Many FB2 fixes

[librarian.git] / librarian / html.py

diff --git a/librarian/html.py b/librarian/html.py
index b126638..c1a5e5b 100644 (file)
--- a/librarian/html.py
+++ b/librarian/html.py
@@ -221,16 +221,23 @@ def add_anchors(root):
             counter += 1
 
 
             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
     for element in root.iterdescendants():
         if element.tag in ('h2', 'h3'):
 def add_table_of_contents(root):
     sections = []
     counter = 1
     for element in root.iterdescendants():
         if element.tag in ('h2', 'h3'):

-            if any_ancestor(element, lambda e: e.get('id') in ('footnotes',) or e.get('class') in ('person-list',)):
+            if any_ancestor(element, lambda e: e.get('id') in ('footnotes', 'nota_red') or e.get('class') in ('person-list',)):

                 continue
 
                 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:
             if element.tag == 'h3' and len(sections) and sections[-1][1] == 'h2':
                 sections[-1][3].append((counter, element.tag, element_text, []))
             else: