code.delx.au - monosys/blob - scripts/youtube.cgi

   1 #!/usr/bin/env python
   2
   3 import cgi
   4 from lxml.html import document_fromstring
   5 import os
   6 import re
   7 import shutil
   8 import subprocess
   9 import sys
  10 import urllib
  11
  12
  13 urllib.URLopener.version = "Mozilla/5.0 (Macintosh; Intel Mac OS X 10.6; rv:2.0.1) Gecko/20100101 Firefox/4.0.1"
  14
  15 fmt_quality = [
  16         (38, ".mp4"),  # 4096x3072
  17         (37, ".mp4"),  # 1920x1080
  18         (22, ".mp4"),  # 1280x720
  19         (45, ".webm"), # 1280x720
  20         (43, ".webm"), # 640x360
  21         (35, ".flv"),  # 854x480
  22         (34, ".flv"),  # 640x360
  23         (18, ".mp4"),  # 480x360
  24         (5,  ".flv"),  # 400x240
  25         (17, ".3gp"),  # 176x144
  26 ]
  27
  28
  29 def print_form(url="", msg=""):
  30         print "Content-Type: application/xhtml+xml\r\n\r\n"
  31         print """
  32 <!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.1//EN" "http://www.w3.org/TR/xhtml11/DTD/xhtml11.dtd">
  33 <html xmlns="http://www.w3.org/1999/xhtml">
  34 <head>
  35         <title>delx.net.au - YouTube Scraper</title>
  36         <link rel="stylesheet" type="text/css" href="/style.css"/>
  37         <style type="text/css">
  38                 input[type="text"] {
  39                         width: 100%;
  40                 }
  41                 .error {
  42                         color: red;
  43                 }
  44         </style>
  45 </head>
  46 <body>
  47         <h1>delx.net.au - YouTube Scraper</h1>
  48         {0}
  49         <form action="" method="get">
  50         <p>This page will let you easily download YouTube videos to watch offline. It
  51         will automatically grab the highest quality version.</p>
  52         <div><input type="text" name="url" value="{1}"/></div>
  53         <div><input type="submit" value="Download!"/></div>
  54         </form>
  55         <p>Tip! Use this bookmarklet: <a href="javascript:(function(){window.location='http://delx.net.au/utils/youtube.cgi?url='+escape(location);})()">YouTube Download</a>
  56         to easily download videos. Right-click the link and add it to bookmarks,
  57         then when you're looking at a YouTube page select that bookmark from your
  58         browser's bookmarks menu to download the video straight away.</p>
  59 </body>
  60 </html>
  61 """.replace("{0}", msg).replace("{1}", url)
  62
  63 def parse_url(url):
  64         f = urllib.urlopen(url)
  65         doc = document_fromstring(f.read())
  66         f.close()
  67         return doc
  68
  69 def get_video_url(doc):
  70         embed = doc.xpath("//embed")[0]
  71         flashvars = embed.attrib["flashvars"]
  72         flashvars = cgi.parse_qs(flashvars)
  73         fmt_url_map = {}
  74         for pair in flashvars["fmt_url_map"][0].split(","):
  75                 key, value = pair.split("|")
  76                 key = int(key)
  77                 fmt_url_map[key] = value
  78         for fmt, extension in fmt_quality:
  79                 try:
  80                         video_url = fmt_url_map[fmt]
  81                         break
  82                 except KeyError:
  83                         continue
  84         else:
  85                 return None, None, None
  86
  87         title = doc.xpath("/html/head/title/text()")[0]
  88         title = re.sub("\s+", " ", title.strip())
  89         valid_chars = frozenset("-_.() abcdefghijklmnopqrstuvwxyzABCDEFGHIJKLMNOPQRSTUVWXYZ0123456789")
  90         filename = "".join(c for c in title.encode("ascii", "ignore") if c in valid_chars)
  91         filename += extension
  92
  93         return video_url, filename
  94
  95 def cgimain():
  96         args = cgi.parse()
  97         try:
  98                 url = args["url"][0]
  99         except:
 100                 print_form(url="http://www.youtube.com/watch?v=FOOBAR")
 101                 return
 102
 103         try:
 104                 doc = parse_url(url)
 105                 video_url, filename = get_video_url(doc)
 106                 data = urllib.urlopen(video_url)
 107                 httpinfo = data.info()
 108                 sys.stdout.write("Content-Disposition: attachment; filename=\"%s\"\r\n" % filename)
 109                 sys.stdout.write("Content-Length: %s\r\n" % httpinfo.getheader("Content-Length"))
 110                 sys.stdout.write("\r\n")
 111                 shutil.copyfileobj(data, sys.stdout)
 112                 data.close()
 113         except Exception, e:
 114                 print_form(
 115                         url=url,
 116                         msg="<p class='error'>Sorry, there was an error. Check your URL?</p>"
 117                 )
 118                 return
 119
 120 def main():
 121         try:
 122                 url = sys.argv[1]
 123         except:
 124                 print >>sys.stderr, "Usage: %s http://youtube.com/watch?v=FOOBAR" % sys.argv[0]
 125                 sys.exit(1)
 126         doc = parse_url(url)
 127         video_url, filename = get_video_url(doc)
 128         data = urllib.urlopen(video_url)
 129         outfile = open(filename, "w")
 130         shutil.copyfileobj(data, outfile)
 131         data.close()
 132         outfile.close()
 133
 134
 135 if __name__ == "__main__":
 136         if os.environ.has_key("SCRIPT_NAME"):
 137                 cgimain()
 138         else:
 139                 main()
 140