remove.py

    class RemoveReferencesFromBodyPipe(CustomPipe):
        def transform(self, data):
            raw, xml = data
            if len(self.super_obj.ref_items) > 0:
                reference_header, references = self._find_references_to_delete(xml)
                if len(self.super_obj.ref_items) == len(references):
                    self._mark_references_to_be_deleted(references + [reference_header])
                else:
                    logger.info("RemoveReferencesFromBodyPipe: FALHOU")
            return data

        def _find_references_to_delete(self, xml):
            """Seleciona os parágrafos a serem removidos com base nos
            delimitadores de referência."""
            reference_header = None
            references = []

            for comment in xml.xpath("//comment()"):
                parent = comment.getparent()

                if reference_header is None and comment.text.strip() == "ref":
                    reference_header = comment.getprevious()
                elif comment.text.strip() == "end-ref":
                    references.append(parent)

            return (reference_header, references)

        def _mark_references_to_be_deleted(self, references):
            """Marca referências a serem removidas durante a execução da pipeline
            de conversão do `body`."""
            for reference in references:
                _remove_tag(reference, True)