From e476250ea9326b2550847b301c265115ff375a31 Mon Sep 17 00:00:00 2001
From: Sam Atman
Date: Wed, 4 Feb 2026 18:36:18 -0500
Subject: Rest of the 'easy' stuff

This gets us up to feature parity with Jacob's work.  I want to
eliminate that last allocation using the comptime hash map, and then
see about eliminating allocations from case comparisons as well.

That should just about do it.
---
 src/GeneralCategories.zig | 102 ++++++++++++++--------------------------------
 1 file changed, 30 insertions(+), 72 deletions(-)

(limited to 'src/GeneralCategories.zig')

diff --git a/src/GeneralCategories.zig b/src/GeneralCategories.zig
index eee7e56..9a383bf 100644
--- a/src/GeneralCategories.zig
+++ b/src/GeneralCategories.zig
@@ -1,8 +1,19 @@
 //! General Categories
 
-s1: []u16 = undefined,
-s2: []u5 = undefined,
-s3: []u5 = undefined,
+const Data = struct {
+    s1: []const u16 = undefined,
+    s2: []const u5 = undefined,
+    s3: []const u5 = undefined,
+};
+
+const general_categories = general_categories: {
+    const data = @import("gencat");
+    break :general_categories Data{
+        .s1 = &data.s1,
+        .s2 = &data.s2,
+        .s3 = &data.s3,
+    };
+};
 
 /// General Category
 pub const Gc = enum {
@@ -38,51 +49,14 @@ pub const Gc = enum {
     Zs, // Separator, Space
 };
 
-const GeneralCategories = @This();
-
-pub fn init(allocator: Allocator) Allocator.Error!GeneralCategories {
-    var gencat = GeneralCategories{};
-    try gencat.setup(allocator);
-    return gencat;
-}
-
-pub fn setup(gencat: *GeneralCategories, allocator: Allocator) Allocator.Error!void {
-    const in_bytes = @embedFile("gencat");
-    var in_fbs = std.io.fixedBufferStream(in_bytes);
-    var reader = in_fbs.reader();
-
-    const endian = builtin.cpu.arch.endian();
-
-    const s1_len: u16 = reader.readInt(u16, endian) catch unreachable;
-    gencat.s1 = try allocator.alloc(u16, s1_len);
-    errdefer allocator.free(gencat.s1);
-    for (0..s1_len) |i| gencat.s1[i] = reader.readInt(u16, endian) catch unreachable;
-
-    const s2_len: u16 = reader.readInt(u16, endian) catch unreachable;
-    gencat.s2 = try allocator.alloc(u5, s2_len);
-    errdefer allocator.free(gencat.s2);
-    for (0..s2_len) |i| gencat.s2[i] = @intCast(reader.readInt(u8, endian) catch unreachable);
-
-    const s3_len: u16 = reader.readInt(u8, endian) catch unreachable;
-    gencat.s3 = try allocator.alloc(u5, s3_len);
-    errdefer allocator.free(gencat.s3);
-    for (0..s3_len) |i| gencat.s3[i] = @intCast(reader.readInt(u8, endian) catch unreachable);
-}
-
-pub fn deinit(gencat: *const GeneralCategories, allocator: mem.Allocator) void {
-    allocator.free(gencat.s1);
-    allocator.free(gencat.s2);
-    allocator.free(gencat.s3);
-}
-
 /// Lookup the General Category for `cp`.
-pub fn gc(gencat: GeneralCategories, cp: u21) Gc {
-    return @enumFromInt(gencat.s3[gencat.s2[gencat.s1[cp >> 8] + (cp & 0xff)]]);
+pub fn gc(cp: u21) Gc {
+    return @enumFromInt(general_categories.s3[general_categories.s2[general_categories.s1[cp >> 8] + (cp & 0xff)]]);
 }
 
 /// True if `cp` has an C general category.
-pub fn isControl(gencat: GeneralCategories, cp: u21) bool {
-    return switch (gencat.gc(cp)) {
+pub fn isControl(cp: u21) bool {
+    return switch (gc(cp)) {
         .Cc,
         .Cf,
         .Cn,
@@ -94,8 +68,8 @@ pub fn isControl(gencat: GeneralCategories, cp: u21) bool {
 }
 
 /// True if `cp` has an L general category.
-pub fn isLetter(gencat: GeneralCategories, cp: u21) bool {
-    return switch (gencat.gc(cp)) {
+pub fn isLetter(cp: u21) bool {
+    return switch (gc(cp)) {
         .Ll,
         .Lm,
         .Lo,
@@ -107,8 +81,8 @@ pub fn isLetter(gencat: GeneralCategories, cp: u21) bool {
 }
 
 /// True if `cp` has an M general category.
-pub fn isMark(gencat: GeneralCategories, cp: u21) bool {
-    return switch (gencat.gc(cp)) {
+pub fn isMark(cp: u21) bool {
+    return switch (gc(cp)) {
         .Mc,
         .Me,
         .Mn,
@@ -118,8 +92,8 @@ pub fn isMark(gencat: GeneralCategories, cp: u21) bool {
 }
 
 /// True if `cp` has an N general category.
-pub fn isNumber(gencat: GeneralCategories, cp: u21) bool {
-    return switch (gencat.gc(cp)) {
+pub fn isNumber(cp: u21) bool {
+    return switch (gc(cp)) {
         .Nd,
         .Nl,
         .No,
@@ -129,8 +103,8 @@ pub fn isNumber(gencat: GeneralCategories, cp: u21) bool {
 }
 
 /// True if `cp` has an P general category.
-pub fn isPunctuation(gencat: GeneralCategories, cp: u21) bool {
-    return switch (gencat.gc(cp)) {
+pub fn isPunctuation(cp: u21) bool {
+    return switch (gc(cp)) {
         .Pc,
         .Pd,
         .Pe,
@@ -144,8 +118,8 @@ pub fn isPunctuation(gencat: GeneralCategories, cp: u21) bool {
 }
 
 /// True if `cp` has an S general category.
-pub fn isSymbol(gencat: GeneralCategories, cp: u21) bool {
-    return switch (gencat.gc(cp)) {
+pub fn isSymbol(cp: u21) bool {
+    return switch (gc(cp)) {
         .Sc,
         .Sk,
         .Sm,
@@ -156,8 +130,8 @@ pub fn isSymbol(gencat: GeneralCategories, cp: u21) bool {
 }
 
 /// True if `cp` has an Z general category.
-pub fn isSeparator(gencat: GeneralCategories, cp: u21) bool {
-    return switch (gencat.gc(cp)) {
+pub fn isSeparator(cp: u21) bool {
+    return switch (gc(cp)) {
         .Zl,
         .Zp,
         .Zs,
@@ -165,19 +139,3 @@ pub fn isSeparator(gencat: GeneralCategories, cp: u21) bool {
         else => false,
     };
 }
-
-fn testAllocator(allocator: Allocator) !void {
-    var gen_cat = try GeneralCategories.init(allocator);
-    gen_cat.deinit(allocator);
-}
-
-test "Allocation failure" {
-    try testing.checkAllAllocationFailures(testing.allocator, testAllocator, .{});
-}
-
-const std = @import("std");
-const builtin = @import("builtin");
-const compress = std.compress;
-const mem = std.mem;
-const testing = std.testing;
-const Allocator = mem.Allocator;
-- 
cgit v1.2.3