compare two (sorted) csv files and print differences

csvdiff.py

import sys

def mycmp(a, b):
    if a.isdigit() and b.isdigit():
        a = int(a)
        b = int(b)
    return cmp(a, b)

def cmpk(a, b, h):
    for i in h:
        c = mycmp(a[i], b[i])
        if c != 0:
            return c
    return 0

class Reader:
    def __init__(self, filename):
        self.f = open(filename)
        self.first = True
        self.next()
        self.filename = filename
    def next(self):
        if self.first:
            self.first = False
            self.line = self.f.readline()[:-1]
            self.headers = self.line.split('\t')
        self.line = self.f.readline()[:-1]
        self.s = dict(zip(self.headers, self.line.split('\t')))
        return self.line

class SQL:
    def where(self, keys, frm):
        where = []
        for k in keys:
            where.append(k+'='+frm.s[k])
        return " and ".join(where)
    def delete(self, keys, frm):
        print 'delete from `%s` where %s;' % (frm.filename, self.where(keys, frm))
    def update(self, keys, col, frm, to):
        print 'update `%s` set `%s` = \'%s\' where %s; # %s' % (frm.filename, col, to.s[col].replace("'", "\\'"), self.where(keys, frm), frm.s[col])
    def insert(self, frm, to):
        print 'insert into `%s` (%s) VALUES (%s);' % (frm.filename, ", ".join(to.s), ", ".join(to.s.values()))

output = SQL()
frm = Reader(sys.argv[1])
to  = Reader(sys.argv[2])
key = sys.argv[3].split(',')
ignore = sys.argv[4].split(',')
while 1:
    if not(frm.line or to.line): break
    if not(frm.line):
        while to.line:
            output.insert(frm, to)
            to.next()
        break
    if not(to.line):
        while frm.line:
            output.delete(key, frm)
            frm.next()
        break
    c = cmpk(frm.s, to.s, key)
    if (c < 0): # from < to
        output.delete(key, frm)
        frm.next()
        continue
    elif (c > 0): # from > to
        values = []
        for h in to.headers:
            if h in ignore: values.append('NULL')
            else: values.append(to.s[h])
        print 'insert into %s (%s) values (%s);' % (sys.argv[1], ", ".join(to.headers),", ".join(values))
        to.next()
        continue
    else: #from == to
        for i in frm.s.iterkeys():
            if (i in key): continue
            if (i in ignore): continue
            if frm.s[i] != to.s[i]: output.update(key, i, frm, to)
    to.next()
    frm.next()