Censor out ascii art, made for Candy Box

censor_ascii.py

import pprint
import os
from collections import Counter



ascii_markers = "|\\/_\" ()n"

def ascii_prob(line):
    line = line.strip()
    counts = Counter(line)
    total = len(line)
    if total == 0:
        return 0
    not_ascii = ['break', 'return', 'else', 'case']
    for phrase in not_ascii:
        if phrase in line:
            return 0
    
    ascii_count = 0
    for c in ascii_markers:
        if c in counts:
            ascii_count += counts[c]
            
    return ascii_count * 1.0 / total



def censor_line(line):
    if ascii_prob(line) < 0.7:
        return line
    
    for mark in ascii_markers:
        line = line.replace(mark, '#')
    
    return line
    
    

def main():
    src_dir = 'cbox123'
    dst_dir = src_dir + '_censored'
    probs = []
    for root, dirs, files in os.walk(src_dir):
        tdir = root.replace(src_dir, dst_dir)
        if not os.path.exists(tdir):
            os.makedirs(tdir)
        for fname in files:
            print fname
            fpath = os.path.join(root, fname)
            out = open(fpath.replace(src_dir, dst_dir), 'w')
            for line in open(fpath):
                probs.append((ascii_prob(line), line))
                out.write(censor_line(line))
    
    pprint.pprint(sorted(probs)[-2500:-2000])

if __name__ == "__main__":
    main()