Log a warning when the HTML reader encounters a nonconformant meta tag

Instead of throwing an exception and skipping the HTML file, log a warning with a message which makes it more obvious as to what happened.
getpelican · Nov 8, 2013 · 35375b1 · 35375b1
1 parent 455c159
commit 35375b1
Show file tree

Hide file tree

Showing 4 changed files with 29 additions and 1 deletion.
diff --git a/pelican/readers.py b/pelican/readers.py
@@ -302,7 +302,12 @@ def build_tag(self, tag, attrs, close_tag):
             return result + '>'
 
         def _handle_meta_tag(self, attrs):
-            name = self._attr_value(attrs, 'name').lower()
+            name = self._attr_value(attrs, 'name')
+            if name is None:
+                attr_serialized = ', '.join(['{}="{}"'.format(k, v) for k, v in attrs])
+                logger.warning("Meta tag in file %s does not have a 'name' attribute, skipping. Attributes: %s", self._filename, attr_serialized)
+                return
+            name = name.lower()
             contents = self._attr_value(attrs, 'content', '')
             if not contents:
                 contents = self._attr_value(attrs, 'contents', '')

diff --git a/pelican/tests/content/article_with_nonconformant_meta_tags.html b/pelican/tests/content/article_with_nonconformant_meta_tags.html
@@ -0,0 +1,12 @@
+<html>
+    <head>
+        <meta charset="utf-8" />
+        <meta http-equiv="X-UA-Compatible" content="IE=edge" />
+        <title>Article with Nonconformant HTML meta tags</title>
+        <meta name="summary" content="Summary and stuff" />
+    </head>
+    <body>
+        Multi-line metadata should be supported
+        as well as <strong>inline markup</strong>.
+    </body>
+</html>
diff --git a/pelican/tests/test_generators.py b/pelican/tests/test_generators.py
@@ -86,6 +86,7 @@ def test_generate_context(self):
             ['Test mkd File', 'published', 'test', 'article'],
             ['This is a super article !', 'published', 'Yeah', 'article'],
             ['This is a super article !', 'published', 'Yeah', 'article'],
+            ['Article with Nonconformant HTML meta tags', 'published', 'Default', 'article'],
             ['This is a super article !', 'published', 'yeah', 'article'],
             ['This is a super article !', 'published', 'yeah', 'article'],
             ['This is a super article !', 'published', 'yeah', 'article'],

diff --git a/pelican/tests/test_readers.py b/pelican/tests/test_readers.py
@@ -384,3 +384,13 @@ def test_article_metadata_key_lowercase(self):
         self.assertIn('category', page.metadata, 'Key should be lowercase.')
         self.assertEqual('Yeah', page.metadata.get('category'),
                          'Value keeps cases.')
+
+    def test_article_with_nonconformant_meta_tags(self):
+        page = self.read_file(path='article_with_nonconformant_meta_tags.html')
+        expected = {
+            'summary': 'Summary and stuff',
+            'title': 'Article with Nonconformant HTML meta tags',
+        }
+
+        for key, value in expected.items():
+            self.assertEqual(value, page.metadata[key], key)