Automatically sort headers in files affected by move_source_file.py

tools/git/move_source_file.py

 #!/usr/bin/env python
 # Copyright (c) 2012 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.
 
 """Moves a C++ file to a new location, updating any include paths that
-point to it.  Updates include guards in moved header files.  Assumes
-Chromium coding style.
+point to it, and re-ordering headers as needed.  Updates include
+guards in moved header files.  Assumes Chromium coding style.
 
-Does not reorder headers; instead, use this after committing all of
-your moves:
-  ./tools/git/for-all-touched-files.py -c "tools/sort-headers.py [[FILENAME]]"
+Attempts to update paths used in .gyp(i) files, but does not reorder
+or restructure .gyp(i) files in any way.
 
-Updates paths used in .gyp(i) files, but does not reorder or
-restructure .gyp(i) files in any way.
+Updates full-path references to files in // comments in source files.
 
-Must run in a git checkout, as it relies on git for a fast way to find
-files that reference the moved file.
+Must run in a git checkout, as it relies on git grep for a fast way to
+find files that reference the moved file.
 """
 
 
 import os
 import re
 import subprocess
 import sys
 
+if __name__ == '__main__':
+  # Need to add the directory containing sort-headers.py to the Python
+  # classpath.
+  sys.path.append(os.path.abspath(os.path.join(sys.path[0], '..')))
+sort_headers = __import__('sort-headers')
+
+
 HANDLED_EXTENSIONS = ['.cc', '.mm', '.h', '.hh']
 
 
 def MakeDestinationPath(from_path, to_path):
   """Given the from and to paths, return a correct destination path.
 
   The initial destination path may either a full path or a directory,
   in which case the path must end with /.  Also does basic sanity
   checks.
   """
@@ skipping 10 matching lines @@
 
 def MoveFile(from_path, to_path):
   """Performs a git mv command to move a file from |from_path| to |to_path|.
   """
   if not os.system('git mv %s %s' % (from_path, to_path)) == 0:
     raise Exception('Fatal: Failed to run git mv command.')
 
 
 def MultiFileFindReplace(original,
                          replacement,
-                         grep_pattern,
-                         file_globs,
-                         guard_formats):
-  """Implements fast multi-file find and replace with optional guards.
+                         file_globs):
+  """Implements fast multi-file find and replace.
 
-  Given an |original| string and a |replacement| string, search for
-  them by formatting |grep_pattern| with |original| and running
-  git grep on the result, for files matching any of |file_globs|.
+  Given an |original| string and a |replacement| string, find matching
+  files by running git grep on |original| in files matching any
+  pattern in |file_globs|.
 
-  Once files are found, the function searches for any of
-  |guard_formats| formatted with |original| and replaces each match
-  with the same guard format as matched, formatted with |replacement|.
+  Once files are found, |re.sub| is run to replace |original| with
+  |replacement|.  |replacement| may use capture group back-references.
 
   Args:
-    original: 'chrome/browser/ui/browser.h'
-    replacement: 'chrome/browser/ui/browser/browser.h'
-    grep_pattern: r'#(include|import)\s*["<]%s[>"]'
+    original: '(#(include|import)\s*["<])chrome/browser/ui/browser.h([>"])'
+    replacement: '\1chrome/browser/ui/browser/browser.h\3'
     file_globs: ['*.cc', '*.h', '*.m', '*.mm']
-    guard_formats: None or ('"%s"', '<%s>')
+
+  Returns the list of files modified.
 
   Raises an exception on error.
   """
   out, err = subprocess.Popen(
-    ['git', 'grep', '-E', '--name-only',
-     grep_pattern % re.escape(original), '--'] + file_globs,
-    stdout=subprocess.PIPE).communicate()
+      ['git', 'grep', '-E', '--name-only', original, '--'] + file_globs,
+      stdout=subprocess.PIPE).communicate()
   referees = out.splitlines()
 
   for referee in referees:
     with open(referee) as f:
       original_contents = f.read()
-      contents = original_contents
-      for guard_format in guard_formats or []:
-        contents = contents.replace(guard_format % original,
-                                    guard_format % replacement)
+      contents = re.sub(original, replacement, original_contents)
       if contents == original_contents:
         raise Exception('No change in file %s although matched in grep' %
                         referee)
       with open(referee, 'w') as f:
         f.write(contents)
 
+  return referees
+
 
 def UpdatePostMove(from_path, to_path):
   """Given a file that has moved from |from_path| to |to_path|,
   updates the moved file's include guard to match the new path and
   updates all references to the file in other source files. Also tries
   to update references in .gyp(i) files using a heuristic.
   """
   # Include paths always use forward slashes.
   from_path = from_path.replace('\\', '/')
   to_path = to_path.replace('\\', '/')
 
   if os.path.splitext(from_path)[1] in ['.h', '.hh']:
     UpdateIncludeGuard(from_path, to_path)
 
     # Update include/import references.
-    MultiFileFindReplace(
-      from_path,
-      to_path,
-      r'#(include|import)\s*["<]%s[>"]',
-      ['*.cc', '*.h', '*.m', '*.mm'],
-      ['"%s"', '<%s>'])
+    files_with_changed_includes = MultiFileFindReplace(
+        r'(#(include|import)\s*["<])%s([>"])' % re.escape(from_path),
+        r'\1%s\3' % to_path,
+        ['*.cc', '*.h', '*.m', '*.mm'])
+
+    # Reorder headers in files that changed.
+    for changed_file in files_with_changed_includes:

[Nico, 2012/11/21 18:35:36]

I'd do `def auto_confirm(a, b): return true; FixFileWithConfrim(changed_file,
always_confirm)`, that's a bit more readable than the lambda imo.

[Jói, 2012/11/22 09:42:50]

Done.

+      sort_headers.FixFileWithConfirmFunction(changed_file, lambda a, b: True)
+
+  # Update comments; only supports // comments, which are primarily
+  # used in our code.
+  MultiFileFindReplace(
+      r'(//.*)%s' % re.escape(from_path),
+      r'\1%s' % to_path,
+      ['*.cc', '*.h', '*.m', '*.mm'])

[Nico, 2012/11/21 18:35:36]

I'd omit this as I said. It only finds comments that contain the full file path,
and comments with filenames aren't super common. Imho this isn't worth 20%.

(But searching crbug.com for "//.*[/].*\w\.cc -file:v8 -file:third_party" finds
more than I thought it would, so up to you.)

[Jói, 2012/11/22 09:42:50]

I'll leave it in since it's safe and convenient and updates something you might
not remember to search for and update yourself.  I've added a comment
documenting that comment handling is somewhat slow and the script could be sped
up by making it optional under a flag; time will tell if it is fast enough or
not.

 
   # Update references in .gyp(i) files.
   def PathMinusFirstComponent(path):
     """foo/bar/baz -> bar/baz"""
     parts = re.split(r"[/\\]", path, 1)
     if len(parts) == 2:
       return parts[1]
     else:
       return parts[0]
-  MultiFileFindReplace(PathMinusFirstComponent(from_path),
-                       PathMinusFirstComponent(to_path),
-                       r'[\'"]%s[\'"]',
-                       ['*.gyp*'],
-                       ["'%s'", '"%s"'])
+  MultiFileFindReplace(
+      r'([\'"])%s([\'"])' % re.escape(PathMinusFirstComponent(from_path)),
+      r'\1%s\2' % PathMinusFirstComponent(to_path),
+      ['*.gyp*'])
 
 
 def MakeIncludeGuardName(path_from_root):
   """Returns an include guard name given a path from root."""
   guard = path_from_root.replace('/', '_')
   guard = guard.replace('\\', '_')
   guard = guard.replace('.', '_')
   guard += '_'
   return guard.upper()
 
@@ skipping 40 matching lines @@
 
   to_path = MakeDestinationPath(from_path, to_path)
   if not already_moved:
     MoveFile(from_path, to_path)
   UpdatePostMove(from_path, to_path)
   return 0
 
 
 if __name__ == '__main__':
   sys.exit(main())