From ca71d8997d6144fd7bc770ff458d469a989d576c Mon Sep 17 00:00:00 2001
From: Maxime Coste <mawww@kakoune.org>
Date: Fri, 5 Aug 2022 20:29:43 +1000
Subject: [PATCH] Reuse existing character classes when possible in regex

---
 src/regex_impl.cc | 13 +++++++++++--
 src/regex_impl.hh |  8 +++++++-
 2 files changed, 18 insertions(+), 3 deletions(-)

diff --git a/src/regex_impl.cc b/src/regex_impl.cc
index 256afcc1c..40701e482 100644
--- a/src/regex_impl.cc
+++ b/src/regex_impl.cc
@@ -119,6 +119,7 @@ struct Children
         Index operator*() const { return m_pos; }
         bool operator!=(Sentinel) const { return m_pos != m_end; }
 
+    private:
         Index find_prev(Index parent, Index pos) const
         {
             Index child = parent+1;
@@ -544,8 +545,10 @@ private:
             character_class.ranges.empty())
             return add_node(ParsedRegex::CharType, (Codepoint)character_class.ctypes);
 
-        auto class_id = m_parsed_regex.character_classes.size();
-        m_parsed_regex.character_classes.push_back(std::move(character_class));
+        auto it = std::find(m_parsed_regex.character_classes.begin(), m_parsed_regex.character_classes.end(), character_class);
+        auto class_id = it - m_parsed_regex.character_classes.begin();
+        if (it == m_parsed_regex.character_classes.end())
+            m_parsed_regex.character_classes.push_back(std::move(character_class));
 
         return add_node(ParsedRegex::CharClass, class_id);
     }
@@ -1536,6 +1539,12 @@ auto test_regex = UnitTest{[]{
         kak_assert(vm.exec("\t\n\v\f\r"));
     }
 
+    {
+        TestVM<> vm{R"([\t-\r]\h+[\t-\r])"};
+        kak_assert(vm.character_classes.size() == 1);
+        kak_assert(vm.exec("\n  \f"));
+    }
+
     {
         TestVM<> vm{R"([^\x00-\x7F]+)"};
         kak_assert(not vm.exec("ascii"));
diff --git a/src/regex_impl.hh b/src/regex_impl.hh
index ca859592d..73d697904 100644
--- a/src/regex_impl.hh
+++ b/src/regex_impl.hh
@@ -33,12 +33,18 @@ constexpr bool with_bit_ops(Meta::Type<CharacterType>) { return true; }
 
 struct CharacterClass
 {
-    struct Range { Codepoint min, max; };
+    struct Range
+    {
+        Codepoint min, max;
+        friend bool operator==(const Range&, const Range&) = default;
+    };
 
     Vector<Range, MemoryDomain::Regex> ranges;
     CharacterType ctypes = CharacterType::None;
     bool negative = false;
     bool ignore_case = false;
+
+    friend bool operator==(const CharacterClass&, const CharacterClass&) = default;
 };
 
 bool is_character_class(const CharacterClass& character_class, Codepoint cp);