Function decode_entities

pattern/db/__init__.py:249–265

Decodes HTML entities in the given string ("&lt;" => "<").

(string)

    return string

def decode_entities(string):
    """ Decodes HTML entities in the given string ("&lt;" => "<").
    """
    # http://snippets.dzone.com/posts/show/4569
    def replace_entity(match):
        hash, hex, name = match.group(1), match.group(2), match.group(3)
        if hash == "#" or name.isdigit():
            if hex == '' : 
                return unichr(int(name))                 # "&#38;" => "&"
            if hex in ("x","X"):
                return unichr(int('0x'+name, 16))        # "&#x0026;" = > "&"
        else:
            cp = htmlentitydefs.name2codepoint.get(name) # "&amp;" => "&"
            return cp and unichr(cp) or match.group()    # "&foo;" => "&foo;"
    if isinstance(string, (str, unicode)):
        return RE_UNICODE.subn(replace_entity, string)[0]
    return string

class _Binary:
    """ A wrapper for BLOB data with engine-specific encoding.