patch: support diff data loss detection and upgrade

In worst case, generating diff in upgrade mode can be two times more expensive
than generating it in git mode directly: we may have to regenerate the whole
diff again whenever a git feature is detected. Also, the first diff attempt is
completely buffered instead of being streamed. That said, even without having
profiled it yet, I am convinced we can fast-path the upgrade mode if necessary
were it to be used in regular diff commands, and not only in mq where avoiding
data loss is worth the price.

tests/autodiff.py

@@ -0,0 +1,46
+# Extension dedicated to test patch.diff() upgrade modes
+#
+#
+from mercurial import cmdutil, patch, util
+
+def autodiff(ui, repo, *pats, **opts):
+    diffopts = patch.diffopts(ui, opts)
+    git = opts.get('git', 'no')
+    brokenfiles = set()
+    losedatafn = None
+    if git in ('yes', 'no'):
+        diffopts.git = git == 'yes'
+        diffopts.upgrade = False
+    elif git == 'auto':
+        diffopts.git = False
+        diffopts.upgrade = True
+    elif git == 'warn':
+        diffopts.git = False
+        diffopts.upgrade = True
+        def losedatafn(fn=None, **kwargs):
+            brokenfiles.add(fn)
+            return True
+    elif git == 'abort':
+        diffopts.git = False
+        diffopts.upgrade = True
+        def losedatafn(fn=None, **kwargs):
+            raise util.Abort('losing data for %s' % fn)
+    else:
+        raise util.Abort('--git must be yes, no or auto')
+
+    node1, node2 = cmdutil.revpair(repo, [])
+    m = cmdutil.match(repo, pats, opts)
+    it = patch.diff(repo, node1, node2, match=m, opts=diffopts,
+                    losedatafn=losedatafn)
+    for chunk in it:
+        ui.write(chunk)
+    for fn in sorted(brokenfiles):
+        ui.write('data lost for: %s\n' % fn)
+
+cmdtable = {
+    "autodiff":
+        (autodiff,
+         [('', 'git', '', 'git upgrade mode (yes/no/auto/warn/abort)'),
+          ],
+         '[OPTION]... [FILE]...'),
+}

tests/test-diff-upgrade

@@ -0,0 +1,63
+#!/bin/sh
+
+echo "[extensions]" >> $HGRCPATH
+echo "autodiff=$TESTDIR/autodiff.py" >> $HGRCPATH
+echo "[diff]" >> $HGRCPATH
+echo "nodates=1" >> $HGRCPATH
+
+hg init repo
+cd repo
+echo '% make a combination of new, changed and deleted file'
+echo regular > regular
+echo rmregular > rmregular
+touch rmempty
+echo exec > exec
+chmod +x exec
+echo rmexec > rmexec
+chmod +x rmexec
+echo setexec > setexec
+echo unsetexec > unsetexec
+chmod +x unsetexec
+echo binary > binary
+python -c "file('rmbinary', 'wb').write('\0')"
+hg ci -Am addfiles
+echo regular >> regular
+echo newregular >> newregular
+rm rmempty
+touch newempty
+rm rmregular
+echo exec >> exec
+echo newexec > newexec
+chmod +x newexec
+rm rmexec
+chmod +x setexec
+chmod -x unsetexec
+python -c "file('binary', 'wb').write('\0\0')"
+python -c "file('newbinary', 'wb').write('\0')"
+rm rmbinary
+hg addremove
+
+echo '% git=no: regular diff for all files'
+hg autodiff --git=no
+
+echo '% git=no: git diff for single regular file'
+hg autodiff --git=yes regular
+
+echo '% git=auto: regular diff for regular files and removals'
+hg autodiff --git=auto regular newregular rmregular rmbinary rmexec
+
+for f in exec newexec setexec unsetexec binary newbinary newempty rmempty; do
+    echo '% git=auto: git diff for' $f
+    hg autodiff --git=auto $f
+done
+
+echo '% git=warn: regular diff with data loss warnings'
+hg autodiff --git=warn
+
+echo '% git=abort: fail on execute bit change'
+hg autodiff --git=abort regular setexec
+
+echo '% git=abort: succeed on regular file'
+hg autodiff --git=abort regular
+
+cd ..

tests/test-diff-upgrade.out

@@ -0,0 +1,186
+% make a combination of new, changed and deleted file
+adding binary
+adding exec
+adding regular
+adding rmbinary
+adding rmempty
+adding rmexec
+adding rmregular
+adding setexec
+adding unsetexec
+adding newbinary
+adding newempty
+adding newexec
+adding newregular
+removing rmbinary
+removing rmempty
+removing rmexec
+removing rmregular
+% git=no: regular diff for all files
+diff -r b3f053cd7c7f binary
+Binary file binary has changed
+diff -r b3f053cd7c7f exec
+--- a/exec
++++ b/exec
+@@ -1,1 +1,2 @@
+ exec
++exec
+diff -r b3f053cd7c7f newbinary
+Binary file newbinary has changed
+diff -r b3f053cd7c7f newexec
+--- /dev/null
++++ b/newexec
+@@ -0,0 +1,1 @@
++newexec
+diff -r b3f053cd7c7f newregular
+--- /dev/null
++++ b/newregular
+@@ -0,0 +1,1 @@
++newregular
+diff -r b3f053cd7c7f regular
+--- a/regular
++++ b/regular
+@@ -1,1 +1,2 @@
+ regular
++regular
+diff -r b3f053cd7c7f rmbinary
+Binary file rmbinary has changed
+diff -r b3f053cd7c7f rmexec
+--- a/rmexec
++++ /dev/null
+@@ -1,1 +0,0 @@
+-rmexec
+diff -r b3f053cd7c7f rmregular
+--- a/rmregular
++++ /dev/null
+@@ -1,1 +0,0 @@
+-rmregular
+% git=no: git diff for single regular file
+diff --git a/regular b/regular
+--- a/regular
++++ b/regular
+@@ -1,1 +1,2 @@
+ regular
++regular
+% git=auto: regular diff for regular files and removals
+diff -r b3f053cd7c7f newregular
+--- /dev/null
++++ b/newregular
+@@ -0,0 +1,1 @@
++newregular
+diff -r b3f053cd7c7f regular
+--- a/regular
++++ b/regular
+@@ -1,1 +1,2 @@
+ regular
++regular
+diff -r b3f053cd7c7f rmbinary
+Binary file rmbinary has changed
+diff -r b3f053cd7c7f rmexec
+--- a/rmexec
++++ /dev/null
+@@ -1,1 +0,0 @@
+-rmexec
+diff -r b3f053cd7c7f rmregular
+--- a/rmregular
++++ /dev/null
+@@ -1,1 +0,0 @@
+-rmregular
+% git=auto: git diff for exec
+diff -r b3f053cd7c7f exec
+--- a/exec
++++ b/exec
+@@ -1,1 +1,2 @@
+ exec
++exec
+% git=auto: git diff for newexec
+diff --git a/newexec b/newexec
+new file mode 100755
+--- /dev/null
++++ b/newexec
+@@ -0,0 +1,1 @@
++newexec
+% git=auto: git diff for setexec
+diff --git a/setexec b/setexec
+old mode 100644
+new mode 100755
+% git=auto: git diff for unsetexec
+diff --git a/unsetexec b/unsetexec
+old mode 100755
+new mode 100644
+% git=auto: git diff for binary
+diff --git a/binary b/binary
+index a9128c283485202893f5af379dd9beccb6e79486..09f370e38f498a462e1ca0faa724559b6630c04f
+GIT binary patch
+literal 2
+Jc${Nk0000200961
+
+% git=auto: git diff for newbinary
+diff --git a/newbinary b/newbinary
+new file mode 100644
+index 0000000000000000000000000000000000000000..f76dd238ade08917e6712764a16a22005a50573d
+GIT binary patch
+literal 1
+Ic${MZ000310RR91
+
+% git=auto: git diff for newempty
+diff --git a/newempty b/newempty
+new file mode 100644
+% git=auto: git diff for rmempty
+diff --git a/rmempty b/rmempty
+deleted file mode 100644
+% git=warn: regular diff with data loss warnings
+diff -r b3f053cd7c7f binary
+Binary file binary has changed
+diff -r b3f053cd7c7f exec
+--- a/exec
++++ b/exec
+@@ -1,1 +1,2 @@
+ exec
++exec
+diff -r b3f053cd7c7f newbinary
+Binary file newbinary has changed
+diff -r b3f053cd7c7f newexec
+--- /dev/null
++++ b/newexec
+@@ -0,0 +1,1 @@
++newexec
+diff -r b3f053cd7c7f newregular
+--- /dev/null
++++ b/newregular
+@@ -0,0 +1,1 @@
++newregular
+diff -r b3f053cd7c7f regular
+--- a/regular
++++ b/regular
+@@ -1,1 +1,2 @@
+ regular
++regular
+diff -r b3f053cd7c7f rmbinary
+Binary file rmbinary has changed
+diff -r b3f053cd7c7f rmexec
+--- a/rmexec
++++ /dev/null
+@@ -1,1 +0,0 @@
+-rmexec
+diff -r b3f053cd7c7f rmregular
+--- a/rmregular
++++ /dev/null
+@@ -1,1 +0,0 @@
+-rmregular
+data lost for: binary
+data lost for: newbinary
+data lost for: newempty
+data lost for: newexec
+data lost for: rmempty
+data lost for: setexec
+data lost for: unsetexec
+% git=abort: fail on execute bit change
+abort: losing data for setexec
+% git=abort: succeed on regular file
+diff -r b3f053cd7c7f regular
+--- a/regular
++++ b/regular
+@@ -1,1 +1,2 @@
+ regular
++regular

mercurial/mdiff.py

@@ -1,278 +1,281
 # mdiff.py - diff and patch routines for mercurial
 #
 # Copyright 2005, 2006 Matt Mackall <[e-mail]>
 #
 # This software may be used and distributed according to the terms of the
 # GNU General Public License version 2, incorporated herein by reference.
 
 from i18n import _
 import bdiff, mpatch, util
 import re, struct
 
 def splitnewlines(text):
     '''like str.splitlines, but only split on newlines.'''
     lines = [l + '\n' for l in text.split('\n')]
     if lines:
         if lines[-1] == '\n':
             lines.pop()
         else:
             lines[-1] = lines[-1][:-1]
     return lines
 
 class diffopts(object):
     '''context is the number of context lines
     text treats all files as text
     showfunc enables diff -p output
     git enables the git extended patch format
     nodates removes dates from diff headers
     ignorews ignores all whitespace changes in the diff
     ignorewsamount ignores changes in the amount of whitespace
-    ignoreblanklines ignores changes whose lines are all blank'''
+    ignoreblanklines ignores changes whose lines are all blank
+    upgrade generates git diffs to avoid data loss
+    '''
 
     defaults = {
         'context': 3,
         'text': False,
         'showfunc': False,
         'git': False,
         'nodates': False,
         'ignorews': False,
         'ignorewsamount': False,
         'ignoreblanklines': False,
+        'upgrade': False,
         }
 
     __slots__ = defaults.keys()
 
     def __init__(self, **opts):
         for k in self.__slots__:
             v = opts.get(k)
             if v is None:
                 v = self.defaults[k]
             setattr(self, k, v)
 
         try:
             self.context = int(self.context)
         except ValueError:
             raise util.Abort(_('diff context lines count must be '
                                'an integer, not %r') % self.context)
 
     def copy(self, **kwargs):
         opts = dict((k, getattr(self, k)) for k in self.defaults)
         opts.update(kwargs)
         return diffopts(**opts)
 
 defaultopts = diffopts()
 
 def wsclean(opts, text, blank=True):
     if opts.ignorews:
         text = re.sub('[ \t]+', '', text)
     elif opts.ignorewsamount:
         text = re.sub('[ \t]+', ' ', text)
         text = re.sub('[ \t]+\n', '\n', text)
     if blank and opts.ignoreblanklines:
         text = re.sub('\n+', '', text)
     return text
 
 def diffline(revs, a, b, opts):
     parts = ['diff']
     if opts.git:
         parts.append('--git')
     if revs and not opts.git:
         parts.append(' '.join(["-r %s" % rev for rev in revs]))
     if opts.git:
         parts.append('a/%s' % a)
         parts.append('b/%s' % b)
     else:
         parts.append(a)
     return ' '.join(parts) + '\n'
 
 def unidiff(a, ad, b, bd, fn1, fn2, r=None, opts=defaultopts):
     def datetag(date, addtab=True):
         if not opts.git and not opts.nodates:
             return '\t%s\n' % date
         if addtab and ' ' in fn1:
             return '\t\n'
         return '\n'
 
     if not a and not b: return ""
     epoch = util.datestr((0, 0))
 
     if not opts.text and (util.binary(a) or util.binary(b)):
         if a and b and len(a) == len(b) and a == b:
             return ""
         l = ['Binary file %s has changed\n' % fn1]
     elif not a:
         b = splitnewlines(b)
         if a is None:
             l1 = '--- /dev/null%s' % datetag(epoch, False)
         else:
             l1 = "--- %s%s" % ("a/" + fn1, datetag(ad))
         l2 = "+++ %s%s" % ("b/" + fn2, datetag(bd))
         l3 = "@@ -0,0 +1,%d @@\n" % len(b)
         l = [l1, l2, l3] + ["+" + e for e in b]
     elif not b:
         a = splitnewlines(a)
         l1 = "--- %s%s" % ("a/" + fn1, datetag(ad))
         if b is None:
             l2 = '+++ /dev/null%s' % datetag(epoch, False)
         else:
             l2 = "+++ %s%s" % ("b/" + fn2, datetag(bd))
         l3 = "@@ -1,%d +0,0 @@\n" % len(a)
         l = [l1, l2, l3] + ["-" + e for e in a]
     else:
         al = splitnewlines(a)
         bl = splitnewlines(b)
         l = list(bunidiff(a, b, al, bl, "a/" + fn1, "b/" + fn2, opts=opts))
         if not l: return ""
         # difflib uses a space, rather than a tab
         l[0] = "%s%s" % (l[0][:-2], datetag(ad))
         l[1] = "%s%s" % (l[1][:-2], datetag(bd))
 
     for ln in xrange(len(l)):
         if l[ln][-1] != '\n':
             l[ln] += "\n\ No newline at end of file\n"
 
     if r:
         l.insert(0, diffline(r, fn1, fn2, opts))
 
     return "".join(l)
 
 # somewhat self contained replacement for difflib.unified_diff
 # t1 and t2 are the text to be diffed
 # l1 and l2 are the text broken up into lines
 # header1 and header2 are the filenames for the diff output
 def bunidiff(t1, t2, l1, l2, header1, header2, opts=defaultopts):
     def contextend(l, len):
         ret = l + opts.context
         if ret > len:
             ret = len
         return ret
 
     def contextstart(l):
         ret = l - opts.context
         if ret < 0:
             return 0
         return ret
 
     def yieldhunk(hunk, header):
         if header:
             for x in header:
                 yield x
         (astart, a2, bstart, b2, delta) = hunk
         aend = contextend(a2, len(l1))
         alen = aend - astart
         blen = b2 - bstart + aend - a2
 
         func = ""
         if opts.showfunc:
             # walk backwards from the start of the context
             # to find a line starting with an alphanumeric char.
             for x in xrange(astart - 1, -1, -1):
                 t = l1[x].rstrip()
                 if funcre.match(t):
                     func = ' ' + t[:40]
                     break
 
         yield "@@ -%d,%d +%d,%d @@%s\n" % (astart + 1, alen,
                                            bstart + 1, blen, func)
         for x in delta:
             yield x
         for x in xrange(a2, aend):
             yield ' ' + l1[x]
 
     header = [ "--- %s\t\n" % header1, "+++ %s\t\n" % header2 ]
 
     if opts.showfunc:
         funcre = re.compile('\w')
 
     # bdiff.blocks gives us the matching sequences in the files.  The loop
     # below finds the spaces between those matching sequences and translates
     # them into diff output.
     #
     if opts.ignorews or opts.ignorewsamount:
         t1 = wsclean(opts, t1, False)
         t2 = wsclean(opts, t2, False)
 
     diff = bdiff.blocks(t1, t2)
     hunk = None
     for i, s1 in enumerate(diff):
         # The first match is special.
         # we've either found a match starting at line 0 or a match later
         # in the file.  If it starts later, old and new below will both be
         # empty and we'll continue to the next match.
         if i > 0:
             s = diff[i-1]
         else:
             s = [0, 0, 0, 0]
         delta = []
         a1 = s[1]
         a2 = s1[0]
         b1 = s[3]
         b2 = s1[2]
 
         old = l1[a1:a2]
         new = l2[b1:b2]
 
         # bdiff sometimes gives huge matches past eof, this check eats them,
         # and deals with the special first match case described above
         if not old and not new:
             continue
 
         if opts.ignoreblanklines:
             if wsclean(opts, "".join(old)) == wsclean(opts, "".join(new)):
                 continue
 
         astart = contextstart(a1)
         bstart = contextstart(b1)
         prev = None
         if hunk:
             # join with the previous hunk if it falls inside the context
             if astart < hunk[1] + opts.context + 1:
                 prev = hunk
                 astart = hunk[1]
                 bstart = hunk[3]
             else:
                 for x in yieldhunk(hunk, header):
                     yield x
                 # we only want to yield the header if the files differ, and
                 # we only want to yield it once.
                 header = None
         if prev:
             # we've joined the previous hunk, record the new ending points.
             hunk[1] = a2
             hunk[3] = b2
             delta = hunk[4]
         else:
             # create a new hunk
             hunk = [ astart, a2, bstart, b2, delta ]
 
         delta[len(delta):] = [ ' ' + x for x in l1[astart:a1] ]
         delta[len(delta):] = [ '-' + x for x in old ]
         delta[len(delta):] = [ '+' + x for x in new ]
 
     if hunk:
         for x in yieldhunk(hunk, header):
             yield x
 
 def patchtext(bin):
     pos = 0
     t = []
     while pos < len(bin):
         p1, p2, l = struct.unpack(">lll", bin[pos:pos + 12])
         pos += 12
         t.append(bin[pos:pos + l])
         pos += l
     return "".join(t)
 
 def patch(a, bin):
     return mpatch.patches(a, [bin])
 
 # similar to difflib.SequenceMatcher.get_matching_blocks
 def get_matching_blocks(a, b):
     return [(d[0], d[2], d[1] - d[0]) for d in bdiff.blocks(a, b)]
 
 def trivialdiffheader(length):
     return struct.pack(">lll", 0, 0, length)
 
 patches = mpatch.patches
 patchedsize = mpatch.patchedsize
 textdiff = bdiff.bdiff

mercurial/patch.py

@@ -1,1473 +1,1529
 # 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, incorporated herein by reference.
 
 from i18n import _
 from node import hex, nullid, short
 import base85, cmdutil, mdiff, util, diffhelpers, copies
 import cStringIO, email.Parser, os, re
 import sys, tempfile, zlib
 
 gitre = re.compile('diff --git a/(.*) b/(.*)')
 
 class PatchError(Exception):
     pass
 
 class NoHunks(PatchError):
     pass
 
 # helper functions
 
 def copyfile(src, dst, basedir):
     abssrc, absdst = [util.canonpath(basedir, basedir, x) for x in [src, dst]]
     if os.path.exists(absdst):
         raise util.Abort(_("cannot create %s: destination already exists") %
                          dst)
 
     dstdir = os.path.dirname(absdst)
     if dstdir and not os.path.isdir(dstdir):
         try:
             os.makedirs(dstdir)
         except IOError:
             raise util.Abort(
                 _("cannot create %s: unable to create destination directory")
                 % dst)
 
     util.copyfile(abssrc, absdst)
 
 # public functions
 
 def extract(ui, fileobj):
     '''extract patch from data read from fileobj.
 
     patch can be a normal patch or contained in an email message.
 
     return tuple (filename, message, user, date, node, p1, p2).
     Any item in the returned tuple can be None. If filename is None,
     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])', re.MULTILINE)
 
     fd, tmpname = tempfile.mkstemp(prefix='hg-patch-')
     tmpfp = os.fdopen(fd, 'w')
     try:
         msg = email.Parser.Parser().parse(fileobj)
 
         subject = msg['Subject']
         user = msg['From']
         if not subject and not user:
             # Not an email, restore parsed headers if any
             subject = '\n'.join(': '.join(h) for h in msg.items()) + '\n'
 
         gitsendmail = 'git-send-email' in msg.get('X-Mailer', '')
         # should try to parse msg['Date']
         date = None
         nodeid = None
         branch = None
         parents = []
 
         if subject:
             if subject.startswith('[PATCH'):
                 pend = subject.find(']')
                 if pend >= 0:
                     subject = subject[pend+1:].lstrip()
             subject = subject.replace('\n\t', ' ')
             ui.debug('Subject: %s\n' % subject)
         if user:
             ui.debug('From: %s\n' % 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
                 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'):
                         ui.debug('patch generated by hg export\n')
                         hgpatch = True
                         # drop earlier commit message content
                         cfp.seek(0)
                         cfp.truncate()
                         subject = None
                     elif hgpatch:
                         if line.startswith('# User '):
                             user = line[7:]
                             ui.debug('From: %s\n' % user)
                         elif line.startswith("# Date "):
                             date = line[7:]
                         elif line.startswith("# Branch "):
                             branch = line[9:]
                         elif line.startswith("# Node ID "):
                             nodeid = line[10:]
                         elif line.startswith("# Parent "):
                             parents.append(line[10:])
                     elif line == '---' and gitsendmail:
                         ignoretext = True
                     if not line.startswith('# ') 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:
         tmpfp.close()
         os.unlink(tmpname)
         raise
 
     if subject and not message.startswith(subject):
         message = '%s\n%s' % (subject, message)
     tmpfp.close()
     if not diffs_seen:
         os.unlink(tmpname)
         return None, message, user, date, branch, None, None, None
     p1 = parents and parents.pop(0) or None
     p2 = parents and parents.pop(0) or None
     return tmpname, message, user, date, branch, nodeid, p1, p2
 
 GP_PATCH  = 1 << 0  # we have to run patch
 GP_FILTER = 1 << 1  # there's some copy/rename operation
 GP_BINARY = 1 << 2  # there's a binary patch
 
 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.lineno = 0
         self.binary = False
 
     def setmode(self, mode):
         islink = mode & 020000
         isexec = mode & 0100
         self.mode = (islink, isexec)
 
 def readgitpatch(lr):
     """extract git-style metadata about patches from <patchname>"""
 
     # Filter patch for git information
     gp = None
     gitpatches = []
     # Can have a git patch with only metadata, causing patch to complain
     dopatch = 0
 
     lineno = 0
     for line in lr:
         lineno += 1
         line = line.rstrip(' \r\n')
         if line.startswith('diff --git'):
             m = gitre.match(line)
             if m:
                 if gp:
                     gitpatches.append(gp)
                 dst = m.group(2)
                 gp = patchmeta(dst)
                 gp.lineno = lineno
         elif gp:
             if line.startswith('--- '):
                 if gp.op in ('COPY', 'RENAME'):
                     dopatch |= GP_FILTER
                 gitpatches.append(gp)
                 gp = None
                 dopatch |= GP_PATCH
                 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'
                 # is the deleted file a symlink?
                 gp.setmode(int(line[-6:], 8))
             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'):
                 dopatch |= GP_BINARY
                 gp.binary = True
     if gp:
         gitpatches.append(gp)
 
     if not gitpatches:
         dopatch = GP_PATCH
 
     return (dopatch, gitpatches)
 
 class linereader(object):
     # simple class to allow pushing lines back into the input stream
     def __init__(self, fp, textmode=False):
         self.fp = fp
         self.buf = []
         self.textmode = textmode
         self.eol = None
 
     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
         l = self.fp.readline()
         if not self.eol:
             if l.endswith('\r\n'):
                 self.eol = '\r\n'
             elif l.endswith('\n'):
                 self.eol = '\n'
         if self.textmode and l.endswith('\r\n'):
             l = l[:-2] + '\n'
         return l
 
     def __iter__(self):
         while 1:
             l = self.readline()
             if not l:
                 break
             yield l
 
 # @@ -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, fname, opener, missing=False, eolmode='strict'):
         self.fname = fname
         self.eolmode = eolmode
         self.eol = None
         self.opener = opener
         self.ui = ui
         self.lines = []
         self.exists = False
         self.missing = missing
         if not missing:
             try:
                 self.lines = self.readlines(fname)
                 self.exists = True
             except IOError:
                 pass
         else:
             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 readlines(self, fname):
         if os.path.islink(fname):
             return [os.readlink(fname)]
         fp = self.opener(fname, 'r')
         try:
             lr = linereader(fp, self.eolmode != 'strict')
             lines = list(lr)
             self.eol = lr.eol
             return lines
         finally:
             fp.close()
 
     def writelines(self, fname, lines):
         # Ensure supplied data ends in fname, being a regular file or
         # a symlink. updatedir() will -too magically- take care of
         # setting it to the proper type afterwards.
         islink = os.path.islink(fname)
         if islink:
             fp = cStringIO.StringIO()
         else:
             fp = self.opener(fname, 'w')
         try:
             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':
                 for l in lines:
                     if l and l[-1] == '\n':
                         l = l[:-1] + eol
                     fp.write(l)
             else:
                 fp.writelines(lines)
             if islink:
                 self.opener.symlink(fp.getvalue(), fname)
         finally:
             fp.close()
 
     def unlink(self, fname):
         os.unlink(fname)
 
     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 hashlines(self):
         self.hash = {}
         for x, s in enumerate(self.lines):
             self.hash.setdefault(s, []).append(x)
 
     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
 
         fname = self.fname + ".rej"
         self.ui.warn(
             _("%d out of %d hunks FAILED -- saving rejects to file %s\n") %
             (len(self.rej), self.hunks, fname))
 
         def rejlines():
             base = os.path.basename(self.fname)
             yield "--- %s\n+++ %s\n" % (base, base)
             for x in self.rej:
                 for l in x.hunk:
                     yield l
                     if l[-1] != '\n':
                         yield "\n\ No newline at end of file\n"
 
         self.writelines(fname, rejlines())
 
     def write(self, dest=None):
         if not self.dirty:
             return
         if not dest:
             dest = self.fname
         self.writelines(dest, self.lines)
 
     def close(self):
         self.write()
         self.write_rej()
 
     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 h.createfile():
             self.ui.warn(_("file %s already exists\n") % self.fname)
             self.rej.append(h)
             return -1
 
         if isinstance(h, binhunk):
             if h.rmfile():
                 self.unlink(self.fname)
             else:
                 self.lines[:] = h.new()
                 self.offset += len(h.new())
                 self.dirty = 1
             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 = h.old()
         # patch starts counting at 1 unless we are adding the file
         if h.starta == 0:
             start = 0
         else:
             start = h.starta + self.offset - 1
         orig_start = start
         # 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, start) == 0:
             if h.rmfile():
                 self.unlink(self.fname)
             else:
                 self.lines[start : start + h.lena] = h.new()
                 self.offset += h.lenb - h.lena
                 self.dirty = 1
             return 0
 
         # ok, we couldn't match the hunk.  Lets look for offsets and fuzz it
         self.hashlines()
         if h.hunk[-1][0] != ' ':
             # if the hunk tried to put something at the bottom of the file
             # override the start line and use eof here
             search_start = len(self.lines)
         else:
             search_start = orig_start + self.skew
 
         for fuzzlen in xrange(3):
             for toponly in [ True, False ]:
                 old = h.old(fuzzlen, toponly)
 
                 cand = self.findlines(old[0][1:], search_start)
                 for l in cand:
                     if diffhelpers.testhunk(old, self.lines, l) == 0:
                         newlines = h.new(fuzzlen, toponly)
                         self.lines[l : l + len(old)] = newlines
                         self.offset += len(newlines) - len(old)
                         self.skew = l - orig_start
                         self.dirty = 1
                         if fuzzlen:
                             fuzzstr = "with fuzz %d " % fuzzlen
                             f = self.ui.warn
                             self.printfile(True)
                         else:
                             fuzzstr = ""
                             f = self.ui.note
                         offset = l - orig_start - fuzzlen
                         if offset == 1:
                             msg = _("Hunk #%d succeeded at %d %s"
                                     "(offset %d line).\n")
                         else:
                             msg = _("Hunk #%d succeeded at %d %s"
                                     "(offset %d lines).\n")
                         f(msg % (h.number, l+1, fuzzstr, 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
 
 class hunk(object):
     def __init__(self, desc, num, lr, context, create=False, remove=False):
         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)
         self.create = create
         self.remove = remove and not create
 
     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, False, False)
         nh.number = self.number
         nh.desc = self.desc
         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
         nh.create = self.create
         nh.remove = self.remove
         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, foo, self.lena, self.startb, foo2, 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
 
     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)
         foo, self.starta, foo2, aend, foo3 = 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('---'):
                 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)
         foo, self.startb, foo2, bend, foo3 = 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('\ '):
                 s = self.b[-1][:-1]
                 self.b[-1] = s
                 self.hunk[hunki-1] = s
                 continue
             if not l:
                 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:
                 # this can happen when the hunk does not add any lines
                 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
 
     def fix_newline(self):
         diffhelpers.fix_newline(self.hunk, self.a, self.b)
 
     def complete(self):
         return len(self.a) == self.lena and len(self.b) == self.lenb
 
     def createfile(self):
         return self.starta == 0 and self.lena == 0 and self.create
 
     def rmfile(self):
         return self.startb == 0 and self.lenb == 0 and self.remove
 
     def fuzzit(self, l, 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(l)-1)
         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
 
             # top and bot now count context in the hunk
             # adjust them if either one is short
             context = max(top, bot, 3)
             if bot < context:
                 bot = max(0, fuzz - (context - bot))
             else:
                 bot = min(fuzz, bot)
             if top < context:
                 top = max(0, fuzz - (context - top))
             else:
                 top = min(fuzz, top)
 
             return l[top:len(l)-bot]
         return l
 
     def old(self, fuzz=0, toponly=False):
         return self.fuzzit(self.a, fuzz, toponly)
 
     def new(self, fuzz=0, toponly=False):
         return self.fuzzit(self.b, fuzz, toponly)
 
 class binhunk:
     'A binary patch file. Only understands literals so far.'
     def __init__(self, gitpatch):
         self.gitpatch = gitpatch
         self.text = None
         self.hunk = ['GIT binary patch\n']
 
     def createfile(self):
         return self.gitpatch.op in ('ADD', 'RENAME', 'COPY')
 
     def rmfile(self):
         return self.gitpatch.op == 'DELETE'
 
     def complete(self):
         return self.text is not None
 
     def new(self):
         return [self.text]
 
     def extract(self, lr):
         line = lr.readline()
         self.hunk.append(line)
         while line and not line.startswith('literal '):
             line = lr.readline()
             self.hunk.append(line)
         if not line:
             raise PatchError(_('could not extract binary patch'))
         size = int(line[8:].rstrip())
         dec = []
         line = lr.readline()
         self.hunk.append(line)
         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
             dec.append(base85.b85decode(line[1:-1])[:l])
             line = lr.readline()
             self.hunk.append(line)
         text = zlib.decompress(''.join(dec))
         if len(text) != size:
             raise PatchError(_('binary patch is %d bytes, not %d') %
                              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 selectfile(afile_orig, bfile_orig, hunk, strip):
     def pathstrip(path, count=1):
         pathlen = len(path)
         i = 0
         if count == 0:
             return '', path.rstrip()
         while count > 0:
             i = path.find('/', i)
             if i == -1:
                 raise PatchError(_("unable to strip away %d dirs from %s") %
                                  (count, path))
             i += 1
             # consume '//' in the path
             while i < pathlen - 1 and path[i] == '/':
                 i += 1
             count -= 1
         return path[:i].lstrip(), path[i:].rstrip()
 
     nulla = afile_orig == "/dev/null"
     nullb = bfile_orig == "/dev/null"
     abase, afile = pathstrip(afile_orig, strip)
     gooda = not nulla and util.lexists(afile)
     bbase, bfile = pathstrip(bfile_orig, strip)
     if afile == bfile:
         goodb = gooda
     else:
         goodb = not nullb and os.path.exists(bfile)
     createfunc = hunk.createfile
     missing = not goodb and not gooda and not createfunc()
 
     # some diff programs apparently produce create patches where the
     # afile is not /dev/null, but rather the same name as the bfile
     if missing and afile == bfile:
         # this isn't very pretty
         hunk.create = True
         if createfunc():
             missing = False
         else:
             hunk.create = 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:
             fname = isbackup and afile or bfile
         elif gooda:
             fname = afile
 
     if not fname:
         if not nullb:
             fname = isbackup and afile or bfile
         elif not nulla:
             fname = afile
         else:
             raise PatchError(_("undefined source and destination files"))
 
     return fname, missing
 
 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, lr.textmode)
     gitlr.push(firstline)
     (dopatch, gitpatches) = readgitpatch(gitlr)
     fp.seek(pos)
     return dopatch, gitpatches
 
 def iterhunks(ui, fp, sourcefile=None):
     """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.
     """
     changed = {}
     current_hunk = None
     afile = ""
     bfile = ""
     state = None
     hunknum = 0
     emitfile = False
     git = False
 
     # our states
     BFILE = 1
     context = None
     lr = linereader(fp)
     dopatch = True
     # gitworkdone is True if a git operation (copy, rename, ...) was
     # performed already for the current file. Useful when the file
     # section may have no hunk.
     gitworkdone = False
 
     while True:
         newfile = False
         x = lr.readline()
         if not x:
             break
         if current_hunk:
             if x.startswith('\ '):
                 current_hunk.fix_newline()
             yield 'hunk', current_hunk
             current_hunk = None
             gitworkdone = False
         if ((sourcefile or state == BFILE) and ((not context and x[0] == '@') or
             ((context is not False) and x.startswith('***************')))):
             try:
                 if context is None and x.startswith('***************'):
                     context = True
                 gpatch = changed.get(bfile)
                 create = afile == '/dev/null' or gpatch and gpatch.op == 'ADD'
                 remove = bfile == '/dev/null' or gpatch and gpatch.op == 'DELETE'
                 current_hunk = hunk(x, hunknum + 1, lr, context, create, remove)
             except PatchError, err:
                 ui.debug(err)
                 current_hunk = None
                 continue
             hunknum += 1
             if emitfile:
                 emitfile = False
                 yield 'file', (afile, bfile, current_hunk)
         elif state == BFILE and x.startswith('GIT binary patch'):
             current_hunk = binhunk(changed[bfile])
             hunknum += 1
             if emitfile:
                 emitfile = False
                 yield 'file', ('a/' + afile, 'b/' + bfile, current_hunk)
             current_hunk.extract(lr)
         elif x.startswith('diff --git'):
             # check for git diff, scanning the whole patch file if needed
             m = gitre.match(x)
             if m:
                 afile, bfile = m.group(1, 2)
                 if not git:
                     git = True
                     dopatch, gitpatches = scangitpatch(lr, x)
                     yield 'git', gitpatches
                     for gp in gitpatches:
                         changed[gp.path] = gp
                 # else error?
                 # copy/rename + modify should modify target, not source
                 gp = changed.get(bfile)
                 if gp and gp.op in ('COPY', 'DELETE', 'RENAME', 'ADD'):
                     afile = bfile
                     gitworkdone = True
             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:
             emitfile = True
             state = BFILE
             hunknum = 0
     if current_hunk:
         if current_hunk.complete():
             yield 'hunk', current_hunk
         else:
             raise PatchError(_("malformed patch %s %s") % (afile,
                              current_hunk.desc))
 
     if hunknum == 0 and dopatch and not gitworkdone:
         raise NoHunks
 
 def applydiff(ui, fp, changed, strip=1, sourcefile=None, eolmode='strict'):
     """
     Reads a patch from fp and tries to apply it.
 
     The dict 'changed' is filled in with all of the filenames changed
     by the patch. 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'.
     """
     rejects = 0
     err = 0
     current_file = None
     gitpatches = None
     opener = util.opener(os.getcwd())
 
     def closefile():
         if not current_file:
             return 0
         current_file.close()
         return len(current_file.rej)
 
     for state, values in iterhunks(ui, fp, sourcefile):
         if state == 'hunk':
             if not current_file:
                 continue
             current_hunk = values
             ret = current_file.apply(current_hunk)
             if ret >= 0:
                 changed.setdefault(current_file.fname, None)
                 if ret > 0:
                     err = 1
         elif state == 'file':
             rejects += closefile()
             afile, bfile, first_hunk = values
             try:
                 if sourcefile:
                     current_file = patchfile(ui, sourcefile, opener, eolmode=eolmode)
                 else:
                     current_file, missing = selectfile(afile, bfile, first_hunk,
                                             strip)
                     current_file = patchfile(ui, current_file, opener, missing, eolmode)
             except PatchError, err:
                 ui.warn(str(err) + '\n')
                 current_file, current_hunk = None, None
                 rejects += 1
                 continue
         elif state == 'git':
             gitpatches = values
             cwd = os.getcwd()
             for gp in gitpatches:
                 if gp.op in ('COPY', 'RENAME'):
                     copyfile(gp.oldpath, gp.path, cwd)
                 changed[gp.path] = gp
         else:
             raise util.Abort(_('unsupported parser state: %s') % state)
 
     rejects += closefile()
 
     if rejects:
         return -1
     return err
 
 def diffopts(ui, opts=None, untrusted=False):
     def get(key, name=None, getter=ui.configbool):
         return ((opts and opts.get(key)) or
                 getter('diff', name or key, None, untrusted=untrusted))
     return mdiff.diffopts(
         text=opts and opts.get('text'),
         git=get('git'),
         nodates=get('nodates'),
         showfunc=get('show_function', 'showfunc'),
         ignorews=get('ignore_all_space', 'ignorews'),
         ignorewsamount=get('ignore_space_change', 'ignorewsamount'),
         ignoreblanklines=get('ignore_blank_lines', 'ignoreblanklines'),
         context=get('unified', getter=ui.config))
 
 def updatedir(ui, repo, patches, similarity=0):
     '''Update dirstate after patch application according to metadata'''
     if not patches:
         return
     copies = []
     removes = set()
     cfiles = patches.keys()
     cwd = repo.getcwd()
     if cwd:
         cfiles = [util.pathto(repo.root, cwd, f) for f in patches.keys()]
     for f in patches:
         gp = patches[f]
         if not gp:
             continue
         if gp.op == 'RENAME':
             copies.append((gp.oldpath, gp.path))
             removes.add(gp.oldpath)
         elif gp.op == 'COPY':
             copies.append((gp.oldpath, gp.path))
         elif gp.op == 'DELETE':
             removes.add(gp.path)
     for src, dst in copies:
         repo.copy(src, dst)
     if (not similarity) and removes:
         repo.remove(sorted(removes), True)
     for f in patches:
         gp = patches[f]
         if gp and gp.mode:
             islink, isexec = gp.mode
             dst = repo.wjoin(gp.path)
             # patch won't create empty files
             if gp.op == 'ADD' and not os.path.exists(dst):
                 flags = (isexec and 'x' or '') + (islink and 'l' or '')
                 repo.wwrite(gp.path, '', flags)
             elif gp.op != 'DELETE':
                 util.set_flags(dst, islink, isexec)
     cmdutil.addremove(repo, cfiles, similarity=similarity)
     files = patches.keys()
     files.extend([r for r in removes if r not in files])
     return sorted(files)
 
 def externalpatch(patcher, args, patchname, ui, strip, cwd, files):
     """use <patcher> to apply <patchname> to the working directory.
     returns whether patch was applied with fuzz factor."""
 
     fuzz = False
     if cwd:
         args.append('-d %s' % util.shellquote(cwd))
     fp = util.popen('%s %s -p%d < %s' % (patcher, ' '.join(args), strip,
                                        util.shellquote(patchname)))
 
     for line in fp:
         line = line.rstrip()
         ui.note(line + '\n')
         if line.startswith('patching file '):
             pf = util.parse_patch_output(line)
             printed_file = False
             files.setdefault(pf, None)
         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')
     code = fp.close()
     if code:
         raise PatchError(_("patch command failed: %s") %
                          util.explain_exit(code)[0])
     return fuzz
 
 def internalpatch(patchobj, ui, strip, cwd, files=None, eolmode='strict'):
     """use builtin patch to apply <patchobj> to the working directory.
     returns whether patch was applied with fuzz factor."""
 
     if files is None:
         files = {}
     if eolmode is None:
         eolmode = ui.config('patch', 'eol', 'strict')
     if eolmode.lower() not in eolmodes:
         raise util.Abort(_('Unsupported line endings type: %s') % eolmode)
     eolmode = eolmode.lower()
 
     try:
         fp = open(patchobj, 'rb')
     except TypeError:
         fp = patchobj
     if cwd:
         curdir = os.getcwd()
         os.chdir(cwd)
     try:
         ret = applydiff(ui, fp, files, strip=strip, eolmode=eolmode)
     finally:
         if cwd:
             os.chdir(curdir)
     if ret < 0:
         raise PatchError
     return ret > 0
 
 def patch(patchname, ui, strip=1, cwd=None, files=None, eolmode='strict'):
     """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')
     args = []
     if files is None:
         files = {}
     try:
         if patcher:
             return externalpatch(patcher, args, patchname, ui, strip, cwd,
                                  files)
         else:
             try:
                 return internalpatch(patchname, ui, strip, cwd, files, eolmode)
             except NoHunks:
                 patcher = util.find_exe('gpatch') or util.find_exe('patch') or 'patch'
                 ui.debug('no valid hunks found; trying with %r instead\n' %
                          patcher)
                 if util.needbinarypatch():
                     args.append('--binary')
                 return externalpatch(patcher, args, patchname, ui, strip, cwd,
                                      files)
     except PatchError, err:
         s = str(err)
         if s:
             raise util.Abort(s)
         else:
             raise util.Abort(_('patch failed to apply'))
 
 def b85diff(to, tn):
     '''print base85-encoded binary diff'''
     def gitindex(text):
         if not text:
             return '0' * 40
         l = len(text)
         s = util.sha1('blob %d\0' % l)
         s.update(text)
         return s.hexdigest()
 
     def fmtline(line):
         l = len(line)
         if l <= 26:
             l = chr(ord('A') + l - 1)
         else:
             l = chr(l - 26 + ord('a') - 1)
         return '%c%s\n' % (l, base85.b85encode(line, True))
 
     def chunk(text, csize=52):
         l = len(text)
         i = 0
         while i < l:
             yield text[i:i+csize]
             i += csize
 
     tohash = gitindex(to)
     tnhash = gitindex(tn)
     if tohash == tnhash:
         return ""
 
     # TODO: deltas
     ret = ['index %s..%s\nGIT binary patch\nliteral %s\n' %
            (tohash, tnhash, len(tn))]
     for l in chunk(zlib.compress(tn)):
         ret.append(fmtline(l))
     ret.append('\n')
     return ''.join(ret)
 
-def _addmodehdr(header, omode, nmode):
-    if omode != nmode:
-        header.append('old mode %s\n' % omode)
-        header.append('new mode %s\n' % nmode)
+class GitDiffRequired(Exception):
+    pass
 
-def diff(repo, node1=None, node2=None, match=None, changes=None, opts=None):
+def diff(repo, node1=None, node2=None, match=None, changes=None, opts=None,
+         losedatafn=None):
     '''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.'''
+    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.
+    '''
 
     if opts is None:
         opts = mdiff.defaultopts
 
     if not node1 and not node2:
         node1 = repo.dirstate.parents()[0]
 
     def lrugetfilectx():
         cache = {}
         order = []
         def getfilectx(f, ctx):
             fctx = ctx.filectx(f, filelog=cache.get(f))
             if f not in cache:
                 if len(cache) > 20:
                     del cache[order.pop(0)]
                 cache[f] = fctx.filelog()
             else:
                 order.remove(f)
             order.append(f)
             return fctx
         return getfilectx
     getfilectx = lrugetfilectx()
 
     ctx1 = repo[node1]
     ctx2 = repo[node2]
 
     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
+        return []
+
+    revs = None
+    if not repo.ui.quiet:
+        hexfunc = repo.ui.debugflag and hex or short
+        revs = [hexfunc(node) for node in [node1, node2] if node]
+
+    copy = {}
+    if opts.git or opts.upgrade:
+        copy = copies.copies(repo, ctx1, ctx2, repo[nullid])[0]
+        copy = copy.copy()
+        for k, v in copy.items():
+            copy[v] = k
+
+    difffn = lambda opts, losedata: trydiff(repo, revs, ctx1, ctx2,
+                 modified, added, removed, copy, getfilectx, opts, losedata)
+    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 _addmodehdr(header, omode, nmode):
+    if omode != nmode:
+        header.append('old mode %s\n' % omode)
+        header.append('new mode %s\n' % nmode)
+
+def trydiff(repo, revs, ctx1, ctx2, modified, added, removed,
+            copy, getfilectx, opts, losedatafn):
 
     date1 = util.datestr(ctx1.date())
     man1 = ctx1.manifest()
 
-    revs = None
-    if not repo.ui.quiet and not opts.git:
-        hexfunc = repo.ui.debugflag and hex or short
-        revs = [hexfunc(node) for node in [node1, node2] if node]
+    gone = set()
+    gitmode = {'l': '120000', 'x': '100755', '': '100644'}
 
     if opts.git:
-        copy, diverge = copies.copies(repo, ctx1, ctx2, repo[nullid])
-        copy = copy.copy()
-        for k, v in copy.items():
-            copy[v] = k
-
-    gone = set()
-    gitmode = {'l': '120000', 'x': '100755', '': '100644'}
+        revs = None
 
     for f in sorted(modified + added + removed):
         to = None
         tn = None
         dodiff = True
         header = []
         if f in man1:
             to = getfilectx(f, ctx1).data()
         if f not in removed:
             tn = getfilectx(f, ctx2).data()
         a, b = f, f
-        if opts.git:
+        if opts.git or losedatafn:
             if f in added:
                 mode = gitmode[ctx2.flags(f)]
                 if f in copy:
+                    if opts.git:
                     a = copy[f]
                     omode = gitmode[man1.flags(a)]
                     _addmodehdr(header, omode, mode)
                     if a in removed and a not in gone:
                         op = 'rename'
                         gone.add(a)
                     else:
                         op = 'copy'
                     header.append('%s from %s\n' % (op, a))
                     header.append('%s to %s\n' % (op, f))
                     to = getfilectx(a, ctx1).data()
                 else:
+                        losedatafn(f)
+                else:
+                    if opts.git:
                     header.append('new file mode %s\n' % mode)
+                    elif ctx2.flags(f):
+                        losedatafn(f)
                 if util.binary(tn):
+                    if opts.git:
                     dodiff = 'binary'
+                    else:
+                        losedatafn(f)
+                if not opts.git and not tn:
+                    # regular diffs cannot represent new empty file
+                    losedatafn(f)
             elif f in removed:
+                if opts.git:
                 # have we already reported a copy above?
                 if f in copy and copy[f] in added and copy[copy[f]] == f:
                     dodiff = False
                 else:
                     header.append('deleted file mode %s\n' %
                                   gitmode[man1.flags(f)])
+                elif not to:
+                    # regular diffs cannot represent empty file deletion
+                    losedatafn(f)
             else:
-                omode = gitmode[man1.flags(f)]
-                nmode = gitmode[ctx2.flags(f)]
-                _addmodehdr(header, omode, nmode)
-                if util.binary(to) or util.binary(tn):
+                oflag = man1.flags(f)
+                nflag = ctx2.flags(f)
+                binary = util.binary(to) or util.binary(tn)
+                if opts.git:
+                    _addmodehdr(header, gitmode[oflag], gitmode[nflag])
+                    if binary:
                     dodiff = 'binary'
+                elif binary or nflag != oflag:
+                    losedatafn(f)
+            if opts.git:
             header.insert(0, mdiff.diffline(revs, a, b, opts))
+
         if dodiff:
             if dodiff == 'binary':
                 text = b85diff(to, tn)
             else:
                 text = mdiff.unidiff(to, date1,
                                     # ctx2 date may be dynamic
                                     tn, util.datestr(ctx2.date()),
                                     a, b, revs, opts=opts)
             if header and (text or len(header) > 1):
                 yield ''.join(header)
             if text:
                 yield text
 
 def export(repo, revs, template='hg-%h.patch', fp=None, switch_parent=False,
            opts=None):
     '''export changesets as hg patches.'''
 
     total = len(revs)
     revwidth = max([len(str(rev)) for rev in revs])
 
     def single(rev, seqno, fp):
         ctx = repo[rev]
         node = ctx.node()
         parents = [p.node() for p in ctx.parents() if p]
         branch = ctx.branch()
         if switch_parent:
             parents.reverse()
         prev = (parents and parents[0]) or nullid
 
         if not fp:
             fp = cmdutil.make_file(repo, template, node, total=total,
                                    seqno=seqno, revwidth=revwidth,
                                    mode='ab')
         if fp != sys.stdout and hasattr(fp, 'name'):
             repo.ui.note("%s\n" % fp.name)
 
         fp.write("# HG changeset patch\n")
         fp.write("# User %s\n" % ctx.user())
         fp.write("# Date %d %d\n" % ctx.date())
         if branch and (branch != 'default'):
             fp.write("# Branch %s\n" % branch)
         fp.write("# Node ID %s\n" % hex(node))
         fp.write("# Parent  %s\n" % hex(prev))
         if len(parents) > 1:
             fp.write("# Parent  %s\n" % hex(parents[1]))
         fp.write(ctx.description().rstrip())
         fp.write("\n\n")
 
         for chunk in diff(repo, prev, node, opts=opts):
             fp.write(chunk)
 
     for seqno, rev in enumerate(revs):
         single(rev, seqno+1, fp)
 
 def diffstatdata(lines):
     filename, adds, removes = None, 0, 0
     for line in lines:
         if line.startswith('diff'):
             if filename:
                 isbinary = adds == 0 and removes == 0
                 yield (filename, adds, removes, isbinary)
             # set numbers to 0 anyway when starting new file
             adds, removes = 0, 0
             if line.startswith('diff --git'):
                 filename = gitre.search(line).group(1)
             else:
                 # format: "diff -r ... -r ... filename"
                 filename = line.split(None, 5)[-1]
         elif line.startswith('+') and not line.startswith('+++'):
             adds += 1
         elif line.startswith('-') and not line.startswith('---'):
             removes += 1
     if filename:
         isbinary = adds == 0 and removes == 0
         yield (filename, adds, removes, isbinary)
 
 def diffstat(lines, width=80, git=False):
     output = []
     stats = list(diffstatdata(lines))
 
     maxtotal, maxname = 0, 0
     totaladds, totalremoves = 0, 0
     hasbinary = False
     for filename, adds, removes, isbinary in stats:
         totaladds += adds
         totalremoves += removes
         maxname = max(maxname, len(filename))
         maxtotal = max(maxtotal, adds+removes)
         if isbinary:
             hasbinary = True
 
     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 git and isbinary:
             count = 'Bin'
         else:
             count = adds + removes
         pluses = '+' * scale(adds)
         minuses = '-' * scale(removes)
         output.append(' %-*s |  %*s %s%s\n' % (maxname, filename, countwidth,
                                                count, pluses, minuses))
 
     if stats:
         output.append(_(' %d files changed, %d insertions(+), %d deletions(-)\n')
                       % (len(stats), totaladds, totalremoves))
 
     return ''.join(output)