Omnibus commited on
Commit
13951ed
·
1 Parent(s): ca288cd

Update app.py

Browse files
Files changed (1) hide show
  1. app.py +12 -18
app.py CHANGED
@@ -7,30 +7,24 @@ from IPython.display import IFrame
7
 
8
  def scrape(instring):
9
 
10
- r = requests.get(instring)
11
-
12
- #soup = bs(r.content,'html5lib')
13
- #soup.prettify(encoding=None, formatter="minimal")
14
- #print(soup)
15
- #divs = soup.find_all("li")
16
- #for dive in divs:
17
-
18
- #mydivs = soup.find(f"{root}", {f"{trunk}": f"{branch}"})
19
- #divs = soup.find_all("li",{"class":"ng-border-bottom-neutral-30 ng-font-neutral-60 ng-flex ng-space-between ng-pa24"})
20
 
21
- #div data-react-class="components/OrderedList" data-react-props="
 
 
 
22
  try:
23
- out = r.content
24
- #print(f'DIVS: {divs}')
25
- #print (mydivs)
26
- #out = soup.get_text()
27
- #print ("Divs")
28
  except Exception:
29
  #out=copy.copy(soup)
30
  print ("No Divs")
31
 
32
- out = IFrame(src={instring}, width=700, height=600)
33
- return gr.HTML.update(f'''<iframe src={instring}></iframe''')
34
  with gr.Blocks() as app:
35
  inp=gr.Textbox()
36
  go_btn = gr.Button()
 
7
 
8
  def scrape(instring):
9
 
10
+ #r = requests.get(instring)
11
+
12
+ url = f'{instring}'
13
+ r = requests.get(url, stream=True)
 
 
 
 
 
 
14
 
15
+ with open('/tmp/metadata.pdf', 'wb') as fd:
16
+ for chunk in r.iter_content(chunk_size):
17
+ fd.write(chunk)
18
+
19
  try:
20
+ out = r.raw
21
+
 
 
 
22
  except Exception:
23
  #out=copy.copy(soup)
24
  print ("No Divs")
25
 
26
+ #out = IFrame(src={instring}, width=700, height=600)
27
+ return gr.HTML.update(f'''<iframe src={out}></iframe''')
28
  with gr.Blocks() as app:
29
  inp=gr.Textbox()
30
  go_btn = gr.Button()