Better unicode output processing.

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([str(content) for content in contents])
+    return ''.join(unicode(content).encode('ascii', 'replace')
+                   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 50 matching lines @@
 
   def ParseRow(self, row):
     cells = row.findAll('td', recursive=False)
     # Figure out which row this is.
     for attrname, attrvalue in cells[0].attrs:
       if attrname != 'class':
         continue
       attrvalue = re.sub(r'^(\S+).*', r'\1', attrvalue)
       if attrvalue == 'DevRev':
         revision = cells[0]
-        self.SawRevision(revision=revision.findAll('a')[0].contents[0])
-        self.SetLink(revlink=revision.findAll('a')[0].attrs[0][1])
+        self.SawRevision(self.ContentsToHtml(
+            revision.findAll('a')[0].contents[0]))
+        self.SetLink(self.ContentsToHtml(revision.findAll('a')[0].attrs[0][1]))
         nameparts = cells[1].contents
-        self.SetName(who=re.sub(r'^\s+(.*)\s*$',
-                                r'\1',
-                                self.ContentsToHtml(nameparts)))
+        self.SetName(re.sub(r'^\s+(.*)\s*$',
+                            r'\1',
+                            self.ContentsToHtml(nameparts)))
         for i, bs in enumerate(cells[2:]):
-          self.SetStatus(category=self.category_order[self.lastMasterSeen][i],
-                         status=str(bs.findAll('table', recursive=False)[0]))
+          self.SetStatus(self.category_order[self.lastMasterSeen][i],
+                         self.ContentsToHtml(bs.findAll('table',
+                                                        recursive=False)[0]))
       if attrvalue == 'DevComment':
         self.SetComment(comment=self.ContentsToHtml(cells[0].contents))
       if attrvalue == 'DevDetails':
         self.SetDetail(detail=self.ContentsToHtml(cells[0].contents))
 
   def Finish(self):
     self.row_orderedkeys = sorted(self.row_orderedkeys, key=int, reverse=True)
     # TODO(cmp): Look for row/master/categories that are unset.  If they are
     #            at the latest revisions, leave them unset.  If they are at
     #            the earliest revisions, set them to ''.
@@ skipping 803 matching lines @@
   # LKGR JSON.
   {
     'remoteurl':
         ('http://build.chromium.org/p/chromium.lkgr/json/builders/Linux%20x64/'
          'builds/-1?as_text=1'),
     'localpath':
         'chromium.lkgr/json/builders/Linux%20x64/builds/-1/as_text=1.json',
     'maxage': 2*60,  # 2 mins
   },
 ]