Spaces:
Runtime error
Runtime error
Update app.py
Browse files
app.py
CHANGED
|
@@ -16,37 +16,13 @@ def run_lora(prompt,site,start,end):
|
|
| 16 |
soup = BeautifulSoup(page.content,"html.parser")
|
| 17 |
import re
|
| 18 |
links = soup.findAll("a")
|
|
|
|
| 19 |
mLink=""
|
| 20 |
#for link in soup.find_all("a", "html.parser", parse_only=SoupStrainer('a')):
|
| 21 |
for link in soup.find_all("a","html.parser",href=re.compile("(?<=/url\?q=)(htt.*://.*)")):
|
| 22 |
print (re.split(":(?=http)",link["href"].replace("/url?q=","")))
|
| 23 |
mLink+=str(re.split(":(?=http)",link["href"].replace("/url?q=","").replace("'","")))
|
| 24 |
-
def extract_results(soup):
|
| 25 |
-
main = soup.select_one("#main")
|
| 26 |
|
| 27 |
-
res = []
|
| 28 |
-
for gdiv in main.select('.e9EfHf, .s6JM6d'):
|
| 29 |
-
res.append(extract_section(gdiv))
|
| 30 |
-
return res
|
| 31 |
-
|
| 32 |
-
def extract_section(gdiv):
|
| 33 |
-
# Getting our elements
|
| 34 |
-
title = gdiv.select_one('h3')
|
| 35 |
-
link = gdiv.select_one('a')
|
| 36 |
-
description = gdiv.find('.BNeawe')
|
| 37 |
-
return {
|
| 38 |
-
# Extract title's text only if text is found
|
| 39 |
-
'title': title.text if title else None,
|
| 40 |
-
|
| 41 |
-
'link': link['href'] if link else None,
|
| 42 |
-
'description': description.text if description else None
|
| 43 |
-
}
|
| 44 |
-
def extract_href(href):
|
| 45 |
-
url = urlparse(href)
|
| 46 |
-
query = parse_qs(url.query)
|
| 47 |
-
if not ('q' in query and query['q'] and len(query['q']) > 0):
|
| 48 |
-
return None
|
| 49 |
-
return query['q'][0]
|
| 50 |
|
| 51 |
with gr.Blocks() as app:
|
| 52 |
gr.HTML("""<html>
|
|
|
|
| 16 |
soup = BeautifulSoup(page.content,"html.parser")
|
| 17 |
import re
|
| 18 |
links = soup.findAll("a")
|
| 19 |
+
print(links)
|
| 20 |
mLink=""
|
| 21 |
#for link in soup.find_all("a", "html.parser", parse_only=SoupStrainer('a')):
|
| 22 |
for link in soup.find_all("a","html.parser",href=re.compile("(?<=/url\?q=)(htt.*://.*)")):
|
| 23 |
print (re.split(":(?=http)",link["href"].replace("/url?q=","")))
|
| 24 |
mLink+=str(re.split(":(?=http)",link["href"].replace("/url?q=","").replace("'","")))
|
|
|
|
|
|
|
| 25 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 26 |
|
| 27 |
with gr.Blocks() as app:
|
| 28 |
gr.HTML("""<html>
|