Simplified app.py for HF Spaces

app.py

@@ -4,30 +4,24 @@ Upload PDF → Chat with AI → Export Code
 """
 
 import os
-import tempfile
-from pathlib import Path
-
 import gradio as gr
 from agno.agent import Agent
 from agno.knowledge.embedder.openai import OpenAIEmbedder
 from agno.knowledge.knowledge import Knowledge
 from agno.knowledge.reader.pdf_reader import PDFReader
-from agno.vectordb.lancedb import LanceDb, SearchType
+from agno.vectordb.lancedb import LanceDb
 from agno.models.neosantara import Neosantara
 
-# Environment
 NEOSANTARAA_API_KEY = os.getenv("NEOSANTARAA_API_KEY", "")
 MODEL_ID = os.getenv("MODEL_ID", "grok-4.1-fast")
 EMBEDDER_ID = os.getenv("EMBEDDER_ID", "nusa-embedding-0001")
 
-# Global state
 knowledge_base = None
 agent = None
-uploaded_file_url = None
+current_file_path = None
 
 
 def get_embedder():
-    """Get embedder configured for Neosantara."""
     return OpenAIEmbedder(
         id=EMBEDDER_ID,
         api_key=NEOSANTARAA_API_KEY,
@@ -35,66 +29,50 @@ def get_embedder():
     )
 
 
-def upload_pdf(file):
-    """Handle PDF upload and create knowledge base."""
-    global knowledge_base, agent, uploaded_file_url
+def process_pdf(file_obj):
+    global knowledge_base, agent, current_file_path
 
-    if file is None:
+    if file_obj is None:
         return "Please upload a PDF file.", None
 
-    # Handle Gradio file upload (can be string path or UploadFile object)
-    if hasattr(file, "name"):
-        # Gradio UploadFile object
-        tmp_path = file.name
-    elif isinstance(file, str):
-        # Already a path
-        tmp_path = file
-    else:
-        # Raw bytes - save to temp
-        with tempfile.NamedTemporaryFile(delete=False, suffix=".pdf") as tmp:
-            tmp.write(file)
-            tmp_path = tmp.name
-
-    # For local files, we'll use file path directly
-    # In production, you'd upload to cloud storage
-    uploaded_file_url = tmp_path
-
     try:
-        # Create embedder
+        if hasattr(file_obj, "name"):
+            file_path = file_obj.name
+        else:
+            file_path = str(file_obj)
+
+        current_file_path = file_path
+
         embedder = get_embedder()
 
-        # Create vector DB
         vector_db = LanceDb(
             table_name="rag_demo", uri="/tmp/lancedb", embedder=embedder
         )
 
-        # Create knowledge base from local file
-        knowledge = Knowledge(vector_db=vector_db, reader=PDFReader(chunk=True))
-        knowledge.insert(path=tmp_path)
+        knowledge_base = Knowledge(vector_db=vector_db, reader=PDFReader(chunk=True))
+        knowledge_base.insert(path=file_path)
 
-        # Create agent
         agent = Agent(
             model=Neosantara(
                 id=MODEL_ID,
                 api_key=NEOSANTARAA_API_KEY,
                 base_url="https://api.neosantara.xyz/v1",
             ),
-            knowledge=knowledge,
+            knowledge=knowledge_base,
             search_knowledge=True,
             markdown=True,
         )
 
         return (
-            f"✅ PDF loaded successfully!",
-            tmp_path,
+            "✅ PDF loaded successfully! You can now chat with the document.",
+            file_path,
         )
 
     except Exception as e:
         return f"❌ Error: {str(e)}", None
 
 
-def chat(message, history):
-    """Handle chat with RAG agent."""
+def chat_fn(message, history):
     global agent
 
     if agent is None:
@@ -107,115 +85,93 @@ def chat(message, history):
         return history + [(message, f"Error: {str(e)}")]
 
 
-def export_code():
-    """Generate ready-to-use Python code."""
-    global uploaded_file_url, MODEL_ID, EMBEDDER_ID
+def generate_code():
+    global current_file_path
 
-    if uploaded_file_url is None:
+    if current_file_path is None:
         return "Please upload a PDF first!"
 
-    code = f'''# Neosantara RAG Agent
+    return f'''# Neosantara RAG Agent
 # Generated from RAG Demo Tool
 
 from agno.agent import Agent
-from agno.knowledge.pdf import PDFKnowledgeBase
-from agno.vectordb.lancedb import LanceDb, SearchType
-from agno.embedder.openai import OpenAIEmbedder
+from agno.knowledge.knowledge import Knowledge
+from agno.knowledge.reader.pdf_reader import PDFReader
+from agno.vectordb.lancedb import LanceDb
+from agno.knowledge.embedder.openai import OpenAIEmbedder
 from agno.models.neosantara import Neosantara
 
-# Configuration
-NEOSANTARA_API_KEY = "your-api-key-here"
+NEOSANTARAA_API_KEY = "your-api-key-here"
 MODEL_ID = "{MODEL_ID}"
 EMBEDDER_ID = "{EMBEDDER_ID}"
 
-# Setup Embedder
 embedder = OpenAIEmbedder(
     id=EMBEDDER_ID,
-    api_key=NEOSANTARA_API_KEY,
+    api_key=NEOSANTARAA_API_KEY,
     base_url="https://api.neosantara.xyz/v1"
 )
 
-# Setup Vector DB
 vector_db = LanceDb(
     table_name="my_doc",
     uri="/tmp/lancedb",
     embedder=embedder
 )
 
-# Create Knowledge Base
-knowledge_base = PDFKnowledgeBase(
-    path="{uploaded_file_url}",
-    vector_db=vector_db
+knowledge = Knowledge(
+    vector_db=vector_db,
+    reader=PDFReader(chunk=True)
 )
-knowledge_base.load()
+knowledge.insert(path="{current_file_path}")
 
-# Create Agent
 agent = Agent(
     model=Neosantara(
         id=MODEL_ID,
-        api_key=NEOSANTARA_API_KEY,
+        api_key=NEOSANTARAA_API_KEY,
         base_url="https://api.neosantara.xyz/v1"
     ),
-    knowledge=knowledge_base,
+    knowledge=knowledge,
     search_knowledge=True,
     markdown=True
 )
 
-# Chat!
 agent.print_response("Apa isi dokumen ini?")
 '''
-    return code
 
 
-def clear_session():
-    """Clear the session."""
-    global knowledge_base, agent, uploaded_file_url
+def clear_fn():
+    global knowledge_base, agent, current_file_path
     knowledge_base = None
     agent = None
-    uploaded_file_url = None
-    return None, [], "Session cleared. Please upload a PDF."
+    current_file_path = None
+    return None, [], "Session cleared. Upload a new PDF."
 
 
-# Build Gradio UI
 with gr.Blocks(title="Neosantara RAG Demo") as demo:
     gr.Markdown("# 📄 Neosantara RAG Demo")
-    gr.Markdown("Upload a PDF, chat with AI about it, and export the code!")
+    gr.Markdown("Upload a PDF, chat with AI, and export the code!")
 
     with gr.Row():
         with gr.Column(scale=1):
-            file_input = gr.File(
-                label="Upload PDF", file_count="single", file_types=[".pdf"]
-            )
-            upload_btn = gr.Button("Process PDF", variant="primary")
-            status_output = gr.Textbox(label="Status", interactive=False)
+            file_input = gr.File(label="Upload PDF", file_types=[".pdf"])
+            process_btn = gr.Button("Process PDF", variant="primary")
+            status = gr.Textbox(label="Status", interactive=False)
             clear_btn = gr.Button("Clear Session")
 
         with gr.Column(scale=2):
             chatbot = gr.Chatbot(label="Chat", height=400)
-            msg_input = gr.Textbox(
-                label="Message", placeholder="Ask about your document...", scale=3
-            )
+            msg = gr.Textbox(label="Message", placeholder="Ask about your document...")
             send_btn = gr.Button("Send", variant="primary")
+            export_btn = gr.Button("📋 Export Code")
+            code_output = gr.Code(label="Generated Code", language="python")
 
-            with gr.Row():
-                export_btn = gr.Button("📋 Export Code")
-
-            code_output = gr.Code(label="Generated Code", language="python", lines=20)
+    process_btn.click(fn=process_pdf, inputs=[file_input], outputs=[status, chatbot])
 
-    # Events
-    upload_btn.click(
-        fn=upload_pdf, inputs=[file_input], outputs=[status_output, chatbot]
-    )
-
-    send_btn.click(fn=chat, inputs=[msg_input, chatbot], outputs=[chatbot])
-    msg_input.submit(fn=chat, inputs=[msg_input, chatbot], outputs=[chatbot])
+    send_btn.click(fn=chat_fn, inputs=[msg, chatbot], outputs=[chatbot])
+    msg.submit(fn=chat_fn, inputs=[msg, chatbot], outputs=[chatbot])
 
-    export_btn.click(fn=export_code, inputs=[], outputs=[code_output])
+    export_btn.click(fn=generate_code, inputs=[], outputs=[code_output])
 
-    clear_btn.click(
-        fn=clear_session, inputs=[], outputs=[file_input, chatbot, status_output]
-    )
+    clear_btn.click(fn=clear_fn, inputs=[], outputs=[file_input, chatbot, status])
 
-# Launch
 if __name__ == "__main__":
-    demo.launch(server_name="0.0.0.0", server_port=7860, share=True)
+    demo.launch()