.configitems('keyword'))
    demoitems('keywordmaps', kwmaps.iteritems())
    keywords = '$' + '$\n$'.join(kwmaps.keys()) + '$\n'
    repo.wopener(fn, 'w').write(keywords)
    repo.add([fn])
    path = repo.wjoin(fn)
    ui.note(_('\n%s keywords written to %s:\n') % (kwstatus, path))
    ui.note(keywords)
    ui.note('\nhg -R "%s" branch "%s"\n' % (tmpdir, branchname))
    # silence branch command if not verbose
    quiet = ui.quiet
    ui.quiet = not ui.verbose
    commands.branch(ui, repo, branchname)
    ui.quiet = quiet
    for name, cmd in ui.configitems('hooks'):
        if name.split('.', 1)[0].find('commit') > -1:
            repo.ui.setconfig('hooks', name, '')
    ui.note(_('unhooked all commit hooks\n'))
    ui.note('hg -R "%s" ci -m "%s"\n' % (tmpdir, msg))
    repo.commit(text=msg)
    format = ui.verbose and ' in %s' % path or ''
    demostatus('%s keywords expanded%s' % (kwstatus, format))
    ui.write(repo.wread(fn))
    ui.debug(_('\nremoving temporary repo %s\n') % tmpdir)
    shutil.rmtree(tmpdir, ignore_errors=True)

def expand(ui, repo, *pats, **opts):
    '''expand keywords in working directory

    Run after (re)enabling keyword expansion.

    kwexpand refuses to run if given files contain local changes.
    '''
    # 3rd argument sets expansion to True
    _kwfwrite(ui, repo, True, *pats, **opts)

def files(ui, repo, *pats, **opts):
    '''print files currently configured for keyword expansion

    Crosscheck which files in working directory are potential targets for
    keyword expansion.
    That is, files matched by [keyword] config patterns but not symlinks.
    '''
    status = _status(ui, repo, *pats, **opts)
    modified, added, removed, deleted, unknown, ignored, clean = status
    files = modified + added + clean
    if opts.get('untracked'):
        files += unknown
    files.sort()
    # use the full definition of repo._link for backwards compatibility
    kwfiles = [f for f in files if _kwtemplater.matcher(f)
               and not os.path.islink(repo.wjoin(f))]
    cwd = pats and repo.getcwd() or ''
    kwfstats = not opts.get('ignore') and (('K', kwfiles),) or ()
    if opts.get('all') or opts.get('ignore'):
        kwfstats += (('I', [f for f in files if f not in kwfiles]),)
    for char, filenames in kwfstats:
        format = (opts.get('all') or ui.verbose) and '%s %%s\n' % char or '%s\n'
        for f in filenames:
            ui.write(format % _pathto(repo, f, cwd))

def shrink(ui, repo, *pats, **opts):
    '''revert expanded keywords in working directory

    Run before changing/disabling active keywords
    or if you experience problems with "hg import" or "hg merge".

    kwshrink refuses to run if given files contain local changes.
    '''
    # 3rd argument sets expansion to False
    _kwfwrite(ui, repo, False, *pats, **opts)


def reposetup(ui, repo):
    '''Sets up repo as kwrepo for keyword substitution.
    Overrides file method to return kwfilelog instead of filelog
    if file matches user configuration.
    Wraps commit to overwrite configured files with updated
    keyword substitutions.
    This is done for local repos only, and only if there are
    files configured at all for keyword substitution.'''

    global _kwtemplater
    hgcmd, hgcmdopts = _cmd, _cmdoptions

    try:
        if (not repo.local() or hgcmd in nokwcommands.split() 
            or '.hg' in repo.root.split('/')
            or repo._url.startswith('bundle:')):
            return
    except AttributeError:
        pass

    inc, exc = [], ['.hg*']
    for pat, opt in ui.configitems('keyword'):
        if opt != 'ignore':
            inc.append(pat)
        else:
            exc.append(pat)
    if not inc:
        return

    if hgcmd == 'diff':
        # only expand if comparing against working dir
        node1, node2 = cmdutil.revpair(repo, hgcmdopts.get('rev'))
        if node2 is not None:
            return
        # shrink if rev is not current node
        if node1 is not None and node1 != repo.changectx().node():
            hgcmd = 'diff1'

    restrict = hgcmd in restricted.split()
    _kwtemplater = kwtemplater(ui, repo, inc, exc, restrict)

    class kwrepo(repo.__class__):
        def file(self, f, kwmatch=False):
            if f[0] == '/':
                f = f[1:]
            if kwmatch or _kwtemplater.matcher(f):
                return kwfilelog(self.sopener, f)
            return filelog.filelog(self.sopener, f)

        def wread(self, filename):
            data = super(kwrepo, self).wread(filename)
            if restrict and _kwtemplater.matcher(filename):
                return _kwtemplater.shrink(data)
            return data

        def _commit(self, files, text, user, date, match, force, lock, wlock,
                    force_editor, p1, p2, extra, empty_ok):
            '''Private commit wrapper for backwards compatibility.'''
            try:
                return super(kwrepo, self).commit(files=files, text=text,
                                                  user=user, date=date,
                                                  match=match, force=force,
                                                  lock=lock, wlock=wlock,
                                                  force_editor=force_editor,
                                                  p1=p1, p2=p2, extra=extra)
            except TypeError:
                try:
                    return super(kwrepo, self).commit(files=files, text=text,
                                                      user=user, date=date,
                                                      match=match, force=force,
                                                      force_editor=force_editor,
                                                      p1=p1, p2=p2,
                                                      extra=extra,
                                                      empty_ok=empty_ok)
                except TypeError:
                    return super(kwrepo, self).commit(files=files, text=text,
                                                      user=user, date=date,
                                                      match=match, force=force,
                                                      force_editor=force_editor,
                                                      p1=p1, p2=p2, extra=extra)

        def commit(self, files=None, text='', user=None, date=None,
                   match=util.always, force=False, lock=None, wlock=None,
                   force_editor=False, p1=None, p2=None, extra={},
                   empty_ok=False):
            # (w)lock arguments removed in 126f527b3ba3
            # so they are None or what was passed to commit
            # use private _(w)lock for deletion
            _lock = lock
            _wlock = wlock
            del wlock, lock
            _p1 = _p2 = None
            try:
                if not _wlock:
                    _wlock = self.wlock()
                if not _lock:
                    _lock = self.lock()
                # store and postpone commit hooks
                commithooks = {}
                for name, cmd in ui.configitems('hooks'):
                    if name.split('.', 1)[0] == 'commit':
                        commithooks[name] = cmd
                        ui.setconfig('hooks', name, '')
                if commithooks:
                    # store parents for commit hook environment
                    if p1 is None:
                        _p1, _p2 = repo.dirstate.parents()
                    else:
                        _p1, _p2 = p1, p2 or nullid
                    _p1 = hex(_p1)
                    if _p2 == nullid:
                        _p2 = ''
                    else:
                        _p2 = hex(_p2)

                node = self._commit(files, text, user, date, match, force,
                                    _lock, _wlock, force_editor, p1, p2, extra,
                                    empty_ok)

                # restore commit hooks
                for name, cmd in commithooks.iteritems():
                    ui.setconfig('hooks', name, cmd)
                if node is not None:
                    _overwrite(ui, self, node=node)
                    repo.hook('commit', node=node, parent1=_p1, parent2=_p2)
                return node
            finally:
                del _wlock, _lock

    repo.__class__ = kwrepo
    try:
        patchfile.__init__ = _kwpatchfile_init
    except NameError:
        pass


cmdtable = {
    'kwdemo':
        (demo,
         [('d', 'default', None, _('show default keyword template maps')),
          ('f', 'rcfile', [], _('read maps from rcfile'))],
         _('hg kwdemo [-d] [-f RCFILE] [TEMPLATEMAP]...')),
    'kwexpand': (expand, commands.walkopts,
                 _('hg kwexpand [OPTION]... [FILE]...')),
    'kwfiles':
        (files,
         [('a', 'all', None, _('show keyword status flags of all files')),
          ('i', 'ignore', None, _('show files excluded from expansion')),
          ('u', 'untracked', None, _('additionally show untracked files')),
         ] + commands.walkopts,
         _('hg kwfiles [OPTION]... [FILE]...')),
    'kwshrink': (shrink, commands.walkopts,
                 _('hg kwshrink [OPTION]... [FILE]...')),
}