cleaning bibtex file for correct upper case title and remove url record. Running by: PYTHONIOENCODING=utf-8 python cleanBib.py

cleanBib.py

#!/bin/env python

import bibtexparser
import argparse
import sys
import re
from bibtexparser.bibdatabase import BibDatabase
from bibtexparser.bwriter import BibTexWriter

def modifiedTitle(title):
    newtitle = ''
    for word in title.split(' '):
        if re.match('.*[A-Z]', word) and word[0] != '{' and word[-1] != '}':
            newtitle += ' {' + word + '}'
        else:
            newtitle += ' ' + word
    return newtitle.strip(' ')

def parseEntry(bibEntry, outBib):
    for key in bibEntry.iterkeys():
        if 'url' in key or 'Url' in key or 'link' in key:
            del bibEntry[key]
    bibEntry['title'] = modifiedTitle(bibEntry['title'])
    return bibEntry

def main(inBib, outBib):
    bib_database = bibtexparser.load(inBib)
    db = BibDatabase()
    db.entries = [parseEntry(bibEntry, outBib) for bibEntry in bib_database.entries]
    inBib.close()
    writer = BibTexWriter()
    outBib.write(writer.write(db))
    outBib.close()
    return 0


if __name__ == '__main__':
    parser = argparse.ArgumentParser(description='Cleaning bibtex file for upper case title and remove url')
    parser.add_argument('-i','--inBib',required=True,
            type=argparse.FileType('r'), default=sys.stdin,
            help='input bibtex file')
    parser.add_argument('-o','--outBib', required=True,
            type=argparse.FileType('w'), default=sys.stdout, help ='output file')
    args = parser.parse_args()
    inBib = args.inBib
    outBib = args.outBib
    main(inBib, outBib)