From 914750c4531877d7b86e997152ed37d965c79151 Mon Sep 17 00:00:00 2001
From: Nick Sweeting <git@nicksweeting.com>
Date: Tue, 19 Mar 2019 18:09:06 -0400
Subject: [PATCH] better title regex to match titles surrounded by newlines

---
 archivebox/util.py | 4 ++--
 1 file changed, 2 insertions(+), 2 deletions(-)

diff --git a/archivebox/util.py b/archivebox/util.py
index fb803732..cd7e9651 100644
--- a/archivebox/util.py
+++ b/archivebox/util.py
@@ -66,9 +66,9 @@ URL_REGEX = re.compile(
     re.IGNORECASE,
 )
 HTML_TITLE_REGEX = re.compile(
-    r'<title>'                         # start matching text after <title> tag
+    r'<title.*?>'                      # start matching text after <title> tag
     r'(.[^<>]+)',                      # get everything up to these symbols
-    re.IGNORECASE,
+    re.IGNORECASE | re.MULTILINE | re.DOTALL | re.UNICODE,
 )
 
 ### Checks & Tests