From: Radek Czajka <radoslaw.czajka@nowoczesnapolska.org.pl>
Date: Wed, 4 Jan 2012 12:34:57 +0000 (+0100)
Subject: #1913: tags in section titles break table of contents
X-Git-Tag: 1.7~172
X-Git-Url: https://git.mdrn.pl/librarian.git/commitdiff_plain/e394602de9243608d1e99a3de448a75646f1a77f?hp=--cc

#1913: tags in section titles break table of contents
---

e394602de9243608d1e99a3de448a75646f1a77f
diff --git a/librarian/html.py b/librarian/html.py
index 997f904..39e5a01 100644
--- a/librarian/html.py
+++ b/librarian/html.py
@@ -228,10 +228,12 @@ def add_table_of_contents(root):
             if any_ancestor(element, lambda e: e.get('id') in ('footnotes',) or e.get('class') in ('person-list',)):
                 continue
 
+            element_text = etree.tostring(element, method='text',
+                    encoding=unicode).strip()
             if element.tag == 'h3' and len(sections) and sections[-1][1] == 'h2':
-                sections[-1][3].append((counter, element.tag, ''.join(element.xpath('text()')), []))
+                sections[-1][3].append((counter, element.tag, element_text, []))
             else:
-                sections.append((counter, element.tag, ''.join(element.xpath('text()')), []))
+                sections.append((counter, element.tag, element_text, []))
             add_anchor(element, "s%d" % counter, with_link=False)
             counter += 1