From ca1e60b138bb13485e202e7c65f2f01f695ed019 Mon Sep 17 00:00:00 2001
From: girst <tobi@isticktoit.net>
Date: Thu, 17 Jun 2021 15:06:03 +0200
Subject: [PATCH] move to new api again, use new api for age-gate bypass

get_video_info?html5=1 is now also being shut down, so we do really need
to move to the new innertube api. :/
---
 app/common/common.py    | 43 ++++++++++++++++++++---------------------
 app/youtube/__init__.py |  1 +
 2 files changed, 22 insertions(+), 22 deletions(-)

diff --git a/app/common/common.py b/app/common/common.py
index acf3c5c..d0989ce 100644
--- a/app/common/common.py
+++ b/app/common/common.py
@@ -201,10 +201,10 @@ def update_channel(db, xmldata, from_webhook=False):
 
     return True
 
-def get_video_info(video_id, sts=0, algo=""):
+def get_video_info(video_id, sts=0, algo="", _embed=False):
     """
     returns: best-quality muxed video stream, stream map, player_response, error-type/mesage
-    error types: player, malformed, livestream, geolocked, exhausted
+    error types: player, malformed, livestream, geolocked, agegated, no-url, exhausted
     """
     player_error, metadata = None, None # for 'exhausted'
     with sqlite3.connect(cf['global']['database']) as conn:
@@ -214,14 +214,14 @@ def get_video_info(video_id, sts=0, algo=""):
     today = datetime.now(timezone.utc).strftime("%Y%m%d")
     # XXX: anticaptcha hasn't been adapted
     # XXX: this is not cached any more!
-    # XXX: age-gated now broken: HtVdAasjOgU (embed ok), XgnwCQzjau8 (no embed)
+    # note: age-gated works as long as it's embeddable (HtVdAasjOgU ok, XgnwCQzjau8 bad)
     r = requests.post("https://www.youtube-nocookie.com/youtubei/v1/player?key=AIzaSyAO_FJ2SlqU8Q4STEHLGCilw_Y9_11qcW8", json={
         'videoId': video_id,
         'context': {
             'client': {
                 'gl': 'US',
                 'hl': 'en',
-                'clientName': 'WEB',
+                'clientName': 'WEB_EMBEDDED_PLAYER' if _embed else 'WEB',
                 'clientVersion': f'2.{today}.01.01',
             }
         },
@@ -232,30 +232,29 @@ def get_video_info(video_id, sts=0, algo=""):
         return None, None, None, 'banned', 'possible IP ban'
 
     metadata = r.json()
+    if "error" in metadata:
+        return None, None, metadata, "malformed", metadata.get("error",{}).get("message","")
     playabilityStatus = metadata['playabilityStatus']['status']
     if playabilityStatus != "OK":
         playabilityReason = metadata['playabilityStatus'].get('reason',
                 '//'.join(metadata['playabilityStatus'].get('messages',[])))
         player_error = f"{playabilityStatus}: {playabilityReason}"
         #if playabilityStatus == "UNPLAYABLE": XXX: do we need that still?
-        if playabilityStatus == "LOGIN_REQUIRED" and metadata['playabilityStatus'].get('reason') == "Sign in to confirm your age" and sts != 0:
-            r = requests.get("https://www.youtube.com/get_video_info?html5=1&video_id="+video_id, {
-                "video_id": video_id,
-                "eurl": f"https://youtube.googleapis.com/v/{video_id}",
-                "el": "embedded",
-                "sts": sts,
-                "hl": "en_US",
-            })
-            params = parse_qs(r.text)
-            if 'errorcode' in params: # status=fail
-                return None, None, None, 'malformed', params['reason'][0]
-            from flask import current_app
-            current_app.logger.error(r.text)
-            metadata = json.loads(params.get('player_response')[0])
-
-        # without videoDetails, there's only the error message
-        maybe_metadata = metadata if 'videoDetails' in metadata else None
-        return None, None, maybe_metadata, 'player', player_error
+        if (playabilityStatus == "LOGIN_REQUIRED"
+                and "confirm your age" in metadata['playabilityStatus'].get('reason')
+                and sts != 0 # only need metadata when no sts (via pubsubhubbub)
+                and not _embed
+        ):
+            _, _, metadata_embed, error_embed, _ = get_video_info(video_id, sts, algo, True)
+            if not error_embed:
+                metadata['streamingData'] = metadata_embed['streamingData']
+                metadata['playabilityStatus'] = metadata_embed['playabilityStatus']
+            else:
+                return None, None, metadata, 'agegated', player_error
+        else:
+            # without videoDetails, there's only the error message
+            maybe_metadata = metadata if 'videoDetails' in metadata else None
+            return None, None, maybe_metadata, 'player', player_error
 
     # livestreams have no adaptive/muxed formats:
     is_live = metadata['videoDetails'].get('isLive', False)
diff --git a/app/youtube/__init__.py b/app/youtube/__init__.py
index 46b1475..52d8e28 100644
--- a/app/youtube/__init__.py
+++ b/app/youtube/__init__.py
@@ -85,6 +85,7 @@ def watch():
         'malformed':  "Video ID is invalid.",
         'geolocked':  "This video is geolocked.",
         'livestream': "Livestreams not supported on this instance.",
+        'agegated':   "Unable to bypass age-restriction.",
         'exhausted':  errdetails or "Couldn't extract video URLs.",
         'player':     errdetails,
     }.get(error)
-- 
2.39.3