From ba5d9081b479e95ffa7f3baf751beedd370cec14 Mon Sep 17 00:00:00 2001
From: Sam Atman
Date: Wed, 4 Feb 2026 18:01:36 -0500
Subject: Normalization and case folding

Both of which deserve some further attention.
---
 src/CombiningData.zig | 49 ++++++++++++++++---------------------------------
 1 file changed, 16 insertions(+), 33 deletions(-)

(limited to 'src/CombiningData.zig')

diff --git a/src/CombiningData.zig b/src/CombiningData.zig
index f58e0de..660dc14 100644
--- a/src/CombiningData.zig
+++ b/src/CombiningData.zig
@@ -1,45 +1,28 @@
 //! Combining Class Data
 
-s1: []u16 = undefined,
-s2: []u8 = undefined,
+const Data = struct {
+    s1: []const u16 = undefined,
+    s2: []const u8 = undefined,
+};
+
+const combining_data = combining_data: {
+    const data = @import("ccc");
+    break :combining_data Data{
+        .s1 = &data.s1,
+        .s2 = &data.s2,
+    };
+};
 
 const CombiningData = @This();
 
-pub fn init(allocator: mem.Allocator) !CombiningData {
-    const in_bytes = @embedFile("ccc");
-    var in_fbs = std.io.fixedBufferStream(in_bytes);
-    var reader = in_fbs.reader();
-
-    const endian = builtin.cpu.arch.endian();
-
-    var cbdata = CombiningData{};
-
-    const stage_1_len: u16 = try reader.readInt(u16, endian);
-    cbdata.s1 = try allocator.alloc(u16, stage_1_len);
-    errdefer allocator.free(cbdata.s1);
-    for (0..stage_1_len) |i| cbdata.s1[i] = try reader.readInt(u16, endian);
-
-    const stage_2_len: u16 = try reader.readInt(u16, endian);
-    cbdata.s2 = try allocator.alloc(u8, stage_2_len);
-    errdefer allocator.free(cbdata.s2);
-    _ = try reader.readAll(cbdata.s2);
-
-    return cbdata;
-}
-
-pub fn deinit(cbdata: *const CombiningData, allocator: mem.Allocator) void {
-    allocator.free(cbdata.s1);
-    allocator.free(cbdata.s2);
-}
-
 /// Returns the canonical combining class for a code point.
-pub fn ccc(cbdata: CombiningData, cp: u21) u8 {
-    return cbdata.s2[cbdata.s1[cp >> 8] + (cp & 0xff)];
+pub fn ccc(cp: u21) u8 {
+    return combining_data.s2[combining_data.s1[cp >> 8] + (cp & 0xff)];
 }
 
 /// True if `cp` is a starter code point, not a combining character.
-pub fn isStarter(cbdata: CombiningData, cp: u21) bool {
-    return cbdata.s2[cbdata.s1[cp >> 8] + (cp & 0xff)] == 0;
+pub fn isStarter(cp: u21) bool {
+    return combining_data.s2[combining_data.s1[cp >> 8] + (cp & 0xff)] == 0;
 }
 
 const std = @import("std");
-- 
cgit v1.2.3