Update app.py

app.py

@@ -3,6 +3,8 @@ import openai
 import fitz  # PyMuPDF for PDF processing
 import base64
 import io
+import numpy as np
+import soundfile as sf
 
 # Variable to store API key
 api_key = ""
@@ -19,217 +21,148 @@ def query_openai(messages, temperature, top_p, max_output_tokens):
         return "Please enter your OpenAI API key first."
 
     try:
-        openai.api_key = api_key  # Set API key dynamically
-
-        # Ensure numeric values for OpenAI parameters
-        temperature = float(temperature) if temperature else 1.0
-        top_p = float(top_p) if top_p else 1.0
-        max_output_tokens = int(max_output_tokens) if max_output_tokens else 2048
+        openai.api_key = api_key  
 
         response = openai.ChatCompletion.create(
             model="gpt-4.5-preview",
             messages=messages,
-            temperature=temperature,
-            top_p=top_p,
-            max_tokens=max_output_tokens
+            temperature=float(temperature),
+            top_p=float(top_p),
+            max_tokens=int(max_output_tokens)
         )
         return response["choices"][0]["message"]["content"]
     except Exception as e:
         return f"Error: {str(e)}"
 
-# Function to process image URL input
-def image_url_chat(image_url, text_query, temperature, top_p, max_output_tokens):
-    if not image_url or not text_query:
-        return "Please provide an image URL and a query."
-
-    messages = [
-        {"role": "user", "content": [
-            {"type": "image_url", "image_url": {"url": image_url}},  # Corrected format
-            {"type": "text", "text": text_query}
-        ]},
-    ]
-    return query_openai(messages, temperature, top_p, max_output_tokens)
-
-# Function to process text input
-def text_chat(text_query, temperature, top_p, max_output_tokens):
-    if not text_query:
-        return "Please enter a query."
-
-    messages = [{"role": "user", "content": [{"type": "text", "text": text_query}]}]
-    return query_openai(messages, temperature, top_p, max_output_tokens)
-
-# Function to process uploaded image input
-def image_chat(image_file, text_query, temperature, top_p, max_output_tokens):
-    if image_file is None or not text_query:
-        return "Please upload an image and provide a query."
-
-    # Encode image as base64
-    with open(image_file, "rb") as img:
-        base64_image = base64.b64encode(img.read()).decode("utf-8")
-
-    image_data = f"data:image/jpeg;base64,{base64_image}"
-
-    messages = [
-        {"role": "user", "content": [
-            {"type": "image_url", "image_url": {"url": image_data}},  # Fixed format
-            {"type": "text", "text": text_query}
-        ]},
-    ]
-    return query_openai(messages, temperature, top_p, max_output_tokens)
-
-# Function to process uploaded PDF input
-def pdf_chat(pdf_file, text_query, temperature, top_p, max_output_tokens):
-    if pdf_file is None or not text_query:
-        return "Please upload a PDF and provide a query."
-
-    try:
-        # Extract text from all pages of the PDF
-        doc = fitz.open(pdf_file.name)
-        text = "\n".join([page.get_text("text") for page in doc])  # Extract text from all pages
-
-        # If no text found, return an error
-        if not text.strip():
-            return "No text found in the PDF."
-
-        # Create the query message with the extracted text and the user's query
-        messages = [
-            {"role": "user", "content": [
-                {"type": "text", "text": text},  # The extracted text from the PDF
-                {"type": "text", "text": text_query}
-            ]},
-        ]
-        return query_openai(messages, temperature, top_p, max_output_tokens)
-    
-    except Exception as e:
-        return f"Error processing the PDF: {str(e)}"
-
-# Function to transcribe audio to text using OpenAI Whisper API
-def transcribe_audio(audio_binary, openai_api_key):
-    if not openai_api_key:
+# Function to transcribe audio
+def transcribe_audio(audio_input):
+    if not api_key:
         return "Error: No API key provided."
     
-    openai.api_key = openai_api_key
+    openai.api_key = api_key
     
     try:
-        # Use the correct transcription API call
-        audio_file_obj = io.BytesIO(audio_binary)
-        audio_file_obj.name = 'audio.wav'  # Set a name for the file object (as OpenAI expects it)
-
-        # Transcribe the audio to text using OpenAI's whisper model
-        audio_file_transcription = openai.Audio.transcribe(file=audio_file_obj, model="whisper-1")
-        return audio_file_transcription.text
+        if isinstance(audio_input, np.ndarray):
+            wav_io = io.BytesIO()
+            sf.write(wav_io, audio_input, samplerate=16000, format="WAV")
+            wav_io.seek(0)
+            audio_file_obj = wav_io
+            audio_file_obj.name = "recorded_audio.wav"
+        else:
+            audio_file_obj = io.BytesIO(audio_input)
+            audio_file_obj.name = "uploaded_audio.wav"
+
+        transcription = openai.Audio.transcribe(file=audio_file_obj, model="whisper-1")
+        return transcription["text"]
     except Exception as e:
         return f"Error transcribing audio: {str(e)}"
 
-# Function to clear the chat (Fix: Returns the correct number of outputs)
+# Function to clear chat
 def clear_chat():
-    return "", "", "", "", "", "", "", None, "", None, "", 1.0, 1.0, 2048
+    return "", "", "", "", "", "", "", None, "", None, "", None, "", 1.0, 1.0, 2048
 
 # Gradio UI Layout
 with gr.Blocks() as demo:
-    gr.Markdown("## GPT-4.5 Preview Chatbot")
-
-    # Accordion for explaining hyperparameters
-    with gr.Accordion("Hyperparameters", open=False):
-        gr.Markdown("""
-        ### Temperature: 
-        Controls the randomness of the model's output. A lower temperature makes the model more deterministic, while a higher temperature makes it more creative and varied.
-
-        ### Top-P (Nucleus Sampling): 
-        Controls the cumulative probability distribution from which the model picks the next word. A lower value makes the model more focused and deterministic, while a higher value increases randomness.
-
-        ### Max Output Tokens: 
-        Limits the number of tokens (words or subwords) the model can generate in its response. You can use this to control the length of the response.
-        """)
+    gr.Markdown("## 🔥 GPT-4.5 AI Chatbot: Text, Image, PDF, & Voice Support")
 
+    # Custom CSS for buttons
     gr.HTML("""
     <style>
         #api_key_button {
-            margin-top: 27px; /* Add margin-top to the button */
-            background: linear-gradient(135deg, #4a00e0 0%, #8e2de2 100%); /* Purple gradient */
+            margin-top: 27px; 
+            background: linear-gradient(135deg, #4a00e0 0%, #8e2de2 100%);
+            color: white;
+            font-weight: bold;
+            border-radius: 5px;
         }
         #api_key_button:hover {
-            background: linear-gradient(135deg, #5b10f1 0%, #9f3ef3 100%); /* Slightly lighter */
+            background: linear-gradient(135deg, #5b10f1 0%, #9f3ef3 100%);
         }
         #clear_chat_button {
-            background: linear-gradient(135deg, #e53e3e 0%, #f56565 100%); /* Red gradient */
+            background: linear-gradient(135deg, #e53e3e 0%, #f56565 100%);
+            color: white;
+            font-weight: bold;
+            border-radius: 5px;
         }
         #clear_chat_button:hover {
-            background: linear-gradient(135deg, #c53030 0%, #e53e3e 100%); /* Slightly darker red gradient on hover */
+            background: linear-gradient(135deg, #c53030 0%, #e53e3e 100%);
         }
     </style>
     """)
-    
+
     # API Key Input
     with gr.Row():
         api_key_input = gr.Textbox(label="Enter OpenAI API Key", type="password")
         api_key_button = gr.Button("Set API Key", elem_id="api_key_button")
         api_key_output = gr.Textbox(label="API Key Status", interactive=False)
 
-    with gr.Row():
-        temperature = gr.Slider(0, 2, value=1.0, step=0.1, label="Temperature")
-        top_p = gr.Slider(0, 1, value=1.0, step=0.1, label="Top-P")
-        max_output_tokens = gr.Slider(0, 16384, value=2048, step=512, label="Max Output Tokens")  # Changed default to 2048
-    
+    # Accordion for Hyperparameters
+    with gr.Accordion("🔧 Advanced Settings (Hyperparameters)", open=False):
+        gr.Markdown("""
+        - **Temperature**: Controls randomness. Lower values make responses more predictable.
+        - **Top-P (Nucleus Sampling)**: Determines how many top probable words can be chosen.
+        - **Max Output Tokens**: Limits the length of the response.
+        """)
+        with gr.Row():
+            temperature = gr.Slider(0, 2, value=1.0, step=0.1, label="Temperature")
+            top_p = gr.Slider(0, 1, value=1.0, step=0.1, label="Top-P")
+            max_output_tokens = gr.Slider(0, 16384, value=2048, step=512, label="Max Output Tokens")  
+
     with gr.Tabs():
-        with gr.Tab("Image URL Chat"):
+        with gr.Tab("💬 Text Chat"):
+            text_query = gr.Textbox(label="Enter your query")
+            text_output = gr.Textbox(label="Response", interactive=False)
+            text_button = gr.Button("Ask")
+
+        with gr.Tab("🖼️ Image URL Chat"):
             image_url = gr.Textbox(label="Enter Image URL")
             image_query = gr.Textbox(label="Ask about the Image")
             image_url_output = gr.Textbox(label="Response", interactive=False)
             image_url_button = gr.Button("Ask")
-        
-        with gr.Tab("Text Chat"):
-            text_query = gr.Textbox(label="Enter your query")
-            text_output = gr.Textbox(label="Response", interactive=False)
-            text_button = gr.Button("Ask")
-        
-        with gr.Tab("Image Chat"):
+
+        with gr.Tab("📸 Image Upload Chat"):
             image_upload = gr.File(label="Upload an Image", type="filepath")
             image_text_query = gr.Textbox(label="Ask about the uploaded image")
             image_output = gr.Textbox(label="Response", interactive=False)
             image_button = gr.Button("Ask")
-        
-        with gr.Tab("PDF Chat"):
+
+        with gr.Tab("📄 PDF Chat"):
             pdf_upload = gr.File(label="Upload a PDF", type="filepath")
             pdf_text_query = gr.Textbox(label="Ask about the uploaded PDF")
             pdf_output = gr.Textbox(label="Response", interactive=False)
             pdf_button = gr.Button("Ask")
 
-        with gr.Tab("Voice Chat"):
-            audio_upload = gr.File(label="Upload an Audio File", type="binary")
-            audio_query = gr.Textbox(label="Ask about the transcription")
+        with gr.Tab("🎤 Voice Chat"):
+            audio_record = gr.Audio(source="microphone", type="numpy", label="🎙️ Record Audio")  
+            audio_upload = gr.File(label="📂 Upload an Audio File", type="binary")  
+            audio_query = gr.Textbox(label="Ask a question about the transcription")
             audio_output = gr.Textbox(label="Response", interactive=False)
             audio_button = gr.Button("Ask")
 
     # Clear chat button
-    clear_button = gr.Button("Clear Chat",elem_id="clear_chat_button")
+    clear_button = gr.Button("🧹 Clear Chat", elem_id="clear_chat_button")
 
     # Button Click Actions
     api_key_button.click(set_api_key, inputs=[api_key_input], outputs=[api_key_output])
-    image_url_button.click(image_url_chat, [image_url, image_query, temperature, top_p, max_output_tokens], image_url_output)
-    text_button.click(text_chat, [text_query, temperature, top_p, max_output_tokens], text_output)
-    image_button.click(image_chat, [image_upload, image_text_query, temperature, top_p, max_output_tokens], image_output)
-    pdf_button.click(pdf_chat, [pdf_upload, pdf_text_query, temperature, top_p, max_output_tokens], pdf_output)
+    text_button.click(lambda q, t, p, m: query_openai([{"role": "user", "content": [{"type": "text", "text": q}]}], t, p, m), 
+                      inputs=[text_query, temperature, top_p, max_output_tokens], 
+                      outputs=[text_output])
     
-    # For Voice Chat
-    audio_button.click(
-        lambda audio_binary, query, temperature, top_p, max_output_tokens: query_openai(
-            [{"role": "user", "content": [{"type": "text", "text": transcribe_audio(audio_binary, api_key)}, {"type": "text", "text": query}]}],
-            temperature, top_p, max_output_tokens
-        ), [audio_upload, audio_query, temperature, top_p, max_output_tokens], audio_output
-    )
-
-    # Fix: Clear button resets all necessary fields correctly
-    clear_button.click(
-        clear_chat,
-        outputs=[
-            image_url, image_query, image_url_output, 
-            text_query, text_output, 
-            image_text_query, image_output, 
-            pdf_upload, pdf_text_query, pdf_output, 
-            temperature, top_p, max_output_tokens
-        ]
-    )
+    image_url_button.click(lambda u, q, t, p, m: query_openai([{"role": "user", "content": [{"type": "image_url", "image_url": {"url": u}}, {"type": "text", "text": q}]}], t, p, m), 
+                           inputs=[image_url, image_query, temperature, top_p, max_output_tokens], 
+                           outputs=[image_url_output])
+
+    image_button.click(lambda f, q, t, p, m: query_openai([{"role": "user", "content": [{"type": "image_url", "image_url": {"url": f}}, {"type": "text", "text": q}]}], t, p, m), 
+                       inputs=[image_upload, image_text_query, temperature, top_p, max_output_tokens], 
+                       outputs=[image_output])
+
+    pdf_button.click(lambda f, q, t, p, m: query_openai([{"role": "user", "content": [{"type": "text", "text": f.read()}, {"type": "text", "text": q}]}], t, p, m), 
+                     inputs=[pdf_upload, pdf_text_query, temperature, top_p, max_output_tokens], 
+                     outputs=[pdf_output])
+
+    audio_button.click(lambda a, q, t, p, m: query_openai([{"role": "user", "content": [{"type": "text", "text": transcribe_audio(a)}, {"type": "text", "text": q}]}], t, p, m), 
+                       inputs=[audio_record, audio_query, temperature, top_p, max_output_tokens], 
+                       outputs=[audio_output])
 
 # Launch Gradio App
 if __name__ == "__main__":