return script[len(PREFIX):-len(SUFFIX)]
-def find_func_name(script):
+def find_cipher_func(script):
FUNC_NAME = R"([a-zA-Z0-9$]+)"
+ DECODE_URI_COMPONENT = R"(\(decodeURIComponent)?"
FUNC_PARAMS = R"(\([a-zA-Z,\.]+\.s\))"
TERMINATOR = R"[,;\)]"
- PATTERN = FUNC_NAME + FUNC_PARAMS + TERMINATOR
+ PATTERN = FUNC_NAME + DECODE_URI_COMPONENT + FUNC_PARAMS + TERMINATOR
match = re.search(PATTERN, script)
func_name = match.groups()[0]
return func_name
-def decode_signature(js_url, signature):
+def find_url_func(script):
+ FUNC_NAME = R"([a-zA-Z0-9$]+)"
+ PATTERN = R"this\.url\s*=\s*" + FUNC_NAME + R"\s*\(\s*this\s*\)"
+
+ match = re.search(PATTERN, script)
+ func_name = match.groups()[0]
+ return func_name
+
+def decode_cipher_url(js_url, cipher):
+ cipher = urllib.parse.parse_qs(cipher)
+ args = [
+ cipher["url"][0],
+ cipher["sp"][0],
+ cipher["s"][0],
+ ]
+
f = urlopen(js_url)
script = f.read().decode("utf-8")
f.close()
- func_name = find_func_name(script)
+ cipher_func_name = find_cipher_func(script)
+ url_func_name = find_url_func(script)
params = {
- "func_name": func_name,
- "signature": json.dumps(signature),
+ "cipher_func_name": cipher_func_name,
+ "url_func_name": url_func_name,
+ "args": json.dumps(args),
"code": json.dumps(extract_js(script)),
}
p = subprocess.Popen(
js_decode_script = ("""
const vm = require('vm');
- const sandbox = {
- location: {
- hash: '',
- href: '',
- protocol: 'http:'
- },
- history: {
- pushState: function(){}
- },
- document: {},
- navigator: {
- userAgent: ''
- },
- XMLHttpRequest: class XMLHttpRequest {},
- matchMedia: () => ({matches: () => {}, media: ''}),
- signature: %(signature)s,
- transformed_signature: null,
- g: function(){} // this is _yt_player
+ const fakeGlobal = {};
+ fakeGlobal.window = fakeGlobal;
+ fakeGlobal.location = {
+ hash: '',
+ host: 'www.youtube.com',
+ hostname: 'www.youtube.com',
+ href: 'https://www.youtube.com',
+ origin: 'https://www.youtube.com',
+ pathname: '/',
+ protocol: 'https:'
+ };
+ fakeGlobal.history = {
+ pushState: function(){}
};
- sandbox.window = sandbox;
+ fakeGlobal.document = {
+ location: fakeGlobal.location
+ };
+ fakeGlobal.document = {};
+ fakeGlobal.navigator = {
+ userAgent: ''
+ };
+ fakeGlobal.XMLHttpRequest = class XMLHttpRequest {};
+ fakeGlobal.matchMedia = () => ({matches: () => {}, media: ''});
+ fakeGlobal.result_url = null;
+ fakeGlobal.g = function(){}; // this is _yt_player
const code_string = %(code)s + ';';
- const exec_string = 'transformed_signature = %(func_name)s("", "MARKER", signature);';
- vm.runInNewContext(code_string + exec_string, sandbox);
-
- function findSignature(obj) {
- if (typeof obj !== 'object') {
- return;
- }
- for (const [key, value] of Object.entries(obj)) {
- if (key === 'MARKER') {
- return value;
- }
- const result = findSignature(value);
- if (result) {
- return result;
- }
- }
- }
- console.log(findSignature(sandbox.transformed_signature));
+ const exec_string = 'result_url = %(url_func_name)s(%(cipher_func_name)s(...%(args)s));';
+ vm.runInNewContext(code_string + exec_string, fakeGlobal);
+
+ console.log(fakeGlobal.result_url);
""" % params)
p.stdin.write(js_decode_script.encode("utf-8"))
p.stdin.close()
- transformed_signature = p.stdout.read().decode("utf-8").strip()
- transformed_signature = urllib.parse.unquote(transformed_signature)
+ result_url = p.stdout.read().decode("utf-8").strip()
if p.wait() != 0:
raise Exception("js failed to execute: %d" % p.returncode)
- return transformed_signature
+ return result_url
def get_best_video(player_config):
- url_data_list = player_config["args"]["url_encoded_fmt_stream_map"].split(",")
js_url = player_config["assets"]["js"]
+ player_args = player_config["args"]
+ player_response = json.loads(player_args["player_response"])
+ formats = player_response["streamingData"]["formats"]
+
best_url = None
best_quality = None
best_extension = None
- for url_data in url_data_list:
- url_data = urllib.parse.parse_qs(url_data)
- mimetype = url_data["type"][0].split(";")[0]
- quality = url_data["quality"][0]
+ for format_data in formats:
+ mimetype = format_data["mimeType"].split(";")[0]
+ quality = format_data["quality"]
- if "stereo3d" in url_data:
- continue
if quality not in QUALITIES:
continue
if mimetype not in MIMETYPES:
if best_quality is not None and quality < best_quality:
continue
- video_url = url_data["url"][0]
- if "sig" in url_data:
- signature = url_data["sig"][0]
- elif "s" in url_data:
- signature = decode_signature(js_url, url_data["s"][0])
+ if "signatureCipher" in format_data:
+ video_url = decode_cipher_url(js_url, format_data["signatureCipher"])
else:
- signature = None
-
- if signature:
- sp = url_data.get("sp", ["signature"])[0]
- video_url = append_to_qs(video_url, {sp: signature})
+ video_url = format_data["url"]
best_url = video_url
best_quality = quality
if not video_url:
return None, None
- filename = sanitize_filename(player_config["args"]["title"])
- filename += "." + extension
+ title = player_config["args"].get("title", None)
+ if not title:
+ title = json.loads(player_config["args"]["player_response"])["videoDetails"]["title"]
+ if not title:
+ title = "Unknown title"
+
+ filename = sanitize_filename(title) + "." + extension
return video_url, filename