Method handle_entityref

pattern/web/soup/BeautifulSoup.py:1403–1444

Handle entity references as data, possibly converting known HTML and/or XML entity references to the corresponding Unicode characters.

(self, ref)

        self.handle_data(data)

    def handle_entityref(self, ref):
        """Handle entity references as data, possibly converting known
        HTML and/or XML entity references to the corresponding Unicode
        characters."""
        data = None
        if self.convertHTMLEntities:
            try:
                data = unichr(name2codepoint[ref])
            except KeyError:
                pass

        if not data and self.convertXMLEntities:
                data = self.XML_ENTITIES_TO_SPECIAL_CHARS.get(ref)

        if not data and self.convertHTMLEntities and \
            not self.XML_ENTITIES_TO_SPECIAL_CHARS.get(ref):
                # TODO: We've got a problem here. We're told this is
                # an entity reference, but it's not an XML entity
                # reference or an HTML entity reference. Nonetheless,
                # the logical thing to do is to pass it through as an
                # unrecognized entity reference.
                #
                # Except: when the input is "&carol;" this function
                # will be called with input "carol". When the input is
                # "AT&T", this function will be called with input
                # "T". We have no way of knowing whether a semicolon
                # was present originally, so we don't know whether
                # this is an unknown entity or just a misplaced
                # ampersand.
                #
                # The more common case is a misplaced ampersand, so I
                # escape the ampersand and omit the trailing semicolon.
                data = "&%s" % ref
        if not data:
            # This case is different from the one above, because we
            # haven't already gone through a supposedly comprehensive
            # mapping of entities to Unicode characters. We might not
            # have gone through any mapping at all. So the chances are
            # very high that this is a real entity, and not a
            # misplaced ampersand.
            data = "&%s;" % ref
        self.handle_data(data)

    def handle_decl(self, data):
        "Handle DOCTYPEs and the like as Declaration objects."