From 7643a962bd29504bf48a23383109b7642700a349 Mon Sep 17 00:00:00 2001
From: jesopo <github@lolnerd.net>
Date: Thu, 25 Apr 2019 15:58:58 +0100
Subject: [PATCH] Refuse to get the title for any url that points locall

---
 modules/title.py  | 10 +++++++++-
 src/utils/http.py | 32 ++++++++++++++++++++++++++++++--
 2 files changed, 39 insertions(+), 3 deletions(-)

diff --git a/modules/title.py b/modules/title.py
index ddc0a50f..f1ea2b30 100644
--- a/modules/title.py
+++ b/modules/title.py
@@ -1,4 +1,4 @@
-import hashlib, re
+import hashlib, re, urllib.parse
 from src import EventManager, ModuleManager, utils
 
 @utils.export("channelset", {"setting": "auto-title",
@@ -14,6 +14,14 @@ class Module(ModuleManager.BaseModule):
             ).hexdigest()
 
     def _get_title(self, url):
+        if not urllib.parse.urlparse(url).scheme:
+            url = "http://%s" % url
+
+        hostname = urllib.parse.urlparse(url).hostname
+        if utils.http.is_localhost(hostname):
+            self.log.warn("tried to get title of localhost: %s", [url])
+            return None
+
         try:
             page = utils.http.request(url, soup=True)
         except utils.http.HTTPWrongContentTypeException:
diff --git a/src/utils/http.py b/src/utils/http.py
index 19f7fc20..09a47607 100644
--- a/src/utils/http.py
+++ b/src/utils/http.py
@@ -1,6 +1,7 @@
-import re, signal, traceback, typing, urllib.error, urllib.parse
+import ipaddress, re, signal, socket, traceback, typing
+import urllib.error, urllib.parse
 import json as _json
-import bs4, requests
+import bs4, netifaces, requests
 from src import utils
 
 REGEX_URL = re.compile("https?://\S+", re.I)
@@ -87,3 +88,30 @@ def request(url: str, method: str="GET", get_params: dict={},
 def strip_html(s: str) -> str:
     return bs4.BeautifulSoup(s, "lxml").get_text()
 
+def resolve_hostname(hostname: str) -> typing.List[str]:
+    try:
+        addresses = socket.getaddrinfo(hostname, None, 0, socket.SOCK_STREAM)
+    except:
+        return []
+    return [address[-1][0] for address in addresses]
+
+def is_ip(addr: str) -> bool:
+    try:
+        ipaddress.ip_address(addr)
+    except ValueError:
+        return False
+    return True
+
+def is_localhost(hostname: str) -> bool:
+    if is_ip(hostname):
+        ips = [ipaddress.ip_address(hostname)]
+    else:
+        ips = [ipaddress.ip_address(ip) for ip in resolve_hostname(hostname)]
+
+    for interface in netifaces.interfaces():
+        links = netifaces.ifaddresses(interface)
+        for link in links[netifaces.AF_INET]+links[netifaces.AF_INET6]:
+            address = ipaddress.ip_address(link["addr"].split("%", 1)[0])
+            if address in ips:
+                return True
+    return False