prepare PDFs and matching MARC records for uploading to the Internet Archive

preparePDFupload.py

# ##################################################
#
# python3 program to prepare for Internet Archive upload of PDFs and MARC records
# this will do the following:
#       1. create a list of identifiers from a folder of PDFs;
#       2. make sure MARC records exist with the same identifiers (stopping on mismatches)
#       3. pull MARC records into the same directory and rename them to IA convention
#       4. generate a spreadsheet called upload.csv to use with IA command-line tool
# 
# TO DO:
#       - query database for volume information and also update database on completion
#       - move to pure Python uploading to eliminate extra steps (no spreadsheet even needed), 
#          and enable file cleanup when done
#       - generally clean up and make more robust (support XML for MARC, etc)
#
# ##################################################

from tkinter.filedialog import askdirectory
import os
import shutil

dirname = askdirectory(title='Choose a directory of PDFs',initialdir='Z:\\holding\\books\\IA_upload\\')

def listPDFs(dirname):
        list = []
        for file in os.listdir(dirname):
                if file.endswith('.pdf'):
                        list.append(file[:-4])
        return list 

def checkmrcs(list):
        rtn = True
        for file in list:
                if not os.path.isfile('Z:\\holding\\books\\MARC_records\\' + file + '.mrc'):
                        print('No matching MARC record for ' + file)
                        rtn = False
        return rtn                

def prepareupload(list):
        filename = dirname + '\\upload.csv'
        csvremainder=".pdf,,Digitized at Georgetown University Law Library,georgetown-university-law-library-rr,true,texts\n"
        with open(filename,mode = 'w',encoding = 'utf-8') as fh:
                fh.write('identifier,file,volume,description,collection,noindex,mediatype\n')
                for f in list:
                        shutil.copy('Z:\\holding\\books\\MARC_records\\' + f + '.mrc',dirname + '\\' + f + '_meta.mrc')
                        fh.write(f + ',' + f + csvremainder)
                        fh.write(',' + f + '_meta.mrc,,,,,\n')
        
list = listPDFs(dirname)
if checkmrcs(list):
        accessfile = dirname + '\\_filelist.txt'
        with open (accessfile,mode = 'w') as fh:
                for file in list:
                        fh.write(file + '\n')
        prepareupload(list)
        print('Ready to upload. Load filelist.txt into Access and manually add any volume numbers to the spreadshet before running uploader.\n\nWhen ready, the IA command-line tool should work (if installed) by running this command: ia upload --spreadsheet=upload.csv')

else:
        print('Fix all MARC record/PDF mismatches before continuing')