Update app.py

app.py

@@ -127,7 +127,7 @@ def compress_data(c,purpose, task, history, result):
         resp = run_gpt(
             COMPRESS_DATA_PROMPT,
             stop_tokens=["observation:", "task:", "action:", "thought:"],
-            max_tokens=1024,
+            max_tokens=2048,
             seed=seed,
             purpose=purpose,
             task=task,
@@ -162,7 +162,7 @@ def compress_history(purpose, task, history):
     resp = run_gpt(
         COMPRESS_HISTORY_PROMPT,
         stop_tokens=["observation:", "task:", "action:", "thought:"],
-        max_tokens=512,
+        max_tokens=1024,
         seed=random.randint(1,1000000000),
         purpose=purpose,
         task=task,
@@ -176,7 +176,7 @@ def call_main(purpose, task, history, action_input, result):
     resp = run_gpt(
         FINDER,
         stop_tokens=["observation:", "task:"],
-        max_tokens=2096,
+        max_tokens=2048,
         seed=random.randint(1,1000000000),
         purpose=purpose,
         task=task,
@@ -230,58 +230,47 @@ def search_all(url):
 def find_all(purpose,task,history, url, result):
     return_list=[]
     print (url)
-    #if action_input in query.tasks:
     print (f"trying URL:: {url}")        
     try:
         if url != "" and url != None:    
-            #rawp = []
             out = []
             source = requests.get(url)
-            #source = urllib.request.urlopen(url).read()
-            soup = bs4.BeautifulSoup(source.content,'lxml')
-            # title of the page
-            print(soup.title)
-            # get attributes:
-            print(soup.title.name)
-            # get values:
-            #print(soup.title.string)
-            # beginning navigation:
-            #print(soup.title.parent.name)
-            #rawp.append([tag.name for tag in soup.find_all()] )
-            #print([tag.name for tag in soup.find_all()])
-            rawp=(f'RAW TEXT RETURNED: {soup.text}')
-            cnt=0
-            cnt+=len(rawp)
-            out.append(rawp)
-            out.append("HTML fragments: ")
-            q=("a","p","span","content","article")
-            for p in soup.find_all("a"):
-                #cnt+=len(p.string)
-                #cnt+=len(p)                
-                
-                out.append([{"LINK TITLE":p.get('title'),"URL":p.get('href'),"STRING":p.string}])
-            c=0
-            out = str(out)
-            rl = len(out)
-            #print (cnt)
-            print(f'rl:: {rl}')
-            #for ea in out:
-            for i in str(out):
-                if i == " " or i=="," or i=="\n" or i=="/" or i=="." or i=="<":
-                    c +=1
-            print (f'c:: {c}')
-            if c > MAX_HISTORY:
-                print("compressing...")
-                rawp = compress_data(c,purpose,task,out,result)  
+            if source.status_code ==200:
+                soup = bs4.BeautifulSoup(source.content,'lxml')
+               
+                rawp=(f'RAW TEXT RETURNED: {soup.text}')
+                cnt=0
+                cnt+=len(rawp)
+                out.append(rawp)
+                out.append("HTML fragments: ")
+                q=("a","p","span","content","article")
+                for p in soup.find_all("a"):
+                    out.append([{"LINK TITLE":p.get('title'),"URL":p.get('href'),"STRING":p.string}])
+                c=0
+                out = str(out)
+                rl = len(out)
+                print(f'rl:: {rl}')
+                for i in str(out):
+                    if i == " " or i=="," or i=="\n" or i=="/" or i=="." or i=="<":
+                        c +=1
+                print (f'c:: {c}')
+                if c > MAX_HISTORY:
+                    print("compressing...")
+                    rawp = compress_data(c,purpose,task,out,result)  
+                    result += rawp
+                else:
+                    rawp = out 
+                    
+                #print (rawp)
+                #print (f'out:: {out}')
+                history += "observation: the search results are:\n {}\n".format(rawp)
+                task = "compile report or complete?"
+                return "MAIN", None, history, task, result
             else:
-                rawp = out 
-            result += rawp
-                
-            print (rawp)
-            print (f'out:: {out}')
-            history += "observation: the search results are:\n {}\n".format(rawp)
-            task = "compile report or complete?"
-            return "MAIN", None, history, task, result
+                history += f"observation: That URL string returned an error: {source.status_code}, I should try a different URL string\n"
+                result="Still Working..."
+                return "MAIN", None, history, task, result
+            
         else: 
             history += "observation: An Error occured\nI need to trigger a search using the following syntax:\naction: SCRAPE_WEBSITE action_input=URL\n"
             return "MAIN", None, history, task, result
@@ -330,6 +319,7 @@ def run_action(purpose, task, history, action_name, action_input,result):
         return "MAIN", None, history, task, result
 
 def run(purpose,history):
+    yield "Searching..."
     task=None
     result=""
     #history = ""
@@ -357,7 +347,11 @@ def run(purpose,history):
             action_input,
             result
         )
-        yield result
+            
+        if not result:
+            yield "More Searching..."
+        else:
+            yield result
         if action_name == "COMPLETE":
             break
     return result
@@ -401,5 +395,4 @@ with gr.Blocks() as app:
     find_btn.click(find_it,[inp,q,num],[outp,outl])
     
 app.launch()
-
 '''