manifest: repurpose flagsdiff() into (node-and-flag)diff()

The manifestdict class already has a method for diff flags between two
manifests (presumably because there is no full access to the private
_flags field). The only caller is merge.manifestmerge(), which also
wants a diff of files between the same manifests. Let's combine the
code for diffing files and flags into a single method on
manifestdict. This puts all the manifest diffing in one place and will
allow for further simplification. It might also be useful for it to be
encapsulated in manifestdict if we later decide to to shard
manifests. The docstring is intentionally unclear about missing
entries for now.

mercurial/manifest.py

@@ -1,245 +1,252
 # manifest.py - manifest revision class for mercurial
 #
 # Copyright 2005-2007 Matt Mackall <[e-mail]>
 #
 # This software may be used and distributed according to the terms of the
 # GNU General Public License version 2 or any later version.
 
 from i18n import _
 import mdiff, parsers, error, revlog, util, dicthelpers
 import array, struct
 
 class manifestdict(dict):
     def __init__(self, mapping=None, flags=None):
         if mapping is None:
             mapping = {}
         if flags is None:
             flags = {}
         dict.__init__(self, mapping)
         self._flags = flags
     def flags(self, f):
         return self._flags.get(f, "")
     def withflags(self):
         return set(self._flags.keys())
     def setflag(self, f, flags):
         """Set the flags (symlink, executable) for path f."""
         self._flags[f] = flags
     def copy(self):
         return manifestdict(self, dict.copy(self._flags))
     def intersectfiles(self, files):
         '''make a new manifestdict with the intersection of self with files
 
         The algorithm assumes that files is much smaller than self.'''
         ret = manifestdict()
         for fn in files:
             if fn in self:
                 ret[fn] = self[fn]
                 flags = self._flags.get(fn, None)
                 if flags:
                     ret._flags[fn] = flags
         return ret
-    def flagsdiff(self, d2):
-        return dicthelpers.diff(self._flags, d2._flags, "")
+
+    def diff(self, m2):
+        '''Finds changes between the current manifest and m2. The result is
+        returned as a dict with filename as key and values of the form
+        ((n1,n2),(fl1,fl2)), where n1/n2 is the nodeid in the current/other
+        manifest and fl1/fl2 is the flag in the current/other manifest.'''
+        flagsdiff = dicthelpers.diff(self._flags, m2._flags, "")
+        fdiff = dicthelpers.diff(self, m2)
+        return dicthelpers.join(fdiff, flagsdiff)
 
     def text(self):
         """Get the full data of this manifest as a bytestring."""
         fl = sorted(self)
         _checkforbidden(fl)
 
         hex, flags = revlog.hex, self.flags
         # if this is changed to support newlines in filenames,
         # be sure to check the templates/ dir again (especially *-raw.tmpl)
         return ''.join("%s\0%s%s\n" % (f, hex(self[f]), flags(f)) for f in fl)
 
     def fastdelta(self, base, changes):
         """Given a base manifest text as an array.array and a list of changes
         relative to that text, compute a delta that can be used by revlog.
         """
         delta = []
         dstart = None
         dend = None
         dline = [""]
         start = 0
         # zero copy representation of base as a buffer
         addbuf = util.buffer(base)
 
         # start with a readonly loop that finds the offset of
         # each line and creates the deltas
         for f, todelete in changes:
             # bs will either be the index of the item or the insert point
             start, end = _msearch(addbuf, f, start)
             if not todelete:
                 l = "%s\0%s%s\n" % (f, revlog.hex(self[f]), self.flags(f))
             else:
                 if start == end:
                     # item we want to delete was not found, error out
                     raise AssertionError(
                             _("failed to remove %s from manifest") % f)
                 l = ""
             if dstart is not None and dstart <= start and dend >= start:
                 if dend < end:
                     dend = end
                 if l:
                     dline.append(l)
             else:
                 if dstart is not None:
                     delta.append([dstart, dend, "".join(dline)])
                 dstart = start
                 dend = end
                 dline = [l]
 
         if dstart is not None:
             delta.append([dstart, dend, "".join(dline)])
         # apply the delta to the base, and get a delta for addrevision
         deltatext, arraytext = _addlistdelta(base, delta)
         return arraytext, deltatext
 
 def _msearch(m, s, lo=0, hi=None):
     '''return a tuple (start, end) that says where to find s within m.
 
     If the string is found m[start:end] are the line containing
     that string.  If start == end the string was not found and
     they indicate the proper sorted insertion point.
 
     m should be a buffer or a string
     s is a string'''
     def advance(i, c):
         while i < lenm and m[i] != c:
             i += 1
         return i
     if not s:
         return (lo, lo)
     lenm = len(m)
     if not hi:
         hi = lenm
     while lo < hi:
         mid = (lo + hi) // 2
         start = mid
         while start > 0 and m[start - 1] != '\n':
             start -= 1
         end = advance(start, '\0')
         if m[start:end] < s:
             # we know that after the null there are 40 bytes of sha1
             # this translates to the bisect lo = mid + 1
             lo = advance(end + 40, '\n') + 1
         else:
             # this translates to the bisect hi = mid
             hi = start
     end = advance(lo, '\0')
     found = m[lo:end]
     if s == found:
         # we know that after the null there are 40 bytes of sha1
         end = advance(end + 40, '\n')
         return (lo, end + 1)
     else:
         return (lo, lo)
 
 def _checkforbidden(l):
     """Check filenames for illegal characters."""
     for f in l:
         if '\n' in f or '\r' in f:
             raise error.RevlogError(
                 _("'\\n' and '\\r' disallowed in filenames: %r") % f)
 
 
 # apply the changes collected during the bisect loop to our addlist
 # return a delta suitable for addrevision
 def _addlistdelta(addlist, x):
     # for large addlist arrays, building a new array is cheaper
     # than repeatedly modifying the existing one
     currentposition = 0
     newaddlist = array.array('c')
 
     for start, end, content in x:
         newaddlist += addlist[currentposition:start]
         if content:
             newaddlist += array.array('c', content)
 
         currentposition = end
 
     newaddlist += addlist[currentposition:]
 
     deltatext = "".join(struct.pack(">lll", start, end, len(content))
                    + content for start, end, content in x)
     return deltatext, newaddlist
 
 def _parse(lines):
     mfdict = manifestdict()
     parsers.parse_manifest(mfdict, mfdict._flags, lines)
     return mfdict
 
 class manifest(revlog.revlog):
     def __init__(self, opener):
         # we expect to deal with not more than four revs at a time,
         # during a commit --amend
         self._mancache = util.lrucachedict(4)
         revlog.revlog.__init__(self, opener, "00manifest.i")
 
     def readdelta(self, node):
         r = self.rev(node)
         return _parse(mdiff.patchtext(self.revdiff(self.deltaparent(r), r)))
 
     def readfast(self, node):
         '''use the faster of readdelta or read'''
         r = self.rev(node)
         deltaparent = self.deltaparent(r)
         if deltaparent != revlog.nullrev and deltaparent in self.parentrevs(r):
             return self.readdelta(node)
         return self.read(node)
 
     def read(self, node):
         if node == revlog.nullid:
             return manifestdict() # don't upset local cache
         if node in self._mancache:
             return self._mancache[node][0]
         text = self.revision(node)
         arraytext = array.array('c', text)
         mapping = _parse(text)
         self._mancache[node] = (mapping, arraytext)
         return mapping
 
     def find(self, node, f):
         '''look up entry for a single file efficiently.
         return (node, flags) pair if found, (None, None) if not.'''
         if node in self._mancache:
             mapping = self._mancache[node][0]
             return mapping.get(f), mapping.flags(f)
         text = self.revision(node)
         start, end = _msearch(text, f)
         if start == end:
             return None, None
         l = text[start:end]
         f, n = l.split('\0')
         return revlog.bin(n[:40]), n[40:-1]
 
     def add(self, map, transaction, link, p1, p2, added, removed):
         if p1 in self._mancache:
             # If our first parent is in the manifest cache, we can
             # compute a delta here using properties we know about the
             # manifest up-front, which may save time later for the
             # revlog layer.
 
             _checkforbidden(added)
             # combine the changed lists into one list for sorting
             work = [(x, False) for x in added]
             work.extend((x, True) for x in removed)
             # this could use heapq.merge() (from Python 2.6+) or equivalent
             # since the lists are already sorted
             work.sort()
 
             arraytext, deltatext = map.fastdelta(self._mancache[p1][1], work)
             cachedelta = self.rev(p1), deltatext
             text = util.buffer(arraytext)
         else:
             # The first parent manifest isn't already loaded, so we'll
             # just encode a fulltext of the manifest and pass that
             # through to the revlog layer, and let it handle the delta
             # process.
             text = map.text()
             arraytext = array.array('c', text)
             cachedelta = None
 
         n = self.addrevision(text, transaction, link, p1, p2, cachedelta)
         self._mancache[n] = (map, arraytext)
 
         return n

mercurial/merge.py

@@ -1,1177 +1,1175
 # merge.py - directory-level update/merge handling for Mercurial
 #
 # Copyright 2006, 2007 Matt Mackall <[e-mail]>
 #
 # This software may be used and distributed according to the terms of the
 # GNU General Public License version 2 or any later version.
 
 import struct
 
 from node import nullid, nullrev, hex, bin
 from i18n import _
 from mercurial import obsolete
-import error as errormod, util, filemerge, copies, subrepo, worker, dicthelpers
+import error as errormod, util, filemerge, copies, subrepo, worker
 import errno, os, shutil
 
 _pack = struct.pack
 _unpack = struct.unpack
 
 def _droponode(data):
     # used for compatibility for v1
     bits = data.split("\0")
     bits = bits[:-2] + bits[-1:]
     return "\0".join(bits)
 
 class mergestate(object):
     '''track 3-way merge state of individual files
 
     it is stored on disk when needed. Two file are used, one with an old
     format, one with a new format. Both contains similar data, but the new
     format can store new kind of field.
 
     Current new format is a list of arbitrary record of the form:
 
         [type][length][content]
 
     Type is a single character, length is a 4 bytes integer, content is an
     arbitrary suites of bytes of length `length`.
 
     Type should be a letter. Capital letter are mandatory record, Mercurial
     should abort if they are unknown. lower case record can be safely ignored.
 
     Currently known record:
 
     L: the node of the "local" part of the merge (hexified version)
     O: the node of the "other" part of the merge (hexified version)
     F: a file to be merged entry
     '''
     statepathv1 = "merge/state"
     statepathv2 = "merge/state2"
 
     def __init__(self, repo):
         self._repo = repo
         self._dirty = False
         self._read()
 
     def reset(self, node=None, other=None):
         self._state = {}
         self._local = None
         self._other = None
         if node:
             self._local = node
             self._other = other
         shutil.rmtree(self._repo.join("merge"), True)
         self._dirty = False
 
     def _read(self):
         """Analyse each record content to restore a serialized state from disk
 
         This function process "record" entry produced by the de-serialization
         of on disk file.
         """
         self._state = {}
         self._local = None
         self._other = None
         records = self._readrecords()
         for rtype, record in records:
             if rtype == 'L':
                 self._local = bin(record)
             elif rtype == 'O':
                 self._other = bin(record)
             elif rtype == "F":
                 bits = record.split("\0")
                 self._state[bits[0]] = bits[1:]
             elif not rtype.islower():
                 raise util.Abort(_('unsupported merge state record: %s')
                                    % rtype)
         self._dirty = False
 
     def _readrecords(self):
         """Read merge state from disk and return a list of record (TYPE, data)
 
         We read data from both v1 and v2 files and decide which one to use.
 
         V1 has been used by version prior to 2.9.1 and contains less data than
         v2. We read both versions and check if no data in v2 contradicts
         v1. If there is not contradiction we can safely assume that both v1
         and v2 were written at the same time and use the extract data in v2. If
         there is contradiction we ignore v2 content as we assume an old version
         of Mercurial has overwritten the mergestate file and left an old v2
         file around.
 
         returns list of record [(TYPE, data), ...]"""
         v1records = self._readrecordsv1()
         v2records = self._readrecordsv2()
         oldv2 = set() # old format version of v2 record
         for rec in v2records:
             if rec[0] == 'L':
                 oldv2.add(rec)
             elif rec[0] == 'F':
                 # drop the onode data (not contained in v1)
                 oldv2.add(('F', _droponode(rec[1])))
         for rec in v1records:
             if rec not in oldv2:
                 # v1 file is newer than v2 file, use it
                 # we have to infer the "other" changeset of the merge
                 # we cannot do better than that with v1 of the format
                 mctx = self._repo[None].parents()[-1]
                 v1records.append(('O', mctx.hex()))
                 # add place holder "other" file node information
                 # nobody is using it yet so we do no need to fetch the data
                 # if mctx was wrong `mctx[bits[-2]]` may fails.
                 for idx, r in enumerate(v1records):
                     if r[0] == 'F':
                         bits = r[1].split("\0")
                         bits.insert(-2, '')
                         v1records[idx] = (r[0], "\0".join(bits))
                 return v1records
         else:
             return v2records
 
     def _readrecordsv1(self):
         """read on disk merge state for version 1 file
 
         returns list of record [(TYPE, data), ...]
 
         Note: the "F" data from this file are one entry short
               (no "other file node" entry)
         """
         records = []
         try:
             f = self._repo.opener(self.statepathv1)
             for i, l in enumerate(f):
                 if i == 0:
                     records.append(('L', l[:-1]))
                 else:
                     records.append(('F', l[:-1]))
             f.close()
         except IOError, err:
             if err.errno != errno.ENOENT:
                 raise
         return records
 
     def _readrecordsv2(self):
         """read on disk merge state for version 2 file
 
         returns list of record [(TYPE, data), ...]
         """
         records = []
         try:
             f = self._repo.opener(self.statepathv2)
             data = f.read()
             off = 0
             end = len(data)
             while off < end:
                 rtype = data[off]
                 off += 1
                 length = _unpack('>I', data[off:(off + 4)])[0]
                 off += 4
                 record = data[off:(off + length)]
                 off += length
                 records.append((rtype, record))
             f.close()
         except IOError, err:
             if err.errno != errno.ENOENT:
                 raise
         return records
 
     def active(self):
         """Whether mergestate is active.
 
         Returns True if there appears to be mergestate. This is a rough proxy
         for "is a merge in progress."
         """
         # Check local variables before looking at filesystem for performance
         # reasons.
         return bool(self._local) or bool(self._state) or \
                self._repo.opener.exists(self.statepathv1) or \
                self._repo.opener.exists(self.statepathv2)
 
     def commit(self):
         """Write current state on disk (if necessary)"""
         if self._dirty:
             records = []
             records.append(("L", hex(self._local)))
             records.append(("O", hex(self._other)))
             for d, v in self._state.iteritems():
                 records.append(("F", "\0".join([d] + v)))
             self._writerecords(records)
             self._dirty = False
 
     def _writerecords(self, records):
         """Write current state on disk (both v1 and v2)"""
         self._writerecordsv1(records)
         self._writerecordsv2(records)
 
     def _writerecordsv1(self, records):
         """Write current state on disk in a version 1 file"""
         f = self._repo.opener(self.statepathv1, "w")
         irecords = iter(records)
         lrecords = irecords.next()
         assert lrecords[0] == 'L'
         f.write(hex(self._local) + "\n")
         for rtype, data in irecords:
             if rtype == "F":
                 f.write("%s\n" % _droponode(data))
         f.close()
 
     def _writerecordsv2(self, records):
         """Write current state on disk in a version 2 file"""
         f = self._repo.opener(self.statepathv2, "w")
         for key, data in records:
             assert len(key) == 1
             format = ">sI%is" % len(data)
             f.write(_pack(format, key, len(data), data))
         f.close()
 
     def add(self, fcl, fco, fca, fd):
         """add a new (potentially?) conflicting file the merge state
         fcl: file context for local,
         fco: file context for remote,
         fca: file context for ancestors,
         fd:  file path of the resulting merge.
 
         note: also write the local version to the `.hg/merge` directory.
         """
         hash = util.sha1(fcl.path()).hexdigest()
         self._repo.opener.write("merge/" + hash, fcl.data())
         self._state[fd] = ['u', hash, fcl.path(),
                            fca.path(), hex(fca.filenode()),
                            fco.path(), hex(fco.filenode()),
                            fcl.flags()]
         self._dirty = True
 
     def __contains__(self, dfile):
         return dfile in self._state
 
     def __getitem__(self, dfile):
         return self._state[dfile][0]
 
     def __iter__(self):
         return iter(sorted(self._state))
 
     def files(self):
         return self._state.keys()
 
     def mark(self, dfile, state):
         self._state[dfile][0] = state
         self._dirty = True
 
     def unresolved(self):
         """Obtain the paths of unresolved files."""
 
         for f, entry in self._state.items():
             if entry[0] == 'u':
                 yield f
 
     def resolve(self, dfile, wctx, labels=None):
         """rerun merge process for file path `dfile`"""
         if self[dfile] == 'r':
             return 0
         stateentry = self._state[dfile]
         state, hash, lfile, afile, anode, ofile, onode, flags = stateentry
         octx = self._repo[self._other]
         fcd = wctx[dfile]
         fco = octx[ofile]
         fca = self._repo.filectx(afile, fileid=anode)
         # "premerge" x flags
         flo = fco.flags()
         fla = fca.flags()
         if 'x' in flags + flo + fla and 'l' not in flags + flo + fla:
             if fca.node() == nullid:
                 self._repo.ui.warn(_('warning: cannot merge flags for %s\n') %
                                    afile)
             elif flags == fla:
                 flags = flo
         # restore local
         f = self._repo.opener("merge/" + hash)
         self._repo.wwrite(dfile, f.read(), flags)
         f.close()
         r = filemerge.filemerge(self._repo, self._local, lfile, fcd, fco, fca,
                                 labels=labels)
         if r is None:
             # no real conflict
             del self._state[dfile]
             self._dirty = True
         elif not r:
             self.mark(dfile, 'r')
         return r
 
 def _checkunknownfile(repo, wctx, mctx, f):
     return (not repo.dirstate._ignore(f)
         and os.path.isfile(repo.wjoin(f))
         and repo.wopener.audit.check(f)
         and repo.dirstate.normalize(f) not in repo.dirstate
         and mctx[f].cmp(wctx[f]))
 
 def _checkunknown(repo, wctx, mctx):
     "check for collisions between unknown files and files in mctx"
 
     error = False
     for f in mctx:
         if f not in wctx and _checkunknownfile(repo, wctx, mctx, f):
             error = True
             wctx._repo.ui.warn(_("%s: untracked file differs\n") % f)
     if error:
         raise util.Abort(_("untracked files in working directory differ "
                            "from files in requested revision"))
 
 def _forgetremoved(wctx, mctx, branchmerge):
     """
     Forget removed files
 
     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 we're merging, and the other revision has removed a file
     that is not present in the working directory, we need to mark it
     as removed.
     """
 
     ractions = []
     factions = xactions = []
     if branchmerge:
         xactions = ractions
     for f in wctx.deleted():
         if f not in mctx:
             xactions.append((f, None, "forget deleted"))
 
     if not branchmerge:
         for f in wctx.removed():
             if f not in mctx:
                 factions.append((f, None, "forget removed"))
 
     return ractions, factions
 
 def _checkcollision(repo, wmf, actions):
     # build provisional merged manifest up
     pmmf = set(wmf)
 
     if actions:
         # k, dr, e and rd are no-op
         for m in 'a', 'f', 'g', 'cd', 'dc':
             for f, args, msg in actions[m]:
                 pmmf.add(f)
         for f, args, msg in actions['r']:
             pmmf.discard(f)
         for f, args, msg in actions['dm']:
             f2, flags = args
             pmmf.discard(f2)
             pmmf.add(f)
         for f, args, msg in actions['dg']:
             f2, flags = args
             pmmf.add(f)
         for f, args, msg in actions['m']:
             f1, f2, fa, move, anc = args
             if move:
                 pmmf.discard(f1)
             pmmf.add(f)
 
     # check case-folding collision in provisional merged manifest
     foldmap = {}
     for f in sorted(pmmf):
         fold = util.normcase(f)
         if fold in foldmap:
             raise util.Abort(_("case-folding collision between %s and %s")
                              % (f, foldmap[fold]))
         foldmap[fold] = f
 
 def manifestmerge(repo, wctx, p2, pa, branchmerge, force, partial,
                   acceptremote, followcopies):
     """
     Merge p1 and p2 with ancestor pa and generate merge action list
 
     branchmerge and force are as passed in to update
     partial = function to filter file lists
     acceptremote = accept the incoming changes without prompting
     """
 
     actions = dict((m, []) for m in 'a f g cd dc r dm dg m dr e rd k'.split())
     copy, movewithdir = {}, {}
 
     # manifests fetched in order are going to be faster, so prime the caches
     [x.manifest() for x in
      sorted(wctx.parents() + [p2, pa], key=lambda x: x.rev())]
 
     if followcopies:
         ret = copies.mergecopies(repo, wctx, p2, pa)
         copy, movewithdir, diverge, renamedelete = ret
         for of, fl in diverge.iteritems():
             actions['dr'].append((of, (fl,), "divergent renames"))
         for of, fl in renamedelete.iteritems():
             actions['rd'].append((of, (fl,), "rename and delete"))
 
     repo.ui.note(_("resolving manifests\n"))
     repo.ui.debug(" branchmerge: %s, force: %s, partial: %s\n"
                   % (bool(branchmerge), bool(force), bool(partial)))
     repo.ui.debug(" ancestor: %s, local: %s, remote: %s\n" % (pa, wctx, p2))
 
     m1, m2, ma = wctx.manifest(), p2.manifest(), pa.manifest()
     copied = set(copy.values())
     copied.update(movewithdir.values())
 
     if '.hgsubstate' in m1:
         # check whether sub state is modified
         for s in sorted(wctx.substate):
             if wctx.sub(s).dirty():
                 m1['.hgsubstate'] += "+"
                 break
 
     aborts = []
     # Compare manifests
-    fdiff = dicthelpers.diff(m1, m2)
-    flagsdiff = m1.flagsdiff(m2)
-    diff12 = dicthelpers.join(fdiff, flagsdiff)
+    diff = m1.diff(m2)
 
-    for f, (n12, fl12) in diff12.iteritems():
+    for f, (n12, fl12) in diff.iteritems():
         if n12:
             n1, n2 = n12
         else: # file contents didn't change, but flags did
             n1 = n2 = m1.get(f, None)
             if n1 is None:
                 # Since n1 == n2, the file isn't present in m2 either. This
                 # means that the file was removed or deleted locally and
                 # removed remotely, but that residual entries remain in flags.
                 # This can happen in manifests generated by workingctx.
                 continue
         if fl12:
             fl1, fl2 = fl12
         else: # flags didn't change, file contents did
             fl1 = fl2 = m1.flags(f)
 
         if partial and not partial(f):
             continue
         if n1 and n2:
             fa = f
             a = ma.get(f, nullid)
             if a == nullid:
                 fa = copy.get(f, f)
                 # Note: f as default is wrong - we can't really make a 3-way
                 # merge without an ancestor file.
             fla = ma.flags(fa)
             nol = 'l' not in fl1 + fl2 + fla
             if n2 == a and fl2 == fla:
                 actions['k'].append((f, (), "keep")) # remote unchanged
             elif n1 == a and fl1 == fla: # local unchanged - use remote
                 if n1 == n2: # optimization: keep local content
                     actions['e'].append((f, (fl2,), "update permissions"))
                 else:
                     actions['g'].append((f, (fl2,), "remote is newer"))
             elif nol and n2 == a: # remote only changed 'x'
                 actions['e'].append((f, (fl2,), "update permissions"))
             elif nol and n1 == a: # local only changed 'x'
                 actions['g'].append((f, (fl1,), "remote is newer"))
             else: # both changed something
                 actions['m'].append((f, (f, f, fa, False, pa.node()),
                                "versions differ"))
         elif f in copied: # files we'll deal with on m2 side
             pass
         elif n1 and f in movewithdir: # directory rename, move local
             f2 = movewithdir[f]
             actions['dm'].append((f2, (f, fl1),
                             "remote directory rename - move from " + f))
         elif n1 and f in copy:
             f2 = copy[f]
             actions['m'].append((f, (f, f2, f2, False, pa.node()),
                             "local copied/moved from " + f2))
         elif n1 and f in ma: # clean, a different, no remote
             if n1 != ma[f]:
                 if acceptremote:
                     actions['r'].append((f, None, "remote delete"))
                 else:
                     actions['cd'].append((f, None, "prompt changed/deleted"))
             elif n1[20:] == "a": # added, no remote
                 actions['f'].append((f, None, "remote deleted"))
             else:
                 actions['r'].append((f, None, "other deleted"))
         elif n2 and f in movewithdir:
             f2 = movewithdir[f]
             actions['dg'].append((f2, (f, fl2),
                             "local directory rename - get from " + f))
         elif n2 and f in copy:
             f2 = copy[f]
             if f2 in m2:
                 actions['m'].append((f, (f2, f, f2, False, pa.node()),
                                 "remote copied from " + f2))
             else:
                 actions['m'].append((f, (f2, f, f2, True, pa.node()),
                                 "remote moved from " + f2))
         elif n2 and f not in ma:
             # local unknown, remote created: the logic is described by the
             # following table:
             #
             # force  branchmerge  different  |  action
             #   n         *           n      |    get
             #   n         *           y      |   abort
             #   y         n           *      |    get
             #   y         y           n      |    get
             #   y         y           y      |   merge
             #
             # Checking whether the files are different is expensive, so we
             # don't do that when we can avoid it.
             if force and not branchmerge:
                 actions['g'].append((f, (fl2,), "remote created"))
             else:
                 different = _checkunknownfile(repo, wctx, p2, f)
                 if force and branchmerge and different:
                     # FIXME: This is wrong - f is not in ma ...
                     actions['m'].append((f, (f, f, f, False, pa.node()),
                                     "remote differs from untracked local"))
                 elif not force and different:
                     aborts.append((f, "ud"))
                 else:
                     actions['g'].append((f, (fl2,), "remote created"))
         elif n2 and n2 != ma[f]:
             different = _checkunknownfile(repo, wctx, p2, f)
             if not force and different:
                 aborts.append((f, "ud"))
             else:
                 # if different: old untracked f may be overwritten and lost
                 if acceptremote:
                     actions['g'].append((f, (m2.flags(f),),
                                    "remote recreating"))
                 else:
                     actions['dc'].append((f, (m2.flags(f),),
                                    "prompt deleted/changed"))
 
     for f, m in sorted(aborts):
         if m == "ud":
             repo.ui.warn(_("%s: untracked file differs\n") % f)
         else: assert False, m
     if aborts:
         raise util.Abort(_("untracked files in working directory differ "
                            "from files in requested revision"))
 
     if not util.checkcase(repo.path):
         # check collision between files only in p2 for clean update
         if (not branchmerge and
             (force or not wctx.dirty(missing=True, branch=False))):
             _checkcollision(repo, m2, None)
         else:
             _checkcollision(repo, m1, actions)
 
     return actions
 
 def batchremove(repo, actions):
     """apply removes to the working directory
 
     yields tuples for progress updates
     """
     verbose = repo.ui.verbose
     unlink = util.unlinkpath
     wjoin = repo.wjoin
     audit = repo.wopener.audit
     i = 0
     for f, args, msg in actions:
         repo.ui.debug(" %s: %s -> r\n" % (f, msg))
         if verbose:
             repo.ui.note(_("removing %s\n") % f)
         audit(f)
         try:
             unlink(wjoin(f), ignoremissing=True)
         except OSError, inst:
             repo.ui.warn(_("update failed to remove %s: %s!\n") %
                          (f, inst.strerror))
         if i == 100:
             yield i, f
             i = 0
         i += 1
     if i > 0:
         yield i, f
 
 def batchget(repo, mctx, actions):
     """apply gets to the working directory
 
     mctx is the context to get from
 
     yields tuples for progress updates
     """
     verbose = repo.ui.verbose
     fctx = mctx.filectx
     wwrite = repo.wwrite
     i = 0
     for f, args, msg in actions:
         repo.ui.debug(" %s: %s -> g\n" % (f, msg))
         if verbose:
             repo.ui.note(_("getting %s\n") % f)
         wwrite(f, fctx(f).data(), args[0])
         if i == 100:
             yield i, f
             i = 0
         i += 1
     if i > 0:
         yield i, f
 
 def applyupdates(repo, actions, wctx, mctx, overwrite, labels=None):
     """apply the merge action list to the working directory
 
     wctx is the working copy context
     mctx is the context to be merged into the working copy
 
     Return a tuple of counts (updated, merged, removed, unresolved) that
     describes how many files were affected by the update.
     """
 
     updated, merged, removed, unresolved = 0, 0, 0, 0
     ms = mergestate(repo)
     ms.reset(wctx.p1().node(), mctx.node())
     moves = []
     for m, l in actions.items():
         l.sort()
 
     # prescan for merges
     for f, args, msg in actions['m']:
         f1, f2, fa, move, anc = args
         if f == '.hgsubstate': # merged internally
             continue
         repo.ui.debug(" preserving %s for resolve of %s\n" % (f1, f))
         fcl = wctx[f1]
         fco = mctx[f2]
         actx = repo[anc]
         if fa in actx:
             fca = actx[fa]
         else:
             fca = repo.filectx(f1, fileid=nullrev)
         ms.add(fcl, fco, fca, f)
         if f1 != f and move:
             moves.append(f1)
 
     audit = repo.wopener.audit
     _updating = _('updating')
     _files = _('files')
     progress = repo.ui.progress
 
     # remove renamed files after safely stored
     for f in moves:
         if os.path.lexists(repo.wjoin(f)):
             repo.ui.debug("removing %s\n" % f)
             audit(f)
             util.unlinkpath(repo.wjoin(f))
 
     numupdates = sum(len(l) for m, l in actions.items() if m != 'k')
 
     if [a for a in actions['r'] if a[0] == '.hgsubstate']:
         subrepo.submerge(repo, wctx, mctx, wctx, overwrite)
 
     # remove in parallel (must come first)
     z = 0
     prog = worker.worker(repo.ui, 0.001, batchremove, (repo,), actions['r'])
     for i, item in prog:
         z += i
         progress(_updating, z, item=item, total=numupdates, unit=_files)
     removed = len(actions['r'])
 
     # get in parallel
     prog = worker.worker(repo.ui, 0.001, batchget, (repo, mctx), actions['g'])
     for i, item in prog:
         z += i
         progress(_updating, z, item=item, total=numupdates, unit=_files)
     updated = len(actions['g'])
 
     if [a for a in actions['g'] if a[0] == '.hgsubstate']:
         subrepo.submerge(repo, wctx, mctx, wctx, overwrite)
 
     # forget (manifest only, just log it) (must come first)
     for f, args, msg in actions['f']:
         repo.ui.debug(" %s: %s -> f\n" % (f, msg))
         z += 1
         progress(_updating, z, item=f, total=numupdates, unit=_files)
 
     # re-add (manifest only, just log it)
     for f, args, msg in actions['a']:
         repo.ui.debug(" %s: %s -> a\n" % (f, msg))
         z += 1
         progress(_updating, z, item=f, total=numupdates, unit=_files)
 
     # keep (noop, just log it)
     for f, args, msg in actions['k']:
         repo.ui.debug(" %s: %s -> k\n" % (f, msg))
         # no progress
 
     # merge
     for f, args, msg in actions['m']:
         repo.ui.debug(" %s: %s -> m\n" % (f, msg))
         z += 1
         progress(_updating, z, item=f, total=numupdates, unit=_files)
         f1, f2, fa, move, anc = args
         if f == '.hgsubstate': # subrepo states need updating
             subrepo.submerge(repo, wctx, mctx, wctx.ancestor(mctx),
                              overwrite)
             continue
         audit(f)
         r = ms.resolve(f, wctx, labels=labels)
         if r is not None and r > 0:
             unresolved += 1
         else:
             if r is None:
                 updated += 1
             else:
                 merged += 1
 
     # directory rename, move local
     for f, args, msg in actions['dm']:
         repo.ui.debug(" %s: %s -> dm\n" % (f, msg))
         z += 1
         progress(_updating, z, item=f, total=numupdates, unit=_files)
         f0, flags = args
         repo.ui.note(_("moving %s to %s\n") % (f0, f))
         audit(f)
         repo.wwrite(f, wctx.filectx(f0).data(), flags)
         util.unlinkpath(repo.wjoin(f0))
         updated += 1
 
     # local directory rename, get
     for f, args, msg in actions['dg']:
         repo.ui.debug(" %s: %s -> dg\n" % (f, msg))
         z += 1
         progress(_updating, z, item=f, total=numupdates, unit=_files)
         f0, flags = args
         repo.ui.note(_("getting %s to %s\n") % (f0, f))
         repo.wwrite(f, mctx.filectx(f0).data(), flags)
         updated += 1
 
     # divergent renames
     for f, args, msg in actions['dr']:
         repo.ui.debug(" %s: %s -> dr\n" % (f, msg))
         z += 1
         progress(_updating, z, item=f, total=numupdates, unit=_files)
         fl, = args
         repo.ui.warn(_("note: possible conflict - %s was renamed "
                        "multiple times to:\n") % f)
         for nf in fl:
             repo.ui.warn(" %s\n" % nf)
 
     # rename and delete
     for f, args, msg in actions['rd']:
         repo.ui.debug(" %s: %s -> rd\n" % (f, msg))
         z += 1
         progress(_updating, z, item=f, total=numupdates, unit=_files)
         fl, = args
         repo.ui.warn(_("note: possible conflict - %s was deleted "
                        "and renamed to:\n") % f)
         for nf in fl:
             repo.ui.warn(" %s\n" % nf)
 
     # exec
     for f, args, msg in actions['e']:
         repo.ui.debug(" %s: %s -> e\n" % (f, msg))
         z += 1
         progress(_updating, z, item=f, total=numupdates, unit=_files)
         flags, = args
         audit(f)
         util.setflags(repo.wjoin(f), 'l' in flags, 'x' in flags)
         updated += 1
 
     ms.commit()
     progress(_updating, None, total=numupdates, unit=_files)
 
     return updated, merged, removed, unresolved
 
 def calculateupdates(repo, wctx, mctx, ancestors, branchmerge, force, partial,
                      acceptremote, followcopies):
     "Calculate the actions needed to merge mctx into wctx using ancestors"
 
     if len(ancestors) == 1: # default
         actions = manifestmerge(repo, wctx, mctx, ancestors[0],
                                 branchmerge, force,
                                 partial, acceptremote, followcopies)
 
     else: # only when merge.preferancestor=* - the default
         repo.ui.note(
             _("note: merging %s and %s using bids from ancestors %s\n") %
             (wctx, mctx, _(' and ').join(str(anc) for anc in ancestors)))
 
         # Call for bids
         fbids = {} # mapping filename to bids (action method to list af actions)
         for ancestor in ancestors:
             repo.ui.note(_('\ncalculating bids for ancestor %s\n') % ancestor)
             actions = manifestmerge(repo, wctx, mctx, ancestor,
                                     branchmerge, force,
                                     partial, acceptremote, followcopies)
             for m, l in sorted(actions.items()):
                 for a in l:
                     f, args, msg = a
                     repo.ui.debug(' %s: %s -> %s\n' % (f, msg, m))
                     if f in fbids:
                         d = fbids[f]
                         if m in d:
                             d[m].append(a)
                         else:
                             d[m] = [a]
                     else:
                         fbids[f] = {m: [a]}
 
         # Pick the best bid for each file
         repo.ui.note(_('\nauction for merging merge bids\n'))
         actions = dict((m, []) for m in actions.keys())
         for f, bids in sorted(fbids.items()):
             # bids is a mapping from action method to list af actions
             # Consensus?
             if len(bids) == 1: # all bids are the same kind of method
                 m, l = bids.items()[0]
                 if util.all(a == l[0] for a in l[1:]): # len(bids) is > 1
                     repo.ui.note(" %s: consensus for %s\n" % (f, m))
                     actions[m].append(l[0])
                     continue
             # If keep is an option, just do it.
             if "k" in bids:
                 repo.ui.note(" %s: picking 'keep' action\n" % f)
                 actions['k'].append(bids["k"][0])
                 continue
             # If there are gets and they all agree [how could they not?], do it.
             if "g" in bids:
                 ga0 = bids["g"][0]
                 if util.all(a == ga0 for a in bids["g"][1:]):
                     repo.ui.note(" %s: picking 'get' action\n" % f)
                     actions['g'].append(ga0)
                     continue
             # TODO: Consider other simple actions such as mode changes
             # Handle inefficient democrazy.
             repo.ui.note(_(' %s: multiple bids for merge action:\n') % f)
             for m, l in sorted(bids.items()):
                 for _f, args, msg in l:
                     repo.ui.note('  %s -> %s\n' % (msg, m))
             # Pick random action. TODO: Instead, prompt user when resolving
             m, l = bids.items()[0]
             repo.ui.warn(_(' %s: ambiguous merge - picked %s action\n') %
                          (f, m))
             actions[m].append(l[0])
             continue
         repo.ui.note(_('end of auction\n\n'))
 
     # Prompt and create actions. TODO: Move this towards resolve phase.
     for f, args, msg in actions['cd']:
         if repo.ui.promptchoice(
             _("local changed %s which remote deleted\n"
               "use (c)hanged version or (d)elete?"
               "$$ &Changed $$ &Delete") % f, 0):
             actions['r'].append((f, None, "prompt delete"))
         else:
             actions['a'].append((f, None, "prompt keep"))
     del actions['cd'][:]
 
     for f, args, msg in actions['dc']:
         flags, = args
         if repo.ui.promptchoice(
             _("remote changed %s which local deleted\n"
               "use (c)hanged version or leave (d)eleted?"
               "$$ &Changed $$ &Deleted") % f, 0) == 0:
             actions['g'].append((f, (flags,), "prompt recreating"))
     del actions['dc'][:]
 
     if wctx.rev() is None:
         ractions, factions = _forgetremoved(wctx, mctx, branchmerge)
         actions['r'].extend(ractions)
         actions['f'].extend(factions)
 
     return actions
 
 def recordupdates(repo, actions, branchmerge):
     "record merge actions to the dirstate"
     # remove (must come first)
     for f, args, msg in actions['r']:
         if branchmerge:
             repo.dirstate.remove(f)
         else:
             repo.dirstate.drop(f)
 
     # forget (must come first)
     for f, args, msg in actions['f']:
         repo.dirstate.drop(f)
 
     # re-add
     for f, args, msg in actions['a']:
         if not branchmerge:
             repo.dirstate.add(f)
 
     # exec change
     for f, args, msg in actions['e']:
         repo.dirstate.normallookup(f)
 
     # keep
     for f, args, msg in actions['k']:
         pass
 
     # get
     for f, args, msg in actions['g']:
         if branchmerge:
             repo.dirstate.otherparent(f)
         else:
             repo.dirstate.normal(f)
 
     # merge
     for f, args, msg in actions['m']:
         f1, f2, fa, move, anc = args
         if branchmerge:
             # We've done a branch merge, mark this file as merged
             # so that we properly record the merger later
             repo.dirstate.merge(f)
             if f1 != f2: # copy/rename
                 if move:
                     repo.dirstate.remove(f1)
                 if f1 != f:
                     repo.dirstate.copy(f1, f)
                 else:
                     repo.dirstate.copy(f2, f)
         else:
             # We've update-merged a locally modified file, so
             # we set the dirstate to emulate a normal checkout
             # of that file some time in the past. Thus our
             # merge will appear as a normal local file
             # modification.
             if f2 == f: # file not locally copied/moved
                 repo.dirstate.normallookup(f)
             if move:
                 repo.dirstate.drop(f1)
 
     # directory rename, move local
     for f, args, msg in actions['dm']:
         f0, flag = args
         if f0 not in repo.dirstate:
             # untracked file moved
             continue
         if branchmerge:
             repo.dirstate.add(f)
             repo.dirstate.remove(f0)
             repo.dirstate.copy(f0, f)
         else:
             repo.dirstate.normal(f)
             repo.dirstate.drop(f0)
 
     # directory rename, get
     for f, args, msg in actions['dg']:
         f0, flag = args
         if branchmerge:
             repo.dirstate.add(f)
             repo.dirstate.copy(f0, f)
         else:
             repo.dirstate.normal(f)
 
 def update(repo, node, branchmerge, force, partial, ancestor=None,
            mergeancestor=False, labels=None):
     """
     Perform a merge between the working directory and the given node
 
     node = the node to update to, or None if unspecified
     branchmerge = whether to merge between branches
     force = whether to force branch merging or file overwriting
     partial = a function to filter file lists (dirstate not updated)
     mergeancestor = whether it is merging with an ancestor. If true,
       we should accept the incoming changes for any prompts that occur.
       If false, merging with an ancestor (fast-forward) is only allowed
       between different named branches. This flag is used by rebase extension
       as a temporary fix and should be avoided in general.
 
     The table below shows all the behaviors of the update command
     given the -c and -C or no options, whether the working directory
     is dirty, whether a revision is specified, and the relationship of
     the parent rev to the target rev (linear, on the same named
     branch, or on another named branch).
 
     This logic is tested by test-update-branches.t.
 
     -c  -C  dirty  rev  |  linear   same  cross
      n   n    n     n   |    ok     (1)     x
      n   n    n     y   |    ok     ok     ok
      n   n    y     n   |   merge   (2)    (2)
      n   n    y     y   |   merge   (3)    (3)
      n   y    *     *   |    ---  discard  ---
      y   n    y     *   |    ---    (4)    ---
      y   n    n     *   |    ---    ok     ---
      y   y    *     *   |    ---    (5)    ---
 
     x = can't happen
     * = don't-care
     1 = abort: not a linear update (merge or update --check to force update)
     2 = abort: uncommitted changes (commit and merge, or update --clean to
                  discard changes)
     3 = abort: uncommitted changes (commit or update --clean to discard changes)
     4 = abort: uncommitted changes (checked in commands.py)
     5 = incompatible options (checked in commands.py)
 
     Return the same tuple as applyupdates().
     """
 
     onode = node
     wlock = repo.wlock()
     try:
         wc = repo[None]
         pl = wc.parents()
         p1 = pl[0]
         pas = [None]
         if ancestor:
             pas = [repo[ancestor]]
 
         if node is None:
             # Here is where we should consider bookmarks, divergent bookmarks,
             # foreground changesets (successors), and tip of current branch;
             # but currently we are only checking the branch tips.
             try:
                 node = repo.branchtip(wc.branch())
             except errormod.RepoLookupError:
                 if wc.branch() == "default": # no default branch!
                     node = repo.lookup("tip") # update to tip
                 else:
                     raise util.Abort(_("branch %s not found") % wc.branch())
 
             if p1.obsolete() and not p1.children():
                 # allow updating to successors
                 successors = obsolete.successorssets(repo, p1.node())
 
                 # behavior of certain cases is as follows,
                 #
                 # divergent changesets: update to highest rev, similar to what
                 #     is currently done when there are more than one head
                 #     (i.e. 'tip')
                 #
                 # replaced changesets: same as divergent except we know there
                 # is no conflict
                 #
                 # pruned changeset: no update is done; though, we could
                 #     consider updating to the first non-obsolete parent,
                 #     similar to what is current done for 'hg prune'
 
                 if successors:
                     # flatten the list here handles both divergent (len > 1)
                     # and the usual case (len = 1)
                     successors = [n for sub in successors for n in sub]
 
                     # get the max revision for the given successors set,
                     # i.e. the 'tip' of a set
                     node = repo.revs("max(%ln)", successors).first()
                     pas = [p1]
 
         overwrite = force and not branchmerge
 
         p2 = repo[node]
         if pas[0] is None:
             if repo.ui.config("merge", "preferancestor", '*') == '*':
                 cahs = repo.changelog.commonancestorsheads(p1.node(), p2.node())
                 pas = [repo[anc] for anc in (sorted(cahs) or [nullid])]
             else:
                 pas = [p1.ancestor(p2, warn=branchmerge)]
 
         fp1, fp2, xp1, xp2 = p1.node(), p2.node(), str(p1), str(p2)
 
         ### check phase
         if not overwrite and len(pl) > 1:
             raise util.Abort(_("outstanding uncommitted merge"))
         if branchmerge:
             if pas == [p2]:
                 raise util.Abort(_("merging with a working directory ancestor"
                                    " has no effect"))
             elif pas == [p1]:
                 if not mergeancestor and p1.branch() == p2.branch():
                     raise util.Abort(_("nothing to merge"),
                                      hint=_("use 'hg update' "
                                             "or check 'hg heads'"))
             if not force and (wc.files() or wc.deleted()):
                 raise util.Abort(_("uncommitted changes"),
                                  hint=_("use 'hg status' to list changes"))
             for s in sorted(wc.substate):
                 if wc.sub(s).dirty():
                     raise util.Abort(_("uncommitted changes in "
                                        "subrepository '%s'") % s)
 
         elif not overwrite:
             if p1 == p2: # no-op update
                 # call the hooks and exit early
                 repo.hook('preupdate', throw=True, parent1=xp2, parent2='')
                 repo.hook('update', parent1=xp2, parent2='', error=0)
                 return 0, 0, 0, 0
 
             if pas not in ([p1], [p2]):  # nonlinear
                 dirty = wc.dirty(missing=True)
                 if dirty or onode is None:
                     # Branching is a bit strange to ensure we do the minimal
                     # amount of call to obsolete.background.
                     foreground = obsolete.foreground(repo, [p1.node()])
                     # note: the <node> variable contains a random identifier
                     if repo[node].node() in foreground:
                         pas = [p1]  # allow updating to successors
                     elif dirty:
                         msg = _("uncommitted changes")
                         if onode is None:
                             hint = _("commit and merge, or update --clean to"
                                      " discard changes")
                         else:
                             hint = _("commit or update --clean to discard"
                                      " changes")
                         raise util.Abort(msg, hint=hint)
                     else:  # node is none
                         msg = _("not a linear update")
                         hint = _("merge or update --check to force update")
                         raise util.Abort(msg, hint=hint)
                 else:
                     # Allow jumping branches if clean and specific rev given
                     pas = [p1]
 
         followcopies = False
         if overwrite:
             pas = [wc]
         elif pas == [p2]: # backwards
             pas = [wc.p1()]
         elif not branchmerge and not wc.dirty(missing=True):
             pass
         elif pas[0] and repo.ui.configbool("merge", "followcopies", True):
             followcopies = True
 
         ### calculate phase
         actions = calculateupdates(repo, wc, p2, pas, branchmerge, force,
                                    partial, mergeancestor, followcopies)
 
         ### apply phase
         if not branchmerge: # just jump to the new rev
             fp1, fp2, xp1, xp2 = fp2, nullid, xp2, ''
         if not partial:
             repo.hook('preupdate', throw=True, parent1=xp1, parent2=xp2)
             # note that we're in the middle of an update
             repo.vfs.write('updatestate', p2.hex())
 
         stats = applyupdates(repo, actions, wc, p2, overwrite, labels=labels)
 
         if not partial:
             repo.dirstate.beginparentchange()
             repo.setparents(fp1, fp2)
             recordupdates(repo, actions, branchmerge)
             # update completed, clear state
             util.unlink(repo.join('updatestate'))
 
             if not branchmerge:
                 repo.dirstate.setbranch(p2.branch())
             repo.dirstate.endparentchange()
     finally:
         wlock.release()
 
     if not partial:
         repo.hook('update', parent1=xp1, parent2=xp2, error=stats[3])
     return stats
 
 def graft(repo, ctx, pctx, labels):
     """Do a graft-like merge.
 
     This is a merge where the merge ancestor is chosen such that one
     or more changesets are grafted onto the current changeset. In
     addition to the merge, this fixes up the dirstate to include only
     a single parent and tries to duplicate any renames/copies
     appropriately.
 
     ctx - changeset to rebase
     pctx - merge base, usually ctx.p1()
     labels - merge labels eg ['local', 'graft']
 
     """
 
     stats = update(repo, ctx.node(), True, True, False, pctx.node(),
                    labels=labels)
     # drop the second merge parent
     repo.dirstate.beginparentchange()
     repo.setparents(repo['.'].node(), nullid)
     repo.dirstate.write()
     # fix up dirstate for copies and renames
     copies.duplicatecopies(repo, ctx.rev(), pctx.rev())
     repo.dirstate.endparentchange()
     return stats