From 8ab7d0f6556cb9585ecf24c3beeaeffa7cad65dc Mon Sep 17 00:00:00 2001
From: tastytea <tastytea@tastytea.de>
Date: Mon, 24 May 2021 17:18:10 +0200
Subject: [PATCH] Extract headlines.

---
 src/search.cpp | 26 ++++++++++++++++++++++++++
 src/search.hpp |  3 +++
 2 files changed, 29 insertions(+)

diff --git a/src/search.cpp b/src/search.cpp
index 8061566..f27eb63 100644
--- a/src/search.cpp
+++ b/src/search.cpp
@@ -71,6 +71,8 @@ std::vector<match> search(const fs::path &filepath, std::string_view regex,
         std::string::const_iterator begin{document.begin()};
         std::string::const_iterator end{document.end()};
         boost::match_results<std::string::const_iterator> match_result;
+        std::string last_headline;
+
         while (boost::regex_search(begin, end, match_result, re,
                                    boost::match_default))
         {
@@ -78,6 +80,12 @@ std::vector<match> search(const fs::path &filepath, std::string_view regex,
             match.filepath = entry;
             match.text = match_result[0];
             match.context = context(match_result, opts.context);
+            const auto current_headline{headline(match_result.prefix().str())};
+            if (!current_headline.empty())
+            {
+                last_headline = current_headline;
+            }
+            match.headline = last_headline;
 
             matches.emplace_back(match);
             begin = match_result[0].second;
@@ -94,6 +102,7 @@ void cleanup_text(std::string &text)
         // Don't strip headlines. We need them later on.
         if (text[pos + 1] == 'h' || text.substr(pos + 1, 2) == "/h")
         {
+            ++pos;
             continue;
         }
         text.erase(pos, text.find('>', pos) + 1 - pos);
@@ -142,4 +151,21 @@ context(const boost::match_results<std::string::const_iterator> &match,
     return {prefix.substr(pos_before + 2), suffix.substr(0, pos_after - 1)};
 }
 
+[[nodiscard]] std::string headline(const std::string_view prefix)
+{
+    size_t pos{prefix.length()};
+    while ((pos = prefix.rfind("<h", pos)) != std::string_view::npos)
+    {
+        if (boost::regex_match(std::string{prefix.substr(pos, 4)},
+                               boost::regex{"<h[1-6][> ]"}))
+        {
+            pos = prefix.find('>', pos) + 1;
+            return std::string{prefix.substr(pos, prefix.find('<', pos) - pos)};
+        }
+        pos -= 2;
+    }
+
+    return {};
+}
+
 } // namespace epubgrep::search
diff --git a/src/search.hpp b/src/search.hpp
index 70572cf..6510bb3 100644
--- a/src/search.hpp
+++ b/src/search.hpp
@@ -70,6 +70,9 @@ void cleanup_text(std::string &text);
 context(const boost::match_results<std::string::const_iterator> &match,
         std::uint64_t words);
 
+//! Return last headline if possible.
+[[nodiscard]] std::string headline(std::string_view prefix);
+
 } // namespace epubgrep::search
 
 #endif // EPUBGREP_SEARCH_HPP