From c07e52d10d215b54704ac4be7243214ef24d39f6 Mon Sep 17 00:00:00 2001
From: Ivan Chikish <nekotekina@gmail.com>
Date: Tue, 17 Dec 2024 16:17:59 +0300
Subject: [PATCH] [C++] Optimize unicode.cpp with binary search

---
 Makefile                  |  2 +-
 scripts/generate_xcompose | 30 ++++++++++++++++++++++--------
 2 files changed, 23 insertions(+), 9 deletions(-)

diff --git a/Makefile b/Makefile
index 03f5e2a..ebe7493 100644
--- a/Makefile
+++ b/Makefile
@@ -53,7 +53,7 @@ endif
 all: compose man
 	mkdir -p bin
 	cp scripts/keyd-application-mapper bin/
-	$(CXX) $(CXXFLAGS) -O3 $(COMPAT_FILES) src/*.c src/*.cpp src/vkbd/$(VKBD).cpp -lpthread -o bin/keyd $(LDFLAGS)
+	$(CXX) $(CXXFLAGS) -O3 $(COMPAT_FILES) src/*.cpp src/vkbd/$(VKBD).cpp -lpthread -o bin/keyd $(LDFLAGS)
 debug:
 	CFLAGS="-g -fsanitize=address -Wunused" $(MAKE)
 compose:
diff --git a/scripts/generate_xcompose b/scripts/generate_xcompose
index d1edb6f..6084b3b 100755
--- a/scripts/generate_xcompose
+++ b/scripts/generate_xcompose
@@ -44,23 +44,37 @@ open('data/keyd.compose', 'w').write(data)
 # table to capitalize on codepoint contiguity, but 35k is small enough to
 # warrant keeping the entire thing in memory.
 
-open('src/unicode.c', 'w').write(f'''
+open('src/unicode.cpp', 'w').write(f'''
 	/* GENERATED BY {sys.argv[0]}, DO NOT MODIFY BY HAND. */
 
 	#include <stdint.h>
 	#include <stdlib.h>
 	#include "keys.h"
+	#include <algorithm>
+	#include <vector>
 
-	uint32_t unicode_table[] = {{ {','.join(map(str, codes))} }};
+	constexpr uint32_t unicode_table[] = {{ {','.join(map(str, codes))} }};
 
 	int unicode_lookup_index(uint32_t codepoint)
 	{{
-		size_t i = 0;
-
-		for(i = 0; i < sizeof(unicode_table)/sizeof(unicode_table[0]); i++) {{
-			if (unicode_table[i] == codepoint)
-				return i;
-		}}
+		// Slight search optimizations: prevent CPU from fetching data from unlikely areas
+		constexpr auto cb = std::cbegin(unicode_table);
+		constexpr auto ce = std::cend(unicode_table);
+		constexpr auto ce2 = std::lower_bound(cb, ce, 0x1100); // before hangul
+		constexpr auto ce3 = std::lower_bound(cb, ce2, 0x531); // after cyrillic
+		constexpr auto ce4 = std::lower_bound(cb, ce3, 0x300); // after latin, modifiers
+
+		auto beg = cb;
+		auto end = ce;
+		if (codepoint < *ce4)
+			end = ce4;
+		else if (codepoint < *ce3)
+			beg = ce4, end = ce3;
+		else if (codepoint < *ce2)
+			beg = ce3, end = ce4;
+		auto res = std::lower_bound(beg, end, codepoint);
+		if (res != end && *res == codepoint)
+			return res - cb;
 
 		return -1;
 	}}