import pandas as pd
import re
import phone_processor # type: ignore

def is_valid_phone_number(phone):
    if isinstance(phone, str):
        return bool(re.match(r'^254\d{9}$', phone))
    return False

data = pd.read_csv('./10million2_bulk_message_KCS.csv', usecols=['destination'])

data['is_valid'] = data['destination'].apply(is_valid_phone_number)

valid_numbers = data[data['is_valid']]['destination'].tolist()

processed_numbers = phone_processor.process_phone_numbers(valid_numbers)

# Save to file
output_file = 'processed_phone_numbers.txt'
with open(output_file, 'w') as f:
    for number in processed_numbers:
        f.write(f"{number}\n")

# Print results
print(f"Processed {len(processed_numbers)} valid phone numbers. Saved to {output_file}.")
print('First 10 processed numbers')
print(processed_numbers[:10])