Send user agent in URL parser requests.

[plomlombot-irc.git] / plomlombot.py

diff --git a/plomlombot.py b/plomlombot.py
index 5fcd517a95f563cfba7a00146d1e6359780a0472..6031d50b3167eb62dfbe16865e1c8278efa44e41 100755 (executable)
--- a/plomlombot.py
+++ b/plomlombot.py
@@ -7,6 +7,7 @@ import select
 import time
 import re
 import urllib.request
+import http.client
 import html
 
 # Defaults, may be overwritten by command line arguments.
@@ -99,13 +100,16 @@ def lineparser_loop(io, nickname):
     def act_on_privmsg(tokens):
 
         def url_check(msg):
-            matches = re.findall("(https?://[^\s]+)", msg)
+            matches = re.findall("(https?://[^\s>]+)", msg)
             for i in range(len(matches)):
                 url = matches[i]
+                request = urllib.request.Request(url, headers={
+                    "User-Agent": "plomlombot"
+                })
                 try:
-                    webpage = urllib.request.urlopen(url, timeout=15)
+                    webpage = urllib.request.urlopen(request, timeout=15)
                 except (urllib.error.HTTPError, urllib.error.URLError,
-                        UnicodeError) as error:
+                        UnicodeError, http.client.BadStatusLine) as error:
                     print("TROUBLE FOLLOWING URL: " + str(error))
                     continue
                 charset = webpage.info().get_content_charset()