convert: look up copies in getchanges instead of getcommit

svn: defer path expansion until getchanges to reduce latency, as well as memory
usage when converting incrementally.

hgext/convert/__init__.py

@@ -1,441 +1,442
 # convert.py Foreign SCM converter
 #
 # Copyright 2005-2007 Matt Mackall <[e-mail]>
 #
 # This software may be used and distributed according to the terms
 # of the GNU General Public License, incorporated herein by reference.
 
 from common import NoRepo, converter_source, converter_sink
 from cvs import convert_cvs
 from git import convert_git
 from hg import mercurial_source, mercurial_sink
 from subversion import convert_svn
 
 import os, shlex, shutil
 from mercurial import hg, ui, util, commands
 from mercurial.i18n import _
 
 commands.norepo += " convert"
 
 converters = [convert_cvs, convert_git, convert_svn, mercurial_source,
               mercurial_sink]
 
 def convertsource(ui, path, **opts):
     for c in converters:
         try:
             return c.getcommit and c(ui, path, **opts)
         except (AttributeError, NoRepo):
             pass
     raise util.Abort('%s: unknown repository type' % path)
 
 def convertsink(ui, path):
     if not os.path.isdir(path):
         raise util.Abort("%s: not a directory" % path)
     for c in converters:
         try:
             return c.putcommit and c(ui, path)
         except (AttributeError, NoRepo):
             pass
     raise util.Abort('%s: unknown repository type' % path)
 
 class convert(object):
     def __init__(self, ui, source, dest, revmapfile, filemapper, opts):
 
         self.source = source
         self.dest = dest
         self.ui = ui
         self.opts = opts
         self.commitcache = {}
         self.revmapfile = revmapfile
         self.revmapfilefd = None
         self.authors = {}
         self.authorfile = None
         self.mapfile = filemapper
 
         self.map = {}
         try:
             origrevmapfile = open(self.revmapfile, 'r')
             for l in origrevmapfile:
                 sv, dv = l[:-1].split()
                 self.map[sv] = dv
             origrevmapfile.close()
         except IOError:
             pass
 
         # 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('authors'):
             self.readauthormap(opts.get('authors'))
             self.authorfile = self.dest.authorfile()
 
     def walktree(self, heads):
         '''Return a mapping that identifies the uncommitted parents of every
         uncommitted changeset.'''
         visit = heads
         known = {}
         parents = {}
         while visit:
             n = visit.pop(0)
             if n in known or n in self.map: continue
             known[n] = 1
             self.commitcache[n] = self.source.getcommit(n)
             cp = self.commitcache[n].parents
             parents[n] = []
             for p in cp:
                 parents[n].append(p)
                 visit.append(p)
 
         return parents
 
     def toposort(self, parents):
         '''Return an ordering such that every uncommitted changeset is
         preceeded by all its uncommitted ancestors.'''
         visit = parents.keys()
         seen = {}
         children = {}
 
         while visit:
             n = visit.pop(0)
             if n in seen: continue
             seen[n] = 1
             # Ensure that nodes without parents are present in the 'children'
             # mapping.
             children.setdefault(n, [])
             for p in parents[n]:
                 if not p in self.map:
                     visit.append(p)
                 children.setdefault(p, []).append(n)
 
         s = []
         removed = {}
         visit = children.keys()
         while visit:
             n = visit.pop(0)
             if n in removed: continue
             dep = 0
             if n in parents:
                 for p in parents[n]:
                     if p in self.map: continue
                     if p not in removed:
                         # we're still dependent
                         visit.append(n)
                         dep = 1
                         break
 
             if not dep:
                 # all n's parents are in the list
                 removed[n] = 1
                 if n not in self.map:
                     s.append(n)
                 if n in children:
                     for c in children[n]:
                         visit.insert(0, c)
 
         if self.opts.get('datesort'):
             depth = {}
             for n in s:
                 depth[n] = 0
                 pl = [p for p in self.commitcache[n].parents
                       if p not in self.map]
                 if pl:
                     depth[n] = max([depth[p] for p in pl]) + 1
 
             s = [(depth[n], self.commitcache[n].date, n) for n in s]
             s.sort()
             s = [e[2] for e in s]
 
         return s
 
     def mapentry(self, src, dst):
         if self.revmapfilefd is None:
             try:
                 self.revmapfilefd = open(self.revmapfile, "a")
             except IOError, (errno, strerror):
                 raise util.Abort("Could not open map file %s: %s, %s\n" % (self.revmapfile, errno, strerror))
         self.map[src] = dst
         self.revmapfilefd.write("%s %s\n" % (src, dst))
         self.revmapfilefd.flush()
 
     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:
             try:
                 srcauthor = line.split('=')[0].strip()
                 dstauthor = line.split('=')[1].strip()
                 if srcauthor in self.authors and dstauthor != self.authors[srcauthor]:
                     self.ui.status(
                         'Overriding mapping for author %s, was %s, will be %s\n'
                         % (srcauthor, self.authors[srcauthor], dstauthor))
                 else:
                     self.ui.debug('Mapping author %s to %s\n'
                                   % (srcauthor, dstauthor))
                     self.authors[srcauthor] = dstauthor
             except IndexError:
                 self.ui.warn(
                     'Ignoring bad line in author file map %s: %s\n'
                     % (authorfile, line))
         afile.close()
 
     def copy(self, rev):
         commit = self.commitcache[rev]
         do_copies = hasattr(self.dest, 'copyfile')
         filenames = []
 
-        for f, v in self.source.getchanges(rev):
+        files, copies = self.source.getchanges(rev)
+        for f, v in files:
             newf = self.mapfile(f)
             if not newf:
                 continue
             filenames.append(newf)
             try:
                 data = self.source.getfile(f, v)
             except IOError, inst:
                 self.dest.delfile(newf)
             else:
                 e = self.source.getmode(f, v)
                 self.dest.putfile(newf, e, data)
                 if do_copies:
-                    if f in commit.copies:
-                        copyf = self.mapfile(commit.copies[f])
+                    if f in copies:
+                        copyf = self.mapfile(copies[f])
                         if copyf:
                             # Merely marks that a copy happened.
                             self.dest.copyfile(copyf, newf)
 
         parents = [self.map[r] for r in commit.parents]
         newnode = self.dest.putcommit(filenames, parents, commit)
         self.mapentry(rev, newnode)
 
     def convert(self):
         try:
             self.dest.before()
             self.source.setrevmap(self.map)
             self.ui.status("scanning source...\n")
             heads = self.source.getheads()
             parents = self.walktree(heads)
             self.ui.status("sorting...\n")
             t = self.toposort(parents)
             num = len(t)
             c = None
 
             self.ui.status("converting...\n")
             for c in t:
                 num -= 1
                 desc = self.commitcache[c].desc
                 if "\n" in desc:
                     desc = desc.splitlines()[0]
                 author = self.commitcache[c].author
                 author = self.authors.get(author, author)
                 self.commitcache[c].author = author
                 self.ui.status("%d %s\n" % (num, desc))
                 self.copy(c)
 
             tags = self.source.gettags()
             ctags = {}
             for k in tags:
                 v = tags[k]
                 if v in self.map:
                     ctags[k] = self.map[v]
 
             if c and ctags:
                 nrev = self.dest.puttags(ctags)
                 # write another hash correspondence to override the previous
                 # one so we don't end up with extra tag heads
                 if nrev:
                     self.mapentry(c, nrev)
 
             self.writeauthormap()
         finally:
             self.cleanup()
 
     def cleanup(self):
         self.dest.after()
         if self.revmapfilefd:
             self.revmapfilefd.close()
 
 def rpairs(name):
     e = len(name)
     while e != -1:
         yield name[:e], name[e+1:]
         e = name.rfind('/', 0, e)
 
 class filemapper(object):
     '''Map and filter filenames when importing.
     A name can be mapped to itself, a new name, or None (omit from new
     repository).'''
 
     def __init__(self, ui, path=None):
         self.ui = ui
         self.include = {}
         self.exclude = {}
         self.rename = {}
         if path:
             if self.parse(path):
                 raise util.Abort(_('errors in filemap'))
 
     def parse(self, path):
         errs = 0
         def check(name, mapping, listname):
             if name in mapping:
                 self.ui.warn(_('%s:%d: %r already in %s list\n') %
                              (lex.infile, lex.lineno, name, listname))
                 return 1
             return 0
         lex = shlex.shlex(open(path), path, True)
         lex.wordchars += '!@#$%^&*()-=+[]{}|;:,./<>?'
         cmd = lex.get_token()
         while cmd:
             if cmd == 'include':
                 name = lex.get_token()
                 errs += check(name, self.exclude, 'exclude')
                 self.include[name] = name
             elif cmd == 'exclude':
                 name = lex.get_token()
                 errs += check(name, self.include, 'include')
                 errs += check(name, self.rename, 'rename')
                 self.exclude[name] = name
             elif cmd == 'rename':
                 src = lex.get_token()
                 dest = lex.get_token()
                 errs += check(src, self.exclude, 'exclude')
                 self.rename[src] = dest
             elif cmd == 'source':
                 errs += self.parse(lex.get_token())
             else:
                 self.ui.warn(_('%s:%d: unknown directive %r\n') %
                              (lex.infile, lex.lineno, cmd))
                 errs += 1
             cmd = lex.get_token()
         return errs
 
     def lookup(self, name, mapping):
         for pre, suf in rpairs(name):
             try:
                 return mapping[pre], pre, suf
             except KeyError, err:
                 pass
         return '', name, ''
         
     def __call__(self, name):
         if self.include:
             inc = self.lookup(name, self.include)[0]
         else:
             inc = name
         if self.exclude:
             exc = self.lookup(name, self.exclude)[0]
         else:
             exc = ''
         if not inc or exc:
             return None
         newpre, pre, suf = self.lookup(name, self.rename)
         if newpre:
             if suf:
                 return newpre + '/' + suf
             return newpre
         return name
 
 def _convert(ui, src, dest=None, revmapfile=None, **opts):
     """Convert a foreign SCM repository to a Mercurial one.
 
     Accepted source formats:
     - GIT
     - CVS
     - SVN
 
     Accepted destination formats:
     - Mercurial
 
     If no revision is given, all revisions will be converted. Otherwise,
     convert will only import up to the named revision (given in a format
     understood by the source).
 
     If no destination directory name is specified, it defaults to the
     basename of the source with '-hg' appended.  If the destination
     repository doesn't exist, it will be created.
 
     If <revmapfile> isn't given, it will be put in a default location
     (<dest>/.hg/shamap by default).  The <revmapfile> is a simple text
     file that maps each source commit ID to the destination ID for
     that revision, like so:
     <source ID> <destination ID>
 
     If the file doesn't exist, it's automatically created.  It's updated
     on each commit copied, so convert-repo can be interrupted and can
     be run repeatedly to copy new commits.
 
     The [username mapping] file is a simple text file that maps each source
     commit author to a destination commit author. It is handy for source SCMs
     that use unix logins to identify authors (eg: CVS). One line per author
     mapping and the line format is:
     srcauthor=whatever string you want
     """
 
     util._encoding = 'UTF-8'
 
     if not dest:
         dest = hg.defaultdest(src) + "-hg"
         ui.status("assuming destination %s\n" % dest)
 
     # Try to be smart and initalize things when required
     created = False
     if os.path.isdir(dest):
         if len(os.listdir(dest)) > 0:
             try:
                 hg.repository(ui, dest)
                 ui.status("destination %s is a Mercurial repository\n" % dest)
             except hg.RepoError:
                 raise util.Abort(
                     "destination directory %s is not empty.\n"
                     "Please specify an empty directory to be initialized\n"
                     "or an already initialized mercurial repository"
                     % dest)
         else:
             ui.status("initializing destination %s repository\n" % dest)
             hg.repository(ui, dest, create=True)
             created = True
     elif os.path.exists(dest):
         raise util.Abort("destination %s exists and is not a directory" % dest)
     else:
         ui.status("initializing destination %s repository\n" % dest)
         hg.repository(ui, dest, create=True)
         created = True
 
     destc = convertsink(ui, dest)
 
     try:
         srcc = convertsource(ui, src, rev=opts.get('rev'))
     except Exception:
         if created:
             shutil.rmtree(dest, True)
         raise
 
     if not revmapfile:
         try:
             revmapfile = destc.revmapfile()
         except:
             revmapfile = os.path.join(destc, "map")
 
 
     c = convert(ui, srcc, destc, revmapfile, filemapper(ui, opts['filemap']),
                 opts)
     c.convert()
 
 cmdtable = {
     "convert":
         (_convert,
          [('A', 'authors', '', 'username mapping filename'),
           ('', 'filemap', '', 'remap file names using contents of file'),
           ('r', 'rev', '', 'import up to target revision REV'),
           ('', 'datesort', None, 'try to sort changesets by date')],
          'hg convert [OPTION]... SOURCE [DEST [MAPFILE]]'),
 }

hgext/convert/common.py

@@ -1,120 +1,120
 # common code for the convert extension
 
 class NoRepo(Exception): pass
 
 class commit(object):
-    def __init__(self, author, date, desc, parents, branch=None, rev=None,
-                 copies={}):
+    def __init__(self, author, date, desc, parents, branch=None, rev=None):
         self.author = author
         self.date = date
         self.desc = desc
         self.parents = parents
         self.branch = branch
         self.rev = rev
-        self.copies = copies
 
 class converter_source(object):
     """Conversion source interface"""
 
     def __init__(self, ui, path, rev=None):
         """Initialize conversion source (or raise NoRepo("message")
         exception if path is not a valid repository)"""
         self.ui = ui
         self.path = path
         self.rev = rev
 
         self.encoding = 'utf-8'
 
     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 file contents as a string"""
         raise NotImplementedError()
 
     def getmode(self, name, rev):
         """Return file mode, eg. '', 'x', or 'l'"""
         raise NotImplementedError()
 
     def getchanges(self, version):
-        """Return sorted list of (filename, id) tuples for all files changed in rev.
+        """Returns a tuple of (files, copies)
+        Files is a sorted list of (filename, id) tuples for all files changed
+        in version, where id is the source revision id of the file.
 
-        id just tells us which revision to return in getfile(), e.g. in
-        git it's an object hash."""
+        copies is a dictionary of dest: source
+        """
         raise NotImplementedError()
 
     def getcommit(self, version):
         """Return the commit object for version"""
         raise NotImplementedError()
 
     def gettags(self):
         """Return the tags as a dictionary of name: revision"""
         raise NotImplementedError()
 
     def recode(self, s, encoding=None):
         if not encoding:
             encoding = self.encoding or 'utf-8'
 
         try:
             return s.decode(encoding).encode("utf-8")
         except:
             try:
                 return s.decode("latin-1").encode("utf-8")
             except:
                 return s.decode(encoding, "replace").encode("utf-8")
 
 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)"""
         raise NotImplementedError()
 
     def getheads(self):
         """Return a list of this repository's heads"""
         raise NotImplementedError()
 
     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 putfile(self, f, e, data):
         """Put file for next putcommit().
         f: path to file
         e: '', 'x', or 'l' (regular file, executable, or symlink)
         data: file contents"""
         raise NotImplementedError()
 
     def delfile(self, f):
         """Delete file for next putcommit().
         f: path to file"""
         raise NotImplementedError()
 
     def putcommit(self, files, parents, commit):
         """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.
         Called after putfile() and delfile() calls. 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, ...}"""
         raise NotImplementedError()

hgext/convert/cvs.py

@@ -1,259 +1,259
 # CVS conversion code inspired by hg-cvs-import and git-cvsimport
 
 import os, locale, re, socket
 from mercurial import util
 
 from common import NoRepo, commit, converter_source
 
 class convert_cvs(converter_source):
     def __init__(self, ui, path, rev=None):
         super(convert_cvs, self).__init__(ui, path, rev=rev)
 
         cvs = os.path.join(path, "CVS")
         if not os.path.exists(cvs):
             raise NoRepo("couldn't open CVS repo %s" % path)
 
         self.changeset = {}
         self.files = {}
         self.tags = {}
         self.lastbranch = {}
         self.parent = {}
         self.socket = None
         self.cvsroot = file(os.path.join(cvs, "Root")).read()[:-1]
         self.cvsrepo = file(os.path.join(cvs, "Repository")).read()[:-1]
         self.encoding = locale.getpreferredencoding()
         self._parse()
         self._connect()
 
     def _parse(self):
         if self.changeset:
             return
 
         maxrev = 0
         cmd = 'cvsps -A -u --cvs-direct -q'
         if self.rev:
             # TODO: handle tags
             try:
                 # patchset number?
                 maxrev = int(self.rev)
             except ValueError:
                 try:
                     # date
                     util.parsedate(self.rev, ['%Y/%m/%d %H:%M:%S'])
                     cmd = "%s -d '1970/01/01 00:00:01' -d '%s'" % (cmd, self.rev)
                 except util.Abort:
                     raise util.Abort('revision %s is not a patchset number or date' % self.rev)
 
         d = os.getcwd()
         try:
             os.chdir(self.path)
             id = None
             state = 0
             for l in os.popen(cmd):
                 if state == 0: # header
                     if l.startswith("PatchSet"):
                         id = l[9:-2]
                         if maxrev and int(id) > maxrev:
                             state = 3
                     elif l.startswith("Date"):
                         date = util.parsedate(l[6:-1], ["%Y/%m/%d %H:%M:%S"])
                         date = util.datestr(date)
                     elif l.startswith("Branch"):
                         branch = l[8:-1]
                         self.parent[id] = self.lastbranch.get(branch, 'bad')
                         self.lastbranch[branch] = id
                     elif l.startswith("Ancestor branch"):
                         ancestor = l[17:-1]
                         self.parent[id] = self.lastbranch[ancestor]
                     elif l.startswith("Author"):
                         author = self.recode(l[8:-1])
                     elif l.startswith("Tag:") or l.startswith("Tags:"):
                         t = l[l.index(':')+1:]
                         t = [ut.strip() for ut in t.split(',')]
                         if (len(t) > 1) or (t[0] and (t[0] != "(none)")):
                             self.tags.update(dict.fromkeys(t, id))
                     elif l.startswith("Log:"):
                         state = 1
                         log = ""
                 elif state == 1: # log
                     if l == "Members: \n":
                         files = {}
                         log = self.recode(log[:-1])
                         state = 2
                     else:
                         log += l
                 elif state == 2:
                     if l == "\n": #
                         state = 0
                         p = [self.parent[id]]
                         if id == "1":
                             p = []
                         if branch == "HEAD":
                             branch = ""
                         c = commit(author=author, date=date, parents=p,
                                    desc=log, branch=branch)
                         self.changeset[id] = c
                         self.files[id] = files
                     else:
                         colon = l.rfind(':')
                         file = l[1:colon]
                         rev = l[colon+1:-2]
                         rev = rev.split("->")[1]
                         files[file] = rev
                 elif state == 3:
                     continue
 
             self.heads = self.lastbranch.values()
         finally:
             os.chdir(d)
 
     def _connect(self):
         root = self.cvsroot
         conntype = None
         user, host = None, None
         cmd = ['cvs', 'server']
 
         self.ui.status("connecting to %s\n" % root)
 
         if root.startswith(":pserver:"):
             root = root[9:]
             m = re.match(r'(?:(.*?)(?::(.*?))?@)?([^:\/]*)(?::(\d*))?(.*)',
                          root)
             if m:
                 conntype = "pserver"
                 user, passw, serv, port, root = m.groups()
                 if not user:
                     user = "anonymous"
                 rr = ":pserver:" + user + "@" + serv + ":" +  root
                 if port:
                     rr2, port = "-", int(port)
                 else:
                     rr2, port = rr, 2401
                 rr += str(port)
 
                 if not passw:
                     passw = "A"
                     pf = open(os.path.join(os.environ["HOME"], ".cvspass"))
                     for l in pf:
                         # :pserver:[e-mail]:/cvsroot/zmailer Ah<Z
                         m = re.match(r'(/\d+\s+/)?(.*)', l)
                         l = m.group(2)
                         w, p = l.split(' ', 1)
                         if w in [rr, rr2]:
                             passw = p
                             break
                     pf.close()
 
                 sck = socket.socket()
                 sck.connect((serv, port))
                 sck.send("\n".join(["BEGIN AUTH REQUEST", root, user, passw,
                                     "END AUTH REQUEST", ""]))
                 if sck.recv(128) != "I LOVE YOU\n":
                     raise NoRepo("CVS pserver authentication failed")
 
                 self.writep = self.readp = sck.makefile('r+')
 
         if not conntype and root.startswith(":local:"):
             conntype = "local"
             root = root[7:]
 
         if not conntype:
             # :ext:user@host/home/user/path/to/cvsroot
             if root.startswith(":ext:"):
                 root = root[5:]
             m = re.match(r'(?:([^@:/]+)@)?([^:/]+):?(.*)', root)
             if not m:
                 conntype = "local"
             else:
                 conntype = "rsh"
                 user, host, root = m.group(1), m.group(2), m.group(3)
 
         if conntype != "pserver":
             if conntype == "rsh":
                 rsh = os.environ.get("CVS_RSH" or "rsh")
                 if user:
                     cmd = [rsh, '-l', user, host] + cmd
                 else:
                     cmd = [rsh, host] + cmd
 
             self.writep, self.readp = os.popen2(cmd)
 
         self.realroot = root
 
         self.writep.write("Root %s\n" % root)
         self.writep.write("Valid-responses ok error Valid-requests Mode"
                           " M Mbinary E Checked-in Created Updated"
                           " Merged Removed\n")
         self.writep.write("valid-requests\n")
         self.writep.flush()
         r = self.readp.readline()
         if not r.startswith("Valid-requests"):
             raise util.Abort("server sucks")
         if "UseUnchanged" in r:
             self.writep.write("UseUnchanged\n")
             self.writep.flush()
             r = self.readp.readline()
 
     def getheads(self):
         return self.heads
 
     def _getfile(self, name, rev):
         if rev.endswith("(DEAD)"):
             raise IOError
 
         args = ("-N -P -kk -r %s --" % rev).split()
         args.append(os.path.join(self.cvsrepo, name))
         for x in args:
             self.writep.write("Argument %s\n" % x)
         self.writep.write("Directory .\n%s\nco\n" % self.realroot)
         self.writep.flush()
 
         data = ""
         while 1:
             line = self.readp.readline()
             if line.startswith("Created ") or line.startswith("Updated "):
                 self.readp.readline() # path
                 self.readp.readline() # entries
                 mode = self.readp.readline()[:-1]
                 count = int(self.readp.readline()[:-1])
                 data = self.readp.read(count)
             elif line.startswith(" "):
                 data += line[1:]
             elif line.startswith("M "):
                 pass
             elif line.startswith("Mbinary "):
                 count = int(self.readp.readline()[:-1])
                 data = self.readp.read(count)
             else:
                 if line == "ok\n":
                     return (data, "x" in mode and "x" or "")
                 elif line.startswith("E "):
                     self.ui.warn("cvs server: %s\n" % line[2:])
                 elif line.startswith("Remove"):
                     l = self.readp.readline()
                     l = self.readp.readline()
                     if l != "ok\n":
                         raise util.Abort("unknown CVS response: %s" % l)
                 else:
                     raise util.Abort("unknown CVS response: %s" % line)
 
     def getfile(self, file, rev):
         data, mode = self._getfile(file, rev)
         self.modecache[(file, rev)] = mode
         return data
 
     def getmode(self, file, rev):
         return self.modecache[(file, rev)]
 
     def getchanges(self, rev):
         self.modecache = {}
         files = self.files[rev]
         cl = files.items()
         cl.sort()
-        return cl
+        return (cl, {})
 
     def getcommit(self, rev):
         return self.changeset[rev]
 
     def gettags(self):
         return self.tags

hgext/convert/git.py

@@ -1,101 +1,101
 # git support for the convert extension
 
 import os
 
 from common import NoRepo, commit, converter_source
 
 class convert_git(converter_source):
     def gitcmd(self, s):
         return os.popen('GIT_DIR=%s %s' % (self.path, s))
 
     def __init__(self, ui, path, rev=None):
         super(convert_git, self).__init__(ui, path, rev=rev)
 
         if os.path.isdir(path + "/.git"):
             path += "/.git"
         if not os.path.exists(path + "/objects"):
             raise NoRepo("couldn't open GIT repo %s" % path)
         self.path = path
 
     def getheads(self):
         if not self.rev:
             return self.gitcmd('git-rev-parse --branches').read().splitlines()
         else:
             fh = self.gitcmd("git-rev-parse --verify %s" % self.rev)
             return [fh.read()[:-1]]
 
     def catfile(self, rev, type):
         if rev == "0" * 40: raise IOError()
         fh = self.gitcmd("git-cat-file %s %s 2>/dev/null" % (type, rev))
         return fh.read()
 
     def getfile(self, name, rev):
         return self.catfile(rev, "blob")
 
     def getmode(self, name, rev):
         return self.modecache[(name, rev)]
 
     def getchanges(self, version):
         self.modecache = {}
         fh = self.gitcmd("git-diff-tree --root -m -r %s" % version)
         changes = []
         for l in fh:
             if "\t" not in l: continue
             m, f = l[:-1].split("\t")
             m = m.split()
             h = m[3]
             p = (m[1] == "100755")
             s = (m[1] == "120000")
             self.modecache[(f, h)] = (p and "x") or (s and "l") or ""
             changes.append((f, h))
-        return changes
+        return (changes, {})
 
     def getcommit(self, version):
         c = self.catfile(version, "commit") # read the commit hash
         end = c.find("\n\n")
         message = c[end+2:]
         message = self.recode(message)
         l = c[:end].splitlines()
         manifest = l[0].split()[1]
         parents = []
         for e in l[1:]:
             n, v = e.split(" ", 1)
             if n == "author":
                 p = v.split()
                 tm, tz = p[-2:]
                 author = " ".join(p[:-2])
                 if author[0] == "<": author = author[1:-1]
                 author = self.recode(author)
             if n == "committer":
                 p = v.split()
                 tm, tz = p[-2:]
                 committer = " ".join(p[:-2])
                 if committer[0] == "<": committer = committer[1:-1]
                 committer = self.recode(committer)
                 message += "\ncommitter: %s\n" % committer
             if n == "parent": parents.append(v)
 
         tzs, tzh, tzm = tz[-5:-4] + "1", tz[-4:-2], tz[-2:]
         tz = -int(tzs) * (int(tzh) * 3600 + int(tzm))
         date = tm + " " + str(tz)
         author = author or "unknown"
 
         c = commit(parents=parents, date=date, author=author, desc=message,
                    rev=version)
         return c
 
     def gettags(self):
         tags = {}
         fh = self.gitcmd('git-ls-remote --tags "%s" 2>/dev/null' % self.path)
         prefix = 'refs/tags/'
         for line in fh:
             line = line.strip()
             if not line.endswith("^{}"):
                 continue
             node, tag = line.split(None, 1)
             if not tag.startswith(prefix):
                 continue
             tag = tag[len(prefix):-3]
             tags[tag] = node
 
         return tags

hgext/convert/hg.py

@@ -1,175 +1,175
 # hg backend for convert extension
 
 # Note 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.
 
 
 import os, time
 from mercurial.i18n import _
 from mercurial.node import *
 from mercurial import hg, lock, revlog, util
 
 from common import NoRepo, commit, converter_source, converter_sink
 
 class mercurial_sink(converter_sink):
     def __init__(self, ui, path):
         self.path = path
         self.ui = ui
         try:
             self.repo = hg.repository(self.ui, path)
         except:
             raise NoRepo("could not open hg repo %s as sink" % path)
         self.lock = None
         self.wlock = None
         self.branchnames = ui.configbool('convert', 'hg.usebranchnames', True)
 
     def before(self):
         self.wlock = self.repo.wlock()
         self.lock = self.repo.lock()
 
     def after(self):
         self.lock = None
         self.wlock = None
 
     def revmapfile(self):
         return os.path.join(self.path, ".hg", "shamap")
 
     def authorfile(self):
         return os.path.join(self.path, ".hg", "authormap")
 
     def getheads(self):
         h = self.repo.changelog.heads()
         return [ hex(x) for x in h ]
 
     def putfile(self, f, e, data):
         self.repo.wwrite(f, data, e)
         if f not in self.repo.dirstate:
             self.repo.dirstate.add(f)
 
     def copyfile(self, source, dest):
         self.repo.copy(source, dest)
 
     def delfile(self, f):
         try:
             os.unlink(self.repo.wjoin(f))
             #self.repo.remove([f])
         except:
             pass
 
     def putcommit(self, files, parents, commit):
         if not files:
             return hex(self.repo.changelog.tip())
 
         seen = {hex(nullid): 1}
         pl = []
         for p in parents:
             if p not in seen:
                 pl.append(p)
                 seen[p] = 1
         parents = pl
 
         if len(parents) < 2: parents.append("0" * 40)
         if len(parents) < 2: parents.append("0" * 40)
         p2 = parents.pop(0)
 
         text = commit.desc
         extra = {}
         if self.branchnames and commit.branch:
             extra['branch'] = commit.branch
         if commit.rev:
             extra['convert_revision'] = commit.rev
 
         while parents:
             p1 = p2
             p2 = parents.pop(0)
             a = self.repo.rawcommit(files, text, commit.author, commit.date,
                                     bin(p1), bin(p2), extra=extra)
             self.repo.dirstate.invalidate()
             text = "(octopus merge fixup)\n"
             p2 = hg.hex(self.repo.changelog.tip())
 
         return p2
 
     def puttags(self, tags):
         try:
             old = self.repo.wfile(".hgtags").read()
             oldlines = old.splitlines(1)
             oldlines.sort()
         except:
             oldlines = []
 
         k = tags.keys()
         k.sort()
         newlines = []
         for tag in k:
             newlines.append("%s %s\n" % (tags[tag], tag))
 
         newlines.sort()
 
         if newlines != oldlines:
             self.ui.status("updating tags\n")
             f = self.repo.wfile(".hgtags", "w")
             f.write("".join(newlines))
             f.close()
             if not oldlines: self.repo.add([".hgtags"])
             date = "%s 0" % int(time.mktime(time.gmtime()))
             self.repo.rawcommit([".hgtags"], "update tags", "convert-repo",
                                 date, self.repo.changelog.tip(), nullid)
             return hex(self.repo.changelog.tip())
 
 class mercurial_source(converter_source):
     def __init__(self, ui, path, rev=None):
         converter_source.__init__(self, ui, path, rev)
         self.repo = hg.repository(self.ui, path)
         self.lastrev = None
         self.lastctx = None
 
     def changectx(self, rev):
         if self.lastrev != rev:
             self.lastctx = self.repo.changectx(rev)
             self.lastrev = rev
         return self.lastctx
 
     def getheads(self):
         return [hex(node) for node in self.repo.heads()]
 
     def getfile(self, name, rev):
         try:
             return self.changectx(rev).filectx(name).data()
         except revlog.LookupError, err:
             raise IOError(err)
 
     def getmode(self, name, rev):
         m = self.changectx(rev).manifest()
         return (m.execf(name) and 'x' or '') + (m.linkf(name) and 'l' or '')
 
     def getchanges(self, rev):
         ctx = self.changectx(rev)
         m, a, r = self.repo.status(ctx.parents()[0].node(), ctx.node())[:3]
         changes = [(name, rev) for name in m + a + r]
         changes.sort()
-        return changes
+        return (changes, self.getcopies(ctx))
 
     def getcopies(self, ctx):
         added = self.repo.status(ctx.parents()[0].node(), ctx.node())[1]
         copies = {}
         for name in added:
             try:
                 copies[name] = ctx.filectx(name).renamed()[0]
             except TypeError:
                 pass
         return copies
         
     def getcommit(self, rev):
         ctx = self.changectx(rev)
         parents = [hex(p.node()) for p in ctx.parents() if p.node() != nullid]
         return commit(author=ctx.user(), date=util.datestr(ctx.date()),
                       desc=ctx.description(), parents=parents,
-                      branch=ctx.branch(), copies=self.getcopies(ctx))
+                      branch=ctx.branch())
 
     def gettags(self):
         tags = [t for t in self.repo.tagslist() if t[0] != 'tip']
         return dict([(name, hex(node)) for name, node in tags])

hgext/convert/subversion.py

@@ -1,641 +1,643
 # Subversion 1.4/1.5 Python API backend
 #
 # Copyright(C) 2007 Daniel Holth et al
 #
 # Configuration options:
 #
 # convert.svn.trunk
 #   Relative path to the trunk (default: "trunk")
 # convert.svn.branches
 #   Relative path to tree of branches (default: "branches")
 #
 # Set these in a hgrc, or on the command line as follows:
 #
 #   hg convert --config convert.svn.trunk=wackoname [...]
 
 import locale
 import os
 import cPickle as pickle
 from mercurial import util
 
 # Subversion stuff. Works best with very recent Python SVN bindings
 # e.g. SVN 1.5 or backports. Thanks to the bzr folks for enhancing
 # these bindings.
 
 from cStringIO import StringIO
 
 from common import NoRepo, commit, converter_source
 
 try:
     from svn.core import SubversionException, Pool
     import svn
     import svn.client
     import svn.core
     import svn.ra
     import svn.delta
     import transport
 except ImportError:
     pass
 
 def geturl(path):
     try:
         return svn.client.url_from_path(svn.core.svn_path_canonicalize(path))
     except SubversionException:
         pass
     if os.path.isdir(path):
         return 'file://%s' % os.path.normpath(os.path.abspath(path))
     return path
 
 def optrev(number):
     optrev = svn.core.svn_opt_revision_t()
     optrev.kind = svn.core.svn_opt_revision_number
     optrev.value.number = number
     return optrev
 
 class changedpath(object):
     def __init__(self, p):
         self.copyfrom_path = p.copyfrom_path
         self.copyfrom_rev = p.copyfrom_rev
         self.action = p.action
 
 # SVN conversion code stolen from bzr-svn and tailor
 class convert_svn(converter_source):
     def __init__(self, ui, url, rev=None):
         super(convert_svn, self).__init__(ui, url, rev=rev)
 
         try:
             SubversionException
         except NameError:
             msg = 'subversion python bindings could not be loaded\n'
             ui.warn(msg)
             raise NoRepo(msg)
 
         self.encoding = locale.getpreferredencoding()
         self.lastrevs = {}
 
         latest = None
         if rev:
             try:
                 latest = int(rev)
             except ValueError:
                 raise util.Abort('svn: revision %s is not an integer' % rev)
         try:
             # Support file://path@rev syntax. Useful e.g. to convert
             # deleted branches.
             at = url.rfind('@')
             if at >= 0:
                 latest = int(url[at+1:])
                 url = url[:at]
         except ValueError, e:
             pass
         self.url = geturl(url)
         self.encoding = 'UTF-8' # Subversion is always nominal UTF-8
         try:
             self.transport = transport.SvnRaTransport(url=self.url)
             self.ra = self.transport.ra
             self.ctx = self.transport.client
             self.base = svn.ra.get_repos_root(self.ra)
             self.module = self.url[len(self.base):]
             self.modulemap = {} # revision, module
             self.commits = {}
-            self.files = {}
+            self.paths = {}
             self.uuid = svn.ra.get_uuid(self.ra).decode(self.encoding)
         except SubversionException, e:
             raise NoRepo("couldn't open SVN repo %s" % self.url)
 
         try:
             self.get_blacklist()
         except IOError, e:
             pass
 
         self.last_changed = self.latest(self.module, latest)
 
         self.head = self.revid(self.last_changed)
 
     def setrevmap(self, revmap):
         lastrevs = {}
         for revid in revmap.keys():
             uuid, module, revnum = self.revsplit(revid)
             lastrevnum = lastrevs.setdefault(module, revnum)
             if revnum > lastrevnum:
                 lastrevs[module] = revnum
         self.lastrevs = lastrevs
 
     def exists(self, path, optrev):
         try:
             return svn.client.ls(self.url.rstrip('/') + '/' + path,
                                  optrev, False, self.ctx)
         except SubversionException, err:
             return []
 
     def getheads(self):
         # detect standard /branches, /tags, /trunk layout
         rev = optrev(self.last_changed)
         rpath = self.url.strip('/')
         cfgtrunk = self.ui.config('convert', 'svn.trunk')
         cfgbranches = self.ui.config('convert', 'svn.branches')
         trunk = (cfgtrunk or 'trunk').strip('/')
         branches = (cfgbranches or 'branches').strip('/')
         if self.exists(trunk, rev) and self.exists(branches, rev):
             self.ui.note('found trunk at %r and branches at %r\n' %
                          (trunk, branches))
             oldmodule = self.module
             self.module += '/' + trunk
             lt = self.latest(self.module, self.last_changed)
             self.head = self.revid(lt)
             self.heads = [self.head]
             branchnames = svn.client.ls(rpath + '/' + branches, rev, False,
                                         self.ctx)
             for branch in branchnames.keys():
                 if oldmodule:
                     module = '/' + oldmodule + '/' + branches + '/' + branch
                 else:
                     module = '/' + branches + '/' + branch
                 brevnum = self.latest(module, self.last_changed)
                 brev = self.revid(brevnum, module)
                 self.ui.note('found branch %s at %d\n' % (branch, brevnum))
                 self.heads.append(brev)
         elif cfgtrunk or cfgbranches:
             raise util.Abort(_('trunk/branch layout expected, '
                                'but not found'))
         else:
             self.ui.note('working with one branch\n')
             self.heads = [self.head]
         return self.heads
 
     def getfile(self, file, rev):
         data, mode = self._getfile(file, rev)
         self.modecache[(file, rev)] = mode
         return data
 
     def getmode(self, file, rev):
         return self.modecache[(file, rev)]
 
     def getchanges(self, rev):
         self.modecache = {}
-        files = self.files[rev]
-        cl = files
-        cl.sort()
+        (paths, parents) = self.paths[rev]
+        files, copies = self.expandpaths(rev, paths, parents)
+        files.sort()
+        files = zip(files, [rev] * len(files))
+
         # caller caches the result, so free it here to release memory
-        del self.files[rev]
-        return cl
+        del self.paths[rev]
+        return (files, copies)
 
     def getcommit(self, rev):
         if rev not in self.commits:
             uuid, module, revnum = self.revsplit(rev)
             self.module = module
             self.reparent(module)
             stop = self.lastrevs.get(module, 0)
             self._fetch_revisions(from_revnum=revnum, to_revnum=stop)
         commit = self.commits[rev]
         # caller caches the result, so free it here to release memory
         del self.commits[rev]
         return commit
 
     def get_log(self, paths, start, end, limit=0, discover_changed_paths=True,
                 strict_node_history=False):
         '''wrapper for svn.ra.get_log.
         on a large repository, svn.ra.get_log pins huge amounts of
         memory that cannot be recovered.  work around it by forking
         and writing results over a pipe.'''
 
         def child(fp):
             protocol = -1
             def receiver(orig_paths, revnum, author, date, message, pool):
                 if orig_paths is not None:
                     for k, v in orig_paths.iteritems():
                         orig_paths[k] = changedpath(v)
                 pickle.dump((orig_paths, revnum, author, date, message),
                             fp, protocol)
 
             try:
                 # Use an ra of our own so that our parent can consume
                 # our results without confusing the server.
                 t = transport.SvnRaTransport(url=self.url)
                 svn.ra.get_log(t.ra, paths, start, end, limit,
                                discover_changed_paths,
                                strict_node_history,
                                receiver)
             except SubversionException, (_, num):
                 self.ui.print_exc()
                 pickle.dump(num, fp, protocol)
             else:
                 pickle.dump(None, fp, protocol)
             fp.close()
 
         def parent(fp):
             while True:
                 entry = pickle.load(fp)
                 try:
                     orig_paths, revnum, author, date, message = entry
                 except:
                     if entry is None:
                         break
                     raise SubversionException("child raised exception", entry)
                 yield entry
 
         rfd, wfd = os.pipe()
         pid = os.fork()
         if pid:
             os.close(wfd)
             for p in parent(os.fdopen(rfd, 'rb')):
                 yield p
             ret = os.waitpid(pid, 0)[1]
             if ret:
                 raise util.Abort(_('get_log %s') % util.explain_exit(ret))
         else:
             os.close(rfd)
             child(os.fdopen(wfd, 'wb'))
             os._exit(0)
 
     def gettags(self):
         tags = {}
         start = self.revnum(self.head)
         try:
             for entry in self.get_log(['/tags'], 0, start):
                 orig_paths, revnum, author, date, message = entry
                 for path in orig_paths:
                     if not path.startswith('/tags/'):
                         continue
                     ent = orig_paths[path]
                     source = ent.copyfrom_path
                     rev = ent.copyfrom_rev
                     tag = path.split('/', 2)[2]
                     tags[tag] = self.revid(rev, module=source)
         except SubversionException, (_, num):
             self.ui.note('no tags found at revision %d\n' % start)
         return tags
 
     # -- helper functions --
 
     def revid(self, revnum, module=None):
         if not module:
             module = self.module
         return (u"svn:%s%s@%s" % (self.uuid, module, revnum)).decode(self.encoding)
 
     def revnum(self, rev):
         return int(rev.split('@')[-1])
 
     def revsplit(self, rev):
         url, revnum = rev.encode(self.encoding).split('@', 1)
         revnum = int(revnum)
         parts = url.split('/', 1)
         uuid = parts.pop(0)[4:]
         mod = ''
         if parts:
             mod = '/' + parts[0]
         return uuid, mod, revnum
 
     def latest(self, path, stop=0):
         'find the latest revision affecting path, up to stop'
         if not stop:
             stop = svn.ra.get_latest_revnum(self.ra)
         try:
             self.reparent('')
             dirent = svn.ra.stat(self.ra, path.strip('/'), stop)
             self.reparent(self.module)
         except SubversionException:
             dirent = None
         if not dirent:
             print self.base, path
             raise util.Abort('%s not found up to revision %d' % (path, stop))
 
         return dirent.created_rev
 
     def get_blacklist(self):
         """Avoid certain revision numbers.
         It is not uncommon for two nearby revisions to cancel each other
         out, e.g. 'I copied trunk into a subdirectory of itself instead
         of making a branch'. The converted repository is significantly
         smaller if we ignore such revisions."""
         self.blacklist = set()
         blacklist = self.blacklist
         for line in file("blacklist.txt", "r"):
             if not line.startswith("#"):
                 try:
                     svn_rev = int(line.strip())
                     blacklist.add(svn_rev)
                 except ValueError, e:
                     pass # not an integer or a comment
 
     def is_blacklisted(self, svn_rev):
         return svn_rev in self.blacklist
 
     def reparent(self, module):
         svn_url = self.base + module
         self.ui.debug("reparent to %s\n" % svn_url.encode(self.encoding))
         svn.ra.reparent(self.ra, svn_url.encode(self.encoding))
 
     def expandpaths(self, rev, paths, parents):
         def get_entry_from_path(path, module=self.module):
             # Given the repository url of this wc, say
             #   "http://server/plone/CMFPlone/branches/Plone-2_0-branch"
             # extract the "entry" portion (a relative path) from what
             # svn log --xml says, ie
             #   "/CMFPlone/branches/Plone-2_0-branch/tests/PloneTestCase.py"
             # that is to say "tests/PloneTestCase.py"
             if path.startswith(module):
                 relative = path[len(module):]
                 if relative.startswith('/'):
                     return relative[1:]
                 else:
                     return relative
 
             # The path is outside our tracked tree...
             self.ui.debug('%r is not under %r, ignoring\n' % (path, module))
             return None
 
         entries = []
         copyfrom = {} # Map of entrypath, revision for finding source of deleted revisions.
         copies = {}
         revnum = self.revnum(rev)
 
+        if revnum in self.modulemap:
+            new_module = self.modulemap[revnum]
+            if new_module != self.module:
+                self.module = new_module
+                self.reparent(self.module)
+
         for path, ent in paths:
-            # self.ui.write("path %s\n" % path)
+            self.ui.write("path %s\n" % path)
             entrypath = get_entry_from_path(path, module=self.module)
             entry = entrypath.decode(self.encoding)
 
             kind = svn.ra.check_path(self.ra, entrypath, revnum)
             if kind == svn.core.svn_node_file:
                 if ent.copyfrom_path:
                     copyfrom_path = get_entry_from_path(ent.copyfrom_path)
                     if copyfrom_path:
                         self.ui.debug("Copied to %s from %s@%s\n" % (entry, copyfrom_path, ent.copyfrom_rev))
                         # It's probably important for hg that the source
                         # exists in the revision's parent, not just the
                         # ent.copyfrom_rev
                         fromkind = svn.ra.check_path(self.ra, copyfrom_path, ent.copyfrom_rev)
                         if fromkind != 0:
                             copies[self.recode(entry)] = self.recode(copyfrom_path)
                 entries.append(self.recode(entry))
             elif kind == 0: # gone, but had better be a deleted *file*
                 self.ui.debug("gone from %s\n" % ent.copyfrom_rev)
 
                 # if a branch is created but entries are removed in the same
                 # changeset, get the right fromrev
                 if parents:
                     uuid, old_module, fromrev = self.revsplit(parents[0])
                 else:
                     fromrev = revnum - 1
                     # might always need to be revnum - 1 in these 3 lines?
                     old_module = self.modulemap.get(fromrev, self.module)
 
                 basepath = old_module + "/" + get_entry_from_path(path, module=self.module)
                 entrypath = old_module + "/" + get_entry_from_path(path, module=self.module)
 
                 def lookup_parts(p):
                     rc = None
                     parts = p.split("/")
                     for i in range(len(parts)):
                         part = "/".join(parts[:i])
                         info = part, copyfrom.get(part, None)
                         if info[1] is not None:
                             self.ui.debug("Found parent directory %s\n" % info[1])
                             rc = info
                     return rc
 
                 self.ui.debug("base, entry %s %s\n" % (basepath, entrypath))
 
                 frompath, froment = lookup_parts(entrypath) or (None, revnum - 1)
 
                 # need to remove fragment from lookup_parts and replace with copyfrom_path
                 if frompath is not None:
                     self.ui.debug("munge-o-matic\n")
                     self.ui.debug(entrypath + '\n')
                     self.ui.debug(entrypath[len(frompath):] + '\n')
                     entrypath = froment.copyfrom_path + entrypath[len(frompath):]
                     fromrev = froment.copyfrom_rev
                     self.ui.debug("Info: %s %s %s %s\n" % (frompath, froment, ent, entrypath))
 
                 fromkind = svn.ra.check_path(self.ra, entrypath, fromrev)
                 if fromkind == svn.core.svn_node_file:   # a deleted file
                     entries.append(self.recode(entry))
                 elif fromkind == svn.core.svn_node_dir:
                     # print "Deleted/moved non-file:", revnum, path, ent
                     # children = self._find_children(path, revnum - 1)
                     # print "find children %s@%d from %d action %s" % (path, revnum, ent.copyfrom_rev, ent.action)
                     # Sometimes this is tricky. For example: in
                     # The Subversion Repository revision 6940 a dir
                     # was copied and one of its files was deleted
                     # from the new location in the same commit. This
                     # code can't deal with that yet.
                     if ent.action == 'C':
                         children = self._find_children(path, fromrev)
                     else:
                         oroot = entrypath.strip('/')
                         nroot = path.strip('/')
                         children = self._find_children(oroot, fromrev)
                         children = [s.replace(oroot,nroot) for s in children]
                     # Mark all [files, not directories] as deleted.
                     for child in children:
                         # Can we move a child directory and its
                         # parent in the same commit? (probably can). Could
                         # cause problems if instead of revnum -1,
                         # we have to look in (copyfrom_path, revnum - 1)
                         entrypath = get_entry_from_path("/" + child, module=old_module)
                         if entrypath:
                             entry = self.recode(entrypath.decode(self.encoding))
                             if entry in copies:
                                 # deleted file within a copy
                                 del copies[entry]
                             else:
                                 entries.append(entry)
                 else:
                     self.ui.debug('unknown path in revision %d: %s\n' % \
                                   (revnum, path))
             elif kind == svn.core.svn_node_dir:
                 # Should probably synthesize normal file entries
                 # and handle as above to clean up copy/rename handling.
 
                 # If the directory just had a prop change,
                 # then we shouldn't need to look for its children.
                 # Also this could create duplicate entries. Not sure
                 # whether this will matter. Maybe should make entries a set.
                 # print "Changed directory", revnum, path, ent.action, ent.copyfrom_path, ent.copyfrom_rev
                 # This will fail if a directory was copied
                 # from another branch and then some of its files
                 # were deleted in the same transaction.
                 children = self._find_children(path, revnum)
                 children.sort()
                 for child in children:
                     # Can we move a child directory and its
                     # parent in the same commit? (probably can). Could
                     # cause problems if instead of revnum -1,
                     # we have to look in (copyfrom_path, revnum - 1)
                     entrypath = get_entry_from_path("/" + child, module=self.module)
                     # print child, self.module, entrypath
                     if entrypath:
                         # Need to filter out directories here...
                         kind = svn.ra.check_path(self.ra, entrypath, revnum)
                         if kind != svn.core.svn_node_dir:
                             entries.append(self.recode(entrypath))
 
                 # Copies here (must copy all from source)
                 # Probably not a real problem for us if
                 # source does not exist
 
                 # Can do this with the copy command "hg copy"
                 # if ent.copyfrom_path:
                 #     copyfrom_entry = get_entry_from_path(ent.copyfrom_path.decode(self.encoding),
                 #             module=self.module)
                 #     copyto_entry = entrypath
                 #
                 #     print "copy directory", copyfrom_entry, 'to', copyto_entry
                 #
                 #     copies.append((copyfrom_entry, copyto_entry))
 
                 if ent.copyfrom_path:
                     copyfrom_path = ent.copyfrom_path.decode(self.encoding)
                     copyfrom_entry = get_entry_from_path(copyfrom_path, module=self.module)
                     if copyfrom_entry:
                         copyfrom[path] = ent
                         self.ui.debug("mark %s came from %s\n" % (path, copyfrom[path]))
 
                         # Good, /probably/ a regular copy. Really should check
                         # to see whether the parent revision actually contains
                         # the directory in question.
                         children = self._find_children(self.recode(copyfrom_path), ent.copyfrom_rev)
                         children.sort()
                         for child in children:
                             entrypath = get_entry_from_path("/" + child, module=self.module)
                             if entrypath:
                                 entry = entrypath.decode(self.encoding)
                                 # print "COPY COPY From", copyfrom_entry, entry
                                 copyto_path = path + entry[len(copyfrom_entry):]
                                 copyto_entry =  get_entry_from_path(copyto_path, module=self.module)
                                 # print "COPY", entry, "COPY To", copyto_entry
                                 copies[self.recode(copyto_entry)] = self.recode(entry)
                                 # copy from quux splort/quuxfile
 
         return (entries, copies)
 
     def _fetch_revisions(self, from_revnum = 0, to_revnum = 347):
         self.child_cset = None
         def parselogentry(orig_paths, revnum, author, date, message):
             self.ui.debug("parsing revision %d (%d changes)\n" %
                           (revnum, len(orig_paths)))
 
             if revnum in self.modulemap:
                 new_module = self.modulemap[revnum]
                 if new_module != self.module:
                     self.module = new_module
                     self.reparent(self.module)
 
             rev = self.revid(revnum)
             # branch log might return entries for a parent we already have
             if (rev in self.commits or
                 (revnum < self.lastrevs.get(self.module, 0))):
                 return
 
             parents = []
             orig_paths = orig_paths.items()
             orig_paths.sort()
             
             # check whether this revision is the start of a branch
             path, ent = orig_paths and orig_paths[0] or (None, None)
             if ent and path == self.module:
                 if ent.copyfrom_path:
                     # ent.copyfrom_rev may not be the actual last revision
                     prev = self.latest(ent.copyfrom_path, ent.copyfrom_rev)
                     self.modulemap[prev] = ent.copyfrom_path
                     parents = [self.revid(prev, ent.copyfrom_path)]
                     self.ui.note('found parent of branch %s at %d: %s\n' % \
                                      (self.module, prev, ent.copyfrom_path))
                 else:
                     self.ui.debug("No copyfrom path, don't know what to do.\n")
 
             self.modulemap[revnum] = self.module # track backwards in time
 
             paths = []
             # filter out unrelated paths
             for path, ent in orig_paths:
                 if not path.startswith(self.module):
                     self.ui.debug("boring@%s: %s\n" % (revnum, path))
                     continue
                 paths.append((path, ent))
 
-            entries, copies = self.expandpaths(rev, paths, parents)
-            # a list of (filename, id) where id lets us retrieve the file.
-            # eg in git, id is the object hash. for svn it'll be the
-            self.files[rev] = zip(entries, [rev] * len(entries))
-            if not entries:
-                return
+            self.paths[rev] = (paths, parents)
 
             # Example SVN datetime. Includes microseconds.
             # ISO-8601 conformant
             # '2007-01-04T17:35:00.902377Z'
             date = util.parsedate(date[:18] + " UTC", ["%Y-%m-%dT%H:%M:%S"])
 
             log = message and self.recode(message)
             author = author and self.recode(author) or ''
             try:
                 branch = self.module.split("/")[-1]
                 if branch == 'trunk':
                     branch = ''
             except IndexError:
                 branch = None
 
             cset = commit(author=author,
                           date=util.datestr(date),
                           desc=log,
                           parents=parents,
-                          copies=copies,
                           branch=branch,
                           rev=rev.encode('utf-8'))
 
             self.commits[rev] = cset
             if self.child_cset and not self.child_cset.parents:
                 self.child_cset.parents = [rev]
             self.child_cset = cset
 
         self.ui.note('fetching revision log for "%s" from %d to %d\n' %
                      (self.module, from_revnum, to_revnum))
 
         try:
             for entry in self.get_log([self.module], from_revnum, to_revnum):
                 orig_paths, revnum, author, date, message = entry
                 if self.is_blacklisted(revnum):
                     self.ui.note('skipping blacklisted revision %d\n' % revnum)
                     continue
                 if orig_paths is None:
                     self.ui.debug('revision %d has no entries\n' % revnum)
                     continue
                 parselogentry(orig_paths, revnum, author, date, message)
         except SubversionException, (_, num):
             if num == svn.core.SVN_ERR_FS_NO_SUCH_REVISION:
                 raise NoSuchRevision(branch=self,
                     revision="Revision number %d" % to_revnum)
             raise
 
     def _getfile(self, file, rev):
         io = StringIO()
         # TODO: ra.get_file transmits the whole file instead of diffs.
         mode = ''
         try:
             revnum = self.revnum(rev)
             if self.module != self.modulemap[revnum]:
                 self.module = self.modulemap[revnum]
                 self.reparent(self.module)
             info = svn.ra.get_file(self.ra, file, revnum, io)
             if isinstance(info, list):
                 info = info[-1]
             mode = ("svn:executable" in info) and 'x' or ''
             mode = ("svn:special" in info) and 'l' or mode
         except SubversionException, e:
             notfound = (svn.core.SVN_ERR_FS_NOT_FOUND,
                 svn.core.SVN_ERR_RA_DAV_PATH_NOT_FOUND)
             if e.apr_err in notfound: # File not found
                 raise IOError()
             raise
         data = io.getvalue()
         if mode == 'l':
             link_prefix = "link "
             if data.startswith(link_prefix):
                 data = data[len(link_prefix):]
         return data, mode
 
     def _find_children(self, path, revnum):
         path = path.strip('/')
         pool = Pool()
         rpath = '/'.join([self.base, path]).strip('/')
         return ['%s/%s' % (path, x) for x in svn.client.ls(rpath, optrev(revnum), True, self.ctx, pool).keys()]