pdf-reader / app.py
Omnibus's picture
Update app.py
563ca5d
raw
history blame
1.02 kB
import gradio as gr
#from bs4 import BeautifulSoup as bs
#import html5lib
#import copy
import requests
def scrape(instring):
r = requests.get(instring)
#soup = bs(r.content,'html5lib')
#soup.prettify(encoding=None, formatter="minimal")
#print(soup)
#divs = soup.find_all("li")
#for dive in divs:
#mydivs = soup.find(f"{root}", {f"{trunk}": f"{branch}"})
#divs = soup.find_all("li",{"class":"ng-border-bottom-neutral-30 ng-font-neutral-60 ng-flex ng-space-between ng-pa24"})
#div data-react-class="components/OrderedList" data-react-props="
try:
out = r.content
#print(f'DIVS: {divs}')
#print (mydivs)
#out = soup.get_text()
#print ("Divs")
except Exception:
#out=copy.copy(soup)
print ("No Divs")
#out =
return out
with gr.Blocks() as app:
inp=gr.Textbox()
go_btn = gr.Button()
outp = gr.HTML()
go_btn.click(scrape,inp,outp)
app.queue(concurrency_count=10).launch()