From bdf9a866519df1307700c31ddf4584aff6ffb2f4 Mon Sep 17 00:00:00 2001
From: tastytea <tastytea@tastytea.de>
Date: Sat, 5 Jun 2021 15:20:40 +0200
Subject: [PATCH] Fix pagebreak-regex and range in which pagebreaks are
 searched.

---
 src/search.cpp | 9 +++++----
 1 file changed, 5 insertions(+), 4 deletions(-)

diff --git a/src/search.cpp b/src/search.cpp
index 99f69e6..75ebfc8 100644
--- a/src/search.cpp
+++ b/src/search.cpp
@@ -129,8 +129,9 @@ void cleanup_text(string &text)
 {
     static const boost::regex re_header_start{"<[hH][1-6]"};
     static const boost::regex re_header_end{"</[hH][1-6]"};
-    static const boost::regex re_pagebreak{".+pagebreak.+(title|aria-label)"
-                                           "=\"([[:alnum:]]+)\".*"};
+    static const boost::regex re_pagebreak{"[^>]+pagebreak[^>]+"
+                                           "(title|aria-label)"
+                                           "=\"([[:alnum:]]+)\""};
 
     size_t pos{};
     while ((pos = text.find('<', pos)) != string::npos)
@@ -149,8 +150,8 @@ void cleanup_text(string &text)
         {
             auto endpos{text.find('>')};
             boost::match_results<const char *> match;
-            if (boost::regex_search(text.substr(pos, endpos).data(), match,
-                                    re_pagebreak))
+            if (boost::regex_search(text.substr(pos, endpos - pos).data(),
+                                    match, re_pagebreak))
             {
                 // FIXME: -fsanitize=address is complaining about this. ↓ 🤷
                 //        Could not reproduce it.