Convert ISI Web of Science Plaintext to bibtex

isi2bibtex.py

#!/usr/bin/env python2.7
"""
translate ISI bibliography (in plaintext) to bibtex

  isi2bib  savedrecs.txt  > savedrecs.bib

Notes:
 1. this script tries very hard to make sensible lists
    of author names, but the task is hard, and results
    should be checked carefully.
   
 2. every entry is assumed to be a journal article.

 3. no attempt is made to typeset article titles.
 
"""

import sys
from collections import OrderedDict

            
ISINAMES = {'VL': 'vol', 'IS': 'issue', 'PD': 'month',
            'PY': 'year', 'BP': 'page1', 'EP': 'page2'}

def parse_isi(fname):
    "convert ISI text to a list of record dictionaries"
    fileh = open(fname, 'r')
    lines = fileh.readlines()
    fileh.close()

    citations = []
    last_key = ''
    for line in lines:
        key = line[0:2]
        val = line[3:-1]
        if key == '  ':
            key = last_key
        if key == 'PT': # start record
            rec = {'title':' ', 'source':' '}
            auths = []
        elif key == 'ER': # end record
            rec['authors'] = auths
            citations.append(rec)
        elif key == 'TI':
            rec['title'] = "%s %s" % (rec['title'], val)
        elif key == 'SO':
            rec['source'] = "%s %s" % (rec['source'], val)
        elif key == 'AU':
            auths.append(val.split(','))
        elif key in ISINAMES:
            rec[ISINAMES[key]] = val.split()[0]
        last_key = key
    return citations

def generate_bibtex(citations):
    "convert list of dictionaries (from parse_isi) to bibtex"
    cite_labels = []
    output = OrderedDict()

    for cite in citations:
        auths = cite['authors']
        year  = cite['year']

        nauthors = len(auths)
        iauthor = 0
        authorlist = ''
        citename = None
        for authnames in auths:
            iauthor += 1
            if len(authnames) == 1:
                firstname, suffix = '', ''
                lastname = authnames[0]
                if ' ' in lastname:
                    firstname, lastname = lastname.split(' ', 1)

            else:
                lastname = authnames[0].strip().title()
                firstname = authnames[1].strip()
                suffix =  ''
                if len(authnames) > 2:
                    suffix = ' '.join(authnames[2:])
            if firstname == firstname.upper():
                fnamex = []
                for letter in firstname:
                    if letter != '.' and letter != ' ':
                        fnamex.append(letter) 
                firstname  = '.~'.join(fnamex) + '.'
            if citename is None:
                citename = lastname
            aname = "%s %s" % (firstname, lastname)
            if suffix != '':
                aname = "%s, %s" % (aname, suffix.strip())
            if iauthor < nauthors:
                aname = aname + " and"
            authorlist = authorlist + ' ' + aname

        label = _label = "%s%s" % (citename, year)
        app = 'a'
        while label in cite_labels:
            label = "%s%s" % (_label, app)
            app = chr(1+ord(app))

        cite_labels.append(label)


        para ="""@article{%s,
        author  = "{%s}",
        title   = "{%s}", 
        journal = "{%s}",
        year    = "%s",
        volume  = "%s", """ % (label,  authorlist.strip(),
                               cite['title'].strip().title(),
                               cite['source'].strip().title(),
                               year, cite['vol'])

        if 'issue' in cite:
            para = "%s\n        number  = \"%s\","  % (para, cite['issue'])
        if 'page1' in cite:
            if 'page2' in cite and cite['page2'] != cite['page1']:
                pageline = "pages   = \"%s--%s\"," % (cite['page1'],
                                                      cite['page2'])
            else:
                pageline = "pages   = \"%s\"," % (cite['page1'])
            para = "%s\n        %s" % (para, pageline)
        para = "%s\n      }\n" % para

        output[label] = para
    return output


if __name__ == '__main__':
    for recfile in sys.argv[1:]:
        citation = parse_isi(recfile)
        for refname, bibtex in generate_bibtex(citation).items():
            print bibtex