Merge bundle -r work from Eric Hopper

tests/test-clone-r

@@ -0,0 +1,59 b''
+#!/bin/bash
+
+hg init test
+cd test
+cat >>afile <<EOF
+0
+EOF
+hg add afile
+hg commit -m "0.0"
+cat >>afile <<EOF
+1
+EOF
+hg commit -m "0.1"
+cat >>afile <<EOF
+2
+EOF
+hg commit -m "0.2"
+cat >>afile <<EOF
+3
+EOF
+hg commit -m "0.3"
+hg update -C 0
+cat >>afile <<EOF
+1
+EOF
+hg commit -m "1.1"
+cat >>afile <<EOF
+2
+EOF
+hg commit -m "1.2"
+cat >fred <<EOF
+a line
+EOF
+cat >>afile <<EOF
+3
+EOF
+hg add fred
+hg commit -m "1.3"
+hg mv afile adifferentfile
+hg commit -m "1.3m"
+hg update -C 3
+hg mv afile anotherfile
+hg commit -m "0.3m"
+hg debugindex .hg/data/afile.i
+hg debugindex .hg/data/adifferentfile.i
+hg debugindex .hg/data/anotherfile.i
+hg debugindex .hg/data/fred.i
+hg debugindex .hg/00manifest.i
+hg verify
+cd ..
+for i in 0 1 2 3 4 5 6 7 8; do
+   hg clone -r "$i" test test-"$i"
+   cd test-"$i"
+   hg verify
+   cd ..
+done
+cd test-8
+hg pull ../test-7
+hg verify

tests/test-clone-r.out

@@ -0,0 +1,126 b''
+   rev    offset  length   base linkrev nodeid       p1           p2
+     0         0       3      0       0 362fef284ce2 000000000000 000000000000
+     1         3       5      1       1 125144f7e028 362fef284ce2 000000000000
+     2         8       7      2       2 4c982badb186 125144f7e028 000000000000
+     3        15       9      3       3 19b1fc555737 4c982badb186 000000000000
+   rev    offset  length   base linkrev nodeid       p1           p2
+     0         0      75      0       7 905359268f77 000000000000 000000000000
+   rev    offset  length   base linkrev nodeid       p1           p2
+     0         0      75      0       8 905359268f77 000000000000 000000000000
+   rev    offset  length   base linkrev nodeid       p1           p2
+     0         0       8      0       6 12ab3bcc5ea4 000000000000 000000000000
+   rev    offset  length   base linkrev nodeid       p1           p2
+     0         0      48      0       0 43eadb1d2d06 000000000000 000000000000
+     1        48      48      1       1 8b89697eba2c 43eadb1d2d06 000000000000
+     2        96      48      2       2 626a32663c2f 8b89697eba2c 000000000000
+     3       144      48      3       3 f54c32f13478 626a32663c2f 000000000000
+     4       192      58      3       6 de68e904d169 626a32663c2f 000000000000
+     5       250      68      3       7 3b45cc2ab868 de68e904d169 000000000000
+     6       318      54      6       8 24d86153a002 f54c32f13478 000000000000
+checking changesets
+checking manifests
+crosschecking files in changesets and manifests
+checking files
+4 files, 9 changesets, 7 total revisions
+requesting all changes
+adding changesets
+adding manifests
+adding file changes
+added 1 changesets with 1 changes to 1 files
+checking changesets
+checking manifests
+crosschecking files in changesets and manifests
+checking files
+1 files, 1 changesets, 1 total revisions
+requesting all changes
+adding changesets
+adding manifests
+adding file changes
+added 2 changesets with 2 changes to 1 files
+checking changesets
+checking manifests
+crosschecking files in changesets and manifests
+checking files
+1 files, 2 changesets, 2 total revisions
+requesting all changes
+adding changesets
+adding manifests
+adding file changes
+added 3 changesets with 3 changes to 1 files
+checking changesets
+checking manifests
+crosschecking files in changesets and manifests
+checking files
+1 files, 3 changesets, 3 total revisions
+requesting all changes
+adding changesets
+adding manifests
+adding file changes
+added 4 changesets with 4 changes to 1 files
+checking changesets
+checking manifests
+crosschecking files in changesets and manifests
+checking files
+1 files, 4 changesets, 4 total revisions
+requesting all changes
+adding changesets
+adding manifests
+adding file changes
+added 2 changesets with 2 changes to 1 files
+checking changesets
+checking manifests
+crosschecking files in changesets and manifests
+checking files
+1 files, 2 changesets, 2 total revisions
+requesting all changes
+adding changesets
+adding manifests
+adding file changes
+added 3 changesets with 3 changes to 1 files
+checking changesets
+checking manifests
+crosschecking files in changesets and manifests
+checking files
+1 files, 3 changesets, 3 total revisions
+requesting all changes
+adding changesets
+adding manifests
+adding file changes
+added 4 changesets with 5 changes to 2 files
+checking changesets
+checking manifests
+crosschecking files in changesets and manifests
+checking files
+2 files, 4 changesets, 5 total revisions
+requesting all changes
+adding changesets
+adding manifests
+adding file changes
+added 5 changesets with 6 changes to 3 files
+checking changesets
+checking manifests
+crosschecking files in changesets and manifests
+checking files
+3 files, 5 changesets, 6 total revisions
+requesting all changes
+adding changesets
+adding manifests
+adding file changes
+added 5 changesets with 5 changes to 2 files
+checking changesets
+checking manifests
+crosschecking files in changesets and manifests
+checking files
+2 files, 5 changesets, 5 total revisions
+pulling from ../test-7
+searching for changes
+adding changesets
+adding manifests
+adding file changes
+added 4 changesets with 2 changes to 3 files (+1 heads)
+(run 'hg update' to get a working copy)
+checking changesets
+checking manifests
+crosschecking files in changesets and manifests
+checking files
+4 files, 9 changesets, 7 total revisions

mercurial/commands.py

@@ -1,2653 +1,2666 b''
 # commands.py - command processing for mercurial
 #
 # Copyright 2005 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 demandload import demandload
 from node import *
 from i18n import gettext as _
 demandload(globals(), "os re sys signal shutil imp urllib pdb")
 demandload(globals(), "fancyopts ui hg util lock revlog")
 demandload(globals(), "fnmatch hgweb mdiff random signal time traceback")
 demandload(globals(), "errno socket version struct atexit sets bz2")
 
 class UnknownCommand(Exception):
     """Exception raised if command is not in the command table."""
 
 def filterfiles(filters, files):
     l = [x for x in files if x in filters]
 
     for t in filters:
         if t and t[-1] != "/":
             t += "/"
         l += [x for x in files if x.startswith(t)]
     return l
 
 def relpath(repo, args):
     cwd = repo.getcwd()
     if cwd:
         return [util.normpath(os.path.join(cwd, x)) for x in args]
     return args
 
 def matchpats(repo, cwd, pats=[], opts={}, head=''):
     return util.cmdmatcher(repo.root, cwd, pats or ['.'], opts.get('include'),
                         opts.get('exclude'), head)
 
 def makewalk(repo, pats, opts, head=''):
     cwd = repo.getcwd()
     files, matchfn, anypats = matchpats(repo, cwd, pats, opts, head)
     exact = dict(zip(files, files))
     def walk():
         for src, fn in repo.walk(files=files, match=matchfn):
             yield src, fn, util.pathto(cwd, fn), fn in exact
     return files, matchfn, walk()
 
 def walk(repo, pats, opts, head=''):
     files, matchfn, results = makewalk(repo, pats, opts, head)
     for r in results:
         yield r
 
 def walkchangerevs(ui, repo, cwd, pats, opts):
     '''Iterate over files and the revs they changed in.
 
     Callers most commonly need to iterate backwards over the history
     it is interested in.  Doing so has awful (quadratic-looking)
     performance, so we use iterators in a "windowed" way.
 
     We walk a window of revisions in the desired order.  Within the
     window, we first walk forwards to gather data, then in the desired
     order (usually backwards) to display it.
 
     This function returns an (iterator, getchange) pair.  The
     getchange function returns the changelog entry for a numeric
     revision.  The iterator yields 3-tuples.  They will be of one of
     the following forms:
 
     "window", incrementing, lastrev: stepping through a window,
     positive if walking forwards through revs, last rev in the
     sequence iterated over - use to reset state for the current window
 
     "add", rev, fns: out-of-order traversal of the given file names
     fns, which changed during revision rev - use to gather data for
     possible display
 
     "iter", rev, None: in-order traversal of the revs earlier iterated
     over with "add" - use to display data'''
 
     if repo.changelog.count() == 0:
         return [], False
 
     cwd = repo.getcwd()
     if not pats and cwd:
         opts['include'] = [os.path.join(cwd, i) for i in opts['include']]
         opts['exclude'] = [os.path.join(cwd, x) for x in opts['exclude']]
     files, matchfn, anypats = matchpats(repo, (pats and cwd) or '',
                                         pats, opts)
     revs = map(int, revrange(ui, repo, opts['rev'] or ['tip:0']))
     wanted = {}
     slowpath = anypats
     window = 300
     fncache = {}
 
     chcache = {}
     def getchange(rev):
         ch = chcache.get(rev)
         if ch is None:
             chcache[rev] = ch = repo.changelog.read(repo.lookup(str(rev)))
         return ch
 
     if not slowpath and not files:
         # No files, no patterns.  Display all revs.
         wanted = dict(zip(revs, revs))
     if not slowpath:
         # Only files, no patterns.  Check the history of each file.
         def filerevgen(filelog):
             for i in xrange(filelog.count() - 1, -1, -window):
                 revs = []
                 for j in xrange(max(0, i - window), i + 1):
                     revs.append(filelog.linkrev(filelog.node(j)))
                 revs.reverse()
                 for rev in revs:
                     yield rev
 
         minrev, maxrev = min(revs), max(revs)
         for file in files:
             filelog = repo.file(file)
             # A zero count may be a directory or deleted file, so
             # try to find matching entries on the slow path.
             if filelog.count() == 0:
                 slowpath = True
                 break
             for rev in filerevgen(filelog):
                 if rev <= maxrev:
                     if rev < minrev:
                         break
                     fncache.setdefault(rev, [])
                     fncache[rev].append(file)
                     wanted[rev] = 1
     if slowpath:
         # The slow path checks files modified in every changeset.
         def changerevgen():
             for i in xrange(repo.changelog.count() - 1, -1, -window):
                 for j in xrange(max(0, i - window), i + 1):
                     yield j, getchange(j)[3]
 
         for rev, changefiles in changerevgen():
             matches = filter(matchfn, changefiles)
             if matches:
                 fncache[rev] = matches
                 wanted[rev] = 1
 
     def iterate():
         for i in xrange(0, len(revs), window):
             yield 'window', revs[0] < revs[-1], revs[-1]
             nrevs = [rev for rev in revs[i:min(i+window, len(revs))]
                      if rev in wanted]
             srevs = list(nrevs)
             srevs.sort()
             for rev in srevs:
                 fns = fncache.get(rev) or filter(matchfn, getchange(rev)[3])
                 yield 'add', rev, fns
             for rev in nrevs:
                 yield 'iter', rev, None
     return iterate(), getchange
 
 revrangesep = ':'
 
 def revrange(ui, repo, revs, revlog=None):
     """Yield revision as strings from a list of revision specifications."""
     if revlog is None:
         revlog = repo.changelog
     revcount = revlog.count()
     def fix(val, defval):
         if not val:
             return defval
         try:
             num = int(val)
             if str(num) != val:
                 raise ValueError
             if num < 0: num += revcount
             if num < 0: num = 0
             elif num >= revcount:
                 raise ValueError
         except ValueError:
             try:
                 num = repo.changelog.rev(repo.lookup(val))
             except KeyError:
                 try:
                     num = revlog.rev(revlog.lookup(val))
                 except KeyError:
                     raise util.Abort(_('invalid revision identifier %s'), val)
         return num
     seen = {}
     for spec in revs:
         if spec.find(revrangesep) >= 0:
             start, end = spec.split(revrangesep, 1)
             start = fix(start, 0)
             end = fix(end, revcount - 1)
             step = start > end and -1 or 1
             for rev in xrange(start, end+step, step):
                 if rev in seen: continue
                 seen[rev] = 1
                 yield str(rev)
         else:
             rev = fix(spec, None)
             if rev in seen: continue
             seen[rev] = 1
             yield str(rev)
 
 def make_filename(repo, r, pat, node=None,
                   total=None, seqno=None, revwidth=None, pathname=None):
     node_expander = {
         'H': lambda: hex(node),
         'R': lambda: str(r.rev(node)),
         'h': lambda: short(node),
         }
     expander = {
         '%': lambda: '%',
         'b': lambda: os.path.basename(repo.root),
         }
 
     try:
         if node:
             expander.update(node_expander)
         if node and revwidth is not None:
             expander['r'] = lambda: str(r.rev(node)).zfill(revwidth)
         if total is not None:
             expander['N'] = lambda: str(total)
         if seqno is not None:
             expander['n'] = lambda: str(seqno)
         if total is not None and seqno is not None:
             expander['n'] = lambda:str(seqno).zfill(len(str(total)))
         if pathname is not None:
             expander['s'] = lambda: os.path.basename(pathname)
             expander['d'] = lambda: os.path.dirname(pathname) or '.'
             expander['p'] = lambda: pathname
 
         newname = []
         patlen = len(pat)
         i = 0
         while i < patlen:
             c = pat[i]
             if c == '%':
                 i += 1
                 c = pat[i]
                 c = expander[c]()
             newname.append(c)
             i += 1
         return ''.join(newname)
     except KeyError, inst:
         raise util.Abort(_("invalid format spec '%%%s' in output file name"),
                     inst.args[0])
 
 def make_file(repo, r, pat, node=None,
               total=None, seqno=None, revwidth=None, mode='wb', pathname=None):
     if not pat or pat == '-':
         return 'w' in mode and sys.stdout or sys.stdin
     if hasattr(pat, 'write') and 'w' in mode:
         return pat
     if hasattr(pat, 'read') and 'r' in mode:
         return pat
     return open(make_filename(repo, r, pat, node, total, seqno, revwidth,
                               pathname),
                 mode)
 
 def dodiff(fp, ui, repo, node1, node2, files=None, match=util.always,
            changes=None, text=False):
     if not changes:
         (c, a, d, u) = repo.changes(node1, node2, files, match=match)
     else:
         (c, a, d, u) = changes
     if files:
         c, a, d = map(lambda x: filterfiles(files, x), (c, a, d))
 
     if not c and not a and not d:
         return
 
     if node2:
         change = repo.changelog.read(node2)
         mmap2 = repo.manifest.read(change[0])
         date2 = util.datestr(change[2])
         def read(f):
             return repo.file(f).read(mmap2[f])
     else:
         date2 = util.datestr()
         if not node1:
             node1 = repo.dirstate.parents()[0]
         def read(f):
             return repo.wfile(f).read()
 
     if ui.quiet:
         r = None
     else:
         hexfunc = ui.verbose and hex or short
         r = [hexfunc(node) for node in [node1, node2] if node]
 
     change = repo.changelog.read(node1)
     mmap = repo.manifest.read(change[0])
     date1 = util.datestr(change[2])
 
     for f in c:
         to = None
         if f in mmap:
             to = repo.file(f).read(mmap[f])
         tn = read(f)
         fp.write(mdiff.unidiff(to, date1, tn, date2, f, r, text=text))
     for f in a:
         to = None
         tn = read(f)
         fp.write(mdiff.unidiff(to, date1, tn, date2, f, r, text=text))
     for f in d:
         to = repo.file(f).read(mmap[f])
         tn = None
         fp.write(mdiff.unidiff(to, date1, tn, date2, f, r, text=text))
 
 def trimuser(ui, name, rev, revcache):
     """trim the name of the user who committed a change"""
     user = revcache.get(rev)
     if user is None:
         user = revcache[rev] = ui.shortuser(name)
     return user
 
 def show_changeset(ui, repo, rev=0, changenode=None, brinfo=None):
     """show a single changeset or file revision"""
     log = repo.changelog
     if changenode is None:
         changenode = log.node(rev)
     elif not rev:
         rev = log.rev(changenode)
 
     if ui.quiet:
         ui.write("%d:%s\n" % (rev, short(changenode)))
         return
 
     changes = log.read(changenode)
     date = util.datestr(changes[2])
 
     parents = [(log.rev(p), ui.verbose and hex(p) or short(p))
                for p in log.parents(changenode)
                if ui.debugflag or p != nullid]
     if not ui.debugflag and len(parents) == 1 and parents[0][0] == rev-1:
         parents = []
 
     if ui.verbose:
         ui.write(_("changeset:   %d:%s\n") % (rev, hex(changenode)))
     else:
         ui.write(_("changeset:   %d:%s\n") % (rev, short(changenode)))
 
     for tag in repo.nodetags(changenode):
         ui.status(_("tag:         %s\n") % tag)
     for parent in parents:
         ui.write(_("parent:      %d:%s\n") % parent)
 
     if brinfo and changenode in brinfo:
         br = brinfo[changenode]
         ui.write(_("branch:      %s\n") % " ".join(br))
 
     ui.debug(_("manifest:    %d:%s\n") % (repo.manifest.rev(changes[0]),
                                       hex(changes[0])))
     ui.status(_("user:        %s\n") % changes[1])
     ui.status(_("date:        %s\n") % date)
 
     if ui.debugflag:
         files = repo.changes(log.parents(changenode)[0], changenode)
         for key, value in zip([_("files:"), _("files+:"), _("files-:")], files):
             if value:
                 ui.note("%-12s %s\n" % (key, " ".join(value)))
     else:
         ui.note(_("files:       %s\n") % " ".join(changes[3]))
 
     description = changes[4].strip()
     if description:
         if ui.verbose:
             ui.status(_("description:\n"))
             ui.status(description)
             ui.status("\n\n")
         else:
             ui.status(_("summary:     %s\n") % description.splitlines()[0])
     ui.status("\n")
 
 def show_version(ui):
     """output version and copyright information"""
     ui.write(_("Mercurial Distributed SCM (version %s)\n")
              % version.get_version())
     ui.status(_(
         "\nCopyright (C) 2005 Matt Mackall <[e-mail]>\n"
         "This is free software; see the source for copying conditions. "
         "There is NO\nwarranty; "
         "not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.\n"
     ))
 
 def help_(ui, cmd=None, with_version=False):
     """show help for a given command or all commands"""
     option_lists = []
     if cmd and cmd != 'shortlist':
         if with_version:
             show_version(ui)
             ui.write('\n')
         key, i = find(cmd)
         # synopsis
         ui.write("%s\n\n" % i[2])
 
         # description
         doc = i[0].__doc__
         if ui.quiet:
             doc = doc.splitlines(0)[0]
         ui.write("%s\n" % doc.rstrip())
 
         if not ui.quiet:
             # aliases
             aliases = ', '.join(key.split('|')[1:])
             if aliases:
                 ui.write(_("\naliases: %s\n") % aliases)
 
             # options
             if i[1]:
                 option_lists.append(("options", i[1]))
 
     else:
         # program name
         if ui.verbose or with_version:
             show_version(ui)
         else:
             ui.status(_("Mercurial Distributed SCM\n"))
         ui.status('\n')
 
         # list of commands
         if cmd == "shortlist":
             ui.status(_('basic commands (use "hg help" '
                         'for the full list or option "-v" for details):\n\n'))
         elif ui.verbose:
             ui.status(_('list of commands:\n\n'))
         else:
             ui.status(_('list of commands (use "hg help -v" '
                         'to show aliases and global options):\n\n'))
 
         h = {}
         cmds = {}
         for c, e in table.items():
             f = c.split("|")[0]
             if cmd == "shortlist" and not f.startswith("^"):
                 continue
             f = f.lstrip("^")
             if not ui.debugflag and f.startswith("debug"):
                 continue
             d = ""
             if e[0].__doc__:
                 d = e[0].__doc__.splitlines(0)[0].rstrip()
             h[f] = d
             cmds[f]=c.lstrip("^")
 
         fns = h.keys()
         fns.sort()
         m = max(map(len, fns))
         for f in fns:
             if ui.verbose:
                 commands = cmds[f].replace("|",", ")
                 ui.write(" %s:\n      %s\n"%(commands,h[f]))
             else:
                 ui.write(' %-*s   %s\n' % (m, f, h[f]))
 
     # global options
     if ui.verbose:
         option_lists.append(("global options", globalopts))
 
     # list all option lists
     opt_output = []
     for title, options in option_lists:
         opt_output.append(("\n%s:\n" % title, None))
         for shortopt, longopt, default, desc in options:
             opt_output.append(("%2s%s" % (shortopt and "-%s" % shortopt,
                                           longopt and " --%s" % longopt),
                                "%s%s" % (desc,
                                          default and _(" (default: %s)") % default
                                          or "")))
 
     if opt_output:
         opts_len = max([len(line[0]) for line in opt_output if line[1]])
         for first, second in opt_output:
             if second:
                 ui.write(" %-*s  %s\n" % (opts_len, first, second))
             else:
                 ui.write("%s\n" % first)
 
 # Commands start here, listed alphabetically
 
 def add(ui, repo, *pats, **opts):
     """add the specified files on the next commit
 
     Schedule files to be version controlled and added to the repository.
 
     The files will be added to the repository at the next commit.
 
     If no names are given, add all files in the current directory and
     its subdirectories.
     """
 
     names = []
     for src, abs, rel, exact in walk(repo, pats, opts):
         if exact:
             if ui.verbose: ui.status(_('adding %s\n') % rel)
             names.append(abs)
         elif repo.dirstate.state(abs) == '?':
             ui.status(_('adding %s\n') % rel)
             names.append(abs)
     repo.add(names)
 
 def addremove(ui, repo, *pats, **opts):
     """add all new files, delete all missing files
 
     Add all new files and remove all missing files from the repository.
 
     New files are ignored if they match any of the patterns in .hgignore. As
     with add, these changes take effect at the next commit.
     """
     add, remove = [], []
     for src, abs, rel, exact in walk(repo, pats, opts):
         if src == 'f' and repo.dirstate.state(abs) == '?':
             add.append(abs)
             if ui.verbose or not exact:
                 ui.status(_('adding %s\n') % rel)
         if repo.dirstate.state(abs) != 'r' and not os.path.exists(rel):
             remove.append(abs)
             if ui.verbose or not exact:
                 ui.status(_('removing %s\n') % rel)
     repo.add(add)
     repo.remove(remove)
 
 def annotate(ui, repo, *pats, **opts):
     """show changeset information per file line
 
     List changes in files, showing the revision id responsible for each line
 
     This command is useful to discover who did a change or when a change took
     place.
 
     Without the -a option, annotate will avoid processing files it
     detects as binary. With -a, annotate will generate an annotation
     anyway, probably with undesirable results.
     """
     def getnode(rev):
         return short(repo.changelog.node(rev))
 
     ucache = {}
     def getname(rev):
         cl = repo.changelog.read(repo.changelog.node(rev))
         return trimuser(ui, cl[1], rev, ucache)
 
     if not pats:
         raise util.Abort(_('at least one file name or pattern required'))
 
     opmap = [['user', getname], ['number', str], ['changeset', getnode]]
     if not opts['user'] and not opts['changeset']:
         opts['number'] = 1
 
     if opts['rev']:
         node = repo.changelog.lookup(opts['rev'])
     else:
         node = repo.dirstate.parents()[0]
     change = repo.changelog.read(node)
     mmap = repo.manifest.read(change[0])
 
     for src, abs, rel, exact in walk(repo, pats, opts):
         if abs not in mmap:
             ui.warn(_("warning: %s is not in the repository!\n") % rel)
             continue
 
         f = repo.file(abs)
         if not opts['text'] and util.binary(f.read(mmap[abs])):
             ui.write(_("%s: binary file\n") % rel)
             continue
 
         lines = f.annotate(mmap[abs])
         pieces = []
 
         for o, f in opmap:
             if opts[o]:
                 l = [f(n) for n, dummy in lines]
                 if l:
                     m = max(map(len, l))
                     pieces.append(["%*s" % (m, x) for x in l])
 
         if pieces:
             for p, l in zip(zip(*pieces), lines):
                 ui.write("%s: %s" % (" ".join(p), l[1]))
 
 def bundle(ui, repo, fname, dest="default-push", **opts):
     """create a changegroup file
 
     Generate a compressed changegroup file collecting all changesets
     not found in the other repository.
 
     This file can then be transferred using conventional means and
     applied to another repository with the unbundle command. This is
     useful when native push and pull are not available or when
     exporting an entire repository is undesirable. The standard file
     extension is ".hg".
 
     Unlike import/export, this exactly preserves all changeset
     contents including permissions, rename data, and revision history.
     """
     f = open(fname, "wb")
     dest = ui.expandpath(dest, repo.root)
     other = hg.repository(ui, dest)
     o = repo.findoutgoing(other)
     cg = repo.changegroup(o)
 
     try:
         f.write("HG10")
         z = bz2.BZ2Compressor(9)
         while 1:
             chunk = cg.read(4096)
             if not chunk:
                 break
             f.write(z.compress(chunk))
         f.write(z.flush())
     except:
         os.unlink(fname)
         raise
 
 def cat(ui, repo, file1, *pats, **opts):
     """output the latest or given revisions of files
 
     Print the specified files as they were at the given revision.
     If no revision is given then the tip is used.
 
     Output may be to a file, in which case the name of the file is
     given using a format string.  The formatting rules are the same as
     for the export command, with the following additions:
 
     %s   basename of file being printed
     %d   dirname of file being printed, or '.' if in repo root
     %p   root-relative path name of file being printed
     """
     mf = {}
     if opts['rev']:
         change = repo.changelog.read(repo.lookup(opts['rev']))
         mf = repo.manifest.read(change[0])
     for src, abs, rel, exact in walk(repo, (file1,) + pats, opts):
         r = repo.file(abs)
         if opts['rev']:
             try:
                 n = mf[abs]
             except (hg.RepoError, KeyError):
                 try:
                     n = r.lookup(rev)
                 except KeyError, inst:
                     raise util.Abort(_('cannot find file %s in rev %s'), rel, rev)
         else:
             n = r.tip()
         fp = make_file(repo, r, opts['output'], node=n, pathname=abs)
         fp.write(r.read(n))
 
 def clone(ui, source, dest=None, **opts):
     """make a copy of an existing repository
 
     Create a copy of an existing repository in a new directory.
 
     If no destination directory name is specified, it defaults to the
     basename of the source.
 
     The location of the source is added to the new repository's
     .hg/hgrc file, as the default to be used for future pulls.
 
     For efficiency, hardlinks are used for cloning whenever the source
     and destination are on the same filesystem.  Some filesystems,
     such as AFS, implement hardlinking incorrectly, but do not report
     errors.  In these cases, use the --pull option to avoid
     hardlinking.
     """
     if dest is None:
         dest = os.path.basename(os.path.normpath(source))
 
     if os.path.exists(dest):
         raise util.Abort(_("destination '%s' already exists"), dest)
 
     dest = os.path.realpath(dest)
 
     class Dircleanup:
         def __init__(self, dir_):
             self.rmtree = shutil.rmtree
             self.dir_ = dir_
             os.mkdir(dir_)
         def close(self):
             self.dir_ = None
         def __del__(self):
             if self.dir_:
                 self.rmtree(self.dir_, True)
 
     if opts['ssh']:
         ui.setconfig("ui", "ssh", opts['ssh'])
     if opts['remotecmd']:
         ui.setconfig("ui", "remotecmd", opts['remotecmd'])
 
     if not os.path.exists(source):
         source = ui.expandpath(source)
 
     d = Dircleanup(dest)
     abspath = source
     other = hg.repository(ui, source)
 
     copy = False
     if other.dev() != -1:
         abspath = os.path.abspath(source)
-        if not opts['pull']:
+        if not opts['pull'] and not opts['rev']:
             copy = True
 
     if copy:
         try:
             # we use a lock here because if we race with commit, we
             # can end up with extra data in the cloned revlogs that's
             # not pointed to by changesets, thus causing verify to
             # fail
             l1 = lock.lock(os.path.join(source, ".hg", "lock"))
         except OSError:
             copy = False
 
     if copy:
         # we lock here to avoid premature writing to the target
         os.mkdir(os.path.join(dest, ".hg"))
         l2 = lock.lock(os.path.join(dest, ".hg", "lock"))
 
         files = "data 00manifest.d 00manifest.i 00changelog.d 00changelog.i"
         for f in files.split():
             src = os.path.join(source, ".hg", f)
             dst = os.path.join(dest, ".hg", f)
             try:
                 util.copyfiles(src, dst)
             except OSError, inst:
                 if inst.errno != errno.ENOENT: raise
 
         repo = hg.repository(ui, dest)
 
     else:
+        revs = None
+        if opts['rev']:
+            if not other.local():
+                raise util.Abort("clone -r not supported yet for remote repositories.")
+            else:
+                revs = [other.lookup(rev) for rev in opts['rev']]
         repo = hg.repository(ui, dest, create=1)
-        repo.pull(other)
+        repo.pull(other, heads = revs)
 
     f = repo.opener("hgrc", "w", text=True)
     f.write("[paths]\n")
     f.write("default = %s\n" % abspath)
 
     if not opts['noupdate']:
         update(ui, repo)
 
     d.close()
 
 def commit(ui, repo, *pats, **opts):
     """commit the specified files or all outstanding changes
 
     Commit changes to the given files into the repository.
 
     If a list of files is omitted, all changes reported by "hg status"
     from the root of the repository will be commited.
 
     The HGEDITOR or EDITOR environment variables are used to start an
     editor to add a commit comment.
     """
     message = opts['message']
     logfile = opts['logfile']
 
     if message and logfile:
         raise util.Abort(_('options --message and --logfile are mutually '
                            'exclusive'))
     if not message and logfile:
         try:
             if logfile == '-':
                 message = sys.stdin.read()
             else:
                 message = open(logfile).read()
         except IOError, inst:
             raise util.Abort(_("can't read commit message '%s': %s") %
                              (logfile, inst.strerror))
 
     if opts['addremove']:
         addremove(ui, repo, *pats, **opts)
     cwd = repo.getcwd()
     if not pats and cwd:
         opts['include'] = [os.path.join(cwd, i) for i in opts['include']]
         opts['exclude'] = [os.path.join(cwd, x) for x in opts['exclude']]
     fns, match, anypats = matchpats(repo, (pats and repo.getcwd()) or '',
                                     pats, opts)
     if pats:
         c, a, d, u = repo.changes(files=fns, match=match)
         files = c + a + [fn for fn in d if repo.dirstate.state(fn) == 'r']
     else:
         files = []
     try:
         repo.commit(files, message, opts['user'], opts['date'], match)
     except ValueError, inst:
         raise util.Abort(str(inst))
 
 def docopy(ui, repo, pats, opts):
     if not pats:
         raise util.Abort(_('no source or destination specified'))
     elif len(pats) == 1:
         raise util.Abort(_('no destination specified'))
     pats = list(pats)
     dest = pats.pop()
     sources = []
     dir2dir = len(pats) == 1 and os.path.isdir(pats[0])
 
     def okaytocopy(abs, rel, exact):
         reasons = {'?': _('is not managed'),
                    'a': _('has been marked for add')}
         reason = reasons.get(repo.dirstate.state(abs))
         if reason:
             if exact: ui.warn(_('%s: not copying - file %s\n') % (rel, reason))
         else:
             return True
 
     for src, abs, rel, exact in walk(repo, pats, opts):
         if okaytocopy(abs, rel, exact):
             sources.append((abs, rel, exact))
     if not sources:
         raise util.Abort(_('no files to copy'))
 
     cwd = repo.getcwd()
     absdest = util.canonpath(repo.root, cwd, dest)
     reldest = util.pathto(cwd, absdest)
     if os.path.exists(reldest):
         destisfile = not os.path.isdir(reldest)
     else:
         destisfile = not dir2dir and (len(sources) == 1
                                       or repo.dirstate.state(absdest) != '?')
 
     if destisfile and len(sources) > 1:
         raise util.Abort(_('with multiple sources, destination must be a '
                            'directory'))
 
     srcpfxlen = 0
     if dir2dir:
         srcpfx = util.pathto(cwd, util.canonpath(repo.root, cwd, pats[0]))
         if os.path.exists(reldest):
             srcpfx = os.path.split(srcpfx)[0]
         if srcpfx:
             srcpfx += os.sep
         srcpfxlen = len(srcpfx)
 
     errs, copied = 0, []
     for abs, rel, exact in sources:
         if destisfile:
             mydest = reldest
         elif dir2dir:
             mydest = os.path.join(dest, rel[srcpfxlen:])
         else:
             mydest = os.path.join(dest, os.path.basename(rel))
         myabsdest = util.canonpath(repo.root, cwd, mydest)
         myreldest = util.pathto(cwd, myabsdest)
         if not opts['force'] and repo.dirstate.state(myabsdest) not in 'a?':
             ui.warn(_('%s: not overwriting - file already managed\n') % myreldest)
             continue
         mydestdir = os.path.dirname(myreldest) or '.'
         if not opts['after']:
             try:
                 if dir2dir: os.makedirs(mydestdir)
                 elif not destisfile: os.mkdir(mydestdir)
             except OSError, inst:
                 if inst.errno != errno.EEXIST: raise
         if ui.verbose or not exact:
             ui.status(_('copying %s to %s\n') % (rel, myreldest))
         if not opts['after']:
             try:
                 shutil.copyfile(rel, myreldest)
                 shutil.copymode(rel, myreldest)
             except shutil.Error, inst:
                 raise util.Abort(str(inst))
             except IOError, inst:
                 if inst.errno == errno.ENOENT:
                     ui.warn(_('%s: deleted in working copy\n') % rel)
                 else:
                     ui.warn(_('%s: cannot copy - %s\n') % (rel, inst.strerror))
                 errs += 1
                 continue
         repo.copy(abs, myabsdest)
         copied.append((abs, rel, exact))
     if errs:
         ui.warn(_('(consider using --after)\n'))
     return errs, copied
 
 def copy(ui, repo, *pats, **opts):
     """mark files as copied for the next commit
 
     Mark dest as having copies of source files.  If dest is a
     directory, copies are put in that directory.  If dest is a file,
     there can only be one source.
 
     By default, this command copies the contents of files as they
     stand in the working directory.  If invoked with --after, the
     operation is recorded, but no copying is performed.
 
     This command takes effect in the next commit.
 
     NOTE: This command should be treated as experimental. While it
     should properly record copied files, this information is not yet
     fully used by merge, nor fully reported by log.
     """
     errs, copied = docopy(ui, repo, pats, opts)
     return errs
 
 def debugancestor(ui, index, rev1, rev2):
     """find the ancestor revision of two revisions in a given index"""
     r = revlog.revlog(file, index, "")
     a = r.ancestor(r.lookup(rev1), r.lookup(rev2))
     ui.write("%d:%s\n" % (r.rev(a), hex(a)))
 
 def debugcheckstate(ui, repo):
     """validate the correctness of the current dirstate"""
     parent1, parent2 = repo.dirstate.parents()
     repo.dirstate.read()
     dc = repo.dirstate.map
     keys = dc.keys()
     keys.sort()
     m1n = repo.changelog.read(parent1)[0]
     m2n = repo.changelog.read(parent2)[0]
     m1 = repo.manifest.read(m1n)
     m2 = repo.manifest.read(m2n)
     errors = 0
     for f in dc:
         state = repo.dirstate.state(f)
         if state in "nr" and f not in m1:
             ui.warn(_("%s in state %s, but not in manifest1\n") % (f, state))
             errors += 1
         if state in "a" and f in m1:
             ui.warn(_("%s in state %s, but also in manifest1\n") % (f, state))
             errors += 1
         if state in "m" and f not in m1 and f not in m2:
             ui.warn(_("%s in state %s, but not in either manifest\n") %
                     (f, state))
             errors += 1
     for f in m1:
         state = repo.dirstate.state(f)
         if state not in "nrm":
             ui.warn(_("%s in manifest1, but listed as state %s") % (f, state))
             errors += 1
     if errors:
         raise util.Abort(_(".hg/dirstate inconsistent with current parent's manifest"))
 
 def debugconfig(ui):
     """show combined config settings from all hgrc files"""
     try:
         repo = hg.repository(ui)
     except hg.RepoError:
         pass
     for section, name, value in ui.walkconfig():
         ui.write('%s.%s=%s\n' % (section, name, value))
 
 def debugsetparents(ui, repo, rev1, rev2=None):
     """
     manually set the parents of the current working directory
 
     This is useful for writing repository conversion tools, but should
     be used with care.
     """
 
     if not rev2:
         rev2 = hex(nullid)
 
     repo.dirstate.setparents(repo.lookup(rev1), repo.lookup(rev2))
 
 def debugstate(ui, repo):
     """show the contents of the current dirstate"""
     repo.dirstate.read()
     dc = repo.dirstate.map
     keys = dc.keys()
     keys.sort()
     for file_ in keys:
         ui.write("%c %3o %10d %s %s\n"
                  % (dc[file_][0], dc[file_][1] & 0777, dc[file_][2],
                     time.strftime("%x %X",
                                   time.localtime(dc[file_][3])), file_))
     for f in repo.dirstate.copies:
         ui.write(_("copy: %s -> %s\n") % (repo.dirstate.copies[f], f))
 
 def debugdata(ui, file_, rev):
     """dump the contents of an data file revision"""
     r = revlog.revlog(file, file_[:-2] + ".i", file_)
     try:
         ui.write(r.revision(r.lookup(rev)))
     except KeyError:
         raise util.Abort(_('invalid revision identifier %s'), rev)
 
 def debugindex(ui, file_):
     """dump the contents of an index file"""
     r = revlog.revlog(file, file_, "")
     ui.write("   rev    offset  length   base linkrev" +
              " nodeid       p1           p2\n")
     for i in range(r.count()):
         e = r.index[i]
         ui.write("% 6d % 9d % 7d % 6d % 7d %s %s %s\n" % (
                 i, e[0], e[1], e[2], e[3],
             short(e[6]), short(e[4]), short(e[5])))
 
 def debugindexdot(ui, file_):
     """dump an index DAG as a .dot file"""
     r = revlog.revlog(file, file_, "")
     ui.write("digraph G {\n")
     for i in range(r.count()):
         e = r.index[i]
         ui.write("\t%d -> %d\n" % (r.rev(e[4]), i))
         if e[5] != nullid:
             ui.write("\t%d -> %d\n" % (r.rev(e[5]), i))
     ui.write("}\n")
 
 def debugrename(ui, repo, file, rev=None):
     """dump rename information"""
     r = repo.file(relpath(repo, [file])[0])
     if rev:
         try:
             # assume all revision numbers are for changesets
             n = repo.lookup(rev)
             change = repo.changelog.read(n)
             m = repo.manifest.read(change[0])
             n = m[relpath(repo, [file])[0]]
         except hg.RepoError, KeyError:
             n = r.lookup(rev)
     else:
         n = r.tip()
     m = r.renamed(n)
     if m:
         ui.write(_("renamed from %s:%s\n") % (m[0], hex(m[1])))
     else:
         ui.write(_("not renamed\n"))
 
 def debugwalk(ui, repo, *pats, **opts):
     """show how files match on given patterns"""
     items = list(walk(repo, pats, opts))
     if not items:
         return
     fmt = '%%s  %%-%ds  %%-%ds  %%s' % (
         max([len(abs) for (src, abs, rel, exact) in items]),
         max([len(rel) for (src, abs, rel, exact) in items]))
     for src, abs, rel, exact in items:
         line = fmt % (src, abs, rel, exact and 'exact' or '')
         ui.write("%s\n" % line.rstrip())
 
 def diff(ui, repo, *pats, **opts):
     """diff working directory (or selected files)
 
     Show differences between revisions for the specified files.
 
     Differences between files are shown using the unified diff format.
 
     When two revision arguments are given, then changes are shown
     between those revisions. If only one revision is specified then
     that revision is compared to the working directory, and, when no
     revisions are specified, the working directory files are compared
     to its parent.
 
     Without the -a option, diff will avoid generating diffs of files
     it detects as binary. With -a, diff will generate a diff anyway,
     probably with undesirable results.
     """
     node1, node2 = None, None
     revs = [repo.lookup(x) for x in opts['rev']]
 
     if len(revs) > 0:
         node1 = revs[0]
     if len(revs) > 1:
         node2 = revs[1]
     if len(revs) > 2:
         raise util.Abort(_("too many revisions to diff"))
 
     fns, matchfn, anypats = matchpats(repo, repo.getcwd(), pats, opts)
 
     dodiff(sys.stdout, ui, repo, node1, node2, fns, match=matchfn,
            text=opts['text'])
 
 def doexport(ui, repo, changeset, seqno, total, revwidth, opts):
     node = repo.lookup(changeset)
     prev, other = repo.changelog.parents(node)
     change = repo.changelog.read(node)
 
     fp = make_file(repo, repo.changelog, opts['output'],
                    node=node, total=total, seqno=seqno,
                    revwidth=revwidth)
     if fp != sys.stdout:
         ui.note("%s\n" % fp.name)
 
     fp.write("# HG changeset patch\n")
     fp.write("# User %s\n" % change[1])
     fp.write("# Node ID %s\n" % hex(node))
     fp.write("# Parent  %s\n" % hex(prev))
     if other != nullid:
         fp.write("# Parent  %s\n" % hex(other))
     fp.write(change[4].rstrip())
     fp.write("\n\n")
 
     dodiff(fp, ui, repo, prev, node, text=opts['text'])
     if fp != sys.stdout:
         fp.close()
 
 def export(ui, repo, *changesets, **opts):
     """dump the header and diffs for one or more changesets
 
     Print the changeset header and diffs for one or more revisions.
 
     The information shown in the changeset header is: author,
     changeset hash, parent and commit comment.
 
     Output may be to a file, in which case the name of the file is
     given using a format string.  The formatting rules are as follows:
 
     %%   literal "%" character
     %H   changeset hash (40 bytes of hexadecimal)
     %N   number of patches being generated
     %R   changeset revision number
     %b   basename of the exporting repository
     %h   short-form changeset hash (12 bytes of hexadecimal)
     %n   zero-padded sequence number, starting at 1
     %r   zero-padded changeset revision number
 
     Without the -a option, export will avoid generating diffs of files
     it detects as binary. With -a, export will generate a diff anyway,
     probably with undesirable results.
     """
     if not changesets:
         raise util.Abort(_("export requires at least one changeset"))
     seqno = 0
     revs = list(revrange(ui, repo, changesets))
     total = len(revs)
     revwidth = max(map(len, revs))
     ui.note(len(revs) > 1 and _("Exporting patches:\n") or _("Exporting patch:\n"))
     for cset in revs:
         seqno += 1
         doexport(ui, repo, cset, seqno, total, revwidth, opts)
 
 def forget(ui, repo, *pats, **opts):
     """don't add the specified files on the next commit
 
     Undo an 'hg add' scheduled for the next commit.
     """
     forget = []
     for src, abs, rel, exact in walk(repo, pats, opts):
         if repo.dirstate.state(abs) == 'a':
             forget.append(abs)
             if ui.verbose or not exact:
                 ui.status(_('forgetting %s\n') % rel)
     repo.forget(forget)
 
 def grep(ui, repo, pattern, *pats, **opts):
     """search for a pattern in specified files and revisions
 
     Search revisions of files for a regular expression.
 
     This command behaves differently than Unix grep.  It only accepts
     Python/Perl regexps.  It searches repository history, not the
     working directory.  It always prints the revision number in which
     a match appears.
 
     By default, grep only prints output for the first revision of a
     file in which it finds a match.  To get it to print every revision
     that contains a change in match status ("-" for a match that
     becomes a non-match, or "+" for a non-match that becomes a match),
     use the --all flag.
     """
     reflags = 0
     if opts['ignore_case']:
         reflags |= re.I
     regexp = re.compile(pattern, reflags)
     sep, eol = ':', '\n'
     if opts['print0']:
         sep = eol = '\0'
 
     fcache = {}
     def getfile(fn):
         if fn not in fcache:
             fcache[fn] = repo.file(fn)
         return fcache[fn]
 
     def matchlines(body):
         begin = 0
         linenum = 0
         while True:
             match = regexp.search(body, begin)
             if not match:
                 break
             mstart, mend = match.span()
             linenum += body.count('\n', begin, mstart) + 1
             lstart = body.rfind('\n', begin, mstart) + 1 or begin
             lend = body.find('\n', mend)
             yield linenum, mstart - lstart, mend - lstart, body[lstart:lend]
             begin = lend + 1
 
     class linestate:
         def __init__(self, line, linenum, colstart, colend):
             self.line = line
             self.linenum = linenum
             self.colstart = colstart
             self.colend = colend
         def __eq__(self, other):
             return self.line == other.line
         def __hash__(self):
             return hash(self.line)
 
     matches = {}
     def grepbody(fn, rev, body):
         matches[rev].setdefault(fn, {})
         m = matches[rev][fn]
         for lnum, cstart, cend, line in matchlines(body):
             s = linestate(line, lnum, cstart, cend)
             m[s] = s
 
     prev = {}
     ucache = {}
     def display(fn, rev, states, prevstates):
         diff = list(sets.Set(states).symmetric_difference(sets.Set(prevstates)))
         diff.sort(lambda x, y: cmp(x.linenum, y.linenum))
         counts = {'-': 0, '+': 0}
         filerevmatches = {}
         for l in diff:
             if incrementing or not opts['all']:
                 change = ((l in prevstates) and '-') or '+'
                 r = rev
             else:
                 change = ((l in states) and '-') or '+'
                 r = prev[fn]
             cols = [fn, str(rev)]
             if opts['line_number']: cols.append(str(l.linenum))
             if opts['all']: cols.append(change)
             if opts['user']: cols.append(trimuser(ui, getchange(rev)[1], rev,
                                                   ucache))
             if opts['files_with_matches']:
                 c = (fn, rev)
                 if c in filerevmatches: continue
                 filerevmatches[c] = 1
             else:
                 cols.append(l.line)
             ui.write(sep.join(cols), eol)
             counts[change] += 1
         return counts['+'], counts['-']
 
     fstate = {}
     skip = {}
     changeiter, getchange = walkchangerevs(ui, repo, repo.getcwd(), pats, opts)
     count = 0
     incrementing = False
     for st, rev, fns in changeiter:
         if st == 'window':
             incrementing = rev
             matches.clear()
         elif st == 'add':
             change = repo.changelog.read(repo.lookup(str(rev)))
             mf = repo.manifest.read(change[0])
             matches[rev] = {}
             for fn in fns:
                 if fn in skip: continue
                 fstate.setdefault(fn, {})
                 try:
                     grepbody(fn, rev, getfile(fn).read(mf[fn]))
                 except KeyError:
                     pass
         elif st == 'iter':
             states = matches[rev].items()
             states.sort()
             for fn, m in states:
                 if fn in skip: continue
                 if incrementing or not opts['all'] or fstate[fn]:
                     pos, neg = display(fn, rev, m, fstate[fn])
                     count += pos + neg
                     if pos and not opts['all']:
                         skip[fn] = True
                 fstate[fn] = m
                 prev[fn] = rev
 
     if not incrementing:
         fstate = fstate.items()
         fstate.sort()
         for fn, state in fstate:
             if fn in skip: continue
             display(fn, rev, {}, state)
     return (count == 0 and 1) or 0
 
 def heads(ui, repo, **opts):
     """show current repository heads
 
     Show all repository head changesets.
 
     Repository "heads" are changesets that don't have children
     changesets. They are where development generally takes place and
     are the usual targets for update and merge operations.
     """
     heads = repo.changelog.heads()
     br = None
     if opts['branches']:
         br = repo.branchlookup(heads)
     for n in repo.changelog.heads():
         show_changeset(ui, repo, changenode=n, brinfo=br)
 
 def identify(ui, repo):
     """print information about the working copy
     Print a short summary of the current state of the repo.
 
     This summary identifies the repository state using one or two parent
     hash identifiers, followed by a "+" if there are uncommitted changes
     in the working directory, followed by a list of tags for this revision.
     """
     parents = [p for p in repo.dirstate.parents() if p != nullid]
     if not parents:
         ui.write(_("unknown\n"))
         return
 
     hexfunc = ui.verbose and hex or short
     (c, a, d, u) = repo.changes()
     output = ["%s%s" % ('+'.join([hexfunc(parent) for parent in parents]),
                         (c or a or d) and "+" or "")]
 
     if not ui.quiet:
         # multiple tags for a single parent separated by '/'
         parenttags = ['/'.join(tags)
                       for tags in map(repo.nodetags, parents) if tags]
         # tags for multiple parents separated by ' + '
         if parenttags:
             output.append(' + '.join(parenttags))
 
     ui.write("%s\n" % ' '.join(output))
 
 def import_(ui, repo, patch1, *patches, **opts):
     """import an ordered set of patches
 
     Import a list of patches and commit them individually.
 
     If there are outstanding changes in the working directory, import
     will abort unless given the -f flag.
 
     If a patch looks like a mail message (its first line starts with
     "From " or looks like an RFC822 header), it will not be applied
     unless the -f option is used.  The importer neither parses nor
     discards mail headers, so use -f only to override the "mailness"
     safety check, not to import a real mail message.
     """
     patches = (patch1,) + patches
 
     if not opts['force']:
         (c, a, d, u) = repo.changes()
         if c or a or d:
             raise util.Abort(_("outstanding uncommitted changes"))
 
     d = opts["base"]
     strip = opts["strip"]
 
     mailre = re.compile(r'(?:From |[\w-]+:)')
 
     # attempt to detect the start of a patch
     # (this heuristic is borrowed from quilt)
     diffre = re.compile(r'(?:Index:[ \t]|diff[ \t]|RCS file: |' +
                         'retrieving revision [0-9]+(\.[0-9]+)*$|' +
                         '(---|\*\*\*)[ \t])')
 
     for patch in patches:
         ui.status(_("applying %s\n") % patch)
         pf = os.path.join(d, patch)
 
         message = []
         user = None
         hgpatch = False
         for line in file(pf):
             line = line.rstrip()
             if (not message and not hgpatch and
                    mailre.match(line) and not opts['force']):
                 if len(line) > 35: line = line[:32] + '...'
                 raise util.Abort(_('first line looks like a '
                                    'mail header: ') + line)
             if diffre.match(line):
                 break
             elif hgpatch:
                 # parse values when importing the result of an hg export
                 if line.startswith("# User "):
                     user = line[7:]
                     ui.debug(_('User: %s\n') % user)
                 elif not line.startswith("# ") and line:
                     message.append(line)
                     hgpatch = False
             elif line == '# HG changeset patch':
                 hgpatch = True
                 message = []       # We may have collected garbage
             else:
                 message.append(line)
 
         # make sure message isn't empty
         if not message:
             message = _("imported patch %s\n") % patch
         else:
             message = "%s\n" % '\n'.join(message)
         ui.debug(_('message:\n%s\n') % message)
 
         files = util.patch(strip, pf, ui)
 
         if len(files) > 0:
             addremove(ui, repo, *files)
         repo.commit(files, message, user)
 
 def incoming(ui, repo, source="default", **opts):
     """show new changesets found in source
 
     Show new changesets found in the specified repo or the default
     pull repo. These are the changesets that would be pulled if a pull
     was requested.
 
     Currently only local repositories are supported.
     """
     source = ui.expandpath(source, repo.root)
     other = hg.repository(ui, source)
     if not other.local():
         raise util.Abort(_("incoming doesn't work for remote repositories yet"))
     o = repo.findincoming(other)
     if not o:
         return
-    o = other.newer(o)
+    o = other.changelog.nodesbetween(o)[0]
     if opts['newest_first']:
         o.reverse()
     for n in o:
         parents = [p for p in other.changelog.parents(n) if p != nullid]
         if opts['no_merges'] and len(parents) == 2:
             continue
         show_changeset(ui, other, changenode=n)
         if opts['patch']:
             prev = (parents and parents[0]) or nullid
             dodiff(ui, ui, other, prev, n)
             ui.write("\n")
 
 def init(ui, dest="."):
     """create a new repository in the given directory
 
     Initialize a new repository in the given directory.  If the given
     directory does not exist, it is created.
 
     If no directory is given, the current directory is used.
     """
     if not os.path.exists(dest):
         os.mkdir(dest)
     hg.repository(ui, dest, create=1)
 
 def locate(ui, repo, *pats, **opts):
     """locate files matching specific patterns
 
     Print all files under Mercurial control whose names match the
     given patterns.
 
     This command searches the current directory and its
     subdirectories.  To search an entire repository, move to the root
     of the repository.
 
     If no patterns are given to match, this command prints all file
     names.
 
     If you want to feed the output of this command into the "xargs"
     command, use the "-0" option to both this command and "xargs".
     This will avoid the problem of "xargs" treating single filenames
     that contain white space as multiple filenames.
     """
     end = opts['print0'] and '\0' or '\n'
 
     for src, abs, rel, exact in walk(repo, pats, opts, '(?:.*/|)'):
         if repo.dirstate.state(abs) == '?':
             continue
         if opts['fullpath']:
             ui.write(os.path.join(repo.root, abs), end)
         else:
             ui.write(rel, end)
 
 def log(ui, repo, *pats, **opts):
     """show revision history of entire repository or files
 
     Print the revision history of the specified files or the entire project.
 
     By default this command outputs: changeset id and hash, tags,
     parents, user, date and time, and a summary for each commit. The
     -v switch adds some more detail, such as changed files, manifest
     hashes or message signatures.
     """
     class dui:
         # Implement and delegate some ui protocol.  Save hunks of
         # output for later display in the desired order.
         def __init__(self, ui):
             self.ui = ui
             self.hunk = {}
         def bump(self, rev):
             self.rev = rev
             self.hunk[rev] = []
         def note(self, *args):
             if self.verbose:
                 self.write(*args)
         def status(self, *args):
             if not self.quiet:
                 self.write(*args)
         def write(self, *args):
             self.hunk[self.rev].append(args)
         def debug(self, *args):
             if self.debugflag:
                 self.write(*args)
         def __getattr__(self, key):
             return getattr(self.ui, key)
     cwd = repo.getcwd()
     if not pats and cwd:
         opts['include'] = [os.path.join(cwd, i) for i in opts['include']]
         opts['exclude'] = [os.path.join(cwd, x) for x in opts['exclude']]
     changeiter, getchange = walkchangerevs(ui, repo, (pats and cwd) or '',
                                            pats, opts)
     for st, rev, fns in changeiter:
         if st == 'window':
             du = dui(ui)
         elif st == 'add':
             du.bump(rev)
             changenode = repo.changelog.node(rev)
             parents = [p for p in repo.changelog.parents(changenode)
                        if p != nullid]
             if opts['no_merges'] and len(parents) == 2:
                  continue
             if opts['only_merges'] and len(parents) != 2:
                  continue
 
             br = None
             if opts['keyword']:
                 changes = repo.changelog.read(repo.changelog.node(rev))
                 miss = 0
                 for k in [kw.lower() for kw in opts['keyword']]:
                     if not (k in changes[1].lower() or
                             k in changes[4].lower() or
                             k in " ".join(changes[3][:20]).lower()):
                         miss = 1
                         break
                 if miss:
                     continue
 
             if opts['branch']:
                 br = repo.branchlookup([repo.changelog.node(rev)])
 
             show_changeset(du, repo, rev, brinfo=br)
             if opts['patch']:
                 prev = (parents and parents[0]) or nullid
                 dodiff(du, du, repo, prev, changenode, fns)
                 du.write("\n\n")
         elif st == 'iter':
             for args in du.hunk[rev]:
                 ui.write(*args)
 
 def manifest(ui, repo, rev=None):
     """output the latest or given revision of the project manifest
 
     Print a list of version controlled files for the given revision.
 
     The manifest is the list of files being version controlled. If no revision
     is given then the tip is used.
     """
     if rev:
         try:
             # assume all revision numbers are for changesets
             n = repo.lookup(rev)
             change = repo.changelog.read(n)
             n = change[0]
         except hg.RepoError:
             n = repo.manifest.lookup(rev)
     else:
         n = repo.manifest.tip()
     m = repo.manifest.read(n)
     mf = repo.manifest.readflags(n)
     files = m.keys()
     files.sort()
 
     for f in files:
         ui.write("%40s %3s %s\n" % (hex(m[f]), mf[f] and "755" or "644", f))
 
 def outgoing(ui, repo, dest="default-push", **opts):
     """show changesets not found in destination
 
     Show changesets not found in the specified destination repo or the
     default push repo. These are the changesets that would be pushed
     if a push was requested.
     """
     dest = ui.expandpath(dest, repo.root)
     other = hg.repository(ui, dest)
     o = repo.findoutgoing(other)
-    o = repo.newer(o)
+    o = repo.changelog.nodesbetween(o)[0]
     if opts['newest_first']:
         o.reverse()
     for n in o:
         parents = [p for p in repo.changelog.parents(n) if p != nullid]
         if opts['no_merges'] and len(parents) == 2:
             continue
         show_changeset(ui, repo, changenode=n)
         if opts['patch']:
             prev = (parents and parents[0]) or nullid
             dodiff(ui, ui, repo, prev, n)
             ui.write("\n")
 
 def parents(ui, repo, rev=None):
     """show the parents of the working dir or revision
 
     Print the working directory's parent revisions.
     """
     if rev:
         p = repo.changelog.parents(repo.lookup(rev))
     else:
         p = repo.dirstate.parents()
 
     for n in p:
         if n != nullid:
             show_changeset(ui, repo, changenode=n)
 
 def paths(ui, search=None):
     """show definition of symbolic path names
 
     Show definition of symbolic path name NAME. If no name is given, show
     definition of available names.
 
     Path names are defined in the [paths] section of /etc/mercurial/hgrc
     and $HOME/.hgrc.  If run inside a repository, .hg/hgrc is used, too.
     """
     try:
         repo = hg.repository(ui=ui)
     except hg.RepoError:
         pass
 
     if search:
         for name, path in ui.configitems("paths"):
             if name == search:
                 ui.write("%s\n" % path)
                 return
         ui.warn(_("not found!\n"))
         return 1
     else:
         for name, path in ui.configitems("paths"):
             ui.write("%s = %s\n" % (name, path))
 
 def pull(ui, repo, source="default", **opts):
     """pull changes from the specified source
 
     Pull changes from a remote repository to a local one.
 
     This finds all changes from the repository at the specified path
     or URL and adds them to the local repository. By default, this
     does not update the copy of the project in the working directory.
 
     Valid URLs are of the form:
 
       local/filesystem/path
       http://[user@]host[:port][/path]
       https://[user@]host[:port][/path]
       ssh://[user@]host[:port][/path]
 
     SSH requires an accessible shell account on the destination machine
     and a copy of hg in the remote path.  With SSH, paths are relative
     to the remote user's home directory by default; use two slashes at
     the start of a path to specify it as relative to the filesystem root.
     """
     source = ui.expandpath(source, repo.root)
     ui.status(_('pulling from %s\n') % (source))
 
     if opts['ssh']:
         ui.setconfig("ui", "ssh", opts['ssh'])
     if opts['remotecmd']:
         ui.setconfig("ui", "remotecmd", opts['remotecmd'])
 
     other = hg.repository(ui, source)
-    r = repo.pull(other)
+    revs = None
+    if opts['rev'] and not other.local():
+        raise util.Abort("pull -r doesn't work for remote repositories yet")
+    elif opts['rev']:
+        revs = [other.lookup(rev) for rev in opts['rev']]
+    r = repo.pull(other, heads=revs)
     if not r:
         if opts['update']:
             return update(ui, repo)
         else:
             ui.status(_("(run 'hg update' to get a working copy)\n"))
 
     return r
 
 def push(ui, repo, dest="default-push", force=False, ssh=None, remotecmd=None):
     """push changes to the specified destination
 
     Push changes from the local repository to the given destination.
 
     This is the symmetrical operation for pull. It helps to move
     changes from the current repository to a different one. If the
     destination is local this is identical to a pull in that directory
     from the current one.
 
     By default, push will refuse to run if it detects the result would
     increase the number of remote heads. This generally indicates the
     the client has forgotten to sync and merge before pushing.
 
     Valid URLs are of the form:
 
       local/filesystem/path
       ssh://[user@]host[:port][/path]
 
     SSH requires an accessible shell account on the destination
     machine and a copy of hg in the remote path.
     """
     dest = ui.expandpath(dest, repo.root)
     ui.status('pushing to %s\n' % (dest))
 
     if ssh:
         ui.setconfig("ui", "ssh", ssh)
     if remotecmd:
         ui.setconfig("ui", "remotecmd", remotecmd)
 
     other = hg.repository(ui, dest)
     r = repo.push(other, force)
     return r
 
 def rawcommit(ui, repo, *flist, **rc):
     """raw commit interface
 
     Lowlevel commit, for use in helper scripts.
 
     This command is not intended to be used by normal users, as it is
     primarily useful for importing from other SCMs.
     """
     if rc['text']:
         ui.warn(_("Warning: -t and --text is deprecated,"
                   " please use -m or --message instead.\n"))
     message = rc['message'] or rc['text']
     if not message and rc['logfile']:
         try:
             message = open(rc['logfile']).read()
         except IOError:
             pass
     if not message and not rc['logfile']:
         raise util.Abort(_("missing commit message"))
 
     files = relpath(repo, list(flist))
     if rc['files']:
         files += open(rc['files']).read().splitlines()
 
     rc['parent'] = map(repo.lookup, rc['parent'])
 
     try:
         repo.rawcommit(files, message, rc['user'], rc['date'], *rc['parent'])
     except ValueError, inst:
         raise util.Abort(str(inst))
 
 def recover(ui, repo):
     """roll back an interrupted transaction
 
     Recover from an interrupted commit or pull.
 
     This command tries to fix the repository status after an interrupted
     operation. It should only be necessary when Mercurial suggests it.
     """
     repo.recover()
 
 def remove(ui, repo, pat, *pats, **opts):
     """remove the specified files on the next commit
 
     Schedule the indicated files for removal from the repository.
 
     This command schedules the files to be removed at the next commit.
     This only removes files from the current branch, not from the
     entire project history.  If the files still exist in the working
     directory, they will be deleted from it.
     """
     names = []
     def okaytoremove(abs, rel, exact):
         c, a, d, u = repo.changes(files = [abs])
         reason = None
         if c: reason = _('is modified')
         elif a: reason = _('has been marked for add')
         elif u: reason = _('is not managed')
         if reason:
             if exact: ui.warn(_('not removing %s: file %s\n') % (rel, reason))
         else:
             return True
     for src, abs, rel, exact in walk(repo, (pat,) + pats, opts):
         if okaytoremove(abs, rel, exact):
             if ui.verbose or not exact: ui.status(_('removing %s\n') % rel)
             names.append(abs)
     repo.remove(names, unlink=True)
 
 def rename(ui, repo, *pats, **opts):
     """rename files; equivalent of copy + remove
 
     Mark dest as copies of sources; mark sources for deletion.  If
     dest is a directory, copies are put in that directory.  If dest is
     a file, there can only be one source.
 
     By default, this command copies the contents of files as they
     stand in the working directory.  If invoked with --after, the
     operation is recorded, but no copying is performed.
 
     This command takes effect in the next commit.
 
     NOTE: This command should be treated as experimental. While it
     should properly record rename files, this information is not yet
     fully used by merge, nor fully reported by log.
     """
     errs, copied = docopy(ui, repo, pats, opts)
     names = []
     for abs, rel, exact in copied:
         if ui.verbose or not exact: ui.status(_('removing %s\n') % rel)
         names.append(abs)
     repo.remove(names, unlink=True)
     return errs
 
 def revert(ui, repo, *names, **opts):
     """revert modified files or dirs back to their unmodified states
 
     Revert any uncommitted modifications made to the named files or
     directories.  This restores the contents of the affected files to
     an unmodified state.
 
     If a file has been deleted, it is recreated.  If the executable
     mode of a file was changed, it is reset.
 
     If a directory is given, all files in that directory and its
     subdirectories are reverted.
 
     If no arguments are given, all files in the current directory and
     its subdirectories are reverted.
     """
     node = opts['rev'] and repo.lookup(opts['rev']) or \
            repo.dirstate.parents()[0]
     root = os.path.realpath(repo.root)
 
     def trimpath(p):
         p = os.path.realpath(p)
         if p.startswith(root):
             rest = p[len(root):]
             if not rest:
                 return rest
             if p.startswith(os.sep):
                 return rest[1:]
             return p
 
     relnames = map(trimpath, names or [os.getcwd()])
     chosen = {}
 
     def choose(name):
         def body(name):
             for r in relnames:
                 if not name.startswith(r):
                     continue
                 rest = name[len(r):]
                 if not rest:
                     return r, True
                 depth = rest.count(os.sep)
                 if not r:
                     if depth == 0 or not opts['nonrecursive']:
                         return r, True
                 elif rest[0] == os.sep:
                     if depth == 1 or not opts['nonrecursive']:
                         return r, True
             return None, False
         relname, ret = body(name)
         if ret:
             chosen[relname] = 1
         return ret
 
     (c, a, d, u) = repo.changes()
     repo.forget(filter(choose, a))
     repo.undelete(filter(choose, d))
 
     r = repo.update(node, False, True, choose, False)
     for n in relnames:
         if n not in chosen:
             ui.warn(_('error: no matches for %s\n') % n)
             r = 1
     sys.stdout.flush()
     return r
 
 def root(ui, repo):
     """print the root (top) of the current working dir
 
     Print the root directory of the current repository.
     """
     ui.write(repo.root + "\n")
 
 def serve(ui, repo, **opts):
     """export the repository via HTTP
 
     Start a local HTTP repository browser and pull server.
 
     By default, the server logs accesses to stdout and errors to
     stderr.  Use the "-A" and "-E" options to log to files.
     """
 
     if opts["stdio"]:
         fin, fout = sys.stdin, sys.stdout
         sys.stdout = sys.stderr
 
         # Prevent insertion/deletion of CRs
         util.set_binary(fin)
         util.set_binary(fout)
 
         def getarg():
             argline = fin.readline()[:-1]
             arg, l = argline.split()
             val = fin.read(int(l))
             return arg, val
         def respond(v):
             fout.write("%d\n" % len(v))
             fout.write(v)
             fout.flush()
 
         lock = None
 
         while 1:
             cmd = fin.readline()[:-1]
             if cmd == '':
                 return
             if cmd == "heads":
                 h = repo.heads()
                 respond(" ".join(map(hex, h)) + "\n")
             if cmd == "lock":
                 lock = repo.lock()
                 respond("")
             if cmd == "unlock":
                 if lock:
                     lock.release()
                 lock = None
                 respond("")
             elif cmd == "branches":
                 arg, nodes = getarg()
                 nodes = map(bin, nodes.split(" "))
                 r = []
                 for b in repo.branches(nodes):
                     r.append(" ".join(map(hex, b)) + "\n")
                 respond("".join(r))
             elif cmd == "between":
                 arg, pairs = getarg()
                 pairs = [map(bin, p.split("-")) for p in pairs.split(" ")]
                 r = []
                 for b in repo.between(pairs):
                     r.append(" ".join(map(hex, b)) + "\n")
                 respond("".join(r))
             elif cmd == "changegroup":
                 nodes = []
                 arg, roots = getarg()
                 nodes = map(bin, roots.split(" "))
 
                 cg = repo.changegroup(nodes)
                 while 1:
                     d = cg.read(4096)
                     if not d:
                         break
                     fout.write(d)
 
                 fout.flush()
 
             elif cmd == "addchangegroup":
                 if not lock:
                     respond("not locked")
                     continue
                 respond("")
 
                 r = repo.addchangegroup(fin)
                 respond("")
 
     optlist = "name templates style address port ipv6 accesslog errorlog"
     for o in optlist.split():
         if opts[o]:
             ui.setconfig("web", o, opts[o])
 
     try:
         httpd = hgweb.create_server(repo)
     except socket.error, inst:
         raise util.Abort('cannot start server: ' + inst.args[1])
 
     if ui.verbose:
         addr, port = httpd.socket.getsockname()
         if addr == '0.0.0.0':
             addr = socket.gethostname()
         else:
             try:
                 addr = socket.gethostbyaddr(addr)[0]
             except socket.error:
                 pass
         if port != 80:
             ui.status(_('listening at http://%s:%d/\n') % (addr, port))
         else:
             ui.status(_('listening at http://%s/\n') % addr)
     httpd.serve_forever()
 
 def status(ui, repo, *pats, **opts):
     """show changed files in the working directory
 
     Show changed files in the working directory.  If no names are
     given, all files are shown.  Otherwise, only files matching the
     given names are shown.
 
     The codes used to show the status of files are:
     M = modified
     A = added
     R = removed
     ? = not tracked
     """
 
     cwd = repo.getcwd()
     files, matchfn, anypats = matchpats(repo, cwd, pats, opts)
     (c, a, d, u) = [[util.pathto(cwd, x) for x in n]
                     for n in repo.changes(files=files, match=matchfn)]
 
     changetypes = [(_('modified'), 'M', c),
                    (_('added'), 'A', a),
                    (_('removed'), 'R', d),
                    (_('unknown'), '?', u)]
 
     end = opts['print0'] and '\0' or '\n'
 
     for opt, char, changes in ([ct for ct in changetypes if opts[ct[0]]]
                                or changetypes):
         if opts['no_status']:
             format = "%%s%s" % end
         else:
             format = "%s %%s%s" % (char, end);
 
         for f in changes:
             ui.write(format % f)
 
 def tag(ui, repo, name, rev=None, **opts):
     """add a tag for the current tip or a given revision
 
     Name a particular revision using <name>.
 
     Tags are used to name particular revisions of the repository and are
     very useful to compare different revision, to go back to significant
     earlier versions or to mark branch points as releases, etc.
 
     If no revision is given, the tip is used.
 
     To facilitate version control, distribution, and merging of tags,
     they are stored as a file named ".hgtags" which is managed
     similarly to other project files and can be hand-edited if
     necessary.
     """
     if opts['text']:
         ui.warn(_("Warning: -t and --text is deprecated,"
                   " please use -m or --message instead.\n"))
     if name == "tip":
         raise util.Abort(_("the name 'tip' is reserved"))
     if rev:
         r = hex(repo.lookup(rev))
     else:
         r = hex(repo.changelog.tip())
 
     if name.find(revrangesep) >= 0:
         raise util.Abort(_("'%s' cannot be used in a tag name") % revrangesep)
 
     if opts['local']:
         repo.opener("localtags", "a").write("%s %s\n" % (r, name))
         return
 
     (c, a, d, u) = repo.changes()
     for x in (c, a, d, u):
         if ".hgtags" in x:
             raise util.Abort(_("working copy of .hgtags is changed "
                                "(please commit .hgtags manually)"))
 
     repo.wfile(".hgtags", "ab").write("%s %s\n" % (r, name))
     if repo.dirstate.state(".hgtags") == '?':
         repo.add([".hgtags"])
 
     message = (opts['message'] or opts['text'] or
                _("Added tag %s for changeset %s") % (name, r))
     try:
         repo.commit([".hgtags"], message, opts['user'], opts['date'])
     except ValueError, inst:
         raise util.Abort(str(inst))
 
 def tags(ui, repo):
     """list repository tags
 
     List the repository tags.
 
     This lists both regular and local tags.
     """
 
     l = repo.tagslist()
     l.reverse()
     for t, n in l:
         try:
             r = "%5d:%s" % (repo.changelog.rev(n), hex(n))
         except KeyError:
             r = "    ?:?"
         ui.write("%-30s %s\n" % (t, r))
 
 def tip(ui, repo):
     """show the tip revision
 
     Show the tip revision.
     """
     n = repo.changelog.tip()
     show_changeset(ui, repo, changenode=n)
 
 def unbundle(ui, repo, fname):
     """apply a changegroup file
 
     Apply a compressed changegroup file generated by the bundle
     command.
     """
     f = urllib.urlopen(fname)
 
     if f.read(4) != "HG10":
         raise util.Abort(_("%s: not a Mercurial bundle file") % fname)
 
     def bzgenerator(f):
         zd = bz2.BZ2Decompressor()
         for chunk in f:
             yield zd.decompress(chunk)
 
     bzgen = bzgenerator(util.filechunkiter(f, 4096))
     repo.addchangegroup(util.chunkbuffer(bzgen))
 
 def undo(ui, repo):
     """undo the last commit or pull
 
     Roll back the last pull or commit transaction on the
     repository, restoring the project to its earlier state.
 
     This command should be used with care. There is only one level of
     undo and there is no redo.
 
     This command is not intended for use on public repositories. Once
     a change is visible for pull by other users, undoing it locally is
     ineffective.
     """
     repo.undo()
 
 def update(ui, repo, node=None, merge=False, clean=False, branch=None):
     """update or merge working directory
 
     Update the working directory to the specified revision.
 
     If there are no outstanding changes in the working directory and
     there is a linear relationship between the current version and the
     requested version, the result is the requested version.
 
     Otherwise the result is a merge between the contents of the
     current working directory and the requested version. Files that
     changed between either parent are marked as changed for the next
     commit and a commit must be performed before any further updates
     are allowed.
 
     By default, update will refuse to run if doing so would require
     merging or discarding local changes.
     """
     if branch:
         br = repo.branchlookup(branch=branch)
         found = []
         for x in br:
             if branch in br[x]:
                 found.append(x)
         if len(found) > 1:
             ui.warn(_("Found multiple heads for %s\n") % branch)
             for x in found:
                 show_changeset(ui, repo, changenode=x, brinfo=br)
             return 1
         if len(found) == 1:
             node = found[0]
             ui.warn(_("Using head %s for branch %s\n") % (short(node), branch))
         else:
             ui.warn(_("branch %s not found\n") % (branch))
             return 1
     else:
         node = node and repo.lookup(node) or repo.changelog.tip()
     return repo.update(node, allow=merge, force=clean)
 
 def verify(ui, repo):
     """verify the integrity of the repository
 
     Verify the integrity of the current repository.
 
     This will perform an extensive check of the repository's
     integrity, validating the hashes and checksums of each entry in
     the changelog, manifest, and tracked files, as well as the
     integrity of their crosslinks and indices.
     """
     return repo.verify()
 
 # Command options and aliases are listed here, alphabetically
 
 table = {
     "^add":
         (add,
          [('I', 'include', [], _('include names matching the given patterns')),
           ('X', 'exclude', [], _('exclude names matching the given patterns'))],
          "hg add [OPTION]... [FILE]..."),
     "addremove":
         (addremove,
          [('I', 'include', [], _('include names matching the given patterns')),
           ('X', 'exclude', [], _('exclude names matching the given patterns'))],
          "hg addremove [OPTION]... [FILE]..."),
     "^annotate":
         (annotate,
          [('r', 'rev', '', _('annotate the specified revision')),
           ('a', 'text', None, _('treat all files as text')),
           ('u', 'user', None, _('list the author')),
           ('n', 'number', None, _('list the revision number (default)')),
           ('c', 'changeset', None, _('list the changeset')),
           ('I', 'include', [], _('include names matching the given patterns')),
           ('X', 'exclude', [], _('exclude names matching the given patterns'))],
          _('hg annotate [OPTION]... FILE...')),
     "bundle":
         (bundle,
          [],
          _('hg bundle FILE DEST')),
     "cat":
         (cat,
          [('I', 'include', [], _('include names matching the given patterns')),
           ('X', 'exclude', [], _('exclude names matching the given patterns')),
           ('o', 'output', "", _('print output to file with formatted name')),
           ('r', 'rev', '', _('print the given revision'))],
          _('hg cat [OPTION]... FILE...')),
     "^clone":
         (clone,
          [('U', 'noupdate', None, _('do not update the new working directory')),
           ('e', 'ssh', "", _('specify ssh command to use')),
           ('', 'pull', None, _('use pull protocol to copy metadata')),
+          ('r', 'rev', [], _('a changeset you would like to have after cloning')),
           ('', 'remotecmd', "", _('specify hg command to run on the remote side'))],
          _('hg clone [OPTION]... SOURCE [DEST]')),
     "^commit|ci":
         (commit,
          [('A', 'addremove', None, _('run addremove during commit')),
           ('I', 'include', [], _('include names matching the given patterns')),
           ('X', 'exclude', [], _('exclude names matching the given patterns')),
           ('m', 'message', "", _('use <text> as commit message')),
           ('l', 'logfile', "", _('read the commit message from <file>')),
           ('d', 'date', "", _('record datecode as commit date')),
           ('u', 'user', "", _('record user as commiter'))],
          _('hg commit [OPTION]... [FILE]...')),
     "copy|cp": (copy,
              [('I', 'include', [], _('include names matching the given patterns')),
               ('X', 'exclude', [], _('exclude names matching the given patterns')),
               ('A', 'after', None, _('record a copy that has already occurred')),
               ('f', 'force', None, _('forcibly copy over an existing managed file'))],
              _('hg copy [OPTION]... [SOURCE]... DEST')),
     "debugancestor": (debugancestor, [], _('debugancestor INDEX REV1 REV2')),
     "debugcheckstate": (debugcheckstate, [], _('debugcheckstate')),
     "debugconfig": (debugconfig, [], _('debugconfig')),
     "debugsetparents": (debugsetparents, [], _('debugsetparents REV1 [REV2]')),
     "debugstate": (debugstate, [], _('debugstate')),
     "debugdata": (debugdata, [], _('debugdata FILE REV')),
     "debugindex": (debugindex, [], _('debugindex FILE')),
     "debugindexdot": (debugindexdot, [], _('debugindexdot FILE')),
     "debugrename": (debugrename, [], _('debugrename FILE [REV]')),
     "debugwalk":
         (debugwalk,
          [('I', 'include', [], _('include names matching the given patterns')),
           ('X', 'exclude', [], _('exclude names matching the given patterns'))],
          _('debugwalk [OPTION]... [FILE]...')),
     "^diff":
         (diff,
          [('r', 'rev', [], _('revision')),
           ('a', 'text', None, _('treat all files as text')),
           ('I', 'include', [], _('include names matching the given patterns')),
           ('X', 'exclude', [], _('exclude names matching the given patterns'))],
          _('hg diff [-a] [-I] [-X] [-r REV1 [-r REV2]] [FILE]...')),
     "^export":
         (export,
          [('o', 'output', "", _('print output to file with formatted name')),
           ('a', 'text', None, _('treat all files as text'))],
          "hg export [-a] [-o OUTFILE] REV..."),
     "forget":
         (forget,
          [('I', 'include', [], _('include names matching the given patterns')),
           ('X', 'exclude', [], _('exclude names matching the given patterns'))],
          "hg forget [OPTION]... FILE..."),
     "grep":
         (grep,
          [('0', 'print0', None, _('end fields with NUL')),
           ('I', 'include', [], _('include names matching the given patterns')),
           ('X', 'exclude', [], _('include names matching the given patterns')),
           ('', 'all', None, _('print all revisions that match')),
           ('i', 'ignore-case', None, _('ignore case when matching')),
           ('l', 'files-with-matches', None, _('print only filenames and revs that match')),
           ('n', 'line-number', None, _('print matching line numbers')),
           ('r', 'rev', [], _('search in given revision range')),
           ('u', 'user', None, _('print user who committed change'))],
          "hg grep [OPTION]... PATTERN [FILE]..."),
     "heads":
         (heads,
          [('b', 'branches', None, _('find branch info'))],
          _('hg heads [-b]')),
     "help": (help_, [], _('hg help [COMMAND]')),
     "identify|id": (identify, [], _('hg identify')),
     "import|patch":
         (import_,
          [('p', 'strip', 1, _('directory strip option for patch. This has the same\n') +
                             _('meaning as the corresponding patch option')),
           ('f', 'force', None, _('skip check for outstanding uncommitted changes')),
           ('b', 'base', "", _('base path'))],
          "hg import [-f] [-p NUM] [-b BASE] PATCH..."),
     "incoming|in": (incoming,
          [('M', 'no-merges', None, _("do not show merges")),
           ('p', 'patch', None, _('show patch')),
           ('n', 'newest-first', None, _('show newest record first'))],
          _('hg incoming [-p] [-n] [-M] [SOURCE]')),
     "^init": (init, [], _('hg init [DEST]')),
     "locate":
         (locate,
          [('r', 'rev', '', _('search the repository as it stood at rev')),
           ('0', 'print0', None, _('end filenames with NUL, for use with xargs')),
           ('f', 'fullpath', None, _('print complete paths from the filesystem root')),
           ('I', 'include', [], _('include names matching the given patterns')),
           ('X', 'exclude', [], _('exclude names matching the given patterns'))],
          _('hg locate [OPTION]... [PATTERN]...')),
     "^log|history":
         (log,
          [('I', 'include', [], _('include names matching the given patterns')),
           ('X', 'exclude', [], _('exclude names matching the given patterns')),
           ('b', 'branch', None, _('show branches')),
           ('k', 'keyword', [], _('search for a keyword')),
           ('r', 'rev', [], _('show the specified revision or range')),
           ('M', 'no-merges', None, _("do not show merges")),
           ('m', 'only-merges', None, _("show only merges")),
           ('p', 'patch', None, _('show patch'))],
          _('hg log [-I] [-X] [-r REV]... [-p] [FILE]')),
     "manifest": (manifest, [], _('hg manifest [REV]')),
     "outgoing|out": (outgoing,
          [('M', 'no-merges', None, _("do not show merges")),
           ('p', 'patch', None, _('show patch')),
           ('n', 'newest-first', None, _('show newest record first'))],
          _('hg outgoing [-p] [-n] [-M] [DEST]')),
     "parents": (parents, [], _('hg parents [REV]')),
     "paths": (paths, [], _('hg paths [NAME]')),
     "^pull":
         (pull,
          [('u', 'update', None, _('update the working directory to tip after pull')),
           ('e', 'ssh', "", _('specify ssh command to use')),
+          ('r', 'rev', [], _('a specific revision you would like to pull')),
           ('', 'remotecmd', "", _('specify hg command to run on the remote side'))],
-         _('hg pull [-u] [-e FILE] [--remotecmd FILE] [SOURCE]')),
+         _('hg pull [-u] [-e FILE] [-r rev] [--remotecmd FILE] [SOURCE]')),
     "^push":
         (push,
          [('f', 'force', None, _('force push')),
           ('e', 'ssh', "", _('specify ssh command to use')),
           ('', 'remotecmd', "", _('specify hg command to run on the remote side'))],
          _('hg push [-f] [-e FILE] [--remotecmd FILE] [DEST]')),
     "rawcommit":
         (rawcommit,
          [('p', 'parent', [], _('parent')),
           ('d', 'date', "", _('date code')),
           ('u', 'user', "", _('user')),
           ('F', 'files', "", _('file list')),
           ('m', 'message', "", _('commit message')),
           ('t', 'text', "", _('commit message (deprecated: use -m)')),
           ('l', 'logfile', "", _('commit message file'))],
          _('hg rawcommit [OPTION]... [FILE]...')),
     "recover": (recover, [], _("hg recover")),
     "^remove|rm": (remove,
                    [('I', 'include', [], _('include names matching the given patterns')),
                     ('X', 'exclude', [], _('exclude names matching the given patterns'))],
                    _("hg remove [OPTION]... FILE...")),
     "rename|mv": (rename,
                   [('I', 'include', [], _('include names matching the given patterns')),
                    ('X', 'exclude', [], _('exclude names matching the given patterns')),
                    ('A', 'after', None, _('record a rename that has already occurred')),
                    ('f', 'force', None, _('forcibly copy over an existing managed file'))],
                   _('hg rename [OPTION]... [SOURCE]... DEST')),
     "^revert":
         (revert,
          [("n", "nonrecursive", None, _("do not recurse into subdirectories")),
           ("r", "rev", "", _("revision to revert to"))],
          _("hg revert [-n] [-r REV] [NAME]...")),
     "root": (root, [], _("hg root")),
     "^serve":
         (serve,
          [('A', 'accesslog', '', _('name of access log file to write to')),
           ('E', 'errorlog', '', _('name of error log file to write to')),
           ('p', 'port', 0, _('port to use (default: 8000)')),
           ('a', 'address', '', _('address to use')),
           ('n', 'name', "", _('name to show in web pages (default: working dir)')),
           ('', 'stdio', None, _('for remote clients')),
           ('t', 'templates', "", _('web templates to use')),
           ('', 'style', "", _('template style to use')),
           ('6', 'ipv6', None, _('use IPv6 in addition to IPv4'))],
          _("hg serve [OPTION]...")),
     "^status":
         (status,
          [('m', 'modified', None, _('show only modified files')),
           ('a', 'added', None, _('show only added files')),
           ('r', 'removed', None, _('show only removed files')),
           ('u', 'unknown', None, _('show only unknown (not tracked) files')),
           ('n', 'no-status', None, _('hide status prefix')),
           ('0', 'print0', None, _('end filenames with NUL, for use with xargs')),
           ('I', 'include', [], _('include names matching the given patterns')),
           ('X', 'exclude', [], _('exclude names matching the given patterns'))],
          _("hg status [OPTION]... [FILE]...")),
     "tag":
         (tag,
          [('l', 'local', None, _('make the tag local')),
           ('m', 'message', "", _('message for tag commit log entry')),
           ('t', 'text', "", _('commit message (deprecated: use -m)')),
           ('d', 'date', "", _('record datecode as commit date')),
           ('u', 'user', "", _('record user as commiter'))],
          _('hg tag [OPTION]... NAME [REV]')),
     "tags": (tags, [], _('hg tags')),
     "tip": (tip, [], _('hg tip')),
     "unbundle":
         (unbundle,
          [],
          _('hg unbundle FILE')),
     "undo": (undo, [], _('hg undo')),
     "^update|up|checkout|co":
         (update,
          [('b', 'branch', "", _('checkout the head of a specific branch')),
           ('m', 'merge', None, _('allow merging of branches')),
           ('C', 'clean', None, _('overwrite locally modified files'))],
          _('hg update [-b TAG] [-m] [-C] [REV]')),
     "verify": (verify, [], _('hg verify')),
     "version": (show_version, [], _('hg version')),
 }
 
 globalopts = [
     ('R', 'repository', "", _("repository root directory")),
     ('', 'cwd', '', _("change working directory")),
     ('y', 'noninteractive', None, _("do not prompt, assume 'yes' for any required answers")),
     ('q', 'quiet', None, _("suppress output")),
     ('v', 'verbose', None, _("enable additional output")),
     ('', 'debug', None, _("enable debugging output")),
     ('', 'debugger', None, _("start debugger")),
     ('', 'traceback', None, _("print traceback on exception")),
     ('', 'time', None, _("time how long the command takes")),
     ('', 'profile', None, _("print command execution profile")),
     ('', 'version', None, _("output version information and exit")),
     ('h', 'help', None, _("display help and exit")),
 ]
 
 norepo = ("clone init version help debugancestor debugconfig debugdata"
           " debugindex debugindexdot paths")
 
 def find(cmd):
     for e in table.keys():
         if re.match("(%s)$" % e, cmd):
             return e, table[e]
 
     raise UnknownCommand(cmd)
 
 class SignalInterrupt(Exception):
     """Exception raised on SIGTERM and SIGHUP."""
 
 def catchterm(*args):
     raise SignalInterrupt
 
 def run():
     sys.exit(dispatch(sys.argv[1:]))
 
 class ParseError(Exception):
     """Exception raised on errors in parsing the command line."""
 
 def parse(ui, args):
     options = {}
     cmdoptions = {}
 
     try:
         args = fancyopts.fancyopts(args, globalopts, options)
     except fancyopts.getopt.GetoptError, inst:
         raise ParseError(None, inst)
 
     if args:
         cmd, args = args[0], args[1:]
         defaults = ui.config("defaults", cmd)
         if defaults:
             # reparse with command defaults added
             args = [cmd] + defaults.split() + args
             try:
                 args = fancyopts.fancyopts(args, globalopts, options)
             except fancyopts.getopt.GetoptError, inst:
                 raise ParseError(None, inst)
 
             cmd, args = args[0], args[1:]
 
         i = find(cmd)[1]
         c = list(i[1])
     else:
         cmd = None
         c = []
 
     # combine global options into local
     for o in globalopts:
         c.append((o[0], o[1], options[o[1]], o[3]))
 
     try:
         args = fancyopts.fancyopts(args, c, cmdoptions)
     except fancyopts.getopt.GetoptError, inst:
         raise ParseError(cmd, inst)
 
     # separate global options back out
     for o in globalopts:
         n = o[1]
         options[n] = cmdoptions[n]
         del cmdoptions[n]
 
     return (cmd, cmd and i[0] or None, args, options, cmdoptions)
 
 def dispatch(args):
     signal.signal(signal.SIGTERM, catchterm)
     try:
         signal.signal(signal.SIGHUP, catchterm)
     except AttributeError:
         pass
 
     u = ui.ui()
     external = []
     for x in u.extensions():
         def on_exception(Exception, inst):
             u.warn(_("*** failed to import extension %s\n") % x[1])
             u.warn("%s\n" % inst)
             if "--traceback" in sys.argv[1:]:
                 traceback.print_exc()
         if x[1]:
             try:
                 mod = imp.load_source(x[0], x[1])
             except Exception, inst:
                 on_exception(Exception, inst)
                 continue
         else:
             def importh(name):
                 mod = __import__(name)
                 components = name.split('.')
                 for comp in components[1:]:
                     mod = getattr(mod, comp)
                 return mod
             try:
                 mod = importh(x[0])
             except Exception, inst:
                 on_exception(Exception, inst)
                 continue
 
         external.append(mod)
     for x in external:
         cmdtable = getattr(x, 'cmdtable', {})
         for t in cmdtable:
             if t in table:
                 u.warn(_("module %s overrides %s\n") % (x.__name__, t))
         table.update(cmdtable)
 
     try:
         cmd, func, args, options, cmdoptions = parse(u, args)
     except ParseError, inst:
         if inst.args[0]:
             u.warn(_("hg %s: %s\n") % (inst.args[0], inst.args[1]))
             help_(u, inst.args[0])
         else:
             u.warn(_("hg: %s\n") % inst.args[1])
             help_(u, 'shortlist')
         sys.exit(-1)
     except UnknownCommand, inst:
         u.warn(_("hg: unknown command '%s'\n") % inst.args[0])
         help_(u, 'shortlist')
         sys.exit(1)
 
     if options["time"]:
         def get_times():
             t = os.times()
             if t[4] == 0.0: # Windows leaves this as zero, so use time.clock()
                 t = (t[0], t[1], t[2], t[3], time.clock())
             return t
         s = get_times()
         def print_time():
             t = get_times()
             u.warn(_("Time: real %.3f secs (user %.3f+%.3f sys %.3f+%.3f)\n") %
                 (t[4]-s[4], t[0]-s[0], t[2]-s[2], t[1]-s[1], t[3]-s[3]))
         atexit.register(print_time)
 
     u.updateopts(options["verbose"], options["debug"], options["quiet"],
               not options["noninteractive"])
 
     # enter the debugger before command execution
     if options['debugger']:
         pdb.set_trace()
 
     try:
         try:
             if options['help']:
                 help_(u, cmd, options['version'])
                 sys.exit(0)
             elif options['version']:
                 show_version(u)
                 sys.exit(0)
             elif not cmd:
                 help_(u, 'shortlist')
                 sys.exit(0)
 
             if options['cwd']:
                 try:
                     os.chdir(options['cwd'])
                 except OSError, inst:
                     raise util.Abort('%s: %s' %
                                      (options['cwd'], inst.strerror))
 
             if cmd not in norepo.split():
                 path = options["repository"] or ""
                 repo = hg.repository(ui=u, path=path)
                 for x in external:
                     if hasattr(x, 'reposetup'): x.reposetup(u, repo)
                 d = lambda: func(u, repo, *args, **cmdoptions)
             else:
                 d = lambda: func(u, *args, **cmdoptions)
 
             if options['profile']:
                 import hotshot, hotshot.stats
                 prof = hotshot.Profile("hg.prof")
                 r = prof.runcall(d)
                 prof.close()
                 stats = hotshot.stats.load("hg.prof")
                 stats.strip_dirs()
                 stats.sort_stats('time', 'calls')
                 stats.print_stats(40)
                 return r
             else:
                 return d()
         except:
             # enter the debugger when we hit an exception
             if options['debugger']:
                 pdb.post_mortem(sys.exc_info()[2])
             if options['traceback']:
                 traceback.print_exc()
             raise
     except hg.RepoError, inst:
         u.warn(_("abort: "), inst, "!\n")
     except revlog.RevlogError, inst:
         u.warn(_("abort: "), inst, "!\n")
     except SignalInterrupt:
         u.warn(_("killed!\n"))
     except KeyboardInterrupt:
         try:
             u.warn(_("interrupted!\n"))
         except IOError, inst:
             if inst.errno == errno.EPIPE:
                 if u.debugflag:
                     u.warn(_("\nbroken pipe\n"))
             else:
                 raise
     except IOError, inst:
         if hasattr(inst, "code"):
             u.warn(_("abort: %s\n") % inst)
         elif hasattr(inst, "reason"):
             u.warn(_("abort: error: %s\n") % inst.reason[1])
         elif hasattr(inst, "args") and inst[0] == errno.EPIPE:
             if u.debugflag:
                 u.warn(_("broken pipe\n"))
         elif getattr(inst, "strerror", None):
             if getattr(inst, "filename", None):
                 u.warn(_("abort: %s - %s\n") % (inst.strerror, inst.filename))
             else:
                 u.warn(_("abort: %s\n") % inst.strerror)
         else:
             raise
     except OSError, inst:
         if hasattr(inst, "filename"):
             u.warn(_("abort: %s: %s\n") % (inst.strerror, inst.filename))
         else:
             u.warn(_("abort: %s\n") % inst.strerror)
     except util.Abort, inst:
         u.warn(_('abort: '), inst.args[0] % inst.args[1:], '\n')
         sys.exit(1)
     except TypeError, inst:
         # was this an argument error?
         tb = traceback.extract_tb(sys.exc_info()[2])
         if len(tb) > 2: # no
             raise
         u.debug(inst, "\n")
         u.warn(_("%s: invalid arguments\n") % cmd)
         help_(u, cmd)
     except UnknownCommand, inst:
         u.warn(_("hg: unknown command '%s'\n") % inst.args[0])
         help_(u, 'shortlist')
     except SystemExit:
         # don't catch this in the catch-all below
         raise
     except:
         u.warn(_("** unknown exception encountered, details follow\n"))
         u.warn(_("** report bug details to [e-mail]\n"))
         raise
 
     sys.exit(-1)

mercurial/localrepo.py

@@ -1,1472 +1,1736 b''
 # localrepo.py - read/write repository class for mercurial
 #
 # Copyright 2005 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.
 
 import struct, os, util
 import filelog, manifest, changelog, dirstate, repo
 from node import *
 from i18n import gettext as _
 from demandload import *
 demandload(globals(), "re lock transaction tempfile stat mdiff errno")
 
 class localrepository:
     def __init__(self, ui, path=None, create=0):
         if not path:
             p = os.getcwd()
             while not os.path.isdir(os.path.join(p, ".hg")):
                 oldp = p
                 p = os.path.dirname(p)
                 if p == oldp: raise repo.RepoError(_("no repo found"))
             path = p
         self.path = os.path.join(path, ".hg")
 
         if not create and not os.path.isdir(self.path):
             raise repo.RepoError(_("repository %s not found") % self.path)
 
         self.root = os.path.abspath(path)
         self.ui = ui
         self.opener = util.opener(self.path)
         self.wopener = util.opener(self.root)
         self.manifest = manifest.manifest(self.opener)
         self.changelog = changelog.changelog(self.opener)
         self.tagscache = None
         self.nodetagscache = None
         self.encodepats = None
         self.decodepats = None
 
         if create:
             os.mkdir(self.path)
             os.mkdir(self.join("data"))
 
         self.dirstate = dirstate.dirstate(self.opener, ui, self.root)
         try:
             self.ui.readconfig(self.opener("hgrc"))
         except IOError: pass
 
     def hook(self, name, **args):
         s = self.ui.config("hooks", name)
         if s:
             self.ui.note(_("running hook %s: %s\n") % (name, s))
             old = {}
             for k, v in args.items():
                 k = k.upper()
                 old[k] = os.environ.get(k, None)
                 os.environ[k] = v
 
             # Hooks run in the repository root
             olddir = os.getcwd()
             os.chdir(self.root)
             r = os.system(s)
             os.chdir(olddir)
 
             for k, v in old.items():
                 if v != None:
                     os.environ[k] = v
                 else:
                     del os.environ[k]
 
             if r:
                 self.ui.warn(_("abort: %s hook failed with status %d!\n") %
                              (name, r))
                 return False
         return True
 
     def tags(self):
         '''return a mapping of tag to node'''
         if not self.tagscache:
             self.tagscache = {}
             def addtag(self, k, n):
                 try:
                     bin_n = bin(n)
                 except TypeError:
                     bin_n = ''
                 self.tagscache[k.strip()] = bin_n
 
             try:
                 # read each head of the tags file, ending with the tip
                 # and add each tag found to the map, with "newer" ones
                 # taking precedence
                 fl = self.file(".hgtags")
                 h = fl.heads()
                 h.reverse()
                 for r in h:
                     for l in fl.read(r).splitlines():
                         if l:
                             n, k = l.split(" ", 1)
                             addtag(self, k, n)
             except KeyError:
                 pass
 
             try:
                 f = self.opener("localtags")
                 for l in f:
                     n, k = l.split(" ", 1)
                     addtag(self, k, n)
             except IOError:
                 pass
 
             self.tagscache['tip'] = self.changelog.tip()
 
         return self.tagscache
 
     def tagslist(self):
         '''return a list of tags ordered by revision'''
         l = []
         for t, n in self.tags().items():
             try:
                 r = self.changelog.rev(n)
             except:
                 r = -2 # sort to the beginning of the list if unknown
             l.append((r,t,n))
         l.sort()
         return [(t,n) for r,t,n in l]
 
     def nodetags(self, node):
         '''return the tags associated with a node'''
         if not self.nodetagscache:
             self.nodetagscache = {}
             for t,n in self.tags().items():
                 self.nodetagscache.setdefault(n,[]).append(t)
         return self.nodetagscache.get(node, [])
 
     def lookup(self, key):
         try:
             return self.tags()[key]
         except KeyError:
             try:
                 return self.changelog.lookup(key)
             except:
                 raise repo.RepoError(_("unknown revision '%s'") % key)
 
     def dev(self):
         return os.stat(self.path).st_dev
 
     def local(self):
         return True
 
     def join(self, f):
         return os.path.join(self.path, f)
 
     def wjoin(self, f):
         return os.path.join(self.root, f)
 
     def file(self, f):
         if f[0] == '/': f = f[1:]
         return filelog.filelog(self.opener, f)
 
     def getcwd(self):
         return self.dirstate.getcwd()
 
     def wfile(self, f, mode='r'):
         return self.wopener(f, mode)
 
     def wread(self, filename):
         if self.encodepats == None:
             l = []
             for pat, cmd in self.ui.configitems("encode"):
                 mf = util.matcher("", "/", [pat], [], [])[1]
                 l.append((mf, cmd))
             self.encodepats = l
 
         data = self.wopener(filename, 'r').read()
 
         for mf, cmd in self.encodepats:
             if mf(filename):
                 self.ui.debug(_("filtering %s through %s\n") % (filename, cmd))
                 data = util.filter(data, cmd)
                 break
 
         return data
 
     def wwrite(self, filename, data, fd=None):
         if self.decodepats == None:
             l = []
             for pat, cmd in self.ui.configitems("decode"):
                 mf = util.matcher("", "/", [pat], [], [])[1]
                 l.append((mf, cmd))
             self.decodepats = l
 
         for mf, cmd in self.decodepats:
             if mf(filename):
                 self.ui.debug(_("filtering %s through %s\n") % (filename, cmd))
                 data = util.filter(data, cmd)
                 break
 
         if fd:
             return fd.write(data)
         return self.wopener(filename, 'w').write(data)
 
     def transaction(self):
         # save dirstate for undo
         try:
             ds = self.opener("dirstate").read()
         except IOError:
             ds = ""
         self.opener("journal.dirstate", "w").write(ds)
 
         def after():
             util.rename(self.join("journal"), self.join("undo"))
             util.rename(self.join("journal.dirstate"),
                         self.join("undo.dirstate"))
 
         return transaction.transaction(self.ui.warn, self.opener,
                                        self.join("journal"), after)
 
     def recover(self):
         lock = self.lock()
         if os.path.exists(self.join("journal")):
             self.ui.status(_("rolling back interrupted transaction\n"))
             return transaction.rollback(self.opener, self.join("journal"))
         else:
             self.ui.warn(_("no interrupted transaction available\n"))
 
     def undo(self):
         lock = self.lock()
         if os.path.exists(self.join("undo")):
             self.ui.status(_("rolling back last transaction\n"))
             transaction.rollback(self.opener, self.join("undo"))
             self.dirstate = None
             util.rename(self.join("undo.dirstate"), self.join("dirstate"))
             self.dirstate = dirstate.dirstate(self.opener, self.ui, self.root)
         else:
             self.ui.warn(_("no undo information available\n"))
 
     def lock(self, wait=1):
         try:
             return lock.lock(self.join("lock"), 0)
         except lock.LockHeld, inst:
             if wait:
                 self.ui.warn(_("waiting for lock held by %s\n") % inst.args[0])
                 return lock.lock(self.join("lock"), wait)
             raise inst
 
     def rawcommit(self, files, text, user, date, p1=None, p2=None):
         orig_parent = self.dirstate.parents()[0] or nullid
         p1 = p1 or self.dirstate.parents()[0] or nullid
         p2 = p2 or self.dirstate.parents()[1] or nullid
         c1 = self.changelog.read(p1)
         c2 = self.changelog.read(p2)
         m1 = self.manifest.read(c1[0])
         mf1 = self.manifest.readflags(c1[0])
         m2 = self.manifest.read(c2[0])
         changed = []
 
         if orig_parent == p1:
             update_dirstate = 1
         else:
             update_dirstate = 0
 
         tr = self.transaction()
         mm = m1.copy()
         mfm = mf1.copy()
         linkrev = self.changelog.count()
         for f in files:
             try:
                 t = self.wread(f)
                 tm = util.is_exec(self.wjoin(f), mfm.get(f, False))
                 r = self.file(f)
                 mfm[f] = tm
 
                 fp1 = m1.get(f, nullid)
                 fp2 = m2.get(f, nullid)
 
                 # is the same revision on two branches of a merge?
                 if fp2 == fp1:
                     fp2 = nullid
 
                 if fp2 != nullid:
                     # is one parent an ancestor of the other?
                     fpa = r.ancestor(fp1, fp2)
                     if fpa == fp1:
                         fp1, fp2 = fp2, nullid
                     elif fpa == fp2:
                         fp2 = nullid
 
                     # is the file unmodified from the parent?
                     if t == r.read(fp1):
                         # record the proper existing parent in manifest
                         # no need to add a revision
                         mm[f] = fp1
                         continue
 
                 mm[f] = r.add(t, {}, tr, linkrev, fp1, fp2)
                 changed.append(f)
                 if update_dirstate:
                     self.dirstate.update([f], "n")
             except IOError:
                 try:
                     del mm[f]
                     del mfm[f]
                     if update_dirstate:
                         self.dirstate.forget([f])
                 except:
                     # deleted from p2?
                     pass
 
         mnode = self.manifest.add(mm, mfm, tr, linkrev, c1[0], c2[0])
         user = user or self.ui.username()
         n = self.changelog.add(mnode, changed, text, tr, p1, p2, user, date)
         tr.close()
         if update_dirstate:
             self.dirstate.setparents(n, nullid)
 
     def commit(self, files = None, text = "", user = None, date = None,
                match = util.always, force=False):
         commit = []
         remove = []
         changed = []
 
         if files:
             for f in files:
                 s = self.dirstate.state(f)
                 if s in 'nmai':
                     commit.append(f)
                 elif s == 'r':
                     remove.append(f)
                 else:
                     self.ui.warn(_("%s not tracked!\n") % f)
         else:
             (c, a, d, u) = self.changes(match=match)
             commit = c + a
             remove = d
 
         p1, p2 = self.dirstate.parents()
         c1 = self.changelog.read(p1)
         c2 = self.changelog.read(p2)
         m1 = self.manifest.read(c1[0])
         mf1 = self.manifest.readflags(c1[0])
         m2 = self.manifest.read(c2[0])
 
         if not commit and not remove and not force and p2 == nullid:
             self.ui.status(_("nothing changed\n"))
             return None
 
         if not self.hook("precommit"):
             return None
 
         lock = self.lock()
         tr = self.transaction()
 
         # check in files
         new = {}
         linkrev = self.changelog.count()
         commit.sort()
         for f in commit:
             self.ui.note(f + "\n")
             try:
                 mf1[f] = util.is_exec(self.wjoin(f), mf1.get(f, False))
                 t = self.wread(f)
             except IOError:
                 self.ui.warn(_("trouble committing %s!\n") % f)
                 raise
 
             r = self.file(f)
 
             meta = {}
             cp = self.dirstate.copied(f)
             if cp:
                 meta["copy"] = cp
                 meta["copyrev"] = hex(m1.get(cp, m2.get(cp, nullid)))
                 self.ui.debug(_(" %s: copy %s:%s\n") % (f, cp, meta["copyrev"]))
                 fp1, fp2 = nullid, nullid
             else:
                 fp1 = m1.get(f, nullid)
                 fp2 = m2.get(f, nullid)
 
             # is the same revision on two branches of a merge?
             if fp2 == fp1:
                 fp2 = nullid
 
             if fp2 != nullid:
                 # is one parent an ancestor of the other?
                 fpa = r.ancestor(fp1, fp2)
                 if fpa == fp1:
                     fp1, fp2 = fp2, nullid
                 elif fpa == fp2:
                     fp2 = nullid
 
                 # is the file unmodified from the parent?
                 if not meta and t == r.read(fp1):
                     # record the proper existing parent in manifest
                     # no need to add a revision
                     new[f] = fp1
                     continue
 
             new[f] = r.add(t, meta, tr, linkrev, fp1, fp2)
             # remember what we've added so that we can later calculate
             # the files to pull from a set of changesets
             changed.append(f)
 
         # update manifest
         m1.update(new)
         for f in remove:
             if f in m1:
                 del m1[f]
         mn = self.manifest.add(m1, mf1, tr, linkrev, c1[0], c2[0],
                                (new, remove))
 
         # add changeset
         new = new.keys()
         new.sort()
 
         if not text:
             edittext = ""
             if p2 != nullid:
                 edittext += "HG: branch merge\n"
             edittext += "\n" + "HG: manifest hash %s\n" % hex(mn)
             edittext += "".join(["HG: changed %s\n" % f for f in changed])
             edittext += "".join(["HG: removed %s\n" % f for f in remove])
             if not changed and not remove:
                 edittext += "HG: no files changed\n"
             edittext = self.ui.edit(edittext)
             if not edittext.rstrip():
                 return None
             text = edittext
 
         user = user or self.ui.username()
         n = self.changelog.add(mn, changed, text, tr, p1, p2, user, date)
         tr.close()
 
         self.dirstate.setparents(n)
         self.dirstate.update(new, "n")
         self.dirstate.forget(remove)
 
         if not self.hook("commit", node=hex(n)):
             return None
         return n
 
     def walk(self, node=None, files=[], match=util.always):
         if node:
             for fn in self.manifest.read(self.changelog.read(node)[0]):
                 if match(fn): yield 'm', fn
         else:
             for src, fn in self.dirstate.walk(files, match):
                 yield src, fn
 
     def changes(self, node1 = None, node2 = None, files = [],
                 match = util.always):
         mf2, u = None, []
 
         def fcmp(fn, mf):
             t1 = self.wread(fn)
             t2 = self.file(fn).read(mf.get(fn, nullid))
             return cmp(t1, t2)
 
         def mfmatches(node):
             mf = dict(self.manifest.read(node))
             for fn in mf.keys():
                 if not match(fn):
                     del mf[fn]
             return mf
 
         # are we comparing the working directory?
         if not node2:
             l, c, a, d, u = self.dirstate.changes(files, match)
 
             # are we comparing working dir against its parent?
             if not node1:
                 if l:
                     # do a full compare of any files that might have changed
                     change = self.changelog.read(self.dirstate.parents()[0])
                     mf2 = mfmatches(change[0])
                     for f in l:
                         if fcmp(f, mf2):
                             c.append(f)
 
                 for l in c, a, d, u:
                     l.sort()
 
                 return (c, a, d, u)
 
         # are we comparing working dir against non-tip?
         # generate a pseudo-manifest for the working dir
         if not node2:
             if not mf2:
                 change = self.changelog.read(self.dirstate.parents()[0])
                 mf2 = mfmatches(change[0])
             for f in a + c + l:
                 mf2[f] = ""
             for f in d:
                 if f in mf2: del mf2[f]
         else:
             change = self.changelog.read(node2)
             mf2 = mfmatches(change[0])
 
         # flush lists from dirstate before comparing manifests
         c, a = [], []
 
         change = self.changelog.read(node1)
         mf1 = mfmatches(change[0])
 
         for fn in mf2:
             if mf1.has_key(fn):
                 if mf1[fn] != mf2[fn]:
                     if mf2[fn] != "" or fcmp(fn, mf1):
                         c.append(fn)
                 del mf1[fn]
             else:
                 a.append(fn)
 
         d = mf1.keys()
 
         for l in c, a, d, u:
             l.sort()
 
         return (c, a, d, u)
 
     def add(self, list):
         for f in list:
             p = self.wjoin(f)
             if not os.path.exists(p):
                 self.ui.warn(_("%s does not exist!\n") % f)
             elif not os.path.isfile(p):
                 self.ui.warn(_("%s not added: only files supported currently\n") % f)
             elif self.dirstate.state(f) in 'an':
                 self.ui.warn(_("%s already tracked!\n") % f)
             else:
                 self.dirstate.update([f], "a")
 
     def forget(self, list):
         for f in list:
             if self.dirstate.state(f) not in 'ai':
                 self.ui.warn(_("%s not added!\n") % f)
             else:
                 self.dirstate.forget([f])
 
     def remove(self, list, unlink=False):
         if unlink:
             for f in list:
                 try:
                     util.unlink(self.wjoin(f))
                 except OSError, inst:
                     if inst.errno != errno.ENOENT: raise
         for f in list:
             p = self.wjoin(f)
             if os.path.exists(p):
                 self.ui.warn(_("%s still exists!\n") % f)
             elif self.dirstate.state(f) == 'a':
                 self.ui.warn(_("%s never committed!\n") % f)
                 self.dirstate.forget([f])
             elif f not in self.dirstate:
                 self.ui.warn(_("%s not tracked!\n") % f)
             else:
                 self.dirstate.update([f], "r")
 
     def undelete(self, list):
         p = self.dirstate.parents()[0]
         mn = self.changelog.read(p)[0]
         mf = self.manifest.readflags(mn)
         m = self.manifest.read(mn)
         for f in list:
             if self.dirstate.state(f) not in  "r":
                 self.ui.warn("%s not removed!\n" % f)
             else:
                 t = self.file(f).read(m[f])
                 try:
                     self.wwrite(f, t)
                 except IOError, e:
                     if e.errno != errno.ENOENT:
                         raise
                     os.makedirs(os.path.dirname(self.wjoin(f)))
                     self.wwrite(f, t)
                 util.set_exec(self.wjoin(f), mf[f])
                 self.dirstate.update([f], "n")
 
     def copy(self, source, dest):
         p = self.wjoin(dest)
         if not os.path.exists(p):
             self.ui.warn(_("%s does not exist!\n") % dest)
         elif not os.path.isfile(p):
             self.ui.warn(_("copy failed: %s is not a file\n") % dest)
         else:
             if self.dirstate.state(dest) == '?':
                 self.dirstate.update([dest], "a")
             self.dirstate.copy(source, dest)
 
     def heads(self):
         return self.changelog.heads()
 
     # branchlookup returns a dict giving a list of branches for
     # each head.  A branch is defined as the tag of a node or
     # the branch of the node's parents.  If a node has multiple
     # branch tags, tags are eliminated if they are visible from other
     # branch tags.
     #
     # So, for this graph:  a->b->c->d->e
     #                       \         /
     #                        aa -----/
     # a has tag 2.6.12
     # d has tag 2.6.13
     # e would have branch tags for 2.6.12 and 2.6.13.  Because the node
     # for 2.6.12 can be reached from the node 2.6.13, that is eliminated
     # from the list.
     #
     # It is possible that more than one head will have the same branch tag.
     # callers need to check the result for multiple heads under the same
     # branch tag if that is a problem for them (ie checkout of a specific
     # branch).
     #
     # passing in a specific branch will limit the depth of the search
     # through the parents.  It won't limit the branches returned in the
     # result though.
     def branchlookup(self, heads=None, branch=None):
         if not heads:
             heads = self.heads()
         headt = [ h for h in heads ]
         chlog = self.changelog
         branches = {}
         merges = []
         seenmerge = {}
 
         # traverse the tree once for each head, recording in the branches
         # dict which tags are visible from this head.  The branches
         # dict also records which tags are visible from each tag
         # while we traverse.
         while headt or merges:
             if merges:
                 n, found = merges.pop()
                 visit = [n]
             else:
                 h = headt.pop()
                 visit = [h]
                 found = [h]
                 seen = {}
             while visit:
                 n = visit.pop()
                 if n in seen:
                     continue
                 pp = chlog.parents(n)
                 tags = self.nodetags(n)
                 if tags:
                     for x in tags:
                         if x == 'tip':
                             continue
                         for f in found:
                             branches.setdefault(f, {})[n] = 1
                         branches.setdefault(n, {})[n] = 1
                         break
                     if n not in found:
                         found.append(n)
                     if branch in tags:
                         continue
                 seen[n] = 1
                 if pp[1] != nullid and n not in seenmerge:
                     merges.append((pp[1], [x for x in found]))
                     seenmerge[n] = 1
                 if pp[0] != nullid:
                     visit.append(pp[0])
         # traverse the branches dict, eliminating branch tags from each
         # head that are visible from another branch tag for that head.
         out = {}
         viscache = {}
         for h in heads:
             def visible(node):
                 if node in viscache:
                     return viscache[node]
                 ret = {}
                 visit = [node]
                 while visit:
                     x = visit.pop()
                     if x in viscache:
                         ret.update(viscache[x])
                     elif x not in ret:
                         ret[x] = 1
                         if x in branches:
                             visit[len(visit):] = branches[x].keys()
                 viscache[node] = ret
                 return ret
             if h not in branches:
                 continue
             # O(n^2), but somewhat limited.  This only searches the
             # tags visible from a specific head, not all the tags in the
             # whole repo.
             for b in branches[h]:
                 vis = False
                 for bb in branches[h].keys():
                     if b != bb:
                         if b in visible(bb):
                             vis = True
                             break
                 if not vis:
                     l = out.setdefault(h, [])
                     l[len(l):] = self.nodetags(b)
         return out
 
     def branches(self, nodes):
         if not nodes: nodes = [self.changelog.tip()]
         b = []
         for n in nodes:
             t = n
             while n:
                 p = self.changelog.parents(n)
                 if p[1] != nullid or p[0] == nullid:
                     b.append((t, n, p[0], p[1]))
                     break
                 n = p[0]
         return b
 
     def between(self, pairs):
         r = []
 
         for top, bottom in pairs:
             n, l, i = top, [], 0
             f = 1
 
             while n != bottom:
                 p = self.changelog.parents(n)[0]
                 if i == f:
                     l.append(n)
                     f = f * 2
                 n = p
                 i += 1
 
             r.append(l)
 
         return r
 
-    def newer(self, nodes):
-        m = {}
-        nl = []
-        pm = {}
-        cl = self.changelog
-        t = l = cl.count()
-
-        # find the lowest numbered node
-        for n in nodes:
-            l = min(l, cl.rev(n))
-            m[n] = 1
-
-        for i in xrange(l, t):
-            n = cl.node(i)
-            if n in m: # explicitly listed
-                pm[n] = 1
-                nl.append(n)
-                continue
-            for p in cl.parents(n):
-                if p in pm: # parent listed
-                    pm[n] = 1
-                    nl.append(n)
-                    break
-
-        return nl
-
     def findincoming(self, remote, base=None, heads=None):
         m = self.changelog.nodemap
         search = []
         fetch = {}
         seen = {}
         seenbranch = {}
         if base == None:
             base = {}
 
         # assume we're closer to the tip than the root
         # and start by examining the heads
         self.ui.status(_("searching for changes\n"))
 
         if not heads:
             heads = remote.heads()
 
         unknown = []
         for h in heads:
             if h not in m:
                 unknown.append(h)
             else:
                 base[h] = 1
 
         if not unknown:
             return None
 
         rep = {}
         reqcnt = 0
 
         # search through remote branches
         # a 'branch' here is a linear segment of history, with four parts:
         # head, root, first parent, second parent
         # (a branch always has two parents (or none) by definition)
         unknown = remote.branches(unknown)
         while unknown:
             r = []
             while unknown:
                 n = unknown.pop(0)
                 if n[0] in seen:
                     continue
 
                 self.ui.debug(_("examining %s:%s\n") % (short(n[0]), short(n[1])))
                 if n[0] == nullid:
                     break
                 if n in seenbranch:
                     self.ui.debug(_("branch already found\n"))
                     continue
                 if n[1] and n[1] in m: # do we know the base?
                     self.ui.debug(_("found incomplete branch %s:%s\n")
                                   % (short(n[0]), short(n[1])))
                     search.append(n) # schedule branch range for scanning
                     seenbranch[n] = 1
                 else:
                     if n[1] not in seen and n[1] not in fetch:
                         if n[2] in m and n[3] in m:
                             self.ui.debug(_("found new changeset %s\n") %
                                           short(n[1]))
                             fetch[n[1]] = 1 # earliest unknown
                             base[n[2]] = 1 # latest known
                             continue
 
                     for a in n[2:4]:
                         if a not in rep:
                             r.append(a)
                             rep[a] = 1
 
                 seen[n[0]] = 1
 
             if r:
                 reqcnt += 1
                 self.ui.debug(_("request %d: %s\n") %
                             (reqcnt, " ".join(map(short, r))))
                 for p in range(0, len(r), 10):
                     for b in remote.branches(r[p:p+10]):
                         self.ui.debug(_("received %s:%s\n") %
                                       (short(b[0]), short(b[1])))
                         if b[0] in m:
                             self.ui.debug(_("found base node %s\n") % short(b[0]))
                             base[b[0]] = 1
                         elif b[0] not in seen:
                             unknown.append(b)
 
         # do binary search on the branches we found
         while search:
             n = search.pop(0)
             reqcnt += 1
             l = remote.between([(n[0], n[1])])[0]
             l.append(n[1])
             p = n[0]
             f = 1
             for i in l:
                 self.ui.debug(_("narrowing %d:%d %s\n") % (f, len(l), short(i)))
                 if i in m:
                     if f <= 2:
                         self.ui.debug(_("found new branch changeset %s\n") %
                                           short(p))
                         fetch[p] = 1
                         base[i] = 1
                     else:
                         self.ui.debug(_("narrowed branch search to %s:%s\n")
                                       % (short(p), short(i)))
                         search.append((p, i))
                     break
                 p, f = i, f * 2
 
         # sanity check our fetch list
         for f in fetch.keys():
             if f in m:
                 raise repo.RepoError(_("already have changeset ") + short(f[:4]))
 
         if base.keys() == [nullid]:
             self.ui.warn(_("warning: pulling from an unrelated repository!\n"))
 
         self.ui.note(_("found new changesets starting at ") +
                      " ".join([short(f) for f in fetch]) + "\n")
 
         self.ui.debug(_("%d total queries\n") % reqcnt)
 
         return fetch.keys()
 
     def findoutgoing(self, remote, base=None, heads=None):
         if base == None:
             base = {}
             self.findincoming(remote, base, heads)
 
         self.ui.debug(_("common changesets up to ")
                       + " ".join(map(short, base.keys())) + "\n")
 
         remain = dict.fromkeys(self.changelog.nodemap)
 
         # prune everything remote has from the tree
         del remain[nullid]
         remove = base.keys()
         while remove:
             n = remove.pop(0)
             if n in remain:
                 del remain[n]
                 for p in self.changelog.parents(n):
                     remove.append(p)
 
         # find every node whose parents have been pruned
         subset = []
         for n in remain:
             p1, p2 = self.changelog.parents(n)
             if p1 not in remain and p2 not in remain:
                 subset.append(n)
 
         # this is the set of all roots we have to push
         return subset
 
-    def pull(self, remote):
+    def pull(self, remote, heads = None):
         lock = self.lock()
 
         # if we have an empty repo, fetch everything
         if self.changelog.tip() == nullid:
             self.ui.status(_("requesting all changes\n"))
             fetch = [nullid]
         else:
             fetch = self.findincoming(remote)
 
         if not fetch:
             self.ui.status(_("no changes found\n"))
             return 1
 
+        if heads is None:
         cg = remote.changegroup(fetch)
+        else:
+            cg = remote.changegroupsubset(fetch, heads)
         return self.addchangegroup(cg)
 
     def push(self, remote, force=False):
         lock = remote.lock()
 
         base = {}
         heads = remote.heads()
         inc = self.findincoming(remote, base, heads)
         if not force and inc:
             self.ui.warn(_("abort: unsynced remote changes!\n"))
             self.ui.status(_("(did you forget to sync? use push -f to force)\n"))
             return 1
 
         update = self.findoutgoing(remote, base)
         if not update:
             self.ui.status(_("no changes found\n"))
             return 1
         elif not force:
             if len(heads) < len(self.changelog.heads()):
                 self.ui.warn(_("abort: push creates new remote branches!\n"))
                 self.ui.status(_("(did you forget to merge?"
                                  " use push -f to force)\n"))
                 return 1
 
         cg = self.changegroup(update)
         return remote.addchangegroup(cg)
 
+    def changegroupsubset(self, bases, heads):
+        """This function generates a changegroup consisting of all the nodes
+        that are descendents of any of the bases, and ancestors of any of
+        the heads.
+
+        It is fairly complex as determining which filenodes and which
+        manifest nodes need to be included for the changeset to be complete
+        is non-trivial.
+
+        Another wrinkle is doing the reverse, figuring out which changeset in
+        the changegroup a particular filenode or manifestnode belongs to."""
+
+        # Set up some initial variables
+        # Make it easy to refer to self.changelog
+        cl = self.changelog
+        # msng is short for missing - compute the list of changesets in this
+        # changegroup.
+        msng_cl_lst, bases, heads = cl.nodesbetween(bases, heads)
+        # Some bases may turn out to be superfluous, and some heads may be
+        # too.  nodesbetween will return the minimal set of bases and heads
+        # necessary to re-create the changegroup.
+
+        # Known heads are the list of heads that it is assumed the recipient
+        # of this changegroup will know about.
+        knownheads = {}
+        # We assume that all parents of bases are known heads.
+        for n in bases:
+            for p in cl.parents(n):
+                if p != nullid:
+                    knownheads[p] = 1
+        knownheads = knownheads.keys()
+        if knownheads:
+            # Now that we know what heads are known, we can compute which
+            # changesets are known.  The recipient must know about all
+            # changesets required to reach the known heads from the null
+            # changeset.
+            has_cl_set, junk, junk = cl.nodesbetween(None, knownheads)
+            junk = None
+            # Transform the list into an ersatz set.
+            has_cl_set = dict.fromkeys(has_cl_set)
+        else:
+            # If there were no known heads, the recipient cannot be assumed to
+            # know about any changesets.
+            has_cl_set = {}
+
+        # Make it easy to refer to self.manifest
+        mnfst = self.manifest
+        # We don't know which manifests are missing yet
+        msng_mnfst_set = {}
+        # Nor do we know which filenodes are missing.
+        msng_filenode_set = {}
+
+        junk = mnfst.index[mnfst.count() - 1] # Get around a bug in lazyindex
+        junk = None
+
+        # A changeset always belongs to itself, so the changenode lookup
+        # function for a changenode is identity.
+        def identity(x):
+            return x
+
+        # A function generating function.  Sets up an environment for the
+        # inner function.
+        def cmp_by_rev_func(revlog):
+            # Compare two nodes by their revision number in the environment's
+            # revision history.  Since the revision number both represents the
+            # most efficient order to read the nodes in, and represents a
+            # topological sorting of the nodes, this function is often useful.
+            def cmp_by_rev(a, b):
+                return cmp(revlog.rev(a), revlog.rev(b))
+            return cmp_by_rev
+
+        # If we determine that a particular file or manifest node must be a
+        # node that the recipient of the changegroup will already have, we can
+        # also assume the recipient will have all the parents.  This function
+        # prunes them from the set of missing nodes.
+        def prune_parents(revlog, hasset, msngset):
+            haslst = hasset.keys()
+            haslst.sort(cmp_by_rev_func(revlog))
+            for node in haslst:
+                parentlst = [p for p in revlog.parents(node) if p != nullid]
+                while parentlst:
+                    n = parentlst.pop()
+                    if n not in hasset:
+                        hasset[n] = 1
+                        p = [p for p in revlog.parents(n) if p != nullid]
+                        parentlst.extend(p)
+            for n in hasset:
+                msngset.pop(n, None)
+
+        # This is a function generating function used to set up an environment
+        # for the inner function to execute in.
+        def manifest_and_file_collector(changedfileset):
+            # This is an information gathering function that gathers
+            # information from each changeset node that goes out as part of
+            # the changegroup.  The information gathered is a list of which
+            # manifest nodes are potentially required (the recipient may
+            # already have them) and total list of all files which were
+            # changed in any changeset in the changegroup.
+            #
+            # We also remember the first changenode we saw any manifest
+            # referenced by so we can later determine which changenode 'owns'
+            # the manifest.
+            def collect_manifests_and_files(clnode):
+                c = cl.read(clnode)
+                for f in c[3]:
+                    # This is to make sure we only have one instance of each
+                    # filename string for each filename.
+                    changedfileset.setdefault(f, f)
+                msng_mnfst_set.setdefault(c[0], clnode)
+            return collect_manifests_and_files
+
+        # Figure out which manifest nodes (of the ones we think might be part
+        # of the changegroup) the recipient must know about and remove them
+        # from the changegroup.
+        def prune_manifests():
+            has_mnfst_set = {}
+            for n in msng_mnfst_set:
+                # If a 'missing' manifest thinks it belongs to a changenode
+                # the recipient is assumed to have, obviously the recipient
+                # must have that manifest.
+                linknode = cl.node(mnfst.linkrev(n))
+                if linknode in has_cl_set:
+                    has_mnfst_set[n] = 1
+            prune_parents(mnfst, has_mnfst_set, msng_mnfst_set)
+
+        # Use the information collected in collect_manifests_and_files to say
+        # which changenode any manifestnode belongs to.
+        def lookup_manifest_link(mnfstnode):
+            return msng_mnfst_set[mnfstnode]
+
+        # A function generating function that sets up the initial environment
+        # the inner function.
+        def filenode_collector(changedfiles):
+            next_rev = [0]
+            # This gathers information from each manifestnode included in the
+            # changegroup about which filenodes the manifest node references
+            # so we can include those in the changegroup too.
+            #
+            # It also remembers which changenode each filenode belongs to.  It
+            # does this by assuming the a filenode belongs to the changenode
+            # the first manifest that references it belongs to.
+            def collect_msng_filenodes(mnfstnode):
+                r = mnfst.rev(mnfstnode)
+                if r == next_rev[0]:
+                    # If the last rev we looked at was the one just previous,
+                    # we only need to see a diff.
+                    delta = mdiff.patchtext(mnfst.delta(mnfstnode))
+                    # For each line in the delta
+                    for dline in delta.splitlines():
+                        # get the filename and filenode for that line
+                        f, fnode = dline.split('\0')
+                        fnode = bin(fnode[:40])
+                        f = changedfiles.get(f, None)
+                        # And if the file is in the list of files we care
+                        # about.
+                        if f is not None:
+                            # Get the changenode this manifest belongs to
+                            clnode = msng_mnfst_set[mnfstnode]
+                            # Create the set of filenodes for the file if
+                            # there isn't one already.
+                            ndset = msng_filenode_set.setdefault(f, {})
+                            # And set the filenode's changelog node to the
+                            # manifest's if it hasn't been set already.
+                            ndset.setdefault(fnode, clnode)
+                else:
+                    # Otherwise we need a full manifest.
+                    m = mnfst.read(mnfstnode)
+                    # For every file in we care about.
+                    for f in changedfiles:
+                        fnode = m.get(f, None)
+                        # If it's in the manifest
+                        if fnode is not None:
+                            # See comments above.
+                            clnode = msng_mnfst_set[mnfstnode]
+                            ndset = msng_filenode_set.setdefault(f, {})
+                            ndset.setdefault(fnode, clnode)
+                # Remember the revision we hope to see next.
+                next_rev[0] = r + 1
+            return collect_msng_filenodes
+
+        # We have a list of filenodes we think we need for a file, lets remove
+        # all those we now the recipient must have.
+        def prune_filenodes(f, filerevlog):
+            msngset = msng_filenode_set[f]
+            hasset = {}
+            # If a 'missing' filenode thinks it belongs to a changenode we
+            # assume the recipient must have, then the recipient must have
+            # that filenode.
+            for n in msngset:
+                clnode = cl.node(filerevlog.linkrev(n))
+                if clnode in has_cl_set:
+                    hasset[n] = 1
+            prune_parents(filerevlog, hasset, msngset)
+
+        # A function generator function that sets up the a context for the
+        # inner function.
+        def lookup_filenode_link_func(fname):
+            msngset = msng_filenode_set[fname]
+            # Lookup the changenode the filenode belongs to.
+            def lookup_filenode_link(fnode):
+                return msngset[fnode]
+            return lookup_filenode_link
+
+        # Now that we have all theses utility functions to help out and
+        # logically divide up the task, generate the group.
+        def gengroup():
+            # The set of changed files starts empty.
+            changedfiles = {}
+            # Create a changenode group generator that will call our functions
+            # back to lookup the owning changenode and collect information.
+            group = cl.group(msng_cl_lst, identity,
+                             manifest_and_file_collector(changedfiles))
+            for chnk in group:
+                yield chnk
+
+            # The list of manifests has been collected by the generator
+            # calling our functions back.
+            prune_manifests()
+            msng_mnfst_lst = msng_mnfst_set.keys()
+            # Sort the manifestnodes by revision number.
+            msng_mnfst_lst.sort(cmp_by_rev_func(mnfst))
+            # Create a generator for the manifestnodes that calls our lookup
+            # and data collection functions back.
+            group = mnfst.group(msng_mnfst_lst, lookup_manifest_link,
+                                filenode_collector(changedfiles))
+            for chnk in group:
+                yield chnk
+
+            # These are no longer needed, dereference and toss the memory for
+            # them.
+            msng_mnfst_lst = None
+            msng_mnfst_set.clear()
+
+            changedfiles = changedfiles.keys()
+            changedfiles.sort()
+            # Go through all our files in order sorted by name.
+            for fname in changedfiles:
+                filerevlog = self.file(fname)
+                # Toss out the filenodes that the recipient isn't really
+                # missing.
+                prune_filenodes(fname, filerevlog)
+                msng_filenode_lst = msng_filenode_set[fname].keys()
+                # If any filenodes are left, generate the group for them,
+                # otherwise don't bother.
+                if len(msng_filenode_lst) > 0:
+                    yield struct.pack(">l", len(fname) + 4) + fname
+                    # Sort the filenodes by their revision #
+                    msng_filenode_lst.sort(cmp_by_rev_func(filerevlog))
+                    # Create a group generator and only pass in a changenode
+                    # lookup function as we need to collect no information
+                    # from filenodes.
+                    group = filerevlog.group(msng_filenode_lst,
+                                             lookup_filenode_link_func(fname))
+                    for chnk in group:
+                        yield chnk
+                # Don't need this anymore, toss it to free memory.
+                del msng_filenode_set[fname]
+            # Signal that no more groups are left.
+            yield struct.pack(">l", 0)
+
+        return util.chunkbuffer(gengroup())
+
     def changegroup(self, basenodes):
-        genread = util.chunkbuffer
+        """Generate a changegroup of all nodes that we have that a recipient
+        doesn't.
+
+        This is much easier than the previous function as we can assume that
+        the recipient has any changenode we aren't sending them."""
+        cl = self.changelog
+        nodes = cl.nodesbetween(basenodes, None)[0]
+        revset = dict.fromkeys([cl.rev(n) for n in nodes])
+
+        def identity(x):
+            return x
+
+        def gennodelst(revlog):
+            for r in xrange(0, revlog.count()):
+                n = revlog.node(r)
+                if revlog.linkrev(n) in revset:
+                    yield n
+
+        def changed_file_collector(changedfileset):
+            def collect_changed_files(clnode):
+                c = cl.read(clnode)
+                for fname in c[3]:
+                    changedfileset[fname] = 1
+            return collect_changed_files
+
+        def lookuprevlink_func(revlog):
+            def lookuprevlink(n):
+                return cl.node(revlog.linkrev(n))
+            return lookuprevlink
 
         def gengroup():
-            nodes = self.newer(basenodes)
+            # construct a list of all changed files
+            changedfiles = {}
 
-            # construct the link map
-            linkmap = {}
-            for n in nodes:
-                linkmap[self.changelog.rev(n)] = n
+            for chnk in cl.group(nodes, identity,
+                                 changed_file_collector(changedfiles)):
+                yield chnk
+            changedfiles = changedfiles.keys()
+            changedfiles.sort()
 
-            # construct a list of all changed files
-            changed = {}
-            for n in nodes:
-                c = self.changelog.read(n)
-                for f in c[3]:
-                    changed[f] = 1
-            changed = changed.keys()
-            changed.sort()
+            mnfst = self.manifest
+            nodeiter = gennodelst(mnfst)
+            for chnk in mnfst.group(nodeiter, lookuprevlink_func(mnfst)):
+                yield chnk
 
-            # the changegroup is changesets + manifests + all file revs
-            revs = [ self.changelog.rev(n) for n in nodes ]
-
-            for y in self.changelog.group(linkmap): yield y
-            for y in self.manifest.group(linkmap): yield y
-            for f in changed:
-                yield struct.pack(">l", len(f) + 4) + f
-                g = self.file(f).group(linkmap)
-                for y in g:
-                    yield y
+            for fname in changedfiles:
+                filerevlog = self.file(fname)
+                nodeiter = gennodelst(filerevlog)
+                nodeiter = list(nodeiter)
+                if nodeiter:
+                    yield struct.pack(">l", len(fname) + 4) + fname
+                    lookup = lookuprevlink_func(filerevlog)
+                    for chnk in filerevlog.group(nodeiter, lookup):
+                        yield chnk
 
             yield struct.pack(">l", 0)
 
-        return genread(gengroup())
+        return util.chunkbuffer(gengroup())
 
     def addchangegroup(self, source):
 
         def getchunk():
             d = source.read(4)
             if not d: return ""
             l = struct.unpack(">l", d)[0]
             if l <= 4: return ""
             d = source.read(l - 4)
             if len(d) < l - 4:
                 raise repo.RepoError(_("premature EOF reading chunk"
                                        " (got %d bytes, expected %d)")
                                      % (len(d), l - 4))
             return d
 
         def getgroup():
             while 1:
                 c = getchunk()
                 if not c: break
                 yield c
 
         def csmap(x):
             self.ui.debug(_("add changeset %s\n") % short(x))
             return self.changelog.count()
 
         def revmap(x):
             return self.changelog.rev(x)
 
         if not source: return
         changesets = files = revisions = 0
 
         tr = self.transaction()
 
         oldheads = len(self.changelog.heads())
 
         # pull off the changeset group
         self.ui.status(_("adding changesets\n"))
         co = self.changelog.tip()
         cn = self.changelog.addgroup(getgroup(), csmap, tr, 1) # unique
         cnr, cor = map(self.changelog.rev, (cn, co))
         if cn == nullid:
             cnr = cor
         changesets = cnr - cor
 
         # pull off the manifest group
         self.ui.status(_("adding manifests\n"))
         mm = self.manifest.tip()
         mo = self.manifest.addgroup(getgroup(), revmap, tr)
 
         # process the files
         self.ui.status(_("adding file changes\n"))
         while 1:
             f = getchunk()
             if not f: break
             self.ui.debug(_("adding %s revisions\n") % f)
             fl = self.file(f)
             o = fl.count()
             n = fl.addgroup(getgroup(), revmap, tr)
             revisions += fl.count() - o
             files += 1
 
         newheads = len(self.changelog.heads())
         heads = ""
         if oldheads and newheads > oldheads:
             heads = _(" (+%d heads)") % (newheads - oldheads)
 
         self.ui.status(_("added %d changesets"
                          " with %d changes to %d files%s\n")
                          % (changesets, revisions, files, heads))
 
         tr.close()
 
         if changesets > 0:
             if not self.hook("changegroup",
                               node=hex(self.changelog.node(cor+1))):
                 self.ui.warn(_("abort: changegroup hook returned failure!\n"))
                 return 1
 
             for i in range(cor + 1, cnr + 1):
                 self.hook("commit", node=hex(self.changelog.node(i)))
 
         return
 
     def update(self, node, allow=False, force=False, choose=None,
                moddirstate=True):
         pl = self.dirstate.parents()
         if not force and pl[1] != nullid:
             self.ui.warn(_("aborting: outstanding uncommitted merges\n"))
             return 1
 
         p1, p2 = pl[0], node
         pa = self.changelog.ancestor(p1, p2)
         m1n = self.changelog.read(p1)[0]
         m2n = self.changelog.read(p2)[0]
         man = self.manifest.ancestor(m1n, m2n)
         m1 = self.manifest.read(m1n)
         mf1 = self.manifest.readflags(m1n)
         m2 = self.manifest.read(m2n)
         mf2 = self.manifest.readflags(m2n)
         ma = self.manifest.read(man)
         mfa = self.manifest.readflags(man)
 
         (c, a, d, u) = self.changes()
 
         # is this a jump, or a merge?  i.e. is there a linear path
         # from p1 to p2?
         linear_path = (pa == p1 or pa == p2)
 
         # resolve the manifest to determine which files
         # we care about merging
         self.ui.note(_("resolving manifests\n"))
         self.ui.debug(_(" force %s allow %s moddirstate %s linear %s\n") %
                       (force, allow, moddirstate, linear_path))
         self.ui.debug(_(" ancestor %s local %s remote %s\n") %
                       (short(man), short(m1n), short(m2n)))
 
         merge = {}
         get = {}
         remove = []
 
         # construct a working dir manifest
         mw = m1.copy()
         mfw = mf1.copy()
         umap = dict.fromkeys(u)
 
         for f in a + c + u:
             mw[f] = ""
             mfw[f] = util.is_exec(self.wjoin(f), mfw.get(f, False))
 
         for f in d:
             if f in mw: del mw[f]
 
             # If we're jumping between revisions (as opposed to merging),
             # and if neither the working directory nor the target rev has
             # the file, then we need to remove it from the dirstate, to
             # prevent the dirstate from listing the file when it is no
             # longer in the manifest.
             if moddirstate and linear_path and f not in m2:
                 self.dirstate.forget((f,))
 
         # Compare manifests
         for f, n in mw.iteritems():
             if choose and not choose(f): continue
             if f in m2:
                 s = 0
 
                 # is the wfile new since m1, and match m2?
                 if f not in m1:
                     t1 = self.wread(f)
                     t2 = self.file(f).read(m2[f])
                     if cmp(t1, t2) == 0:
                         n = m2[f]
                     del t1, t2
 
                 # are files different?
                 if n != m2[f]:
                     a = ma.get(f, nullid)
                     # are both different from the ancestor?
                     if n != a and m2[f] != a:
                         self.ui.debug(_(" %s versions differ, resolve\n") % f)
                         # merge executable bits
                         # "if we changed or they changed, change in merge"
                         a, b, c = mfa.get(f, 0), mfw[f], mf2[f]
                         mode = ((a^b) | (a^c)) ^ a
                         merge[f] = (m1.get(f, nullid), m2[f], mode)
                         s = 1
                     # are we clobbering?
                     # is remote's version newer?
                     # or are we going back in time?
                     elif force or m2[f] != a or (p2 == pa and mw[f] == m1[f]):
                         self.ui.debug(_(" remote %s is newer, get\n") % f)
                         get[f] = m2[f]
                         s = 1
                 elif f in umap:
                     # this unknown file is the same as the checkout
                     get[f] = m2[f]
 
                 if not s and mfw[f] != mf2[f]:
                     if force:
                         self.ui.debug(_(" updating permissions for %s\n") % f)
                         util.set_exec(self.wjoin(f), mf2[f])
                     else:
                         a, b, c = mfa.get(f, 0), mfw[f], mf2[f]
                         mode = ((a^b) | (a^c)) ^ a
                         if mode != b:
                             self.ui.debug(_(" updating permissions for %s\n") % f)
                             util.set_exec(self.wjoin(f), mode)
                 del m2[f]
             elif f in ma:
                 if n != ma[f]:
                     r = _("d")
                     if not force and (linear_path or allow):
                         r = self.ui.prompt(
                             (_(" local changed %s which remote deleted\n") % f) +
                              _("(k)eep or (d)elete?"), _("[kd]"), _("k"))
                     if r == _("d"):
                         remove.append(f)
                 else:
                     self.ui.debug(_("other deleted %s\n") % f)
                     remove.append(f) # other deleted it
             else:
                 # file is created on branch or in working directory
                 if force and f not in umap:
                     self.ui.debug(_("remote deleted %s, clobbering\n") % f)
                     remove.append(f)
                 elif n == m1.get(f, nullid): # same as parent
                     if p2 == pa: # going backwards?
                         self.ui.debug(_("remote deleted %s\n") % f)
                         remove.append(f)
                     else:
                         self.ui.debug(_("local modified %s, keeping\n") % f)
                 else:
                     self.ui.debug(_("working dir created %s, keeping\n") % f)
 
         for f, n in m2.iteritems():
             if choose and not choose(f): continue
             if f[0] == "/": continue
             if f in ma and n != ma[f]:
                 r = _("k")
                 if not force and (linear_path or allow):
                     r = self.ui.prompt(
                         (_("remote changed %s which local deleted\n") % f) +
                          _("(k)eep or (d)elete?"), _("[kd]"), _("k"))
                 if r == _("k"): get[f] = n
             elif f not in ma:
                 self.ui.debug(_("remote created %s\n") % f)
                 get[f] = n
             else:
                 if force or p2 == pa: # going backwards?
                     self.ui.debug(_("local deleted %s, recreating\n") % f)
                     get[f] = n
                 else:
                     self.ui.debug(_("local deleted %s\n") % f)
 
         del mw, m1, m2, ma
 
         if force:
             for f in merge:
                 get[f] = merge[f][1]
             merge = {}
 
         if linear_path or force:
             # we don't need to do any magic, just jump to the new rev
             branch_merge = False
             p1, p2 = p2, nullid
         else:
             if not allow:
                 self.ui.status(_("this update spans a branch"
                                  " affecting the following files:\n"))
                 fl = merge.keys() + get.keys()
                 fl.sort()
                 for f in fl:
                     cf = ""
                     if f in merge: cf = _(" (resolve)")
                     self.ui.status(" %s%s\n" % (f, cf))
                 self.ui.warn(_("aborting update spanning branches!\n"))
                 self.ui.status(_("(use update -m to merge across branches"
                                  " or -C to lose changes)\n"))
                 return 1
             branch_merge = True
 
         if moddirstate:
             self.dirstate.setparents(p1, p2)
 
         # get the files we don't need to change
         files = get.keys()
         files.sort()
         for f in files:
             if f[0] == "/": continue
             self.ui.note(_("getting %s\n") % f)
             t = self.file(f).read(get[f])
             try:
                 self.wwrite(f, t)
             except IOError, e:
                 if e.errno != errno.ENOENT:
                     raise
                 os.makedirs(os.path.dirname(self.wjoin(f)))
                 self.wwrite(f, t)
             util.set_exec(self.wjoin(f), mf2[f])
             if moddirstate:
                 if branch_merge:
                     self.dirstate.update([f], 'n', st_mtime=-1)
                 else:
                     self.dirstate.update([f], 'n')
 
         # merge the tricky bits
         files = merge.keys()
         files.sort()
         for f in files:
             self.ui.status(_("merging %s\n") % f)
             my, other, flag = merge[f]
             self.merge3(f, my, other)
             util.set_exec(self.wjoin(f), flag)
             if moddirstate:
                 if branch_merge:
                     # We've done a branch merge, mark this file as merged
                     # so that we properly record the merger later
                     self.dirstate.update([f], 'm')
                 else:
                     # We've update-merged a locally modified file, so
                     # we set the dirstate to emulate a normal checkout
                     # of that file some time in the past. Thus our
                     # merge will appear as a normal local file
                     # modification.
                     f_len = len(self.file(f).read(other))
                     self.dirstate.update([f], 'n', st_size=f_len, st_mtime=-1)
 
         remove.sort()
         for f in remove:
             self.ui.note(_("removing %s\n") % f)
             try:
                 util.unlink(self.wjoin(f))
             except OSError, inst:
                 if inst.errno != errno.ENOENT:
                     self.ui.warn(_("update failed to remove %s: %s!\n") %
                                  (f, inst.strerror))
         if moddirstate:
             if branch_merge:
                 self.dirstate.update(remove, 'r')
             else:
                 self.dirstate.forget(remove)
 
     def merge3(self, fn, my, other):
         """perform a 3-way merge in the working directory"""
 
         def temp(prefix, node):
             pre = "%s~%s." % (os.path.basename(fn), prefix)
             (fd, name) = tempfile.mkstemp("", pre)
             f = os.fdopen(fd, "wb")
             self.wwrite(fn, fl.read(node), f)
             f.close()
             return name
 
         fl = self.file(fn)
         base = fl.ancestor(my, other)
         a = self.wjoin(fn)
         b = temp("base", base)
         c = temp("other", other)
 
         self.ui.note(_("resolving %s\n") % fn)
         self.ui.debug(_("file %s: my %s other %s ancestor %s\n") %
                               (fn, short(my), short(other), short(base)))
 
         cmd = (os.environ.get("HGMERGE") or self.ui.config("ui", "merge")
                or "hgmerge")
         r = os.system('%s "%s" "%s" "%s"' % (cmd, a, b, c))
         if r:
             self.ui.warn(_("merging %s failed!\n") % fn)
 
         os.unlink(b)
         os.unlink(c)
 
     def verify(self):
         filelinkrevs = {}
         filenodes = {}
         changesets = revisions = files = 0
         errors = [0]
         neededmanifests = {}
 
         def err(msg):
             self.ui.warn(msg + "\n")
             errors[0] += 1
 
         seen = {}
         self.ui.status(_("checking changesets\n"))
         for i in range(self.changelog.count()):
             changesets += 1
             n = self.changelog.node(i)
             l = self.changelog.linkrev(n)
             if l != i:
                 err(_("incorrect link (%d) for changeset revision %d") %(l, i))
             if n in seen:
                 err(_("duplicate changeset at revision %d") % i)
             seen[n] = 1
 
             for p in self.changelog.parents(n):
                 if p not in self.changelog.nodemap:
                     err(_("changeset %s has unknown parent %s") %
                                  (short(n), short(p)))
             try:
                 changes = self.changelog.read(n)
             except Exception, inst:
                 err(_("unpacking changeset %s: %s") % (short(n), inst))
 
             neededmanifests[changes[0]] = n
 
             for f in changes[3]:
                 filelinkrevs.setdefault(f, []).append(i)
 
         seen = {}
         self.ui.status(_("checking manifests\n"))
         for i in range(self.manifest.count()):
             n = self.manifest.node(i)
             l = self.manifest.linkrev(n)
 
             if l < 0 or l >= self.changelog.count():
                 err(_("bad manifest link (%d) at revision %d") % (l, i))
 
             if n in neededmanifests:
                 del neededmanifests[n]
 
             if n in seen:
                 err(_("duplicate manifest at revision %d") % i)
 
             seen[n] = 1
 
             for p in self.manifest.parents(n):
                 if p not in self.manifest.nodemap:
                     err(_("manifest %s has unknown parent %s") %
                         (short(n), short(p)))
 
             try:
                 delta = mdiff.patchtext(self.manifest.delta(n))
             except KeyboardInterrupt:
                 self.ui.warn(_("interrupted"))
                 raise
             except Exception, inst:
                 err(_("unpacking manifest %s: %s") % (short(n), inst))
 
             ff = [ l.split('\0') for l in delta.splitlines() ]
             for f, fn in ff:
                 filenodes.setdefault(f, {})[bin(fn[:40])] = 1
 
         self.ui.status(_("crosschecking files in changesets and manifests\n"))
 
         for m,c in neededmanifests.items():
             err(_("Changeset %s refers to unknown manifest %s") %
                 (short(m), short(c)))
         del neededmanifests
 
         for f in filenodes:
             if f not in filelinkrevs:
                 err(_("file %s in manifest but not in changesets") % f)
 
         for f in filelinkrevs:
             if f not in filenodes:
                 err(_("file %s in changeset but not in manifest") % f)
 
         self.ui.status(_("checking files\n"))
         ff = filenodes.keys()
         ff.sort()
         for f in ff:
             if f == "/dev/null": continue
             files += 1
             fl = self.file(f)
             nodes = { nullid: 1 }
             seen = {}
             for i in range(fl.count()):
                 revisions += 1
                 n = fl.node(i)
 
                 if n in seen:
                     err(_("%s: duplicate revision %d") % (f, i))
                 if n not in filenodes[f]:
                     err(_("%s: %d:%s not in manifests") % (f, i, short(n)))
                 else:
                     del filenodes[f][n]
 
                 flr = fl.linkrev(n)
                 if flr not in filelinkrevs[f]:
                     err(_("%s:%s points to unexpected changeset %d")
                             % (f, short(n), flr))
                 else:
                     filelinkrevs[f].remove(flr)
 
                 # verify contents
                 try:
                     t = fl.read(n)
                 except Exception, inst:
                     err(_("unpacking file %s %s: %s") % (f, short(n), inst))
 
                 # verify parents
                 (p1, p2) = fl.parents(n)
                 if p1 not in nodes:
                     err(_("file %s:%s unknown parent 1 %s") %
                         (f, short(n), short(p1)))
                 if p2 not in nodes:
                     err(_("file %s:%s unknown parent 2 %s") %
                             (f, short(n), short(p1)))
                 nodes[n] = 1
 
             # cross-check
             for node in filenodes[f]:
                 err(_("node %s in manifests not in %s") % (hex(node), f))
 
         self.ui.status(_("%d files, %d changesets, %d total revisions\n") %
                        (files, changesets, revisions))
 
         if errors[0]:
             self.ui.warn(_("%d integrity errors encountered!\n") % errors[0])
             return 1

mercurial/revlog.py

@@ -1,677 +1,825 b''
 """
 revlog.py - storage back-end for mercurial
 
 This provides efficient delta storage with O(1) retrieve and append
 and O(changes) merge between branches
 
 Copyright 2005 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 node import *
 from i18n import gettext as _
 from demandload import demandload
 demandload(globals(), "binascii errno heapq mdiff sha struct zlib")
 
 def hash(text, p1, p2):
     """generate a hash from the given text and its parent hashes
 
     This hash combines both the current file contents and its history
     in a manner that makes it easy to distinguish nodes with the same
     content in the revision graph.
     """
     l = [p1, p2]
     l.sort()
     s = sha.new(l[0])
     s.update(l[1])
     s.update(text)
     return s.digest()
 
 def compress(text):
     """ generate a possibly-compressed representation of text """
     if not text: return text
     if len(text) < 44:
         if text[0] == '\0': return text
         return 'u' + text
     bin = zlib.compress(text)
     if len(bin) > len(text):
         if text[0] == '\0': return text
         return 'u' + text
     return bin
 
 def decompress(bin):
     """ decompress the given input """
     if not bin: return bin
     t = bin[0]
     if t == '\0': return bin
     if t == 'x': return zlib.decompress(bin)
     if t == 'u': return bin[1:]
     raise RevlogError(_("unknown compression type %s") % t)
 
 indexformat = ">4l20s20s20s"
 
 class lazyparser:
     """
     this class avoids the need to parse the entirety of large indices
 
     By default we parse and load 1000 entries at a time.
 
     If no position is specified, we load the whole index, and replace
     the lazy objects in revlog with the underlying objects for
     efficiency in cases where we look at most of the nodes.
     """
     def __init__(self, data, revlog):
         self.data = data
         self.s = struct.calcsize(indexformat)
         self.l = len(data)/self.s
         self.index = [None] * self.l
         self.map = {nullid: -1}
         self.all = 0
         self.revlog = revlog
 
     def load(self, pos=None):
         if self.all: return
         if pos is not None:
             block = pos / 1000
             i = block * 1000
             end = min(self.l, i + 1000)
         else:
             self.all = 1
             i = 0
             end = self.l
             self.revlog.index = self.index
             self.revlog.nodemap = self.map
 
         while i < end:
             d = self.data[i * self.s: (i + 1) * self.s]
             e = struct.unpack(indexformat, d)
             self.index[i] = e
             self.map[e[6]] = i
             i += 1
 
 class lazyindex:
     """a lazy version of the index array"""
     def __init__(self, parser):
         self.p = parser
     def __len__(self):
         return len(self.p.index)
     def load(self, pos):
         if pos < 0:
             pos += len(self.p.index)
         self.p.load(pos)
         return self.p.index[pos]
     def __getitem__(self, pos):
         return self.p.index[pos] or self.load(pos)
     def append(self, e):
         self.p.index.append(e)
 
 class lazymap:
     """a lazy version of the node map"""
     def __init__(self, parser):
         self.p = parser
     def load(self, key):
         if self.p.all: return
         n = self.p.data.find(key)
         if n < 0:
             raise KeyError(key)
         pos = n / self.p.s
         self.p.load(pos)
     def __contains__(self, key):
         self.p.load()
         return key in self.p.map
     def __iter__(self):
         yield nullid
         for i in xrange(self.p.l):
             try:
                 yield self.p.index[i][6]
             except:
                 self.p.load(i)
                 yield self.p.index[i][6]
     def __getitem__(self, key):
         try:
             return self.p.map[key]
         except KeyError:
             try:
                 self.load(key)
                 return self.p.map[key]
             except KeyError:
                 raise KeyError("node " + hex(key))
     def __setitem__(self, key, val):
         self.p.map[key] = val
 
 class RevlogError(Exception): pass
 
 class revlog:
     """
     the underlying revision storage object
 
     A revlog consists of two parts, an index and the revision data.
 
     The index is a file with a fixed record size containing
     information on each revision, includings its nodeid (hash), the
     nodeids of its parents, the position and offset of its data within
     the data file, and the revision it's based on. Finally, each entry
     contains a linkrev entry that can serve as a pointer to external
     data.
 
     The revision data itself is a linear collection of data chunks.
     Each chunk represents a revision and is usually represented as a
     delta against the previous chunk. To bound lookup time, runs of
     deltas are limited to about 2 times the length of the original
     version data. This makes retrieval of a version proportional to
     its size, or O(1) relative to the number of revisions.
 
     Both pieces of the revlog are written to in an append-only
     fashion, which means we never need to rewrite a file to insert or
     remove data, and can use some simple techniques to avoid the need
     for locking while reading.
     """
     def __init__(self, opener, indexfile, datafile):
         """
         create a revlog object
 
         opener is a function that abstracts the file opening operation
         and can be used to implement COW semantics or the like.
         """
         self.indexfile = indexfile
         self.datafile = datafile
         self.opener = opener
         self.cache = None
 
         try:
             i = self.opener(self.indexfile).read()
         except IOError, inst:
             if inst.errno != errno.ENOENT:
                 raise
             i = ""
 
         if len(i) > 10000:
             # big index, let's parse it on demand
             parser = lazyparser(i, self)
             self.index = lazyindex(parser)
             self.nodemap = lazymap(parser)
         else:
             s = struct.calcsize(indexformat)
             l = len(i) / s
             self.index = [None] * l
             m = [None] * l
 
             n = 0
             for f in xrange(0, len(i), s):
                 # offset, size, base, linkrev, p1, p2, nodeid
                 e = struct.unpack(indexformat, i[f:f + s])
                 m[n] = (e[6], n)
                 self.index[n] = e
                 n += 1
 
             self.nodemap = dict(m)
             self.nodemap[nullid] = -1
 
     def tip(self): return self.node(len(self.index) - 1)
     def count(self): return len(self.index)
     def node(self, rev): return (rev < 0) and nullid or self.index[rev][6]
     def rev(self, node):
         try:
             return self.nodemap[node]
         except KeyError:
             raise RevlogError(_('%s: no node %s') % (self.indexfile, hex(node)))
     def linkrev(self, node): return self.index[self.rev(node)][3]
     def parents(self, node):
         if node == nullid: return (nullid, nullid)
         return self.index[self.rev(node)][4:6]
 
     def start(self, rev): return self.index[rev][0]
     def length(self, rev): return self.index[rev][1]
     def end(self, rev): return self.start(rev) + self.length(rev)
     def base(self, rev): return self.index[rev][2]
 
     def reachable(self, rev, stop=None):
         reachable = {}
         visit = [rev]
         reachable[rev] = 1
         if stop:
             stopn = self.rev(stop)
         else:
             stopn = 0
         while visit:
             n = visit.pop(0)
             if n == stop:
                 continue
             if n == nullid:
                 continue
             for p in self.parents(n):
                 if self.rev(p) < stopn:
                     continue
                 if p not in reachable:
                     reachable[p] = 1
                     visit.append(p)
         return reachable
 
+    def nodesbetween(self, roots=None, heads=None):
+        """Return a tuple containing three elements. Elements 1 and 2 contain
+        a final list bases and heads after all the unreachable ones have been
+        pruned.  Element 0 contains a topologically sorted list of all
+
+        nodes that satisfy these constraints:
+        1. All nodes must be descended from a node in roots (the nodes on
+           roots are considered descended from themselves).
+        2. All nodes must also be ancestors of a node in heads (the nodes in
+           heads are considered to be their own ancestors).
+
+        If roots is unspecified, nullid is assumed as the only root.
+        If heads is unspecified, it is taken to be the output of the
+        heads method (i.e. a list of all nodes in the repository that
+        have no children)."""
+        nonodes = ([], [], [])
+        if roots is not None:
+            roots = list(roots)
+            if not roots:
+                return nonodes
+            lowestrev = min([self.rev(n) for n in roots])
+        else:
+            roots = [nullid] # Everybody's a descendent of nullid
+            lowestrev = -1
+        if (lowestrev == -1) and (heads is None):
+            # We want _all_ the nodes!
+            return ([self.node(r) for r in xrange(0, self.count())],
+                    [nullid], list(self.heads()))
+        if heads is None:
+            # All nodes are ancestors, so the latest ancestor is the last
+            # node.
+            highestrev = self.count() - 1
+            # Set ancestors to None to signal that every node is an ancestor.
+            ancestors = None
+            # Set heads to an empty dictionary for later discovery of heads
+            heads = {}
+        else:
+            heads = list(heads)
+            if not heads:
+                return nonodes
+            ancestors = {}
+            # Start at the top and keep marking parents until we're done.
+            nodestotag = heads[:]
+            # Turn heads into a dictionary so we can remove 'fake' heads.
+            # Also, later we will be using it to filter out the heads we can't
+            # find from roots.
+            heads = dict.fromkeys(heads, 0)
+            # Remember where the top was so we can use it as a limit later.
+            highestrev = max([self.rev(n) for n in nodestotag])
+            while nodestotag:
+                # grab a node to tag
+                n = nodestotag.pop()
+                # Never tag nullid
+                if n == nullid:
+                    continue
+                # A node's revision number represents its place in a
+                # topologically sorted list of nodes.
+                r = self.rev(n)
+                if r >= lowestrev:
+                    if n not in ancestors:
+                        # If we are possibly a descendent of one of the roots
+                        # and we haven't already been marked as an ancestor
+                        ancestors[n] = 1 # Mark as ancestor
+                        # Add non-nullid parents to list of nodes to tag.
+                        nodestotag.extend([p for p in self.parents(n) if
+                                           p != nullid])
+                    elif n in heads: # We've seen it before, is it a fake head?
+                        # So it is, real heads should not be the ancestors of
+                        # any other heads.
+                        heads.pop(n)
+            if not ancestors:
+                return nonodes
+            # Now that we have our set of ancestors, we want to remove any
+            # roots that are not ancestors.
+
+            # If one of the roots was nullid, everything is included anyway.
+            if lowestrev > -1:
+                # But, since we weren't, let's recompute the lowest rev to not
+                # include roots that aren't ancestors.
+
+                # Filter out roots that aren't ancestors of heads
+                roots = [n for n in roots if n in ancestors]
+                # Recompute the lowest revision
+                if roots:
+                    lowestrev = min([self.rev(n) for n in roots])
+                else:
+                    # No more roots?  Return empty list
+                    return nonodes
+            else:
+                # We are descending from nullid, and don't need to care about
+                # any other roots.
+                lowestrev = -1
+                roots = [nullid]
+        # Transform our roots list into a 'set' (i.e. a dictionary where the
+        # values don't matter.
+        descendents = dict.fromkeys(roots, 1)
+        # Also, keep the original roots so we can filter out roots that aren't
+        # 'real' roots (i.e. are descended from other roots).
+        roots = descendents.copy()
+        # Our topologically sorted list of output nodes.
+        orderedout = []
+        # Don't start at nullid since we don't want nullid in our output list,
+        # and if nullid shows up in descedents, empty parents will look like
+        # they're descendents.
+        for r in xrange(max(lowestrev, 0), highestrev + 1):
+            n = self.node(r)
+            isdescendent = False
+            if lowestrev == -1:  # Everybody is a descendent of nullid
+                isdescendent = True
+            elif n in descendents:
+                # n is already a descendent
+                isdescendent = True
+                # This check only needs to be done here because all the roots
+                # will start being marked is descendents before the loop.
+                if n in roots:
+                    # If n was a root, check if it's a 'real' root.
+                    p = tuple(self.parents(n))
+                    # If any of its parents are descendents, it's not a root.
+                    if (p[0] in descendents) or (p[1] in descendents):
+                        roots.pop(n)
+            else:
+                p = tuple(self.parents(n))
+                # A node is a descendent if either of its parents are
+                # descendents.  (We seeded the dependents list with the roots
+                # up there, remember?)
+                if (p[0] in descendents) or (p[1] in descendents):
+                    descendents[n] = 1
+                    isdescendent = True
+            if isdescendent and ((ancestors is None) or (n in ancestors)):
+                # Only include nodes that are both descendents and ancestors.
+                orderedout.append(n)
+                if (ancestors is not None) and (n in heads):
+                    # We're trying to figure out which heads are reachable
+                    # from roots.
+                    # Mark this head as having been reached
+                    heads[n] = 1
+                elif ancestors is None:
+                    # Otherwise, we're trying to discover the heads.
+                    # Assume this is a head because if it isn't, the next step
+                    # will eventually remove it.
+                    heads[n] = 1
+                    # But, obviously its parents aren't.
+                    for p in self.parents(n):
+                        heads.pop(p, None)
+        heads = [n for n in heads.iterkeys() if heads[n] != 0]
+        roots = roots.keys()
+        assert orderedout
+        assert roots
+        assert heads
+        return (orderedout, roots, heads)
+
     def heads(self, stop=None):
         """return the list of all nodes that have no children"""
         p = {}
         h = []
         stoprev = 0
         if stop and stop in self.nodemap:
             stoprev = self.rev(stop)
 
         for r in range(self.count() - 1, -1, -1):
             n = self.node(r)
             if n not in p:
                 h.append(n)
             if n == stop:
                 break
             if r < stoprev:
                 break
             for pn in self.parents(n):
                 p[pn] = 1
         return h
 
     def children(self, node):
         """find the children of a given node"""
         c = []
         p = self.rev(node)
         for r in range(p + 1, self.count()):
             n = self.node(r)
             for pn in self.parents(n):
                 if pn == node:
                     c.append(n)
                     continue
                 elif pn == nullid:
                     continue
         return c
 
     def lookup(self, id):
         """locate a node based on revision number or subset of hex nodeid"""
         try:
             rev = int(id)
             if str(rev) != id: raise ValueError
             if rev < 0: rev = self.count() + rev
             if rev < 0 or rev >= self.count(): raise ValueError
             return self.node(rev)
         except (ValueError, OverflowError):
             c = []
             for n in self.nodemap:
                 if hex(n).startswith(id):
                     c.append(n)
             if len(c) > 1: raise RevlogError(_("Ambiguous identifier"))
             if len(c) < 1: raise RevlogError(_("No match found"))
             return c[0]
 
         return None
 
     def diff(self, a, b):
         """return a delta between two revisions"""
         return mdiff.textdiff(a, b)
 
     def patches(self, t, pl):
         """apply a list of patches to a string"""
         return mdiff.patches(t, pl)
 
     def delta(self, node):
         """return or calculate a delta between a node and its predecessor"""
         r = self.rev(node)
         b = self.base(r)
         if r == b:
             return self.diff(self.revision(self.node(r - 1)),
                              self.revision(node))
         else:
             f = self.opener(self.datafile)
             f.seek(self.start(r))
             data = f.read(self.length(r))
         return decompress(data)
 
     def revision(self, node):
         """return an uncompressed revision of a given"""
         if node == nullid: return ""
         if self.cache and self.cache[0] == node: return self.cache[2]
 
         # look up what we need to read
         text = None
         rev = self.rev(node)
         start, length, base, link, p1, p2, node = self.index[rev]
         end = start + length
         if base != rev: start = self.start(base)
 
         # do we have useful data cached?
         if self.cache and self.cache[1] >= base and self.cache[1] < rev:
             base = self.cache[1]
             start = self.start(base + 1)
             text = self.cache[2]
             last = 0
 
         f = self.opener(self.datafile)
         f.seek(start)
         data = f.read(end - start)
 
         if text is None:
             last = self.length(base)
             text = decompress(data[:last])
 
         bins = []
         for r in xrange(base + 1, rev + 1):
             s = self.length(r)
             bins.append(decompress(data[last:last + s]))
             last = last + s
 
         text = mdiff.patches(text, bins)
 
         if node != hash(text, p1, p2):
             raise RevlogError(_("integrity check failed on %s:%d")
                           % (self.datafile, rev))
 
         self.cache = (node, rev, text)
         return text
 
     def addrevision(self, text, transaction, link, p1=None, p2=None, d=None):
         """add a revision to the log
 
         text - the revision data to add
         transaction - the transaction object used for rollback
         link - the linkrev data to add
         p1, p2 - the parent nodeids of the revision
         d - an optional precomputed delta
         """
         if text is None: text = ""
         if p1 is None: p1 = self.tip()
         if p2 is None: p2 = nullid
 
         node = hash(text, p1, p2)
 
         if node in self.nodemap:
             return node
 
         n = self.count()
         t = n - 1
 
         if n:
             base = self.base(t)
             start = self.start(base)
             end = self.end(t)
             if not d:
                 prev = self.revision(self.tip())
                 d = self.diff(prev, text)
             data = compress(d)
             dist = end - start + len(data)
 
         # full versions are inserted when the needed deltas
         # become comparable to the uncompressed text
         if not n or dist > len(text) * 2:
             data = compress(text)
             base = n
         else:
             base = self.base(t)
 
         offset = 0
         if t >= 0:
             offset = self.end(t)
 
         e = (offset, len(data), base, link, p1, p2, node)
 
         self.index.append(e)
         self.nodemap[node] = n
         entry = struct.pack(indexformat, *e)
 
         transaction.add(self.datafile, e[0])
         self.opener(self.datafile, "a").write(data)
         transaction.add(self.indexfile, n * len(entry))
         self.opener(self.indexfile, "a").write(entry)
 
         self.cache = (node, n, text)
         return node
 
     def ancestor(self, a, b):
         """calculate the least common ancestor of nodes a and b"""
         # calculate the distance of every node from root
         dist = {nullid: 0}
         for i in xrange(self.count()):
             n = self.node(i)
             p1, p2 = self.parents(n)
             dist[n] = max(dist[p1], dist[p2]) + 1
 
         # traverse ancestors in order of decreasing distance from root
         def ancestors(node):
             # we store negative distances because heap returns smallest member
             h = [(-dist[node], node)]
             seen = {}
             earliest = self.count()
             while h:
                 d, n = heapq.heappop(h)
                 if n not in seen:
                     seen[n] = 1
                     r = self.rev(n)
                     yield (-d, n)
                     for p in self.parents(n):
                         heapq.heappush(h, (-dist[p], p))
 
         def generations(node):
             sg, s = None, {}
             for g,n in ancestors(node):
                 if g != sg:
                     if sg:
                         yield sg, s
                     sg, s = g, {n:1}
                 else:
                     s[n] = 1
             yield sg, s
 
         x = generations(a)
         y = generations(b)
         gx = x.next()
         gy = y.next()
 
         # increment each ancestor list until it is closer to root than
         # the other, or they match
         while 1:
             #print "ancestor gen %s %s" % (gx[0], gy[0])
             if gx[0] == gy[0]:
                 # find the intersection
                 i = [ n for n in gx[1] if n in gy[1] ]
                 if i:
                     return i[0]
                 else:
                     #print "next"
                     gy = y.next()
                     gx = x.next()
             elif gx[0] < gy[0]:
                 #print "next y"
                 gy = y.next()
             else:
                 #print "next x"
                 gx = x.next()
 
-    def group(self, linkmap):
+    def group(self, nodelist, lookup, infocollect = None):
         """calculate a delta group
 
         Given a list of changeset revs, return a set of deltas and
         metadata corresponding to nodes. the first delta is
         parent(nodes[0]) -> nodes[0] the receiver is guaranteed to
         have this parent as it has all history before these
         changesets. parent is parent[0]
         """
-        revs = []
-        needed = {}
-
-        # find file nodes/revs that match changeset revs
-        for i in xrange(0, self.count()):
-            if self.index[i][3] in linkmap:
-                revs.append(i)
-                needed[i] = 1
+        revs = [self.rev(n) for n in nodelist]
+        needed = dict.fromkeys(revs, 1)
 
         # if we don't have any revisions touched by these changesets, bail
         if not revs:
             yield struct.pack(">l", 0)
             return
 
         # add the parent of the first rev
         p = self.parents(self.node(revs[0]))[0]
         revs.insert(0, self.rev(p))
 
         # for each delta that isn't contiguous in the log, we need to
         # reconstruct the base, reconstruct the result, and then
         # calculate the delta. We also need to do this where we've
         # stored a full version and not a delta
         for i in xrange(0, len(revs) - 1):
             a, b = revs[i], revs[i + 1]
             if a + 1 != b or self.base(b) == b:
                 for j in xrange(self.base(a), a + 1):
                     needed[j] = 1
                 for j in xrange(self.base(b), b + 1):
                     needed[j] = 1
 
         # calculate spans to retrieve from datafile
         needed = needed.keys()
         needed.sort()
         spans = []
         oo = -1
         ol = 0
         for n in needed:
             if n < 0: continue
             o = self.start(n)
             l = self.length(n)
             if oo + ol == o: # can we merge with the previous?
                 nl = spans[-1][2]
                 nl.append((n, l))
                 ol += l
                 spans[-1] = (oo, ol, nl)
             else:
                 oo = o
                 ol = l
                 spans.append((oo, ol, [(n, l)]))
 
         # read spans in, divide up chunks
         chunks = {}
         for span in spans:
             # we reopen the file for each span to make http happy for now
             f = self.opener(self.datafile)
             f.seek(span[0])
             data = f.read(span[1])
 
             # divide up the span
             pos = 0
             for r, l in span[2]:
                 chunks[r] = decompress(data[pos: pos + l])
                 pos += l
 
         # helper to reconstruct intermediate versions
         def construct(text, base, rev):
             bins = [chunks[r] for r in xrange(base + 1, rev + 1)]
             return mdiff.patches(text, bins)
 
         # build deltas
         deltas = []
         for d in xrange(0, len(revs) - 1):
             a, b = revs[d], revs[d + 1]
             n = self.node(b)
 
+            if infocollect is not None:
+                infocollect(n)
+
             # do we need to construct a new delta?
             if a + 1 != b or self.base(b) == b:
                 if a >= 0:
                     base = self.base(a)
                     ta = chunks[self.base(a)]
                     ta = construct(ta, base, a)
                 else:
                     ta = ""
 
                 base = self.base(b)
                 if a > base:
                     base = a
                     tb = ta
                 else:
                     tb = chunks[self.base(b)]
                 tb = construct(tb, base, b)
                 d = self.diff(ta, tb)
             else:
                 d = chunks[b]
 
             p = self.parents(n)
-            meta = n + p[0] + p[1] + linkmap[self.linkrev(n)]
+            meta = n + p[0] + p[1] + lookup(n)
             l = struct.pack(">l", len(meta) + len(d) + 4)
             yield l
             yield meta
             yield d
 
         yield struct.pack(">l", 0)
 
     def addgroup(self, revs, linkmapper, transaction, unique=0):
         """
         add a delta group
 
         given a set of deltas, add them to the revision log. the
         first delta is against its parent, which should be in our
         log, the rest are against the previous delta.
         """
 
         #track the base of the current delta log
         r = self.count()
         t = r - 1
         node = nullid
 
         base = prev = -1
         start = end = measure = 0
         if r:
             start = self.start(self.base(t))
             end = self.end(t)
             measure = self.length(self.base(t))
             base = self.base(t)
             prev = self.tip()
 
         transaction.add(self.datafile, end)
         transaction.add(self.indexfile, r * struct.calcsize(indexformat))
         dfh = self.opener(self.datafile, "a")
         ifh = self.opener(self.indexfile, "a")
 
         # loop through our set of deltas
         chain = None
         for chunk in revs:
             node, p1, p2, cs = struct.unpack("20s20s20s20s", chunk[:80])
             link = linkmapper(cs)
             if node in self.nodemap:
                 # this can happen if two branches make the same change
                 # if unique:
                 #    raise RevlogError(_("already have %s") % hex(node[:4]))
                 chain = node
                 continue
             delta = chunk[80:]
 
             if not chain:
                 # retrieve the parent revision of the delta chain
                 chain = p1
                 if not chain in self.nodemap:
                     raise RevlogError(_("unknown base %s") % short(chain[:4]))
 
             # full versions are inserted when the needed deltas become
             # comparable to the uncompressed text or when the previous
             # version is not the one we have a delta against. We use
             # the size of the previous full rev as a proxy for the
             # current size.
 
             if chain == prev:
                 cdelta = compress(delta)
 
             if chain != prev or (end - start + len(cdelta)) > measure * 2:
                 # flush our writes here so we can read it in revision
                 dfh.flush()
                 ifh.flush()
                 text = self.revision(chain)
                 text = self.patches(text, [delta])
                 chk = self.addrevision(text, transaction, link, p1, p2)
                 if chk != node:
                     raise RevlogError(_("consistency error adding group"))
                 measure = len(text)
             else:
                 e = (end, len(cdelta), self.base(t), link, p1, p2, node)
                 self.index.append(e)
                 self.nodemap[node] = r
                 dfh.write(cdelta)
                 ifh.write(struct.pack(indexformat, *e))
 
             t, r, chain, prev = r, r + 1, node, node
             start = self.start(self.base(t))
             end = self.end(t)
 
         dfh.close()
         ifh.close()
         return node