Forgot admin imports

README.md

@@ -48,7 +48,7 @@ Extra functionality for further development of features.
 - [x] Video reporting functionality
 - [x] Ability (for external applications) to queue up video ids for download
 - [x] Add websub requesting and receiving ability. (not fully usable yet without celery tasks)
-- [] OIDC or Webauthn logins instead of static argon2 passwords
+- [x] OIDC or Webauthn logins instead of static argon2 passwords
 
 ### Stage 3
 Mainly focused on retiring the cronjob based scripts and moving it to celery based tasks

ayta/__init__.py

@@ -9,15 +9,24 @@ def create_app(test_config=None):
     config = {'MONGO_CONNECTION': os.environ.get('AYTA_MONGOCONNECTION', 'mongodb://root:example@192.168.66.140:27017'),
               'OIDC_PROVIDER': os.environ.get('AYTA_OIDC_PROVIDER', 'https://auth.ventilaar.nl'),
               'OIDC_ID': os.environ.get('AYTA_OIDC_ID', 'ayta'),
-              'CACHE_TYPE': os.environ.get('AYTA_CACHETYPE', 'SimpleCache'),
               'CACHE_DEFAULT_TIMEOUT': int(os.environ.get('AYTA_CACHETIMEOUT', 6)),
-              'SECRET_KEY': os.environ.get('AYTA_SECRETKEY', secrets.token_hex(32)),
               'DEBUG': bool(os.environ.get('AYTA_DEBUG', False)),
               'DOMAIN': os.environ.get('AYTA_DOMAIN', 'https://testing.mashallah.nl'),
-              'CELERY': dict(broker_url=str(os.environ.get('AYTA_CELERYBROKER', 'amqp://guest:guest@192.168.66.140:5672/')), 
+              'CELERY': {'broker_url': str(os.environ.get('AYTA_CELERYBROKER', 'amqp://guest:guest@192.168.66.140:5672/'))}
-                             task_ignore_result=True,)
              }
     
+    # Static Flask configuration options
+    
+    config['CELERY']['task_ignore_result'] = True
+    config['CACHE_TYPE'] = 'SimpleCache'
+    config['SECRET_KEY'] = secrets.token_bytes(32)
+    
+    # Celery Periodic tasks
+    
+    config['CELERY']['beat_schedule'] = {}
+    config['CELERY']['beat_schedule']['Renew WebSub endpoints'] = {'task': 'ayta.tasks.websub_renew_expiring', 'schedule': 4000}
+    #config['CELERY']['beat_schedule']['Process WebSub data'] = {'task': 'ayta.tasks.websub_process_data', 'schedule': 6}
+    
     app = Flask(__name__)
     app.config.from_mapping(config)
     

ayta/blueprints/admin.py

@@ -2,7 +2,7 @@ from flask import Blueprint, render_template, request, redirect, url_for, flash
 from ..nosql import get_nosql
 from ..dlp import checkChannelId, getChannelInfo
 from ..decorators import login_required
-from ..tasks import subscribe_websub_callback, unsubscribe_websub_callback
+from ..tasks import websub_subscribe_callback, websub_unsubscribe_callback
 from datetime import datetime
 from secrets import token_urlsafe
 
@@ -71,7 +71,7 @@ def channel(channelId):
         value = request.form.get('value', None)
         
         if task == 'subscribe-websub':
-            task = subscribe_websub_callback.delay(channelId)
+            task = websub_subscribe_callback.delay(channelId)
             flash(f"Started task {task.id}")
             return redirect(url_for('admin.channel', channelId=channelId))
         
@@ -114,9 +114,7 @@ def websub():
         value = request.form.get('value', None)
         
         if task == 'unsubscribe':
-            channelId = get_nosql().websub_getCallback(value).get('channel')
+            task = websub_unsubscribe_callback.delay(value)
-            
-            task = unsubscribe_websub_callback.delay(value, channelId)
             
             flash(f"Started task {task.id}")
             return redirect(url_for('admin.websub'))

ayta/blueprints/api.py

@@ -33,7 +33,7 @@ def websub(cap):
         return challenge
     
     if get_nosql().websub_existsCallback(cap):
-        if not get_nosql().websub_savePost(cap, str(request.data)):
+        if not get_nosql().websub_savePost(cap, request.data):
             return abort(500)
         return '', 202
     

ayta/blueprints/channel.py

@@ -34,7 +34,7 @@ def channel(channelId):
     for videoId in videoIds:
         videos.append(get_nosql().get_video_info(videoId, limited=True))
         
-    videos = sorted(videos, key=lambda x: x.get('upload_date'), reverse=True)
+    videos = sorted(videos, key=lambda x: x.get('upload_date', '19700101'), reverse=True)
 
     return render_template('channel/channel.html', channel=channelInfo, videos=videos)
     

ayta/blueprints/watch.py

@@ -36,4 +36,9 @@ def base():
 
     render['info'] = get_nosql().get_video_info(vGet)
     render['params'] = request.args.get('v')
+    
+    if render['info']['_status'] != 'available':
+        flash(render['info'].get('_status_description', 'Video unavailable because of technical errors. Come back later.'))
+        return redirect(url_for('index.base'))
+        
     return render_template('watch/index.html', render=render)

ayta/nosql.py

@@ -205,7 +205,7 @@ class Mango:
     
     def get_recent_videos(self, count=99):
         """ Returns a SET of YouTube video ID's which have been added last to the info_json collection """
-        result = self.info_json.find({}, {'_id': 0, 'id': 1}, sort=[('_id', pymongo.DESCENDING)]).limit(count)
+        result = self.info_json.find({'_status': 'available'}, {'_id': 0, 'id': 1}, sort=[('_id', pymongo.DESCENDING)]).limit(count)
 
         ids = []
         
@@ -216,7 +216,7 @@ class Mango:
         
     def get_video_info(self, videoId, limited=False):
         if limited:
-            projection = {'_id': 1, 'id': 1, 'title': 1, 'upload_date': 1, 'description': 1, 'channel_id': 1, 'uploader': 1, 'epoch': 1, 'title_slug': 1}
+            projection = {'_id': 1, 'id': 1, 'title': 1, 'upload_date': 1, 'description': 1, 'channel_id': 1, 'uploader': 1, 'epoch': 1, '_title_slug': 1}
         else:
             projection = {}
         
@@ -294,21 +294,24 @@ class Mango:
         status = status.get('status')
         
         if status in ['requesting']:
-            self.websub_callbacks.update_one({'id': callbackId}, {'$set': {'status': 'active', 'activation_time': current_time(object=True), 'lease': lease}})
+            self.websub_callbacks.update_one({'id': callbackId}, {'$set': {'status': 'active', 'activation_time': current_time(object=True), 'lease': int(lease)}})
             return True
         
         return False
 
-    def websub_existsCallback(self, callbackId):
+    def websub_existsCallback(self, callbackId, channel=False):
-        status = self.websub_callbacks.find_one({'id': callbackId}, {'status': 1})
+        if channel:
+            query = {'channel': callbackId}
+        else:
+            query = {'id': callbackId}
+            
+        status = self.websub_callbacks.find_one(query, {'id': 1, 'status': 1})
 
         if not status:
             return False
         
-        status = status.get('status')
+        if status.get('status') in ['requesting', 'active', 'retiring']:
-        
+            return status.get('id')
-        if status in ['requesting', 'active', 'retiring']:
-            return True
             
         return False
         
@@ -347,10 +350,24 @@ class Mango:
     def websub_savePost(self, callbackId, data):
         return self.websub_data.insert_one({'callback_id': callbackId, 'state': 'unprocessed', 'received_time': current_time(object=True), 'raw_data': data}).inserted_id
         
+    def websub_getFirstPostData(self):
+        data = self.websub_data.find_one({'state': 'unprocessed'}, {'_id': 1, 'raw_data': 1}, sort=[('received_time', 1)])
+        
+        if not data:
+            return None
+        
+        self.websub_data.update_one({'_id': data['_id']}, {'$set': {'state': 'processing'}})
+        
+        return (data.get('_id'), data.get('raw_data'))
+        
+    def websub_deletePostProcessing(self, _id):
+        self.websub_data.delete_one({'_id': _id})
+        
     def websub_cleanRetired(self, days=3):
         days = self.datetime.utcnow() - self.timedelta(days=days)
 
         self.websub_callbacks.delete_many({'status': 'retired', 'retired_time': {'$lt': days}})
+        self.websub_callbacks.delete_many({'status': 'requesting', 'requesting_time': {'$lt': days}})
         
         return True
 

ayta/tasks.py

@@ -1,22 +1,32 @@
 from celery import shared_task
 from flask import current_app
 
+##########################################
+#              CELERY TASKS              #
+##########################################
+
 @shared_task()
-def subscribe_websub_callback(channelId):
+def websub_subscribe_callback(channelId):
     import requests
     from .nosql import get_nosql
     
+    # check if a callback already exists for channel
+    answer = get_nosql().websub_existsCallback(channelId, channel=True)
+    
+    if not answer:
         callbackId = get_nosql().websub_newCallback(channelId)
+    else:
+        callbackId = answer
     
     url = 'https://pubsubhubbub.appspot.com/subscribe'
     data = {
-        'hub.callback': f'https://{current_app.config["DOMAIN"]}/api/websub//{callbackId}',
+        'hub.callback': f'{current_app.config["DOMAIN"]}/api/websub/{callbackId}',
         'hub.topic': f'https://www.youtube.com/xml/feeds/videos.xml?channel_id={channelId}',
         'hub.verify': 'async',
         'hub.mode': 'subscribe',
         'hub.verify_token': '',
         'hub.secret': '',
-        'hub.lease_numbers': '86400',
+        'hub.lease_numbers': '432000',
     }
     
     get_nosql().websub_requestingCallback(callbackId)
@@ -27,12 +37,19 @@ def subscribe_websub_callback(channelId):
     return False
     
 @shared_task()
-def unsubscribe_websub_callback(callbackId, channelId):
+def websub_unsubscribe_callback(callbackId):
     import requests
     from .nosql import get_nosql
     
+    answer = get_nosql().websub_existsCallback(callbackId)
+    
+    if not answer:
+        return False
+        
+    channelId = get_nosql().websub_getCallback(callbackId).get('channel')
+    
     url = 'https://pubsubhubbub.appspot.com/subscribe'
-    data = {'hub.callback': f'https://{current_app.config["DOMAIN"]}/api/websub/{callbackId}',
+    data = {'hub.callback': f'{current_app.config["DOMAIN"]}/api/websub/{callbackId}',
             'hub.topic': f'https://www.youtube.com/xml/feeds/videos.xml?channel_id={channelId}',
             'hub.verify': 'async',
             'hub.mode': 'unsubscribe'
@@ -45,3 +62,78 @@ def unsubscribe_websub_callback(callbackId, channelId):
         return True
         
     return False
+    
+@shared_task()
+def websub_process_data():
+    from .nosql import get_nosql
+    
+    while True:
+        data = get_nosql().websub_getFirstPostData()
+        if not data:
+            break
+        
+        _id, data = data
+        
+        parsed = do_parse_data(data)
+        if not parsed:
+            get_nosql().websub_deletePostProcessing(_id)
+                
+        state, channelId, videoId = parsed
+
+        get_nosql().websub_deletePostProcessing(_id)
+
+@shared_task()
+def websub_renew_expiring(hours=6):
+    from .nosql import get_nosql
+    from datetime import datetime, timedelta
+
+    for callbackId in get_nosql().websub_getCallbacks():
+        data = get_nosql().websub_getCallback(callbackId)
+        
+        pivot = datetime.utcnow() - timedelta(hours=hours)
+        expires = data.get('activation_time') + timedelta(seconds=data.get('lease'))
+        
+        if pivot <= expires:  # if expiration happens after the calculation time pass the loop
+            continue
+        
+        print(f'{callbackId} should be renewed')
+        
+        websub_subscribe_callback.delay(data.get('channel'))
+
+##########################################
+#              TASK MODULES              #
+##########################################
+
+def do_parse_data(data):
+    import xml.etree.ElementTree as ET
+    
+    data = data.decode('utf-8')
+
+    try:
+        root = ET.fromstring(data)
+    except ET.ParseError:
+        print('Not XML')
+        return False
+
+    yt = any(child.tag.startswith('{http://www.youtube.com/xml/schemas/2015}') for child in root.iter())
+    at = any(child.tag.startswith('{http://purl.org/atompub/tombstones/1.0}') for child in root.iter())
+
+    if yt and not at:
+        # Video published
+        state = 'added'
+        ns = {'yt': 'http://www.youtube.com/xml/schemas/2015', '': 'http://www.w3.org/2005/Atom'}
+        entry = root.find('.//{http://www.w3.org/2005/Atom}entry')
+        videoId = entry.find('./yt:videoId', ns).text
+        channelId = entry.find('./yt:channelId', ns).text
+    elif not yt and at:
+        # Video hidden
+        state = 'removed'
+        ns = {'at': 'http://purl.org/atompub/tombstones/1.0', '': 'http://www.w3.org/2005/Atom'}
+        deleted_entry = root.find('.//{http://purl.org/atompub/tombstones/1.0}deleted-entry')
+        videoId = deleted_entry.attrib['ref'].split(':')[-1]
+        channelId = deleted_entry.find('./at:by/uri', ns).text.split('/')[-1]
+    else:
+        print('Unknown xml')
+        return False
+        
+    return (state, channelId, videoId)

ayta/templates/admin/websub.html

@@ -50,6 +50,7 @@
 	    {% for callback in callbacks %}
         <tr class="filterable">
 		  <td>
+			<a target="_blank" rel="noopener noreferrer" href="https://pubsubhubbub.appspot.com/subscription-details?hub.callback={{ config['DOMAIN'] }}/api/websub/{{ callbacks[callback].get('id') }}&hub.topic=https://www.youtube.com/xml/feeds/videos.xml?channel_id={{ callbacks[callback].get('channel') }}"><button class="btn-small waves-effect waves-light" title="Information on Pubsubhubbub (external link)">ℹ️</button></a>
 		    <form method="post">
 			  <input type="text" value="{{ callbacks[callback].get('id') }}" name="value" hidden>
 		      <button class="btn-small waves-effect waves-light" type="submit" name="task" value="unsubscribe" title="Send unsubscribe request to hub" {% if callbacks[callback].get('status') != 'active' %}disabled{% endif %}>🗑️</button>

ayta/templates/channel/channel.html

@@ -25,7 +25,7 @@
 	<div class="card medium black-text">
 	  <a href="{{ url_for('watch.base') }}?v={{ video.get('id') }}">
 	    <div class="card-image">
-		  <img loading="lazy" src="https://archive.ventilaar.net/videos/automatic/{{ video.get('channel_id') }}/{{ video.get('id') }}/{{ video.get('title_slug') }}.jpg">
+		  <img loading="lazy" src="https://archive.ventilaar.net/videos/automatic/{{ video.get('channel_id') }}/{{ video.get('id') }}/{{ video.get('_title_slug') }}.jpg">
         </div>
 	  </a>
       <div class="card-content activator">

ayta/templates/channel/orphaned.html

@@ -25,7 +25,7 @@
 	<div class="card medium black-text">
 	  <a href="{{ url_for('watch.base') }}?v={{ video.get('id') }}">
 	    <div class="card-image">
-		  <img loading="lazy" src="https://archive.ventilaar.net/videos/automatic/{{ video.get('channel_id') }}/{{ video.get('id') }}/{{ video.get('title_slug') }}.jpg">
+		  <img loading="lazy" src="https://archive.ventilaar.net/videos/automatic/{{ video.get('channel_id') }}/{{ video.get('id') }}/{{ video.get('_title_slug') }}.jpg">
         </div>
 	  </a>
       <div class="card-content activator">

ayta/templates/channel/recent.html

@@ -25,7 +25,7 @@
 	<div class="card medium black-text">
 	  <a href="{{ url_for('watch.base') }}?v={{ video.get('id') }}">
 	    <div class="card-image">
-		  <img loading="lazy" src="https://archive.ventilaar.net/videos/automatic/{{ video.get('channel_id') }}/{{ video.get('id') }}/{{ video.get('title_slug') }}.jpg">
+		  <img loading="lazy" src="https://archive.ventilaar.net/videos/automatic/{{ video.get('channel_id') }}/{{ video.get('id') }}/{{ video.get('_title_slug') }}.jpg">
         </div>
 	  </a>
       <div class="card-content activator">

ayta/templates/watch/index.html

@@ -14,8 +14,8 @@
 <div class="row">
   <div class="col s12 mt-4 center-align">
     <video controls class="responsive-video">
-    <source src="https://archive.ventilaar.net/videos/automatic/{{ render.get('info').get('channel_id') }}/{{ render.get('info').get('id') }}/{{ render.get('info').get('title_slug') }}.mp4">
+    <source src="https://archive.ventilaar.net/videos/automatic/{{ render.get('info').get('channel_id') }}/{{ render.get('info').get('id') }}/{{ render.get('info').get('_title_slug') }}.mp4">
-    <source src="https://archive.ventilaar.net/videos/automatic/{{ render.get('info').get('channel_id') }}/{{ render.get('info').get('id') }}/{{ render.get('info').get('title_slug') }}.webm">
+    <source src="https://archive.ventilaar.net/videos/automatic/{{ render.get('info').get('channel_id') }}/{{ render.get('info').get('id') }}/{{ render.get('info').get('_title_slug') }}.webm">
     Your browser does not support the video tag.
     </video>
   </div>

requirements.txt

@@ -5,8 +5,7 @@ flask-caching
 flask-limiter
 pymongo
 yt-dlp
-argon2-cffi
 gunicorn
 celery
 sqlalchemy
-pyjwt
+pyjwt[crypto]