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

   1 #!/usr/bin/env python3
   2
   3 import cgi
   4 import html.parser
   5 import http.cookiejar
   6 import json
   7 import os
   8 import re
   9 import shutil
  10 import subprocess
  11 import sys
  12 import time
  13 import urllib.error
  14 import urllib.parse
  15 import urllib.request
  16
  17
  18 MAX_MEMORY_BYTES = 128 * 1024*1024
  19 USER_AGENT = "Mozilla/5.0 (X11; Ubuntu; Linux x86_64; rv:15.0) Gecko/20100101 Firefox/15.0.1"
  20
  21 MIMETYPES = {
  22     "video/mp4": "mp4",
  23     "video/x-flv": "flv",
  24     "video/3gpp": "3gp",
  25 }
  26
  27 QUALITIES = {
  28     "hd1080": 5,
  29     "hd720": 4,
  30     "large": 3,
  31     "medium": 2,
  32     "small": 1,
  33 }
  34
  35
  36 class VideoUnavailable(Exception):
  37     pass
  38
  39 def print_form(url="", msg=""):
  40     script_url = "http://%s%s" % (os.environ["HTTP_HOST"], os.environ["REQUEST_URI"])
  41     sys.stdout.write("Content-Type: text/html\r\n\r\n")
  42     sys.stdout.write("""
  43 <!DOCTYPE html>
  44 <html>
  45 <head>
  46     <title>delx.net.au - YouTube Scraper</title>
  47     <link rel="stylesheet" type="text/css" href="/style.css">
  48     <style type="text/css">
  49         input[type="text"] {
  50             width: 100%;
  51         }
  52         .error {
  53             color: red;
  54         }
  55     </style>
  56 </head>
  57 <body>
  58     <h1>delx.net.au - YouTube Scraper</h1>
  59     {0}
  60     <form action="" method="get">
  61     <p>This page will let you easily download YouTube videos to watch offline. It
  62     will automatically grab the highest quality version.</p>
  63     <div><input type="text" name="url" value="{1}"/></div>
  64     <div><input type="submit" value="Download!"/></div>
  65     </form>
  66     <p>Tip! Use this bookmarklet: <a href="javascript:(function(){window.location='{2}?url='+escape(location);})()">YouTube Download</a>
  67     to easily download videos. Right-click the link and add it to bookmarks,
  68     then when you're looking at a YouTube page select that bookmark from your
  69     browser's bookmarks menu to download the video straight away.</p>
  70 </body>
  71 </html>
  72 """.replace("{0}", msg).replace("{1}", url).replace("{2}", script_url))
  73
  74 cookiejar = http.cookiejar.CookieJar()
  75 urlopener = urllib.request.build_opener(urllib.request.HTTPCookieProcessor(cookiejar))
  76 referrer = ""
  77
  78 def urlopen(url, offset=None):
  79     if url.startswith("//"):
  80         url = "https:" + url
  81     if not url.startswith("http://") and not url.startswith("https://"):
  82         url = "https://www.youtube.com" + url
  83
  84     global referrer
  85     req = urllib.request.Request(url)
  86     if not referrer:
  87         referrer = url
  88     else:
  89         req.add_header("Referer", referrer)
  90
  91     req.add_header("User-Agent", USER_AGENT)
  92
  93     if offset:
  94         req.add_header("Range", "bytes=%d-" % offset)
  95
  96     res = urlopener.open(req)
  97
  98     content_range = res.getheader("Content-Range")
  99     if content_range:
 100         tokens = content_range.split()
 101         assert tokens[0] == "bytes"
 102         start = int(tokens[1].split("-")[0])
 103         assert start == offset
 104     return res
 105
 106 def parse_url(url, parser):
 107     f = urlopen(url)
 108     parser.feed(f.read().decode("utf-8"))
 109     parser.close()
 110     f.close()
 111
 112 def append_to_qs(url, params):
 113     r = list(urllib.parse.urlsplit(url))
 114     qs = urllib.parse.parse_qs(r[3])
 115     qs.update(params)
 116     r[3] = urllib.parse.urlencode(qs, True)
 117     url = urllib.parse.urlunsplit(r)
 118     return url
 119
 120 def get_player_config(scripts):
 121     player_config = None
 122     for script in scripts:
 123         for line in script.split("\n"):
 124             s = "ytplayer.config = {"
 125             if s in line:
 126                 p1 = line.find(s) + len(s) - 1
 127                 p2 = line.find("};", p1) + 1
 128                 if p1 >= 0 and p2 > 0:
 129                     return json.loads(line[p1:p2])
 130
 131 def extract_js(script):
 132     PREFIX = "var _yt_player={};(function(g){var window=this;"
 133     SUFFIX = ";})(_yt_player);\n"
 134     assert script.startswith(PREFIX)
 135     assert script.endswith(SUFFIX)
 136
 137     return script[len(PREFIX):-len(SUFFIX)]
 138
 139 def find_func_name(script):
 140     FUNC_NAME = R"([a-zA-Z0-9$]+)"
 141     FUNC_PARAMS = R"(\([a-zA-Z,\.]+\.s\))"
 142     TERMINATOR = R"[,;\)]"
 143     PATTERN = FUNC_NAME + FUNC_PARAMS + TERMINATOR
 144
 145     match = re.search(PATTERN, script)
 146     func_name = match.groups()[0]
 147     return func_name
 148
 149 def decode_signature(js_url, signature):
 150     f = urlopen(js_url)
 151     script = f.read().decode("utf-8")
 152     f.close()
 153
 154     func_name = find_func_name(script)
 155
 156     params = {
 157         "func_name": func_name,
 158         "signature": json.dumps(signature),
 159         "code": json.dumps(extract_js(script)),
 160     }
 161     p = subprocess.Popen(
 162         "node",
 163         shell=True,
 164         close_fds=True,
 165         stdin=subprocess.PIPE,
 166         stdout=subprocess.PIPE
 167     )
 168     js_decode_script = ("""
 169         const vm = require('vm');
 170
 171         const sandbox = {
 172             location: {
 173                 hash: '',
 174                 href: '',
 175                 protocol: 'http:'
 176             },
 177             history: {
 178                 pushState: function(){}
 179             },
 180             document: {},
 181             navigator: {
 182                 userAgent: ''
 183             },
 184             matchMedia: () => ({matches: () => {}, media: ''}),
 185             signature: %(signature)s,
 186             transformed_signature: null,
 187             g: function(){} // this is _yt_player
 188         };
 189         sandbox.window = sandbox;
 190
 191         const code_string = %(code)s + ';';
 192         const exec_string = 'transformed_signature = %(func_name)s("", "MARKER", signature);';
 193         vm.runInNewContext(code_string + exec_string, sandbox);
 194
 195         function findSignature(obj) {
 196             if (typeof obj !== 'object') {
 197                 return;
 198             }
 199             for (const [key, value] of Object.entries(obj)) {
 200                 if (key === 'MARKER') {
 201                     return value;
 202                 }
 203                 const result = findSignature(value);
 204                 if (result) {
 205                     return result;
 206                 }
 207             }
 208         }
 209         console.log(findSignature(sandbox.transformed_signature));
 210     """ % params)
 211
 212     p.stdin.write(js_decode_script.encode("utf-8"))
 213     p.stdin.close()
 214
 215     transformed_signature = p.stdout.read().decode("utf-8").strip()
 216     if p.wait() != 0:
 217         raise Exception("js failed to execute: %d" % p.returncode)
 218
 219     return transformed_signature
 220
 221 def get_best_video(player_config):
 222     url_data_list = player_config["args"]["url_encoded_fmt_stream_map"].split(",")
 223     js_url = player_config["assets"]["js"]
 224
 225     best_url = None
 226     best_quality = None
 227     best_extension = None
 228     for url_data in url_data_list:
 229         url_data = urllib.parse.parse_qs(url_data)
 230         mimetype = url_data["type"][0].split(";")[0]
 231         quality = url_data["quality"][0]
 232
 233         if "stereo3d" in url_data:
 234             continue
 235         if quality not in QUALITIES:
 236             continue
 237         if mimetype not in MIMETYPES:
 238             continue
 239
 240         extension = MIMETYPES[mimetype]
 241         quality = QUALITIES.get(quality, -1)
 242
 243         if best_quality is not None and quality < best_quality:
 244             continue
 245
 246         video_url = url_data["url"][0]
 247         if "sig" in url_data:
 248             signature = url_data["sig"][0]
 249         elif "s" in url_data:
 250             signature = decode_signature(js_url, url_data["s"][0])
 251         else:
 252             signature = None
 253
 254         if signature:
 255             video_url = append_to_qs(video_url, {"signature": signature})
 256
 257         best_url = video_url
 258         best_quality = quality
 259         best_extension = extension
 260
 261     return best_url, best_extension
 262
 263 def sanitize_filename(filename):
 264     return (
 265         re.sub("\s+", " ", filename.strip())
 266         .replace("\\", "-")
 267         .replace("/", "-")
 268         .replace("\0", " ")
 269     )
 270
 271 def get_video_url(page):
 272     player_config = get_player_config(page.scripts)
 273     if not player_config:
 274         raise VideoUnavailable(page.unavailable_message or "Could not find video URL")
 275
 276     video_url, extension = get_best_video(player_config)
 277     if not video_url:
 278         return None, None
 279
 280     filename = sanitize_filename(page.title)
 281     filename += "." + extension
 282
 283     return video_url, filename
 284
 285 class YouTubeVideoPageParser(html.parser.HTMLParser):
 286     def __init__(self):
 287         super().__init__()
 288         self.title = None
 289         self.unavailable_message = None
 290         self.scripts = []
 291
 292     def handle_starttag(self, tag, attrs):
 293         attrs = dict(attrs)
 294         self._handle_title(tag, attrs)
 295         self._handle_unavailable_message(tag, attrs)
 296         self._handle_script(tag, attrs)
 297
 298     def handle_endtag(self, tag):
 299         self.handle_data = self._ignore_data
 300
 301     def _ignore_data(self, _):
 302         pass
 303
 304     def _handle_title(self, tag, attrs):
 305         if tag == "title":
 306             self.handle_data = self._handle_title_data
 307
 308     def _handle_title_data(self, data):
 309         self.title = data.strip()
 310
 311     def _handle_unavailable_message(self, tag, attrs):
 312         if attrs.get("id", None) == "unavailable-message":
 313             self.handle_data = self._handle_unavailable_message_data
 314
 315     def _handle_unavailable_message_data(self, data):
 316         self.unavailable_message = data.strip()
 317
 318     def _handle_script(self, tag, attrs):
 319         if tag == "script":
 320             self.handle_data = self._handle_script_data
 321
 322     def _handle_script_data(self, data):
 323         if data:
 324             self.scripts.append(data)
 325
 326 def write_video(filename, video_data):
 327     quoted_filename = urllib.parse.quote(filename.encode("utf-8"))
 328     sys.stdout.buffer.write(
 329         b"Content-Disposition: attachment; filename*=UTF-8''{0}\r\n"
 330         .replace(b"{0}", quoted_filename.encode("utf-8"))
 331     )
 332     sys.stdout.buffer.write(
 333         b"Content-Length: {0}\r\n"
 334         .replace(b"{0}", video_data.getheader("Content-Length").encode("utf-8"))
 335     )
 336     sys.stdout.buffer.write(b"\r\n")
 337     shutil.copyfileobj(video_data, sys.stdout.buffer)
 338     video_data.close()
 339
 340 def cgimain():
 341     args = cgi.parse()
 342     try:
 343         url = args["url"][0]
 344     except:
 345         print_form(url="http://www.youtube.com/watch?v=FOOBAR")
 346         return
 347
 348     try:
 349         page = YouTubeVideoPageParser()
 350         parse_url(url, page)
 351         video_url, filename = get_video_url(page)
 352         video_data = urlopen(video_url)
 353     except VideoUnavailable as e:
 354         print_form(
 355             url=url,
 356             msg="<p class='error'>Sorry, there was an error: %s</p>" % cgi.escape(e.args[0])
 357         )
 358     except Exception as e:
 359         print_form(
 360             url=url,
 361             msg="<p class='error'>Sorry, there was an error. Check your URL?</p>"
 362         )
 363         return
 364
 365     write_video(filename, video_data)
 366
 367 def pp_size(size):
 368     suffixes = ["", "KiB", "MiB", "GiB"]
 369     for i, suffix in enumerate(suffixes):
 370         if size < 1024:
 371             break
 372         size /= 1024
 373     return "%.2f %s" % (size, suffix)
 374
 375 def copy_with_progress(content_length, infile, outfile):
 376     def print_status():
 377         rate = 0
 378         if now != last_ts:
 379             rate = last_bytes_read / (now - last_ts)
 380         sys.stdout.write("\33[2K\r")
 381         sys.stdout.write("%s / %s (%s/sec)" % (
 382             pp_size(bytes_read),
 383             pp_size(content_length),
 384             pp_size(rate),
 385         ))
 386         sys.stdout.flush()
 387
 388     last_ts = 0
 389     last_bytes_read = 0
 390     bytes_read = 0
 391     while True:
 392         now = time.time()
 393         if now - last_ts > 0.5:
 394             print_status()
 395             last_ts = now
 396             last_bytes_read = 0
 397
 398         buf = infile.read(32768)
 399         if not buf:
 400             break
 401         outfile.write(buf)
 402         last_bytes_read += len(buf)
 403         bytes_read += len(buf)
 404
 405     # Newline at the end
 406     print_status()
 407     print()
 408
 409 def main():
 410     try:
 411         url = sys.argv[1]
 412     except:
 413         print("Usage: %s http://youtube.com/watch?v=FOOBAR" % sys.argv[0], file=sys.stderr)
 414         sys.exit(1)
 415
 416     page = YouTubeVideoPageParser()
 417     parse_url(url, page)
 418     video_url, filename = get_video_url(page)
 419     print("Downloading", filename)
 420
 421     outfile = open(filename, "ab")
 422     offset = outfile.tell()
 423     if offset > 0:
 424         print("Resuming download from", pp_size(offset))
 425     total_size = None
 426
 427     while True:
 428         try:
 429             video_data = urlopen(video_url, offset)
 430         except urllib.error.HTTPError as e:
 431             if e.code == 416:
 432                 print("File is complete!")
 433                 break
 434             else:
 435                 raise
 436
 437         content_length = int(video_data.getheader("Content-Length"))
 438         if total_size is None:
 439             total_size = content_length
 440
 441         try:
 442             copy_with_progress(content_length, video_data, outfile)
 443         except IOError as e:
 444             print()
 445
 446         video_data.close()
 447         if outfile.tell() != total_size:
 448             old_offset = offset
 449             offset = outfile.tell()
 450             if old_offset == offset:
 451                 time.sleep(1)
 452             print("Restarting download from", pp_size(offset))
 453         else:
 454             break
 455
 456     outfile.close()
 457
 458
 459 if __name__ == "__main__":
 460     if "SCRIPT_NAME" in os.environ:
 461         cgimain()
 462     else:
 463         try:
 464             main()
 465         except KeyboardInterrupt:
 466             print("\nExiting...")
 467             sys.exit(1)
 468