[grit] Avoid intermediate files for HTML inlining when building .pak files

This avoids problems where multiple invocations of grit try to write to
the same intermediate file.

BUG=69633
TEST=Manually

Review URL: http://codereview.chromium.org/6245002

git-svn-id: svn://svn.chromium.org/chrome/trunk/src@71491 0039d316-1c4b-4281-b951-d872f2087c98

6 files changed, 88 insertions, 33 deletions

diff --git a/tools/grit/grit/format/data_pack.py b/tools/grit/grit/format/data_pack.py
index 664cbaec..20a8f73 100755
--- a/tools/grit/grit/format/data_pack.py
+++ b/tools/grit/grit/format/data_pack.py
@@ -29,7 +29,7 @@ class DataPack(interface.ItemFormatter):
     nodes = DataPack.GetDataNodes(item)
     data = {}
     for node in nodes:
-      id, value = node.GetDataPackPair(output_dir, lang)
+      id, value = node.GetDataPackPair(lang)
       data[id] = value
     return DataPack.WriteDataPack(data)
 
diff --git a/tools/grit/grit/format/html_inline.py b/tools/grit/grit/format/html_inline.py
index 9979bca..c5cb074 100755
--- a/tools/grit/grit/format/html_inline.py
+++ b/tools/grit/grit/format/html_inline.py
@@ -69,19 +69,30 @@ def SrcInlineAsDataURL(src_match, base_path, distribution, inlined_files):
   prefix = src_match.string[src_match.start():src_match.start('filename')-1]
   return "%s\"data:%s;base64,%s\"" % (prefix, mimetype, inline_data)
 
-def InlineFile(input_filename, output_filename, grd_node):
-  """Inlines the resources in a specified file.
+
+class InlinedData:
+  """Helper class holding the results from DoInline().
+
+  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):
+  """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 writes the result
-  to output_filename.
+  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
-    output_filename: name of file to be written to
     grd_node: html node from the grd file for this include tag
   Returns:
-    a set of filenames of all the inlined files
+    a tuple of the inlined data as a string and the set of filenames
+    of all the inlined files
   """
   input_filepath = os.path.dirname(input_filename)
 
@@ -198,16 +209,45 @@ def InlineFile(input_filename, output_filename, grd_node):
   flat_text = re.sub('<link rel="icon".+?href="(?P<filename>[^"\']*)"',
                      SrcReplace,
                      flat_text)
-  if output_filename:
-    out_file = open(output_filename, 'wb')
-    out_file.writelines(flat_text)
-    out_file.close()
-  return inlined_files
+
+  return InlinedData(flat_text, inlined_files)
+
+
+def InlineToString(input_filename, grd_node):
+  """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
+  """
+  return DoInline(input_filename, grd_node).inlined_data
+
+
+def InlineToFile(input_filename, output_filename, grd_node):
+  """Inlines the resources in a specified file and writes it.
+
+  Reads input_filename, finds all the src attributes and attempts to
+  inline the files they are referring to, then writes the result
+  to output_filename.
+
+  Args:
+    input_filename: name of file to read in
+    output_filename: name of file to be written to
+    grd_node: html node from the grd file for this include tag
+  Returns:
+    a set of filenames of all the inlined files
+  """
+  inlined_data = InlineToString(input_filename, grd_node)
+  out_file = open(output_filename, 'wb')
+  out_file.writelines(inlined_data)
+  out_file.close()
 
 
 def GetResourceFilenames(filename):
   """For a grd file, returns a set of all the files that would be inline."""
-  return InlineFile(filename, None, None)
+  return DoInline(filename, None).inlined_files
 
 
 def main():
@@ -215,7 +255,7 @@ def main():
     print "Flattens a HTML file by inlining its external resources.\n"
     print "html_inline.py inputfile outputfile"
   else:
-    InlineFile(sys.argv[1], sys.argv[2], None)
+    InlineToFile(sys.argv[1], sys.argv[2], None)
 
 if __name__ == '__main__':
   main()
diff --git a/tools/grit/grit/format/rc.py b/tools/grit/grit/format/rc.py
index 13834b7..3e999c4 100644
--- a/tools/grit/grit/format/rc.py
+++ b/tools/grit/grit/format/rc.py
@@ -424,6 +424,8 @@ class RcInclude(interface.ItemFormatter):
     # if needed (e.g. if it is an HTML file include).
     filename = os.path.abspath(item.FileForLanguage(lang, output_dir))
     if self.flatten_html:
+      # TODO(akalin): Flatten to a subdirectory of output_dir, lest we
+      # run into bugs like http://crbug.com/69633.
       item.Flatten(output_dir)
       # The flattened file is in the output dir.
       filename = os.path.basename(filename)
diff --git a/tools/grit/grit/node/include.py b/tools/grit/grit/node/include.py
index 4f96cdf..cb64866 100644
--- a/tools/grit/grit/node/include.py
+++ b/tools/grit/grit/node/include.py
@@ -20,13 +20,23 @@ class IncludeNode(base.Node):
   def __init__(self):
     base.Node.__init__(self)
 
-    # Keep track of whether we've flattened the file or not.  We don't
-    # want to flatten the same file multiple times.
-    self._is_flattened = False
+    # Cache flattened data so that we don't flatten the same file
+    # multiple times.
+    self._flattened_data = None
+    # Also keep track of the last filename we flattened to, so we can
+    # avoid doing it more than once.
+    self._last_flat_filename = None
 
   def _IsValidChild(self, child):
     return False
 
+  def _GetFlattenedData(self):
+    if not self._flattened_data:
+      filename = self.FilenameToOpen()
+      self._flattened_data = (
+          grit.format.html_inline.InlineToString(filename, self))
+    return self._flattened_data
+
   def MandatoryAttributes(self):
     return ['name', 'type', 'file']
 
@@ -62,33 +72,36 @@ class IncludeNode(base.Node):
     '''
     return self.FilenameToOpen()
 
-  def GetDataPackPair(self, output_dir, lang):
+  def GetDataPackPair(self, lang):
     '''Returns a (id, string) pair that represents the resource id and raw
     bytes of the data.  This is used to generate the data pack data file.
     '''
     from grit.format import rc_header
     id_map = rc_header.Item.tids_
     id = id_map[self.GetTextualIds()[0]]
-    filename = self.FilenameToOpen()
     if self.attrs['flattenhtml'] == 'true':
-      self.Flatten(output_dir)
-      # The flattened file is in the output dir.
-      filename = os.path.join(output_dir, os.path.split(filename)[1])
-
-    file = open(filename, 'rb')
-    data = file.read()
-    file.close()
+      data = self._GetFlattenedData()
+    else:
+      filename = self.FilenameToOpen()
+      infile = open(filename, 'rb')
+      data = infile.read()
+      infile.close()
 
     return id, data
 
   def Flatten(self, output_dir):
-    if self._is_flattened:
-      return
-
     filename = self.FilenameToOpen()
     flat_filename = os.path.join(output_dir, os.path.split(filename)[1])
-    grit.format.html_inline.InlineFile(filename, flat_filename, self)
-    self._is_flattened = True
+
+    if self._last_flat_filename == flat_filename:
+      return
+
+    outfile = open(flat_filename, 'wb')
+    outfile.write(self._GetFlattenedData())
+    outfile.close()
+
+    self._last_flat_filename = flat_filename
+
 
   def GetHtmlResourceFilenames(self):
     """Returns a set of all filenames inlined by this file."""
diff --git a/tools/grit/grit/node/message.py b/tools/grit/grit/node/message.py
index 2dfd0ec..a48a645 100644
--- a/tools/grit/grit/node/message.py
+++ b/tools/grit/grit/node/message.py
@@ -186,7 +186,7 @@ class MessageNode(base.ContentNode):
     else:
       return self.attrs['offset']
 
-  def GetDataPackPair(self, output_dir, lang):
+  def GetDataPackPair(self, lang):
     '''Returns a (id, string) pair that represents the string id and the string
     in utf8.  This is used to generate the data pack data file.
     '''
diff --git a/tools/grit/grit_info.py b/tools/grit/grit_info.py
index 042f457..cf80954 100755
--- a/tools/grit/grit_info.py
+++ b/tools/grit/grit_info.py
@@ -109,7 +109,7 @@ def main(argv):
       outputs = [posixpath.join(args[0], f) for f in Outputs(f, defines)]
       print '\n'.join(outputs)
   else:
-    PringUsage()
+    PrintUsage()
     return 1
   return 0