tppmodbot_format.py

import codecs
def format_modbot_bans(chatlogfile,outputfile,emotelist=None):
    chatlog = [codecs.decode(x,'utf-8').rstrip().split('\t') for x in open(chatlogfile).readlines() if len(x)>1]
    if emotelist is not None:emotes=[x.rstrip() for x in open(emotelist).readlines()]
    for i,x in enumerate(chatlog):
        if x[1]=='-*-':
            tmp=[y for y in x]
            tmp.pop(1)
            tmp.append('-*- '+' '.join(tmp[1].split(' ')[1:]))
            tmp[1] = tmp[1].split(' ')[0]
            chatlog[i]=tmp
        if emotelist is not None:
            tmp = x[2].split(' ')
            KappaCounter = 0
            ind = 0
            LastKappa = None
            for j,y in enumerate(tmp):
                if y in emotes:
                    if y != LastKappa:
                        KappaCounter=1
                        LastKappa = y
                        ind = j
                    else:
                        KappaCounter+=1
                    tmp[j] = ''
                    tmp[ind] = '({})x{}'.format(LastKappa,KappaCounter)
                else:
                    KappaCounter = 0
                    LastKappa = None
            while '' in tmp:tmp.remove('')
            chatlog[i][2] = ' '.join(tmp)

    # First pass: Find all the TPPModbot lines
    print("Parsing modbot lines...")
    modbot = {i:[x[2].split(' ')[0][1:], ' '.join(x[2].split(' ')[5:])] for i,x in enumerate(chatlog) if x[1].lower() == "tppmodbot"}
    inds = [i for i in modbot.keys()]
    inds.sort()
    # Second pass: Find the offending post and format it into a file, in order.
    op = [["Time","User","Ban","Post"]]
    print("Finding the offenders...")
    for i in inds:
        for j in range(i):
            if chatlog[i-j][1].lower()==modbot[i][0].lower():break
        if chatlog[i-j][1].lower()==modbot[i][0].lower():
            op.append(chatlog[i-j][0:2]+[modbot[i][1]]+[chatlog[i-j][2]])
    with open(outputfile,'w+') as O:
        stats = {}
        for x in op:
            O.write('|'.join([codecs.encode(y,'utf-8') for y in x])+'\n')
            if x[2] not in stats.keys():stats[x[2]]=1
            else:stats[x[2]]+=1
    stats.pop('Ban')
    return stats