import sys, re
d = {}
l = []
n = 1
regx0 = re.compile(r'\(([0-9]+)\)[A-Z]+')
regx1 = re.compile(r'[A-Z]+')
repl = lambda g: g.groups()[0]
for line in open(sys.argv[1], 'r'):
    tmpl = []
    if line.startswith('Col'):
        l.append(line.strip())
        continue
    line = regx0.sub(repl, line)
    for i in line.strip().split('\t'):
        if regx1.match(i):
            if not i in d.keys():
                d[i] = 'NA%s' %n
                n += 1
            tmpl.append(d[i])
        else:
            tmpl.append(i)
    l.append('\t'.join(tmpl))
print '\n'.join(l)

2楼: Originally posted by jjdg at 2012-07-07 01:15:44
还是不太清楚你的意图