#############################################################
# csv2xml.py
# A simple python script which converts the UNSPSC CSV
# format file to an XML form suitable for XSLT processing
# into a topic map.
#
# (c) 2002 Khalil Ahmed
# Permission is granted to use, copy and redistribute this
# file as long as this notice remains intact.
# The author requests to be notified of any significant
# enhancements or bug fixes made to this file. Notification
# may be made by email to kal@techquila.com
##############################################################

f = open('11UNSPSC.csv', 'r')
lines = f.readlines()
o = open('11UNSPSC.xml', 'w+')
o.write('<?xml version="1.0"?>')
o.write('<unspsc>')

inseg = 0
infam = 0
incls = 0

def write_meta(o, toks):
    o.write('<title>%s</title>\n' % toks[4])
    o.write('<add ver="%s" date="%s"/>\n' % (toks[5], toks[6][:-1]))
    o.write('<entry>%s.%s.%s.%s</entry>\n' % (toks[0], toks[1], toks[2], toks[3]))
for l in lines[1:]:
    toks = l.split(",")
    if (len(toks) != 7):
        print("Error - found a line with more than 7 columns.")
    if (toks[1] == "00"):
        if (incls):
            o.write('</class>\n')
            incls = 0
        if (infam):
            o.write('</family>\n')
            infam = 0
        if (inseg):
            o.write('</segment>\n')
        inseg = 1
        o.write('<segment id="%s">\n' % toks[0])
        write_meta(o, toks)
    elif(toks[2] == "00"):
        if (incls):
            o.write('</class>\n')
            incls = 0
        if (infam):
            o.write('</family>\n')
        infam = 1
        o.write('<family id="%s">\n' % toks[1])
        write_meta(o, toks)
    elif(toks[3] == "00"):
        if (incls):
            o.write('</class>\n')
        incls = 1
        o.write('<class id="%s">\n' % toks[2])
        write_meta(o, toks)
    else:
        o.write('<commodity id="%s">' % toks[3])
        write_meta(o, toks)
        o.write('</commodity>\n')
o.write('</class>\n')
o.write('</family>\n')
o.write('</segment>\n')
o.write('</unspsc>')
o.close()
f.close()