X-Git-Url: https://git.mdrn.pl/librarian.git/blobdiff_plain/6c0f71f581d5b8d9a519a6645f571f54b38afca8..a6ee2dd83d3c4d5d2d3e8cb3401734ced2b12c22:/librarian/html.py diff --git a/librarian/html.py b/librarian/html.py index 5f832e3..5974d93 100644 --- a/librarian/html.py +++ b/librarian/html.py @@ -263,3 +263,17 @@ def add_table_of_contents(root): root.insert(0, toc) + +def extract_annotations(html_path): + """For each annotation, yields a tuple: anchor, text, html.""" + parser = etree.HTMLParser(encoding='utf-8') + tree = etree.parse(html_path, parser) + footnotes = tree.find('//*[@id="footnotes"]') + if footnotes is not None: + for footnote in footnotes.findall('div'): + anchor = footnote.find('a[@name]').get('name') + del footnote[:2] + text_str = etree.tostring(footnote, method='text', encoding='utf-8').strip() + html_str = etree.tostring(footnote, method='html', encoding='utf-8') + yield anchor, text_str, html_str +