bgneal@849: """Common HTML related functions"""
bgneal@849: import bleach
bgneal@9: 
bgneal@9: 
bgneal@849: # Each entry in the _CLEAN_PROFILES dict is a profile name -> 3-tuple pair. The
bgneal@849: # tuple consists of (allowed_tags_list, allowed_attributes_dict,
bgneal@849: # allowed_styles_list)
bgneal@849: #
bgneal@849: _CLEAN_PROFILES = {
bgneal@849:     'comments': (
bgneal@849:         [
bgneal@849:             'a', 'b', 'blockquote', 'br', 'code', 'del', 'em',
bgneal@864:             'h1', 'h2', 'h3', 'h4', 'h5', 'h6', 'hr',
bgneal@849:             'i', 'img', 'li', 'ol', 'p', 'pre', 'strong', 'ul',
bgneal@849:         ],
bgneal@849:         {
bgneal@849:             'a': ['href'],
bgneal@849:             'img': ['src', 'alt', 'title'],
bgneal@849:         },
bgneal@849:         [],
bgneal@849:     ),
bgneal@849:     'news': (
bgneal@849:         [
bgneal@849:             'a', 'b', 'blockquote', 'br', 'caption', 'center', 'code', 'col',
bgneal@849:             'colgroup', 'dd', 'del', 'div', 'dl', 'dt', 'em',
bgneal@864:             'h1', 'h2', 'h3', 'h4', 'h5', 'h6', 'hr',
bgneal@849:             'i', 'img', 'ins', 'li', 'ol', 'p', 'pre', 'small', 'strike',
bgneal@849:             'strong', 'sub', 'sup', 'table', 'tbody', 'td', 'tfoot', 'th',
bgneal@849:             'thead', 'tr', 'tt', 'u', 'ul',
bgneal@849:         ],
bgneal@849:         {
bgneal@849:             'a': ['href'],
bgneal@849:             'img': ['src', 'alt', 'title', 'width', 'height'],
bgneal@849:         },
bgneal@849:         [],
bgneal@849:     ),
bgneal@849: }
bgneal@849: 
bgneal@849: 
bgneal@849: def clean_html(text, profile='comments'):
bgneal@9:     """Cleans HTML of dangerous tags and content."""
bgneal@849:     text = text.strip()
bgneal@849:     if not text:
bgneal@849:         return text
bgneal@9: 
bgneal@849:     tags, attrs, styles = _CLEAN_PROFILES[profile]
bgneal@9: 
bgneal@849:     return bleach.clean(text, tags=tags, attributes=attrs, styles=styles,
bgneal@849:         strip=True, strip_comments=True)