Fixed deprecation warnings for calling HTMLParser on Python 3.X

The logic here comes from http://bazaar.launchpad.net/~leonardr/beautifulsoup/bs4/view/head:/bs4/builder/_htmlparser.py#L20,
inspecting the stdlib code in different versions, and simplifying.

I haven't included the monkey patch from beautifulsoup for Python 3.2.2
because I don't think the bug affects django-compressor's parsing of script
tags.

compressor/parser/default_htmlparser.py

@@ -1,3 +1,5 @@
+import sys
+
 from django.utils import six
 from django.utils.encoding import smart_text
 
@@ -5,9 +7,26 @@ from compressor.exceptions import ParserError
 from compressor.parser import ParserBase
 
 
+# Starting in Python 3.2, the HTMLParser constructor takes a 'strict'
+# argument which default to True (which we don't want).
+# In Python 3.3, it defaults to False.
+# In Python 3.4, passing it at all raises a deprecation warning.
+# So we only pass it for 3.2.
+# In Python 3.4, it also takes a 'convert_charrefs' argument
+# which raises a warning if we don't pass it.
+major, minor, release = sys.version_info[:3]
+CONSTRUCTOR_TAKES_STRICT = major == 3 and minor == 2
+CONSTRUCTOR_TAKES_CONVERT_CHARREFS = major == 3 and minor >= 4
+HTML_PARSER_ARGS = {}
+if CONSTRUCTOR_TAKES_STRICT:
+    HTML_PARSER_ARGS['strict'] = False
+if CONSTRUCTOR_TAKES_CONVERT_CHARREFS:
+    HTML_PARSER_ARGS['convert_charrefs'] = False
+
+
 class DefaultHtmlParser(ParserBase, six.moves.html_parser.HTMLParser):
     def __init__(self, content):
-        six.moves.html_parser.HTMLParser.__init__(self)
+        six.moves.html_parser.HTMLParser.__init__(self, **HTML_PARSER_ARGS)
         self.content = content
         self._css_elems = []
         self._js_elems = []