From 972ce1d0fe4426baf882f004ed824b7d11d24545 Mon Sep 17 00:00:00 2001
From: tastytea <tastytea@tastytea.de>
Date: Mon, 24 May 2021 16:37:30 +0200
Subject: [PATCH] Don't strip headlines.

---
 man/epubgrep.1.adoc | 3 ++-
 src/search.cpp      | 5 +++++
 2 files changed, 7 insertions(+), 1 deletion(-)

diff --git a/man/epubgrep.1.adoc b/man/epubgrep.1.adoc
index 41f8c08..a90a01c 100644
--- a/man/epubgrep.1.adoc
+++ b/man/epubgrep.1.adoc
@@ -72,7 +72,8 @@ epubgrep does not operate on lines, but on whole files. This means you can
 search for text spanning multiple lines. All newlines will be replaced by
 spaces.
 
-HTML will be stripped unless *--raw* is specified.
+HTML will be stripped (except headlines) and newlines will be removed unless
+*--raw* is specified.
 
 === Configuration
 
diff --git a/src/search.cpp b/src/search.cpp
index 71ff97d..8061566 100644
--- a/src/search.cpp
+++ b/src/search.cpp
@@ -91,6 +91,11 @@ void cleanup_text(std::string &text)
 {
     for (size_t pos{}; pos != std::string::npos; pos = text.find('<', pos))
     {
+        // Don't strip headlines. We need them later on.
+        if (text[pos + 1] == 'h' || text.substr(pos + 1, 2) == "/h")
+        {
+            continue;
+        }
         text.erase(pos, text.find('>', pos) + 1 - pos);
     }