convert: keep converted hg parents that are outside convert.hg.revs (BC)

Before, when converting revisions without also including their already
converted parents in convert.hg.revs, the parents would no longer be parents.

That seems unfortunate and we dare to assume that nobody ever wants that.

Instead, preserve parents that are outside the current convert range but
already have been converted.

The parents returned in getcommit() are unconditionally converted, so we
introduce a separate optparents with optional parents.

hgext/convert/common.py

@@ -1,492 +1,494 b''
 # common.py - common code for the convert extension
 #
 #  Copyright 2005-2009 Matt Mackall <[e-mail]> and others
 #
 # This software may be used and distributed according to the terms of the
 # GNU General Public License version 2 or any later version.
 from __future__ import absolute_import
 
 import base64
 import cPickle as pickle
 import datetime
 import errno
 import os
 import re
 import subprocess
 
 from mercurial import (
     error,
     phases,
     util,
 )
 from mercurial.i18n import _
 
 propertycache = util.propertycache
 
 def encodeargs(args):
     def encodearg(s):
         lines = base64.encodestring(s)
         lines = [l.splitlines()[0] for l in lines]
         return ''.join(lines)
 
     s = pickle.dumps(args)
     return encodearg(s)
 
 def decodeargs(s):
     s = base64.decodestring(s)
     return pickle.loads(s)
 
 class MissingTool(Exception):
     pass
 
 def checktool(exe, name=None, abort=True):
     name = name or exe
     if not util.findexe(exe):
         if abort:
             exc = error.Abort
         else:
             exc = MissingTool
         raise exc(_('cannot find required "%s" tool') % name)
 
 class NoRepo(Exception):
     pass
 
 SKIPREV = 'SKIP'
 
 class commit(object):
     def __init__(self, author, date, desc, parents, branch=None, rev=None,
-                 extra={}, sortkey=None, saverev=True, phase=phases.draft):
+                 extra={}, sortkey=None, saverev=True, phase=phases.draft,
+                 optparents=None):
         self.author = author or 'unknown'
         self.date = date or '0 0'
         self.desc = desc
-        self.parents = parents
+        self.parents = parents # will be converted and used as parents
+        self.optparents = optparents or [] # will be used if already converted
         self.branch = branch
         self.rev = rev
         self.extra = extra
         self.sortkey = sortkey
         self.saverev = saverev
         self.phase = phase
 
 class converter_source(object):
     """Conversion source interface"""
 
     def __init__(self, ui, path=None, revs=None):
         """Initialize conversion source (or raise NoRepo("message")
         exception if path is not a valid repository)"""
         self.ui = ui
         self.path = path
         self.revs = revs
 
         self.encoding = 'utf-8'
 
     def checkhexformat(self, revstr, mapname='splicemap'):
         """ fails if revstr is not a 40 byte hex. mercurial and git both uses
             such format for their revision numbering
         """
         if not re.match(r'[0-9a-fA-F]{40,40}$', revstr):
             raise error.Abort(_('%s entry %s is not a valid revision'
                                ' identifier') % (mapname, revstr))
 
     def before(self):
         pass
 
     def after(self):
         pass
 
     def targetfilebelongstosource(self, targetfilename):
         """Returns true if the given targetfile belongs to the source repo. This
         is useful when only a subdirectory of the target belongs to the source
         repo."""
         # For normal full repo converts, this is always True.
         return True
 
     def setrevmap(self, revmap):
         """set the map of already-converted revisions"""
         pass
 
     def getheads(self):
         """Return a list of this repository's heads"""
         raise NotImplementedError
 
     def getfile(self, name, rev):
         """Return a pair (data, mode) where data is the file content
         as a string and mode one of '', 'x' or 'l'. rev is the
         identifier returned by a previous call to getchanges().
         Data is None if file is missing/deleted in rev.
         """
         raise NotImplementedError
 
     def getchanges(self, version, full):
         """Returns a tuple of (files, copies, cleanp2).
 
         files is a sorted list of (filename, id) tuples for all files
         changed between version and its first parent returned by
         getcommit(). If full, all files in that revision is returned.
         id is the source revision id of the file.
 
         copies is a dictionary of dest: source
 
         cleanp2 is the set of files filenames that are clean against p2.
         (Files that are clean against p1 are already not in files (unless
         full). This makes it possible to handle p2 clean files similarly.)
         """
         raise NotImplementedError
 
     def getcommit(self, version):
         """Return the commit object for version"""
         raise NotImplementedError
 
     def numcommits(self):
         """Return the number of commits in this source.
 
         If unknown, return None.
         """
         return None
 
     def gettags(self):
         """Return the tags as a dictionary of name: revision
 
         Tag names must be UTF-8 strings.
         """
         raise NotImplementedError
 
     def recode(self, s, encoding=None):
         if not encoding:
             encoding = self.encoding or 'utf-8'
 
         if isinstance(s, unicode):
             return s.encode("utf-8")
         try:
             return s.decode(encoding).encode("utf-8")
         except UnicodeError:
             try:
                 return s.decode("latin-1").encode("utf-8")
             except UnicodeError:
                 return s.decode(encoding, "replace").encode("utf-8")
 
     def getchangedfiles(self, rev, i):
         """Return the files changed by rev compared to parent[i].
 
         i is an index selecting one of the parents of rev.  The return
         value should be the list of files that are different in rev and
         this parent.
 
         If rev has no parents, i is None.
 
         This function is only needed to support --filemap
         """
         raise NotImplementedError
 
     def converted(self, rev, sinkrev):
         '''Notify the source that a revision has been converted.'''
         pass
 
     def hasnativeorder(self):
         """Return true if this source has a meaningful, native revision
         order. For instance, Mercurial revisions are store sequentially
         while there is no such global ordering with Darcs.
         """
         return False
 
     def hasnativeclose(self):
         """Return true if this source has ability to close branch.
         """
         return False
 
     def lookuprev(self, rev):
         """If rev is a meaningful revision reference in source, return
         the referenced identifier in the same format used by getcommit().
         return None otherwise.
         """
         return None
 
     def getbookmarks(self):
         """Return the bookmarks as a dictionary of name: revision
 
         Bookmark names are to be UTF-8 strings.
         """
         return {}
 
     def checkrevformat(self, revstr, mapname='splicemap'):
         """revstr is a string that describes a revision in the given
            source control system.  Return true if revstr has correct
            format.
         """
         return True
 
 class converter_sink(object):
     """Conversion sink (target) interface"""
 
     def __init__(self, ui, path):
         """Initialize conversion sink (or raise NoRepo("message")
         exception if path is not a valid repository)
 
         created is a list of paths to remove if a fatal error occurs
         later"""
         self.ui = ui
         self.path = path
         self.created = []
 
     def revmapfile(self):
         """Path to a file that will contain lines
         source_rev_id sink_rev_id
         mapping equivalent revision identifiers for each system."""
         raise NotImplementedError
 
     def authorfile(self):
         """Path to a file that will contain lines
         srcauthor=dstauthor
         mapping equivalent authors identifiers for each system."""
         return None
 
     def putcommit(self, files, copies, parents, commit, source, revmap, full,
                   cleanp2):
         """Create a revision with all changed files listed in 'files'
         and having listed parents. 'commit' is a commit object
         containing at a minimum the author, date, and message for this
         changeset.  'files' is a list of (path, version) tuples,
         'copies' is a dictionary mapping destinations to sources,
         'source' is the source repository, and 'revmap' is a mapfile
         of source revisions to converted revisions. Only getfile() and
         lookuprev() should be called on 'source'. 'full' means that 'files'
         is complete and all other files should be removed.
         'cleanp2' is a set of the filenames that are unchanged from p2
         (only in the common merge case where there two parents).
 
         Note that the sink repository is not told to update itself to
         a particular revision (or even what that revision would be)
         before it receives the file data.
         """
         raise NotImplementedError
 
     def puttags(self, tags):
         """Put tags into sink.
 
         tags: {tagname: sink_rev_id, ...} where tagname is an UTF-8 string.
         Return a pair (tag_revision, tag_parent_revision), or (None, None)
         if nothing was changed.
         """
         raise NotImplementedError
 
     def setbranch(self, branch, pbranches):
         """Set the current branch name. Called before the first putcommit
         on the branch.
         branch: branch name for subsequent commits
         pbranches: (converted parent revision, parent branch) tuples"""
         pass
 
     def setfilemapmode(self, active):
         """Tell the destination that we're using a filemap
 
         Some converter_sources (svn in particular) can claim that a file
         was changed in a revision, even if there was no change.  This method
         tells the destination that we're using a filemap and that it should
         filter empty revisions.
         """
         pass
 
     def before(self):
         pass
 
     def after(self):
         pass
 
     def putbookmarks(self, bookmarks):
         """Put bookmarks into sink.
 
         bookmarks: {bookmarkname: sink_rev_id, ...}
         where bookmarkname is an UTF-8 string.
         """
         pass
 
     def hascommitfrommap(self, rev):
         """Return False if a rev mentioned in a filemap is known to not be
         present."""
         raise NotImplementedError
 
     def hascommitforsplicemap(self, rev):
         """This method is for the special needs for splicemap handling and not
         for general use. Returns True if the sink contains rev, aborts on some
         special cases."""
         raise NotImplementedError
 
 class commandline(object):
     def __init__(self, ui, command):
         self.ui = ui
         self.command = command
 
     def prerun(self):
         pass
 
     def postrun(self):
         pass
 
     def _cmdline(self, cmd, *args, **kwargs):
         cmdline = [self.command, cmd] + list(args)
         for k, v in kwargs.iteritems():
             if len(k) == 1:
                 cmdline.append('-' + k)
             else:
                 cmdline.append('--' + k.replace('_', '-'))
             try:
                 if len(k) == 1:
                     cmdline.append('' + v)
                 else:
                     cmdline[-1] += '=' + v
             except TypeError:
                 pass
         cmdline = [util.shellquote(arg) for arg in cmdline]
         if not self.ui.debugflag:
             cmdline += ['2>', os.devnull]
         cmdline = ' '.join(cmdline)
         return cmdline
 
     def _run(self, cmd, *args, **kwargs):
         def popen(cmdline):
             p = subprocess.Popen(cmdline, shell=True, bufsize=-1,
                     close_fds=util.closefds,
                     stdout=subprocess.PIPE)
             return p
         return self._dorun(popen, cmd, *args, **kwargs)
 
     def _run2(self, cmd, *args, **kwargs):
         return self._dorun(util.popen2, cmd, *args, **kwargs)
 
     def _run3(self, cmd, *args, **kwargs):
         return self._dorun(util.popen3, cmd, *args, **kwargs)
 
     def _dorun(self, openfunc, cmd,  *args, **kwargs):
         cmdline = self._cmdline(cmd, *args, **kwargs)
         self.ui.debug('running: %s\n' % (cmdline,))
         self.prerun()
         try:
             return openfunc(cmdline)
         finally:
             self.postrun()
 
     def run(self, cmd, *args, **kwargs):
         p = self._run(cmd, *args, **kwargs)
         output = p.communicate()[0]
         self.ui.debug(output)
         return output, p.returncode
 
     def runlines(self, cmd, *args, **kwargs):
         p = self._run(cmd, *args, **kwargs)
         output = p.stdout.readlines()
         p.wait()
         self.ui.debug(''.join(output))
         return output, p.returncode
 
     def checkexit(self, status, output=''):
         if status:
             if output:
                 self.ui.warn(_('%s error:\n') % self.command)
                 self.ui.warn(output)
             msg = util.explainexit(status)[0]
             raise error.Abort('%s %s' % (self.command, msg))
 
     def run0(self, cmd, *args, **kwargs):
         output, status = self.run(cmd, *args, **kwargs)
         self.checkexit(status, output)
         return output
 
     def runlines0(self, cmd, *args, **kwargs):
         output, status = self.runlines(cmd, *args, **kwargs)
         self.checkexit(status, ''.join(output))
         return output
 
     @propertycache
     def argmax(self):
         # POSIX requires at least 4096 bytes for ARG_MAX
         argmax = 4096
         try:
             argmax = os.sysconf("SC_ARG_MAX")
         except (AttributeError, ValueError):
             pass
 
         # Windows shells impose their own limits on command line length,
         # down to 2047 bytes for cmd.exe under Windows NT/2k and 2500 bytes
         # for older 4nt.exe. See http://support.microsoft.com/kb/830473 for
         # details about cmd.exe limitations.
 
         # Since ARG_MAX is for command line _and_ environment, lower our limit
         # (and make happy Windows shells while doing this).
         return argmax // 2 - 1
 
     def _limit_arglist(self, arglist, cmd, *args, **kwargs):
         cmdlen = len(self._cmdline(cmd, *args, **kwargs))
         limit = self.argmax - cmdlen
         bytes = 0
         fl = []
         for fn in arglist:
             b = len(fn) + 3
             if bytes + b < limit or len(fl) == 0:
                 fl.append(fn)
                 bytes += b
             else:
                 yield fl
                 fl = [fn]
                 bytes = b
         if fl:
             yield fl
 
     def xargs(self, arglist, cmd, *args, **kwargs):
         for l in self._limit_arglist(arglist, cmd, *args, **kwargs):
             self.run0(cmd, *(list(args) + l), **kwargs)
 
 class mapfile(dict):
     def __init__(self, ui, path):
         super(mapfile, self).__init__()
         self.ui = ui
         self.path = path
         self.fp = None
         self.order = []
         self._read()
 
     def _read(self):
         if not self.path:
             return
         try:
             fp = open(self.path, 'r')
         except IOError as err:
             if err.errno != errno.ENOENT:
                 raise
             return
         for i, line in enumerate(fp):
             line = line.splitlines()[0].rstrip()
             if not line:
                 # Ignore blank lines
                 continue
             try:
                 key, value = line.rsplit(' ', 1)
             except ValueError:
                 raise error.Abort(
                     _('syntax error in %s(%d): key/value pair expected')
                     % (self.path, i + 1))
             if key not in self:
                 self.order.append(key)
             super(mapfile, self).__setitem__(key, value)
         fp.close()
 
     def __setitem__(self, key, value):
         if self.fp is None:
             try:
                 self.fp = open(self.path, 'a')
             except IOError as err:
                 raise error.Abort(_('could not open map file %r: %s') %
                                  (self.path, err.strerror))
         self.fp.write('%s %s\n' % (key, value))
         self.fp.flush()
         super(mapfile, self).__setitem__(key, value)
 
     def close(self):
         if self.fp:
             self.fp.close()
             self.fp = None
 
 def makedatetimestamp(t):
     """Like util.makedate() but for time t instead of current time"""
     delta = (datetime.datetime.utcfromtimestamp(t) -
              datetime.datetime.fromtimestamp(t))
     tz = delta.days * 86400 + delta.seconds
     return t, tz

hgext/convert/convcmd.py

@@ -1,608 +1,611 b''
 # convcmd - convert extension commands definition
 #
 # Copyright 2005-2007 Matt Mackall <[e-mail]>
 #
 # This software may be used and distributed according to the terms of the
 # GNU General Public License version 2 or any later version.
 from __future__ import absolute_import
 
 import os
 import shlex
 import shutil
 
 from mercurial import (
     encoding,
     error,
     hg,
     util,
 )
 from mercurial.i18n import _
 
 from . import (
     bzr,
     common,
     cvs,
     darcs,
     filemap,
     git,
     gnuarch,
     hg as hgconvert,
     monotone,
     p4,
     subversion,
 )
 
 mapfile = common.mapfile
 MissingTool = common.MissingTool
 NoRepo = common.NoRepo
 SKIPREV = common.SKIPREV
 
 bzr_source = bzr.bzr_source
 convert_cvs = cvs.convert_cvs
 convert_git = git.convert_git
 darcs_source = darcs.darcs_source
 gnuarch_source = gnuarch.gnuarch_source
 mercurial_sink = hgconvert.mercurial_sink
 mercurial_source = hgconvert.mercurial_source
 monotone_source = monotone.monotone_source
 p4_source = p4.p4_source
 svn_sink = subversion.svn_sink
 svn_source = subversion.svn_source
 
 orig_encoding = 'ascii'
 
 def recode(s):
     if isinstance(s, unicode):
         return s.encode(orig_encoding, 'replace')
     else:
         return s.decode('utf-8').encode(orig_encoding, 'replace')
 
 def mapbranch(branch, branchmap):
     '''
     >>> bmap = {'default': 'branch1'}
     >>> for i in ['', None]:
     ...     mapbranch(i, bmap)
     'branch1'
     'branch1'
     >>> bmap = {'None': 'branch2'}
     >>> for i in ['', None]:
     ...     mapbranch(i, bmap)
     'branch2'
     'branch2'
     >>> bmap = {'None': 'branch3', 'default': 'branch4'}
     >>> for i in ['None', '', None, 'default', 'branch5']:
     ...     mapbranch(i, bmap)
     'branch3'
     'branch4'
     'branch4'
     'branch4'
     'branch5'
     '''
     # If branch is None or empty, this commit is coming from the source
     # repository's default branch and destined for the default branch in the
     # destination repository. For such commits, using a literal "default"
     # in branchmap below allows the user to map "default" to an alternate
     # default branch in the destination repository.
     branch = branchmap.get(branch or 'default', branch)
     # At some point we used "None" literal to denote the default branch,
     # attempt to use that for backward compatibility.
     if (not branch):
         branch = branchmap.get(str(None), branch)
     return branch
 
 source_converters = [
     ('cvs', convert_cvs, 'branchsort'),
     ('git', convert_git, 'branchsort'),
     ('svn', svn_source, 'branchsort'),
     ('hg', mercurial_source, 'sourcesort'),
     ('darcs', darcs_source, 'branchsort'),
     ('mtn', monotone_source, 'branchsort'),
     ('gnuarch', gnuarch_source, 'branchsort'),
     ('bzr', bzr_source, 'branchsort'),
     ('p4', p4_source, 'branchsort'),
     ]
 
 sink_converters = [
     ('hg', mercurial_sink),
     ('svn', svn_sink),
     ]
 
 def convertsource(ui, path, type, revs):
     exceptions = []
     if type and type not in [s[0] for s in source_converters]:
         raise error.Abort(_('%s: invalid source repository type') % type)
     for name, source, sortmode in source_converters:
         try:
             if not type or name == type:
                 return source(ui, path, revs), sortmode
         except (NoRepo, MissingTool) as inst:
             exceptions.append(inst)
     if not ui.quiet:
         for inst in exceptions:
             ui.write("%s\n" % inst)
     raise error.Abort(_('%s: missing or unsupported repository') % path)
 
 def convertsink(ui, path, type):
     if type and type not in [s[0] for s in sink_converters]:
         raise error.Abort(_('%s: invalid destination repository type') % type)
     for name, sink in sink_converters:
         try:
             if not type or name == type:
                 return sink(ui, path)
         except NoRepo as inst:
             ui.note(_("convert: %s\n") % inst)
         except MissingTool as inst:
             raise error.Abort('%s\n' % inst)
     raise error.Abort(_('%s: unknown repository type') % path)
 
 class progresssource(object):
     def __init__(self, ui, source, filecount):
         self.ui = ui
         self.source = source
         self.filecount = filecount
         self.retrieved = 0
 
     def getfile(self, file, rev):
         self.retrieved += 1
         self.ui.progress(_('getting files'), self.retrieved,
                          item=file, total=self.filecount, unit=_('files'))
         return self.source.getfile(file, rev)
 
     def targetfilebelongstosource(self, targetfilename):
         return self.source.targetfilebelongstosource(targetfilename)
 
     def lookuprev(self, rev):
         return self.source.lookuprev(rev)
 
     def close(self):
         self.ui.progress(_('getting files'), None)
 
 class converter(object):
     def __init__(self, ui, source, dest, revmapfile, opts):
 
         self.source = source
         self.dest = dest
         self.ui = ui
         self.opts = opts
         self.commitcache = {}
         self.authors = {}
         self.authorfile = None
 
         # Record converted revisions persistently: maps source revision
         # ID to target revision ID (both strings).  (This is how
         # incremental conversions work.)
         self.map = mapfile(ui, revmapfile)
 
         # Read first the dst author map if any
         authorfile = self.dest.authorfile()
         if authorfile and os.path.exists(authorfile):
             self.readauthormap(authorfile)
         # Extend/Override with new author map if necessary
         if opts.get('authormap'):
             self.readauthormap(opts.get('authormap'))
             self.authorfile = self.dest.authorfile()
 
         self.splicemap = self.parsesplicemap(opts.get('splicemap'))
         self.branchmap = mapfile(ui, opts.get('branchmap'))
 
     def parsesplicemap(self, path):
         """ check and validate the splicemap format and
             return a child/parents dictionary.
             Format checking has two parts.
             1. generic format which is same across all source types
             2. specific format checking which may be different for
                different source type.  This logic is implemented in
                checkrevformat function in source files like
                hg.py, subversion.py etc.
         """
 
         if not path:
             return {}
         m = {}
         try:
             fp = open(path, 'r')
             for i, line in enumerate(fp):
                 line = line.splitlines()[0].rstrip()
                 if not line:
                     # Ignore blank lines
                     continue
                 # split line
                 lex = shlex.shlex(line, posix=True)
                 lex.whitespace_split = True
                 lex.whitespace += ','
                 line = list(lex)
                 # check number of parents
                 if not (2 <= len(line) <= 3):
                     raise error.Abort(_('syntax error in %s(%d): child parent1'
                                        '[,parent2] expected') % (path, i + 1))
                 for part in line:
                     self.source.checkrevformat(part)
                 child, p1, p2 = line[0], line[1:2], line[2:]
                 if p1 == p2:
                     m[child] = p1
                 else:
                     m[child] = p1 + p2
          # if file does not exist or error reading, exit
         except IOError:
             raise error.Abort(_('splicemap file not found or error reading %s:')
                                % path)
         return m
 
 
     def walktree(self, heads):
         '''Return a mapping that identifies the uncommitted parents of every
         uncommitted changeset.'''
         visit = heads
         known = set()
         parents = {}
         numcommits = self.source.numcommits()
         while visit:
             n = visit.pop(0)
             if n in known:
                 continue
             if n in self.map:
                 m = self.map[n]
                 if m == SKIPREV or self.dest.hascommitfrommap(m):
                     continue
             known.add(n)
             self.ui.progress(_('scanning'), len(known), unit=_('revisions'),
                              total=numcommits)
             commit = self.cachecommit(n)
             parents[n] = []
             for p in commit.parents:
                 parents[n].append(p)
                 visit.append(p)
         self.ui.progress(_('scanning'), None)
 
         return parents
 
     def mergesplicemap(self, parents, splicemap):
         """A splicemap redefines child/parent relationships. Check the
         map contains valid revision identifiers and merge the new
         links in the source graph.
         """
         for c in sorted(splicemap):
             if c not in parents:
                 if not self.dest.hascommitforsplicemap(self.map.get(c, c)):
                     # Could be in source but not converted during this run
                     self.ui.warn(_('splice map revision %s is not being '
                                    'converted, ignoring\n') % c)
                 continue
             pc = []
             for p in splicemap[c]:
                 # We do not have to wait for nodes already in dest.
                 if self.dest.hascommitforsplicemap(self.map.get(p, p)):
                     continue
                 # Parent is not in dest and not being converted, not good
                 if p not in parents:
                     raise error.Abort(_('unknown splice map parent: %s') % p)
                 pc.append(p)
             parents[c] = pc
 
     def toposort(self, parents, sortmode):
         '''Return an ordering such that every uncommitted changeset is
         preceded by all its uncommitted ancestors.'''
 
         def mapchildren(parents):
             """Return a (children, roots) tuple where 'children' maps parent
             revision identifiers to children ones, and 'roots' is the list of
             revisions without parents. 'parents' must be a mapping of revision
             identifier to its parents ones.
             """
             visit = sorted(parents)
             seen = set()
             children = {}
             roots = []
 
             while visit:
                 n = visit.pop(0)
                 if n in seen:
                     continue
                 seen.add(n)
                 # Ensure that nodes without parents are present in the
                 # 'children' mapping.
                 children.setdefault(n, [])
                 hasparent = False
                 for p in parents[n]:
                     if p not in self.map:
                         visit.append(p)
                         hasparent = True
                     children.setdefault(p, []).append(n)
                 if not hasparent:
                     roots.append(n)
 
             return children, roots
 
         # Sort functions are supposed to take a list of revisions which
         # can be converted immediately and pick one
 
         def makebranchsorter():
             """If the previously converted revision has a child in the
             eligible revisions list, pick it. Return the list head
             otherwise. Branch sort attempts to minimize branch
             switching, which is harmful for Mercurial backend
             compression.
             """
             prev = [None]
             def picknext(nodes):
                 next = nodes[0]
                 for n in nodes:
                     if prev[0] in parents[n]:
                         next = n
                         break
                 prev[0] = next
                 return next
             return picknext
 
         def makesourcesorter():
             """Source specific sort."""
             keyfn = lambda n: self.commitcache[n].sortkey
             def picknext(nodes):
                 return sorted(nodes, key=keyfn)[0]
             return picknext
 
         def makeclosesorter():
             """Close order sort."""
             keyfn = lambda n: ('close' not in self.commitcache[n].extra,
                                self.commitcache[n].sortkey)
             def picknext(nodes):
                 return sorted(nodes, key=keyfn)[0]
             return picknext
 
         def makedatesorter():
             """Sort revisions by date."""
             dates = {}
             def getdate(n):
                 if n not in dates:
                     dates[n] = util.parsedate(self.commitcache[n].date)
                 return dates[n]
 
             def picknext(nodes):
                 return min([(getdate(n), n) for n in nodes])[1]
 
             return picknext
 
         if sortmode == 'branchsort':
             picknext = makebranchsorter()
         elif sortmode == 'datesort':
             picknext = makedatesorter()
         elif sortmode == 'sourcesort':
             picknext = makesourcesorter()
         elif sortmode == 'closesort':
             picknext = makeclosesorter()
         else:
             raise error.Abort(_('unknown sort mode: %s') % sortmode)
 
         children, actives = mapchildren(parents)
 
         s = []
         pendings = {}
         while actives:
             n = picknext(actives)
             actives.remove(n)
             s.append(n)
 
             # Update dependents list
             for c in children.get(n, []):
                 if c not in pendings:
                     pendings[c] = [p for p in parents[c] if p not in self.map]
                 try:
                     pendings[c].remove(n)
                 except ValueError:
                     raise error.Abort(_('cycle detected between %s and %s')
                                        % (recode(c), recode(n)))
                 if not pendings[c]:
                     # Parents are converted, node is eligible
                     actives.insert(0, c)
                     pendings[c] = None
 
         if len(s) != len(parents):
             raise error.Abort(_("not all revisions were sorted"))
 
         return s
 
     def writeauthormap(self):
         authorfile = self.authorfile
         if authorfile:
             self.ui.status(_('writing author map file %s\n') % authorfile)
             ofile = open(authorfile, 'w+')
             for author in self.authors:
                 ofile.write("%s=%s\n" % (author, self.authors[author]))
             ofile.close()
 
     def readauthormap(self, authorfile):
         afile = open(authorfile, 'r')
         for line in afile:
 
             line = line.strip()
             if not line or line.startswith('#'):
                 continue
 
             try:
                 srcauthor, dstauthor = line.split('=', 1)
             except ValueError:
                 msg = _('ignoring bad line in author map file %s: %s\n')
                 self.ui.warn(msg % (authorfile, line.rstrip()))
                 continue
 
             srcauthor = srcauthor.strip()
             dstauthor = dstauthor.strip()
             if self.authors.get(srcauthor) in (None, dstauthor):
                 msg = _('mapping author %s to %s\n')
                 self.ui.debug(msg % (srcauthor, dstauthor))
                 self.authors[srcauthor] = dstauthor
                 continue
 
             m = _('overriding mapping for author %s, was %s, will be %s\n')
             self.ui.status(m % (srcauthor, self.authors[srcauthor], dstauthor))
 
         afile.close()
 
     def cachecommit(self, rev):
         commit = self.source.getcommit(rev)
         commit.author = self.authors.get(commit.author, commit.author)
         commit.branch = mapbranch(commit.branch, self.branchmap)
         self.commitcache[rev] = commit
         return commit
 
     def copy(self, rev):
         commit = self.commitcache[rev]
         full = self.opts.get('full')
         changes = self.source.getchanges(rev, full)
         if isinstance(changes, basestring):
             if changes == SKIPREV:
                 dest = SKIPREV
             else:
                 dest = self.map[changes]
             self.map[rev] = dest
             return
         files, copies, cleanp2 = changes
         pbranches = []
         if commit.parents:
             for prev in commit.parents:
                 if prev not in self.commitcache:
                     self.cachecommit(prev)
                 pbranches.append((self.map[prev],
                                   self.commitcache[prev].branch))
         self.dest.setbranch(commit.branch, pbranches)
         try:
             parents = self.splicemap[rev]
             self.ui.status(_('spliced in %s as parents of %s\n') %
                            (_(' and ').join(parents), rev))
             parents = [self.map.get(p, p) for p in parents]
         except KeyError:
             parents = [b[0] for b in pbranches]
+            parents.extend(self.map[x]
+                           for x in commit.optparents
+                           if x in self.map)
         if len(pbranches) != 2:
             cleanp2 = set()
         if len(parents) < 3:
             source = progresssource(self.ui, self.source, len(files))
         else:
             # For an octopus merge, we end up traversing the list of
             # changed files N-1 times. This tweak to the number of
             # files makes it so the progress bar doesn't overflow
             # itself.
             source = progresssource(self.ui, self.source,
                                     len(files) * (len(parents) - 1))
         newnode = self.dest.putcommit(files, copies, parents, commit,
                                       source, self.map, full, cleanp2)
         source.close()
         self.source.converted(rev, newnode)
         self.map[rev] = newnode
 
     def convert(self, sortmode):
         try:
             self.source.before()
             self.dest.before()
             self.source.setrevmap(self.map)
             self.ui.status(_("scanning source...\n"))
             heads = self.source.getheads()
             parents = self.walktree(heads)
             self.mergesplicemap(parents, self.splicemap)
             self.ui.status(_("sorting...\n"))
             t = self.toposort(parents, sortmode)
             num = len(t)
             c = None
 
             self.ui.status(_("converting...\n"))
             for i, c in enumerate(t):
                 num -= 1
                 desc = self.commitcache[c].desc
                 if "\n" in desc:
                     desc = desc.splitlines()[0]
                 # convert log message to local encoding without using
                 # tolocal() because the encoding.encoding convert()
                 # uses is 'utf-8'
                 self.ui.status("%d %s\n" % (num, recode(desc)))
                 self.ui.note(_("source: %s\n") % recode(c))
                 self.ui.progress(_('converting'), i, unit=_('revisions'),
                                  total=len(t))
                 self.copy(c)
             self.ui.progress(_('converting'), None)
 
             if not self.ui.configbool('convert', 'skiptags'):
                 tags = self.source.gettags()
                 ctags = {}
                 for k in tags:
                     v = tags[k]
                     if self.map.get(v, SKIPREV) != SKIPREV:
                         ctags[k] = self.map[v]
 
                 if c and ctags:
                     nrev, tagsparent = self.dest.puttags(ctags)
                     if nrev and tagsparent:
                         # write another hash correspondence to override the
                         # previous one so we don't end up with extra tag heads
                         tagsparents = [e for e in self.map.iteritems()
                                        if e[1] == tagsparent]
                         if tagsparents:
                             self.map[tagsparents[0][0]] = nrev
 
             bookmarks = self.source.getbookmarks()
             cbookmarks = {}
             for k in bookmarks:
                 v = bookmarks[k]
                 if self.map.get(v, SKIPREV) != SKIPREV:
                     cbookmarks[k] = self.map[v]
 
             if c and cbookmarks:
                 self.dest.putbookmarks(cbookmarks)
 
             self.writeauthormap()
         finally:
             self.cleanup()
 
     def cleanup(self):
         try:
             self.dest.after()
         finally:
             self.source.after()
         self.map.close()
 
 def convert(ui, src, dest=None, revmapfile=None, **opts):
     global orig_encoding
     orig_encoding = encoding.encoding
     encoding.encoding = 'UTF-8'
 
     # support --authors as an alias for --authormap
     if not opts.get('authormap'):
         opts['authormap'] = opts.get('authors')
 
     if not dest:
         dest = hg.defaultdest(src) + "-hg"
         ui.status(_("assuming destination %s\n") % dest)
 
     destc = convertsink(ui, dest, opts.get('dest_type'))
 
     try:
         srcc, defaultsort = convertsource(ui, src, opts.get('source_type'),
                                           opts.get('rev'))
     except Exception:
         for path in destc.created:
             shutil.rmtree(path, True)
         raise
 
     sortmodes = ('branchsort', 'datesort', 'sourcesort', 'closesort')
     sortmode = [m for m in sortmodes if opts.get(m)]
     if len(sortmode) > 1:
         raise error.Abort(_('more than one sort mode specified'))
     if sortmode:
         sortmode = sortmode[0]
     else:
         sortmode = defaultsort
 
     if sortmode == 'sourcesort' and not srcc.hasnativeorder():
         raise error.Abort(_('--sourcesort is not supported by this data source')
                          )
     if sortmode == 'closesort' and not srcc.hasnativeclose():
         raise error.Abort(_('--closesort is not supported by this data source'))
 
     fmap = opts.get('filemap')
     if fmap:
         srcc = filemap.filemap_source(ui, srcc, fmap)
         destc.setfilemapmode(True)
 
     if not revmapfile:
         revmapfile = destc.revmapfile()
 
     c = converter(ui, srcc, destc, revmapfile, opts)
     c.convert(sortmode)

hgext/convert/hg.py

@@ -1,652 +1,655 b''
 # hg.py - hg backend for convert extension
 #
 #  Copyright 2005-2009 Matt Mackall <[e-mail]> and others
 #
 # This software may be used and distributed according to the terms of the
 # GNU General Public License version 2 or any later version.
 
 # Notes for hg->hg conversion:
 #
 # * Old versions of Mercurial didn't trim the whitespace from the ends
 #   of commit messages, but new versions do.  Changesets created by
 #   those older versions, then converted, may thus have different
 #   hashes for changesets that are otherwise identical.
 #
 # * Using "--config convert.hg.saverev=true" will make the source
 #   identifier to be stored in the converted revision. This will cause
 #   the converted revision to have a different identity than the
 #   source.
 from __future__ import absolute_import
 
 import os
 import re
 import time
 
 from mercurial import (
     bookmarks,
     context,
     error,
     exchange,
     hg,
     lock as lockmod,
     merge as mergemod,
     node as nodemod,
     phases,
     scmutil,
     util,
 )
 stringio = util.stringio
 
 from mercurial.i18n import _
 from . import common
 mapfile = common.mapfile
 NoRepo = common.NoRepo
 
 sha1re = re.compile(r'\b[0-9a-f]{12,40}\b')
 
 class mercurial_sink(common.converter_sink):
     def __init__(self, ui, path):
         common.converter_sink.__init__(self, ui, path)
         self.branchnames = ui.configbool('convert', 'hg.usebranchnames', True)
         self.clonebranches = ui.configbool('convert', 'hg.clonebranches', False)
         self.tagsbranch = ui.config('convert', 'hg.tagsbranch', 'default')
         self.lastbranch = None
         if os.path.isdir(path) and len(os.listdir(path)) > 0:
             try:
                 self.repo = hg.repository(self.ui, path)
                 if not self.repo.local():
                     raise NoRepo(_('%s is not a local Mercurial repository')
                                  % path)
             except error.RepoError as err:
                 ui.traceback()
                 raise NoRepo(err.args[0])
         else:
             try:
                 ui.status(_('initializing destination %s repository\n') % path)
                 self.repo = hg.repository(self.ui, path, create=True)
                 if not self.repo.local():
                     raise NoRepo(_('%s is not a local Mercurial repository')
                                  % path)
                 self.created.append(path)
             except error.RepoError:
                 ui.traceback()
                 raise NoRepo(_("could not create hg repository %s as sink")
                              % path)
         self.lock = None
         self.wlock = None
         self.filemapmode = False
         self.subrevmaps = {}
 
     def before(self):
         self.ui.debug('run hg sink pre-conversion action\n')
         self.wlock = self.repo.wlock()
         self.lock = self.repo.lock()
 
     def after(self):
         self.ui.debug('run hg sink post-conversion action\n')
         if self.lock:
             self.lock.release()
         if self.wlock:
             self.wlock.release()
 
     def revmapfile(self):
         return self.repo.join("shamap")
 
     def authorfile(self):
         return self.repo.join("authormap")
 
     def setbranch(self, branch, pbranches):
         if not self.clonebranches:
             return
 
         setbranch = (branch != self.lastbranch)
         self.lastbranch = branch
         if not branch:
             branch = 'default'
         pbranches = [(b[0], b[1] and b[1] or 'default') for b in pbranches]
         if pbranches:
             pbranch = pbranches[0][1]
         else:
             pbranch = 'default'
 
         branchpath = os.path.join(self.path, branch)
         if setbranch:
             self.after()
             try:
                 self.repo = hg.repository(self.ui, branchpath)
             except Exception:
                 self.repo = hg.repository(self.ui, branchpath, create=True)
             self.before()
 
         # pbranches may bring revisions from other branches (merge parents)
         # Make sure we have them, or pull them.
         missings = {}
         for b in pbranches:
             try:
                 self.repo.lookup(b[0])
             except Exception:
                 missings.setdefault(b[1], []).append(b[0])
 
         if missings:
             self.after()
             for pbranch, heads in sorted(missings.iteritems()):
                 pbranchpath = os.path.join(self.path, pbranch)
                 prepo = hg.peer(self.ui, {}, pbranchpath)
                 self.ui.note(_('pulling from %s into %s\n') % (pbranch, branch))
                 exchange.pull(self.repo, prepo,
                               [prepo.lookup(h) for h in heads])
             self.before()
 
     def _rewritetags(self, source, revmap, data):
         fp = stringio()
         for line in data.splitlines():
             s = line.split(' ', 1)
             if len(s) != 2:
                 continue
             revid = revmap.get(source.lookuprev(s[0]))
             if not revid:
                 if s[0] == nodemod.nullhex:
                     revid = s[0]
                 else:
                     continue
             fp.write('%s %s\n' % (revid, s[1]))
         return fp.getvalue()
 
     def _rewritesubstate(self, source, data):
         fp = stringio()
         for line in data.splitlines():
             s = line.split(' ', 1)
             if len(s) != 2:
                 continue
 
             revid = s[0]
             subpath = s[1]
             if revid != nodemod.nullhex:
                 revmap = self.subrevmaps.get(subpath)
                 if revmap is None:
                     revmap = mapfile(self.ui,
                                      self.repo.wjoin(subpath, '.hg/shamap'))
                     self.subrevmaps[subpath] = revmap
 
                     # It is reasonable that one or more of the subrepos don't
                     # need to be converted, in which case they can be cloned
                     # into place instead of converted.  Therefore, only warn
                     # once.
                     msg = _('no ".hgsubstate" updates will be made for "%s"\n')
                     if len(revmap) == 0:
                         sub = self.repo.wvfs.reljoin(subpath, '.hg')
 
                         if self.repo.wvfs.exists(sub):
                             self.ui.warn(msg % subpath)
 
                 newid = revmap.get(revid)
                 if not newid:
                     if len(revmap) > 0:
                         self.ui.warn(_("%s is missing from %s/.hg/shamap\n") %
                                      (revid, subpath))
                 else:
                     revid = newid
 
             fp.write('%s %s\n' % (revid, subpath))
 
         return fp.getvalue()
 
     def _calculatemergedfiles(self, source, p1ctx, p2ctx):
         """Calculates the files from p2 that we need to pull in when merging p1
         and p2, given that the merge is coming from the given source.
 
         This prevents us from losing files that only exist in the target p2 and
         that don't come from the source repo (like if you're merging multiple
         repositories together).
         """
         anc = [p1ctx.ancestor(p2ctx)]
         # Calculate what files are coming from p2
         actions, diverge, rename = mergemod.calculateupdates(
             self.repo, p1ctx, p2ctx, anc,
             True,  # branchmerge
             True,  # force
             False, # acceptremote
             False, # followcopies
         )
 
         for file, (action, info, msg) in actions.iteritems():
             if source.targetfilebelongstosource(file):
                 # If the file belongs to the source repo, ignore the p2
                 # since it will be covered by the existing fileset.
                 continue
 
             # If the file requires actual merging, abort. We don't have enough
             # context to resolve merges correctly.
             if action in ['m', 'dm', 'cd', 'dc']:
                 raise error.Abort(_("unable to convert merge commit "
                     "since target parents do not merge cleanly (file "
                     "%s, parents %s and %s)") % (file, p1ctx,
                                                  p2ctx))
             elif action == 'k':
                 # 'keep' means nothing changed from p1
                 continue
             else:
                 # Any other change means we want to take the p2 version
                 yield file
 
     def putcommit(self, files, copies, parents, commit, source, revmap, full,
                   cleanp2):
         files = dict(files)
 
         def getfilectx(repo, memctx, f):
             if p2ctx and f in p2files and f not in copies:
                 self.ui.debug('reusing %s from p2\n' % f)
                 try:
                     return p2ctx[f]
                 except error.ManifestLookupError:
                     # If the file doesn't exist in p2, then we're syncing a
                     # delete, so just return None.
                     return None
             try:
                 v = files[f]
             except KeyError:
                 return None
             data, mode = source.getfile(f, v)
             if data is None:
                 return None
             if f == '.hgtags':
                 data = self._rewritetags(source, revmap, data)
             if f == '.hgsubstate':
                 data = self._rewritesubstate(source, data)
             return context.memfilectx(self.repo, f, data, 'l' in mode,
                                       'x' in mode, copies.get(f))
 
         pl = []
         for p in parents:
             if p not in pl:
                 pl.append(p)
         parents = pl
         nparents = len(parents)
         if self.filemapmode and nparents == 1:
             m1node = self.repo.changelog.read(nodemod.bin(parents[0]))[0]
             parent = parents[0]
 
         if len(parents) < 2:
             parents.append(nodemod.nullid)
         if len(parents) < 2:
             parents.append(nodemod.nullid)
         p2 = parents.pop(0)
 
         text = commit.desc
 
         sha1s = re.findall(sha1re, text)
         for sha1 in sha1s:
             oldrev = source.lookuprev(sha1)
             newrev = revmap.get(oldrev)
             if newrev is not None:
                 text = text.replace(sha1, newrev[:len(sha1)])
 
         extra = commit.extra.copy()
 
         sourcename = self.repo.ui.config('convert', 'hg.sourcename')
         if sourcename:
             extra['convert_source'] = sourcename
 
         for label in ('source', 'transplant_source', 'rebase_source',
                       'intermediate-source'):
             node = extra.get(label)
 
             if node is None:
                 continue
 
             # Only transplant stores its reference in binary
             if label == 'transplant_source':
                 node = nodemod.hex(node)
 
             newrev = revmap.get(node)
             if newrev is not None:
                 if label == 'transplant_source':
                     newrev = nodemod.bin(newrev)
 
                 extra[label] = newrev
 
         if self.branchnames and commit.branch:
             extra['branch'] = commit.branch
         if commit.rev and commit.saverev:
             extra['convert_revision'] = commit.rev
 
         while parents:
             p1 = p2
             p2 = parents.pop(0)
             p1ctx = self.repo[p1]
             p2ctx = None
             if p2 != nodemod.nullid:
                 p2ctx = self.repo[p2]
             fileset = set(files)
             if full:
                 fileset.update(self.repo[p1])
                 fileset.update(self.repo[p2])
 
             if p2ctx:
                 p2files = set(cleanp2)
                 for file in self._calculatemergedfiles(source, p1ctx, p2ctx):
                     p2files.add(file)
                     fileset.add(file)
 
             ctx = context.memctx(self.repo, (p1, p2), text, fileset,
                                  getfilectx, commit.author, commit.date, extra)
 
             # We won't know if the conversion changes the node until after the
             # commit, so copy the source's phase for now.
             self.repo.ui.setconfig('phases', 'new-commit',
                                    phases.phasenames[commit.phase], 'convert')
 
             with self.repo.transaction("convert") as tr:
                 node = nodemod.hex(self.repo.commitctx(ctx))
 
                 # If the node value has changed, but the phase is lower than
                 # draft, set it back to draft since it hasn't been exposed
                 # anywhere.
                 if commit.rev != node:
                     ctx = self.repo[node]
                     if ctx.phase() < phases.draft:
                         phases.retractboundary(self.repo, tr, phases.draft,
                                                [ctx.node()])
 
             text = "(octopus merge fixup)\n"
             p2 = node
 
         if self.filemapmode and nparents == 1:
             man = self.repo.manifest
             mnode = self.repo.changelog.read(nodemod.bin(p2))[0]
             closed = 'close' in commit.extra
             if not closed and not man.cmp(m1node, man.revision(mnode)):
                 self.ui.status(_("filtering out empty revision\n"))
                 self.repo.rollback(force=True)
                 return parent
         return p2
 
     def puttags(self, tags):
         try:
             parentctx = self.repo[self.tagsbranch]
             tagparent = parentctx.node()
         except error.RepoError:
             parentctx = None
             tagparent = nodemod.nullid
 
         oldlines = set()
         for branch, heads in self.repo.branchmap().iteritems():
             for h in heads:
                 if '.hgtags' in self.repo[h]:
                     oldlines.update(
                         set(self.repo[h]['.hgtags'].data().splitlines(True)))
         oldlines = sorted(list(oldlines))
 
         newlines = sorted([("%s %s\n" % (tags[tag], tag)) for tag in tags])
         if newlines == oldlines:
             return None, None
 
         # if the old and new tags match, then there is nothing to update
         oldtags = set()
         newtags = set()
         for line in oldlines:
             s = line.strip().split(' ', 1)
             if len(s) != 2:
                 continue
             oldtags.add(s[1])
         for line in newlines:
             s = line.strip().split(' ', 1)
             if len(s) != 2:
                 continue
             if s[1] not in oldtags:
                 newtags.add(s[1].strip())
 
         if not newtags:
             return None, None
 
         data = "".join(newlines)
         def getfilectx(repo, memctx, f):
             return context.memfilectx(repo, f, data, False, False, None)
 
         self.ui.status(_("updating tags\n"))
         date = "%s 0" % int(time.mktime(time.gmtime()))
         extra = {'branch': self.tagsbranch}
         ctx = context.memctx(self.repo, (tagparent, None), "update tags",
                              [".hgtags"], getfilectx, "convert-repo", date,
                              extra)
         node = self.repo.commitctx(ctx)
         return nodemod.hex(node), nodemod.hex(tagparent)
 
     def setfilemapmode(self, active):
         self.filemapmode = active
 
     def putbookmarks(self, updatedbookmark):
         if not len(updatedbookmark):
             return
         wlock = lock = tr = None
         try:
             wlock = self.repo.wlock()
             lock = self.repo.lock()
             tr = self.repo.transaction('bookmark')
             self.ui.status(_("updating bookmarks\n"))
             destmarks = self.repo._bookmarks
             for bookmark in updatedbookmark:
                 destmarks[bookmark] = nodemod.bin(updatedbookmark[bookmark])
             destmarks.recordchange(tr)
             tr.close()
         finally:
             lockmod.release(lock, wlock, tr)
 
     def hascommitfrommap(self, rev):
         # the exact semantics of clonebranches is unclear so we can't say no
         return rev in self.repo or self.clonebranches
 
     def hascommitforsplicemap(self, rev):
         if rev not in self.repo and self.clonebranches:
             raise error.Abort(_('revision %s not found in destination '
                                'repository (lookups with clonebranches=true '
                                'are not implemented)') % rev)
         return rev in self.repo
 
 class mercurial_source(common.converter_source):
     def __init__(self, ui, path, revs=None):
         common.converter_source.__init__(self, ui, path, revs)
         self.ignoreerrors = ui.configbool('convert', 'hg.ignoreerrors', False)
         self.ignored = set()
         self.saverev = ui.configbool('convert', 'hg.saverev', False)
         try:
             self.repo = hg.repository(self.ui, path)
             # try to provoke an exception if this isn't really a hg
             # repo, but some other bogus compatible-looking url
             if not self.repo.local():
                 raise error.RepoError
         except error.RepoError:
             ui.traceback()
             raise NoRepo(_("%s is not a local Mercurial repository") % path)
         self.lastrev = None
         self.lastctx = None
         self._changescache = None, None
         self.convertfp = None
         # Restrict converted revisions to startrev descendants
         startnode = ui.config('convert', 'hg.startrev')
         hgrevs = ui.config('convert', 'hg.revs')
         if hgrevs is None:
             if startnode is not None:
                 try:
                     startnode = self.repo.lookup(startnode)
                 except error.RepoError:
                     raise error.Abort(_('%s is not a valid start revision')
                                      % startnode)
                 startrev = self.repo.changelog.rev(startnode)
                 children = {startnode: 1}
                 for r in self.repo.changelog.descendants([startrev]):
                     children[self.repo.changelog.node(r)] = 1
                 self.keep = children.__contains__
             else:
                 self.keep = util.always
             if revs:
                 self._heads = [self.repo[r].node() for r in revs]
             else:
                 self._heads = self.repo.heads()
         else:
             if revs or startnode is not None:
                 raise error.Abort(_('hg.revs cannot be combined with '
                                    'hg.startrev or --rev'))
             nodes = set()
             parents = set()
             for r in scmutil.revrange(self.repo, [hgrevs]):
                 ctx = self.repo[r]
                 nodes.add(ctx.node())
                 parents.update(p.node() for p in ctx.parents())
             self.keep = nodes.__contains__
             self._heads = nodes - parents
 
     def _changectx(self, rev):
         if self.lastrev != rev:
             self.lastctx = self.repo[rev]
             self.lastrev = rev
         return self.lastctx
 
     def _parents(self, ctx):
         return [p for p in ctx.parents() if p and self.keep(p.node())]
 
     def getheads(self):
         return [nodemod.hex(h) for h in self._heads if self.keep(h)]
 
     def getfile(self, name, rev):
         try:
             fctx = self._changectx(rev)[name]
             return fctx.data(), fctx.flags()
         except error.LookupError:
             return None, None
 
     def _changedfiles(self, ctx1, ctx2):
         ma, r = [], []
         maappend = ma.append
         rappend = r.append
         d = ctx1.manifest().diff(ctx2.manifest())
         for f, ((node1, flag1), (node2, flag2)) in d.iteritems():
             if node2 is None:
                 rappend(f)
             else:
                 maappend(f)
         return ma, r
 
     def getchanges(self, rev, full):
         ctx = self._changectx(rev)
         parents = self._parents(ctx)
         if full or not parents:
             files = copyfiles = ctx.manifest()
         if parents:
             if self._changescache[0] == rev:
                 ma, r = self._changescache[1]
             else:
                 ma, r = self._changedfiles(parents[0], ctx)
             if not full:
                 files = ma + r
             copyfiles = ma
         # _getcopies() is also run for roots and before filtering so missing
         # revlogs are detected early
         copies = self._getcopies(ctx, parents, copyfiles)
         cleanp2 = set()
         if len(parents) == 2:
             d = parents[1].manifest().diff(ctx.manifest(), clean=True)
             for f, value in d.iteritems():
                 if value is None:
                     cleanp2.add(f)
         changes = [(f, rev) for f in files if f not in self.ignored]
         changes.sort()
         return changes, copies, cleanp2
 
     def _getcopies(self, ctx, parents, files):
         copies = {}
         for name in files:
             if name in self.ignored:
                 continue
             try:
                 copysource, _copynode = ctx.filectx(name).renamed()
                 if copysource in self.ignored:
                     continue
                 # Ignore copy sources not in parent revisions
                 found = False
                 for p in parents:
                     if copysource in p:
                         found = True
                         break
                 if not found:
                     continue
                 copies[name] = copysource
             except TypeError:
                 pass
             except error.LookupError as e:
                 if not self.ignoreerrors:
                     raise
                 self.ignored.add(name)
                 self.ui.warn(_('ignoring: %s\n') % e)
         return copies
 
     def getcommit(self, rev):
         ctx = self._changectx(rev)
-        parents = [p.hex() for p in self._parents(ctx)]
+        _parents = self._parents(ctx)
+        parents = [p.hex() for p in _parents]
+        optparents = [p.hex() for p in ctx.parents() if p and p not in _parents]
         crev = rev
 
         return common.commit(author=ctx.user(),
                              date=util.datestr(ctx.date(),
                                                '%Y-%m-%d %H:%M:%S %1%2'),
                              desc=ctx.description(),
                              rev=crev,
                              parents=parents,
+                             optparents=optparents,
                              branch=ctx.branch(),
                              extra=ctx.extra(),
                              sortkey=ctx.rev(),
                              saverev=self.saverev,
                              phase=ctx.phase())
 
     def gettags(self):
         # This will get written to .hgtags, filter non global tags out.
         tags = [t for t in self.repo.tagslist()
                 if self.repo.tagtype(t[0]) == 'global']
         return dict([(name, nodemod.hex(node)) for name, node in tags
                      if self.keep(node)])
 
     def getchangedfiles(self, rev, i):
         ctx = self._changectx(rev)
         parents = self._parents(ctx)
         if not parents and i is None:
             i = 0
             ma, r = ctx.manifest().keys(), []
         else:
             i = i or 0
             ma, r = self._changedfiles(parents[i], ctx)
         ma, r = [[f for f in l if f not in self.ignored] for l in (ma, r)]
 
         if i == 0:
             self._changescache = (rev, (ma, r))
 
         return ma + r
 
     def converted(self, rev, destrev):
         if self.convertfp is None:
             self.convertfp = open(self.repo.join('shamap'), 'a')
         self.convertfp.write('%s %s\n' % (destrev, rev))
         self.convertfp.flush()
 
     def before(self):
         self.ui.debug('run hg source pre-conversion action\n')
 
     def after(self):
         self.ui.debug('run hg source post-conversion action\n')
 
     def hasnativeorder(self):
         return True
 
     def hasnativeclose(self):
         return True
 
     def lookuprev(self, rev):
         try:
             return nodemod.hex(self.repo.lookup(rev))
         except (error.RepoError, error.LookupError):
             return None
 
     def getbookmarks(self):
         return bookmarks.listbookmarks(self.repo)
 
     def checkrevformat(self, revstr, mapname='splicemap'):
         """ Mercurial, revision string is a 40 byte hex """
         self.checkhexformat(revstr, mapname)

tests/test-convert-hg-startrev.t

@@ -1,245 +1,245 b''
 
   $ cat >> $HGRCPATH <<EOF
   > [extensions]
   > convert =
   > [convert]
   > hg.saverev = yes
   > EOF
 
   $ glog()
   > {
   >     hg -R "$1" log -G --template '{rev} "{desc}" files: {files}\n'
   > }
 
   $ hg init source
   $ cd source
 
   $ echo a > a
   $ echo b > b
   $ echo f > f
   $ hg ci -d '0 0' -qAm '0: add a b f'
   $ echo c > c
   $ hg move f d
   $ hg ci -d '1 0' -qAm '1: add c, move f to d'
   $ hg copy a e
   $ echo b >> b
   $ hg ci -d '2 0' -qAm '2: copy e from a, change b'
   $ hg up -C 0
   2 files updated, 0 files merged, 3 files removed, 0 files unresolved
   $ echo a >> a
   $ hg ci -d '3 0' -qAm '3: change a'
   $ hg merge
   merging a and e to e
   3 files updated, 1 files merged, 1 files removed, 0 files unresolved
   (branch merge, don't forget to commit)
   $ hg ci -d '4 0' -qAm '4: merge 2 and 3'
   $ echo a >> a
   $ hg ci -d '5 0' -qAm '5: change a'
   $ cd ..
 
 Convert from null revision
 
   $ hg convert --config convert.hg.startrev=null source full
   initializing destination full repository
   scanning source...
   sorting...
   converting...
   5 0: add a b f
   4 1: add c, move f to d
   3 2: copy e from a, change b
   2 3: change a
   1 4: merge 2 and 3
   0 5: change a
 
   $ glog full
   o  5 "5: change a" files: a
   |
   o    4 "4: merge 2 and 3" files: e f
   |\
   | o  3 "3: change a" files: a
   | |
   o |  2 "2: copy e from a, change b" files: b e
   | |
   o |  1 "1: add c, move f to d" files: c d f
   |/
   o  0 "0: add a b f" files: a b f
   
   $ rm -Rf full
 
 Convert from zero revision
 
   $ hg convert --config convert.hg.startrev=0 source full
   initializing destination full repository
   scanning source...
   sorting...
   converting...
   5 0: add a b f
   4 1: add c, move f to d
   3 2: copy e from a, change b
   2 3: change a
   1 4: merge 2 and 3
   0 5: change a
 
   $ glog full
   o  5 "5: change a" files: a
   |
   o    4 "4: merge 2 and 3" files: e f
   |\
   | o  3 "3: change a" files: a
   | |
   o |  2 "2: copy e from a, change b" files: b e
   | |
   o |  1 "1: add c, move f to d" files: c d f
   |/
   o  0 "0: add a b f" files: a b f
   
 Convert from merge parent
 
   $ hg convert --config convert.hg.startrev=1 source conv1
   initializing destination conv1 repository
   scanning source...
   sorting...
   converting...
   3 1: add c, move f to d
   2 2: copy e from a, change b
   1 4: merge 2 and 3
   0 5: change a
 
   $ glog conv1
   o  3 "5: change a" files: a
   |
   o  2 "4: merge 2 and 3" files: a e
   |
   o  1 "2: copy e from a, change b" files: b e
   |
   o  0 "1: add c, move f to d" files: a b c d
   
   $ cd conv1
   $ hg up -q
 
 Check copy preservation
 
   $ hg st -C --change 2 e
   M e
   $ hg st -C --change 1 e
   A e
     a
   $ hg st -C --change 0 a
   A a
 
 (It seems like a bug in log that the following doesn't show rev 1.)
 
   $ hg log --follow --copies e
   changeset:   2:82bbac3d2cf4
   user:        test
   date:        Thu Jan 01 00:00:04 1970 +0000
   summary:     4: merge 2 and 3
   
   changeset:   0:23c3be426dce
   user:        test
   date:        Thu Jan 01 00:00:01 1970 +0000
   summary:     1: add c, move f to d
   
 Check copy removal on missing parent
 
   $ hg log --follow --copies d
   changeset:   0:23c3be426dce
   user:        test
   date:        Thu Jan 01 00:00:01 1970 +0000
   summary:     1: add c, move f to d
   
   $ hg cat -r tip a b
   a
   a
   a
   b
   b
   $ hg -q verify
   $ cd ..
 
 Convert from merge
 
   $ hg convert --config convert.hg.startrev=4 source conv4
   initializing destination conv4 repository
   scanning source...
   sorting...
   converting...
   1 4: merge 2 and 3
   0 5: change a
   $ glog conv4
   o  1 "5: change a" files: a
   |
   o  0 "4: merge 2 and 3" files: a b c d e
   
   $ cd conv4
   $ hg up -C
   5 files updated, 0 files merged, 0 files removed, 0 files unresolved
   $ hg cat -r tip a b
   a
   a
   a
   b
   b
   $ hg -q verify
   $ cd ..
 
 Convert from revset in convert.hg.revs
 
   $ hg convert --config convert.hg.revs='3:4+0' source revsetrepo
   initializing destination revsetrepo repository
   scanning source...
   sorting...
   converting...
   2 0: add a b f
   1 3: change a
   0 4: merge 2 and 3
 
   $ glog revsetrepo
   o  2 "4: merge 2 and 3" files: b c d e f
   |
   o  1 "3: change a" files: a
   |
   o  0 "0: add a b f" files: a b f
   
 Convert from specified revs
 
   $ hg convert --rev 3 --rev 2 source multiplerevs
   initializing destination multiplerevs repository
   scanning source...
   sorting...
   converting...
   3 0: add a b f
   2 1: add c, move f to d
   1 2: copy e from a, change b
   0 3: change a
   $ glog multiplerevs
   o  3 "3: change a" files: a
   |
   | o  2 "2: copy e from a, change b" files: b e
   | |
   | o  1 "1: add c, move f to d" files: c d f
   |/
   o  0 "0: add a b f" files: a b f
   
 Convert in multiple steps that doesn't overlap - the link to the parent is
-currently missing
+preserved anyway
 
   $ hg convert --config convert.hg.revs=::1 source multistep
   initializing destination multistep repository
   scanning source...
   sorting...
   converting...
   1 0: add a b f
   0 1: add c, move f to d
   $ hg convert --config convert.hg.revs=2 source multistep
   scanning source...
   sorting...
   converting...
   0 2: copy e from a, change b
   $ glog multistep
-  o  2 "2: copy e from a, change b" files: a b c d e
-  
+  o  2 "2: copy e from a, change b" files: b e
+  |
   o  1 "1: add c, move f to d" files: c d f
   |
   o  0 "0: add a b f" files: a b f