Расчет шоколадок для школьников

chocoCalc.py

#!/usr/bin/python3
# -*- coding: utf-8 -*-
import sys,codecs
import urllib.request
import re

sys.stdout = codecs.getwriter('cp866')(sys.stdout.buffer, 'replace')

url = 'http://informatics.mccme.ru/course/view.php?id=526'
cookies = ''


def readUrl(url):
    print('Retrieving %s...' % url, end='');
    sys.stdout.flush()
    headers = {'Cookie': cookies}
    req = urllib.request.Request(url, None, headers)
    response = urllib.request.urlopen(req)
    res = str(response.read(), encoding='utf-8', errors='replace')
    print('Done')
    return res

def getProbNum(monitor):
    mainTable = re.search(r'<table class="BlueTable".*</table>', monitor, re.DOTALL)
    assert(mainTable)
    contests = mainTable.group(0).split('<tr class="Caption">')
    result = []
    for contest in contests:
        nc = contest.count("<tr>")
        if nc > 0:
            result.append(nc)
    return result
    
def getProbNumNew(monitor):
    mainTable = re.search(r'<table align=center class="BlueTable".*?</table>', monitor, re.DOTALL)
    assert(mainTable)
    contests = re.findall(r'<td colspan=(\d+)>', mainTable.group(0))
    #print("contests: ",contests)
    result = list(map(int, contests))
    return result

def convertPoints(points):
    result = []
    for tmp, bg, sc in points:
        if (bg == 'ffff00'):
            result.append(sc)
        elif (bg == 'e1ffe2'):
            result.append('*')
        elif (bg == 'e10000'):
            result.append('I')
        elif (sc == '&nbsp;'):
            result.append('.')
        else:
            result.append('-')
    return result

def getChildren(monitor):
    mainTable = re.search(r'<table align=center class="BlueTable".*</table>', monitor, re.DOTALL)
    assert(mainTable)
    children = re.findall(r'<tr>\s*<td>[0-9-]+\s*<td><a href="/user/.*?</tr>', mainTable.group(0), re.DOTALL)
    result = {}
    for child in children:
        name = re.search(r'<a href="/user/view.php\?id=\d+">([^<]*)</a>', child)
        assert(name)
        name = name.group(1)
        points = re.findall(r'<td width="25"(  bgcolor="#([0-9a-fA-F]*)")? >([-+0-9]*|&nbsp;)</td>',child)
        if (name in result.keys()):
            print('Contestant %s duplicates!' % name)
            exit(1)
        result[name] = convertPoints(points)
    return result

def appendData(data, probNum, children):
    for name, res in children.items():
        if (not name in data.keys()):
            data[name] = []
        probI = 0
        groupI = 0
        add = []
        #print(name)
        #print(probNum)
        #print(res)
        #print(len(res))
        #print(sum(probNum))
        for prob in res:
            probI = probI + 1
            add.append(prob)
            if (probI == probNum[groupI]):
                #print("switch: ", add)
                data[name].append(add)
                probI = 0
                add = []
                groupI = groupI + 1
            #print(groupI)
        assert(groupI == len(probNum))

def parseContests(monitor, data):
    probNum = getProbNumNew(monitor)
    children = getChildren(monitor)
    appendData(data, probNum, children)


def calcFull(cont):
    result = 0
    for c in cont:
        ok = True
        for p in c:
            if (p[0] != '+'):
                ok = False
        if ok:
            result = result + 1
    fullCont = result
    if (result > 0):
        result = result // 3 + 1
    return (result, 'full%d' % fullCont)

def calcClear(cont):
    result = 0
    for c in cont:
        ok = True
        for p in c:
            if (p != '+'):
                ok = False
        if ok:
            result = result + 1
    return (result, 'clear%d' % result)

def calcSemiClear(cont):
    result = 0
    for c in cont:
        ok = True
        was1 = False
        for p in c:
            if (p == '+'):
                pass
            elif (p == '+1'):
                was1 = True
            else:
                ok = False
        if ok and was1:
            result = result + 1
    choco = result // 2
    return (choco, 'semi%d' % result)

chocoFunc = [calcFull, calcClear, calcSemiClear]

def calcChocoForCont(cont):
    result = []
    for func in chocoFunc:
        result.append(func(cont))
    return result

def calcChoco(data):
    choco = {}
    totalChoco = []
    for name, cont in data.items():
        choco[name] = calcChocoForCont(cont)
        total = 0
        for func in choco[name]:
            total = total + func[0]
        totalChoco.append((total, name))
    totalChoco.sort(reverse = True)
    return choco, totalChoco


mainPage = readUrl(url)
monList = re.findall(r'"(http://informatics.mccme.ru/mod/monitor/view.php\?id=\d+)"', mainPage)
data = {}
for monPage in monList:
    monitor = readUrl(monPage)
    parseContests(monitor, data)

choco, totalChoco = calcChoco(data)

for tmp, name in totalChoco:
    s = "%30s|" % name
    t = ''
    total = 0
    for func in choco[name]:
        total = total + func[0]
        t = t + '%d(%s) ' % func
    s = s + '%d=%s|' % (total, t)
    for c in data[name]:
        for p in c:
            if (p[0]=='+'):
                if (len(p)>2):
                    s = s + 'Z'
                else:
                    s = s + p[-1]
            else:
                s = s + p
        s = s + '|'
    print(s)