#426: Fix encoding issues when saving/loading intersphinx inventories.

sphinx/builders/html.py

@@ -738,7 +738,7 @@ class StandaloneHTMLBuilder(Builder):
         try:
             f.write('# Sphinx inventory version 2\n')
             f.write('# Project: %s\n' % self.config.project.encode('utf-8'))
-            f.write('# Version: %s\n' % self.config.version)
+            f.write('# Version: %s\n' % self.config.version.encode('utf-8'))
             f.write('# The remainder of this file is compressed using zlib.\n')
             compressor = zlib.compressobj(9)
             for domainname, domain in self.env.domains.iteritems():
@@ -749,10 +749,13 @@ class StandaloneHTMLBuilder(Builder):
                         anchor = anchor[:-len(name)] + '$'
                     uri = self.get_target_uri(docname) + '#' + anchor
                     if dispname == name:
-                        dispname = '-'
+                        dispname = u'-'
                     f.write(compressor.compress(
-                        '%s %s:%s %s %s %s\n' % (name, domainname, type, prio,
-                                                 uri, dispname)))
+                        '%s %s:%s %s %s %s\n' % (name.encode('utf-8'),
+                                                 domainname.encode('utf-8'),
+                                                 type.encode('utf-8'), prio,
+                                                 uri.encode('utf-8'),
+                                                 dispname.encode('utf-8'))))
             f.write(compressor.flush())
         finally:
             f.close()

sphinx/ext/intersphinx.py

@@ -67,7 +67,7 @@ def read_inventory_v2(f, uri, join, bufsize=16*1024):
     line = f.readline()
     projname = line.rstrip()[11:].decode('utf-8')
     line = f.readline()
-    version = line.rstrip()[11:]
+    version = line.rstrip()[11:].decode('utf-8')
     line = f.readline()
     if 'zlib' not in line:
         raise ValueError
@@ -84,14 +84,14 @@ def read_inventory_v2(f, uri, join, bufsize=16*1024):
             buf += chunk
             lineend = buf.find('\n')
             while lineend != -1:
-                yield buf[:lineend]
+                yield buf[:lineend].decode('utf-8')
                 buf = buf[lineend+1:]
                 lineend = buf.find('\n')
         assert not buf
 
     for line in split_lines(read_chunks()):
         name, type, prio, location, dispname = line.rstrip().split(None, 4)
-        if location.endswith('$'):
+        if location.endswith(u'$'):
             location = location[:-1] + name
         location = join(uri, location)
         invdata.setdefault(type, {})[name] = (projname, version,

tests/root/markup.txt

@@ -246,6 +246,8 @@ Invalid index markup...
    keyword:
 
 
+.. _ölabel:
+
 Ö... Some strange characters
 ----------------------------