Use typing effect instead of streaming as it causes a lot of issues with telegram rate limits.

main.py

@@ -8,7 +8,9 @@ from dotenv import load_dotenv
 from pydub import AudioSegment
 from telegram import Update
 from functools import wraps
-from telegram.error import BadRequest, RetryAfter
+from telegram.constants import ChatAction
+from functools import wraps
+from telegram.error import BadRequest, RetryAfter, TimedOut
 from telegram.ext import ApplicationBuilder, CommandHandler, ContextTypes, MessageHandler, filters
 
 logging.basicConfig(
@@ -63,6 +65,8 @@ async def start(update: Update, context: ContextTypes.DEFAULT_TYPE):
 @restricted    
 async def imagine(update: Update, context: ContextTypes.DEFAULT_TYPE):
     users[f"{update.effective_chat.id}"]["usage"]['dalle'] += 1
+    await context.bot.send_chat_action(chat_id=update.effective_chat.id, action=ChatAction.TYPING)
+    
     response = openai.Image.create(
         prompt=update.message.text,
         n=1,
@@ -78,7 +82,7 @@ async def imagine(update: Update, context: ContextTypes.DEFAULT_TYPE):
 async def attachment(update: Update, context: ContextTypes.DEFAULT_TYPE):
     # Initialize variables
     chat_id = update.effective_chat.id
-    user_id = update.effective_user.id
+    await context.bot.send_chat_action(chat_id=chat_id, action=ChatAction.TYPING)
     users[f"{chat_id}"]["usage"]['whisper'] = 0
     transcript = {'text': ''}
     
@@ -148,6 +152,7 @@ async def attachment(update: Update, context: ContextTypes.DEFAULT_TYPE):
 @restricted
 async def chat(update: Update, context: ContextTypes.DEFAULT_TYPE):
     chat_id = str(update.effective_chat.id)
+    await context.bot.send_chat_action(chat_id=chat_id, action=ChatAction.TYPING)
     
     # Initialize user if not present
     if chat_id not in users:
@@ -166,60 +171,30 @@ async def chat(update: Update, context: ContextTypes.DEFAULT_TYPE):
         user_context.pop(0)
     
     # Interact with ChatGPT API and stream the response
-    response = openai.ChatCompletion.create(
-        model=MODEL,
-        messages=[{"role": "system", "content": SYSTEM_PROMPT}] + user_context,
-        stream=True,
-        temperature=float(TEMPERATURE)
-    )
+    try:
+        response = openai.ChatCompletion.create(
+            model=MODEL,
+            messages=[{"role": "system", "content": SYSTEM_PROMPT}] + user_context,
+            temperature=float(TEMPERATURE)
+        )
+    except:
+        await context.bot.send_message(chat_id=update.effective_chat.id, text="There was a problem with OpenAI, so I can't answer you.")
     
     # Initialize variables for streaming
     assistant_message = ""
-    message_sent = False
-    sent_message = None
-    
-    # Process response chunks
-    batch = 3 # Batches of 10 to update
-    for chunk in response:
-        batch -= 1
-        if 'choices' in chunk:
-            choice = chunk['choices'][0]
-            if 'delta' in choice and 'content' in choice['delta']:
-                new_content = choice['delta']['content']
-                assistant_message += new_content
+    if 'choices' in response:
+        assistant_message = response['choices'][0]['message']['content']
+        await context.bot.send_message(chat_id=update.effective_chat.id, text=assistant_message)
+    else:
+        await context.bot.send_message(chat_id=update.effective_chat.id, text="There was a problem with OpenAI. Maybe your prompt is forbidden? They like to censor a lot!")
 
-                # Edit the message in real time
-                if not message_sent:
-                    sent_message = await context.bot.send_message(chat_id=update.effective_chat.id, text=assistant_message)
-                    message_sent = True
-                else:
-                    if batch == 0:
-                        try:
-                            if new_content.strip() != "":
-                                await context.bot.edit_message_text(chat_id=update.effective_chat.id, message_id=sent_message.message_id, text=assistant_message)
-                        except BadRequest as e:
-                            if "Message is not modified" not in str(e):
-                                raise e
-                        except RetryAfter as e:
-                            await asyncio.sleep(e.retry_after)
-                        batch = 3
-
-    # Final check to ensure the entire message is displayed
-    try:
-        await context.bot.edit_message_text(chat_id=update.effective_chat.id, message_id=sent_message.message_id, text=assistant_message)
-    except BadRequest as e:
-        if "Message is not modified" not in str(e):
-            raise e
-    except RetryAfter as e:
-        await asyncio.sleep(e.retry_after)
-    
     # Update context
     user_context.append({"role": "assistant", "content": assistant_message})
     if len(user_context) > MAX_USER_CONTEXT:
         user_context.pop(0)
     
     # Update usage
-    users[chat_id]["usage"]['chatgpt'] += round(len(str(user_context))/3, 0)
+    users[chat_id]["usage"]['chatgpt'] += int(response['usage']['total_tokens'])
 
 @restricted
 async def clear(update: Update, context: ContextTypes.DEFAULT_TYPE) -> None: