decode all HTML entities

3 files changed, 24 insertions, 6 deletions

diff --git a/tests/test_click.py b/tests/test_click.py
index 5c27c53..b02179a 100644
--- a/tests/test_click.py
+++ b/tests/test_click.py
@@ -130,3 +130,8 @@ class TestClick(unittest.TestCase):
         self.assertEqual(_parse_attrs("href='foo' id=\"bar\" "), {'href': 'foo', 'id': 'bar'})
         self.assertEqual(_parse_attrs("href='foo' id='bar' "), {'href': 'foo', 'id': 'bar'})
         self.assertEqual(_parse_attrs("tag='foo\"'"), {'tag': 'foo"'})
+        self.assertEqual(
+            _parse_attrs('value="<>&"{"'),
+                {'value': u('<>&"{')})
+        self.assertEqual(_parse_attrs('value="&sum;"'), {'value': u('∑')})
+        self.assertEqual(_parse_attrs('value="&#x20ac;"'), {'value': u('€')})
diff --git a/webtest/app.py b/webtest/app.py
index 406a846..32a0ef9 100644
--- a/webtest/app.py
+++ b/webtest/app.py
@@ -21,6 +21,7 @@ from webtest.compat import StringIO
 from webtest.compat import BytesIO
 from webtest.compat import SimpleCookie, CookieError
 from webtest.compat import cookie_quote
+from webtest.compat import name2codepoint
 from webtest.compat import urlencode
 from webtest.compat import splittype
 from webtest.compat import splithost
@@ -1819,15 +1820,25 @@ def _make_pattern(pat):
         "Cannot make callable pattern object out of %r" % pat)
 
 
+entity_pattern = re.compile(r"&(\w+|#\d+|#[xX][a-fA-F0-9]+);")
 def html_unquote(v):
     """
-    Unquote (some) entities in HTML.  (incomplete)
+    Unquote entities in HTML.
     """
-    for ent, repl in [('&nbsp;', ' '), ('&gt;', '>'),
-                      ('&lt;', '<'), ('&quot;', '"'),
-                      ('&amp;', '&')]:
-        v = v.replace(ent, repl)
-    return v
+    to_chr = chr if PY3 else unichr
+    def repl(match):
+        s = match.group(1)
+        if s.startswith("#"):
+            if s[1].lower() == "x":
+                s = int(s[2:], 16)
+            else:
+                s = int(s[1:])
+        elif s in name2codepoint:
+            s = name2codepoint[s]
+        else:
+            return
+        return to_chr(s)
+    return entity_pattern.sub(repl, v)
 
 
 def encode_params(params, content_type):
diff --git a/webtest/compat.py b/webtest/compat.py
index 1c30a4f..b4d39ac 100644
--- a/webtest/compat.py
+++ b/webtest/compat.py
@@ -8,6 +8,7 @@ if sys.version_info[0] > 2:
     binary_type = bytes
     from json import loads
     from json import dumps
+    from html.entities import name2codepoint
     from io import StringIO
     from io import BytesIO
     from urllib.parse import urlencode
@@ -40,6 +41,7 @@ else:
     string_types = basestring
     text_type = unicode
     binary_type = str
+    from htmlentitydefs import name2codepoint
     from urllib import splittype
     from urllib import splithost
     from urllib import urlencode