From: Radek Czajka <radoslaw.czajka@nowoczesnapolska.org.pl>
Date: Tue, 28 Feb 2012 14:26:23 +0000 (+0100)
Subject: HTML: remove anchors from TOC
X-Git-Tag: 1.7~166
X-Git-Url: https://git.mdrn.pl/librarian.git/commitdiff_plain/6badff0808ac498cca8f405d047316940aac8a03

HTML: remove anchors from TOC
---

diff --git a/librarian/html.py b/librarian/html.py
index b126638..e04790f 100644
--- a/librarian/html.py
+++ b/librarian/html.py
@@ -221,6 +221,14 @@ def add_anchors(root):
             counter += 1
 
 
+def raw_printable_text(element):
+    working = copy.deepcopy(element)
+    for e in working.findall('a'):
+        if e.get('class') == 'annotation':
+            e.text = ''
+    return etree.tostring(working, method='text', encoding=unicode).strip()
+
+
 def add_table_of_contents(root):
     sections = []
     counter = 1
@@ -229,8 +237,7 @@ 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()
+            element_text = raw_printable_text(element)
             if element.tag == 'h3' and len(sections) and sections[-1][1] == 'h2':
                 sections[-1][3].append((counter, element.tag, element_text, []))
             else: