Update app.py

app.py

@@ -9,7 +9,7 @@ app = Flask(__name__)
 GLOBAL_WORKSPACE_ID = None
 GLOBAL_BOT_ID = None
 
-# Empty cookie value (to be filled by the user)
+# Empty cookie value
 AUTH_COOKIE = "pscd=try.botpress.com; _hjSessionUser_2931810=eyJpZCI6ImQ2MGMzYjhkLTlkMjQtNTA0OS1hMzlmLWEzNmI0NzA0NzUxNCIsImNyZWF0ZWQiOjE3MzU3MTg0MDcwNTAsImV4aXN0aW5nIjp0cnVlfQ==; hubspotutk=75739411a4d011b2164c4f3d944ecb94; intercom-device-id-bjzkw2xf=afd0a36b-b229-44e3-828e-60483c80c10c; _hjSessionUser_3339867=eyJpZCI6IjU4ODlmMTY4LWRkNGEtNTJhZS1hZTUzLWZlYWQwM2ZmMTVjNyIsImNyZWF0ZWQiOjE3MzU3MTg1ODM4MDgsImV4aXN0aW5nIjp0cnVlfQ==; __hstc=59821234.75739411a4d011b2164c4f3d944ecb94.1735718442141.1746194848836.1746538539919.88; mp_1195923e954ce61d822842b5832047cd_mixpanel=%7B%22distinct_id%22%3A%20%22d403ad7b-ea73-4d29-b977-5fd95afd585c%22%2C%22%24device_id%22%3A%20%22d403ad7b-ea73-4d29-b977-5fd95afd585c%22%2C%22%24initial_referrer%22%3A%20%22https%3A%2F%2Fapp.botpress.cloud%2F%22%2C%22%24initial_referring_domain%22%3A%20%22app.botpress.cloud%22%2C%22__mps%22%3A%20%7B%7D%2C%22__mpso%22%3A%20%7B%22%24initial_referrer%22%3A%20%22https%3A%2F%2Fapp.botpress.cloud%2F%22%2C%22%24initial_referring_domain%22%3A%20%22app.botpress.cloud%22%7D%2C%22__mpus%22%3A%20%7B%7D%2C%22__mpa%22%3A%20%7B%7D%2C%22__mpu%22%3A%20%7B%7D%2C%22__mpr%22%3A%20%5B%5D%2C%22__mpap%22%3A%20%5B%5D%2C%22%24user_id%22%3A%20%22d403ad7b-ea73-4d29-b977-5fd95afd585c%22%7D; intercom-session-bjzkw2xf=TTdnZGNWUC9xNXMreE80NXhRZFNZS0pyUEdsbkJRc2JMcXdGZmcveVRPYkxZTmVnVnhqMUhJTWlDcEpVcWljeDZVYVVSblN4YnV5S0xBdWxDd2swQjZiaUZTeWl5M1psRmtoUWJwUU9FSFE9LS1Bay9zNldJTmVhUFdwMFNReFRmcXB3PT0=--68abc5394d7aab99748f3e451637cab5d9152a4c; _ga=GA1.2.1726154447.1735718383; _gid=GA1.2.1619749406.1746821934; _gat_UA-226900660-1=1; _gat_UA-226900660-2=1; _ga_W6YT9YSNLH=GS2.2.s1746821948$o89$g0$t1746821948$j0$l0$h0; _ga_CYSS87Q508=GS2.2.s1746821949$o89$g0$t1746821949$j0$l0$h0; _hjSession_2931810=eyJpZCI6IjA1YTlkMzY3LWNiZWEtNGQ3OC04YzNiLTEzNTFjNjkxYzViZSIsImMiOjE3NDY4MjE5NTU2NjYsInMiOjAsInIiOjAsInNiIjowLCJzciI6MCwic2UiOjAsImZzIjowLCJzcCI6MH0=; _ga_PCC6TBWJY6=GS2.1.s1746821926$o120$g1$t1746821960$j0$l0$h0; _ga_HKHSWES9V9=GS2.1.s1746821929$o120$g1$t1746821960$j29$l0$h193244206; csrf_token_bd9ac21c34b9f0915e733c3e5305d737d0722c1168be7376b889426b5ec2a298=pIDxu4npODUGEpo7JHVQrKZ4GFaa3U+3BpgxaV5hcVw=; ory_kratos_session=MTc0NjgyMjAyOHxfUmdMaVlTQXVfSmxlT1lJSGpyU2FhbjVUTHg0R0ZsQWgtVm00M3pHcXZwVG9yNW1qRDJheUFGaFZvNmFEUVdBOThQR014RjJJbmhUMmhIV1I1ME5UVHZkTDNpMUMtQlRjZ1ZTbE55M19Pb2dHTF9vQlJoSGlBQnRRWUp0M1ZUdnVvcENLeVhOTllWNk1zMk11bFVPOWFrTzJMTTdxMmVteUozVVRDMWE5TVIxbDgzU3dUY2VQaDBRWDN4bDJUVm8yUkZQa19sb09GbzlFZHF2MDFQcVR6bVVWVVpDLXVoQ1lXMEh2LV9Sd2VNZXM1cjM4TGZPVTJqdW5xNTBETTBDYkppU0xNU2xicUk3Z2EyMnFkVmdyQT09fEYiv7pXcfXVnpIFi4JLGgDObQAchyJCoAwGDSkFkoX7; ajs_user_id=d403ad7b-ea73-4d29-b977-5fd95afd585c; ajs_anonymous_id=cda6139d-cb82-4906-bfac-adaea115b097"
 
 
@@ -177,64 +177,195 @@ def delete_workspace(workspace_id):
 
 
 # -------------------------------------------------------------------
-# Flask Endpoint
+# Main function that calls the Botpress API endpoint
 # -------------------------------------------------------------------
-@app.route("/chat", methods=["POST"])
-def chat_endpoint():
+def chat_with_assistant(user_input, chat_history, bot_id, workspace_id):
     """
-    Expects JSON with:
-    {
-      "audio_url": "string" (required)
+    Sends the user input and chat history to the Botpress API endpoint,
+    returns the assistant's response and (possibly updated) bot/workspace IDs.
+    """
+    # Prepare the headers
+    headers = {
+        "User-Agent": "Mozilla/5.0",
+        "x-bot-id": bot_id,
+        "Content-Type": "application/json",
+        "Cookie": AUTH_COOKIE
     }
-    Returns JSON with:
-    {
-      "assistant_response": "string" (transcription result)
+
+    # Process chat history into the format expected by the API
+    messages = []
+    system_prompt = ""
+    for msg in chat_history:
+        if msg["role"] == "system":
+            system_prompt = msg["content"]
+        else:
+            messages.append({
+                "role": msg["role"],
+                "content": msg["content"],
+                "type": "text"
+            })
+
+    # Add the latest user input if not already in chat history
+    if user_input and (not messages or messages[-1]["role"] != "user" or messages[-1]["content"] != user_input):
+        messages.append({
+            "role": "user",
+            "content": user_input,
+            "type": "text"
+        })
+
+    # Prepare the payload for the API
+    payload = {
+        "type": "openai:generateContent",
+        "input": {
+            "model": {
+                "id": "gpt-4.1-2025-04-14"
+            },
+            "systemPrompt": system_prompt,
+            "messages": messages,
+            "temperature": 0.9,
+            "topP": 0.95,
+            "debug": False,
+        }
     }
-    """
-    global GLOBAL_WORKSPACE_ID, GLOBAL_BOT_ID
 
-    # Parse JSON from request
-    data = request.get_json(force=True)
-    audio_url = data.get("audio_url", "")
+    botpress_url = "https://api.botpress.cloud/v1/chat/actions"
+    max_retries = 3
+    timeout = 120  # Increased timeout for long messages
 
-    if not audio_url:
-        return jsonify({"assistant_response": "Error: audio_url is required"}), 400
+    # Flag to track if we need to create new IDs
+    need_new_ids = False
 
-    # If we don't yet have a workspace or bot, create them
-    if not GLOBAL_WORKSPACE_ID or not GLOBAL_BOT_ID:
-        print("No existing IDs found. Creating new workspace and bot...")
-        GLOBAL_WORKSPACE_ID = create_workspace()
-        if GLOBAL_WORKSPACE_ID:
-            GLOBAL_BOT_ID = create_bot(GLOBAL_WORKSPACE_ID)
+    # Attempt to send the request
+    for attempt in range(max_retries):
+        try:
+            print(f"Attempt {attempt+1}: Sending request to Botpress API with bot_id={bot_id}, workspace_id={workspace_id}")
+            response = requests.post(botpress_url, json=payload, headers=headers, timeout=timeout)
 
-        # If creation failed
-        if not GLOBAL_WORKSPACE_ID or not GLOBAL_BOT_ID:
-            return jsonify({"assistant_response": "I'm currently unavailable. Please try again later."}), 500
+            # If successful (200)
+            if response.status_code == 200:
+                data = response.json()
+                assistant_content = data.get('output', {}).get('choices', [{}])[0].get('content', '')
+                print(f"Successfully received response from Botpress API")
+                return assistant_content, bot_id, workspace_id
+
+            # Handle 403 error (authentication/authorization issue)
+            elif response.status_code == 403:
+                print(f"Received 403 error. Need to create new IDs.")
+                need_new_ids = True
+                break
+
+            elif response.status_code == 404:
+                print(f"Received 404 error. Need to create new IDs.")
+                need_new_ids = True
+                break
+
+            # Handle network errors or timeouts (just retry)
+            elif response.status_code in [443, 408, 502, 503, 504]:
+                print(f"Received error {response.status_code}. Retrying...")
+                time.sleep(3)  # Wait before retrying
+                continue
+
+            # Any other error status code
+            else:
+                print(f"Received unexpected error: {response.status_code}, {response.text}")
+                if attempt < max_retries - 1:
+                    time.sleep(2)
+                    continue
+                else:
+                    return f"Unable to get a response from the assistant (Error {response.status_code}).", bot_id, workspace_id
+
+        except requests.exceptions.Timeout:
+            print(f"Request timed out. Retrying...")
+            if attempt < max_retries - 1:
+                time.sleep(2)
+                continue
+            else:
+                return "The assistant is taking too long to respond. Please try again with a shorter message.", bot_id, workspace_id
+
+        except Exception as e:
+            print(f"Error during request: {str(e)}")
+            if attempt < max_retries - 1:
+                time.sleep(2)
+                continue
+            else:
+                return f"Unable to get a response from the assistant: {str(e)}", bot_id, workspace_id
+
+    # If we need new IDs, create them and try again
+    if need_new_ids:
+        print("Creating new workspace and bot IDs...")
+
+        # First, try to clean up old resources
+        if bot_id and workspace_id:
+            delete_bot(bot_id, workspace_id)
+            delete_workspace(workspace_id)
+
+        # Create new resources
+        new_workspace_id = create_workspace()
+        if not new_workspace_id:
+            return "Failed to create a new workspace. Please try again later.", bot_id, workspace_id
+
+        new_bot_id = create_bot(new_workspace_id)
+        if not new_bot_id:
+            return "Failed to create a new bot. Please try again later.", new_workspace_id, workspace_id
+
+        # Update headers with new bot ID
+        headers["x-bot-id"] = new_bot_id
 
+        # Try one more time with the new IDs
+        try:
+            print(f"Retrying with new bot_id={new_bot_id}, workspace_id={new_workspace_id}")
+            retry_response = requests.post(botpress_url, json=payload, headers=headers, timeout=timeout)
+
+            if retry_response.status_code == 200:
+                data = retry_response.json()
+                assistant_content = data.get('output', {}).get('choices', [{}])[0].get('content', '')
+                print(f"Successfully received response with new IDs")
+                return assistant_content, new_bot_id, new_workspace_id
+            else:
+                print(f"Failed with new IDs: {retry_response.status_code}, {retry_response.text}")
+                return f"Unable to get a response from the assistant with new credentials.", new_bot_id, new_workspace_id
+
+        except Exception as e:
+            print(f"Error with new IDs: {str(e)}")
+            return f"Unable to get a response from the assistant with new credentials: {str(e)}", new_bot_id, new_workspace_id
+
+    # Should not reach here due to the handling in the loop
+    return "Unable to get a response from the assistant.", bot_id, workspace_id
+
+
+# -------------------------------------------------------------------
+# Function to transcribe audio using the same API and error handling approach
+# -------------------------------------------------------------------
+def transcribe_audio(file_url, language="", prompt="get all text from this Audio", temperature=0, bot_id=None, workspace_id=None):
+    """
+    Sends an audio file URL to the Botpress API for transcription
+    Uses the same bot/workspace management as the chat function
+    """
     # Prepare the headers
     headers = {
         "User-Agent": "Mozilla/5.0",
-        "x-bot-id": GLOBAL_BOT_ID,
         "Content-Type": "application/json",
         "Cookie": AUTH_COOKIE
     }
 
-    # Use the transcription payload
+    # Add bot ID to headers if available
+    if bot_id:
+        headers["x-bot-id"] = bot_id
+
+    # Prepare the payload for audio transcription
     payload = {
         "type": "openai:transcribeAudio",
         "input": {
-            "fileUrl": audio_url,
-            "language": "",
-            "prompt": "get all text from this Audio",
-            "temperature": 0
+            "fileUrl": file_url,
+            "language": language,
+            "prompt": prompt,
+            "temperature": temperature
         }
     }
 
-    print(f"Processing audio transcription request with URL: {audio_url}")
-
     botpress_url = "https://api.botpress.cloud/v1/chat/actions"
     max_retries = 3
-    timeout = 120  # Increased timeout for long messages
+    timeout = 180  # Longer timeout for audio processing
 
     # Flag to track if we need to create new IDs
     need_new_ids = False
@@ -242,20 +373,24 @@ def chat_endpoint():
     # Attempt to send the request
     for attempt in range(max_retries):
         try:
-            print(f"Attempt {attempt+1}: Sending request to Botpress API with bot_id={GLOBAL_BOT_ID}, workspace_id={GLOBAL_WORKSPACE_ID}")
+            print(f"Attempt {attempt+1}: Sending transcription request to Botpress API")
+            if bot_id:
+                print(f"Using bot_id={bot_id}, workspace_id={workspace_id}")
+
             response = requests.post(botpress_url, json=payload, headers=headers, timeout=timeout)
 
             # If successful (200)
             if response.status_code == 200:
                 data = response.json()
 
-                # For audio transcription, extract the full text from all segments
+                # Extract the transcription text from all segments
                 segments = data.get('output', {}).get('segments', [])
-                full_transcript = " ".join([segment.get('text', '').strip() for segment in segments])
+                transcription = " ".join([segment.get('text', '').strip() for segment in segments])
+
                 print(f"Successfully received transcription from Botpress API")
-                return jsonify({"assistant_response": full_transcript})
+                return transcription, data, bot_id, workspace_id
 
-            # Handle 403 error (authentication/authorization issue)
+            # Handle 403/404 errors (authentication/authorization issue)
             elif response.status_code in [403, 404]:
                 print(f"Received {response.status_code} error. Need to create new IDs.")
                 need_new_ids = True
@@ -264,7 +399,7 @@ def chat_endpoint():
             # Handle network errors or timeouts (just retry)
             elif response.status_code in [443, 408, 502, 503, 504]:
                 print(f"Received error {response.status_code}. Retrying...")
-                time.sleep(3)  # Wait before retrying
+                time.sleep(3)
                 continue
 
             # Any other error status code
@@ -274,72 +409,191 @@ def chat_endpoint():
                     time.sleep(2)
                     continue
                 else:
-                    return jsonify({"assistant_response": f"Unable to get a response (Error {response.status_code})."})
+                    return f"Unable to transcribe audio (Error {response.status_code}).", None, bot_id, workspace_id
 
         except requests.exceptions.Timeout:
-            print(f"Request timed out. Retrying...")
+            print(f"Transcription request timed out. Retrying...")
             if attempt < max_retries - 1:
-                time.sleep(2)
+                time.sleep(3)
                 continue
             else:
-                return jsonify({"assistant_response": "The request is taking too long to process. Please try again."})
+                return "The transcription is taking too long. Please try with a shorter audio file.", None, bot_id, workspace_id
 
         except Exception as e:
-            print(f"Error during request: {str(e)}")
+            print(f"Error during transcription request: {str(e)}")
             if attempt < max_retries - 1:
-                time.sleep(2)
+                time.sleep(3)
                 continue
             else:
-                return jsonify({"assistant_response": f"Unable to process the request: {str(e)}"})
+                return f"Unable to transcribe audio: {str(e)}", None, bot_id, workspace_id
 
     # If we need new IDs, create them and try again
     if need_new_ids:
-        print("Creating new workspace and bot IDs...")
+        print("Creating new workspace and bot IDs for transcription...")
 
         # First, try to clean up old resources
-        if GLOBAL_BOT_ID and GLOBAL_WORKSPACE_ID:
-            delete_bot(GLOBAL_BOT_ID, GLOBAL_WORKSPACE_ID)
-            delete_workspace(GLOBAL_WORKSPACE_ID)
+        if bot_id and workspace_id:
+            delete_bot(bot_id, workspace_id)
+            delete_workspace(workspace_id)
 
         # Create new resources
         new_workspace_id = create_workspace()
         if not new_workspace_id:
-            return jsonify({"assistant_response": "Failed to create a new workspace. Please try again later."})
+            return "Failed to create a new workspace for transcription. Please try again later.", None, bot_id, workspace_id
 
         new_bot_id = create_bot(new_workspace_id)
         if not new_bot_id:
-            return jsonify({"assistant_response": "Failed to create a new bot. Please try again later."})
-
-        # Update global IDs
-        GLOBAL_BOT_ID = new_bot_id
-        GLOBAL_WORKSPACE_ID = new_workspace_id
+            return "Failed to create a new bot for transcription. Please try again later.", None, new_workspace_id, workspace_id
 
         # Update headers with new bot ID
         headers["x-bot-id"] = new_bot_id
 
         # Try one more time with the new IDs
         try:
-            print(f"Retrying with new bot_id={new_bot_id}, workspace_id={new_workspace_id}")
+            print(f"Retrying transcription with new bot_id={new_bot_id}, workspace_id={new_workspace_id}")
             retry_response = requests.post(botpress_url, json=payload, headers=headers, timeout=timeout)
 
             if retry_response.status_code == 200:
                 data = retry_response.json()
-
-                # For audio transcription, extract the full text from all segments
                 segments = data.get('output', {}).get('segments', [])
-                full_transcript = " ".join([segment.get('text', '').strip() for segment in segments])
+                transcription = " ".join([segment.get('text', '').strip() for segment in segments])
                 print(f"Successfully received transcription with new IDs")
-                return jsonify({"assistant_response": full_transcript})
+                return transcription, data, new_bot_id, new_workspace_id
             else:
-                print(f"Failed with new IDs: {retry_response.status_code}, {retry_response.text}")
-                return jsonify({"assistant_response": "Unable to get a response with new credentials."})
+                print(f"Transcription failed with new IDs: {retry_response.status_code}, {retry_response.text}")
+                return f"Unable to transcribe audio with new credentials.", None, new_bot_id, new_workspace_id
 
         except Exception as e:
-            print(f"Error with new IDs: {str(e)}")
-            return jsonify({"assistant_response": f"Unable to get a response with new credentials: {str(e)}"})
+            print(f"Error with new IDs during transcription: {str(e)}")
+            return f"Unable to transcribe audio with new credentials: {str(e)}", None, new_bot_id, new_workspace_id
 
     # Should not reach here due to the handling in the loop
-    return jsonify({"assistant_response": "Unable to process the request."})
+    return "Unable to transcribe audio.", None, bot_id, workspace_id
+
+
+# -------------------------------------------------------------------
+# Flask Endpoints
+# -------------------------------------------------------------------
+@app.route("/chat", methods=["POST"])
+def chat_endpoint():
+    """
+    Expects JSON with:
+    {
+      "user_input": "string",
+      "chat_history": [
+         {"role": "system", "content": "..."},
+         {"role": "user",   "content": "..."},
+         ...
+      ]
+    }
+    Returns JSON with:
+    {
+      "assistant_response": "string"
+    }
+    """
+    global GLOBAL_WORKSPACE_ID, GLOBAL_BOT_ID
+
+    # Parse JSON from request
+    data = request.get_json(force=True)
+    user_input = data.get("user_input", "")
+    chat_history = data.get("chat_history", [])
+
+    # If we don't yet have a workspace or bot, create them
+    if not GLOBAL_WORKSPACE_ID or not GLOBAL_BOT_ID:
+        print("No existing IDs found. Creating new workspace and bot...")
+        GLOBAL_WORKSPACE_ID = create_workspace()
+        if GLOBAL_WORKSPACE_ID:
+            GLOBAL_BOT_ID = create_bot(GLOBAL_WORKSPACE_ID)
+
+        # If creation failed
+        if not GLOBAL_WORKSPACE_ID or not GLOBAL_BOT_ID:
+            return jsonify({"assistant_response": "I'm currently unavailable. Please try again later."}), 500
+
+    # Call our function that interacts with Botpress API
+    print(f"Sending chat request with existing bot_id={GLOBAL_BOT_ID}, workspace_id={GLOBAL_WORKSPACE_ID}")
+    assistant_response, updated_bot_id, updated_workspace_id = chat_with_assistant(
+        user_input,
+        chat_history,
+        GLOBAL_BOT_ID,
+        GLOBAL_WORKSPACE_ID
+    )
+
+    # Update global IDs if they changed
+    if updated_bot_id != GLOBAL_BOT_ID or updated_workspace_id != GLOBAL_WORKSPACE_ID:
+        print(f"Updating global IDs: bot_id={updated_bot_id}, workspace_id={updated_workspace_id}")
+        GLOBAL_BOT_ID = updated_bot_id
+        GLOBAL_WORKSPACE_ID = updated_workspace_id
+
+    return jsonify({"assistant_response": assistant_response})
+
+
+@app.route("/transcribe", methods=["POST"])
+def transcribe_endpoint():
+    """
+    Expects JSON with:
+    {
+      "file_url": "string",
+      "language": "string" (optional),
+      "prompt": "string" (optional),
+      "temperature": float (optional),
+      "include_full_response": boolean (optional)
+    }
+    Returns JSON with:
+    {
+      "transcription": "string",
+      "full_response": {} (optional)
+    }
+    """
+    global GLOBAL_WORKSPACE_ID, GLOBAL_BOT_ID
+
+    # Parse JSON from request
+    data = request.get_json(force=True)
+    file_url = data.get("file_url", "")
+    language = data.get("language", "")
+    prompt = data.get("prompt", "get all text from this Audio")
+    temperature = data.get("temperature", 0)
+    include_full_response = data.get("include_full_response", False)
+
+    # Validate input
+    if not file_url:
+        return jsonify({"error": "Missing file_url parameter"}), 400
+
+    # If we don't yet have a workspace or bot, create them
+    if not GLOBAL_WORKSPACE_ID or not GLOBAL_BOT_ID:
+        print("No existing IDs found for transcription. Creating new workspace and bot...")
+        GLOBAL_WORKSPACE_ID = create_workspace()
+        if GLOBAL_WORKSPACE_ID:
+            GLOBAL_BOT_ID = create_bot(GLOBAL_WORKSPACE_ID)
+
+        # If creation failed
+        if not GLOBAL_WORKSPACE_ID or not GLOBAL_BOT_ID:
+            return jsonify({"transcription": "Transcription service is currently unavailable. Please try again later."}), 500
+
+    # Call our function that interacts with Botpress API for transcription
+    print(f"Sending transcription request with existing bot_id={GLOBAL_BOT_ID}, workspace_id={GLOBAL_WORKSPACE_ID}")
+    transcription, full_data, updated_bot_id, updated_workspace_id = transcribe_audio(
+        file_url,
+        language,
+        prompt,
+        temperature,
+        GLOBAL_BOT_ID,
+        GLOBAL_WORKSPACE_ID
+    )
+
+    # Update global IDs if they changed
+    if updated_bot_id != GLOBAL_BOT_ID or updated_workspace_id != GLOBAL_WORKSPACE_ID:
+        print(f"Updating global IDs after transcription: bot_id={updated_bot_id}, workspace_id={updated_workspace_id}")
+        GLOBAL_BOT_ID = updated_bot_id
+        GLOBAL_WORKSPACE_ID = updated_workspace_id
+
+    # Prepare the response
+    response = {"transcription": transcription}
+
+    # Include full response data if requested
+    if include_full_response and full_data:
+        response["full_response"] = full_data
+
+    return jsonify(response)
 
 
 # -------------------------------------------------------------------