From 95f9487f6a7bde2d7266399bdf6843b97cc1b301 Mon Sep 17 00:00:00 2001
From: Sam Atman
Date: Thu, 5 Feb 2026 07:07:40 -0500
Subject: Base units do not allocate

CanonData included.  I may still sort out caseless matching without
allocation, but that's a stretch goal.

Closes #86
Closes #85
---
 src/CanonData.zig | 57 +++++++++++++++++--------------------------------------
 1 file changed, 17 insertions(+), 40 deletions(-)

(limited to 'src/CanonData.zig')

diff --git a/src/CanonData.zig b/src/CanonData.zig
index c972534..5c1ffa6 100644
--- a/src/CanonData.zig
+++ b/src/CanonData.zig
@@ -1,13 +1,23 @@
 //! Canonicalization Data
 
-s1: []const u16 = undefined,
-s2: []const @import("canon").Canonicalization = undefined,
-nfc: std.AutoHashMapUnmanaged([2]u21, u21),
+const Data = struct {
+    s1: []const u16 = undefined,
+    s2: []const @import("canon").Canonicalization = undefined,
+};
+
+const canon_data = canon_data: {
+    const canon_ = @import("canon");
+    break :canon_data Data{
+        .s1 = &canon_.s1,
+        .s2 = &canon_.s2,
+    };
+};
 
 const CanonData = @This();
 
 // There's a bug here, which is down to how static u21 vs. runtime are handled,
-// the "unique representation" claim is not working out.  So we do this:
+// the "unique representation" claim is not working out.  AutoHash casts to bytes,
+// and that won't fly.  So we do this:
 
 const Context = struct {
     pub fn hash(_: Context, cps: [2]u21) u64 {
@@ -22,47 +32,14 @@ const Context = struct {
 
 const c_map = comptime_map.ComptimeHashMap([2]u21, u21, Context, @import("canon").c_map);
 
-pub fn init(allocator: mem.Allocator) !CanonData {
-    var cdata = CanonData{
-        .nfc = .empty,
-    };
-    errdefer cdata.deinit(allocator);
-
-    const data = @import("canon");
-    cdata.s1 = &data.s1;
-    cdata.s2 = &data.s2;
-    var count: usize = 0;
-    for (data.composite) |cp| {
-        count += 1;
-        const cps = cdata.toNfd(cp);
-        std.debug.assert(cps.len == 2);
-        try cdata.nfc.put(allocator, cps[0..2].*, cp);
-    }
-
-    // var keys = cdata.nfc.keyIterator();
-    // while (keys.next()) |key| {
-    //     const c32: [2]u32 = .{ key[0], key[1] };
-    //     if (c_map.get(c32)) |_| {
-    //         std.debug.print("got", .{});
-    //     }
-    // }
-
-    return cdata;
-}
-
-pub fn deinit(cdata: *CanonData, allocator: mem.Allocator) void {
-    cdata.nfc.deinit(allocator);
-}
-
 /// Returns canonical decomposition for `cp`.
-pub fn toNfd(cdata: *const CanonData, cp: u21) []const u21 {
-    const canon = &cdata.s2[cdata.s1[cp >> 8] + (cp & 0xff)];
+pub fn toNfd(cp: u21) []const u21 {
+    const canon = &canon_data.s2[canon_data.s1[cp >> 8] + (cp & 0xff)];
     return canon.cps[0..canon.len];
 }
 
 // Returns the primary composite for the codepoints in `cp`.
-pub fn toNfc(cdata: *const CanonData, cps: [2]u21) ?u21 {
-    _ = cdata;
+pub fn toNfc(cps: [2]u21) ?u21 {
     if (c_map.get(cps)) |cpp| {
         return cpp.*;
     } else {
-- 
cgit v1.2.3