Advertisement
Not a member of Pastebin yet?
Sign Up,
it unlocks many cool features!
- with open('demo1.html','r') as file:
- data = file.read()
- title_pattern = r'<title>(.*)?</title>'
- titles = re.findall(title_pattern,data,re.IGNORECASE)
- print(f'Title: {titles}')
- head_pattern = r'<h[1-6]>(.*)?</h[1-6]>'
- heads = re.findall(head_pattern,data,re.IGNORECASE)
- print(f'Heading: {heads}')
- p_pattern = r'<p>(.*)?</p>'
- ps = re.findall(p_pattern,data,re.IGNORECASE)
- print(f'Paras: {ps}')
- link_pattern = r'<a\s+href=["\'](.*)?["\']>'
- links = re.findall(link_pattern,data,re.IGNORECASE)
- print(f'Links: {links}')
Advertisement
Add Comment
Please, Sign In to add comment
Advertisement
