harvest_reference_numbers.py

#usr/bin/env python

'''
Loop/iterate through all files within  subfolders:
    find the SIP log file
        how to process each line in a textfile
        find the line in the log file that contains the reference number
        extract just the value
        print a list of values
        maybe find some way to insert into the OE register
'''
import os
import sys
import ififuncs

source_dir = sys.argv[1]
for root, dirnames, filenames in sorted(os.walk(source_dir)):
    for filename in filenames:
        if filename.endswith('_sip_log.log'):
            full_path = os.path.join(root, filename)
            with open(full_path) as logfile:
                for line in logfile:
                    if 'Input =' in line:
                        base, ext = os.path.splitext(line.rstrip().split()[-1])
                        original_oe = base
                    if 'eventIdentifierType=UUID' in line:
                        
                        uuid = line.split('value=')[-1].rstrip().split(',')[0]
                    if 'has source' in line:
                        source_tape = line.split('has source=')[-1].rstrip()
                    if 'eventIdentifierType=object entry' in line:
                        new_oe = line.split('value=')[-1].rstrip()
                    if 'eventIdentifierType=Filmographic reference number' in line:
                        ref_no = line.split('value=')[-1].rstrip() 
                    if 'eventIdentifierType=accession number' in line:
                        accession_number = line.split('value=')[-1].rstrip()
                    location = os.path.dirname(os.path.dirname(os.path.dirname(full_path)))
                        
                print '%s,%s,%s,%s,%s,%s,%s ' % (original_oe, uuid, source_tape, new_oe, ref_no, accession_number, location)
                        
                ''''
                if 'eventIdentifierType=Filmographic reference number' in line:
                    print('Representation of ' + line.split('value=')[-1].rstrip() +'|')
                '''
                '''
                if 'Input =' in line:
                    base, ext = os.path.splitext(line.rstrip().split()[-1])
                    #print base, ext
                    old_oe_filename = os.path.join(root.replace('logs', 'objects'), os.path.splitext(line.rstrip().split()[-1])[0]) + '%s' % ext
                    print old_oe_filename
                    objects = os.path.dirname(old_oe_filename)
                    uuid_filename = os.path.join(objects, os.listdir(objects)[0])
                    print uuid_filename
                    os.rename(uuid_filename, old_oe_filename)
                '''