Merge pull request #20 from Earth-Online/master

fix #19 missing start tag error

Author: python273

telegraph/utils.py

@@ -83,6 +83,11 @@ def handle_endtag(self, tag):
         if tag in VOID_ELEMENTS:
             return
 
+        if not len(self.parent_nodes):
+            raise InvalidHTML('"{}" missing start tag'.format(
+                tag
+            ))
+
         self.current_nodes = self.parent_nodes.pop()
 
         last_node = self.current_nodes[-1]

tests/test_html_converter.py

@@ -19,7 +19,7 @@
         'attrs': {'href': 'https://telegra.ph/'},
         'children': ['Test link</a>']
         }]
-    },
+     },
     {'tag': 'figure', 'children': [
         {'tag': 'img', 'attrs': {'src': '/file/6c2ecfdfd6881d37913fa.png'}},
         {'tag': 'figcaption'}
@@ -45,6 +45,8 @@
     ]},
 ]
 
+HTML_NO_STARTTAG = "</a><h1></h1>"
+
 
 class TestHTMLConverter(TestCase):
     def test_html_to_nodes(self):
@@ -130,3 +132,7 @@ def test_clear_whitespace_nodes(self):
         ]
 
         self.assertEqual(clear_whitespace_nodes(nodes)[0], expected)
+
+    def test_no_starttag_node(self):
+        with self.assertRaises(InvalidHTML):
+             html_to_nodes(HTML_NO_STARTTAG)