def _parse_entities()

in bq-connector/docai_bq_connector/connector/BqDocumentMapper.py [0:0]

• 46 lines of code
• 9 McCabe index (conditional complexity)

    def _parse_entities(self, entities) -> DocumentRow:
        row = DocumentRow()
        for entity in entities:
            if len(entity.page_anchor.page_refs) != 1:
                continue
            content = entity.mention_text
            value = content if content is not None and content.strip() != "" else None
            if len(entity.properties) == 0:
                if row.find_field_by_name(entity.type_) is not None:
                    self.errors.append(
                        ConversionError(
                            entity.type_,
                            value,
                            "Duplicate field definition",
                            None,
                            ConversionError.error_type_duplicate_field,
                            identifier=entity.id,
                        )
                    )
                    continue
                row.fields.append(
                    DocumentField(
                        entity.type_,
                        value,
                        entity.normalized_value,
                        entity.confidence,
                        entity.page_anchor.page_refs[0].page + 1,
                    )
                )
            else:
                parent_field = row.find_field_by_name(entity.type_)
                if parent_field is None:
                    parent_field = DocumentField(
                        entity.type_,
                        value,
                        entity.normalized_value,
                        entity.confidence,
                        entity.page_anchor.page_refs[0].page + 1,
                    )
                    row.fields.append(parent_field)
                row_children = self._parse_entities(entity.properties)
                if len(parent_field.children) > 0:
                    parent_field.children[0].fields.extend(row_children.fields)
                else:
                    parent_field.children.append(row_children)
        return row