python统计小脚本

import sys
import urllib
import random
import re


reload(sys)
sys.setdefaultencoding('utf-8')


#正则
r = re.compile(r' elapse (?P<data>[0-9]+)')
#多少分位
marks = [
        0,
        25,
        50,
        75,
        80,
        85,
        90,
        95
        ]


datas = []


for m in r.finditer("".join(open(sys.argv[1]).readlines())):
    v = m.group('data')
    datas.append(float(v.strip()))


datas.sort()
marks.sort()


for mark in marks:
    if ((mark > 100) or (mark < 0)):
        continue
    mark_percent = 1 - float(mark) / 100
    mark_size = int(mark_percent * len(datas))
    print "%d%% 长尾:\t%f" % (mark, (sum(datas[-mark_size:]) / mark_size))