2 files changed, 48 insertions, 0 deletions

diff --git a/lexlib/CharacterCategory.cxx b/lexlib/CharacterCategory.cxx
index bc2fa2336..ca76325df 100644
--- a/lexlib/CharacterCategory.cxx
+++ b/lexlib/CharacterCategory.cxx
@@ -7,6 +7,7 @@
 // Copyright 2013 by Neil Hodgson <neilh@scintilla.org>
 // The License.txt file describes the conditions under which this software may be distributed.
 
+#include <vector>
 #include <algorithm>
 #include <iterator>
 
@@ -3790,6 +3791,7 @@ const int catRanges[] = {
 33554397,
 33554460,
 35651549,
+35651613,
 //--Autogenerated -- end of section automatically generated
 };
 
@@ -3963,4 +3965,33 @@ bool IsXidContinue(int character) {
 	}
 }
 
+CharacterCategoryMap::CharacterCategoryMap() noexcept {
+	Optimize(256);
+}
+
+int CharacterCategoryMap::Size() const noexcept {
+	return static_cast<int>(dense.size());
+}
+
+void CharacterCategoryMap::Optimize(int countCharacters) {
+	const int characters = std::clamp(countCharacters, 256, maxUnicode + 1);
+	dense.resize(characters);
+
+	int end = 0;
+	int index = 0;
+	int current = catRanges[index];
+	++index;
+	do {
+		const int next = catRanges[index];
+		const unsigned char category = current & maskCategory;
+		current >>= 5;
+		end = std::min(characters, next >> 5);
+		while (current < end) {
+			dense[current++] = category;
+		}
+		current = next;
+		++index;
+	} while (characters > end);
+}
+
 }
diff --git a/lexlib/CharacterCategory.h b/lexlib/CharacterCategory.h
index 767d79670..d1ac39152 100644
--- a/lexlib/CharacterCategory.h
+++ b/lexlib/CharacterCategory.h
@@ -28,6 +28,23 @@ bool IsIdContinue(int character);
 bool IsXidStart(int character);
 bool IsXidContinue(int character);
 
+class CharacterCategoryMap {
+private:
+	std::vector<unsigned char> dense;
+public:
+	CharacterCategoryMap() noexcept;
+	CharacterCategory CategoryFor(int character) const {
+		if (static_cast<size_t>(character) < dense.size()) {
+			return static_cast<CharacterCategory>(dense[character]);
+		} else {
+			// binary search through ranges
+			return CategoriseCharacter(character);
+		}
+	}
+	int Size() const noexcept;
+	void Optimize(int countCharacters);
+};
+
 }
 
 #endif