Update app.py

app.py

@@ -14,6 +14,7 @@ from constants import DEFAULT_GLOBAL_CTX
 
 from pingpong import PingPong
 from pingpong.context import CtxLastWindowStrategy
+from pingpong.context import InternetSearchStrategy, SimilaritySearcher
 
 TOKEN = os.getenv('HF_TOKEN')
 MODEL_ID = 'meta-llama/Llama-2-70b-chat-hf'
@@ -109,9 +110,8 @@ def reset_chat(idx, ld, state):
 
 def internet_search(ppmanager, serper_api_key, global_context, ctx_num_lconv, device="cpu"):
     internet_search_ppm = copy.deepcopy(ppm)
-    internet_search_prompt = f"My question is '{user_msg}'. Based on the conversation history, "
-                             f"give me an appropriate query to answer my question for google search. "
-                             f"You should not say more than query. You should not say any words except the query."
+    user_msg = internet_search_ppm.pingpongs[-1].ping
+    internet_search_prompt = f"My question is '{user_msg}'. Based on the conversation history, give me an appropriate query to answer my question for google search. You should not say more than query. You should not say any words except the query."
     
     internet_search_ppm.pingpongs[-1].ping = internet_search_prompt
     internet_search_prompt = build_prompts(internet_search_ppm, "", win_size=ctx_num_lconv)
@@ -161,7 +161,8 @@ async def chat_stream(
             yield "", uis, prompt, str(res)
     
     async for result in gen_text(
-        prompt, hf_model=MODEL_ID, hf_token=TOKEN,
+        search_prompt if internet_option else prompt, 
+        hf_model=MODEL_ID, hf_token=TOKEN,
         parameters={
             'max_new_tokens': res_mnts,
             'do_sample': res_sample,