Strip comments and whitespace from javascript resources

tools/grit/grit/format/html_inline.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.
 
 """Flattens a HTML file by inlining its external resources.
 
 This is a small script that takes a HTML file, looks for src attributes
 and inlines the specified file, producing one HTML file with no external
 dependencies. It recursively inlines the included files.
 """
 
 import os
 import re
 import sys
 import base64
 import mimetypes
 
 from grit import lazy_re
 from grit import util
+from grit.format import minifier
 
 # There is a python bug that makes mimetypes crash if the Windows
 # registry contains non-Latin keys ( http://bugs.python.org/issue9291
 # ). Initing manually and blocking external mime-type databases will
 # prevent that bug and if we add svg manually, it will still give us
 # the data we need.
 mimetypes.init([])
 mimetypes.add_type('image/svg+xml', '.svg')
 
 DIST_DEFAULT = 'chromium'
@@ skipping 91 matching lines @@
 
   Holds the inlined data and the set of filenames of all the inlined
   files.
   """
   def __init__(self, inlined_data, inlined_files):
     self.inlined_data = inlined_data
     self.inlined_files = inlined_files
 
 def DoInline(
     input_filename, grd_node, allow_external_script=False,
-    preprocess_only=False, names_only=False, rewrite_function=None,
-    filename_expansion_function=None):
+    preprocess_only=False, names_only=False, strip_whitespace=False,
+    rewrite_function=None, filename_expansion_function=None):
   """Helper function that inlines the resources in a specified file.
 
   Reads input_filename, finds all the src attributes and attempts to
   inline the files they are referring to, then returns the result and
   the set of inlined files.
 
   Args:
     input_filename: name of file to read in
     grd_node: html node from the grd file for this include tag
     preprocess_only: Skip all HTML processing, only handle <if> and <include>.
     names_only: |nil| will be returned for the inlined contents (faster).
+    strip_whitespace: remove whitespace and comments in the input files.
     rewrite_function: function(filepath, text, distribution) which will be
         called to rewrite html content before inlining images.
     filename_expansion_function: function(filename) which will be called to
         rewrite filenames before attempting to read them.
   Returns:
     a tuple of the inlined data as a string and the set of filenames
     of all the inlined files
   """
   if filename_expansion_function:
     input_filename = filename_expansion_function(input_filename)
@@ skipping 58 matching lines @@
           pos = next_if.end()
 
       content = str[content_start:end_if.start()]
       trailing = str[end_if.end():]
 
       if condition_satisfied:
         str = leading + CheckConditionalElements(content) + trailing
       else:
         str = leading + trailing
 
-  def InlineFileContents(src_match, pattern, inlined_files=inlined_files):
+  def InlineFileContents(src_match,
+                         pattern,
+                         inlined_files=inlined_files,
+                         strip_whitespace=False):
     """Helper function to inline external files of various types"""
     filepath = GetFilepath(src_match)
     if filepath is None:
       return src_match.group(0)
     inlined_files.add(filepath)
 
     if names_only:
       inlined_files.update(GetResourceFilenames(
           filepath,
           allow_external_script,
           rewrite_function,
           filename_expansion_function=filename_expansion_function))
       return ""
 
     return pattern % InlineToString(
         filepath, grd_node, allow_external_script=allow_external_script,
+        strip_whitespace=strip_whitespace,
         filename_expansion_function=filename_expansion_function)
 
   def InlineIncludeFiles(src_match):
     """Helper function to directly inline generic external files (without
        wrapping them with any kind of tags).
     """
     return InlineFileContents(src_match, '%s')
 
   def InlineScript(match):
     """Helper function to inline external script files"""
     attrs = (match.group('attrs1') + match.group('attrs2')).strip()
     if attrs:
-       attrs = ' ' + attrs
-    return InlineFileContents(match, '<script' + attrs + '>%s</script>')
+      attrs = ' ' + attrs
+    return InlineFileContents(match, '<script' + attrs + '>%s</script>',
+                              strip_whitespace=True)
 
   def InlineCSSText(text, css_filepath):
     """Helper function that inlines external resources in CSS text"""
     filepath = os.path.dirname(css_filepath)
     # Allow custom modifications before inlining images.
     if rewrite_function:
       text = rewrite_function(filepath, text, distribution)
     text = InlineCSSImages(text, filepath)
     return InlineCSSImports(text, filepath)
 
@@ skipping 50 matching lines @@
 
 
   flat_text = util.ReadFile(input_filename, util.BINARY)
 
   # Check conditional elements, remove unsatisfied ones from the file. We do
   # this twice. The first pass is so that we don't even bother calling
   # InlineScript, InlineCSSFile and InlineIncludeFiles on text we're eventually
   # going to throw out anyway.
   flat_text = CheckConditionalElements(flat_text)
 
+  flat_text = _INCLUDE_RE.sub(InlineIncludeFiles, flat_text)
+
   if not preprocess_only:
+    if strip_whitespace:
+      flat_text = minifier.Minify(flat_text,
+                                  os.path.splitext(input_filename)[1])
+
     if not allow_external_script:
       # We need to inline css and js before we inline images so that image
       # references gets inlined in the css and js
       flat_text = re.sub('<script (?P<attrs1>.*?)src="(?P<filename>[^"\']*)"' +
                          '(?P<attrs2>.*?)></script>',
                          InlineScript,
                          flat_text)
 
     flat_text = _STYLESHEET_RE.sub(
         lambda m: InlineCSSFile(m, '<style>%s</style>'),
         flat_text)
 
-  flat_text = _INCLUDE_RE.sub(InlineIncludeFiles, flat_text)
-
   # Check conditional elements, second pass. This catches conditionals in any
   # of the text we just inlined.
   flat_text = CheckConditionalElements(flat_text)
 
   if not preprocess_only:
     # Allow custom modifications before inlining images.
     if rewrite_function:
       flat_text = rewrite_function(input_filepath, flat_text, distribution)
     flat_text = _SRC_RE.sub(SrcReplace, flat_text)
 
     # TODO(arv): Only do this inside <style> tags.
     flat_text = InlineCSSImages(flat_text)
 
     flat_text = _ICON_RE.sub(SrcReplace, flat_text)
 
   if names_only:
     flat_text = None  # Will contains garbage if the flag is set anyway.
   return InlinedData(flat_text, inlined_files)
 
 
 def InlineToString(input_filename, grd_node, preprocess_only = False,
-                   allow_external_script=False, rewrite_function=None,
-                   filename_expansion_function=None):
+                   allow_external_script=False, strip_whitespace=False,
+                   rewrite_function=None, filename_expansion_function=None):
   """Inlines the resources in a specified file and returns it as a string.
 
   Args:
     input_filename: name of file to read in
     grd_node: html node from the grd file for this include tag
   Returns:
     the inlined data as a string
   """
   try:
     return DoInline(
         input_filename,
         grd_node,
         preprocess_only=preprocess_only,
         allow_external_script=allow_external_script,
+        strip_whitespace=strip_whitespace,
         rewrite_function=rewrite_function,
         filename_expansion_function=filename_expansion_function).inlined_data
   except IOError, e:
     raise Exception("Failed to open %s while trying to flatten %s. (%s)" %
                     (e.filename, input_filename, e.strerror))
 
 
 def InlineToFile(input_filename, output_filename, grd_node):
   """Inlines the resources in a specified file and writes it.
 
@@ skipping 18 matching lines @@
                          rewrite_function=None,
                          filename_expansion_function=None):
   """For a grd file, returns a set of all the files that would be inline."""
   try:
     return DoInline(
         filename,
         None,
         names_only=True,
         preprocess_only=False,
         allow_external_script=allow_external_script,
+        strip_whitespace=False,
         rewrite_function=rewrite_function,
         filename_expansion_function=filename_expansion_function).inlined_files
   except IOError, e:
     raise Exception("Failed to open %s while trying to flatten %s. (%s)" %
                     (e.filename, filename, e.strerror))
 
 
 def main():
   if len(sys.argv) <= 2:
     print "Flattens a HTML file by inlining its external resources.\n"
     print "html_inline.py inputfile outputfile"
   else:
     InlineToFile(sys.argv[1], sys.argv[2], None)
 
 if __name__ == '__main__':
   main()