submitsCalc.py

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

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

# Лицей 40
#tableBaseUrl = 'http://informatics.mccme.ru/moodle/ajax/ajax.php?problem_id=0&group_id=3696&user_id=0&lang_id=-1&status_id=-1&statement_id=0&objectName=submits&count=10&with_comment=&page=%d&action=getHTMLTable'
# Заоч
tableBaseUrl = 'http://informatics.mccme.ru/moodle/ajax/ajax.php?problem_id=0&group_id=3644&user_id=0&lang_id=-1&status_id=-1&statement_id=0&objectName=submits&count=10&with_comment=&page=%d&action=getHTMLTable'
AC = 'Зачтено/Принято'
startDateStr = '2015-10-04'
endDateStr = '2015-10-10'
startDate = datetime.datetime.strptime(startDateStr + " 00:00:00", "%Y-%m-%d %H:%M:%S")
endDate = datetime.datetime.strptime(endDateStr + " 23:59:59", "%Y-%m-%d %H:%M:%S")

def readUrl(url, cookies=''):
    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 processSubmit(name, prob, date, outcome, childrenResults):
    if date > endDate:  # we do not need this, but continue search
        return True
    if date < startDate:  # stop search
        return False
    #print(name, prob, date, outcome)
    if not (name in childrenResults.keys()):
        childrenResults[name] = {}
    if not (prob in childrenResults[name].keys()):
        childrenResults[name][prob] = ""
    if (childrenResults[name][prob] != AC) and (outcome == AC):
        childrenResults[name][prob] = AC
    return True

def parseSubmits(submitsTable, childrenResults):
    submitsRows = submitsTable.split("<tr>")
    result = False
    for row in submitsRows:
        data = re.search(r'<td>[^<]*</td>\s*<td><a [^>]*>([^<]*)</a></td>\s*<td><a [^>]*>([^<]*)</a></td>\s*<td>([^<]*)</td>\s*<td>[^<]*</td>\s*<td>([^<]*)</td>', row, re.DOTALL)
        if (not data):
            continue
        name = data.group(1)
        prob = data.group(2)
        dateStr = data.group(3)
        outcome = data.group(4).strip()
        date = datetime.datetime.strptime(dateStr, "%Y-%m-%d %H:%M:%S")
        resultSubmit = processSubmit(name, prob, date, outcome, childrenResults)
        result = result or resultSubmit
    return result


childrenResults = {}
page = 0
while True:
    submitsUrl = tableBaseUrl % (page)
    submitsJson = readUrl(submitsUrl)
    submits = json.loads(submitsJson)
    result = parseSubmits(submits["result"]["text"], childrenResults)
    if not result:
         break
    page = page + 1

for name, res in childrenResults.items():
    cnt = 0
    half = 0
    for prob, outcome in res.items():
        if outcome == AC:
            cnt = cnt + 1
        else:
            half = 0.5
    print(name, cnt+half)