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/NormPropsData.zig | 46 +++++++++++++++-------------------------------
 1 file changed, 15 insertions(+), 31 deletions(-)

(limited to 'src/NormPropsData.zig')

diff --git a/src/NormPropsData.zig b/src/NormPropsData.zig
index 7b53542..cca3556 100644
--- a/src/NormPropsData.zig
+++ b/src/NormPropsData.zig
@@ -1,48 +1,32 @@
 //! Normalization Properties Data
 
-s1: []u16 = undefined,
-s2: []u4 = undefined,
+const Data = struct {
+    s1: []const u16 = undefined,
+    s2: []const u3 = undefined,
+};
+
+const norms = norm_props_data: {
+    const data = @import("normp");
+    break :norm_props_data Data{
+        .s1 = &data.s1,
+        .s2 = &data.s2,
+    };
+};
 
 const NormProps = @This();
 
-pub fn init(allocator: mem.Allocator) !NormProps {
-    const in_bytes = @embedFile("normp");
-    var in_fbs = std.io.fixedBufferStream(in_bytes);
-    var reader = in_fbs.reader();
-
-    const endian = builtin.cpu.arch.endian();
-    var norms = NormProps{};
-
-    const stage_1_len: u16 = try reader.readInt(u16, endian);
-    norms.s1 = try allocator.alloc(u16, stage_1_len);
-    errdefer allocator.free(norms.s1);
-    for (0..stage_1_len) |i| norms.s1[i] = try reader.readInt(u16, endian);
-
-    const stage_2_len: u16 = try reader.readInt(u16, endian);
-    norms.s2 = try allocator.alloc(u4, stage_2_len);
-    errdefer allocator.free(norms.s2);
-    for (0..stage_2_len) |i| norms.s2[i] = @intCast(try reader.readInt(u8, endian));
-
-    return norms;
-}
-
-pub fn deinit(norms: *const NormProps, allocator: mem.Allocator) void {
-    allocator.free(norms.s1);
-    allocator.free(norms.s2);
-}
-
 /// Returns true if `cp` is already in NFD form.
-pub fn isNfd(norms: *const NormProps, cp: u21) bool {
+pub fn isNfd(cp: u21) bool {
     return norms.s2[norms.s1[cp >> 8] + (cp & 0xff)] & 1 == 0;
 }
 
 /// Returns true if `cp` is already in NFKD form.
-pub fn isNfkd(norms: *const NormProps, cp: u21) bool {
+pub fn isNfkd(cp: u21) bool {
     return norms.s2[norms.s1[cp >> 8] + (cp & 0xff)] & 2 == 0;
 }
 
 /// Returns true if `cp` is not allowed in any normalized form.
-pub fn isFcx(norms: *const NormProps, cp: u21) bool {
+pub fn isFcx(cp: u21) bool {
     return norms.s2[norms.s1[cp >> 8] + (cp & 0xff)] & 4 == 4;
 }
 
-- 
cgit v1.2.3