Add support for A+ files.

patch.py

 # coding=utf8
 # Copyright (c) 2011 The Chromium Authors. All rights reserved.
 # Use of this source code is governed by a BSD-style license that can be
 # found in the LICENSE file.
 """Utility functions to handle patches."""
 
 import posixpath
 import os
 import re
 
@@ skipping 15 matching lines @@
   """Defines a single file being modified.
 
   '/' is always used instead of os.sep for consistency.
   """
   is_delete = False
   is_binary = False
   is_new = False
 
   def __init__(self, filename):
     self.filename = self._process_filename(filename)
+    # Set when the file is copied or moved.
+    self.source_filename = None
 
   @staticmethod
   def _process_filename(filename):
     filename = filename.replace('\\', '/')
     # Blacklist a few characters for simplicity.
     for i in ('%', '$', '..', '\'', '"'):
       if i in filename:
         raise UnsupportedPatchFormat(
             filename, 'Can\'t use \'%s\' in filename.' % i)
     for i in ('/', 'CON', 'COM'):
       if filename.startswith(i):
         raise UnsupportedPatchFormat(
             filename, 'Filename can\'t start with \'%s\'.' % i)
     return filename
 
   def get(self):  # pragma: no coverage
     raise NotImplementedError('Nothing to grab')
 
   def set_relpath(self, relpath):
     if not relpath:
       return
     relpath = relpath.replace('\\', '/')
     if relpath[0] == '/':
       self._fail('Relative path starts with %s' % relpath[0])
     self.filename = self._process_filename(
         posixpath.join(relpath, self.filename))
+    if self.source_filename:
+      self.source_filename = self._process_filename(
+          posixpath.join(relpath, self.source_filename))
 
   def _fail(self, msg):
     """Shortcut function to raise UnsupportedPatchFormat."""
     raise UnsupportedPatchFormat(self.filename, msg)
 
 
 class FilePatchDelete(FilePatchBase):
   """Deletes a file."""
   is_delete = True
 
@@ skipping 33 matching lines @@
     if self.is_git_diff:
       self._verify_git_header()
     else:
       self._verify_svn_header()
 
   def get(self):
     return self.diff_header + self.diff_hunks
 
   def set_relpath(self, relpath):
     old_filename = self.filename
+    old_source_filename = self.source_filename or self.filename
     super(FilePatchDiff, self).set_relpath(relpath)
     # Update the header too.
-    self.diff_header = self.diff_header.replace(old_filename, self.filename)
+    source_filename = self.source_filename or self.filename
+    lines = self.diff_header.splitlines(True)
+    for i, line in enumerate(lines):
+      if line.startswith('diff --git'):
+        lines[i] = line.replace(
+            'a/' + old_source_filename, source_filename).replace(
+                'b/' + old_filename, self.filename)
+      elif re.match(r'^\w+ from .+$', line) or line.startswith('---'):
+        lines[i] = line.replace(old_source_filename, source_filename)
+      elif re.match(r'^\w+ to .+$', line) or line.startswith('+++'):
+        lines[i] = line.replace(old_filename, self.filename)
+    self.diff_header = ''.join(lines)
 
   def _split_header(self, diff):
     """Splits a diff in two: the header and the hunks."""
     header = []
     hunks = diff.splitlines(True)
     while hunks:
       header.append(hunks.pop(0))
       if header[-1].startswith('--- '):
         break
     else:
@@ skipping 51 matching lines @@
     """
     lines = self.diff_header.splitlines()
 
     # Verify the diff --git line.
     old = None
     new = None
     while lines:
       match = re.match(r'^diff \-\-git (.*?) (.*)$', lines.pop(0))
       if not match:
         continue
-      old = match.group(1).replace('\\', '/')
-      new = match.group(2).replace('\\', '/')
-      if old.startswith('a/') and new.startswith('b/'):
+      if match.group(1).startswith('a/') and match.group(2).startswith('b/'):
         self.patchlevel = 1
-        old = old[2:]
-        new = new[2:]
+      old = self.mangle(match.group(1))
+      new = self.mangle(match.group(2))
+
       # The rename is about the new file so the old file can be anything.
       if new not in (self.filename, 'dev/null'):
         self._fail('Unexpected git diff output name %s.' % new)
       if old == 'dev/null' and new == 'dev/null':
         self._fail('Unexpected /dev/null git diff.')
       break
 
     if not old or not new:
       self._fail('Unexpected git diff; couldn\'t find git header.')
 
+    if old not in (self.filename, 'dev/null'):
+      # Copy or rename.
+      self.source_filename = old
+
     last_line = ''
 
     while lines:
       line = lines.pop(0)
-      # TODO(maruel): old should be replace with self.source_file
-      # TODO(maruel): new == self.filename and remove new
-      self._verify_git_header_process_line(lines, line, last_line, old, new)
+      self._verify_git_header_process_line(lines, line, last_line)
       last_line = line
 
     # Cheap check to make sure the file name is at least mentioned in the
     # 'diff' header. That the only remaining invariant.
     if not self.filename in self.diff_header:
       self._fail('Diff seems corrupted.')
 
-  def _verify_git_header_process_line(self, lines, line, last_line, old, new):
+  def _verify_git_header_process_line(self, lines, line, last_line):
     """Processes a single line of the header.
 
     Returns True if it should continue looping.
 
     Format is described to
     http://www.kernel.org/pub/software/scm/git/docs/git-diff.html
     """
     match = re.match(r'^(rename|copy) from (.+)$', line)
+    old = self.source_filename or self.filename
     if match:
       if old != match.group(2):
         self._fail('Unexpected git diff input name for line %s.' % line)
       if not lines or not lines[0].startswith('%s to ' % match.group(1)):
         self._fail(
             'Confused %s from/to git diff for line %s.' %
                 (match.group(1), line))
       return
 
     match = re.match(r'^(rename|copy) to (.+)$', line)
     if match:
-      if new != match.group(2):
+      if self.filename != match.group(2):
         self._fail('Unexpected git diff output name for line %s.' % line)
       if not last_line.startswith('%s from ' % match.group(1)):
         self._fail(
             'Confused %s from/to git diff for line %s.' %
                 (match.group(1), line))
       return
 
     match = re.match(r'^new(| file) mode (\d{6})$', line)
     if match:
       mode = match.group(2)
       # Only look at owner ACL for executable.
       # TODO(maruel): Add support to remove a property.
       if bool(int(mode[4]) & 1):
         self.svn_properties.append(('svn:executable', '*'))
 
     match = re.match(r'^--- (.*)$', line)
     if match:
       if last_line[:3] in ('---', '+++'):
         self._fail('--- and +++ are reversed')
       self.is_new = match.group(1) == '/dev/null'
       # TODO(maruel): Use self.source_file.
-      if old != self.mangle(match.group(1)) and match.group(1) != '/dev/null':
+      if self.mangle(match.group(1)) not in (old, 'dev/null'):
         self._fail('Unexpected git diff: %s != %s.' % (old, match.group(1)))
       if not lines or not lines[0].startswith('+++'):
         self._fail('Missing git diff output name.')
       return
 
     match = re.match(r'^\+\+\+ (.*)$', line)
     if match:
       if not last_line.startswith('---'):
         self._fail('Unexpected git diff: --- not following +++.')
       # TODO(maruel): new == self.filename.
       if '/dev/null' == match.group(1):
         self.is_delete = True
-      elif new != self.mangle(match.group(1)):
-        self._fail('Unexpected git diff: %s != %s.' % (new, match.group(1)))
+      elif self.filename != self.mangle(match.group(1)):
+        self._fail(
+            'Unexpected git diff: %s != %s.' % (self.filename, match.group(1)))
       if lines:
         self._fail('Crap after +++')
       # We're done.
       return
 
   def _verify_svn_header(self):
     """Sanity checks the header.
 
     A svn diff can contain only property changes, in that case there will be no
     proper header. To make things worse, this property change header is
@@ skipping 15 matching lines @@
   def _verify_svn_header_process_line(self, lines, line, last_line):
     """Processes a single line of the header.
 
     Returns True if it should continue looping.
     """
     match = re.match(r'^--- ([^\t]+).*$', line)
     if match:
       if last_line[:3] in ('---', '+++'):
         self._fail('--- and +++ are reversed')
       self.is_new = match.group(1) == '/dev/null'
-      # For copy and renames, it's possible that the -- line doesn't match
-      # +++, so don't check match.group(1) to match self.filename or
-      # '/dev/null', it can be anything else.
-      # TODO(maruel): Handle rename/copy explicitly.
-      # if (self.mangle(match.group(1)) != self.filename and
-      #     match.group(1) != '/dev/null'):
-      #  self.source_file = match.group(1)
+      if (self.mangle(match.group(1)) != self.filename and
+           match.group(1) != '/dev/null'):
+        self.source_filename = match.group(1)
       if not lines or not lines[0].startswith('+++'):
         self._fail('Nothing after header.')
       return
 
     match = re.match(r'^\+\+\+ ([^\t]+).*$', line)
     if match:
       if not last_line.startswith('---'):
         self._fail('Unexpected diff: --- not following +++.')
       if match.group(1) == '/dev/null':
         self.is_delete = True
@@ skipping 18 matching lines @@
     for patch in self.patches:
       patch.set_relpath(relpath)
 
   def __iter__(self):
     for patch in self.patches:
       yield patch
 
   @property
   def filenames(self):
     return [p.filename for p in self.patches]