Fix another source of utf8 characters being lost.

appengine/chromium_build/app.py

 # 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.
 
 from __future__ import with_statement
 
 import datetime
 import json
 import logging
 import os
@@ skipping 280 matching lines @@
     # Map(k,v): k=Master, v=Dict of category data
     self.category_data = {}
 
     self.category_count = 0
     self.master = ''
     self.lastRevisionSeen = None
     self.lastMasterSeen = None
 
   @staticmethod
   def ContentsToHtml(contents):
-    return ''.join(unicode(content).encode('ascii', 'replace')
-                   for content in contents)
+    return ''.join(str(content).decode('utf-8') for content in contents)
 
   @property
   def last_row(self):
     return self.row_data[self.lastRevisionSeen]
 
   def SawMaster(self, master):
     self.lastMasterSeen = master
     assert(self.lastMasterSeen not in self.category_order)
     self.masters.append(self.lastMasterSeen)
     self.category_order.setdefault(self.lastMasterSeen, [])
@@ skipping 162 matching lines @@
   template_environment.filters['notstarted'] = notstarted
   merged_template = template_environment.from_string(console_template)
   merged_content = merged_template.render(data=mergedconsole)
   # For debugging:
   # print merged_content
   # import code
   # code.interact(local=locals())
 
   # Place merged data at |merged_tag|'s location in |merged_page|, and put the
   # result in |merged_content|.
-  merged_tag.replaceWith(str(merged_content))
+  merged_tag.replaceWith(merged_content)
   # .prettify() may damage the HTML but makes output nicer.  However, that
   # cost is a bunch of extra whitespace.  We reduce page size by not using
   # .prettify().
   merged_content = merged_page.__str__(encoding=None)
   merged_content = re.sub(
       r'\'\<a href="\'', '\'<a \' + attributes + \' href="\'', merged_content)
   merged_content = re.sub(
       r'\'\<table\>\'', r"'<table ' + attributes + '>'", merged_content)
   merged_content = re.sub(
       r'\'\<div\>\'', r"'<div ' + attributes + '>'", merged_content)
@@ skipping 167 matching lines @@
   # exist more than once.  Reverts are examples of commits which can contain
   # multiple Cr-Commit-Position instances.  In those cases, only the last one
   # is correct, so split on the break tag and reverse the result to find the
   # last occurrence of Cr-Commit-Position.
   for line in reversed(commit_msg.split('<br />')):
     if line.startswith('Cr-Commit-Position: '):
       return filter(str.isdigit, str(line.split('@')[-1]))
   return '0'
 
 
-def utf8_convert(bstring):
+def utf8_convert(bstag):

[jrobbins, 2015/02/12 17:55:31]

What does 'bstag' stand for?

[cmp, 2015/02/12 17:59:58]

BeautifulSoup tag.

   # cmp also investigated:
-  #   bstring.__str__(encoding='utf-8').decode('utf-8')
+  #   bstag.__str__(encoding='utf-8').decode('utf-8')
   # He found that the BeautifulSoup() __str__ method when used with a 'utf-8'
   # encoding returned effectively the same thing as str(), a Python built-in.
   # After a handful of tests, he switched to using str() to avoid the add'l
   # complexity of another BeautifulSoup method.
-  return str(bstring).decode('utf-8')
+  return str(bstag).decode('utf-8')
 
 
 # W0613:600,28:parse_master: Unused argument 'remoteurl'
 # pylint: disable=W0613
 def parse_master(localpath, remoteurl, page_data=None):
   """Part of the new pipeline to store individual rows rather than
   whole pages of html. Parses the master data into a set of rows,
   and writes them out to the datastore in an easily retrievable format.
 
   Doesn't modify page_data dict.
@@ skipping 560 matching lines @@
          'builds/-1?as_text=1'),
     'localpath':
         'chromium.lkgr/json/builders/Linux%20x64/builds/-1/as_text=1.json',
     'maxage': 2*60,  # 2 mins
   },
 
 # # Trigger background process update.
 # {
 #     'remoteurl': 'http://chromium-build.appspot.com/backend/update'
 ]