more convenient startup of node.js

[youtube-cgi] / youtube.cgi

diff --git a/youtube.cgi b/youtube.cgi
index 8b103bee7847de97bbaa5ae68350d6c98e358e76..45c8574965dd00947d769c381400fdd14bc854b9 100755 (executable)
--- a/youtube.cgi
+++ b/youtube.cgi
@@ -4,7 +4,6 @@ from __future__ import division
 
 import cookielib
 import cgi
-import itertools
 import json
 from lxml import html
 import os
@@ -134,6 +133,11 @@ 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 +145,63 @@ 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 decode_signature(js_url, s):
+       script = urlopen(js_url).read()
+       func_name = re.search(R"\b([a-z]+)\([a-z]+\.s\);", script).groups()[0]
+       p1 = script.find("function " + func_name)
+       p2 = script.find("}", p1)
+       func_block = script[p1:p2+1]
+
+       p = subprocess.Popen(
+               "js",
+               shell=True,
+               close_fds=True,
+               stdin=subprocess.PIPE,
+               stdout=subprocess.PIPE
        )
+       p.stdin.write(func_block + "\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
 
@@ -238,27 +276,32 @@ def pp_size(size):
 
 def copy_with_progress(content_length, infile, outfile):
        def print_status():
+               rate = 0
+               if now != last_ts:
+                       rate = last_bytes_read / (now - last_ts)
                sys.stdout.write("\33[2K\r")
                sys.stdout.write("%s / %s (%s/sec)" % (
                        pp_size(bytes_read),
                        pp_size(content_length),
-                       pp_size(bytes_read / (now - start_ts)),
+                       pp_size(rate),
                ))
                sys.stdout.flush()
 
-       start_ts = time.time()
        last_ts = 0
+       last_bytes_read = 0
        bytes_read = 0
        while True:
                now = time.time()
                if now - last_ts > 0.5:
-                       last_ts = now
                        print_status()
+                       last_ts = now
+                       last_bytes_read = 0
 
                buf = infile.read(32768)
                if not buf:
                        break
                outfile.write(buf)
+               last_bytes_read += len(buf)
                bytes_read += len(buf)
 
        # Newline at the end
@@ -315,7 +358,7 @@ 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: