Update app.py

app.py

@@ -4,8 +4,9 @@ import requests
 import bs4
 import lxml
 
-def find_it1(url,q1=None,q2=None):
+def find_it1(url,q=None,num=None):
     out = []
+    rawp = []
     source = urllib.request.urlopen(url).read()
     soup = bs4.BeautifulSoup(source,'lxml')
     # title of the page
@@ -23,21 +24,22 @@ def find_it1(url,q1=None,q2=None):
     # getting specific values:
     print(soup.p)
     print(soup.find_all('p'))
-    for paragraph in soup.find_all('p'):
-        
-        print(paragraph.string)
-        print(str(paragraph.text))
-        out.append(paragraph)
-        out.append(paragraph.string)
-        out.append(paragraph.text)
+    for p in soup.find_all(f'{q}'):
         
+        #print(paragraph.string)
+        #print(str(paragraph.text))
+        out.append(p)
+        out.append(p.string)
+        out.append(p.text)
+    print([str(tag) for tag in soup.find_all()])
+    rawp.append([str(tag) for tag in soup.find_all])
     for url in soup.find_all('a'):
         print(url.get('href'))
         
     print(soup.get_text())
 
     
-    return out
+    return rawp, out
     
 def find_it2(url):
     response = requests.get(url,a1=None,q2=None,q3=None)
@@ -54,8 +56,12 @@ def find_it2(url):
 with gr.Blocks() as app:
     with gr.Row():
         inp = gr.Textbox()
-        btn = gr.Button()
-    outp = gr.Textbox()
-    btn.click(find_it1,inp,outp)
+        q = gr.Textbox(value="p")
+        num = gr.Number(value=1)
+    btn = gr.Button()
+    with gr.Row():
+        rawp = gr.Textbox()
+        outp = gr.Textbox()
+    btn.click(find_it1,[inp,q,num],[rawp,outp])
 app.launch()