find the youtube video and downlaod convert to matlab readable format avi from mp4

download_convert.py

'''
This code stupidly take the lined list of movie names, suppossely scraped from imdb
and download them as a first instance of YouTube. It also converts the downloaded video mp4
to avi version to be processed in Matlab.

erogol - erengolge@gmail.com


It needs;
Beautifulsoup 	: easy_install BeautifulSoup -- for scraping
ffmpeg 			: apt-get install ffmpeg -- for video conversion
youtube-dl		: apt0get install youtube-dl -- for youtube download
'''

import urllib2
from BeautifulSoup import BeautifulSoup
import pdb
import time
import os
import shlex
import subprocess	


def get_recent_item(PATH):
	l = [(os.path.getmtime(PATH+x), x) for x in os.listdir(PATH)]
	l.sort()
	return l[0][1]

#RANK_LIST = ['NC-17','PG-13','PG','R','G']
RANK_LIST = ['g', 'nc_17','pg_13','pg','r']
RANK_LIST = ['pg_13','pg','r']
#f = open('movies_download_list.data','r')
f1 = open('g.txt','r')
f2 = open('nc_17.txt','r')
f3= open('pg_13.txt','r')
f4= open('pg.txt','r')
f5= open('r.txt','r')

files = [f3,f4,f5]

counter = 0
already_installed_list = []

for file_obj in files:
	for movie_name in file_obj:
		movie_name = movie_name.replace('\n',' ')
		
		print '***********************************'
		print movie_name
		print '***********************************\n'
		
		FILE_PATH = 'outputs/'
		OUTPUT_PATH = FILE_PATH+RANK_LIST[counter]+'/'
		TEMP_PATH = 'temp/'
		
		# check movie_name whether conveyed already
		
		url = 'http://www.youtube.com/results?search_query='+movie_name.replace(' ','+')+'official+movie+trailer';
		try:
			soup = BeautifulSoup(urllib2.urlopen(url).read())
			#pdb.set_trace()
			link = soup('a',{'dir':'ltr'})[0]
			link = link['href']
			
		except:
			continue
			
		if link not in already_installed_list:
			already_installed_list.append(link)	
			all_link = 'http://www.youtube.com'+link
			#command = "youtube-dl -o '"+OUTPUT_PATH+"/%(title)s-%(id)s.%(ext)s'"+' --write-description --write-info-json --write-thumbnail  -f 18 '+all_link]
			try:
				command = "youtube-dl --restrict-filenames -u erengolge   -p 137555555eren   --min-filesize 3m --max-filesize 50m  -o '"+TEMP_PATH+"%(title)s.%(ext)s'"+' -f 18 '+all_link
				print command
				os.system(command);
				#subprocess.Popen(command, stdout=subprocess.PIPE)
				#output = test.communicate()[0]
				video_name = get_recent_item(TEMP_PATH)
				video_name_alone = video_name.split('.')[0]
				#convert_video to avi format to be processed in Matlab
				if not os.path.exists(OUTPUT_PATH):
				 	os.makedirs(OUTPUT_PATH) 
				command = 'ffmpeg -i '+ TEMP_PATH+video_name+' -an -vb 1000k -r 32 -vf scale=-1:360 ' + OUTPUT_PATH+video_name_alone+'.avi';
				os.system(command)
				os.remove(TEMP_PATH+video_name)
			except:
				pass
		else:
			print movie_name + ' already downloaded !!!!!!!'
		#time.sleep(3);
	counter = counter+1