def make_doc_field_detector(self) -> DocumentFieldDetector: detector = DocumentFieldDetector() detector.exclude_regexps = 'cushion' detector.include_regexps = r'(?<=\D{3,3}\s\D{5,5}\s)\D+' detector.detected_value = 'shall' detector.extraction_hint = None return detector
def make_doc_field_detector( self, exclude_regexps: Optional[str] = None, include_regexps: Optional[str] = None, detected_value: Optional[str] = None) -> DocumentFieldDetector: detector = DocumentFieldDetector() detector.exclude_regexps = exclude_regexps if exclude_regexps is not None else 'cushion' detector.include_regexps = include_regexps if include_regexps is not None else r'(?<=\D{3,3}\s\D{5,5}\s)\D+' detector.detected_value = detected_value if detected_value is not None else 'shall' detector.extraction_hint = 'detected' return detector
def make_doc_field_detector(exclude_regexps: Optional[str] = None, include_regexps: Optional[str] = None, detected_value: Optional[str] = None, regexps_pre_process_lower: bool = True, definition_words: Optional[str] = None) -> DocumentFieldDetector: detector = DocumentFieldDetector() detector.exclude_regexps = exclude_regexps if exclude_regexps is not None else 'cushion' detector.include_regexps = include_regexps if include_regexps is not None else r'(?<=\D{3,3}\s\D{5,5}\s)\D+' if detected_value is not None: detector.detected_value = detected_value detector.extraction_hint = 'TAKE_FIRST' # 'detected' detector.text_part = 'INSIDE_REGEXP' detector.regexps_pre_process_lower = regexps_pre_process_lower detector.definition_words = definition_words return detector