SOJ statistics

SOJStatistics.py

import urllib2
import re
oo = 9000

# userPageAddr = 'http://soj.me/user.php?id=8750'
userPageAddr = raw_input('Please input your user page address on Sicily(http://soj.me/user.php?id=8750, for example): ')
userId = re.findall('id=(\d+)', userPageAddr)[0]
s = urllib2.urlopen(userPageAddr).read()
problems = re.findall(r'show_problem\.php\?pid=(\d+)', s, re.M)
problems.sort()

def get_pass_count(problemID):
    addr = 'http://soj.me/show_problem.php?pid=%s' % problemID
    page = urllib2.urlopen(addr).read()
    try:
        tds = re.findall(r"<td[^>]*>(.*)</td>", page, re.M)
        return int(tds[tds.index('Accepted:')+1])
    except:
        print 'Can not view problem#%d' % problemID
        return 0

levels = [1, 50, 100, 150, 200, 300, oo]
intervals = [(levels[i], levels[i+1]) for i in xrange(len(levels)-1)]
data = {interval:0 for interval in intervals}
total = 0
for pid in problems:
    pid = int(pid)
    # print 'Problem#%d ' % pid,
    cnt = get_pass_count(pid)
    # print 'pass %d people' % cnt
    for l, r in intervals:
        if l <= cnt < r:
            data[(l, r)] += 1
            total += 1
            break
percents = {interval:cnt*100./total for interval, cnt in data.iteritems()}

for l, r in intervals:
    print 'interval [%4d, %4d), accepted %3d(%.2f%%).' %(
        l, r, data[l, r], percents[l, r])
print 'total accepted: %d' % total

with open('save-id-%s' %(userId), 'w') as f:
    print >> f, userPageAddr
    print >> f, total
    print >> f, data

raw_input('Press enter to exit...')