hype streaming take two plus better handling of multiple users (it still doesn't work)

.vscode/launch.json

@@ -10,7 +10,7 @@
             "request": "launch",
             "module": "flask",
             "env": {
-                "FLASK_APP": "app.py",
+                "FLASK_APP": "./app/app.py",
                 "FLASK_DEBUG": "1"
             },
             "args": [

app/app.py

@@ -1,62 +1,40 @@
+import logging
 from flask import Flask, render_template, Response, request
-from main import get_auto_transcript, get_video_id, create_and_stream, log, output_stream, awaiter
+from main import yoink, process
-from asyncio import sleep
-from datetime import datetime
-import threading, pytz
 
 app = Flask(__name__, static_folder="website/static", template_folder="website")
 
+# Configure logging
+logging.basicConfig(
+    filename='./logs/app.log',
+    level=logging.INFO,
+    format='%(asctime)s %(levelname)s: %(message)s',
+    datefmt='%Y-%m-%d %H:%M:%S'
+)
+
 @app.route('/')
 def home():
+    logging.info("Home page accessed.")
     return render_template('index.html')
 
 @app.route('/process_url', methods=['POST'])
 def process_url():
-    url = request.form.get('url', '').strip()
+    global most_recent_thread
-    if not url:
+    url = request.form['url']
-        log("No URL provided.\n")
+    logging.info(f"Received URL for processing: {url}")
-        return "No URL provided.", 400
+    success, msg, status_code, most_recent_thread = process(url)
-
+    if success:
-    log(f"\n\n\n## New Entry at {datetime.now(pytz.timezone('America/New_York')).strftime('%Y-%m-%d %H:%M:%S')}\n")
+        logging.info("Processing started successfully.")
-    log(f"URL: {url}\n")
-
-    # Extract the video ID from the URL
-    video_id = get_video_id(url)
-    if not video_id:
-        log(f"Could not parse video id from URL: {url}\n")
-        return "Couldn't parse video ID from URL. (Ensure it's a valid YouTube.com or YouTu.be URL.)", 400
-    log(f"Video ID: {video_id}\n")
-
-    # Get the transcript for that video ID
-    transcript = get_auto_transcript(video_id)
-    if not transcript:
-        log("## Error: Could not retrieve transcript. Assistant won't be called.\n")
-        return "Parsed video ID, but transcript retrieval failed (might be disabled by the video owner).", 400
-
-    # Start the stream thread
-    thread = threading.Thread(target=create_and_stream, args=(transcript,))
-    thread.start()
-    log("Stream preparation complete, sending reply...\n")
         return Response("Processing started. Check /stream_output for updates.", content_type='text/plain', status=200)
+    else:
+        logging.error(f"Processing failed: {msg}")
+        return Response(msg, content_type='text/plain', status=status_code)
 
 @app.route('/stream_output')
-def stream_output_route():
+def stream_output():
-    def generate():
+    logging.info("Streaming output requested.")
-        log("<details>\n<summary>Starting stream thread...</summary>\n\n")
+    return Response(yoink(most_recent_thread), content_type='text/plain', status=200)
-        
-        # Start streaming output from output_stream
-        log("Starting to stream output.\n")
-        while not output_stream.done:
-            if output_stream.buffer:
-                delta = output_stream.buffer.pop(0)
-                yield delta.encode("utf-8")
-            else:
-                awaiter(sleep(0.05))
-        log(f"\nStream completed.\n</details>\n\n---\n\n### Completed Assistant Response:\n{output_stream.response}\n\n---\n\n")
-        output_stream.reset()
-        log("\n### Task completed successfully without errors!\n")
-    
-    return Response(generate(), content_type='text/plain', status=200)
 
 if __name__ == '__main__':
-    app.run(debug=True, host='0.0.0.0', port=1986)
+    logging.info("Starting Flask application.")
+    app.run(debug=True)

app/main.py

@@ -1,23 +1,39 @@
-# To parse video ids
 import re
+import threading
+import asyncio
+from asyncio import sleep
+from typing_extensions import override
+from datetime import datetime
+import pytz
+import os
+import logging
 
-# Youtube Transcript stuff import
+# Youtube Transcript imports
 import youtube_transcript_api._errors
 from youtube_transcript_api import YouTubeTranscriptApi
 from youtube_transcript_api.formatters import TextFormatter
 
-# OpenAI API stuff import
+# OpenAI API imports
 from openai import AssistantEventHandler
 from openai import OpenAI
 
-### For streaming
+# Load environment variables
-from typing_extensions import override
+from dotenv import load_dotenv
-import asyncio
+load_dotenv()
+
+# Handle async outside of async functions
 awaiter = asyncio.run
 
+# Configure logging
+logging.basicConfig(
+    filename='./logs/main.log',
+    level=logging.INFO,
+    format='%(asctime)s %(levelname)s: %(message)s',
+    datefmt='%Y-%m-%d %H:%M:%S'
+)
+
 # The StreamOutput class to handle streaming
 class StreamOutput:
-  
     def __init__(self):
         self.delta: str = ""
         self.response: str = ""
@@ -28,44 +44,33 @@ class StreamOutput:
         self.delta = ""
         self.response = ""
         self.done = False
-    self.buffer: list = []
+        self.buffer = []
 
-  async def send_delta(self, delta):
+    def send_delta(self, delta):
+      awaiter(self.process_delta(delta))
+
+    async def process_delta(self, delta):
         self.delta = delta
         self.response += delta
-      def get_index(list):
+
-        if len(list) == 0:  
+        def get_index(lst):
+            if len(lst) == 0:
                 return 0
             else:
-          return len(list)-1
+                return len(lst) - 1
-      if self.buffer != []:
+
+        if self.buffer:
             try:
                 if self.delta != self.buffer[get_index(self.buffer)]:
                     self.buffer.append(delta)
             except IndexError as index_error:
-          log(f"\nCaught IndexError: {str(index_error)}")
+                logging.error(f"Caught IndexError: {str(index_error)}")
                 self.buffer.append(delta)
-      else: self.buffer.append(delta)
+        else:
+            self.buffer.append(delta)
+        return
 
-# To get the env var
+# OpenAI Config 
-from dotenv import load_dotenv
-import os
-
-load_dotenv()
-
-# For logging
-import pytz
-from datetime import datetime
-
-def log(message):
-  try:
-    with open("logs/log.md", "a") as file:
-      file.write(message)
-  except FileNotFoundError:
-        with open("logs/log.md", "x+"):
-            log(message)
-
-### OpenAI Config 
 
 # Setting up OpenAI Client with API Key
 client = OpenAI(
@@ -74,23 +79,29 @@ client = OpenAI(
     api_key=os.getenv("OPENAI_API_KEY")
 )
 
-# screw bardo assistant that is configured to make notes and 5Q&A based on any given YouTube Transcript
+# Screw Bardo Assistant ID
 asst_screw_bardo_id = "asst_JGFaX6uOIotqy5mIJnu3Yyp7"
 
-# This is copy and pasted straight up from the quickstart guide, just appending to an output buffer instead of directly printing:
+# Event Handler for OpenAI Assistant
 class EventHandler(AssistantEventHandler):    
     @override
     def on_text_created(self, text) -> None:
-    awaiter(output_stream.send_delta("Response Recieved:\n\nScrew-Bardo:\n\n"))
+        output_stream.send_delta("Response Received:\n\nScrew-Bardo:\n\n")
+        logging.info("Text created event handled.")
 
     @override
     def on_text_delta(self, delta, snapshot):
-      awaiter(output_stream.send_delta(delta.value))
+        output_stream.send_delta(delta.value)
+        logging.debug(f"Text delta received: {delta.value}")
 
     def on_tool_call_created(self, tool_call):
-      raise Exception("Assistant shouldn't be calling tools.")
+        error_msg = "Assistant shouldn't be calling tools."
+        logging.error(error_msg)
+        raise Exception(error_msg)
 
 def create_and_stream(transcript):
+    logging.info("Starting OpenAI stream thread.")
+    try:
         with client.beta.threads.create_and_run_stream(
             assistant_id=asst_screw_bardo_id,
             thread={
@@ -100,35 +111,85 @@ def create_and_stream(transcript):
         ) as stream:
             stream.until_done()
             output_stream.done = True
+            logging.info("OpenAI stream completed.")
+    except Exception as e:
+        logging.exception("Exception occurred during create_and_stream.")
+
+def yoink(thread: threading.Thread):
+    logging.info("Starting stream thread...")
+    thread.start()
+    logging.info("Stream thread started. Beginning to stream output.")
+
+    try:
+        while not output_stream.done:
+            if output_stream.buffer:
+                delta = output_stream.buffer.pop(0)
+                yield bytes(delta, encoding="utf-8")
+            else:
+                asyncio.run(sleep(0.018))
+    except Exception as e:
+        logging.exception("Exception occurred during streaming output.")
+    finally:
+        logging.info("Stream completed successfully.")
+        logging.info(f"Completed Assistant Response:\n{output_stream.response}")
+        output_stream.reset()
+        thread.join()
+        logging.info("Stream thread joined. Task completed successfully without errors.")
+
+def process(url):
+    current_time = datetime.now(pytz.timezone('America/New_York')).strftime('%Y-%m-%d %H:%M:%S')
+    logging.info(f"New Entry at {current_time}")
+    logging.info(f"URL: {url}")
+
+    video_id = get_video_id(url) 
+    if not video_id:
+        logging.warning(f"Could not parse video id from URL: {url}")
+        return (False, "Couldn't parse video ID from URL. (Are you sure you entered a valid YouTube.com or YouTu.be URL?)", 400, None)
+    logging.info(f"Parsed Video ID: {video_id}")
+
+    # Get the transcript for that video ID
+    transcript = get_auto_transcript(video_id)
+    if not transcript:
+        logging.error("Error: could not retrieve transcript. Assistant won't be called.")
+        return (False, "Successfully parsed video ID from URL, however the ID was either invalid, the transcript was disabled by the video owner, or some other error was raised because of YouTube.", 200, None)
+
+    thread = threading.Thread(
+        name="create_stream",
+        target=create_and_stream,
+        args=(transcript,)
+    )
+    logging.info("Stream preparation complete, sending reply.")
+    return (True, None, None, thread)
 
 def get_video_id(url):
     youtu_be = r'(?<=youtu.be/)([A-Za-z0-9_-]{11})'
     youtube_com = r'(?<=youtube\.com\/watch\?v=)([A-Za-z0-9_-]{11})'
 
-  id = re.search(youtu_be, url)
+    id_match = re.search(youtu_be, url)
-  if not id:
+    if not id_match:
-      id = re.search(youtube_com, url)
+        id_match = re.search(youtube_com, url)
 
-  if not id:
+    if not id_match:
         # Couldn't parse video ID from URL
+        logging.warning(f"Failed to parse video ID from URL: {url}")
         return None
 
-  return id.group(1)
+    return id_match.group(1)
 
-# Takes the transcript and formats it in basic text before writing it to auto-transcript.txt
 def get_auto_transcript(video_id):
     trans_api_errors = youtube_transcript_api._errors
     try:
         transcript = YouTubeTranscriptApi.get_transcript(video_id, languages=['en'], proxies=None, cookies=None, preserve_formatting=False)
     except trans_api_errors.TranscriptsDisabled as e:
-    log(f'\n\n# Exception while fetching transcript:\n \n{e}\n')
+        logging.exception(f"Exception while fetching transcript: {e}")
         return None
 
     formatter = TextFormatter()  # Ensure that you create an instance of TextFormatter
-
     txt_transcript = formatter.format_transcript(transcript)
+    logging.info("Transcript successfully retrieved and formatted.")
     return txt_transcript
 
+# Initialize output stream
 output_stream = StreamOutput()
 
-log(f"\n\n# Main initilized at {datetime.now(pytz.timezone('America/New_York')).strftime('%Y-%m-%d %H:%M:%S')}. Presumeably application starting.\n")
+logging.info(f"Main initialized at {datetime.now(pytz.timezone('America/New_York')).strftime('%Y-%m-%d %H:%M:%S')}. Presumably application starting.")

app/website/static/style.css

@@ -40,6 +40,7 @@ body {
     overflow-y: auto;
     font-size: 1rem;
     line-height: 1.5;
+    scroll-behavior: smooth;
 }
 
 .form-section {