852_and_866_remover.py

'''takes a marc21 dump and adds 999$i fields to it'''
import sys
from pymarc import MARCReader
from pymarc import Field
import json


def main():
    stats = {
        'records': 0,
        '866': 0,
        '852': 0,
        '852 and 866': 0,
        'failed': 0,
        'ur': 0
        }
    with open(sys.argv[1], 'rb') as marc_file:
        reader = MARCReader(marc_file, 'rb')
        results_file = sys.argv[2]
        with open(results_file, 'wb+') as res_file:          
            for marc_record in reader:
                stats['records'] += 1
                try:
                    if ('007' in marc_record and marc_record['007'].format_field().startswith('cr')):
                        stats['ur'] += 1  
                    else:
                        f852s = marc_record.get_fields('852')
                        if (f852s):
                            stats['852'] += len(f852s)                  
                        f866s = marc_record.get_fields('866')
                        if (f866s):
                            stats['866'] += len(f866s)
                        if (f866s and f852s):
                            stats['852 and 866'] += 1
                        marc_record.remove_fields('852', '866')
                        mrc = marc_record.as_marc()
                        res_file.write(mrc)                    
                except Exception as ee:
                    print(ee)
                    stats['failed'] += 1
                if stats['records'] % 1000 == 0:
                    print(stats)
        print(stats)


if __name__ == '__main__':
    main()