#!/usr/bin/python
# -*- coding: utf-8 -*-

import multiprocessing as mp
import os
import time
import sys

def parse_argv(argv):
    if len(argv) != 2:
        print "Usage: detect_sponsor.py <input_file>"
        sys.exit(1)

    return argv[1]

def ts2sec(ts):
    h = int(ts[0:2]) * 3600
    m = int(ts[3:5]) * 60
    s = int(ts[6:8])
    return h + m + s
    
def sec2ts(sec):
    s = sec % 60
    sec /= 60    
    m = sec % 60
    sec /= 60
    h = sec
    return '%02d:%02d:%02d' % (h, m, s)
    
def make_time_sequence(file_name, start_time, end_time, timestep):
    st_sec = ts2sec(start_time)
    ed_sec = ts2sec(end_time)
    sequence = [(file_name, sec2ts(x)) for x in range(st_sec, ed_sec, timestep)]
    return sequence
    
def detect_sponsor(file_name):
    sequence = make_time_sequence(file_name, "00:01:20", "00:05:00", 3)
   
    pool = mp.Pool(mp.cpu_count() * 2)
    results = pool.map(core_func, sequence, len(sequence)/4)

    pool.terminate()
    pool.join()

    #results = [core_func(x) for x in sequence]

    for r in results:
        if r[1] == True:
            print "found sponsor text in", r[0]
            return

    print 'no sponsor text found'
    

def core_func(seq):
    # 영상 파일의 1분 20초부터 5분 00초까지
    # 매 3초당 스크린샷을 추출한다.
    file = seq[0]
    ts   = seq[1]
    path = './img/image_%s.png' % ts
    
    cmd = 'ffmpeg -ss %s -i \"%s\" -f image2 -vframes 1 -vcodec png %s' % (ts, file, path)
    os.system(cmd)

    # 추출된 파일의 threshold를 대략 95% 정도 주어서 이진화시킨다.
    thresh_file = './mono_%s.png' % ts
    conv_cmd = 'convert -threshold 95%% %s %s' % (path, thresh_file)
    os.system(conv_cmd)

    # 이진화된 영상에 tasseract 일본어를 적용시킨다.
    result_file = './result_%s' % ts
    ocr_cmd = 'tesseract %s %s -l jpn' % (thresh_file, result_file)
    os.system(ocr_cmd)

    ret = False
    with open(result_file+'.txt', 'r') as f:
        first_line = f.readline()
        if first_line.strip() == '提 供':
            print result_file, 'has sponsor text!'
            ret = True

    os.remove(thresh_file)
    os.remove(result_file+'.txt')
    return (ts, ret)
    
def main(argv = None):
    file_name = parse_argv(argv)
    detect_sponsor(file_name)

if __name__ == '__main__':
    sys.exit(main(sys.argv))