]> code.delx.au - youtube-cgi/blobdiff - youtube.cgi
fixes for youtube changes
[youtube-cgi] / youtube.cgi
index f5845a4e9e95d424abc972895e94e11994f1bf86..6a5a92ee2b59f8c061d7ebbf7b658d6c26d59302 100755 (executable)
@@ -1,10 +1,9 @@
-#!/usr/bin/env python
+#!/usr/bin/python2
 
 from __future__ import division
 
 import cookielib
 import cgi
-import itertools
 import json
 from lxml import html
 import os
@@ -80,6 +79,9 @@ urlopener = urllib2.build_opener(urllib2.HTTPCookieProcessor(cookiejar))
 referrer = ""
 
 def urlopen(url, offset=None):
+       if url.startswith("//"):
+               url = "http:" + url
+
        global referrer
        req = urllib2.Request(url)
        if referrer:
@@ -134,6 +136,11 @@ def get_player_config(doc):
                                p2 = line.rfind(";")
                                if p1 >= 0 and p2 > 0:
                                        return json.loads(line[p1+1:p2])
+                       if "ytplayer.config =" in line:
+                               p1 = line.find("ytplayer.config =")
+                               p2 = line.rfind(";")
+                               if p1 >= 0 and p2 > 0:
+                                       return json.loads(line[p1+18:p2])
                        if "'PLAYER_CONFIG': " in line:
                                p1 = line.find(":")
                                if p1 >= 0:
@@ -141,29 +148,75 @@ def get_player_config(doc):
                                        convert_from_old_itag(player_config)
                                        return player_config
 
-def get_best_video(player_config):
-       url_data = urlparse.parse_qs(player_config["args"]["url_encoded_fmt_stream_map"])
-       url_data = itertools.izip_longest(
-               url_data["url"],
-               url_data["type"],
-               url_data["quality"],
-               url_data.get("sig", []),
+def extract_function(output, script, func_name):
+       p1 = script.find("function " + func_name + "(")
+       p2 = script.find("}", p1)
+       code = script[p1:p2+1]
+       output.append(code)
+       deps = re.findall(R"[^\.][= ]([\$0-9a-zA-Z]+)\(", code)
+       deps = set(deps)
+       deps.remove(func_name)
+       for dep in deps:
+               extract_function(output, script, dep)
+
+def decode_signature(js_url, s):
+       script = urlopen(js_url).read()
+       func_name = re.search(R"\b([a-zA-Z]+)\([a-zA-Z]+\.s\);", script).groups()[0]
+
+       codes = []
+       extract_function(codes, script, func_name)
+
+       p = subprocess.Popen(
+               "js",
+               shell=True,
+               close_fds=True,
+               stdin=subprocess.PIPE,
+               stdout=subprocess.PIPE
        )
+       for code in codes:
+               p.stdin.write(code + "\n")
+       p.stdin.write("console.log(%s('%s'));\n" % (func_name, s))
+       p.stdin.close()
+
+       signature = p.stdout.read().strip()
+       if p.wait() != 0:
+               raise Exception("js failed to execute: %d" % p.returncode)
+
+       return signature
+
+def get_best_video(player_config):
+       url_data_list = player_config["args"]["url_encoded_fmt_stream_map"].split(",")
+       js_url = player_config["assets"]["js"]
+
        best_url = None
        best_quality = None
        best_extension = None
-       for video_url, mimetype, quality, signature in url_data:
-               mimetype = mimetype.split(";")[0]
+       for url_data in url_data_list:
+               url_data = urlparse.parse_qs(url_data)
+               mimetype = url_data["type"][0].split(";")[0]
+               quality = url_data["quality"][0]
+
+               if quality not in QUALITIES:
+                       continue
                if mimetype not in MIMETYPES:
                        continue
+
                extension = MIMETYPES[mimetype]
-               quality = QUALITIES.get(quality.split(",")[0], -1)
-               if best_quality is None or quality > best_quality:
-                       if signature:
-                               video_url = append_to_qs(video_url, {"signature": signature})
-                       best_url = video_url
-                       best_quality = quality
-                       best_extension = extension
+               quality = QUALITIES.get(quality, -1)
+
+               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]
+               else:
+                       signature = decode_signature(js_url, url_data["s"][0])
+               video_url = append_to_qs(video_url, {"signature": signature})
+
+               best_url = video_url
+               best_quality = quality
+               best_extension = extension
 
        return best_url, best_extension
 
@@ -320,7 +373,7 @@ def main():
 
 
 if __name__ == "__main__":
-       resource.setrlimit(resource.RLIMIT_AS, (MAX_MEMORY_BYTES, MAX_MEMORY_BYTES))
+###    resource.setrlimit(resource.RLIMIT_AS, (MAX_MEMORY_BYTES, MAX_MEMORY_BYTES))
        if os.environ.has_key("SCRIPT_NAME"):
                cgimain()
        else: