python3: add polyglot wrapper for html.entities/htmlentitydefs

src/calibre/__init__.py

@@ -614,7 +614,7 @@ def entity_to_unicode(match, exceptions=[], encoding='cp1252',
         return check(html5_entities[ent])
     except KeyError:
         pass
-    from htmlentitydefs import name2codepoint
+    from polyglot.html_entities import name2codepoint
     try:
         return check(my_unichr(name2codepoint[ent]))
     except KeyError:

src/calibre/utils/cleantext.py

@@ -2,8 +2,9 @@ __license__ = 'GPL 3'
 __copyright__ = '2010, sengian <sengian1@gmail.com>'
 __docformat__ = 'restructuredtext en'
 
-import re, htmlentitydefs
+import re
 from polyglot.builtins import codepoint_to_chr, map, range
+from polyglot.html_entities import name2codepoint
 from calibre.constants import plugins, preferred_encoding
 
 try:
@@ -80,7 +81,7 @@ def unescape(text, rm=False, rchar=u''):
         else:
             # named entity
             try:
-                text = codepoint_to_chr(htmlentitydefs.name2codepoint[text[1:-1]])
+                text = codepoint_to_chr(name2codepoint[text[1:-1]])
             except KeyError:
                 pass
         if rm:

src/polyglot/html_entities.py

@@ -0,0 +1,10 @@
+#!/usr/bin/env python2
+# vim:fileencoding=utf-8
+# License: GPL v3 Copyright: 2019, Eli Schwartz <eschwartz@archlinux.org>
+
+from polyglot.builtins import is_py3
+
+if is_py3:
+    from html.entities import name2codepoint
+else:
+    from htmlentitydefs import name2codepoint