Update app.py

app.py

@@ -12,6 +12,8 @@ import datetime
 from pypdf import PdfReader
 import uuid
 #from query import tasks
+from gradio_client import Client
+
 from agent import (
     PREFIX,
     GET_CHART,
@@ -24,6 +26,16 @@ api=HfApi()
 
 client = InferenceClient("mistralai/Mixtral-8x7B-Instruct-v0.1")
 
+
+def sort_fn(inp):
+
+    client_sort = Client("Omnibus/sort_document")
+    sen,nouns = client_sort.predict(
+    		f"{inp}",	# str  in 'Paste Text' Textbox component
+    		api_name="/sort_doc"
+    )
+    return sen
+
 def find_all(url):
     return_list=[]
     print (url)
@@ -58,6 +70,47 @@ def find_all(url):
         
     return "MAIN", None, history, task
 
+FIND_KEYWORDS="""Find keywords from the dictionary of provided keywords that are relevant to the users query.
+Return the keyword:value pairs from the list in the form of a JSON file output.
+dictionary:
+{keywords}
+user query:
+
+"""
+
+def find_keyword_fn(c,inp,data):
+
+    
+    seed=random.randint(1,1000000000)
+    divr=int(c)/MAX_DATA
+    divi=int(divr)+1 if divr != int(divr) else int(divr)
+    chunk = int(int(c)/divr)
+    out = []
+    s=0
+    e=chunk
+    print(f'e:: {e}')
+    new_history=""
+    #task = f'Compile this data to fulfill the task: {task}, and complete the purpose: {purpose}\n'
+    for z in range(divi):
+        print(f's:e :: {s}:{e}')
+        
+        hist = history[s:e]
+        resp = run_gpt(
+            FIND_KEYWORDS,
+            stop_tokens=[],
+            max_tokens=8192,
+            seed=seed,
+            keywords=data,
+        ).strip("\n")    
+        out.append(resp)
+        #new_history = resp
+        print (resp)
+        #out+=resp
+        e=e+chunk
+        s=s+chunk    
+    return out 
+    
+
 def read_txt(txt_path):
     text=""
     with open(txt_path,"r") as f:
@@ -409,8 +462,10 @@ def summarize(inp,history,report_check,chart_check,data=None,files=None,director
             if i == " " or i=="," or i=="\n":
                 c +=1
         print (f'c:: {c}')
-        
-        json_out = compress_data(c,inp,out)  
+
+        json_start = sort_fn(out)
+        json_out = find_keywords_fn(c,inp,json_start)
+        #json_out = compress_data(c,inp,out)  
         #json_box.append(json_out)
 
         #json_object = json.dumps(eval(json_out), indent=4)
@@ -489,6 +544,9 @@ with gr.Blocks() as app:
     #text=gr.JSON()
     #inp_query.change(search_models,inp_query,models_dd)
     clear_btn.click(clear_fn,None,[prompt,chatbot])
+    
+    #go=button.click(summarize,[prompt,chatbot,report_check,chart_check,data,file,directory,url,pdf_url,pdf_batch],[prompt,chatbot,e_box,json_out])
     go=button.click(summarize,[prompt,chatbot,report_check,chart_check,data,file,directory,url,pdf_url,pdf_batch],[prompt,chatbot,e_box,json_out])
+    
     stop_button.click(None,None,None,cancels=[go])
 app.queue(default_concurrency_limit=20).launch(show_api=False)