#!/usr/bin/python
# coding:utf8
# @IgnorePep8

import os
import sys
dirname = os.path.dirname(os.path.abspath(__file__))
sys.path.insert(0, dirname + '/..')

import subprocess
from cdgen.post_func import send_hash
from crawler.utils import server_list, get_play_url, get_hid_list, get_hash_list, get_quote_keyward, write_hid_list, build_pipe, send_start_signal

server_num = len(server_list)
crawler_cmd = ['python', '%s/web_crawler.py' % dirname]
p_crawler = None

def get_hids(keyword, fd):
    global p_crawler
    hid_list = []
    base_url = 'http://www.yezibo.com/SearchPlayFile.aspx?key=%s&OrType=False&taxismode=refweb&lock=True&page=%d'
    page = 1
    while True:
        url = base_url % (keyword, page)
        page += 1
        print url
        page_hid_list = get_hid_list(url)
        if not page_hid_list:
            break
        hid_list += page_hid_list
    hid_list = list(set(hid_list))
    write_hid_list(hid_list)
    send_start_signal(fd)
    p_crawler = subprocess.Popen(crawler_cmd)
    return hid_list

def get_hid_hash(hid):
    hash_list = []
    base_url = '%s/OpenPlayMovie.aspx?hid=%s&modular=15&ntime=%d'
    
    print 'HID: %s' % hid
    count = 0
    tried = {}
    while count < 0.1 * server_num:
        url = get_play_url(base_url % (server_list[count % server_num], hid, count))
        count += 1
        tried[url] = tried.get(url, 0) + 1
        if tried[url] > 3:
            continue
        print 'TRY %02d: %s' % (count, url)
        page_hash_list = get_hash_list(url)
        if not page_hash_list:
            continue
        hash_list += page_hash_list
        break
    hash_list = list(set(hash_list))
    return hash_list

def get_hash(hid_list, failed_hid):
    hash_list = []
    for hid in hid_list:
        hid_hash_list = get_hid_hash(hid)
        hash_list += hid_hash_list
        for hash_s in hid_hash_list:
            send_hash(hash_s)
        if not hid_hash_list:
            failed_hid.append(hid)
    hash_list = list(set(hash_list))
    send_hash('END')
    return hash_list

def run(keyword):
    fd = build_pipe()
    quote_keyward = get_quote_keyward(keyword)

    hid_list = get_hids(quote_keyward, fd)
    failed_hid = []
    print 'HID LIST: %d' % len(hid_list)
    open('/tmp/%s.hid' % keyword, 'w').write(str(hid_list))

    hash_list = get_hash(hid_list, failed_hid)
    print 'HASH LIST: %d' % len(hash_list)
    open('/tmp/%s.hash' % keyword, 'w').write(str(hash_list))

    p_crawler.wait()

if __name__ == '__main__':
    run(sys.argv[1])