Changed path for importing books.

[wolnelektury.git] / lib / librarian / html.py

diff --git a/lib/librarian/html.py b/lib/librarian/html.py
index ea71bbb..3dd70f5 100644 (file)
--- a/lib/librarian/html.py
+++ b/lib/librarian/html.py
@@ -109,7 +109,7 @@ def extract_fragments(input_filename):
 
     for event, element in etree.iterparse(input_filename, events=('start', 'end')):
         # Process begin and end elements
-        if element.tag == 'span' and element.get('class', '') in ('theme-begin', 'theme-end'):
+        if element.get('class', '') in ('theme-begin', 'theme-end'):
             if not event == 'end': continue # Process elements only once, on end event
 
             # Open new fragment
@@ -171,12 +171,18 @@ def add_anchor(element, number):
     element.insert(0, anchor_target)
 
 
+def any_ancestor(element, test):
+    for ancestor in element.iterancestors():
+        if test(ancestor):
+            return True
+    return False
+
+
 def add_anchors(root):
     counter = 1
     for element in root.iterdescendants():
-        if element.getparent().tag in 'div' and 'note' in element.getparent().get('class', ''):
-            continue
-        if element.getparent().tag in 'blockquote':
+        if any_ancestor(element, lambda e: e.get('class') in ('note', 'motto', 'motto_podpis', 'dedication')
+        or e.tag == 'blockquote'):
             continue
         
         if element.tag == 'p' and 'verse' in element.get('class', ''):