OCR / app /extract_dates /date_engine.py
ShahzainHaider's picture
Upload folder using huggingface_hub
7bbae49
raw
history blame
1.02 kB
from app.extract_dates.validate_date import find_smallest_date, validate_date
def date_extractor(ocr_list):
try:
global dob_found
dob_found = False
dob = None
dates_list = []
for index,word in enumerate(ocr_list):
date_valid ,pattern = validate_date(word)
if date_valid:
# valid_pattern = pattern
ocr_list.pop(index) # removing elements from list that is being used.
dates_list.append(word)
if dates_list:
# print(dates_list)
dob = find_smallest_date(dates_list) # smallest date will be DOB
dob_found = True
print("DATE OF BIRTH : ", dob)
else:
print("Date not found")
dob_found = False
dob = ''
return dob_found, dob, ocr_list
except Exception as e:
print("[Exception in date_extractor] : ", str(e))
dob = ''
return dob_found, dob, ocr_list