Update app.py

app.py

@@ -17,11 +17,6 @@ from huggingface_hub import InferenceClient
 import inspect
 import logging
 import shutil
-import soundfile as sf
-from transformers import WhisperProcessor, WhisperForConditionalGeneration
-from transformers import pipeline
-import numpy as np
-
 
 
 # Set up basic configuration for logging
@@ -33,7 +28,8 @@ llama_cloud_api_key = os.environ.get("LLAMA_CLOUD_API_KEY")
 ACCOUNT_ID = os.environ.get("CLOUDFARE_ACCOUNT_ID")
 API_TOKEN = os.environ.get("CLOUDFLARE_AUTH_TOKEN")
 API_BASE_URL = "https://api.cloudflare.com/client/v4/accounts/a17f03e0f049ccae0c15cdcf3b9737ce/ai/run/"
-WHISPER_API_URL = "https://api-inference.huggingface.co/models/openai/whisper-large-v3"
+API_URL = "https://api-inference.huggingface.co/models/openai/whisper-large-v3"
+headers={"Authorization": f"Bearer {huggingface_token}"}
 
 print(f"ACCOUNT_ID: {ACCOUNT_ID}")
 print(f"CLOUDFLARE_AUTH_TOKEN: {API_TOKEN[:5]}..." if API_TOKEN else "Not set")
@@ -402,20 +398,27 @@ def summarize_web_results(query: str, search_results: List[Dict[str, str]], conv
     except Exception as e:
         return f"An error occurred during summarization: {str(e)}"
 
-def process_input(message, history, audio, model, temperature, num_calls, use_web_search, selected_docs):
-    if audio is not None:
-        try:
-            transcribed_text = transcribe(audio)
-            print(f"Transcribed audio: {transcribed_text}")
-            message = transcribed_text
-        except Exception as e:
-            print(f"Error transcribing audio: {str(e)}")
-            return f"Error transcribing audio: {str(e)}", history
+def transcribe_audio(audio):
+    if audio is None:
+        return "No audio file provided."
     
-    return respond(message, history, model, temperature, num_calls, use_web_search, selected_docs)
+    try:
+        sr, y = audio
+        y = y.astype(np.float32)
+        y /= np.max(np.abs(y))
 
-def respond(message, history, model, temperature, num_calls, use_web_search, selected_docs):
+        response = requests.post(API_URL, headers=headers, json={"inputs": {"sampling_rate": sr, "raw": y.tolist()}})
+        response.raise_for_status()
+        return response.json()["text"]
+    except Exception as e:
+        return f"An error occurred during transcription: {str(e)}"
 
+# Modify the existing respond function to handle both PDF and web search
+def respond(message, history, model, temperature, num_calls, use_web_search, selected_docs, audio_input):
+    if audio_input is not None:
+        transcription = transcribe_audio(audio_input)
+        message = f"Transcription: {transcription}\n\nUser query: {message}"
+    
     logging.info(f"User Query: {message}")
     logging.info(f"Model Used: {model}")
     logging.info(f"Selected Documents: {selected_docs}")
@@ -624,14 +627,6 @@ Write a detailed and complete response that answers the following user question:
         
         logging.info("Finished generating response")
 
-def transcribe(audio):
-    if audio is None:
-        return ""
-    
-    headers = {"Authorization": f"Bearer {huggingface_token}"}
-    response = requests.post(WHISPER_API_URL, headers=headers, data=audio)
-    return response.json().get("text", "")
-
 def vote(data: gr.LikeData):
     if data.liked:
         print(f"You upvoted this response: {data.value}")
@@ -659,7 +654,6 @@ def display_documents():
         label="Select documents to query or delete"
     )
 
-# Define initial conversation
 def initial_conversation():
     return [
         (None, "Welcome! I'm your AI assistant for web search and PDF analysis. Here's how you can use me:\n\n"
@@ -682,64 +676,60 @@ use_web_search = gr.Checkbox(label="Use Web Search", value=False)
 
 custom_placeholder = "Ask a question (Note: You can toggle between Web Search and PDF Chat in Additional Inputs below)"
 
-# Create the Gradio interface
-# Create the Gradio interface
-with gr.Blocks() as demo:
-    gr.Markdown("# AI-powered PDF Chat and Web Search Assistant")
-    
-    audio_input = gr.Audio(sources=["microphone"], type="numpy", label="Speak your query")
-    
-    chatbot = gr.Chatbot(
+# Update the demo interface
+# Update the Gradio interface
+demo = gr.ChatInterface(
+    respond,
+    additional_inputs_accordion=gr.Accordion(label="⚙️ Parameters", open=True, render=False),
+    additional_inputs=[
+        gr.Dropdown(choices=MODELS, label="Select Model", value=MODELS[3]),
+        gr.Slider(minimum=0.1, maximum=1.0, value=0.2, step=0.1, label="Temperature"),
+        gr.Slider(minimum=1, maximum=5, value=1, step=1, label="Number of API Calls"),
+        gr.Checkbox(label="Use Web Search", value=True),
+        gr.CheckboxGroup(label="Select documents to query")        
+    ],
+    title="AI-powered PDF Chat and Web Search Assistant",
+    description="Chat with your PDFs or use web search to answer questions.",
+    theme=gr.themes.Soft(
+        primary_hue="orange",
+        secondary_hue="amber",
+        neutral_hue="gray",
+        font=[gr.themes.GoogleFont("Exo"), "ui-sans-serif", "system-ui", "sans-serif"]
+    ).set(
+        body_background_fill_dark="#0c0505",
+        block_background_fill_dark="#0c0505",
+        block_border_width="1px",
+        block_title_background_fill_dark="#1b0f0f",
+        input_background_fill_dark="#140b0b",
+        button_secondary_background_fill_dark="#140b0b",
+        border_color_accent_dark="#1b0f0f",
+        border_color_primary_dark="#1b0f0f",
+        background_fill_secondary_dark="#0c0505",
+        color_accent_soft_dark="transparent",
+        code_background_fill_dark="#140b0b"
+    ),
+    css=css,
+    examples=[
+        ["Tell me about the contents of the uploaded PDFs."],
+        ["What are the main topics discussed in the documents?"],
+        ["Can you summarize the key points from the PDFs?"],
+        ["What's the latest news about artificial intelligence?"]
+    ],
+    cache_examples=False,
+    analytics_enabled=False,
+    textbox=gr.Textbox(placeholder="Ask a question about the uploaded PDFs or any topic", container=False, scale=7),
+    chatbot = gr.Chatbot(  
         show_copy_button=True,
         likeable=True,
         layout="bubble",
         height=400,
         value=initial_conversation()
     )
-    
-    chat_interface = gr.ChatInterface(
-        fn=process_input,
-        chatbot=chatbot,
-        additional_inputs=[
-            gr.Dropdown(choices=MODELS, label="Select Model", value=MODELS[3]),
-            gr.Slider(minimum=0.1, maximum=1.0, value=0.2, step=0.1, label="Temperature"),
-            gr.Slider(minimum=1, maximum=5, value=1, step=1, label="Number of API Calls"),
-            gr.Checkbox(label="Use Web Search", value=True),
-            gr.CheckboxGroup(label="Select documents to query"),
-            audio_input
-        ],
-        title="AI-powered PDF Chat and Web Search Assistant",
-        description="Chat with your PDFs, use web search to answer questions, or speak your query.",
-        theme=gr.themes.Soft(
-            primary_hue="orange",
-            secondary_hue="amber",
-            neutral_hue="gray",
-            font=[gr.themes.GoogleFont("Exo"), "ui-sans-serif", "system-ui", "sans-serif"]
-        ).set(
-            body_background_fill_dark="#0c0505",
-            block_background_fill_dark="#0c0505",
-            block_border_width="1px",
-            block_title_background_fill_dark="#1b0f0f",
-            input_background_fill_dark="#140b0b",
-            button_secondary_background_fill_dark="#140b0b",
-            border_color_accent_dark="#1b0f0f",
-            border_color_primary_dark="#1b0f0f",
-            background_fill_secondary_dark="#0c0505",
-            color_accent_soft_dark="transparent",
-            code_background_fill_dark="#140b0b"
-        ),
-        css=css,
-        examples=[
-            ["Tell me about the contents of the uploaded PDFs."],
-            ["What are the main topics discussed in the documents?"],
-            ["Can you summarize the key points from the PDFs?"],
-            ["What's the latest news about artificial intelligence?"]
-        ],
-        cache_examples=False,
-        analytics_enabled=False,
-        textbox=gr.Textbox(placeholder="Ask a question about the uploaded PDFs or any topic", container=False, scale=7),
-    )
+)
 
+# Add file upload functionality
+# Add file upload functionality
+with demo:
     gr.Markdown("## Upload and Manage PDF Documents")
     with gr.Row():
         file_input = gr.Files(label="Upload your PDF documents", file_types=[".pdf"])
@@ -750,23 +740,25 @@ with gr.Blocks() as demo:
     update_output = gr.Textbox(label="Update Status")
     delete_button = gr.Button("Delete Selected Documents")
     
-    # Set up event handlers for document management
+    # Update both the output text and the document selector
     update_button.click(
         update_vectors, 
         inputs=[file_input, parser_dropdown], 
-        outputs=[update_output, chat_interface.additional_inputs[-2]]  # Assuming the CheckboxGroup is the second-to-last additional input
+        outputs=[update_output, demo.additional_inputs[-1]]  # Use the CheckboxGroup from additional_inputs
     )
     
+    # Add the refresh button functionality
     refresh_button.click(
         refresh_documents, 
         inputs=[], 
-        outputs=[chat_interface.additional_inputs[-2]]  # Assuming the CheckboxGroup is the second-to-last additional input
+        outputs=[demo.additional_inputs[-1]]  # Use the CheckboxGroup from additional_inputs
     )
     
+    # Add the delete button functionality
     delete_button.click(
         delete_documents,
-        inputs=[chat_interface.additional_inputs[-2]],  # Assuming the CheckboxGroup is the second-to-last additional input
-        outputs=[update_output, chat_interface.additional_inputs[-2]]
+        inputs=[demo.additional_inputs[-1]],  # Use the CheckboxGroup from additional_inputs
+        outputs=[update_output, demo.additional_inputs[-1]]
     )
 
     gr.Markdown(
@@ -783,4 +775,4 @@ with gr.Blocks() as demo:
     )
 
 if __name__ == "__main__":
-    demo.launch(share=True)
+    demo.launch(share=True)