From 6f11a6a00cdc7f96b91ded3ff313eabc57cc9f83 Mon Sep 17 00:00:00 2001
From: emkael <emkael@tlen.pl>
Date: Sun, 13 Jun 2021 12:50:35 +0200
Subject: Correctly detecting remote content encoding Remember kids, never
 parse HTML with regex.

---
 jfr_playoff/remote.py | 5 ++++-
 1 file changed, 4 insertions(+), 1 deletion(-)

(limited to 'jfr_playoff/remote.py')

diff --git a/jfr_playoff/remote.py b/jfr_playoff/remote.py
index 74d75b5..b918ea1 100644
--- a/jfr_playoff/remote.py
+++ b/jfr_playoff/remote.py
@@ -16,8 +16,11 @@ class RemoteUrl:
         if url not in cls.url_cache:
             request = requests.get(url)
             encoding_match = re.search(
-                'content=".*;( )?charset=(.*)"',
+                'content=".*;( )?charset=(.*?)"',
                 request.content, re.IGNORECASE)
+            PlayoffLogger.get('remote').debug(
+                'Content encoding: %s',
+                encoding_match.group(2))
             if encoding_match:
                 request.encoding = encoding_match.group(2)
             cls.url_cache[url] = request.text
-- 
cgit v1.2.3