update app.py

app.py

@@ -11,7 +11,7 @@ from langchain.text_splitter import RecursiveCharacterTextSplitter
 
 #Persistent storage is mounted to /data
 DB_URL  = "./data/lancedb" 
-
+TABLE_NAME = "pdf_table"
 
 # define schema for table with embedding api
 
@@ -28,7 +28,7 @@ def lanceDBConnection(df):
     """
     db = lancedb.connect(DB_URL)
     table = db.create_table(
-        "scratch",
+        TABLE_NAME,
         schema=TextModel,
         mode="overwrite",
     )
@@ -48,14 +48,14 @@ def get_pdf(file):
     
     text_splitter=RecursiveCharacterTextSplitter(chunk_size=1500,chunk_overlap=50)
     docs_sp=text_splitter.split_documents(documents)
-    texts = [chunk.page_content for chunk in docs_sp] 
+    texts = [chunk.page_content for chunk in docs_sp]
     df = pd.DataFrame({"text": texts})
     table = lanceDBConnection(df)
     return f"PDF uploaded successfully. Total number of documents: {len(df)}"
     
 def get_nearest_neighbours(query):
     db = lancedb.connect(DB_URL)
-    table = db.open_table("scratch")
+    table = db.open_table(TABLE_NAME)
     result = table.search(query).limit(3).to_list()
     context = [r["text"] for r in result]
     return context
@@ -69,11 +69,11 @@ pdf_interface=gr.Interface(
 question_interface=gr.Interface(
       fn=get_nearest_neighbours,
       inputs=[gr.Textbox(label="Enter your question")],
-      outputs=["text"]*3
+      outputs=[gr.Textbox(label="Relevant chunk")]*3
   )
 
 demo=gr.TabbedInterface(
     interface_list=[pdf_interface, question_interface],
     tab_names=["Upload Pdfs","Get relevant chunks"],
-    title="Your study Partner")
+    title="Save PDF chunks into LanceDB on persitent storage")
 demo.launch(share=True)