diff: don't crash when merged-in addition was removed (issue4786)

During a merge, if the user removes a file that came from parent 2 and
did not exist in parent 1, the file's status will be "removed". This
surprises the diff code, which crashes because it expects removed
files exist in parent 1. This has been broken since 377124ba6b10
(trydiff: use 'not in addedset' for symmetry with 'not in removedset',
2014-12-23).

Fix by fixing up the list of removed file, similar to how we currently
fix up the list of modified and added files during a merge.

mercurial/patch.py

@@ -1,2580 +1,2587 b''
 # patch.py - patch file parsing routines
 #
 # Copyright 2006 Brendan Cully <[e-mail]>
 # Copyright 2007 Chris Mason <[e-mail]>
 #
 # This software may be used and distributed according to the terms of the
 # GNU General Public License version 2 or any later version.
 
 from __future__ import absolute_import
 
 import cStringIO
 import collections
 import copy
 import email
 import errno
 import os
 import posixpath
 import re
 import shutil
 import tempfile
 import zlib
 
 from .i18n import _
 from .node import (
     hex,
     short,
 )
 from . import (
     base85,
     copies,
     diffhelpers,
     encoding,
     error,
     mdiff,
     pathutil,
     scmutil,
     util,
 )
 
 gitre = re.compile('diff --git a/(.*) b/(.*)')
 tabsplitter = re.compile(r'(\t+|[^\t]+)')
 
 class PatchError(Exception):
     pass
 
 
 # public functions
 
 def split(stream):
     '''return an iterator of individual patches from a stream'''
     def isheader(line, inheader):
         if inheader and line[0] in (' ', '\t'):
             # continuation
             return True
         if line[0] in (' ', '-', '+'):
             # diff line - don't check for header pattern in there
             return False
         l = line.split(': ', 1)
         return len(l) == 2 and ' ' not in l[0]
 
     def chunk(lines):
         return cStringIO.StringIO(''.join(lines))
 
     def hgsplit(stream, cur):
         inheader = True
 
         for line in stream:
             if not line.strip():
                 inheader = False
             if not inheader and line.startswith('# HG changeset patch'):
                 yield chunk(cur)
                 cur = []
                 inheader = True
 
             cur.append(line)
 
         if cur:
             yield chunk(cur)
 
     def mboxsplit(stream, cur):
         for line in stream:
             if line.startswith('From '):
                 for c in split(chunk(cur[1:])):
                     yield c
                 cur = []
 
             cur.append(line)
 
         if cur:
             for c in split(chunk(cur[1:])):
                 yield c
 
     def mimesplit(stream, cur):
         def msgfp(m):
             fp = cStringIO.StringIO()
             g = email.Generator.Generator(fp, mangle_from_=False)
             g.flatten(m)
             fp.seek(0)
             return fp
 
         for line in stream:
             cur.append(line)
         c = chunk(cur)
 
         m = email.Parser.Parser().parse(c)
         if not m.is_multipart():
             yield msgfp(m)
         else:
             ok_types = ('text/plain', 'text/x-diff', 'text/x-patch')
             for part in m.walk():
                 ct = part.get_content_type()
                 if ct not in ok_types:
                     continue
                 yield msgfp(part)
 
     def headersplit(stream, cur):
         inheader = False
 
         for line in stream:
             if not inheader and isheader(line, inheader):
                 yield chunk(cur)
                 cur = []
                 inheader = True
             if inheader and not isheader(line, inheader):
                 inheader = False
 
             cur.append(line)
 
         if cur:
             yield chunk(cur)
 
     def remainder(cur):
         yield chunk(cur)
 
     class fiter(object):
         def __init__(self, fp):
             self.fp = fp
 
         def __iter__(self):
             return self
 
         def next(self):
             l = self.fp.readline()
             if not l:
                 raise StopIteration
             return l
 
     inheader = False
     cur = []
 
     mimeheaders = ['content-type']
 
     if not util.safehasattr(stream, 'next'):
         # http responses, for example, have readline but not next
         stream = fiter(stream)
 
     for line in stream:
         cur.append(line)
         if line.startswith('# HG changeset patch'):
             return hgsplit(stream, cur)
         elif line.startswith('From '):
             return mboxsplit(stream, cur)
         elif isheader(line, inheader):
             inheader = True
             if line.split(':', 1)[0].lower() in mimeheaders:
                 # let email parser handle this
                 return mimesplit(stream, cur)
         elif line.startswith('--- ') and inheader:
             # No evil headers seen by diff start, split by hand
             return headersplit(stream, cur)
         # Not enough info, keep reading
 
     # if we are here, we have a very plain patch
     return remainder(cur)
 
 ## Some facility for extensible patch parsing:
 # list of pairs ("header to match", "data key")
 patchheadermap = [('Date', 'date'),
                   ('Branch', 'branch'),
                   ('Node ID', 'nodeid'),
                  ]
 
 def extract(ui, fileobj):
     '''extract patch from data read from fileobj.
 
     patch can be a normal patch or contained in an email message.
 
     return a dictionary. Standard keys are:
       - filename,
       - message,
       - user,
       - date,
       - branch,
       - node,
       - p1,
       - p2.
     Any item can be missing from the dictionary. If filename is missing,
     fileobj did not contain a patch. Caller must unlink filename when done.'''
 
     # attempt to detect the start of a patch
     # (this heuristic is borrowed from quilt)
     diffre = re.compile(r'^(?:Index:[ \t]|diff[ \t]|RCS file: |'
                         r'retrieving revision [0-9]+(\.[0-9]+)*$|'
                         r'---[ \t].*?^\+\+\+[ \t]|'
                         r'\*\*\*[ \t].*?^---[ \t])', re.MULTILINE|re.DOTALL)
 
     data = {}
     fd, tmpname = tempfile.mkstemp(prefix='hg-patch-')
     tmpfp = os.fdopen(fd, 'w')
     try:
         msg = email.Parser.Parser().parse(fileobj)
 
         subject = msg['Subject']
         data['user'] = msg['From']
         if not subject and not data['user']:
             # Not an email, restore parsed headers if any
             subject = '\n'.join(': '.join(h) for h in msg.items()) + '\n'
 
         # should try to parse msg['Date']
         parents = []
 
         if subject:
             if subject.startswith('[PATCH'):
                 pend = subject.find(']')
                 if pend >= 0:
                     subject = subject[pend + 1:].lstrip()
             subject = re.sub(r'\n[ \t]+', ' ', subject)
             ui.debug('Subject: %s\n' % subject)
         if data['user']:
             ui.debug('From: %s\n' % data['user'])
         diffs_seen = 0
         ok_types = ('text/plain', 'text/x-diff', 'text/x-patch')
         message = ''
         for part in msg.walk():
             content_type = part.get_content_type()
             ui.debug('Content-Type: %s\n' % content_type)
             if content_type not in ok_types:
                 continue
             payload = part.get_payload(decode=True)
             m = diffre.search(payload)
             if m:
                 hgpatch = False
                 hgpatchheader = False
                 ignoretext = False
 
                 ui.debug('found patch at byte %d\n' % m.start(0))
                 diffs_seen += 1
                 cfp = cStringIO.StringIO()
                 for line in payload[:m.start(0)].splitlines():
                     if line.startswith('# HG changeset patch') and not hgpatch:
                         ui.debug('patch generated by hg export\n')
                         hgpatch = True
                         hgpatchheader = True
                         # drop earlier commit message content
                         cfp.seek(0)
                         cfp.truncate()
                         subject = None
                     elif hgpatchheader:
                         if line.startswith('# User '):
                             data['user'] = line[7:]
                             ui.debug('From: %s\n' % data['user'])
                         elif line.startswith("# Parent "):
                             parents.append(line[9:].lstrip())
                         elif line.startswith("# "):
                             for header, key in patchheadermap:
                                 prefix = '# %s ' % header
                                 if line.startswith(prefix):
                                     data[key] = line[len(prefix):]
                         else:
                             hgpatchheader = False
                     elif line == '---':
                         ignoretext = True
                     if not hgpatchheader and not ignoretext:
                         cfp.write(line)
                         cfp.write('\n')
                 message = cfp.getvalue()
                 if tmpfp:
                     tmpfp.write(payload)
                     if not payload.endswith('\n'):
                         tmpfp.write('\n')
             elif not diffs_seen and message and content_type == 'text/plain':
                 message += '\n' + payload
     except: # re-raises
         tmpfp.close()
         os.unlink(tmpname)
         raise
 
     if subject and not message.startswith(subject):
         message = '%s\n%s' % (subject, message)
     data['message'] = message
     tmpfp.close()
     if parents:
         data['p1'] = parents.pop(0)
         if parents:
             data['p2'] = parents.pop(0)
 
     if diffs_seen:
         data['filename'] = tmpname
     else:
         os.unlink(tmpname)
     return data
 
 class patchmeta(object):
     """Patched file metadata
 
     'op' is the performed operation within ADD, DELETE, RENAME, MODIFY
     or COPY.  'path' is patched file path. 'oldpath' is set to the
     origin file when 'op' is either COPY or RENAME, None otherwise. If
     file mode is changed, 'mode' is a tuple (islink, isexec) where
     'islink' is True if the file is a symlink and 'isexec' is True if
     the file is executable. Otherwise, 'mode' is None.
     """
     def __init__(self, path):
         self.path = path
         self.oldpath = None
         self.mode = None
         self.op = 'MODIFY'
         self.binary = False
 
     def setmode(self, mode):
         islink = mode & 0o20000
         isexec = mode & 0o100
         self.mode = (islink, isexec)
 
     def copy(self):
         other = patchmeta(self.path)
         other.oldpath = self.oldpath
         other.mode = self.mode
         other.op = self.op
         other.binary = self.binary
         return other
 
     def _ispatchinga(self, afile):
         if afile == '/dev/null':
             return self.op == 'ADD'
         return afile == 'a/' + (self.oldpath or self.path)
 
     def _ispatchingb(self, bfile):
         if bfile == '/dev/null':
             return self.op == 'DELETE'
         return bfile == 'b/' + self.path
 
     def ispatching(self, afile, bfile):
         return self._ispatchinga(afile) and self._ispatchingb(bfile)
 
     def __repr__(self):
         return "<patchmeta %s %r>" % (self.op, self.path)
 
 def readgitpatch(lr):
     """extract git-style metadata about patches from <patchname>"""
 
     # Filter patch for git information
     gp = None
     gitpatches = []
     for line in lr:
         line = line.rstrip(' \r\n')
         if line.startswith('diff --git a/'):
             m = gitre.match(line)
             if m:
                 if gp:
                     gitpatches.append(gp)
                 dst = m.group(2)
                 gp = patchmeta(dst)
         elif gp:
             if line.startswith('--- '):
                 gitpatches.append(gp)
                 gp = None
                 continue
             if line.startswith('rename from '):
                 gp.op = 'RENAME'
                 gp.oldpath = line[12:]
             elif line.startswith('rename to '):
                 gp.path = line[10:]
             elif line.startswith('copy from '):
                 gp.op = 'COPY'
                 gp.oldpath = line[10:]
             elif line.startswith('copy to '):
                 gp.path = line[8:]
             elif line.startswith('deleted file'):
                 gp.op = 'DELETE'
             elif line.startswith('new file mode '):
                 gp.op = 'ADD'
                 gp.setmode(int(line[-6:], 8))
             elif line.startswith('new mode '):
                 gp.setmode(int(line[-6:], 8))
             elif line.startswith('GIT binary patch'):
                 gp.binary = True
     if gp:
         gitpatches.append(gp)
 
     return gitpatches
 
 class linereader(object):
     # simple class to allow pushing lines back into the input stream
     def __init__(self, fp):
         self.fp = fp
         self.buf = []
 
     def push(self, line):
         if line is not None:
             self.buf.append(line)
 
     def readline(self):
         if self.buf:
             l = self.buf[0]
             del self.buf[0]
             return l
         return self.fp.readline()
 
     def __iter__(self):
         while True:
             l = self.readline()
             if not l:
                 break
             yield l
 
 class abstractbackend(object):
     def __init__(self, ui):
         self.ui = ui
 
     def getfile(self, fname):
         """Return target file data and flags as a (data, (islink,
         isexec)) tuple. Data is None if file is missing/deleted.
         """
         raise NotImplementedError
 
     def setfile(self, fname, data, mode, copysource):
         """Write data to target file fname and set its mode. mode is a
         (islink, isexec) tuple. If data is None, the file content should
         be left unchanged. If the file is modified after being copied,
         copysource is set to the original file name.
         """
         raise NotImplementedError
 
     def unlink(self, fname):
         """Unlink target file."""
         raise NotImplementedError
 
     def writerej(self, fname, failed, total, lines):
         """Write rejected lines for fname. total is the number of hunks
         which failed to apply and total the total number of hunks for this
         files.
         """
         pass
 
     def exists(self, fname):
         raise NotImplementedError
 
 class fsbackend(abstractbackend):
     def __init__(self, ui, basedir):
         super(fsbackend, self).__init__(ui)
         self.opener = scmutil.opener(basedir)
 
     def _join(self, f):
         return os.path.join(self.opener.base, f)
 
     def getfile(self, fname):
         if self.opener.islink(fname):
             return (self.opener.readlink(fname), (True, False))
 
         isexec = False
         try:
             isexec = self.opener.lstat(fname).st_mode & 0o100 != 0
         except OSError as e:
             if e.errno != errno.ENOENT:
                 raise
         try:
             return (self.opener.read(fname), (False, isexec))
         except IOError as e:
             if e.errno != errno.ENOENT:
                 raise
             return None, None
 
     def setfile(self, fname, data, mode, copysource):
         islink, isexec = mode
         if data is None:
             self.opener.setflags(fname, islink, isexec)
             return
         if islink:
             self.opener.symlink(data, fname)
         else:
             self.opener.write(fname, data)
             if isexec:
                 self.opener.setflags(fname, False, True)
 
     def unlink(self, fname):
         self.opener.unlinkpath(fname, ignoremissing=True)
 
     def writerej(self, fname, failed, total, lines):
         fname = fname + ".rej"
         self.ui.warn(
             _("%d out of %d hunks FAILED -- saving rejects to file %s\n") %
             (failed, total, fname))
         fp = self.opener(fname, 'w')
         fp.writelines(lines)
         fp.close()
 
     def exists(self, fname):
         return self.opener.lexists(fname)
 
 class workingbackend(fsbackend):
     def __init__(self, ui, repo, similarity):
         super(workingbackend, self).__init__(ui, repo.root)
         self.repo = repo
         self.similarity = similarity
         self.removed = set()
         self.changed = set()
         self.copied = []
 
     def _checkknown(self, fname):
         if self.repo.dirstate[fname] == '?' and self.exists(fname):
             raise PatchError(_('cannot patch %s: file is not tracked') % fname)
 
     def setfile(self, fname, data, mode, copysource):
         self._checkknown(fname)
         super(workingbackend, self).setfile(fname, data, mode, copysource)
         if copysource is not None:
             self.copied.append((copysource, fname))
         self.changed.add(fname)
 
     def unlink(self, fname):
         self._checkknown(fname)
         super(workingbackend, self).unlink(fname)
         self.removed.add(fname)
         self.changed.add(fname)
 
     def close(self):
         wctx = self.repo[None]
         changed = set(self.changed)
         for src, dst in self.copied:
             scmutil.dirstatecopy(self.ui, self.repo, wctx, src, dst)
         if self.removed:
             wctx.forget(sorted(self.removed))
             for f in self.removed:
                 if f not in self.repo.dirstate:
                     # File was deleted and no longer belongs to the
                     # dirstate, it was probably marked added then
                     # deleted, and should not be considered by
                     # marktouched().
                     changed.discard(f)
         if changed:
             scmutil.marktouched(self.repo, changed, self.similarity)
         return sorted(self.changed)
 
 class filestore(object):
     def __init__(self, maxsize=None):
         self.opener = None
         self.files = {}
         self.created = 0
         self.maxsize = maxsize
         if self.maxsize is None:
             self.maxsize = 4*(2**20)
         self.size = 0
         self.data = {}
 
     def setfile(self, fname, data, mode, copied=None):
         if self.maxsize < 0 or (len(data) + self.size) <= self.maxsize:
             self.data[fname] = (data, mode, copied)
             self.size += len(data)
         else:
             if self.opener is None:
                 root = tempfile.mkdtemp(prefix='hg-patch-')
                 self.opener = scmutil.opener(root)
             # Avoid filename issues with these simple names
             fn = str(self.created)
             self.opener.write(fn, data)
             self.created += 1
             self.files[fname] = (fn, mode, copied)
 
     def getfile(self, fname):
         if fname in self.data:
             return self.data[fname]
         if not self.opener or fname not in self.files:
             return None, None, None
         fn, mode, copied = self.files[fname]
         return self.opener.read(fn), mode, copied
 
     def close(self):
         if self.opener:
             shutil.rmtree(self.opener.base)
 
 class repobackend(abstractbackend):
     def __init__(self, ui, repo, ctx, store):
         super(repobackend, self).__init__(ui)
         self.repo = repo
         self.ctx = ctx
         self.store = store
         self.changed = set()
         self.removed = set()
         self.copied = {}
 
     def _checkknown(self, fname):
         if fname not in self.ctx:
             raise PatchError(_('cannot patch %s: file is not tracked') % fname)
 
     def getfile(self, fname):
         try:
             fctx = self.ctx[fname]
         except error.LookupError:
             return None, None
         flags = fctx.flags()
         return fctx.data(), ('l' in flags, 'x' in flags)
 
     def setfile(self, fname, data, mode, copysource):
         if copysource:
             self._checkknown(copysource)
         if data is None:
             data = self.ctx[fname].data()
         self.store.setfile(fname, data, mode, copysource)
         self.changed.add(fname)
         if copysource:
             self.copied[fname] = copysource
 
     def unlink(self, fname):
         self._checkknown(fname)
         self.removed.add(fname)
 
     def exists(self, fname):
         return fname in self.ctx
 
     def close(self):
         return self.changed | self.removed
 
 # @@ -start,len +start,len @@ or @@ -start +start @@ if len is 1
 unidesc = re.compile('@@ -(\d+)(?:,(\d+))? \+(\d+)(?:,(\d+))? @@')
 contextdesc = re.compile('(?:---|\*\*\*) (\d+)(?:,(\d+))? (?:---|\*\*\*)')
 eolmodes = ['strict', 'crlf', 'lf', 'auto']
 
 class patchfile(object):
     def __init__(self, ui, gp, backend, store, eolmode='strict'):
         self.fname = gp.path
         self.eolmode = eolmode
         self.eol = None
         self.backend = backend
         self.ui = ui
         self.lines = []
         self.exists = False
         self.missing = True
         self.mode = gp.mode
         self.copysource = gp.oldpath
         self.create = gp.op in ('ADD', 'COPY', 'RENAME')
         self.remove = gp.op == 'DELETE'
         if self.copysource is None:
             data, mode = backend.getfile(self.fname)
         else:
             data, mode = store.getfile(self.copysource)[:2]
         if data is not None:
             self.exists = self.copysource is None or backend.exists(self.fname)
             self.missing = False
             if data:
                 self.lines = mdiff.splitnewlines(data)
             if self.mode is None:
                 self.mode = mode
             if self.lines:
                 # Normalize line endings
                 if self.lines[0].endswith('\r\n'):
                     self.eol = '\r\n'
                 elif self.lines[0].endswith('\n'):
                     self.eol = '\n'
                 if eolmode != 'strict':
                     nlines = []
                     for l in self.lines:
                         if l.endswith('\r\n'):
                             l = l[:-2] + '\n'
                         nlines.append(l)
                     self.lines = nlines
         else:
             if self.create:
                 self.missing = False
             if self.mode is None:
                 self.mode = (False, False)
         if self.missing:
             self.ui.warn(_("unable to find '%s' for patching\n") % self.fname)
 
         self.hash = {}
         self.dirty = 0
         self.offset = 0
         self.skew = 0
         self.rej = []
         self.fileprinted = False
         self.printfile(False)
         self.hunks = 0
 
     def writelines(self, fname, lines, mode):
         if self.eolmode == 'auto':
             eol = self.eol
         elif self.eolmode == 'crlf':
             eol = '\r\n'
         else:
             eol = '\n'
 
         if self.eolmode != 'strict' and eol and eol != '\n':
             rawlines = []
             for l in lines:
                 if l and l[-1] == '\n':
                     l = l[:-1] + eol
                 rawlines.append(l)
             lines = rawlines
 
         self.backend.setfile(fname, ''.join(lines), mode, self.copysource)
 
     def printfile(self, warn):
         if self.fileprinted:
             return
         if warn or self.ui.verbose:
             self.fileprinted = True
         s = _("patching file %s\n") % self.fname
         if warn:
             self.ui.warn(s)
         else:
             self.ui.note(s)
 
 
     def findlines(self, l, linenum):
         # looks through the hash and finds candidate lines.  The
         # result is a list of line numbers sorted based on distance
         # from linenum
 
         cand = self.hash.get(l, [])
         if len(cand) > 1:
             # resort our list of potentials forward then back.
             cand.sort(key=lambda x: abs(x - linenum))
         return cand
 
     def write_rej(self):
         # our rejects are a little different from patch(1).  This always
         # creates rejects in the same form as the original patch.  A file
         # header is inserted so that you can run the reject through patch again
         # without having to type the filename.
         if not self.rej:
             return
         base = os.path.basename(self.fname)
         lines = ["--- %s\n+++ %s\n" % (base, base)]
         for x in self.rej:
             for l in x.hunk:
                 lines.append(l)
                 if l[-1] != '\n':
                     lines.append("\n\ No newline at end of file\n")
         self.backend.writerej(self.fname, len(self.rej), self.hunks, lines)
 
     def apply(self, h):
         if not h.complete():
             raise PatchError(_("bad hunk #%d %s (%d %d %d %d)") %
                             (h.number, h.desc, len(h.a), h.lena, len(h.b),
                             h.lenb))
 
         self.hunks += 1
 
         if self.missing:
             self.rej.append(h)
             return -1
 
         if self.exists and self.create:
             if self.copysource:
                 self.ui.warn(_("cannot create %s: destination already "
                                "exists\n") % self.fname)
             else:
                 self.ui.warn(_("file %s already exists\n") % self.fname)
             self.rej.append(h)
             return -1
 
         if isinstance(h, binhunk):
             if self.remove:
                 self.backend.unlink(self.fname)
             else:
                 l = h.new(self.lines)
                 self.lines[:] = l
                 self.offset += len(l)
                 self.dirty = True
             return 0
 
         horig = h
         if (self.eolmode in ('crlf', 'lf')
             or self.eolmode == 'auto' and self.eol):
             # If new eols are going to be normalized, then normalize
             # hunk data before patching. Otherwise, preserve input
             # line-endings.
             h = h.getnormalized()
 
         # fast case first, no offsets, no fuzz
         old, oldstart, new, newstart = h.fuzzit(0, False)
         oldstart += self.offset
         orig_start = oldstart
         # if there's skew we want to emit the "(offset %d lines)" even
         # when the hunk cleanly applies at start + skew, so skip the
         # fast case code
         if (self.skew == 0 and
             diffhelpers.testhunk(old, self.lines, oldstart) == 0):
             if self.remove:
                 self.backend.unlink(self.fname)
             else:
                 self.lines[oldstart:oldstart + len(old)] = new
                 self.offset += len(new) - len(old)
                 self.dirty = True
             return 0
 
         # ok, we couldn't match the hunk. Lets look for offsets and fuzz it
         self.hash = {}
         for x, s in enumerate(self.lines):
             self.hash.setdefault(s, []).append(x)
 
         for fuzzlen in xrange(self.ui.configint("patch", "fuzz", 2) + 1):
             for toponly in [True, False]:
                 old, oldstart, new, newstart = h.fuzzit(fuzzlen, toponly)
                 oldstart = oldstart + self.offset + self.skew
                 oldstart = min(oldstart, len(self.lines))
                 if old:
                     cand = self.findlines(old[0][1:], oldstart)
                 else:
                     # Only adding lines with no or fuzzed context, just
                     # take the skew in account
                     cand = [oldstart]
 
                 for l in cand:
                     if not old or diffhelpers.testhunk(old, self.lines, l) == 0:
                         self.lines[l : l + len(old)] = new
                         self.offset += len(new) - len(old)
                         self.skew = l - orig_start
                         self.dirty = True
                         offset = l - orig_start - fuzzlen
                         if fuzzlen:
                             msg = _("Hunk #%d succeeded at %d "
                                     "with fuzz %d "
                                     "(offset %d lines).\n")
                             self.printfile(True)
                             self.ui.warn(msg %
                                 (h.number, l + 1, fuzzlen, offset))
                         else:
                             msg = _("Hunk #%d succeeded at %d "
                                     "(offset %d lines).\n")
                             self.ui.note(msg % (h.number, l + 1, offset))
                         return fuzzlen
         self.printfile(True)
         self.ui.warn(_("Hunk #%d FAILED at %d\n") % (h.number, orig_start))
         self.rej.append(horig)
         return -1
 
     def close(self):
         if self.dirty:
             self.writelines(self.fname, self.lines, self.mode)
         self.write_rej()
         return len(self.rej)
 
 class header(object):
     """patch header
     """
     diffgit_re = re.compile('diff --git a/(.*) b/(.*)$')
     diff_re = re.compile('diff -r .* (.*)$')
     allhunks_re = re.compile('(?:index|deleted file) ')
     pretty_re = re.compile('(?:new file|deleted file) ')
     special_re = re.compile('(?:index|deleted|copy|rename) ')
     newfile_re = re.compile('(?:new file)')
 
     def __init__(self, header):
         self.header = header
         self.hunks = []
 
     def binary(self):
         return any(h.startswith('index ') for h in self.header)
 
     def pretty(self, fp):
         for h in self.header:
             if h.startswith('index '):
                 fp.write(_('this modifies a binary file (all or nothing)\n'))
                 break
             if self.pretty_re.match(h):
                 fp.write(h)
                 if self.binary():
                     fp.write(_('this is a binary file\n'))
                 break
             if h.startswith('---'):
                 fp.write(_('%d hunks, %d lines changed\n') %
                          (len(self.hunks),
                           sum([max(h.added, h.removed) for h in self.hunks])))
                 break
             fp.write(h)
 
     def write(self, fp):
         fp.write(''.join(self.header))
 
     def allhunks(self):
         return any(self.allhunks_re.match(h) for h in self.header)
 
     def files(self):
         match = self.diffgit_re.match(self.header[0])
         if match:
             fromfile, tofile = match.groups()
             if fromfile == tofile:
                 return [fromfile]
             return [fromfile, tofile]
         else:
             return self.diff_re.match(self.header[0]).groups()
 
     def filename(self):
         return self.files()[-1]
 
     def __repr__(self):
         return '<header %s>' % (' '.join(map(repr, self.files())))
 
     def isnewfile(self):
         return any(self.newfile_re.match(h) for h in self.header)
 
     def special(self):
         # Special files are shown only at the header level and not at the hunk
         # level for example a file that has been deleted is a special file.
         # The user cannot change the content of the operation, in the case of
         # the deleted file he has to take the deletion or not take it, he
         # cannot take some of it.
         # Newly added files are special if they are empty, they are not special
         # if they have some content as we want to be able to change it
         nocontent = len(self.header) == 2
         emptynewfile = self.isnewfile() and nocontent
         return emptynewfile or \
                 any(self.special_re.match(h) for h in self.header)
 
 class recordhunk(object):
     """patch hunk
 
     XXX shouldn't we merge this with the other hunk class?
     """
     maxcontext = 3
 
     def __init__(self, header, fromline, toline, proc, before, hunk, after):
         def trimcontext(number, lines):
             delta = len(lines) - self.maxcontext
             if False and delta > 0:
                 return number + delta, lines[:self.maxcontext]
             return number, lines
 
         self.header = header
         self.fromline, self.before = trimcontext(fromline, before)
         self.toline, self.after = trimcontext(toline, after)
         self.proc = proc
         self.hunk = hunk
         self.added, self.removed = self.countchanges(self.hunk)
 
     def __eq__(self, v):
         if not isinstance(v, recordhunk):
             return False
 
         return ((v.hunk == self.hunk) and
                 (v.proc == self.proc) and
                 (self.fromline == v.fromline) and
                 (self.header.files() == v.header.files()))
 
     def __hash__(self):
         return hash((tuple(self.hunk),
             tuple(self.header.files()),
             self.fromline,
             self.proc))
 
     def countchanges(self, hunk):
         """hunk -> (n+,n-)"""
         add = len([h for h in hunk if h[0] == '+'])
         rem = len([h for h in hunk if h[0] == '-'])
         return add, rem
 
     def write(self, fp):
         delta = len(self.before) + len(self.after)
         if self.after and self.after[-1] == '\\ No newline at end of file\n':
             delta -= 1
         fromlen = delta + self.removed
         tolen = delta + self.added
         fp.write('@@ -%d,%d +%d,%d @@%s\n' %
                  (self.fromline, fromlen, self.toline, tolen,
                   self.proc and (' ' + self.proc)))
         fp.write(''.join(self.before + self.hunk + self.after))
 
     pretty = write
 
     def filename(self):
         return self.header.filename()
 
     def __repr__(self):
         return '<hunk %r@%d>' % (self.filename(), self.fromline)
 
 def filterpatch(ui, headers, operation=None):
     """Interactively filter patch chunks into applied-only chunks"""
     if operation is None:
         operation = _('record')
 
     def prompt(skipfile, skipall, query, chunk):
         """prompt query, and process base inputs
 
         - y/n for the rest of file
         - y/n for the rest
         - ? (help)
         - q (quit)
 
         Return True/False and possibly updated skipfile and skipall.
         """
         newpatches = None
         if skipall is not None:
             return skipall, skipfile, skipall, newpatches
         if skipfile is not None:
             return skipfile, skipfile, skipall, newpatches
         while True:
             resps = _('[Ynesfdaq?]'
                       '$$ &Yes, record this change'
                       '$$ &No, skip this change'
                       '$$ &Edit this change manually'
                       '$$ &Skip remaining changes to this file'
                       '$$ Record remaining changes to this &file'
                       '$$ &Done, skip remaining changes and files'
                       '$$ Record &all changes to all remaining files'
                       '$$ &Quit, recording no changes'
                       '$$ &? (display help)')
             r = ui.promptchoice("%s %s" % (query, resps))
             ui.write("\n")
             if r == 8: # ?
                 for c, t in ui.extractchoices(resps)[1]:
                     ui.write('%s - %s\n' % (c, t.lower()))
                 continue
             elif r == 0: # yes
                 ret = True
             elif r == 1: # no
                 ret = False
             elif r == 2: # Edit patch
                 if chunk is None:
                     ui.write(_('cannot edit patch for whole file'))
                     ui.write("\n")
                     continue
                 if chunk.header.binary():
                     ui.write(_('cannot edit patch for binary file'))
                     ui.write("\n")
                     continue
                 # Patch comment based on the Git one (based on comment at end of
                 # https://mercurial-scm.org/wiki/RecordExtension)
                 phelp = '---' + _("""
 To remove '-' lines, make them ' ' lines (context).
 To remove '+' lines, delete them.
 Lines starting with # will be removed from the patch.
 
 If the patch applies cleanly, the edited hunk will immediately be
 added to the record list. If it does not apply cleanly, a rejects
 file will be generated: you can use that when you try again. If
 all lines of the hunk are removed, then the edit is aborted and
 the hunk is left unchanged.
 """)
                 (patchfd, patchfn) = tempfile.mkstemp(prefix="hg-editor-",
                         suffix=".diff", text=True)
                 ncpatchfp = None
                 try:
                     # Write the initial patch
                     f = os.fdopen(patchfd, "w")
                     chunk.header.write(f)
                     chunk.write(f)
                     f.write('\n'.join(['# ' + i for i in phelp.splitlines()]))
                     f.close()
                     # Start the editor and wait for it to complete
                     editor = ui.geteditor()
                     ret = ui.system("%s \"%s\"" % (editor, patchfn),
                                     environ={'HGUSER': ui.username()})
                     if ret != 0:
                         ui.warn(_("editor exited with exit code %d\n") % ret)
                         continue
                     # Remove comment lines
                     patchfp = open(patchfn)
                     ncpatchfp = cStringIO.StringIO()
                     for line in patchfp:
                         if not line.startswith('#'):
                             ncpatchfp.write(line)
                     patchfp.close()
                     ncpatchfp.seek(0)
                     newpatches = parsepatch(ncpatchfp)
                 finally:
                     os.unlink(patchfn)
                     del ncpatchfp
                 # Signal that the chunk shouldn't be applied as-is, but
                 # provide the new patch to be used instead.
                 ret = False
             elif r == 3: # Skip
                 ret = skipfile = False
             elif r == 4: # file (Record remaining)
                 ret = skipfile = True
             elif r == 5: # done, skip remaining
                 ret = skipall = False
             elif r == 6: # all
                 ret = skipall = True
             elif r == 7: # quit
                 raise error.Abort(_('user quit'))
             return ret, skipfile, skipall, newpatches
 
     seen = set()
     applied = {}        # 'filename' -> [] of chunks
     skipfile, skipall = None, None
     pos, total = 1, sum(len(h.hunks) for h in headers)
     for h in headers:
         pos += len(h.hunks)
         skipfile = None
         fixoffset = 0
         hdr = ''.join(h.header)
         if hdr in seen:
             continue
         seen.add(hdr)
         if skipall is None:
             h.pretty(ui)
         msg = (_('examine changes to %s?') %
                _(' and ').join("'%s'" % f for f in h.files()))
         r, skipfile, skipall, np = prompt(skipfile, skipall, msg, None)
         if not r:
             continue
         applied[h.filename()] = [h]
         if h.allhunks():
             applied[h.filename()] += h.hunks
             continue
         for i, chunk in enumerate(h.hunks):
             if skipfile is None and skipall is None:
                 chunk.pretty(ui)
             if total == 1:
                 msg = _("record this change to '%s'?") % chunk.filename()
             else:
                 idx = pos - len(h.hunks) + i
                 msg = _("record change %d/%d to '%s'?") % (idx, total,
                                                            chunk.filename())
             r, skipfile, skipall, newpatches = prompt(skipfile,
                     skipall, msg, chunk)
             if r:
                 if fixoffset:
                     chunk = copy.copy(chunk)
                     chunk.toline += fixoffset
                 applied[chunk.filename()].append(chunk)
             elif newpatches is not None:
                 for newpatch in newpatches:
                     for newhunk in newpatch.hunks:
                         if fixoffset:
                             newhunk.toline += fixoffset
                         applied[newhunk.filename()].append(newhunk)
             else:
                 fixoffset += chunk.removed - chunk.added
     return (sum([h for h in applied.itervalues()
                if h[0].special() or len(h) > 1], []), {})
 class hunk(object):
     def __init__(self, desc, num, lr, context):
         self.number = num
         self.desc = desc
         self.hunk = [desc]
         self.a = []
         self.b = []
         self.starta = self.lena = None
         self.startb = self.lenb = None
         if lr is not None:
             if context:
                 self.read_context_hunk(lr)
             else:
                 self.read_unified_hunk(lr)
 
     def getnormalized(self):
         """Return a copy with line endings normalized to LF."""
 
         def normalize(lines):
             nlines = []
             for line in lines:
                 if line.endswith('\r\n'):
                     line = line[:-2] + '\n'
                 nlines.append(line)
             return nlines
 
         # Dummy object, it is rebuilt manually
         nh = hunk(self.desc, self.number, None, None)
         nh.number = self.number
         nh.desc = self.desc
         nh.hunk = self.hunk
         nh.a = normalize(self.a)
         nh.b = normalize(self.b)
         nh.starta = self.starta
         nh.startb = self.startb
         nh.lena = self.lena
         nh.lenb = self.lenb
         return nh
 
     def read_unified_hunk(self, lr):
         m = unidesc.match(self.desc)
         if not m:
             raise PatchError(_("bad hunk #%d") % self.number)
         self.starta, self.lena, self.startb, self.lenb = m.groups()
         if self.lena is None:
             self.lena = 1
         else:
             self.lena = int(self.lena)
         if self.lenb is None:
             self.lenb = 1
         else:
             self.lenb = int(self.lenb)
         self.starta = int(self.starta)
         self.startb = int(self.startb)
         diffhelpers.addlines(lr, self.hunk, self.lena, self.lenb, self.a,
                              self.b)
         # if we hit eof before finishing out the hunk, the last line will
         # be zero length.  Lets try to fix it up.
         while len(self.hunk[-1]) == 0:
             del self.hunk[-1]
             del self.a[-1]
             del self.b[-1]
             self.lena -= 1
             self.lenb -= 1
         self._fixnewline(lr)
 
     def read_context_hunk(self, lr):
         self.desc = lr.readline()
         m = contextdesc.match(self.desc)
         if not m:
             raise PatchError(_("bad hunk #%d") % self.number)
         self.starta, aend = m.groups()
         self.starta = int(self.starta)
         if aend is None:
             aend = self.starta
         self.lena = int(aend) - self.starta
         if self.starta:
             self.lena += 1
         for x in xrange(self.lena):
             l = lr.readline()
             if l.startswith('---'):
                 # lines addition, old block is empty
                 lr.push(l)
                 break
             s = l[2:]
             if l.startswith('- ') or l.startswith('! '):
                 u = '-' + s
             elif l.startswith('  '):
                 u = ' ' + s
             else:
                 raise PatchError(_("bad hunk #%d old text line %d") %
                                  (self.number, x))
             self.a.append(u)
             self.hunk.append(u)
 
         l = lr.readline()
         if l.startswith('\ '):
             s = self.a[-1][:-1]
             self.a[-1] = s
             self.hunk[-1] = s
             l = lr.readline()
         m = contextdesc.match(l)
         if not m:
             raise PatchError(_("bad hunk #%d") % self.number)
         self.startb, bend = m.groups()
         self.startb = int(self.startb)
         if bend is None:
             bend = self.startb
         self.lenb = int(bend) - self.startb
         if self.startb:
             self.lenb += 1
         hunki = 1
         for x in xrange(self.lenb):
             l = lr.readline()
             if l.startswith('\ '):
                 # XXX: the only way to hit this is with an invalid line range.
                 # The no-eol marker is not counted in the line range, but I
                 # guess there are diff(1) out there which behave differently.
                 s = self.b[-1][:-1]
                 self.b[-1] = s
                 self.hunk[hunki - 1] = s
                 continue
             if not l:
                 # line deletions, new block is empty and we hit EOF
                 lr.push(l)
                 break
             s = l[2:]
             if l.startswith('+ ') or l.startswith('! '):
                 u = '+' + s
             elif l.startswith('  '):
                 u = ' ' + s
             elif len(self.b) == 0:
                 # line deletions, new block is empty
                 lr.push(l)
                 break
             else:
                 raise PatchError(_("bad hunk #%d old text line %d") %
                                  (self.number, x))
             self.b.append(s)
             while True:
                 if hunki >= len(self.hunk):
                     h = ""
                 else:
                     h = self.hunk[hunki]
                 hunki += 1
                 if h == u:
                     break
                 elif h.startswith('-'):
                     continue
                 else:
                     self.hunk.insert(hunki - 1, u)
                     break
 
         if not self.a:
             # this happens when lines were only added to the hunk
             for x in self.hunk:
                 if x.startswith('-') or x.startswith(' '):
                     self.a.append(x)
         if not self.b:
             # this happens when lines were only deleted from the hunk
             for x in self.hunk:
                 if x.startswith('+') or x.startswith(' '):
                     self.b.append(x[1:])
         # @@ -start,len +start,len @@
         self.desc = "@@ -%d,%d +%d,%d @@\n" % (self.starta, self.lena,
                                              self.startb, self.lenb)
         self.hunk[0] = self.desc
         self._fixnewline(lr)
 
     def _fixnewline(self, lr):
         l = lr.readline()
         if l.startswith('\ '):
             diffhelpers.fix_newline(self.hunk, self.a, self.b)
         else:
             lr.push(l)
 
     def complete(self):
         return len(self.a) == self.lena and len(self.b) == self.lenb
 
     def _fuzzit(self, old, new, fuzz, toponly):
         # this removes context lines from the top and bottom of list 'l'.  It
         # checks the hunk to make sure only context lines are removed, and then
         # returns a new shortened list of lines.
         fuzz = min(fuzz, len(old))
         if fuzz:
             top = 0
             bot = 0
             hlen = len(self.hunk)
             for x in xrange(hlen - 1):
                 # the hunk starts with the @@ line, so use x+1
                 if self.hunk[x + 1][0] == ' ':
                     top += 1
                 else:
                     break
             if not toponly:
                 for x in xrange(hlen - 1):
                     if self.hunk[hlen - bot - 1][0] == ' ':
                         bot += 1
                     else:
                         break
 
             bot = min(fuzz, bot)
             top = min(fuzz, top)
             return old[top:len(old) - bot], new[top:len(new) - bot], top
         return old, new, 0
 
     def fuzzit(self, fuzz, toponly):
         old, new, top = self._fuzzit(self.a, self.b, fuzz, toponly)
         oldstart = self.starta + top
         newstart = self.startb + top
         # zero length hunk ranges already have their start decremented
         if self.lena and oldstart > 0:
             oldstart -= 1
         if self.lenb and newstart > 0:
             newstart -= 1
         return old, oldstart, new, newstart
 
 class binhunk(object):
     'A binary patch file.'
     def __init__(self, lr, fname):
         self.text = None
         self.delta = False
         self.hunk = ['GIT binary patch\n']
         self._fname = fname
         self._read(lr)
 
     def complete(self):
         return self.text is not None
 
     def new(self, lines):
         if self.delta:
             return [applybindelta(self.text, ''.join(lines))]
         return [self.text]
 
     def _read(self, lr):
         def getline(lr, hunk):
             l = lr.readline()
             hunk.append(l)
             return l.rstrip('\r\n')
 
         size = 0
         while True:
             line = getline(lr, self.hunk)
             if not line:
                 raise PatchError(_('could not extract "%s" binary data')
                                  % self._fname)
             if line.startswith('literal '):
                 size = int(line[8:].rstrip())
                 break
             if line.startswith('delta '):
                 size = int(line[6:].rstrip())
                 self.delta = True
                 break
         dec = []
         line = getline(lr, self.hunk)
         while len(line) > 1:
             l = line[0]
             if l <= 'Z' and l >= 'A':
                 l = ord(l) - ord('A') + 1
             else:
                 l = ord(l) - ord('a') + 27
             try:
                 dec.append(base85.b85decode(line[1:])[:l])
             except ValueError as e:
                 raise PatchError(_('could not decode "%s" binary patch: %s')
                                  % (self._fname, str(e)))
             line = getline(lr, self.hunk)
         text = zlib.decompress(''.join(dec))
         if len(text) != size:
             raise PatchError(_('"%s" length is %d bytes, should be %d')
                              % (self._fname, len(text), size))
         self.text = text
 
 def parsefilename(str):
     # --- filename \t|space stuff
     s = str[4:].rstrip('\r\n')
     i = s.find('\t')
     if i < 0:
         i = s.find(' ')
         if i < 0:
             return s
     return s[:i]
 
 def reversehunks(hunks):
     '''reverse the signs in the hunks given as argument
 
     This function operates on hunks coming out of patch.filterpatch, that is
     a list of the form: [header1, hunk1, hunk2, header2...]. Example usage:
 
     >>> rawpatch = """diff --git a/folder1/g b/folder1/g
     ... --- a/folder1/g
     ... +++ b/folder1/g
     ... @@ -1,7 +1,7 @@
     ... +firstline
     ...  c
     ...  1
     ...  2
     ... + 3
     ... -4
     ...  5
     ...  d
     ... +lastline"""
     >>> hunks = parsepatch(rawpatch)
     >>> hunkscomingfromfilterpatch = []
     >>> for h in hunks:
     ...     hunkscomingfromfilterpatch.append(h)
     ...     hunkscomingfromfilterpatch.extend(h.hunks)
 
     >>> reversedhunks = reversehunks(hunkscomingfromfilterpatch)
     >>> fp = cStringIO.StringIO()
     >>> for c in reversedhunks:
     ...      c.write(fp)
     >>> fp.seek(0)
     >>> reversedpatch = fp.read()
     >>> print reversedpatch
     diff --git a/folder1/g b/folder1/g
     --- a/folder1/g
     +++ b/folder1/g
     @@ -1,4 +1,3 @@
     -firstline
      c
      1
      2
     @@ -1,6 +2,6 @@
      c
      1
      2
     - 3
     +4
      5
      d
     @@ -5,3 +6,2 @@
      5
      d
     -lastline
 
     '''
 
     from . import crecord as crecordmod
     newhunks = []
     for c in hunks:
         if isinstance(c, crecordmod.uihunk):
             # curses hunks encapsulate the record hunk in _hunk
             c = c._hunk
         if isinstance(c, recordhunk):
             for j, line in enumerate(c.hunk):
                 if line.startswith("-"):
                     c.hunk[j] = "+" + c.hunk[j][1:]
                 elif line.startswith("+"):
                     c.hunk[j] = "-" + c.hunk[j][1:]
             c.added, c.removed = c.removed, c.added
         newhunks.append(c)
     return newhunks
 
 def parsepatch(originalchunks):
     """patch -> [] of headers -> [] of hunks """
     class parser(object):
         """patch parsing state machine"""
         def __init__(self):
             self.fromline = 0
             self.toline = 0
             self.proc = ''
             self.header = None
             self.context = []
             self.before = []
             self.hunk = []
             self.headers = []
 
         def addrange(self, limits):
             fromstart, fromend, tostart, toend, proc = limits
             self.fromline = int(fromstart)
             self.toline = int(tostart)
             self.proc = proc
 
         def addcontext(self, context):
             if self.hunk:
                 h = recordhunk(self.header, self.fromline, self.toline,
                         self.proc, self.before, self.hunk, context)
                 self.header.hunks.append(h)
                 self.fromline += len(self.before) + h.removed
                 self.toline += len(self.before) + h.added
                 self.before = []
                 self.hunk = []
             self.context = context
 
         def addhunk(self, hunk):
             if self.context:
                 self.before = self.context
                 self.context = []
             self.hunk = hunk
 
         def newfile(self, hdr):
             self.addcontext([])
             h = header(hdr)
             self.headers.append(h)
             self.header = h
 
         def addother(self, line):
             pass # 'other' lines are ignored
 
         def finished(self):
             self.addcontext([])
             return self.headers
 
         transitions = {
             'file': {'context': addcontext,
                      'file': newfile,
                      'hunk': addhunk,
                      'range': addrange},
             'context': {'file': newfile,
                         'hunk': addhunk,
                         'range': addrange,
                         'other': addother},
             'hunk': {'context': addcontext,
                      'file': newfile,
                      'range': addrange},
             'range': {'context': addcontext,
                       'hunk': addhunk},
             'other': {'other': addother},
             }
 
     p = parser()
     fp = cStringIO.StringIO()
     fp.write(''.join(originalchunks))
     fp.seek(0)
 
     state = 'context'
     for newstate, data in scanpatch(fp):
         try:
             p.transitions[state][newstate](p, data)
         except KeyError:
             raise PatchError('unhandled transition: %s -> %s' %
                                    (state, newstate))
         state = newstate
     del fp
     return p.finished()
 
 def pathtransform(path, strip, prefix):
     '''turn a path from a patch into a path suitable for the repository
 
     prefix, if not empty, is expected to be normalized with a / at the end.
 
     Returns (stripped components, path in repository).
 
     >>> pathtransform('a/b/c', 0, '')
     ('', 'a/b/c')
     >>> pathtransform('   a/b/c   ', 0, '')
     ('', '   a/b/c')
     >>> pathtransform('   a/b/c   ', 2, '')
     ('a/b/', 'c')
     >>> pathtransform('a/b/c', 0, 'd/e/')
     ('', 'd/e/a/b/c')
     >>> pathtransform('   a//b/c   ', 2, 'd/e/')
     ('a//b/', 'd/e/c')
     >>> pathtransform('a/b/c', 3, '')
     Traceback (most recent call last):
     PatchError: unable to strip away 1 of 3 dirs from a/b/c
     '''
     pathlen = len(path)
     i = 0
     if strip == 0:
         return '', prefix + path.rstrip()
     count = strip
     while count > 0:
         i = path.find('/', i)
         if i == -1:
             raise PatchError(_("unable to strip away %d of %d dirs from %s") %
                              (count, strip, path))
         i += 1
         # consume '//' in the path
         while i < pathlen - 1 and path[i] == '/':
             i += 1
         count -= 1
     return path[:i].lstrip(), prefix + path[i:].rstrip()
 
 def makepatchmeta(backend, afile_orig, bfile_orig, hunk, strip, prefix):
     nulla = afile_orig == "/dev/null"
     nullb = bfile_orig == "/dev/null"
     create = nulla and hunk.starta == 0 and hunk.lena == 0
     remove = nullb and hunk.startb == 0 and hunk.lenb == 0
     abase, afile = pathtransform(afile_orig, strip, prefix)
     gooda = not nulla and backend.exists(afile)
     bbase, bfile = pathtransform(bfile_orig, strip, prefix)
     if afile == bfile:
         goodb = gooda
     else:
         goodb = not nullb and backend.exists(bfile)
     missing = not goodb and not gooda and not create
 
     # some diff programs apparently produce patches where the afile is
     # not /dev/null, but afile starts with bfile
     abasedir = afile[:afile.rfind('/') + 1]
     bbasedir = bfile[:bfile.rfind('/') + 1]
     if (missing and abasedir == bbasedir and afile.startswith(bfile)
         and hunk.starta == 0 and hunk.lena == 0):
         create = True
         missing = False
 
     # If afile is "a/b/foo" and bfile is "a/b/foo.orig" we assume the
     # diff is between a file and its backup. In this case, the original
     # file should be patched (see original mpatch code).
     isbackup = (abase == bbase and bfile.startswith(afile))
     fname = None
     if not missing:
         if gooda and goodb:
             if isbackup:
                 fname = afile
             else:
                 fname = bfile
         elif gooda:
             fname = afile
 
     if not fname:
         if not nullb:
             if isbackup:
                 fname = afile
             else:
                 fname = bfile
         elif not nulla:
             fname = afile
         else:
             raise PatchError(_("undefined source and destination files"))
 
     gp = patchmeta(fname)
     if create:
         gp.op = 'ADD'
     elif remove:
         gp.op = 'DELETE'
     return gp
 
 def scanpatch(fp):
     """like patch.iterhunks, but yield different events
 
     - ('file',    [header_lines + fromfile + tofile])
     - ('context', [context_lines])
     - ('hunk',    [hunk_lines])
     - ('range',   (-start,len, +start,len, proc))
     """
     lines_re = re.compile(r'@@ -(\d+),(\d+) \+(\d+),(\d+) @@\s*(.*)')
     lr = linereader(fp)
 
     def scanwhile(first, p):
         """scan lr while predicate holds"""
         lines = [first]
         while True:
             line = lr.readline()
             if not line:
                 break
             if p(line):
                 lines.append(line)
             else:
                 lr.push(line)
                 break
         return lines
 
     while True:
         line = lr.readline()
         if not line:
             break
         if line.startswith('diff --git a/') or line.startswith('diff -r '):
             def notheader(line):
                 s = line.split(None, 1)
                 return not s or s[0] not in ('---', 'diff')
             header = scanwhile(line, notheader)
             fromfile = lr.readline()
             if fromfile.startswith('---'):
                 tofile = lr.readline()
                 header += [fromfile, tofile]
             else:
                 lr.push(fromfile)
             yield 'file', header
         elif line[0] == ' ':
             yield 'context', scanwhile(line, lambda l: l[0] in ' \\')
         elif line[0] in '-+':
             yield 'hunk', scanwhile(line, lambda l: l[0] in '-+\\')
         else:
             m = lines_re.match(line)
             if m:
                 yield 'range', m.groups()
             else:
                 yield 'other', line
 
 def scangitpatch(lr, firstline):
     """
     Git patches can emit:
     - rename a to b
     - change b
     - copy a to c
     - change c
 
     We cannot apply this sequence as-is, the renamed 'a' could not be
     found for it would have been renamed already. And we cannot copy
     from 'b' instead because 'b' would have been changed already. So
     we scan the git patch for copy and rename commands so we can
     perform the copies ahead of time.
     """
     pos = 0
     try:
         pos = lr.fp.tell()
         fp = lr.fp
     except IOError:
         fp = cStringIO.StringIO(lr.fp.read())
     gitlr = linereader(fp)
     gitlr.push(firstline)
     gitpatches = readgitpatch(gitlr)
     fp.seek(pos)
     return gitpatches
 
 def iterhunks(fp):
     """Read a patch and yield the following events:
     - ("file", afile, bfile, firsthunk): select a new target file.
     - ("hunk", hunk): a new hunk is ready to be applied, follows a
     "file" event.
     - ("git", gitchanges): current diff is in git format, gitchanges
     maps filenames to gitpatch records. Unique event.
     """
     afile = ""
     bfile = ""
     state = None
     hunknum = 0
     emitfile = newfile = False
     gitpatches = None
 
     # our states
     BFILE = 1
     context = None
     lr = linereader(fp)
 
     while True:
         x = lr.readline()
         if not x:
             break
         if state == BFILE and (
             (not context and x[0] == '@')
             or (context is not False and x.startswith('***************'))
             or x.startswith('GIT binary patch')):
             gp = None
             if (gitpatches and
                 gitpatches[-1].ispatching(afile, bfile)):
                 gp = gitpatches.pop()
             if x.startswith('GIT binary patch'):
                 h = binhunk(lr, gp.path)
             else:
                 if context is None and x.startswith('***************'):
                     context = True
                 h = hunk(x, hunknum + 1, lr, context)
             hunknum += 1
             if emitfile:
                 emitfile = False
                 yield 'file', (afile, bfile, h, gp and gp.copy() or None)
             yield 'hunk', h
         elif x.startswith('diff --git a/'):
             m = gitre.match(x.rstrip(' \r\n'))
             if not m:
                 continue
             if gitpatches is None:
                 # scan whole input for git metadata
                 gitpatches = scangitpatch(lr, x)
                 yield 'git', [g.copy() for g in gitpatches
                               if g.op in ('COPY', 'RENAME')]
                 gitpatches.reverse()
             afile = 'a/' + m.group(1)
             bfile = 'b/' + m.group(2)
             while gitpatches and not gitpatches[-1].ispatching(afile, bfile):
                 gp = gitpatches.pop()
                 yield 'file', ('a/' + gp.path, 'b/' + gp.path, None, gp.copy())
             if not gitpatches:
                 raise PatchError(_('failed to synchronize metadata for "%s"')
                                  % afile[2:])
             gp = gitpatches[-1]
             newfile = True
         elif x.startswith('---'):
             # check for a unified diff
             l2 = lr.readline()
             if not l2.startswith('+++'):
                 lr.push(l2)
                 continue
             newfile = True
             context = False
             afile = parsefilename(x)
             bfile = parsefilename(l2)
         elif x.startswith('***'):
             # check for a context diff
             l2 = lr.readline()
             if not l2.startswith('---'):
                 lr.push(l2)
                 continue
             l3 = lr.readline()
             lr.push(l3)
             if not l3.startswith("***************"):
                 lr.push(l2)
                 continue
             newfile = True
             context = True
             afile = parsefilename(x)
             bfile = parsefilename(l2)
 
         if newfile:
             newfile = False
             emitfile = True
             state = BFILE
             hunknum = 0
 
     while gitpatches:
         gp = gitpatches.pop()
         yield 'file', ('a/' + gp.path, 'b/' + gp.path, None, gp.copy())
 
 def applybindelta(binchunk, data):
     """Apply a binary delta hunk
     The algorithm used is the algorithm from git's patch-delta.c
     """
     def deltahead(binchunk):
         i = 0
         for c in binchunk:
             i += 1
             if not (ord(c) & 0x80):
                 return i
         return i
     out = ""
     s = deltahead(binchunk)
     binchunk = binchunk[s:]
     s = deltahead(binchunk)
     binchunk = binchunk[s:]
     i = 0
     while i < len(binchunk):
         cmd = ord(binchunk[i])
         i += 1
         if (cmd & 0x80):
             offset = 0
             size = 0
             if (cmd & 0x01):
                 offset = ord(binchunk[i])
                 i += 1
             if (cmd & 0x02):
                 offset |= ord(binchunk[i]) << 8
                 i += 1
             if (cmd & 0x04):
                 offset |= ord(binchunk[i]) << 16
                 i += 1
             if (cmd & 0x08):
                 offset |= ord(binchunk[i]) << 24
                 i += 1
             if (cmd & 0x10):
                 size = ord(binchunk[i])
                 i += 1
             if (cmd & 0x20):
                 size |= ord(binchunk[i]) << 8
                 i += 1
             if (cmd & 0x40):
                 size |= ord(binchunk[i]) << 16
                 i += 1
             if size == 0:
                 size = 0x10000
             offset_end = offset + size
             out += data[offset:offset_end]
         elif cmd != 0:
             offset_end = i + cmd
             out += binchunk[i:offset_end]
             i += cmd
         else:
             raise PatchError(_('unexpected delta opcode 0'))
     return out
 
 def applydiff(ui, fp, backend, store, strip=1, prefix='', eolmode='strict'):
     """Reads a patch from fp and tries to apply it.
 
     Returns 0 for a clean patch, -1 if any rejects were found and 1 if
     there was any fuzz.
 
     If 'eolmode' is 'strict', the patch content and patched file are
     read in binary mode. Otherwise, line endings are ignored when
     patching then normalized according to 'eolmode'.
     """
     return _applydiff(ui, fp, patchfile, backend, store, strip=strip,
                       prefix=prefix, eolmode=eolmode)
 
 def _applydiff(ui, fp, patcher, backend, store, strip=1, prefix='',
                eolmode='strict'):
 
     if prefix:
         prefix = pathutil.canonpath(backend.repo.root, backend.repo.getcwd(),
                                     prefix)
         if prefix != '':
             prefix += '/'
     def pstrip(p):
         return pathtransform(p, strip - 1, prefix)[1]
 
     rejects = 0
     err = 0
     current_file = None
 
     for state, values in iterhunks(fp):
         if state == 'hunk':
             if not current_file:
                 continue
             ret = current_file.apply(values)
             if ret > 0:
                 err = 1
         elif state == 'file':
             if current_file:
                 rejects += current_file.close()
                 current_file = None
             afile, bfile, first_hunk, gp = values
             if gp:
                 gp.path = pstrip(gp.path)
                 if gp.oldpath:
                     gp.oldpath = pstrip(gp.oldpath)
             else:
                 gp = makepatchmeta(backend, afile, bfile, first_hunk, strip,
                                    prefix)
             if gp.op == 'RENAME':
                 backend.unlink(gp.oldpath)
             if not first_hunk:
                 if gp.op == 'DELETE':
                     backend.unlink(gp.path)
                     continue
                 data, mode = None, None
                 if gp.op in ('RENAME', 'COPY'):
                     data, mode = store.getfile(gp.oldpath)[:2]
                     # FIXME: failing getfile has never been handled here
                     assert data is not None
                 if gp.mode:
                     mode = gp.mode
                     if gp.op == 'ADD':
                         # Added files without content have no hunk and
                         # must be created
                         data = ''
                 if data or mode:
                     if (gp.op in ('ADD', 'RENAME', 'COPY')
                         and backend.exists(gp.path)):
                         raise PatchError(_("cannot create %s: destination "
                                            "already exists") % gp.path)
                     backend.setfile(gp.path, data, mode, gp.oldpath)
                 continue
             try:
                 current_file = patcher(ui, gp, backend, store,
                                        eolmode=eolmode)
             except PatchError as inst:
                 ui.warn(str(inst) + '\n')
                 current_file = None
                 rejects += 1
                 continue
         elif state == 'git':
             for gp in values:
                 path = pstrip(gp.oldpath)
                 data, mode = backend.getfile(path)
                 if data is None:
                     # The error ignored here will trigger a getfile()
                     # error in a place more appropriate for error
                     # handling, and will not interrupt the patching
                     # process.
                     pass
                 else:
                     store.setfile(path, data, mode)
         else:
             raise error.Abort(_('unsupported parser state: %s') % state)
 
     if current_file:
         rejects += current_file.close()
 
     if rejects:
         return -1
     return err
 
 def _externalpatch(ui, repo, patcher, patchname, strip, files,
                    similarity):
     """use <patcher> to apply <patchname> to the working directory.
     returns whether patch was applied with fuzz factor."""
 
     fuzz = False
     args = []
     cwd = repo.root
     if cwd:
         args.append('-d %s' % util.shellquote(cwd))
     fp = util.popen('%s %s -p%d < %s' % (patcher, ' '.join(args), strip,
                                        util.shellquote(patchname)))
     try:
         for line in fp:
             line = line.rstrip()
             ui.note(line + '\n')
             if line.startswith('patching file '):
                 pf = util.parsepatchoutput(line)
                 printed_file = False
                 files.add(pf)
             elif line.find('with fuzz') >= 0:
                 fuzz = True
                 if not printed_file:
                     ui.warn(pf + '\n')
                     printed_file = True
                 ui.warn(line + '\n')
             elif line.find('saving rejects to file') >= 0:
                 ui.warn(line + '\n')
             elif line.find('FAILED') >= 0:
                 if not printed_file:
                     ui.warn(pf + '\n')
                     printed_file = True
                 ui.warn(line + '\n')
     finally:
         if files:
             scmutil.marktouched(repo, files, similarity)
     code = fp.close()
     if code:
         raise PatchError(_("patch command failed: %s") %
                          util.explainexit(code)[0])
     return fuzz
 
 def patchbackend(ui, backend, patchobj, strip, prefix, files=None,
                  eolmode='strict'):
     if files is None:
         files = set()
     if eolmode is None:
         eolmode = ui.config('patch', 'eol', 'strict')
     if eolmode.lower() not in eolmodes:
         raise error.Abort(_('unsupported line endings type: %s') % eolmode)
     eolmode = eolmode.lower()
 
     store = filestore()
     try:
         fp = open(patchobj, 'rb')
     except TypeError:
         fp = patchobj
     try:
         ret = applydiff(ui, fp, backend, store, strip=strip, prefix=prefix,
                         eolmode=eolmode)
     finally:
         if fp != patchobj:
             fp.close()
         files.update(backend.close())
         store.close()
     if ret < 0:
         raise PatchError(_('patch failed to apply'))
     return ret > 0
 
 def internalpatch(ui, repo, patchobj, strip, prefix='', files=None,
                   eolmode='strict', similarity=0):
     """use builtin patch to apply <patchobj> to the working directory.
     returns whether patch was applied with fuzz factor."""
     backend = workingbackend(ui, repo, similarity)
     return patchbackend(ui, backend, patchobj, strip, prefix, files, eolmode)
 
 def patchrepo(ui, repo, ctx, store, patchobj, strip, prefix, files=None,
               eolmode='strict'):
     backend = repobackend(ui, repo, ctx, store)
     return patchbackend(ui, backend, patchobj, strip, prefix, files, eolmode)
 
 def patch(ui, repo, patchname, strip=1, prefix='', files=None, eolmode='strict',
           similarity=0):
     """Apply <patchname> to the working directory.
 
     'eolmode' specifies how end of lines should be handled. It can be:
     - 'strict': inputs are read in binary mode, EOLs are preserved
     - 'crlf': EOLs are ignored when patching and reset to CRLF
     - 'lf': EOLs are ignored when patching and reset to LF
     - None: get it from user settings, default to 'strict'
     'eolmode' is ignored when using an external patcher program.
 
     Returns whether patch was applied with fuzz factor.
     """
     patcher = ui.config('ui', 'patch')
     if files is None:
         files = set()
     if patcher:
         return _externalpatch(ui, repo, patcher, patchname, strip,
                               files, similarity)
     return internalpatch(ui, repo, patchname, strip, prefix, files, eolmode,
                          similarity)
 
 def changedfiles(ui, repo, patchpath, strip=1):
     backend = fsbackend(ui, repo.root)
     with open(patchpath, 'rb') as fp:
         changed = set()
         for state, values in iterhunks(fp):
             if state == 'file':
                 afile, bfile, first_hunk, gp = values
                 if gp:
                     gp.path = pathtransform(gp.path, strip - 1, '')[1]
                     if gp.oldpath:
                         gp.oldpath = pathtransform(gp.oldpath, strip - 1, '')[1]
                 else:
                     gp = makepatchmeta(backend, afile, bfile, first_hunk, strip,
                                        '')
                 changed.add(gp.path)
                 if gp.op == 'RENAME':
                     changed.add(gp.oldpath)
             elif state not in ('hunk', 'git'):
                 raise error.Abort(_('unsupported parser state: %s') % state)
         return changed
 
 class GitDiffRequired(Exception):
     pass
 
 def diffallopts(ui, opts=None, untrusted=False, section='diff'):
     '''return diffopts with all features supported and parsed'''
     return difffeatureopts(ui, opts=opts, untrusted=untrusted, section=section,
                            git=True, whitespace=True, formatchanging=True)
 
 diffopts = diffallopts
 
 def difffeatureopts(ui, opts=None, untrusted=False, section='diff', git=False,
                     whitespace=False, formatchanging=False):
     '''return diffopts with only opted-in features parsed
 
     Features:
     - git: git-style diffs
     - whitespace: whitespace options like ignoreblanklines and ignorews
     - formatchanging: options that will likely break or cause correctness issues
       with most diff parsers
     '''
     def get(key, name=None, getter=ui.configbool, forceplain=None):
         if opts:
             v = opts.get(key)
             if v:
                 return v
         if forceplain is not None and ui.plain():
             return forceplain
         return getter(section, name or key, None, untrusted=untrusted)
 
     # core options, expected to be understood by every diff parser
     buildopts = {
         'nodates': get('nodates'),
         'showfunc': get('show_function', 'showfunc'),
         'context': get('unified', getter=ui.config),
     }
 
     if git:
         buildopts['git'] = get('git')
     if whitespace:
         buildopts['ignorews'] = get('ignore_all_space', 'ignorews')
         buildopts['ignorewsamount'] = get('ignore_space_change',
                                           'ignorewsamount')
         buildopts['ignoreblanklines'] = get('ignore_blank_lines',
                                             'ignoreblanklines')
     if formatchanging:
         buildopts['text'] = opts and opts.get('text')
         buildopts['nobinary'] = get('nobinary', forceplain=False)
         buildopts['noprefix'] = get('noprefix', forceplain=False)
 
     return mdiff.diffopts(**buildopts)
 
 def diff(repo, node1=None, node2=None, match=None, changes=None, opts=None,
          losedatafn=None, prefix='', relroot=''):
     '''yields diff of changes to files between two nodes, or node and
     working directory.
 
     if node1 is None, use first dirstate parent instead.
     if node2 is None, compare node1 with working directory.
 
     losedatafn(**kwarg) is a callable run when opts.upgrade=True and
     every time some change cannot be represented with the current
     patch format. Return False to upgrade to git patch format, True to
     accept the loss or raise an exception to abort the diff. It is
     called with the name of current file being diffed as 'fn'. If set
     to None, patches will always be upgraded to git format when
     necessary.
 
     prefix is a filename prefix that is prepended to all filenames on
     display (used for subrepos).
 
     relroot, if not empty, must be normalized with a trailing /. Any match
     patterns that fall outside it will be ignored.'''
 
     if opts is None:
         opts = mdiff.defaultopts
 
     if not node1 and not node2:
         node1 = repo.dirstate.p1()
 
     def lrugetfilectx():
         cache = {}
         order = collections.deque()
         def getfilectx(f, ctx):
             fctx = ctx.filectx(f, filelog=cache.get(f))
             if f not in cache:
                 if len(cache) > 20:
                     del cache[order.popleft()]
                 cache[f] = fctx.filelog()
             else:
                 order.remove(f)
             order.append(f)
             return fctx
         return getfilectx
     getfilectx = lrugetfilectx()
 
     ctx1 = repo[node1]
     ctx2 = repo[node2]
 
     relfiltered = False
     if relroot != '' and match.always():
         # as a special case, create a new matcher with just the relroot
         pats = [relroot]
         match = scmutil.match(ctx2, pats, default='path')
         relfiltered = True
 
     if not changes:
         changes = repo.status(ctx1, ctx2, match=match)
     modified, added, removed = changes[:3]
 
     if not modified and not added and not removed:
         return []
 
     if repo.ui.debugflag:
         hexfunc = hex
     else:
         hexfunc = short
     revs = [hexfunc(node) for node in [ctx1.node(), ctx2.node()] if node]
 
     copy = {}
     if opts.git or opts.upgrade:
         copy = copies.pathcopies(ctx1, ctx2, match=match)
 
     if relroot is not None:
         if not relfiltered:
             # XXX this would ideally be done in the matcher, but that is
             # generally meant to 'or' patterns, not 'and' them. In this case we
             # need to 'and' all the patterns from the matcher with relroot.
             def filterrel(l):
                 return [f for f in l if f.startswith(relroot)]
             modified = filterrel(modified)
             added = filterrel(added)
             removed = filterrel(removed)
             relfiltered = True
         # filter out copies where either side isn't inside the relative root
         copy = dict(((dst, src) for (dst, src) in copy.iteritems()
                      if dst.startswith(relroot)
                      and src.startswith(relroot)))
 
     modifiedset = set(modified)
     addedset = set(added)
+    removedset = set(removed)
     for f in modified:
         if f not in ctx1:
             # Fix up added, since merged-in additions appear as
             # modifications during merges
             modifiedset.remove(f)
             addedset.add(f)
+    for f in removed:
+        if f not in ctx1:
+            # Merged-in additions that are then removed are reported as removed.
+            # They are not in ctx1, so We don't want to show them in the diff.
+            removedset.remove(f)
     modified = sorted(modifiedset)
     added = sorted(addedset)
+    removed = sorted(removedset)
 
     def difffn(opts, losedata):
         return trydiff(repo, revs, ctx1, ctx2, modified, added, removed,
                        copy, getfilectx, opts, losedata, prefix, relroot)
     if opts.upgrade and not opts.git:
         try:
             def losedata(fn):
                 if not losedatafn or not losedatafn(fn=fn):
                     raise GitDiffRequired
             # Buffer the whole output until we are sure it can be generated
             return list(difffn(opts.copy(git=False), losedata))
         except GitDiffRequired:
             return difffn(opts.copy(git=True), None)
     else:
         return difffn(opts, None)
 
 def difflabel(func, *args, **kw):
     '''yields 2-tuples of (output, label) based on the output of func()'''
     headprefixes = [('diff', 'diff.diffline'),
                     ('copy', 'diff.extended'),
                     ('rename', 'diff.extended'),
                     ('old', 'diff.extended'),
                     ('new', 'diff.extended'),
                     ('deleted', 'diff.extended'),
                     ('---', 'diff.file_a'),
                     ('+++', 'diff.file_b')]
     textprefixes = [('@', 'diff.hunk'),
                     ('-', 'diff.deleted'),
                     ('+', 'diff.inserted')]
     head = False
     for chunk in func(*args, **kw):
         lines = chunk.split('\n')
         for i, line in enumerate(lines):
             if i != 0:
                 yield ('\n', '')
             if head:
                 if line.startswith('@'):
                     head = False
             else:
                 if line and line[0] not in ' +-@\\':
                     head = True
             stripline = line
             diffline = False
             if not head and line and line[0] in '+-':
                 # highlight tabs and trailing whitespace, but only in
                 # changed lines
                 stripline = line.rstrip()
                 diffline = True
 
             prefixes = textprefixes
             if head:
                 prefixes = headprefixes
             for prefix, label in prefixes:
                 if stripline.startswith(prefix):
                     if diffline:
                         for token in tabsplitter.findall(stripline):
                             if '\t' == token[0]:
                                 yield (token, 'diff.tab')
                             else:
                                 yield (token, label)
                     else:
                         yield (stripline, label)
                     break
             else:
                 yield (line, '')
             if line != stripline:
                 yield (line[len(stripline):], 'diff.trailingwhitespace')
 
 def diffui(*args, **kw):
     '''like diff(), but yields 2-tuples of (output, label) for ui.write()'''
     return difflabel(diff, *args, **kw)
 
 def _filepairs(modified, added, removed, copy, opts):
     '''generates tuples (f1, f2, copyop), where f1 is the name of the file
     before and f2 is the the name after. For added files, f1 will be None,
     and for removed files, f2 will be None. copyop may be set to None, 'copy'
     or 'rename' (the latter two only if opts.git is set).'''
     gone = set()
 
     copyto = dict([(v, k) for k, v in copy.items()])
 
     addedset, removedset = set(added), set(removed)
 
     for f in sorted(modified + added + removed):
         copyop = None
         f1, f2 = f, f
         if f in addedset:
             f1 = None
             if f in copy:
                 if opts.git:
                     f1 = copy[f]
                     if f1 in removedset and f1 not in gone:
                         copyop = 'rename'
                         gone.add(f1)
                     else:
                         copyop = 'copy'
         elif f in removedset:
             f2 = None
             if opts.git:
                 # have we already reported a copy above?
                 if (f in copyto and copyto[f] in addedset
                     and copy[copyto[f]] == f):
                     continue
         yield f1, f2, copyop
 
 def trydiff(repo, revs, ctx1, ctx2, modified, added, removed,
             copy, getfilectx, opts, losedatafn, prefix, relroot):
     '''given input data, generate a diff and yield it in blocks
 
     If generating a diff would lose data like flags or binary data and
     losedatafn is not None, it will be called.
 
     relroot is removed and prefix is added to every path in the diff output.
 
     If relroot is not empty, this function expects every path in modified,
     added, removed and copy to start with it.'''
 
     def gitindex(text):
         if not text:
             text = ""
         l = len(text)
         s = util.sha1('blob %d\0' % l)
         s.update(text)
         return s.hexdigest()
 
     if opts.noprefix:
         aprefix = bprefix = ''
     else:
         aprefix = 'a/'
         bprefix = 'b/'
 
     def diffline(f, revs):
         revinfo = ' '.join(["-r %s" % rev for rev in revs])
         return 'diff %s %s' % (revinfo, f)
 
     date1 = util.datestr(ctx1.date())
     date2 = util.datestr(ctx2.date())
 
     gitmode = {'l': '120000', 'x': '100755', '': '100644'}
 
     if relroot != '' and (repo.ui.configbool('devel', 'all')
                           or repo.ui.configbool('devel', 'check-relroot')):
         for f in modified + added + removed + copy.keys() + copy.values():
             if f is not None and not f.startswith(relroot):
                 raise AssertionError(
                     "file %s doesn't start with relroot %s" % (f, relroot))
 
     for f1, f2, copyop in _filepairs(modified, added, removed, copy, opts):
         content1 = None
         content2 = None
         flag1 = None
         flag2 = None
         if f1:
             content1 = getfilectx(f1, ctx1).data()
             if opts.git or losedatafn:
                 flag1 = ctx1.flags(f1)
         if f2:
             content2 = getfilectx(f2, ctx2).data()
             if opts.git or losedatafn:
                 flag2 = ctx2.flags(f2)
         binary = False
         if opts.git or losedatafn:
             binary = util.binary(content1) or util.binary(content2)
 
         if losedatafn and not opts.git:
             if (binary or
                 # copy/rename
                 f2 in copy or
                 # empty file creation
                 (not f1 and not content2) or
                 # empty file deletion
                 (not content1 and not f2) or
                 # create with flags
                 (not f1 and flag2) or
                 # change flags
                 (f1 and f2 and flag1 != flag2)):
                 losedatafn(f2 or f1)
 
         path1 = f1 or f2
         path2 = f2 or f1
         path1 = posixpath.join(prefix, path1[len(relroot):])
         path2 = posixpath.join(prefix, path2[len(relroot):])
         header = []
         if opts.git:
             header.append('diff --git %s%s %s%s' %
                           (aprefix, path1, bprefix, path2))
             if not f1: # added
                 header.append('new file mode %s' % gitmode[flag2])
             elif not f2: # removed
                 header.append('deleted file mode %s' % gitmode[flag1])
             else:  # modified/copied/renamed
                 mode1, mode2 = gitmode[flag1], gitmode[flag2]
                 if mode1 != mode2:
                     header.append('old mode %s' % mode1)
                     header.append('new mode %s' % mode2)
                 if copyop is not None:
                     header.append('%s from %s' % (copyop, path1))
                     header.append('%s to %s' % (copyop, path2))
         elif revs and not repo.ui.quiet:
             header.append(diffline(path1, revs))
 
         if binary and opts.git and not opts.nobinary:
             text = mdiff.b85diff(content1, content2)
             if text:
                 header.append('index %s..%s' %
                               (gitindex(content1), gitindex(content2)))
         else:
             text = mdiff.unidiff(content1, date1,
                                  content2, date2,
                                  path1, path2, opts=opts)
         if header and (text or len(header) > 1):
             yield '\n'.join(header) + '\n'
         if text:
             yield text
 
 def diffstatsum(stats):
     maxfile, maxtotal, addtotal, removetotal, binary = 0, 0, 0, 0, False
     for f, a, r, b in stats:
         maxfile = max(maxfile, encoding.colwidth(f))
         maxtotal = max(maxtotal, a + r)
         addtotal += a
         removetotal += r
         binary = binary or b
 
     return maxfile, maxtotal, addtotal, removetotal, binary
 
 def diffstatdata(lines):
     diffre = re.compile('^diff .*-r [a-z0-9]+\s(.*)$')
 
     results = []
     filename, adds, removes, isbinary = None, 0, 0, False
 
     def addresult():
         if filename:
             results.append((filename, adds, removes, isbinary))
 
     for line in lines:
         if line.startswith('diff'):
             addresult()
             # set numbers to 0 anyway when starting new file
             adds, removes, isbinary = 0, 0, False
             if line.startswith('diff --git a/'):
                 filename = gitre.search(line).group(2)
             elif line.startswith('diff -r'):
                 # format: "diff -r ... -r ... filename"
                 filename = diffre.search(line).group(1)
         elif line.startswith('+') and not line.startswith('+++ '):
             adds += 1
         elif line.startswith('-') and not line.startswith('--- '):
             removes += 1
         elif (line.startswith('GIT binary patch') or
               line.startswith('Binary file')):
             isbinary = True
     addresult()
     return results
 
 def diffstat(lines, width=80, git=False):
     output = []
     stats = diffstatdata(lines)
     maxname, maxtotal, totaladds, totalremoves, hasbinary = diffstatsum(stats)
 
     countwidth = len(str(maxtotal))
     if hasbinary and countwidth < 3:
         countwidth = 3
     graphwidth = width - countwidth - maxname - 6
     if graphwidth < 10:
         graphwidth = 10
 
     def scale(i):
         if maxtotal <= graphwidth:
             return i
         # If diffstat runs out of room it doesn't print anything,
         # which isn't very useful, so always print at least one + or -
         # if there were at least some changes.
         return max(i * graphwidth // maxtotal, int(bool(i)))
 
     for filename, adds, removes, isbinary in stats:
         if isbinary:
             count = 'Bin'
         else:
             count = adds + removes
         pluses = '+' * scale(adds)
         minuses = '-' * scale(removes)
         output.append(' %s%s |  %*s %s%s\n' %
                       (filename, ' ' * (maxname - encoding.colwidth(filename)),
                        countwidth, count, pluses, minuses))
 
     if stats:
         output.append(_(' %d files changed, %d insertions(+), '
                         '%d deletions(-)\n')
                       % (len(stats), totaladds, totalremoves))
 
     return ''.join(output)
 
 def diffstatui(*args, **kw):
     '''like diffstat(), but yields 2-tuples of (output, label) for
     ui.write()
     '''
 
     for line in diffstat(*args, **kw).splitlines():
         if line and line[-1] in '+-':
             name, graph = line.rsplit(' ', 1)
             yield (name + ' ', '')
             m = re.search(r'\++', graph)
             if m:
                 yield (m.group(0), 'diffstat.inserted')
             m = re.search(r'-+', graph)
             if m:
                 yield (m.group(0), 'diffstat.deleted')
         else:
             yield (line, '')
         yield ('\n', '')

tests/test-diffdir.t

@@ -1,40 +1,70 b''
   $ hg init
   $ touch a
   $ hg add a
   $ hg ci -m "a"
 
   $ echo 123 > b
   $ hg add b
   $ hg diff --nodates
   diff -r 3903775176ed b
   --- /dev/null
   +++ b/b
   @@ -0,0 +1,1 @@
   +123
 
   $ hg diff --nodates -r tip
   diff -r 3903775176ed b
   --- /dev/null
   +++ b/b
   @@ -0,0 +1,1 @@
   +123
 
   $ echo foo > a
   $ hg diff --nodates
   diff -r 3903775176ed a
   --- a/a
   +++ b/a
   @@ -0,0 +1,1 @@
   +foo
   diff -r 3903775176ed b
   --- /dev/null
   +++ b/b
   @@ -0,0 +1,1 @@
   +123
 
   $ hg diff -r ""
   hg: parse error: empty query
   [255]
   $ hg diff -r tip -r ""
   hg: parse error: empty query
   [255]
+
+Remove a file that was added via merge. Since the file is not in parent 1,
+it should not be in the diff.
+
+  $ hg ci -m 'a=foo' a
+  $ hg co -Cq null
+  $ echo 123 > b
+  $ hg add b
+  $ hg ci -m "b"
+  created new head
+  $ hg merge 1
+  1 files updated, 0 files merged, 0 files removed, 0 files unresolved
+  (branch merge, don't forget to commit)
+  $ hg rm -f a
+  $ hg diff --nodates
+
+Rename a file that was added via merge. Since the rename source is not in
+parent 1, the diff should be relative to /dev/null
+
+  $ hg co -Cq 2
+  $ hg merge 1
+  1 files updated, 0 files merged, 0 files removed, 0 files unresolved
+  (branch merge, don't forget to commit)
+  $ hg mv a a2
+  $ hg diff --nodates
+  diff -r cf44b38435e5 a2
+  --- /dev/null
+  +++ b/a2
+  @@ -0,0 +1,1 @@
+  +foo