fulvia/modules/url.py

#!/usr/bin/env python3
"""
URL parsing.
"""
import re
from urllib.parse import urlparse
from html.parser import HTMLParser

import requests

from module import hook

HEADERS = {"User-Agent": "bix nood gimme the title", "Range": "bytes=0-4096"}

@hook(True)
def title_auto(bot, trigger):
	"""
	Automatically show titles for URLs. For shortened URLs/redirects, find
	where the URL redirects to and show the title for that.
	"""
	if "http" not in trigger.group(0):
		return
	url_finder = re.compile(r"((?:http|https)(?::\/\/\S+))", re.IGNORECASE)

	urls = re.findall(url_finder, trigger.group(0))
	if len(urls) == 0:
		return

	for url in urls:
		broken = False
		for key in bot.url_callbacks:
			if key in url:
				bot.url_callbacks[key](bot, url)
				broken = True
		if broken:
			continue
		try:
			res = requests.get(url, headers=HEADERS, verify=True, timeout=10)
		except requests.exceptions.ConnectionError:
			continue
		try:
			res.raise_for_status()
		except:
			continue
		if not res.headers.get("Content-Type"):
			continue
		if not res.headers.get("Content-Type").startswith("text/html"):
			continue
		if res.text.find("<title>") == -1:
			continue
		title = res.text[res.text.find("<title>")+7:res.text.find("</title>")]
		title = HTMLParser().unescape(title)
		title = title.replace("\n","").strip()
		title = title.encode("latin-1").decode("utf-8")
		hostname = urlparse(url).hostname
		bot.msg(f"[ \x0310{title} \x03] - \x0304{hostname}")