Convert the spread sheet to an author list

exac_auth_afil.py

import sys
import re
from operator import itemgetter

char_map = { '0':  u'\u2070', \
             '1':  u'\u00B9', \
             '2':  u'\u00B2', \
             '3':  u'\u00B3', \
             '4':  u'\u2074', \
             '5':  u'\u2075', \
             '6':  u'\u2076', \
             '7':  u'\u2077', \
             '8':  u'\u2078', \
             '9':  u'\u2079' }

def to_super(num):
    out = ''
    for c in str(num):
        out += char_map[c]
    return out


start_affil = int(sys.argv[1])
affiliations = {}
for l in open(sys.argv[2]):
    d = l.rstrip().split('\t')
    if d[0] == 'Code' or len(d) < 6:
        continue
    affiliations[d[0]] = (d[1] + ', ' if len(d[1]) > 0 else '') + \
                         (d[2]+ ', ' if len(d[2]) > 0 else '') + \
                          d[3] + ', ' + \
                         (d[4] + ', ' if len(d[4]) > 0 else '') + \
                          d[5]

affil_order = {}

authors = []
for l in open(sys.argv[3]):
    d = l.rstrip().split('\t')
    if d[0] == 'LastName' or len(d) < 5:
        continue
    author_affils = []
    for author_affil in re.split(', ',d[4]):
        if not author_affil in affil_order:
            affil_order[author_affil] = len(affil_order)
        author_affils.append(str(affil_order[author_affil]+ start_affil))

    authors.append( d[1] + ' ' + \
                    (d[2] + ' ' if len(d[2]) > 0 else '') + \
                    d[0] + \
                    ','.join(author_affils) )

print ', '.join(authors)

affil_out = []
for a in affiliations:
    if a in affil_order:
        affil_out.append( [affil_order[a],affiliations[a]])

for affil in sorted(affil_out, key=itemgetter(0)):
    print affil[0]+start_affil,affil[1]