linkify/hashtagify

utils.py

# -*- coding: utf-8 -*-
import re

_linkify_re = re.compile(
    r"""(?i)\b((?:https?://|www\d{0,3}[.]|[a-z0-9.\-]+[.][a-z]{2,4}/)(?:[^\s()<>]+|\(([^\s()<>]+|(\([^\s()<>]+\)))*\))+(?:\(([^\s()<>]+|(\([^\s()<>]+\)))*\)|[^\s`!()\[\]{};:'".,<>?«»“”‘’]))""")

_hashtag_re = re.compile(r'(?: |^)#(\w{2,})', re.UNICODE)

def _link_replace(match):
    href = match.groups(1)[0]
    if not href.startswith('http'):
        href = 'http://' + href
    return '<a href="%s" target="_blank">%s</a>' % (href, match.groups(1)[0])


def linkify(content):
    return _linkify_re.sub(_link_replace, content)


def hashtagify(content, url_template):
    def _hashtag_replace(match):
        tag = match.groups(0)[0]
        url = url_template % tag
        # This might need escaping.
        return ' <a href="%s" target="_blank">#%s</a>' % (url, tag)
    return _hashtag_re.sub(_hashtag_replace, content)