Update app.py

app.py

@@ -48,14 +48,26 @@ audio_language_dict = {
 def index_text(text: str) -> str:
     global indexed_texts, indexed_embeddings
     try:
-        embedding = sentence_model.encode([text])[0]
-        indexed_texts.append(text)
-        indexed_embeddings.append(embedding)
-        return "Text indexed successfully."
+        # Split the text into sentences or smaller chunks
+        chunks = text.split('. ')
+        for chunk in chunks:
+            if chunk:
+                embedding = sentence_model.encode([chunk])[0]
+                indexed_texts.append(chunk)
+                indexed_embeddings.append(embedding)
+        return f"Text indexed successfully. Total indexed chunks: {len(indexed_texts)}"
     except Exception as e:
         return f"Error indexing text: {str(e)}"
 
-def find_most_similar(query: str, top_k: int = 1) -> list:
+def clear_index() -> str:
+    global indexed_texts, indexed_embeddings
+    indexed_texts.clear()
+    indexed_embeddings.clear()
+    return "Index cleared successfully. Ready for new indexing."
+
+def find_most_similar(query: str, top_k: int = 3) -> list:
+    if not indexed_texts:
+        return ["No indexed text available."]
     query_embedding = sentence_model.encode([query])[0]
     similarities = [np.dot(query_embedding, doc_embedding) for doc_embedding in indexed_embeddings]
     top_indices = np.argsort(similarities)[-top_k:][::-1]
@@ -65,7 +77,8 @@ def chat_with_context(question: str, model: str) -> str:
     if not indexed_texts:
         return "Please index some text first."
     
-    context = find_most_similar(question)[0]
+    relevant_contexts = find_most_similar(question, top_k=3)
+    context = " ".join(relevant_contexts)
     
     try:
         prompt = f"Context: {context}\n\nQuestion: {question}\n\nAnswer:"
@@ -116,6 +129,7 @@ with gr.Blocks() as iface:
     
     with gr.Row():
         index_button = gr.Button("Index")
+        clear_index_button = gr.Button("Clear Index")
     
     index_status = gr.Textbox(label="Indexing Status")
     
@@ -159,6 +173,7 @@ with gr.Blocks() as iface:
     
     convert_button.click(convert_text, inputs=[text_input, translation_lang_dropdown], outputs=translated_text)
     index_button.click(index_text, inputs=[translated_text], outputs=[index_status])
+    clear_index_button.click(clear_index, outputs=[index_status])
     use_chat.change(update_chat_visibility, inputs=[use_chat], outputs=[chat_group])
     chat_button.click(chat_with_context, inputs=[chat_input, chat_model], outputs=[chat_output])