Update characters that need be quoted in attributes in the serializer per spec

This also moves to using re, which seems far cleaner than the reduce-based
search previously used.

Author: gsnedders

html5lib/serializer/htmlserializer.py

@@ -1,10 +1,7 @@
 from __future__ import absolute_import, division, unicode_literals
 from six import text_type
 
-try:
-    from functools import reduce
-except ImportError:
-    pass
+import re
 
 from ..constants import voidElements, booleanAttributes, spaceCharacters
 from ..constants import rcdataElements, entities, xmlEntities
@@ -13,6 +10,8 @@
 
 spaceCharacters = "".join(spaceCharacters)
 
+quoteAttributeSpec = re.compile("[" + spaceCharacters + "\"'=<>`]")
+
 try:
     from codecs import register_error, xmlcharrefreplace_errors
 except ImportError:
@@ -240,11 +239,10 @@ def serialize(self, treewalker, encoding=None):
                         (k not in booleanAttributes.get(name, tuple()) and
                          k not in booleanAttributes.get("", tuple())):
                         yield self.encodeStrict("=")
-                        if self.quote_attr_values or not v:
+                        if self.quote_attr_values:
                             quote_attr = True
                         else:
-                            quote_attr = reduce(lambda x, y: x or (y in v),
-                                                spaceCharacters + ">\"'=", False)
+                            quote_attr = len(v) == 0 or quoteAttributeSpec.search(v)
                         v = v.replace("&", "&amp;")
                         if self.escape_lt_in_attrs:
                             v = v.replace("<", "&lt;")

html5lib/tests/serializer-testdata/core.test

@@ -42,7 +42,7 @@
         },
         {
             "expected": [
-                "<span title=foo<bar>"
+                "<span title=\"foo<bar\">"
             ],
             "input": [
                 [