Refactor diffrevs/diffdir into changes

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Refactor diffrevs/diffdir into changes

Add dirstate.changes to replace most of diffdir
Add localrepository.changes to replace diffrevs/diffdir

This code can now efficiently check for changes in single files, and
often without consulting the manifest. This should eventually make 'hg
diff Makefile' in a large project much faster.

This also fixes a bug where 'hg diff -r tip' failed to account for
files that had been added but not committed yet.

manifest hash: 20fde5d4b4cee49a76bcfe50f2dacf58b1f2258b
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.0 (GNU/Linux)

iD8DBQFCxMxpywK+sNU5EO8RAhzOAJ9VLQJoC+hiRYQtTSPbDhXBEJfQZwCgpDx9
GAwQ9jZHNsgXckBfXNCkJV8=
=hMuc
-----END PGP SIGNATURE-----

tests/test-diffdir

@@ -0,0 +1,12 b''
+#!/bin/sh
+
+hg init
+touch a
+hg add a
+hg ci -t "a" -u test -d "0 0"
+
+echo 123 > b
+hg add b
+hg diff | sed "s/\(\(---\|+++\).*\)\t.*/\1/"
+
+hg diff -r tip | sed "s/\(\(---\|+++\).*\)\t.*/\1/"

tests/test-diffdir.out

@@ -0,0 +1,10 b''
+diff -r 3903775176ed b
+--- /dev/null
++++ b/b
+@@ -0,0 +1,1 @@
++123
+diff -r 3903775176ed b
+--- /dev/null
++++ b/b
+@@ -0,0 +1,1 @@
++123

mercurial/commands.py

@@ -1,917 +1,917 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.
 
 import os, re, sys, signal
 import fancyopts, ui, hg, util
 from demandload import *
 demandload(globals(), "mdiff time hgweb traceback random signal errno version")
 
 class UnknownCommand(Exception): pass
 
 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 relfilter(repo, files):
     if os.getcwd() != repo.root:
         p = os.getcwd()[len(repo.root) + 1: ]
         return filterfiles([util.pconvert(p)], files)
     return files
 
 def relpath(repo, args):
     if os.getcwd() != repo.root:
         p = os.getcwd()[len(repo.root) + 1: ]
         return [ util.pconvert(os.path.normpath(os.path.join(p, x))) for x in args ]
     return args
 
 def dodiff(ui, repo, path, files = None, node1 = None, node2 = None):
     def date(c):
         return time.asctime(time.gmtime(float(c[2].split(' ')[0])))
 
     if node2:
         change = repo.changelog.read(node2)
         mmap2 = repo.manifest.read(change[0])
-        (c, a, d) = repo.diffrevs(node1, node2)
+        (c, a, d, u) = repo.changes(node1, node2)
         def read(f): return repo.file(f).read(mmap2[f])
         date2 = date(change)
     else:
         date2 = time.asctime()
-        (c, a, d, u) = repo.diffdir(path, node1)
+        (c, a, d, u) = repo.changes(None, node1, path)
         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 hg.hex or hg.short
         r = [hexfunc(node) for node in [node1, node2] if node]
 
     change = repo.changelog.read(node1)
     mmap = repo.manifest.read(change[0])
     date1 = date(change)
 
     if files:
         c, a, d = map(lambda x: filterfiles(files, x), (c, a, d))
 
     for f in c:
         to = None
         if f in mmap:
             to = repo.file(f).read(mmap[f])
         tn = read(f)
         sys.stdout.write(mdiff.unidiff(to, date1, tn, date2, f, r))
     for f in a:
         to = None
         tn = read(f)
         sys.stdout.write(mdiff.unidiff(to, date1, tn, date2, f, r))
     for f in d:
         to = repo.file(f).read(mmap[f])
         tn = None
         sys.stdout.write(mdiff.unidiff(to, date1, tn, date2, f, r))
 
 def show_changeset(ui, repo, rev=0, changenode=None, filelog=None):
     """show a single changeset or file revision"""
     changelog = repo.changelog
     if filelog:
         log = filelog
         filerev = rev
         node = filenode = filelog.node(filerev)
         changerev = filelog.linkrev(filenode)
         changenode = changenode or changelog.node(changerev)
     else:
         log = changelog
         changerev = rev
         if changenode is None:
             changenode = changelog.node(changerev)
         elif not changerev:
             rev = changerev = changelog.rev(changenode)
         node = changenode
 
     if ui.quiet:
         ui.write("%d:%s\n" % (rev, hg.hex(node)))
         return
 
     changes = changelog.read(changenode)
 
     parents = [(log.rev(parent), hg.hex(parent))
                for parent in log.parents(node)
                if ui.debugflag or parent != hg.nullid]
     if not ui.debugflag and len(parents) == 1 and parents[0][0] == rev-1:
         parents = []
 
     if filelog:
         ui.write("revision:    %d:%s\n" % (filerev, hg.hex(filenode)))
         for parent in parents:
             ui.write("parent:      %d:%s\n" % parent)
         ui.status("changeset:   %d:%s\n" % (changerev, hg.hex(changenode)))
     else:
         ui.write("changeset:   %d:%s\n" % (changerev, hg.hex(changenode)))
         for tag in repo.nodetags(changenode):
             ui.status("tag:         %s\n" % tag)
         for parent in parents:
             ui.write("parent:      %d:%s\n" % parent)
         ui.note("manifest:    %d:%s\n" % (repo.manifest.rev(changes[0]),
                                           hg.hex(changes[0])))
     ui.status("user:        %s\n" % changes[1])
     ui.status("date:        %s\n" % time.asctime(
         time.localtime(float(changes[2].split(' ')[0]))))
     if ui.debugflag:
-        files = repo.diffrevs(changelog.parents(changenode)[0], changenode)
+        files = repo.changes(changelog.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")
         else:
             ui.status("summary:     %s\n" % description.splitlines()[0])
     ui.status("\n")
 
 def show_version(ui):
     """output version and copyright information"""
     ui.write("Mercurial 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):
     '''show help for a given command or all commands'''
     if cmd:
         try:
             i = find(cmd)
             ui.write("%s\n\n" % i[2])
 
             if i[1]:
                 for s, l, d, c in i[1]:
                     opt=' '
                     if s: opt = opt + '-' + s + ' '
                     if l: opt = opt + '--' + l + ' '
                     if d: opt = opt + '(' + str(d) + ')'
                     ui.write(opt, "\n")
                     if c: ui.write('   %s\n' % c)
                 ui.write("\n")
 
             ui.write(i[0].__doc__, "\n")
         except UnknownCommand:
             ui.warn("hg: unknown command %s\n" % cmd)
         sys.exit(0)
     else:
         if not ui.quiet:
             show_version(ui)
             ui.write('\n')
         ui.write('hg commands:\n\n')
 
         h = {}
         for c, e in table.items():
             f = c.split("|")[0]
             if f.startswith("debug"):
                 continue
             d = ""
             if e[0].__doc__:
                 d = e[0].__doc__.splitlines(0)[0].rstrip()
             h[f] = d
 
         fns = h.keys()
         fns.sort()
         m = max(map(len, fns))
         for f in fns:
             ui.write(' %-*s   %s\n' % (m, f, h[f]))
 
 # Commands start here, listed alphabetically
 
 def add(ui, repo, file, *files):
     '''add the specified files on the next commit'''
     repo.add(relpath(repo, (file,) + files))
 
 def addremove(ui, repo, *files):
     """add all new files, delete all missing files"""
     if files:
         files = relpath(repo, files)
         d = []
         u = []
         for f in files:
             p = repo.wjoin(f)
             s = repo.dirstate.state(f)
             isfile = os.path.isfile(p)
             if s != 'r' and not isfile:
                 d.append(f)
             elif s not in 'nmai' and isfile:
                 u.append(f)
     else:
-        (c, a, d, u) = repo.diffdir(repo.root)
+        (c, a, d, u) = repo.changes(None, None)
     repo.add(u)
     repo.remove(d)
 
 def annotate(u, repo, file, *files, **ops):
     """show changeset information per file line"""
     def getnode(rev):
         return hg.short(repo.changelog.node(rev))
 
     def getname(rev):
         try:
             return bcache[rev]
         except KeyError:
             cl = repo.changelog.read(repo.changelog.node(rev))
             name = cl[1]
             f = name.find('@')
             if f >= 0:
                 name = name[:f]
             f = name.find('<')
             if f >= 0:
                 name = name[f+1:]
             bcache[rev] = name
             return name
 
     bcache = {}
     opmap = [['user', getname], ['number', str], ['changeset', getnode]]
     if not ops['user'] and not ops['changeset']:
         ops['number'] = 1
 
     node = repo.dirstate.parents()[0]
     if ops['revision']:
         node = repo.changelog.lookup(ops['revision'])
     change = repo.changelog.read(node)
     mmap = repo.manifest.read(change[0])
     for f in relpath(repo, (file,) + files):
         lines = repo.file(f).annotate(mmap[f])
         pieces = []
 
         for o, f in opmap:
             if ops[o]:
                 l = [ f(n) for n,t in lines ]
                 m = max(map(len, l))
                 pieces.append([ "%*s" % (m, x) for x in l])
 
         for p,l in zip(zip(*pieces), lines):
             u.write(" ".join(p) + ": " + l[1])
 
 def cat(ui, repo, file, rev = []):
     """output the latest or given revision of a file"""
     r = repo.file(relpath(repo, [file])[0])
     n = r.tip()
     if rev: n = r.lookup(rev)
     sys.stdout.write(r.read(n))
 
 def clone(ui, source, dest = None, **opts):
     """make a copy of an existing repository"""
     source = ui.expandpath(source)
 
     if dest is None:
         dest = os.path.basename(os.path.normpath(source))
 
     if os.path.exists(dest):
         ui.warn("abort: destination '%s' already exists\n" % dest)
         return 1
 
     class dircleanup:
         def __init__(self, dir):
             self.dir = dir
             os.mkdir(dir)
         def close(self):
             self.dir = None
         def __del__(self):
             if self.dir:
                 import shutil
                 shutil.rmtree(self.dir, True)
 
     d = dircleanup(dest)
 
     link = 0
     if not source.startswith("http://"):
         d1 = os.stat(dest).st_dev
         d2 = os.stat(source).st_dev
         if d1 == d2: link = 1
 
     if link:
         ui.note("copying by hardlink\n")
         util.system("cp -al '%s'/.hg '%s'/.hg" % (source, dest))
         try:
             os.remove(os.path.join(dest, ".hg", "dirstate"))
         except: pass
 
         repo = hg.repository(ui, dest)
 
     else:
         repo = hg.repository(ui, dest, create=1)
         other = hg.repository(ui, source)
         fetch = repo.findincoming(other)
         if fetch:
             cg = other.changegroup(fetch)
             repo.addchangegroup(cg)
 
     f = repo.opener("hgrc", "w")
     f.write("[paths]\n")
     f.write("default = %s\n" % source)
 
     if not opts['noupdate']:
         update(ui, repo)
 
     d.close()
 
 def commit(ui, repo, *files, **opts):
     """commit the specified files or all outstanding changes"""
     text = opts['text']
     if not text and opts['logfile']:
         try: text = open(opts['logfile']).read()
         except IOError: pass
 
     if opts['addremove']:
         addremove(ui, repo, *files)
     repo.commit(relpath(repo, files), text, opts['user'], opts['date'])
 
 def copy(ui, repo, source, dest):
     """mark a file as copied or renamed for the next commit"""
     return repo.copy(*relpath(repo, (source, dest)))
 
 def debugcheckdirstate(ui, repo):
     parent1, parent2 = repo.dirstate.parents()
     dc = repo.dirstate.dup()
     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:
             print "%s in state %s, but not listed in manifest1" % (f, state)
             errors += 1
         if state in "a" and f in m1:
             print "%s in state %s, but also listed in manifest1" % (f, state)
             errors += 1
         if state in "m" and f not in m1 and f not in m2:
             print "%s in state %s, but not listed in either manifest" % (f, state)
             errors += 1
     for f in m1:
         state = repo.dirstate.state(f)
         if state not in "nrm":
             print "%s in manifest1, but listed as state %s" % (f, state)
             errors += 1
     if errors:
         print ".hg/dirstate inconsistent with current parent's manifest, aborting"
         sys.exit(1)
 
 def debugdumpdirstate(ui, repo):
     dc = repo.dirstate.dup()
     keys = dc.keys()
     keys.sort()
     for file in keys:
         print "%s => %c" % (file, dc[file][0])
 
 def debugindex(ui, file):
     r = hg.revlog(hg.opener(""), file, "")
     print "   rev    offset  length   base linkrev"+\
           " p1           p2           nodeid"
     for i in range(r.count()):
         e = r.index[i]
         print "% 6d % 9d % 7d % 6d % 7d %s.. %s.. %s.." % (
             i, e[0], e[1], e[2], e[3],
             hg.hex(e[4][:5]), hg.hex(e[5][:5]), hg.hex(e[6][:5]))
 
 def debugindexdot(ui, file):
     r = hg.revlog(hg.opener(""), file, "")
     print "digraph G {"
     for i in range(r.count()):
         e = r.index[i]
         print "\t%d -> %d" % (r.rev(e[4]), i)
         if e[5] != hg.nullid:
             print "\t%d -> %d" % (r.rev(e[5]), i)
     print "}"
 
 def diff(ui, repo, *files, **opts):
     """diff working directory (or selected files)"""
     revs = []
     if opts['rev']:
         revs = map(lambda x: repo.lookup(x), opts['rev'])
 
     if len(revs) > 2:
         ui.warn("too many revisions to diff\n")
         sys.exit(1)
 
     if files:
         files = relpath(repo, files)
     else:
         files = relpath(repo, [""])
 
     dodiff(ui, repo, os.getcwd(), files, *revs)
 
 def export(ui, repo, changeset):
     """dump the changeset header and diffs for a revision"""
     node = repo.lookup(changeset)
     prev, other = repo.changelog.parents(node)
     change = repo.changelog.read(node)
     print "# HG changeset patch"
     print "# User %s" % change[1]
     print "# Node ID %s" % hg.hex(node)
     print "# Parent  %s" % hg.hex(prev)
     print
     if other != hg.nullid:
         print "# Parent  %s" % hg.hex(other)
     print change[4].rstrip()
     print
 
     dodiff(ui, repo, "", None, prev, node)
 
 def forget(ui, repo, file, *files):
     """don't add the specified files on the next commit"""
     repo.forget(relpath(repo, (file,) + files))
 
 def heads(ui, repo):
     """show current repository heads"""
     for n in repo.changelog.heads():
         show_changeset(ui, repo, changenode=n)
 
 def identify(ui, repo):
     """print information about the working copy"""
     parents = [p for p in repo.dirstate.parents() if p != hg.nullid]
     if not parents:
         ui.write("unknown\n")
         return
 
     hexfunc = ui.verbose and hg.hex or hg.short
-    (c, a, d, u) = repo.diffdir(repo.root)
+    (c, a, d, u) = repo.changes(None, None)
     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 ' + '
         output.append(' + '.join(parenttags))
 
     ui.write("%s\n" % ' '.join(output))
 
 def import_(ui, repo, patch1, *patches, **opts):
     """import an ordered set of patches"""
     try:
         import psyco
         psyco.full()
     except:
         pass
 
     patches = (patch1,) + patches
 
     d = opts["base"]
     strip = opts["strip"]
 
     for patch in patches:
         ui.status("applying %s\n" % patch)
         pf = os.path.join(d, patch)
 
         text = ""
         for l in file(pf):
             if l[:4] == "--- ": break
             text += l
 
         # make sure text isn't empty
         if not text: text = "imported patch %s\n" % patch
 
         f = os.popen("patch -p%d < %s" % (strip, pf))
         files = []
         for l in f.read().splitlines():
             l.rstrip('\r\n');
             ui.status("%s\n" % l)
             if l[:14] == 'patching file ':
                 pf = l[14:]
                 if pf not in files:
                     files.append(pf)
         patcherr = f.close()
         if patcherr:
             sys.stderr.write("patch failed")
             sys.exit(1)
 
         if len(files) > 0:
             addremove(ui, repo, *files)
         repo.commit(files, text)
 
 def init(ui, source=None):
     """create a new repository in the current directory"""
 
     if source:
         ui.warn("no longer supported: use \"hg clone\" instead\n")
         sys.exit(1)
     repo = hg.repository(ui, ".", create=1)
 
 def log(ui, repo, f = None):
     """show the revision history of the repository or a single file"""
     if f:
         f = relpath(repo, [f])[0]
         r = repo.file(f)
         for i in range(r.count() - 1, -1, -1):
             show_changeset(ui, repo, filelog=r, rev=i)
     else:
         for i in range(repo.changelog.count() - 1, -1, -1):
             show_changeset(ui, repo, rev=i)
 
 def manifest(ui, repo, rev = []):
     """output the latest or given revision of the project manifest"""
     n = repo.manifest.tip()
     if rev:
         n = repo.manifest.lookup(rev)
     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" % (hg.hex(m[f]), mf[f] and "755" or "644", f))
 
 def parents(ui, repo, node = None):
     '''show the parents of the current working dir'''
     if node:
         p = repo.changelog.parents(repo.lookup(hg.bin(node)))
     else:
         p = repo.dirstate.parents()
 
     for n in p:
         if n != hg.nullid:
             show_changeset(ui, repo, changenode=n)
 
 def pull(ui, repo, source="default", **opts):
     """pull changes from the specified source"""
     source = ui.expandpath(source)
 
     ui.status('pulling from %s\n' % (source))
 
     other = hg.repository(ui, source)
     fetch = repo.findincoming(other)
     if not fetch:
         ui.status("no changes found\n")
         return
 
     cg = other.changegroup(fetch)
     r = repo.addchangegroup(cg)
     if cg and 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"):
     """push changes to the specified destination"""
     dest = ui.expandpath(dest)
 
     if not dest.startswith("ssh://"):
         ui.warn("abort: can only push to ssh:// destinations currently\n")
         return 1
 
     m = re.match(r'ssh://(([^@]+)@)?([^:/]+)(:(\d+))?(/(.*))?', dest)
     if not m:
         ui.warn("abort: couldn't parse destination %s\n" % dest)
         return 1
 
     user, host, port, path = map(m.group, (2, 3, 5, 7))
     uhost = user and ("%s@%s" % (user, host)) or host
     port = port and (" -p %s") % port or ""
     path = path or ""
 
     sport = random.randrange(30000, 60000)
     cmd = "ssh %s%s -R %d:localhost:%d 'cd %s; hg pull http://localhost:%d/'"
     cmd = cmd % (uhost, port, sport+1, sport, path, sport+1)
 
     child = os.fork()
     if not child:
         sys.stdout = file("/dev/null", "w")
         sys.stderr = sys.stdout
         hgweb.server(repo.root, "pull", "", "localhost", sport)
     else:
         ui.status("connecting to %s\n" % host)
         r = os.system(cmd)
         os.kill(child, signal.SIGTERM)
         return r
 
 def rawcommit(ui, repo, *flist, **rc):
     "raw commit interface"
 
     text = rc['text']
     if not text and rc['logfile']:
         try: text = open(rc['logfile']).read()
         except IOError: pass
     if not text and not rc['logfile']:
         print "missing commit text"
         return 1
 
     files = relpath(repo, list(flist))
     if rc['files']:
         files += open(rc['files']).read().splitlines()
 
     rc['parent'] = map(repo.lookup, rc['parent'])
 
     repo.rawcommit(files, text, rc['user'], rc['date'], *rc['parent'])
 
 def recover(ui, repo):
     """roll back an interrupted transaction"""
     repo.recover()
 
 def remove(ui, repo, file, *files):
     """remove the specified files on the next commit"""
     repo.remove(relpath(repo, (file,) + files))
 
 def root(ui, repo):
     """print the root (top) of the current working dir"""
     ui.write(repo.root + "\n")
 
 def serve(ui, repo, **opts):
     """export the repository via HTTP"""
     hgweb.server(repo.root, opts["name"], opts["templates"],
                  opts["address"], opts["port"])
 
 def status(ui, repo):
     '''show changed files in the working directory
 
     C = changed
     A = added
     R = removed
     ? = not tracked'''
 
-    (c, a, d, u) = repo.diffdir(os.getcwd())
+    (c, a, d, u) = repo.changes(None, None, os.getcwd())
     (c, a, d, u) = map(lambda x: relfilter(repo, x), (c, a, d, u))
 
     for f in c: print "C", f
     for f in a: print "A", f
     for f in d: print "R", f
     for f in u: print "?", f
 
 def tag(ui, repo, name, rev = None, **opts):
     """add a tag for the current tip or a given revision"""
 
     if name == "tip":
 	ui.warn("abort: 'tip' is a reserved name!\n")
 	return -1
 
-    (c, a, d, u) = repo.diffdir(repo.root)
+    (c, a, d, u) = repo.changes(None, None)
     for x in (c, a, d, u):
 	if ".hgtags" in x:
 	    ui.warn("abort: working copy of .hgtags is changed!\n")
             ui.status("(please commit .hgtags manually)\n")
 	    return -1
 
     if rev:
         r = hg.hex(repo.lookup(rev))
     else:
         r = hg.hex(repo.changelog.tip())
 
     add = 0
     if not os.path.exists(repo.wjoin(".hgtags")): add = 1
     repo.wfile(".hgtags", "a").write("%s %s\n" % (r, name))
     if add: repo.add([".hgtags"])
 
     if not opts['text']:
         opts['text'] = "Added tag %s for changeset %s" % (name, r)
 
     repo.commit([".hgtags"], opts['text'], opts['user'], opts['date'])
 
 def tags(ui, repo):
     """list repository tags"""
 
     l = repo.tagslist()
     l.reverse()
     for t, n in l:
         try:
             r = "%5d:%s" % (repo.changelog.rev(n), hg.hex(n))
         except KeyError:
             r = "    ?:?"
         ui.write("%-30s %s\n" % (t, r))
 
 def tip(ui, repo):
     """show the tip revision"""
     n = repo.changelog.tip()
     show_changeset(ui, repo, changenode=n)
 
 def undo(ui, repo):
     """undo the last transaction"""
     repo.undo()
 
 def update(ui, repo, node=None, merge=False, clean=False):
     '''update or merge working directory
 
     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.
     '''
     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"""
     return repo.verify()
 
 # Command options and aliases are listed here, alphabetically
 
 table = {
     "add": (add, [], "hg add [files]"),
     "addremove": (addremove, [], "hg addremove [files]"),
     "annotate": (annotate,
                      [('r', 'revision', '', 'revision'),
                       ('u', 'user', None, 'show user'),
                       ('n', 'number', None, 'show revision number'),
                       ('c', 'changeset', None, 'show changeset')],
                      'hg annotate [-u] [-c] [-n] [-r id] [files]'),
     "cat": (cat, [], 'hg cat <file> [rev]'),
     "clone": (clone, [('U', 'noupdate', None, 'skip update after cloning')],
               'hg clone [options] <source> [dest]'),
     "commit|ci": (commit,
                   [('t', 'text', "", 'commit text'),
                    ('A', 'addremove', None, 'run add/remove during commit'),
                    ('l', 'logfile', "", 'commit text file'),
                    ('d', 'date', "", 'data'),
                    ('u', 'user', "", 'user')],
                   'hg commit [files]'),
     "copy": (copy, [], 'hg copy <source> <dest>'),
     "debugcheckdirstate": (debugcheckdirstate, [], 'debugcheckdirstate'),
     "debugdumpdirstate": (debugdumpdirstate, [], 'debugdumpdirstate'),
     "debugindex": (debugindex, [], 'debugindex <file>'),
     "debugindexdot": (debugindexdot, [], 'debugindexdot <file>'),
     "diff": (diff, [('r', 'rev', [], 'revision')],
              'hg diff [-r A] [-r B] [files]'),
     "export": (export, [], "hg export <changeset>"),
     "forget": (forget, [], "hg forget [files]"),
     "heads": (heads, [], 'hg heads'),
     "help": (help, [], 'hg help [command]'),
     "identify|id": (identify, [], 'hg identify'),
     "import|patch": (import_,
                      [('p', 'strip', 1, 'path strip'),
                       ('b', 'base', "", 'base path')],
                      "hg import [options] <patches>"),
     "init": (init, [], 'hg init'),
     "log|history": (log, [], 'hg log [file]'),
     "manifest": (manifest, [], 'hg manifest [rev]'),
     "parents": (parents, [], 'hg parents [node]'),
     "pull": (pull,
                   [('u', 'update', None, 'update working directory')],
 		  'hg pull [options] [source]'),
     "push": (push, [], 'hg push <destination>'),
     "rawcommit": (rawcommit,
                   [('p', 'parent', [], 'parent'),
                    ('d', 'date', "", 'data'),
                    ('u', 'user', "", 'user'),
                    ('F', 'files', "", 'file list'),
                    ('t', 'text', "", 'commit text'),
                    ('l', 'logfile', "", 'commit text file')],
                   'hg rawcommit [options] [files]'),
     "recover": (recover, [], "hg recover"),
     "remove|rm": (remove, [], "hg remove [files]"),
     "root": (root, [], "hg root"),
     "serve": (serve, [('p', 'port', 8000, 'listen port'),
                       ('a', 'address', '', 'interface address'),
                       ('n', 'name', os.getcwd(), 'repository name'),
                       ('t', 'templates', "", 'template map')],
               "hg serve [options]"),
     "status": (status, [], 'hg status'),
     "tag": (tag,  [('t', 'text', "", 'commit text'),
                    ('d', 'date', "", 'date'),
                    ('u', 'user', "", 'user')],
             'hg tag [options] <name> [rev]'),
     "tags": (tags, [], 'hg tags'),
     "tip": (tip, [], 'hg tip'),
     "undo": (undo, [], 'hg undo'),
     "update|up|checkout|co":
             (update,
              [('m', 'merge', None, 'allow merging of conflicts'),
               ('C', 'clean', None, 'overwrite locally modified files')],
              'hg update [options] [node]'),
     "verify": (verify, [], 'hg verify'),
     "version": (show_version, [], 'hg version'),
     }
 
 norepo = "clone init version help debugindex debugindexdot"
 
 def find(cmd):
     for e in table.keys():
         if re.match("(%s)$" % e, cmd):
             return table[e]
 
     raise UnknownCommand(cmd)
 
 class SignalInterrupt(Exception): pass
 
 def catchterm(*args):
     raise SignalInterrupt
 
 def run():
     sys.exit(dispatch(sys.argv[1:]))
 
 def dispatch(args):
     options = {}
     opts = [('v', 'verbose', None, 'verbose'),
             ('d', 'debug', None, 'debug'),
             ('q', 'quiet', None, 'quiet'),
             ('p', 'profile', None, 'profile'),
             ('R', 'repository', "", 'repository root directory'),
             ('', 'traceback', None, 'print traceback on exception'),
             ('y', 'noninteractive', None, 'run non-interactively'),
             ('', 'version', None, 'output version information and exit'),
             ]
 
     args = fancyopts.fancyopts(args, opts, options,
                                'hg [options] <command> [options] [files]')
 
     if not args:
         cmd = "help"
     else:
         cmd, args = args[0], args[1:]
 
     u = ui.ui(options["verbose"], options["debug"], options["quiet"],
            not options["noninteractive"])
 
     if options["version"]:
         show_version(u)
         sys.exit(0)
 
     try:
         i = find(cmd)
     except UnknownCommand:
         u.warn("hg: unknown command '%s'\n" % cmd)
         help(u)
         sys.exit(1)
 
     signal.signal(signal.SIGTERM, catchterm)
 
     cmdoptions = {}
     try:
         args = fancyopts.fancyopts(args, i[1], cmdoptions, i[2])
     except fancyopts.getopt.GetoptError, inst:
         u.warn("hg %s: %s\n" % (cmd, inst))
         help(u, cmd)
         sys.exit(-1)
 
     try:
         try:
             if cmd not in norepo.split():
                 path = options["repository"] or ""
                 repo = hg.repository(ui=u, path=path)
                 d = lambda: i[0](u, repo, *args, **cmdoptions)
             else:
                 d = lambda: i[0](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:
             if options['traceback']:
                 traceback.print_exc()
             raise
     except util.CommandError, inst:
         u.warn("abort: %s\n" % inst.args)
     except hg.RepoError, inst:
         u.warn("abort: ", inst, "!\n")
     except SignalInterrupt:
         u.warn("killed!\n")
     except KeyboardInterrupt:
         u.warn("interrupted!\n")
     except IOError, inst:
         if hasattr(inst, "code"):
             u.warn("abort: %s\n" % inst)
         elif hasattr(inst, "reason"):
             u.warn("abort: error %d: %s\n" % (inst.reason[0], inst.reason[1]))
         elif hasattr(inst, "args") and inst[0] == errno.EPIPE:
             u.warn("broken pipe\n")
         else:
             raise
     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" % i[0].__name__)
         help(u, cmd)
 
     sys.exit(-1)
 

mercurial/hg.py

@@ -1,1496 +1,1522 b''
 # hg.py - repository classes 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 sys, struct, os
 import util
 from revlog import *
 from demandload import *
 demandload(globals(), "re lock urllib urllib2 transaction time socket")
 demandload(globals(), "tempfile httprangereader bdiff")
 
 class filelog(revlog):
     def __init__(self, opener, path):
         revlog.__init__(self, opener,
                         os.path.join("data", path + ".i"),
                         os.path.join("data", path + ".d"))
 
     def read(self, node):
         t = self.revision(node)
         if t[:2] != '\1\n':
             return t
         s = t.find('\1\n', 2)
         return t[s+2:]
 
     def readmeta(self, node):
         t = self.revision(node)
         if t[:2] != '\1\n':
             return t
         s = t.find('\1\n', 2)
         mt = t[2:s]
         for l in mt.splitlines():
             k, v = l.split(": ", 1)
             m[k] = v
         return m
 
     def add(self, text, meta, transaction, link, p1=None, p2=None):
         if meta or text[:2] == '\1\n':
             mt = ""
             if meta:
                 mt = [ "%s: %s\n" % (k, v) for k,v in meta.items() ]
             text = "\1\n" + "".join(mt) + "\1\n" + text
         return self.addrevision(text, transaction, link, p1, p2)
 
     def annotate(self, node):
 
         def decorate(text, rev):
             return ([rev] * len(text.splitlines()), text)
 
         def pair(parent, child):
             for a1, a2, b1, b2 in bdiff.blocks(parent[1], child[1]):
                 child[0][b1:b2] = parent[0][a1:a2]
             return child
 
         # find all ancestors
         needed = {node:1}
         visit = [node]
         while visit:
             n = visit.pop(0)
             for p in self.parents(n):
                 if p not in needed:
                     needed[p] = 1
                     visit.append(p)
                 else:
                     # count how many times we'll use this
                     needed[p] += 1
 
         # sort by revision which is a topological order
         visit = [ (self.rev(n), n) for n in needed.keys() ]
         visit.sort()
         hist = {}
 
         for r,n in visit:
             curr = decorate(self.read(n), self.linkrev(n))
             for p in self.parents(n):
                 if p != nullid:
                     curr = pair(hist[p], curr)
                     # trim the history of unneeded revs
                     needed[p] -= 1
                     if not needed[p]:
                         del hist[p]
             hist[n] = curr
 
         return zip(hist[n][0], hist[n][1].splitlines(1))
 
 class manifest(revlog):
     def __init__(self, opener):
         self.mapcache = None
         self.listcache = None
         self.addlist = None
         revlog.__init__(self, opener, "00manifest.i", "00manifest.d")
 
     def read(self, node):
         if node == nullid: return {} # don't upset local cache
         if self.mapcache and self.mapcache[0] == node:
             return self.mapcache[1].copy()
         text = self.revision(node)
         map = {}
         flag = {}
         self.listcache = (text, text.splitlines(1))
         for l in self.listcache[1]:
             (f, n) = l.split('\0')
             map[f] = bin(n[:40])
             flag[f] = (n[40:-1] == "x")
         self.mapcache = (node, map, flag)
         return map
 
     def readflags(self, node):
         if node == nullid: return {} # don't upset local cache
         if not self.mapcache or self.mapcache[0] != node:
             self.read(node)
         return self.mapcache[2]
 
     def diff(self, a, b):
         # this is sneaky, as we're not actually using a and b
         if self.listcache and self.addlist and self.listcache[0] == a:
             d = mdiff.diff(self.listcache[1], self.addlist, 1)
             if mdiff.patch(a, d) != b:
                 sys.stderr.write("*** sortdiff failed, falling back ***\n")
                 return mdiff.textdiff(a, b)
             return d
         else:
             return mdiff.textdiff(a, b)
 
     def add(self, map, flags, transaction, link, p1=None, p2=None):
         files = map.keys()
         files.sort()
 
         self.addlist = ["%s\000%s%s\n" %
                         (f, hex(map[f]), flags[f] and "x" or '')
                         for f in files]
         text = "".join(self.addlist)
 
         n = self.addrevision(text, transaction, link, p1, p2)
         self.mapcache = (n, map, flags)
         self.listcache = (text, self.addlist)
         self.addlist = None
 
         return n
 
 class changelog(revlog):
     def __init__(self, opener):
         revlog.__init__(self, opener, "00changelog.i", "00changelog.d")
 
     def extract(self, text):
         if not text:
             return (nullid, "", "0", [], "")
         last = text.index("\n\n")
         desc = text[last + 2:]
         l = text[:last].splitlines()
         manifest = bin(l[0])
         user = l[1]
         date = l[2]
         files = l[3:]
         return (manifest, user, date, files, desc)
 
     def read(self, node):
         return self.extract(self.revision(node))
 
     def add(self, manifest, list, desc, transaction, p1=None, p2=None,
                   user=None, date=None):
         user = (user or
                 os.environ.get("HGUSER") or
                 os.environ.get("EMAIL") or
                 (os.environ.get("LOGNAME",
                                os.environ.get("USERNAME", "unknown"))
                  + '@' + socket.getfqdn()))
         date = date or "%d %d" % (time.time(), time.timezone)
         list.sort()
         l = [hex(manifest), user, date] + list + ["", desc]
         text = "\n".join(l)
         return self.addrevision(text, transaction, self.count(), p1, p2)
 
 class dirstate:
     def __init__(self, opener, ui, root):
         self.opener = opener
         self.root = root
         self.dirty = 0
         self.ui = ui
         self.map = None
         self.pl = None
         self.copies = {}
 
     def __del__(self):
         if self.dirty:
             self.write()
 
     def __getitem__(self, key):
         try:
             return self.map[key]
         except TypeError:
             self.read()
             return self[key]
 
     def __contains__(self, key):
         if not self.map: self.read()
         return key in self.map
 
     def parents(self):
         if not self.pl:
             self.read()
         return self.pl
 
     def setparents(self, p1, p2 = nullid):
         self.dirty = 1
         self.pl = p1, p2
 
     def state(self, key):
         try:
             return self[key][0]
         except KeyError:
             return "?"
 
     def read(self):
         if self.map is not None: return self.map
 
         self.map = {}
         self.pl = [nullid, nullid]
         try:
             st = self.opener("dirstate").read()
             if not st: return
         except: return
 
         self.pl = [st[:20], st[20: 40]]
 
         pos = 40
         while pos < len(st):
             e = struct.unpack(">cllll", st[pos:pos+17])
             l = e[4]
             pos += 17
             f = st[pos:pos + l]
             if '\0' in f:
                 f, c = f.split('\0')
                 self.copies[f] = c
             self.map[f] = e[:4]
             pos += l
 
     def copy(self, source, dest):
         self.read()
         self.dirty = 1
         self.copies[dest] = source
 
     def copied(self, file):
         return self.copies.get(file, None)
 
     def update(self, files, state):
         ''' current states:
         n  normal
         m  needs merging
         r  marked for removal
         a  marked for addition'''
 
         if not files: return
         self.read()
         self.dirty = 1
         for f in files:
             if state == "r":
                 self.map[f] = ('r', 0, 0, 0)
             else:
                 s = os.stat(os.path.join(self.root, f))
                 self.map[f] = (state, s.st_mode, s.st_size, s.st_mtime)
 
     def forget(self, files):
         if not files: return
         self.read()
         self.dirty = 1
         for f in files:
             try:
                 del self.map[f]
             except KeyError:
                 self.ui.warn("not in dirstate: %s!\n" % f)
                 pass
 
     def clear(self):
         self.map = {}
         self.dirty = 1
 
     def write(self):
         st = self.opener("dirstate", "w")
         st.write("".join(self.pl))
         for f, e in self.map.items():
             c = self.copied(f)
             if c:
                 f = f + "\0" + c
             e = struct.pack(">cllll", e[0], e[1], e[2], e[3], len(f))
             st.write(e + f)
         self.dirty = 0
 
-    def dup(self):
+    def changes(self, files, ignore):
         self.read()
-        return self.map.copy()
+        dc = self.map.copy()
+        lookup, changed, added, unknown = [], [], [], []
+
+        # compare all files by default
+        if not files: files = [self.root]
+
+        def uniq(g):
+            seen = {}
+            for f in g:
+                if f not in seen:
+                    seen[f] = 1
+                    yield f
+
+        # recursive generator of all files listed
+        def walk(files):
+            for f in uniq(files):
+                if os.path.isdir(f):
+                    for dir, subdirs, fl in os.walk(f):
+                        d = dir[len(self.root) + 1:]
+                        if ".hg" in subdirs: subdirs.remove(".hg")
+                        for fn in fl:
+                            fn = util.pconvert(os.path.join(d, fn))
+                            yield fn
+                else:
+                    yield f[len(self.root) + 1:]
+
+        for fn in uniq(walk(files)):
+            try: s = os.stat(os.path.join(self.root, fn))
+            except: continue
+
+            if fn in dc:
+                c = dc[fn]
+                del dc[fn]
+
+                if c[0] == 'm':
+                    changed.append(fn)
+                elif c[0] == 'a':
+                    added.append(fn)
+                elif c[0] == 'r':
+                    unknown.append(fn)
+                elif c[2] != s.st_size or (c[1] ^ s.st_mode) & 0100:
+                    changed.append(fn)
+                elif c[1] != s.st_mode or c[3] != s.st_mtime:
+                    lookup.append(fn)
+            else:
+                if not ignore(fn): unknown.append(fn)
+
+        return (lookup, changed, added, dc.keys(), unknown)
 
 # used to avoid circular references so destructors work
 def opener(base):
     p = base
     def o(path, mode="r"):
         if p[:7] == "http://":
             f = os.path.join(p, urllib.quote(path))
             return httprangereader.httprangereader(f)
 
         f = os.path.join(p, path)
 
         mode += "b" # for that other OS
 
         if mode[0] != "r":
             try:
                 s = os.stat(f)
             except OSError:
                 d = os.path.dirname(f)
                 if not os.path.isdir(d):
                     os.makedirs(d)
             else:
                 if s.st_nlink > 1:
                     file(f + ".tmp", "wb").write(file(f, "rb").read())
                     util.rename(f+".tmp", f)
 
         return file(f, mode)
 
     return o
 
 class RepoError(Exception): pass
 
 class localrepository:
     def __init__(self, ui, path=None, create=0):
         self.remote = 0
         if path and path[:7] == "http://":
             self.remote = 1
             self.path = path
         else:
             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 RepoError("no repo found")
                 path = p
             self.path = os.path.join(path, ".hg")
 
             if not create and not os.path.isdir(self.path):
                 raise RepoError("repository %s not found" % self.path)
 
         self.root = path
         self.ui = ui
 
         if create:
             os.mkdir(self.path)
             os.mkdir(self.join("data"))
 
         self.opener = opener(self.path)
         self.wopener = opener(self.root)
         self.manifest = manifest(self.opener)
         self.changelog = changelog(self.opener)
         self.ignorelist = None
         self.tagscache = None
         self.nodetagscache = None
 
         if not self.remote:
             self.dirstate = dirstate(self.opener, ui, self.root)
             try:
                 self.ui.readconfig(self.opener("hgrc"))
             except IOError: pass
 
     def ignore(self, f):
         if self.ignorelist is None:
             self.ignorelist = []
             try:
                 l = file(self.wjoin(".hgignore"))
                 for pat in l:
                     if pat != "\n":
                         self.ignorelist.append(re.compile(util.pconvert(pat[:-1])))
             except IOError: pass
         for pat in self.ignorelist:
             if pat.search(f): return True
         return False
 
     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
 
             r = os.system(s)
 
             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 = {}
             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.revision(r).splitlines():
                         if l:
                             n, k = l.split(" ", 1)
                             try:
                                 bin_n = bin(n)
                             except TypeError:
                                 bin_n = ''
                             self.tagscache[k.strip()] = bin_n
             except KeyError:
                 pass
             for k, n in self.ui.configitems("tags"):
                 try:
                     bin_n = bin(n)
                 except TypeError:
                     bin_n = ''
                 self.tagscache[k] = bin_n
 
             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:
             return self.changelog.lookup(key)
 
     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(self.opener, f)
 
     def wfile(self, f, mode='r'):
         return self.wopener(f, mode)
 
     def transaction(self):
         # save dirstate for undo
         try:
             ds = self.opener("dirstate").read()
         except IOError:
             ds = ""
         self.opener("undo.dirstate", "w").write(ds)
 
         return transaction.transaction(self.opener, self.join("journal"),
                                        self.join("undo"))
 
     def recover(self):
         lock = self.lock()
         if os.path.exists(self.join("recover")):
             self.ui.status("rolling back interrupted transaction\n")
             return transaction.rollback(self.opener, self.join("recover"))
         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(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])
 
         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.wfile(f).read()
                 tm = util.is_exec(self.wjoin(f), mfm.get(f, False))
                 r = self.file(f)
                 mfm[f] = tm
                 mm[f] = r.add(t, {}, tr, linkrev,
                               m1.get(f, nullid), m2.get(f, nullid))
                 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])
         n = self.changelog.add(mnode, files, 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):
         commit = []
         remove = []
         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.diffdir(self.root)
+            (c, a, d, u) = self.changes(None, None)
             commit = c + a
             remove = d
 
         if not commit and not remove:
             self.ui.status("nothing changed\n")
             return
 
         if not self.hook("precommit"):
             return 1
 
         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])
         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.wfile(f).read()
             except IOError:
                 self.warn("trouble committing %s!\n" % f)
                 raise
 
             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"])) 
 
             r = self.file(f)
             fp1 = m1.get(f, nullid)
             fp2 = m2.get(f, nullid)
             new[f] = r.add(t, meta, tr, linkrev, fp1, fp2)
 
         # 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])
 
         # add changeset
         new = new.keys()
         new.sort()
 
         if not text:
             edittext = "\n" + "HG: manifest hash %s\n" % hex(mn)
             edittext += "".join(["HG: changed %s\n" % f for f in new])
             edittext += "".join(["HG: removed %s\n" % f for f in remove])
             edittext = self.ui.edit(edittext)
             if not edittext.rstrip():
                 return 1
             text = edittext
 
         n = self.changelog.add(mn, new, text, tr, p1, p2, user, date)
 
         if not self.hook("commit", node=hex(n)):
             return 1
 
         tr.close()
 
         self.dirstate.setparents(n)
         self.dirstate.update(new, "n")
         self.dirstate.forget(remove)
 
-    def diffdir(self, path, changeset = None):
-        changed = []
-        added = []
-        unknown = []
-        mf = {}
+    def changes(self, node1, node2, *files):
+        # changed, added, deleted, unknown
+        c, a, d, u, mf1 = [], [], [], [], None
 
-        if changeset:
-            change = self.changelog.read(changeset)
-            mf = self.manifest.read(change[0])
-            dc = dict.fromkeys(mf)
-        else:
-            changeset = self.dirstate.parents()[0]
-            change = self.changelog.read(changeset)
-            mf = self.manifest.read(change[0])
-            dc = self.dirstate.dup()
-
-        def fcmp(fn):
+        def fcmp(fn, mf):
             t1 = self.wfile(fn).read()
             t2 = self.file(fn).revision(mf[fn])
             return cmp(t1, t2)
 
-        for dir, subdirs, files in os.walk(path):
-            d = dir[len(self.root)+1:]
-            if ".hg" in subdirs: subdirs.remove(".hg")
+        # are we comparing the working directory?
+        if not node1:
+            l, c, a, d, u = self.dirstate.changes(files, self.ignore)
+
+            # are we comparing working dir against its parent?
+            if not node2:
+                if l:
+                    # do a full compare of any files that might have changed
+                    change = self.changelog.read(self.dirstate.parents()[0])
+                    mf1 = self.manifest.read(change[0])
+                    for f in lookup:
+                        if fcmp(f, mf):
+                            c.append(f)
+                return (c, a, d, u)
 
-            for f in files:
-                fn = util.pconvert(os.path.join(d, f))
-                try: s = os.stat(os.path.join(self.root, fn))
-                except: continue
-                if fn in dc:
-                    c = dc[fn]
-                    del dc[fn]
-                    if not c:
-                        if fcmp(fn):
-                            changed.append(fn)
-                    elif c[0] == 'm':
-                        changed.append(fn)
-                    elif c[0] == 'a':
-                        added.append(fn)
-                    elif c[0] == 'r':
-                        unknown.append(fn)
-                    elif c[2] != s.st_size or (c[1] ^ s.st_mode) & 0100:
-                        changed.append(fn)
-                    elif c[1] != s.st_mode or c[3] != s.st_mtime:
-                        if fcmp(fn):
-                            changed.append(fn)
-                else:
-                    if self.ignore(fn): continue
-                    unknown.append(fn)
+        # are we comparing working dir against non-tip?
+        # generate a pseudo-manifest for the working dir
+        if not node1:
+            if not mf1:
+                change = self.changelog.read(self.dirstate.parents()[0])
+                mf1 = self.manifest.read(change[0])
+            for f in a + c + l:
+                mf1[f] = ""
+            for f in d:
+                if f in mf1: del mf1[f]
+        else:
+            change = self.changelog.read(node1)
+            mf1 = self.manifest.read(change[0])
 
-        deleted = dc.keys()
-        deleted.sort()
-
-        return (changed, added, deleted, unknown)
-
-    def diffrevs(self, node1, node2):
-        changed, added = [], []
-
-        change = self.changelog.read(node1)
-        mf1 = self.manifest.read(change[0])
         change = self.changelog.read(node2)
         mf2 = self.manifest.read(change[0])
 
         for fn in mf2:
             if mf1.has_key(fn):
                 if mf1[fn] != mf2[fn]:
-                    changed.append(fn)
+                    if mf1[fn] != "" or fcmp(fn, mf2):
+                        c.append(fn)
                 del mf1[fn]
             else:
-                added.append(fn)
+                a.append(fn)
 
-        deleted = mf1.keys()
-        deleted.sort()
+        d = mf1.keys()
+        d.sort()
 
-        return (changed, added, deleted)
+        return (c, a, d, u)
 
     def add(self, list):
         for f in list:
             p = self.wjoin(f)
             if not os.path.isfile(p):
                 self.ui.warn("%s does not exist!\n" % f)
             elif self.dirstate.state(f) == 'n':
                 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):
         for f in list:
             p = self.wjoin(f)
             if os.path.isfile(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 copy(self, source, dest):
         p = self.wjoin(dest)
         if not os.path.isfile(dest):
             self.ui.warn("%s does not exist!\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()
 
     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):
         m = self.changelog.nodemap
         search = []
         fetch = []
         seen = {}
         seenbranch = {}
 
         # if we have an empty repo, fetch everything
         if self.changelog.tip() == nullid:
             self.ui.status("requesting all changes\n")
             return [nullid]
 
         # otherwise, assume we're closer to the tip than the root
         self.ui.status("searching for changes\n")
         heads = remote.heads()
         unknown = []
         for h in heads:
             if h not in m:
                 unknown.append(h)
 
         if not unknown:
             return None
 
         rep = {}
         reqcnt = 0
 
         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.append(n[1]) # earliest unknown
                             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] not in m and b[0] not in seen:
                             unknown.append(b)
 
         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.append(p)
                     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
 
         for f in fetch:
             if f in m:
                 raise RepoError("already have changeset " + short(f[:4]))
 
         if fetch == [nullid]:
             self.ui.warn("warning: pulling from an unrelated repository!\n")
 
         self.ui.note("adding new changesets starting at " +
                      " ".join([short(f) for f in fetch]) + "\n")
 
         self.ui.debug("%d total queries\n" % reqcnt)
 
         return fetch
 
     def changegroup(self, basenodes):
         nodes = self.newer(basenodes)
 
         # construct the link map
         linkmap = {}
         for n in nodes:
             linkmap[self.changelog.rev(n)] = n
 
         # 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()
 
         # 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
 
     def addchangegroup(self, generator):
 
         class genread:
             def __init__(self, generator):
                 self.g = generator
                 self.buf = ""
             def read(self, l):
                 while l > len(self.buf):
                     try:
                         self.buf += self.g.next()
                     except StopIteration:
                         break
                 d, self.buf = self.buf[:l], self.buf[l:]
                 return d
 
         def getchunk():
             d = source.read(4)
             if not d: return ""
             l = struct.unpack(">l", d)[0]
             if l <= 4: return ""
             return source.read(l - 4)
 
         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 generator: return
         changesets = files = revisions = 0
 
         source = genread(generator)
         lock = self.lock()
         tr = self.transaction()
 
         # pull off the changeset group
         self.ui.status("adding changesets\n")
         co = self.changelog.tip()
         cn = self.changelog.addgroup(getgroup(), csmap, tr, 1) # unique
         changesets = self.changelog.rev(cn) - self.changelog.rev(co)
 
         # 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 revisions\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
 
         self.ui.status(("modified %d files, added %d changesets" +
                         " and %d new revisions\n")
                        % (files, changesets, revisions))
 
         tr.close()
         return
 
     def update(self, node, allow=False, force=False):
         pl = self.dirstate.parents()
         if not force and pl[1] != nullid:
             self.ui.warn("aborting: outstanding uncommitted merges\n")
             return
 
         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.diffdir(self.root)
+        (c, a, d, u) = self.changes(None, None)
 
         # 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(" ancestor %s local %s remote %s\n" %
                       (short(man), short(m1n), short(m2n)))
 
         merge = {}
         get = {}
         remove = []
         mark = {}
 
         # construct a working dir manifest
         mw = m1.copy()
         mfw = mf1.copy()
         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 linear_path and f not in m2:
                 self.dirstate.forget((f,))
 
         for f, n in mw.iteritems():
             if f in m2:
                 s = 0
 
                 # is the wfile new since m1, and match m2?
                 if f not in m1:
                     t1 = self.wfile(f).read()
                     t2 = self.file(f).revision(m2[f])
                     if cmp(t1, t2) == 0:
                         mark[f] = 1
                         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
                     else:
                         mark[f] = 1
 
                 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)
                             mark[f] = 1
                 del m2[f]
             elif f in ma:
                 if not force and n != ma[f]:
                     r = ""
                     if 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:
                 if 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 created %s, keeping\n" % f)
                 else:
                     self.ui.debug("working dir created %s, keeping\n" % f)
 
         for f, n in m2.iteritems():
             if f[0] == "/": continue
             if not force and f in ma and n != ma[f]:
                 r = ""
                 if 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 == "d": remove.append(f)
             else:
                 self.ui.debug("remote created %s\n" % f)
                 get[f] = n
 
         del mw, m1, m2, ma
 
         if force:
             for f in merge:
                 get[f] = merge[f][1]
             merge = {}
 
         if linear_path:
             # we don't need to do any magic, just jump to the new rev
             mode = 'n'
             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 perform a branch merge)\n")
                 return 1
             # we have to remember what files we needed to get/change
             # because any file that's different from either one of its
             # parents must be in the changeset
             mode = 'm'
             self.dirstate.update(mark.keys(), "m")
 
         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.wfile(f, "w").write(t)
             except IOError:
                 os.makedirs(os.path.dirname(self.wjoin(f)))
                 self.wfile(f, "w").write(t)
             util.set_exec(self.wjoin(f), mf2[f])
             self.dirstate.update([f], mode)
 
         # merge the tricky bits
         files = merge.keys()
         files.sort()
         for f in files:
             self.ui.status("merging %s\n" % f)
             m, o, flag = merge[f]
             self.merge3(f, m, o)
             util.set_exec(self.wjoin(f), flag)
             self.dirstate.update([f], 'm')
 
         for f in remove:
             self.ui.note("removing %s\n" % f)
             os.unlink(f)
         if mode == 'n':
             self.dirstate.forget(remove)
         else:
             self.dirstate.update(remove, 'r')
 
     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")
             f.write(fl.revision(node))
             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: other %s ancestor %s\n" %
                               (fn, short(other), short(base)))
 
         cmd = os.environ.get("HGMERGE", "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
 
         seen = {}
         self.ui.status("checking changesets\n")
         for i in range(self.changelog.count()):
             changesets += 1
             n = self.changelog.node(i)
             if n in seen:
                 self.ui.warn("duplicate changeset at revision %d\n" % i)
                 errors += 1
             seen[n] = 1
 
             for p in self.changelog.parents(n):
                 if p not in self.changelog.nodemap:
                     self.ui.warn("changeset %s has unknown parent %s\n" %
                                  (short(n), short(p)))
                     errors += 1
             try:
                 changes = self.changelog.read(n)
             except Exception, inst:
                 self.ui.warn("unpacking changeset %s: %s\n" % (short(n), inst))
                 errors += 1
 
             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)
             if n in seen:
                 self.ui.warn("duplicate manifest at revision %d\n" % i)
                 errors += 1
             seen[n] = 1
 
             for p in self.manifest.parents(n):
                 if p not in self.manifest.nodemap:
                     self.ui.warn("manifest %s has unknown parent %s\n" %
                             (short(n), short(p)))
                     errors += 1
 
             try:
                 delta = mdiff.patchtext(self.manifest.delta(n))
             except KeyboardInterrupt:
                 print "aborted"
                 sys.exit(0)
             except Exception, inst:
                 self.ui.warn("unpacking manifest %s: %s\n"
                              % (short(n), inst))
                 errors += 1
 
             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 f in filenodes:
             if f not in filelinkrevs:
                 self.ui.warn("file %s in manifest but not in changesets\n" % f)
                 errors += 1
 
         for f in filelinkrevs:
             if f not in filenodes:
                 self.ui.warn("file %s in changeset but not in manifest\n" % f)
                 errors += 1
 
         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:
                     self.ui.warn("%s: duplicate revision %d\n" % (f, i))
                     errors += 1
 
                 if n not in filenodes[f]:
                     self.ui.warn("%s: %d:%s not in manifests\n"
                                  % (f, i, short(n)))
                     print len(filenodes[f].keys()), fl.count(), f
                     errors += 1
                 else:
                     del filenodes[f][n]
 
                 flr = fl.linkrev(n)
                 if flr not in filelinkrevs[f]:
                     self.ui.warn("%s:%s points to unexpected changeset %d\n"
                             % (f, short(n), fl.linkrev(n)))
                     errors += 1
                 else:
                     filelinkrevs[f].remove(flr)
 
                 # verify contents
                 try:
                     t = fl.read(n)
                 except Exception, inst:
                     self.ui.warn("unpacking file %s %s: %s\n"
                                  % (f, short(n), inst))
                     errors += 1
 
                 # verify parents
                 (p1, p2) = fl.parents(n)
                 if p1 not in nodes:
                     self.ui.warn("file %s:%s unknown parent 1 %s" %
                             (f, short(n), short(p1)))
                     errors += 1
                 if p2 not in nodes:
                     self.ui.warn("file %s:%s unknown parent 2 %s" %
                             (f, short(n), short(p1)))
                     errors += 1
                 nodes[n] = 1
 
             # cross-check
             for node in filenodes[f]:
                 self.ui.warn("node %s in manifests not in %s\n"
                              % (hex(n), f))
                 errors += 1
 
         self.ui.status("%d files, %d changesets, %d total revisions\n" %
                        (files, changesets, revisions))
 
         if errors:
             self.ui.warn("%d integrity errors encountered!\n" % errors)
             return 1
 
 class remoterepository:
     def __init__(self, ui, path):
         self.url = path
         self.ui = ui
         no_list = [ "localhost", "127.0.0.1" ]
         host = ui.config("http_proxy", "host")
         if host is None:
             host = os.environ.get("http_proxy")
         if host and host.startswith('http://'):
             host = host[7:]
         user = ui.config("http_proxy", "user")
         passwd = ui.config("http_proxy", "passwd")
         no = ui.config("http_proxy", "no")
         if no is None:
             no = os.environ.get("no_proxy")
         if no:
             no_list = no_list + no.split(",")
 
         no_proxy = 0
         for h in no_list:
             if (path.startswith("http://" + h + "/") or
                 path.startswith("http://" + h + ":") or
                 path == "http://" + h):
                 no_proxy = 1
 
         # Note: urllib2 takes proxy values from the environment and those will
         # take precedence
         for env in ["HTTP_PROXY", "http_proxy", "no_proxy"]:
             if os.environ.has_key(env):
                 del os.environ[env]
 
         proxy_handler = urllib2.BaseHandler()
         if host and not no_proxy:
             proxy_handler = urllib2.ProxyHandler({"http" : "http://" + host})
 
         authinfo = None
         if user and passwd:
             passmgr = urllib2.HTTPPasswordMgrWithDefaultRealm()
             passmgr.add_password(None, host, user, passwd)
             authinfo = urllib2.ProxyBasicAuthHandler(passmgr)
 
         opener = urllib2.build_opener(proxy_handler, authinfo)
         urllib2.install_opener(opener)
 
     def do_cmd(self, cmd, **args):
         self.ui.debug("sending %s command\n" % cmd)
         q = {"cmd": cmd}
         q.update(args)
         qs = urllib.urlencode(q)
         cu = "%s?%s" % (self.url, qs)
         return urllib2.urlopen(cu)
 
     def heads(self):
         d = self.do_cmd("heads").read()
         try:
             return map(bin, d[:-1].split(" "))
         except:
             self.ui.warn("unexpected response:\n" + d[:400] + "\n...\n")
             raise
 
     def branches(self, nodes):
         n = " ".join(map(hex, nodes))
         d = self.do_cmd("branches", nodes=n).read()
         try:
             br = [ tuple(map(bin, b.split(" "))) for b in d.splitlines() ]
             return br
         except:
             self.ui.warn("unexpected response:\n" + d[:400] + "\n...\n")
             raise
 
     def between(self, pairs):
         n = "\n".join(["-".join(map(hex, p)) for p in pairs])
         d = self.do_cmd("between", pairs=n).read()
         try:
             p = [ l and map(bin, l.split(" ")) or [] for l in d.splitlines() ]
             return p
         except:
             self.ui.warn("unexpected response:\n" + d[:400] + "\n...\n")
             raise
 
     def changegroup(self, nodes):
         n = " ".join(map(hex, nodes))
         zd = zlib.decompressobj()
         f = self.do_cmd("changegroup", roots=n)
         bytes = 0
         while 1:
             d = f.read(4096)
             bytes += len(d)
             if not d:
                 yield zd.flush()
                 break
             yield zd.decompress(d)
         self.ui.note("%d bytes of data transfered\n" % bytes)
 
 def repository(ui, path=None, create=0):
     if path and path[:7] == "http://":
         return remoterepository(ui, path)
     if path and path[:5] == "hg://":
         return remoterepository(ui, path.replace("hg://", "http://"))
     if path and path[:11] == "old-http://":
         return localrepository(ui, path.replace("old-http://", "http://"))
     else:
         return localrepository(ui, path, create)
 

mercurial/hgweb.py

@@ -1,722 +1,722 b''
 # hgweb.py - web interface to a mercurial repository
 #
 # Copyright 21 May 2005 - (c) 2005 Jake Edge <[e-mail]>
 # 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 os, cgi, time, re, difflib, sys, zlib
 from mercurial.hg import *
 from mercurial.ui import *
 
 def templatepath():
     for f in "templates", "../templates":
         p = os.path.join(os.path.dirname(__file__), f)
         if os.path.isdir(p): return p
 
 def age(t):
     def plural(t, c):
         if c == 1: return t
         return t + "s"
     def fmt(t, c):
         return "%d %s" % (c, plural(t, c))
 
     now = time.time()
     delta = max(1, int(now - t))
 
     scales = [["second", 1],
               ["minute", 60],
               ["hour", 3600],
               ["day", 3600 * 24],
               ["week", 3600 * 24 * 7],
               ["month", 3600 * 24 * 30],
               ["year", 3600 * 24 * 365]]
 
     scales.reverse()
 
     for t, s in scales:
         n = delta / s
         if n >= 2 or s == 1: return fmt(t, n)
 
 def nl2br(text):
     return text.replace('\n', '<br/>\n')
 
 def obfuscate(text):
     return ''.join([ '&#%d;' % ord(c) for c in text ])
 
 def up(p):
     if p[0] != "/": p = "/" + p
     if p[-1] == "/": p = p[:-1]
     up = os.path.dirname(p)
     if up == "/":
         return "/"
     return up + "/"
 
 def httphdr(type):
     print 'Content-type: %s\n' % type
 
 def write(*things):
     for thing in things:
         if hasattr(thing, "__iter__"):
             for part in thing:
                 write(part)
         else:
             sys.stdout.write(str(thing))
 
 def template(tmpl, filters = {}, **map):
     while tmpl:
         m = re.search(r"#([a-zA-Z0-9]+)((\|[a-zA-Z0-9]+)*)#", tmpl)
         if m:
             yield tmpl[:m.start(0)]
             v = map.get(m.group(1), "")
             v = callable(v) and v() or v
 
             fl = m.group(2)
             if fl:
                 for f in fl.split("|")[1:]:
                     v = filters[f](v)
 
             yield v
             tmpl = tmpl[m.end(0):]
         else:
             yield tmpl
             return
 
 class templater:
     def __init__(self, mapfile, filters = {}):
         self.cache = {}
         self.map = {}
         self.base = os.path.dirname(mapfile)
         self.filters = filters
 
         for l in file(mapfile):
             m = re.match(r'(\S+)\s*=\s*"(.*)"$', l)
             if m:
                 self.cache[m.group(1)] = m.group(2)
             else:
                 m = re.match(r'(\S+)\s*=\s*(\S+)', l)
                 if m:
                     self.map[m.group(1)] = os.path.join(self.base, m.group(2))
                 else:
                     raise "unknown map entry '%s'"  % l
 
     def __call__(self, t, **map):
         try:
             tmpl = self.cache[t]
         except KeyError:
             tmpl = self.cache[t] = file(self.map[t]).read()
         return template(tmpl, self.filters, **map)
 
 class hgweb:
     maxchanges = 10
     maxfiles = 10
 
     def __init__(self, path, name, templates = ""):
         self.templates = templates or templatepath()
         self.reponame = name
         self.path = path
         self.mtime = -1
         self.viewonly = 0
 
         self.filters = {
             "escape": cgi.escape,
             "age": age,
             "date": (lambda x: time.asctime(time.gmtime(x))),
             "addbreaks": nl2br,
             "obfuscate": obfuscate,
             "short": (lambda x: x[:12]),
             "firstline": (lambda x: x.splitlines(1)[0]),
             "permissions": (lambda x: x and "-rwxr-xr-x" or "-rw-r--r--")
             }
 
     def refresh(self):
         s = os.stat(os.path.join(self.path, ".hg", "00changelog.i"))
         if s.st_mtime != self.mtime:
             self.mtime = s.st_mtime
             self.repo = repository(ui(), self.path)
 
     def date(self, cs):
         return time.asctime(time.gmtime(float(cs[2].split(' ')[0])))
 
     def listfiles(self, files, mf):
         for f in files[:self.maxfiles]:
             yield self.t("filenodelink", node = hex(mf[f]), file = f)
         if len(files) > self.maxfiles:
             yield self.t("fileellipses")
 
     def listfilediffs(self, files, changeset):
         for f in files[:self.maxfiles]:
             yield self.t("filedifflink", node = hex(changeset), file = f)
         if len(files) > self.maxfiles:
             yield self.t("fileellipses")
 
     def parent(self, t1, node=nullid, rev=-1, **args):
         if node != hex(nullid):
             yield self.t(t1, node = node, rev = rev, **args)
 
     def diff(self, node1, node2, files):
         def filterfiles(list, files):
             l = [ x for x in list if x in files ]
 
             for f in files:
                 if f[-1] != os.sep: f += os.sep
                 l += [ x for x in list if x.startswith(f) ]
             return l
 
         parity = [0]
         def diffblock(diff, f, fn):
             yield self.t("diffblock",
                          lines = prettyprintlines(diff),
                          parity = parity[0],
                          file = f,
                          filenode = hex(fn or nullid))
             parity[0] = 1 - parity[0]
 
         def prettyprintlines(diff):
             for l in diff.splitlines(1):
                 if l.startswith('+'):
                     yield self.t("difflineplus", line = l)
                 elif l.startswith('-'):
                     yield self.t("difflineminus", line = l)
                 elif l.startswith('@'):
                     yield self.t("difflineat", line = l)
                 else:
                     yield self.t("diffline", line = l)
 
         r = self.repo
         cl = r.changelog
         mf = r.manifest
         change1 = cl.read(node1)
         change2 = cl.read(node2)
         mmap1 = mf.read(change1[0])
         mmap2 = mf.read(change2[0])
         date1 = self.date(change1)
         date2 = self.date(change2)
 
-        c, a, d = r.diffrevs(node1, node2)
+        c, a, d = r.changes(node1, node2)
         c, a, d = map(lambda x: filterfiles(x, files), (c, a, d))
 
         for f in c:
             to = r.file(f).read(mmap1[f])
             tn = r.file(f).read(mmap2[f])
             yield diffblock(mdiff.unidiff(to, date1, tn, date2, f), f, tn)
         for f in a:
             to = None
             tn = r.file(f).read(mmap2[f])
             yield diffblock(mdiff.unidiff(to, date1, tn, date2, f), f, tn)
         for f in d:
             to = r.file(f).read(mmap1[f])
             tn = None
             yield diffblock(mdiff.unidiff(to, date1, tn, date2, f), f, tn)
 
     def header(self):
         yield self.t("header", repo = self.reponame)
 
     def footer(self):
         yield self.t("footer", repo = self.reponame)
 
     def changelog(self, pos):
         def changenav():
             def seq(factor = 1):
                 yield 1 * factor
                 yield 3 * factor
                 #yield 5 * factor
                 for f in seq(factor * 10):
                     yield f
 
             l = []
             for f in seq():
                 if f < self.maxchanges / 2: continue
                 if f > count: break
                 r = "%d" % f
                 if pos + f < count: l.append(("+" + r, pos + f))
                 if pos - f >= 0: l.insert(0, ("-" + r, pos - f))
 
             yield self.t("naventry", rev = 0, label="(0)")
 
             for label, rev in l:
                 yield self.t("naventry", label = label, rev = rev)
 
             yield self.t("naventry", label="tip")
 
         def changelist():
             parity = (start - end) & 1
             cl = self.repo.changelog
             l = [] # build a list in forward order for efficiency
             for i in range(start, end):
                 n = cl.node(i)
                 changes = cl.read(n)
                 hn = hex(n)
                 p1, p2 = cl.parents(n)
                 t = float(changes[2].split(' ')[0])
 
                 l.insert(0, self.t(
                     'changelogentry',
                     parity = parity,
                     author = changes[1],
                     parent1 = self.parent("changelogparent",
                                           hex(p1), cl.rev(p1)),
                     parent2 = self.parent("changelogparent",
                                           hex(p2), cl.rev(p2)),
                     p1 = hex(p1), p2 = hex(p2),
                     p1rev = cl.rev(p1), p2rev = cl.rev(p2),
                     manifest = hex(changes[0]),
                     desc = changes[4],
                     date = t,
                     files = self.listfilediffs(changes[3], n),
                     rev = i,
                     node = hn))
                 parity = 1 - parity
 
             yield l
 
         cl = self.repo.changelog
         mf = cl.read(cl.tip())[0]
         count = cl.count()
         start = max(0, pos - self.maxchanges + 1)
         end = min(count, start + self.maxchanges)
         pos = end - 1
 
         yield self.t('changelog',
                      header = self.header(),
                      footer = self.footer(),
                      repo = self.reponame,
                      changenav = changenav,
                      manifest = hex(mf),
                      rev = pos, changesets = count, entries = changelist)
 
     def changeset(self, nodeid):
         n = bin(nodeid)
         cl = self.repo.changelog
         changes = cl.read(n)
         p1, p2 = cl.parents(n)
         p1rev, p2rev = cl.rev(p1), cl.rev(p2)
         t = float(changes[2].split(' ')[0])
 
         files = []
         mf = self.repo.manifest.read(changes[0])
         for f in changes[3]:
             files.append(self.t("filenodelink",
                                 filenode = hex(mf.get(f, nullid)), file = f))
 
         def diff():
             yield self.diff(p1, n, changes[3])
 
         yield self.t('changeset',
                      header = self.header(),
                      footer = self.footer(),
                      repo = self.reponame,
                      diff = diff,
                      rev = cl.rev(n),
                      node = nodeid,
                      parent1 = self.parent("changesetparent",
                                            hex(p1), cl.rev(p1)),
                      parent2 = self.parent("changesetparent",
                                            hex(p2), cl.rev(p2)),
                      p1 = hex(p1), p2 = hex(p2),
                      p1rev = cl.rev(p1), p2rev = cl.rev(p2),
                      manifest = hex(changes[0]),
                      author = changes[1],
                      desc = changes[4],
                      date = t,
                      files = files)
 
     def filelog(self, f, filenode):
         cl = self.repo.changelog
         fl = self.repo.file(f)
         count = fl.count()
 
         def entries():
             l = []
             parity = (count - 1) & 1
 
             for i in range(count):
 
                 n = fl.node(i)
                 lr = fl.linkrev(n)
                 cn = cl.node(lr)
                 cs = cl.read(cl.node(lr))
                 p1, p2 = fl.parents(n)
                 t = float(cs[2].split(' ')[0])
 
                 l.insert(0, self.t("filelogentry",
                                    parity = parity,
                                    filenode = hex(n),
                                    filerev = i,
                                    file = f,
                                    node = hex(cn),
                                    author = cs[1],
                                    date = t,
                                    desc = cs[4],
                                    p1 = hex(p1), p2 = hex(p2),
                                    p1rev = fl.rev(p1), p2rev = fl.rev(p2)))
                 parity = 1 - parity
 
             yield l
 
         yield self.t("filelog",
                      header = self.header(),
                      footer = self.footer(),
                      repo = self.reponame,
                      file = f,
                      filenode = filenode,
                      entries = entries)
 
     def filerevision(self, f, node):
         fl = self.repo.file(f)
         n = bin(node)
         text = fl.read(n)
         changerev = fl.linkrev(n)
         cl = self.repo.changelog
         cn = cl.node(changerev)
         cs = cl.read(cn)
         p1, p2 = fl.parents(n)
         t = float(cs[2].split(' ')[0])
         mfn = cs[0]
 
         def lines():
             for l, t in enumerate(text.splitlines(1)):
                 yield self.t("fileline", line = t,
                              linenumber = "% 6d" % (l + 1),
                              parity = l & 1)
 
         yield self.t("filerevision", file = f,
                      header = self.header(),
                      footer = self.footer(),
                      repo = self.reponame,
                      filenode = node,
                      path = up(f),
                      text = lines(),
                      rev = changerev,
                      node = hex(cn),
                      manifest = hex(mfn),
                      author = cs[1],
                      date = t,
                      parent1 = self.parent("filerevparent",
                                            hex(p1), fl.rev(p1), file=f),
                      parent2 = self.parent("filerevparent",
                                            hex(p2), fl.rev(p2), file=f),
                      p1 = hex(p1), p2 = hex(p2),
                      permissions = self.repo.manifest.readflags(mfn)[f],
                      p1rev = fl.rev(p1), p2rev = fl.rev(p2))
 
     def fileannotate(self, f, node):
         bcache = {}
         ncache = {}
         fl = self.repo.file(f)
         n = bin(node)
         changerev = fl.linkrev(n)
 
         cl = self.repo.changelog
         cn = cl.node(changerev)
         cs = cl.read(cn)
         p1, p2 = fl.parents(n)
         t = float(cs[2].split(' ')[0])
         mfn = cs[0]
 
         def annotate():
             parity = 1
             last = None
             for r, l in fl.annotate(n):
                 try:
                     cnode = ncache[r]
                 except KeyError:
                     cnode = ncache[r] = self.repo.changelog.node(r)
 
                 try:
                     name = bcache[r]
                 except KeyError:
                     cl = self.repo.changelog.read(cnode)
                     name = cl[1]
                     f = name.find('@')
                     if f >= 0:
                         name = name[:f]
                     f = name.find('<')
                     if f >= 0:
                         name = name[f+1:]
                     bcache[r] = name
 
                 if last != cnode:
                     parity = 1 - parity
                     last = cnode
 
                 yield self.t("annotateline",
                              parity = parity,
                              node = hex(cnode),
                              rev = r,
                              author = name,
                              file = f,
                              line = l)
 
         yield self.t("fileannotate",
                      header = self.header(),
                      footer = self.footer(),
                      repo = self.reponame,
                      file = f,
                      filenode = node,
                      annotate = annotate,
                      path = up(f),
                      rev = changerev,
                      node = hex(cn),
                      manifest = hex(mfn),
                      author = cs[1],
                      date = t,
                      parent1 = self.parent("fileannotateparent",
                                            hex(p1), fl.rev(p1), file=f),
                      parent2 = self.parent("fileannotateparent",
                                            hex(p2), fl.rev(p2), file=f),
                      p1 = hex(p1), p2 = hex(p2),
                      permissions = self.repo.manifest.readflags(mfn)[f],
                      p1rev = fl.rev(p1), p2rev = fl.rev(p2))
 
     def manifest(self, mnode, path):
         mf = self.repo.manifest.read(bin(mnode))
         rev = self.repo.manifest.rev(bin(mnode))
         node = self.repo.changelog.node(rev)
         mff=self.repo.manifest.readflags(bin(mnode))
 
         files = {}
 
         p = path[1:]
         l = len(p)
 
         for f,n in mf.items():
             if f[:l] != p:
                 continue
             remain = f[l:]
             if "/" in remain:
                 short = remain[:remain.find("/") + 1] # bleah
                 files[short] = (f, None)
             else:
                 short = os.path.basename(remain)
                 files[short] = (f, n)
 
         def filelist():
             parity = 0
             fl = files.keys()
             fl.sort()
             for f in fl:
                 full, fnode = files[f]
                 if fnode:
                     yield self.t("manifestfileentry",
                                  file = full,
                                  manifest = mnode,
                                  filenode = hex(fnode),
                                  parity = parity,
                                  basename = f,
                                  permissions = mff[full])
                 else:
                     yield self.t("manifestdirentry",
                                  parity = parity,
                                  path = os.path.join(path, f),
                                  manifest = mnode, basename = f[:-1])
                 parity = 1 - parity
 
         yield self.t("manifest",
                      header = self.header(),
                      footer = self.footer(),
                      repo = self.reponame,
                      manifest = mnode,
                      rev = rev,
                      node = hex(node),
                      path = path,
                      up = up(path),
                      entries = filelist)
 
     def tags(self):
         cl = self.repo.changelog
         mf = cl.read(cl.tip())[0]
 
         i = self.repo.tagslist()
         i.reverse()
 
         def entries():
             parity = 0
             for k,n in i:
                 yield self.t("tagentry",
                              parity = parity,
                              tag = k,
                              node = hex(n))
                 parity = 1 - parity
 
         yield self.t("tags",
                      header = self.header(),
                      footer = self.footer(),
                      repo = self.reponame,
                      manifest = hex(mf),
                      entries = entries)
 
     def filediff(self, file, changeset):
         n = bin(changeset)
         cl = self.repo.changelog
         p1 = cl.parents(n)[0]
         cs = cl.read(n)
         mf = self.repo.manifest.read(cs[0])
 
         def diff():
             yield self.diff(p1, n, file)
 
         yield self.t("filediff",
                      header = self.header(),
                      footer = self.footer(),
                      repo = self.reponame,
                      file = file,
                      filenode = hex(mf.get(file, nullid)),
                      node = changeset,
                      rev = self.repo.changelog.rev(n),
                      p1 = hex(p1),
                      p1rev = self.repo.changelog.rev(p1),
                      diff = diff)
 
     # add tags to things
     # tags -> list of changesets corresponding to tags
     # find tag, changeset, file
 
     def run(self):
         self.refresh()
         args = cgi.parse()
 
         m = os.path.join(self.templates, "map")
         if args.has_key('style'):
             b = os.path.basename("map-" + args['style'][0])
             p = os.path.join(self.templates, b)
             if os.path.isfile(p): m = p
 
         self.t = templater(m, self.filters)
 
         if not args.has_key('cmd') or args['cmd'][0] == 'changelog':
             hi = self.repo.changelog.count() - 1
             if args.has_key('rev'):
                 hi = args['rev'][0]
                 try:
                     hi = self.repo.changelog.rev(self.repo.lookup(hi))
                 except KeyError: pass
 
             write(self.changelog(hi))
 
         elif args['cmd'][0] == 'changeset':
             write(self.changeset(args['node'][0]))
 
         elif args['cmd'][0] == 'manifest':
             write(self.manifest(args['manifest'][0], args['path'][0]))
 
         elif args['cmd'][0] == 'tags':
             write(self.tags())
 
         elif args['cmd'][0] == 'filediff':
             write(self.filediff(args['file'][0], args['node'][0]))
 
         elif args['cmd'][0] == 'file':
             write(self.filerevision(args['file'][0], args['filenode'][0]))
 
         elif args['cmd'][0] == 'annotate':
             write(self.fileannotate(args['file'][0], args['filenode'][0]))
 
         elif args['cmd'][0] == 'filelog':
             write(self.filelog(args['file'][0], args['filenode'][0]))
 
         elif args['cmd'][0] == 'heads':
             httphdr("text/plain")
             h = self.repo.heads()
             sys.stdout.write(" ".join(map(hex, h)) + "\n")
 
         elif args['cmd'][0] == 'branches':
             httphdr("text/plain")
             nodes = []
             if args.has_key('nodes'):
                 nodes = map(bin, args['nodes'][0].split(" "))
             for b in self.repo.branches(nodes):
                 sys.stdout.write(" ".join(map(hex, b)) + "\n")
 
         elif args['cmd'][0] == 'between':
             httphdr("text/plain")
             nodes = []
             if args.has_key('pairs'):
                 pairs = [ map(bin, p.split("-"))
                           for p in args['pairs'][0].split(" ") ]
             for b in self.repo.between(pairs):
                 sys.stdout.write(" ".join(map(hex, b)) + "\n")
 
         elif args['cmd'][0] == 'changegroup':
             httphdr("application/hg-changegroup")
             nodes = []
             if self.viewonly:
                 return
 
             if args.has_key('roots'):
                 nodes = map(bin, args['roots'][0].split(" "))
 
             z = zlib.compressobj()
             for chunk in self.repo.changegroup(nodes):
                 sys.stdout.write(z.compress(chunk))
 
             sys.stdout.write(z.flush())
 
         else:
             write(self.t("error"))
 
 def server(path, name, templates, address, port):
 
     import BaseHTTPServer
     import sys, os
 
     class hgwebhandler(BaseHTTPServer.BaseHTTPRequestHandler):
         def do_POST(self):
             try:
                 self.do_hgweb()
             except socket.error, inst:
                 if inst.args[0] != 32: raise
 
         def do_GET(self):
             self.do_POST()
 
         def do_hgweb(self):
             query = ""
             p = self.path.find("?")
             if p:
                 query = self.path[p + 1:]
                 query = query.replace('+', ' ')
 
             env = {}
             env['GATEWAY_INTERFACE'] = 'CGI/1.1'
             env['REQUEST_METHOD'] = self.command
             if query:
                 env['QUERY_STRING'] = query
             host = self.address_string()
             if host != self.client_address[0]:
                 env['REMOTE_HOST'] = host
                 env['REMOTE_ADDR'] = self.client_address[0]
 
             if self.headers.typeheader is None:
                 env['CONTENT_TYPE'] = self.headers.type
             else:
                 env['CONTENT_TYPE'] = self.headers.typeheader
             length = self.headers.getheader('content-length')
             if length:
                 env['CONTENT_LENGTH'] = length
             accept = []
             for line in self.headers.getallmatchingheaders('accept'):
                 if line[:1] in "\t\n\r ":
                     accept.append(line.strip())
                 else:
                     accept = accept + line[7:].split(',')
             env['HTTP_ACCEPT'] = ','.join(accept)
 
             os.environ.update(env)
 
             save = sys.argv, sys.stdin, sys.stdout, sys.stderr
             try:
                 sys.stdin = self.rfile
                 sys.stdout = self.wfile
                 sys.argv = ["hgweb.py"]
                 if '=' not in query:
                     sys.argv.append(query)
                 self.send_response(200, "Script output follows")
                 hg.run()
             finally:
                 sys.argv, sys.stdin, sys.stdout, sys.stderr = save
 
     hg = hgweb(path, name, templates)
     httpd = BaseHTTPServer.HTTPServer((address, port), hgwebhandler)
     httpd.serve_forever()

tests/test-bad-pull.out

@@ -1,20 +1,21 b''
 + hg clone http://localhost:20059/ copy
 requesting all changes
 adding changesets
 abort: error 111: Connection refused
 transaction abort!
 rollback completed
 + echo 255
 255
 + ls copy
+ls: copy: No such file or directory
 + cat
 + python dumb.py
 + hg clone http://localhost:20059/foo copy2
 requesting all changes
 adding changesets
 abort: HTTP Error 404: File not found
 transaction abort!
 rollback completed
 + echo 255
 255
 + set +x