Update app.py

app.py

@@ -394,7 +394,7 @@ def format_json(inp):
     return out_json
 
 
-    
+
 def summarize(inp,history,report_check,chart_check,data=None,files=None,directory=None,url=None,pdf_url=None,pdf_batch=None):
     json_box=[]
     error_box=""
@@ -404,6 +404,124 @@ def summarize(inp,history,report_check,chart_check,data=None,files=None,director
     history = [(inp,"Working on it...")] 
     yield "",history,error_box,json_box
 
+    if pdf_batch.startswith("http"):
+        c=0
+        data=""
+        for i in str(pdf_batch):
+            if i==",":
+                c+=1
+        print (f'c:: {c}')
+
+        try:
+            for i in range(c+1):
+                batch_url = pdf_batch.split(",",c)[i]
+                bb = read_pdf_online(batch_url)
+                data=f'{data}\nFile Name URL ({batch_url}):\n{bb}'
+        except Exception as e:
+            print(e)
+            #data=f'{data}\nError reading URL ({batch_url})'
+            
+    if directory:
+        for ea in directory:
+            print(ea)
+        
+    if pdf_url.startswith("http"):
+        print("PDF_URL")
+        out = read_pdf_online(pdf_url)
+        data=out
+    if url.startswith("http"):
+        val, out = find_all(url)
+        if not val:
+            data="Error"
+            rawp = str(out)
+        else:
+            data=out
+    if files:
+        for i, file in enumerate(files):
+            try: 
+                print (file)
+                if file.endswith(".pdf"):
+                    zz=read_pdf(file)
+                    print (zz)
+                    data=f'{data}\nFile Name ({file}):\n{zz}'
+                elif file.endswith(".txt"):
+                    zz=read_txt(file)
+                    print (zz)
+                    data=f'{data}\nFile Name ({file}):\n{zz}'                
+            except Exception as e:
+                data=f'{data}\nError opening File Name ({file})'                
+                print (e) 
+    if data != "Error" and data != "":
+        print(inp)
+        out = str(data)
+        rl = len(out)
+        print(f'rl:: {rl}')
+        c=1
+        for i in str(out):
+            if i == " " or i=="," or i=="\n":
+                c +=1
+        print (f'c:: {c}')
+
+        json_out = sort_fn(out)
+     yield "", history,error_box,json_out,json_out
+
+
+def search_fn(prompt,history,report_check,chart_check,data):
+    if data != "Error" and data != "":
+        print(inp)
+        out = str(data)
+        rl = len(out)
+        print(f'rl:: {rl}')
+        c=1
+        for i in str(out):
+            if i == " " or i=="," or i=="\n":
+                c +=1
+        print (f'c:: {c}')
+
+        #json_start = sort_fn(out)
+        json_out = find_keyword_fn(c,prompt,data)
+        print(f'JSON_BOX:: {json_out}')
+        out = str(json_out)
+        if report_check:
+            rl = len(out)
+            print(f'rl:: {rl}')
+            c=1
+            for i in str(out):
+                if i == " " or i=="," or i=="\n":
+                    c +=1
+            print (f'c2:: {c}')
+            rawp = compress_data_og(c,inp,out)
+        else:
+            rawp = out
+        try:
+            json_out=format_json(json_out)
+            print("JSON FORMATTED")
+        except Exception as e:
+            print (e)
+        if chart_check:
+            print (f"making chart from ::: {rawp}")
+            error_box = get_chart(str(json_out))
+            print(error_box)
+    else:
+        rawp = "Provide a valid data source"
+    #print (rawp)
+    #print (f'out:: {out}')
+    #history += "observation: the search results are:\n {}\n".format(out)
+    #task = "complete?"
+    history.clear()
+    history.append((inp,rawp))
+    yield "", history,error_box,json_out    
+
+
+def summarize_OG(inp,history,report_check,chart_check,data=None,files=None,directory=None,url=None,pdf_url=None,pdf_batch=None):
+    json_box=[]
+    error_box=""
+    if inp == "":
+        inp = "Process this data"
+    history.clear()
+    history = [(inp,"Working on it...")] 
+    yield "",history,error_box,json_box
+
     if pdf_batch.startswith("http"):
         c=0
         data=""
@@ -496,11 +614,19 @@ def summarize(inp,history,report_check,chart_check,data=None,files=None,director
     history.append((inp,rawp))
     yield "", history,error_box,json_out
 
+
+
+
+
+
+
+    
 #################################
 def clear_fn():
     return "",[(None,None)]
 
 with gr.Blocks() as app:
+    doc_state=gr.State()
     gr.HTML("""<center><h1>Mixtral 8x7B TLDR Summarizer + Web</h1><h3>Summarize Data of unlimited length</h3>""")
     chatbot = gr.Chatbot(label="Mixtral 8x7B Chatbot",show_copy_button=True)
     with gr.Row():
@@ -509,7 +635,8 @@ with gr.Blocks() as app:
         with gr.Column(scale=1):
             report_check=gr.Checkbox(label="Return Report", value=False)
             chart_check=gr.Checkbox(label="Return Chart", value=False)
-            button=gr.Button()
+            load_button=gr.Button("Load")
+            button=gr.Button("Run")
         
         #models_dd=gr.Dropdown(choices=[m for m in return_list],interactive=True)
     with gr.Row():
@@ -535,7 +662,8 @@ with gr.Blocks() as app:
     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])
-    
+    find=button.click(search_fn,[prompt,chatbot,report_check,chart_check,doc_state],[prompt,chatbot,e_box,json_out])
+    go=load_button.click(summarize,[prompt,chatbot,report_check,chart_check,data,file,directory,url,pdf_url,pdf_batch],[prompt,chatbot,e_box,json_out,doc_state])
+    #find=button.click()
     stop_button.click(None,None,None,cancels=[go])
 app.queue(default_concurrency_limit=20).launch(show_api=False)