add support for html5lib below 0.95, #168

isso/tests/test_html.py

@@ -62,7 +62,7 @@ class TestHTML(unittest.TestCase):
             print("Hello, World")
             </code></pre>""")
 
-    @unittest.skipIf(html.html5lib_version == "0.95", "backport")
+    @unittest.skipIf(html.HTML5LIB_VERSION <= html.HTML5LIB_SIMPLETREE, "backport")
     def test_sanitizer(self):
         sanitizer = html.Sanitizer(elements=[], attributes=[])
         examples = [
@@ -75,7 +75,7 @@ class TestHTML(unittest.TestCase):
         for (input, expected) in examples:
             self.assertEqual(html.sanitize(sanitizer, input), expected)
 
-    @unittest.skipIf(html.html5lib_version == "0.95", "backport")
+    @unittest.skipIf(html.HTML5LIB_VERSION <= html.HTML5LIB_SIMPLETREE, "backport")
     def test_sanitizer_extensions(self):
         sanitizer = html.Sanitizer(elements=["img"], attributes=["src"])
         examples = [

isso/utils/html.py

@@ -2,14 +2,17 @@
 
 from __future__ import unicode_literals
 
-import pkg_resources
 import operator
+import pkg_resources
+
+from distutils.version import LooseVersion as Version
+
+HTML5LIB_VERSION = Version(pkg_resources.get_distribution("html5lib").version)
+HTML5LIB_SIMPLETREE = Version("0.9.5")
 
 from isso.compat import reduce
 
 import html5lib
-html5lib_version = pkg_resources.get_distribution("html5lib").version
-
 from html5lib.sanitizer import HTMLSanitizer
 from html5lib.serializer import HTMLSerializer
 
@@ -45,7 +48,11 @@ def sanitize(tokenizer, document):
     parser = html5lib.HTMLParser(tokenizer=tokenizer)
     domtree = parser.parseFragment(document)
 
-    builder = "simpletree" if html5lib_version == "0.95" else "etree"
+    if HTML5LIB_VERSION > HTML5LIB_SIMPLETREE:
+        builder = "etree"
+    else:
+        builder = "simpletree"
+
     stream = html5lib.treewalkers.getTreeWalker(builder)(domtree)
     serializer = HTMLSerializer(quote_attr_values=True, omit_optional_tags=False)