utils.py

from typing import List, Tuple
from src.detections import DateDetection, Optional

def _parse_day_or_month(re_entry) -> List[Tuple[int, int, DateDetection]]:
    assert re_entry["day_or_month_year"] is not None
    result = []
    
    if re_entry["day_month1"] is not None:
        if len(re_entry["day_month1"]) == 1:
            result.append((DateDetection.AnnotationPart.TWO_DIGITS_DAY, "0" + re_entry["day_month1"]))
        else:
            result.append((DateDetection.AnnotationPart.TWO_DIGITS_DAY, re_entry["day_month1"]))
        result.append((DateDetection.AnnotationPart.OTHER, re_entry["punct1"]))
        
        if len(re_entry["day_month2"]) == 1:
            result.append((DateDetection.AnnotationPart.TWO_DIGIT_MONTH, "0" + re_entry["day_month2"]))
        else:
            result.append((DateDetection.AnnotationPart.TWO_DIGIT_MONTH, re_entry["day_month2"]))
            
        result.append((DateDetection.AnnotationPart.OTHER, re_entry["punct1"]))
    elif "day_month2" in re_entry:
        if len(re_entry["day_month2"]) == 1:
            result.append((DateDetection.AnnotationPart.TWO_DIGIT_MONTH, "0" + re_entry["day_month2"]))
        else:
            result.append((DateDetection.AnnotationPart.TWO_DIGIT_MONTH, re_entry["day_month2"]))
    
        result.append((DateDetection.AnnotationPart.OTHER, re_entry["punct1"]))
        
    if "year1" in re_entry:
        if len(re_entry["year1"]) == 2:
            result.append((DateDetection.AnnotationPart.TWO_DIGIT_YEAR, re_entry["year1"]))
        else:    
            result.append((DateDetection.AnnotationPart.FOUR_DIGIT_YEAR, re_entry["year1"]))

    return result

def _parse_year_month_or_day(re_entry) -> List[Tuple[int, int, DateDetection]]:
    assert re_entry["year_month_or_day"] is not None
    result = []
    
    if "year2" in re_entry:
        if len(re_entry["year2"]) == 2:
            result.append((DateDetection.AnnotationPart.TWO_DIGIT_YEAR, re_entry["year2"]))
        else:    
            result.append((DateDetection.AnnotationPart.FOUR_DIGIT_YEAR, re_entry["year2"]))
            
        result.append((DateDetection.AnnotationPart.OTHER, re_entry["punct3"]))
        
    if "day_month3" in re_entry:
        if len(re_entry["day_month3"]) == 1:
            result.append((DateDetection.AnnotationPart.TWO_DIGITS_DAY, "0" + re_entry["day_month3"]))
        else:
            result.append((DateDetection.AnnotationPart.TWO_DIGITS_DAY, re_entry["day_month3"]))
        result.append((DateDetection.AnnotationPart.OTHER, re_entry["punct4"]))
        
    if "day_month4" in re_entry:
        if len(re_entry["day_month4"]) == 1:
            result.append((DateDetection.AnnotationPart.TWO_DIGIT_MONTH, "0" + re_entry["day_month4"]))
        else:
            result.append((DateDetection.AnnotationPart.TWO_DIGIT_MONTH, re_entry["day_month4"]))
            
        result.append((DateDetection.AnnotationPart.OTHER, re_entry["punct4"]))
    
    return result

def _parse_month_in_words(re_entry) -> List[Tuple[DateDetection.AnnotationPart, str]]:
    assert re_entry["month_in_words"] is not None
    result = []
    
    if re_entry["day1"] is not None:
        if len(re_entry["day1"]) == 1:
            result.append((DateDetection.AnnotationPart.TWO_DIGITS_DAY, "0" + re_entry["day1"]))
        else:
            result.append((DateDetection.AnnotationPart.TWO_DIGITS_DAY, re_entry["day1"]))
        
        result.append((DateDetection.AnnotationPart.OTHER, re_entry["punct5"]))
        
    if re_entry["month"] is not None:
        result.append((DateDetection.AnnotationPart.TEXT_MONTH, re_entry["month"]))
        
        if re_entry["day1"] is None:
            result.append((DateDetection.AnnotationPart.OTHER, re_entry["punct7"]))
        else:
            result.append((DateDetection.AnnotationPart.OTHER, re_entry["punct6"]))
        
    if re_entry["day2"] is not None:
        if len(re_entry["day2"]) == 1:
            result.append((DateDetection.AnnotationPart.TWO_DIGITS_DAY, "0" + re_entry["day2"]))
        else:
            result.append((DateDetection.AnnotationPart.TWO_DIGITS_DAY, re_entry["day2"]))
        result.append((DateDetection.AnnotationPart.OTHER, re_entry["punct6"]))
        
    if re_entry["year3"] is not None:
        if len(re_entry["year3"]) == 2:
            result.append((DateDetection.AnnotationPart.TWO_DIGIT_YEAR, re_entry["year3"]))
        else:    
            result.append((DateDetection.AnnotationPart.FOUR_DIGIT_YEAR, re_entry["year3"]))
                
    return result

def _parse_date_to_format(re_entry) -> Optional[List[Tuple[DateDetection.AnnotationPart, str]]]:   
    if re_entry["day_or_month_year"] is not None:
        result = _parse_day_or_month(re_entry)
    elif re_entry["year_month_or_day"] is not None:
        result = _parse_year_month_or_day(re_entry)
    elif re_entry["month_in_words"] is not None:
        result = _parse_month_in_words(re_entry)
    else:
        result = None
        
    return result