New single byte props table

5 files changed, 313 insertions, 290 deletions

diff --git a/build.zig b/build.zig
index f552547..8193f0d 100644
--- a/build.zig
+++ b/build.zig
@@ -7,35 +7,14 @@ pub fn build(b: *std.Build) void {
     const ziglyph = b.dependency("ziglyph", .{});
 
     const gbp_gen_exe = b.addExecutable(.{
-        .name = "grapheme_break",
-        .root_source_file = .{ .path = "codegen/grapheme_break.zig" },
+        .name = "gbp",
+        .root_source_file = .{ .path = "codegen/gbp.zig" },
         .target = b.host,
         .optimize = .Debug,
     });
-    gbp_gen_exe.root_module.addImport("ziglyph", ziglyph.module("ziglyph"));
     const run_gbp_gen_exe = b.addRunArtifact(gbp_gen_exe);
     const gbp_gen_out = run_gbp_gen_exe.addOutputFileArg("gbp.zig");
 
-    const emoji_gen_exe = b.addExecutable(.{
-        .name = "emoji",
-        .root_source_file = .{ .path = "codegen/emoji.zig" },
-        .target = b.host,
-        .optimize = .Debug,
-    });
-    emoji_gen_exe.root_module.addImport("ziglyph", ziglyph.module("ziglyph"));
-    const run_emoji_gen_exe = b.addRunArtifact(emoji_gen_exe);
-    const emoji_gen_out = run_emoji_gen_exe.addOutputFileArg("emoji.zig");
-
-    const indic_gen_exe = b.addExecutable(.{
-        .name = "indic",
-        .root_source_file = .{ .path = "codegen/indic.zig" },
-        .target = b.host,
-        .optimize = .Debug,
-    });
-    indic_gen_exe.root_module.addImport("ziglyph", ziglyph.module("ziglyph"));
-    const run_indic_gen_exe = b.addRunArtifact(indic_gen_exe);
-    const indic_gen_out = run_indic_gen_exe.addOutputFileArg("indic.zig");
-
     const exe = b.addExecutable(.{
         .name = "zgbench",
         .root_source_file = .{ .path = "src/main.zig" },
@@ -44,8 +23,6 @@ pub fn build(b: *std.Build) void {
     });
     exe.root_module.addImport("ziglyph", ziglyph.module("ziglyph"));
     exe.root_module.addAnonymousImport("gbp", .{ .root_source_file = gbp_gen_out });
-    exe.root_module.addAnonymousImport("emoji", .{ .root_source_file = emoji_gen_out });
-    exe.root_module.addAnonymousImport("indic", .{ .root_source_file = indic_gen_out });
     b.installArtifact(exe);
 
     const run_cmd = b.addRunArtifact(exe);
@@ -62,8 +39,6 @@ pub fn build(b: *std.Build) void {
     });
     exe_unit_tests.root_module.addImport("ziglyph", ziglyph.module("ziglyph"));
     exe_unit_tests.root_module.addAnonymousImport("gbp", .{ .root_source_file = gbp_gen_out });
-    exe_unit_tests.root_module.addAnonymousImport("emoji", .{ .root_source_file = emoji_gen_out });
-    exe_unit_tests.root_module.addAnonymousImport("indic", .{ .root_source_file = indic_gen_out });
 
     const run_exe_unit_tests = b.addRunArtifact(exe_unit_tests);
 
diff --git a/codegen/emoji.zig b/codegen/emoji.zig
deleted file mode 100644
index 6dbb74f..0000000
--- a/codegen/emoji.zig
+++ /dev/null
@@ -1,82 +0,0 @@
-const std = @import("std");
-
-const emoji = @import("ziglyph").emoji;
-
-const block_size = 256;
-const Block = [block_size]bool;
-
-const BlockMap = std.HashMap(
-    Block,
-    u16,
-    struct {
-        pub fn hash(_: @This(), k: Block) u64 {
-            var hasher = std.hash.Wyhash.init(0);
-            std.hash.autoHashStrat(&hasher, k, .DeepRecursive);
-            return hasher.final();
-        }
-
-        pub fn eql(_: @This(), a: Block, b: Block) bool {
-            return std.mem.eql(bool, &a, &b);
-        }
-    },
-    std.hash_map.default_max_load_percentage,
-);
-
-pub fn main() !void {
-    var arena = std.heap.ArenaAllocator.init(std.heap.page_allocator);
-    defer arena.deinit();
-    const allocator = arena.allocator();
-
-    var blocks_map = BlockMap.init(allocator);
-    defer blocks_map.deinit();
-
-    var stage1 = std.ArrayList(u16).init(allocator);
-    defer stage1.deinit();
-
-    var stage2 = std.ArrayList(bool).init(allocator);
-    defer stage2.deinit();
-
-    var block: Block = [_]bool{false} ** block_size;
-    var block_len: u16 = 0;
-
-    for (0..0x10ffff + 1) |cp| {
-        const isEmoji = emoji.isExtendedPictographic(@intCast(cp));
-
-        block[block_len] = isEmoji;
-        block_len += 1;
-
-        if (block_len < block_size and cp != 0x10ffff) continue;
-
-        const gop = try blocks_map.getOrPut(block);
-        if (!gop.found_existing) {
-            gop.value_ptr.* = @intCast(stage2.items.len);
-            try stage2.appendSlice(&block);
-        }
-
-        try stage1.append(gop.value_ptr.*);
-        block_len = 0;
-    }
-
-    var args_iter = std.process.args();
-    _ = args_iter.skip();
-    const output_path = args_iter.next() orelse @panic("No output file arg!");
-
-    var out_file = try std.fs.cwd().createFile(output_path, .{});
-    defer out_file.close();
-    var out_buf = std.io.bufferedWriter(out_file.writer());
-    const writer = out_buf.writer();
-
-    try writer.print("pub const stage_1 = [{}]u16{{", .{stage1.items.len});
-    for (stage1.items) |v| {
-        _ = try writer.print("{},", .{v});
-    }
-    try writer.writeAll("};\n");
-
-    try writer.print("pub const stage_2 = [{}]bool{{", .{stage2.items.len});
-    for (stage2.items) |v| {
-        _ = try writer.print("{},", .{v});
-    }
-    try writer.writeAll("};\n");
-
-    try out_buf.flush();
-}
diff --git a/codegen/gbp.zig b/codegen/gbp.zig
new file mode 100644
index 0000000..7e45ddf
--- /dev/null
+++ b/codegen/gbp.zig
@@ -0,0 +1,280 @@
+const std = @import("std");
+
+const Indic = enum {
+    none,
+
+    Consonant,
+    Extend,
+    Linker,
+};
+
+const Gbp = enum {
+    none,
+
+    Control,
+    CR,
+    Extend,
+    L,
+    LF,
+    LV,
+    LVT,
+    Prepend,
+    Regional_Indicator,
+    SpacingMark,
+    T,
+    V,
+    ZWJ,
+};
+
+const block_size = 256;
+const Block = [block_size]u16;
+
+const BlockMap = std.HashMap(
+    Block,
+    u16,
+    struct {
+        pub fn hash(_: @This(), k: Block) u64 {
+            var hasher = std.hash.Wyhash.init(0);
+            std.hash.autoHashStrat(&hasher, k, .DeepRecursive);
+            return hasher.final();
+        }
+
+        pub fn eql(_: @This(), a: Block, b: Block) bool {
+            return std.mem.eql(u16, &a, &b);
+        }
+    },
+    std.hash_map.default_max_load_percentage,
+);
+
+pub fn main() !void {
+    var arena = std.heap.ArenaAllocator.init(std.heap.page_allocator);
+    defer arena.deinit();
+    const allocator = arena.allocator();
+
+    var indic_map = std.AutoHashMap(u21, Indic).init(allocator);
+    defer indic_map.deinit();
+
+    var gbp_map = std.AutoHashMap(u21, Gbp).init(allocator);
+    defer gbp_map.deinit();
+
+    var emoji_set = std.AutoHashMap(u21, void).init(allocator);
+    defer emoji_set.deinit();
+
+    var line_buf: [4096]u8 = undefined;
+
+    // Process Indic
+    var indic_file = try std.fs.cwd().openFile("unicode/DerivedCoreProperties.txt", .{});
+    defer indic_file.close();
+    var indic_buf = std.io.bufferedReader(indic_file.reader());
+    const indic_reader = indic_buf.reader();
+
+    while (try indic_reader.readUntilDelimiterOrEof(&line_buf, '\n')) |line| {
+        if (line.len == 0 or line[0] == '#') continue;
+        if (std.mem.indexOf(u8, line, "InCB") == null) continue;
+        const no_comment = if (std.mem.indexOfScalar(u8, line, '#')) |octo| line[0..octo] else line;
+
+        var field_iter = std.mem.tokenizeAny(u8, no_comment, "; ");
+        var current_code: [2]u21 = undefined;
+
+        var i: usize = 0;
+        while (field_iter.next()) |field| : (i += 1) {
+            switch (i) {
+                0 => {
+                    // Code point(s)
+                    if (std.mem.indexOf(u8, field, "..")) |dots| {
+                        current_code = .{
+                            try std.fmt.parseInt(u21, field[0..dots], 16),
+                            try std.fmt.parseInt(u21, field[dots + 2 ..], 16),
+                        };
+                    } else {
+                        const code = try std.fmt.parseInt(u21, field, 16);
+                        current_code = .{ code, code };
+                    }
+                },
+                2 => {
+                    // Prop
+                    const prop = std.meta.stringToEnum(Indic, field) orelse return error.InvalidPorp;
+                    for (current_code[0]..current_code[1] + 1) |cp| try indic_map.put(@intCast(cp), prop);
+                },
+                else => {},
+            }
+        }
+    }
+
+    // Process GBP
+    var gbp_file = try std.fs.cwd().openFile("unicode/auxiliary/GraphemeBreakProperty.txt", .{});
+    defer gbp_file.close();
+    var gbp_buf = std.io.bufferedReader(gbp_file.reader());
+    const gbp_reader = gbp_buf.reader();
+
+    while (try gbp_reader.readUntilDelimiterOrEof(&line_buf, '\n')) |line| {
+        if (line.len == 0 or line[0] == '#') continue;
+        const no_comment = if (std.mem.indexOfScalar(u8, line, '#')) |octo| line[0..octo] else line;
+
+        var field_iter = std.mem.tokenizeAny(u8, no_comment, "; ");
+        var current_code: [2]u21 = undefined;
+
+        var i: usize = 0;
+        while (field_iter.next()) |field| : (i += 1) {
+            switch (i) {
+                0 => {
+                    // Code point(s)
+                    if (std.mem.indexOf(u8, field, "..")) |dots| {
+                        current_code = .{
+                            try std.fmt.parseInt(u21, field[0..dots], 16),
+                            try std.fmt.parseInt(u21, field[dots + 2 ..], 16),
+                        };
+                    } else {
+                        const code = try std.fmt.parseInt(u21, field, 16);
+                        current_code = .{ code, code };
+                    }
+                },
+                1 => {
+                    // Prop
+                    const prop = std.meta.stringToEnum(Gbp, field) orelse return error.InvalidPorp;
+                    for (current_code[0]..current_code[1] + 1) |cp| try gbp_map.put(@intCast(cp), prop);
+                },
+                else => {},
+            }
+        }
+    }
+
+    // Process Emoji
+    var emoji_file = try std.fs.cwd().openFile("unicode/emoji/emoji-data.txt", .{});
+    defer emoji_file.close();
+    var emoji_buf = std.io.bufferedReader(emoji_file.reader());
+    const emoji_reader = emoji_buf.reader();
+
+    while (try emoji_reader.readUntilDelimiterOrEof(&line_buf, '\n')) |line| {
+        if (line.len == 0 or line[0] == '#') continue;
+        if (std.mem.indexOf(u8, line, "Extended_Pictographic") == null) continue;
+        const no_comment = if (std.mem.indexOfScalar(u8, line, '#')) |octo| line[0..octo] else line;
+
+        var field_iter = std.mem.tokenizeAny(u8, no_comment, "; ");
+
+        var i: usize = 0;
+        while (field_iter.next()) |field| : (i += 1) {
+            switch (i) {
+                0 => {
+                    // Code point(s)
+                    if (std.mem.indexOf(u8, field, "..")) |dots| {
+                        const from = try std.fmt.parseInt(u21, field[0..dots], 16);
+                        const to = try std.fmt.parseInt(u21, field[dots + 2 ..], 16);
+                        for (from..to + 1) |cp| try emoji_set.put(@intCast(cp), {});
+                    } else {
+                        const cp = try std.fmt.parseInt(u21, field, 16);
+                        try emoji_set.put(@intCast(cp), {});
+                    }
+                },
+                else => {},
+            }
+        }
+    }
+
+    var blocks_map = BlockMap.init(allocator);
+    defer blocks_map.deinit();
+
+    var stage1 = std.ArrayList(u16).init(allocator);
+    defer stage1.deinit();
+
+    var stage2 = std.ArrayList(u16).init(allocator);
+    defer stage2.deinit();
+
+    var stage3 = std.AutoArrayHashMap(u8, u16).init(allocator);
+    defer stage3.deinit();
+    var stage3_len: u16 = 0;
+
+    var block: Block = [_]u16{0} ** block_size;
+    var block_len: u16 = 0;
+
+    for (0..0x110000) |i| {
+        const cp: u21 = @intCast(i);
+        const gbp_prop: u8 = @intFromEnum(gbp_map.get(cp) orelse .none);
+        const indic_prop: u8 = @intFromEnum(indic_map.get(cp) orelse .none);
+        const emoji_prop: u1 = @intFromBool(emoji_set.contains(cp));
+        var props_byte: u8 = gbp_prop << 4;
+        props_byte |= indic_prop << 1;
+        props_byte |= emoji_prop;
+
+        const stage3_idx = blk: {
+            const gop = try stage3.getOrPut(props_byte);
+            if (!gop.found_existing) {
+                gop.value_ptr.* = stage3_len;
+                stage3_len += 1;
+            }
+
+            break :blk gop.value_ptr.*;
+        };
+
+        block[block_len] = stage3_idx;
+        block_len += 1;
+
+        if (block_len < block_size and cp != 0x10ffff) continue;
+
+        const gop = try blocks_map.getOrPut(block);
+        if (!gop.found_existing) {
+            gop.value_ptr.* = @intCast(stage2.items.len);
+            try stage2.appendSlice(&block);
+        }
+
+        try stage1.append(gop.value_ptr.*);
+        block_len = 0;
+    }
+
+    var args_iter = std.process.args();
+    _ = args_iter.skip();
+    const output_path = args_iter.next() orelse @panic("No output file arg!");
+
+    var out_file = try std.fs.cwd().createFile(output_path, .{});
+    defer out_file.close();
+    var out_buf = std.io.bufferedWriter(out_file.writer());
+    const writer = out_buf.writer();
+
+    const prop_code =
+        \\const std = @import("std");
+        \\
+        \\pub const Indic = enum {
+        \\    none,
+        \\
+        \\    Consonant,
+        \\    Extend,
+        \\    Linker,
+        \\};
+        \\
+        \\pub const Gbp = enum {
+        \\    none,
+        \\    Control,
+        \\    CR,
+        \\    Extend,
+        \\    L,
+        \\    LF,
+        \\    LV,
+        \\    LVT,
+        \\    Prepend,
+        \\    Regional_Indicator,
+        \\    SpacingMark,
+        \\    T,
+        \\    V,
+        \\    ZWJ,
+        \\};
+        \\
+    ;
+
+    try writer.writeAll(prop_code);
+
+    try writer.print("const Stage2Int = std.math.IntFittingRange(0, {});\n", .{stage2.items.len});
+    try writer.print("pub const stage_1 = [{}]Stage2Int{{", .{stage1.items.len});
+    for (stage1.items) |v| try writer.print("{},", .{v});
+    try writer.writeAll("};\n");
+
+    try writer.print("const Stage3Int = std.math.IntFittingRange(0, {});\n", .{stage3_len});
+    try writer.print("pub const stage_2 = [{}]Stage3Int{{", .{stage2.items.len});
+    for (stage2.items) |v| try writer.print("{},", .{v});
+    try writer.writeAll("};\n");
+
+    try writer.print("pub const stage_3 = [{}]u8{{", .{stage3_len});
+    for (stage3.keys()) |v| try writer.print("{},", .{v});
+    try writer.writeAll("};\n");
+
+    try out_buf.flush();
+}
diff --git a/codegen/grapheme_break.zig b/codegen/grapheme_break.zig
deleted file mode 100644
index 95237f3..0000000
--- a/codegen/grapheme_break.zig
+++ /dev/null
@@ -1,154 +0,0 @@
-const std = @import("std");
-
-const gbp = @import("ziglyph").grapheme_break;
-
-const Prop = enum {
-    none,
-
-    control,
-    extend,
-    hangul_l,
-    hangul_lv,
-    hangul_lvt,
-    hangul_v,
-    hangul_t,
-    prepend,
-    regional,
-    spacing,
-    zwj,
-
-    fn forCodePoint(cp: u21) Prop {
-        if (gbp.isControl(cp)) return .control;
-        if (gbp.isExtend(cp)) return .extend;
-        if (gbp.isL(cp)) return .hangul_l;
-        if (gbp.isLv(cp)) return .hangul_lv;
-        if (gbp.isLvt(cp)) return .hangul_lvt;
-        if (gbp.isT(cp)) return .hangul_t;
-        if (gbp.isV(cp)) return .hangul_v;
-        if (gbp.isPrepend(cp)) return .prepend;
-        if (gbp.isRegionalIndicator(cp)) return .regional;
-        if (gbp.isSpacingmark(cp)) return .spacing;
-        if (gbp.isZwj(cp)) return .zwj;
-
-        return .none;
-    }
-};
-
-const block_size = 256;
-const Block = [block_size]u4;
-
-const BlockMap = std.HashMap(
-    Block,
-    u16,
-    struct {
-        pub fn hash(_: @This(), k: Block) u64 {
-            var hasher = std.hash.Wyhash.init(0);
-            std.hash.autoHashStrat(&hasher, k, .DeepRecursive);
-            return hasher.final();
-        }
-
-        pub fn eql(_: @This(), a: Block, b: Block) bool {
-            return std.mem.eql(u4, &a, &b);
-        }
-    },
-    std.hash_map.default_max_load_percentage,
-);
-
-pub fn main() !void {
-    var arena = std.heap.ArenaAllocator.init(std.heap.page_allocator);
-    defer arena.deinit();
-    const allocator = arena.allocator();
-
-    var blocks_map = BlockMap.init(allocator);
-    defer blocks_map.deinit();
-
-    var stage1 = std.ArrayList(u16).init(allocator);
-    defer stage1.deinit();
-
-    var stage2 = std.ArrayList(u4).init(allocator);
-    defer stage2.deinit();
-
-    var stage3 = std.ArrayList(Prop).init(allocator);
-    defer stage3.deinit();
-
-    var block: Block = [_]u4{0} ** block_size;
-    var block_len: u16 = 0;
-
-    for (0..0x10ffff + 1) |cp| {
-        const prop = Prop.forCodePoint(@intCast(cp));
-
-        const block_idx = blk: {
-            for (stage3.items, 0..) |item, i| {
-                if (item == prop) break :blk i;
-            }
-
-            const idx = stage3.items.len;
-            try stage3.append(prop);
-            break :blk idx;
-        };
-
-        block[block_len] = @intCast(block_idx);
-        block_len += 1;
-
-        if (block_len < block_size and cp != 0x10ffff) continue;
-
-        const gop = try blocks_map.getOrPut(block);
-        if (!gop.found_existing) {
-            gop.value_ptr.* = @intCast(stage2.items.len);
-            try stage2.appendSlice(&block);
-        }
-
-        try stage1.append(gop.value_ptr.*);
-        block_len = 0;
-    }
-
-    var args_iter = std.process.args();
-    _ = args_iter.skip();
-    const output_path = args_iter.next() orelse @panic("No output file arg!");
-
-    var out_file = try std.fs.cwd().createFile(output_path, .{});
-    defer out_file.close();
-    var out_buf = std.io.bufferedWriter(out_file.writer());
-    const writer = out_buf.writer();
-
-    const prop_code =
-        \\pub const Prop = enum {
-        \\    none,
-        \\
-        \\    control,
-        \\    extend,
-        \\    hangul_l,
-        \\    hangul_lv,
-        \\    hangul_lvt,
-        \\    hangul_v,
-        \\    hangul_t,
-        \\    prepend,
-        \\    regional,
-        \\    spacing,
-        \\    zwj,
-        \\};
-        \\
-    ;
-
-    try writer.writeAll(prop_code);
-
-    try writer.print("pub const stage_1 = [{}]u16{{", .{stage1.items.len});
-    for (stage1.items) |v| {
-        _ = try writer.print("{},", .{v});
-    }
-    try writer.writeAll("};\n");
-
-    try writer.print("pub const stage_2 = [{}]u4{{", .{stage2.items.len});
-    for (stage2.items) |v| {
-        _ = try writer.print("{},", .{v});
-    }
-    try writer.writeAll("};\n");
-
-    try writer.print("pub const stage_3 = [{}]Prop{{", .{stage3.items.len});
-    for (stage3.items) |v| {
-        _ = try writer.print(".{s},", .{@tagName(v)});
-    }
-    try writer.writeAll("};\n");
-
-    try out_buf.flush();
-}
diff --git a/src/Grapheme.zig b/src/Grapheme.zig
index d868ef7..ba90b24 100644
--- a/src/Grapheme.zig
+++ b/src/Grapheme.zig
@@ -6,11 +6,7 @@ const unicode = std.unicode;
 const ziglyph = @import("ziglyph");
 const CodePoint = @import("CodePoint.zig");
 const CodePointIterator = CodePoint.CodePointIterator;
-// const emoji = ziglyph.emoji;
-// const gbp = ziglyph.grapheme_break;
 const gbp = @import("gbp");
-const emoji = @import("emoji");
-const indic = @import("indic");
 
 pub const Grapheme = @This();
 
@@ -83,7 +79,10 @@ pub const GraphemeIterator = struct {
 
 // Predicates
 inline fn isBreaker(cp: u21) bool {
-    return cp == '\x0d' or cp == '\x0a' or gbp.stage_3[gbp.stage_2[gbp.stage_1[cp >> 8] + (cp & 0xff)]] == .control;
+    // Extract relevant properties.
+    const cp_props_byte = gbp.stage_3[gbp.stage_2[gbp.stage_1[cp >> 8] + (cp & 0xff)]];
+    const cp_gbp_prop: gbp.Gbp = @enumFromInt(cp_props_byte >> 4);
+    return cp == '\x0d' or cp == '\x0a' or cp_gbp_prop == .Control;
 }
 
 inline fn isIgnorable(cp: u21) bool {
@@ -170,13 +169,20 @@ pub fn graphemeBreak(
     cp2: u21,
     state: *u3,
 ) bool {
+    // Extract relevant properties.
+    const cp1_props_byte = gbp.stage_3[gbp.stage_2[gbp.stage_1[cp1 >> 8] + (cp1 & 0xff)]];
+    const cp1_gbp_prop: gbp.Gbp = @enumFromInt(cp1_props_byte >> 4);
+    const cp1_indic_prop: gbp.Indic = @enumFromInt((cp1_props_byte >> 1) & 0x7);
+    const cp1_is_emoji = cp1_props_byte & 1 == 1;
+
+    const cp2_props_byte = gbp.stage_3[gbp.stage_2[gbp.stage_1[cp2 >> 8] + (cp2 & 0xff)]];
+    const cp2_gbp_prop: gbp.Gbp = @enumFromInt(cp2_props_byte >> 4);
+    const cp2_indic_prop: gbp.Indic = @enumFromInt((cp2_props_byte >> 1) & 0x7);
+    const cp2_is_emoji = cp2_props_byte & 1 == 1;
+
     // GB11: Emoji Extend* ZWJ x Emoji
-    const cp1_is_emoji = emoji.stage_2[emoji.stage_1[cp1 >> 8] + (cp1 & 0xff)];
-    const cp2_is_emoji = emoji.stage_2[emoji.stage_1[cp2 >> 8] + (cp2 & 0xff)];
     if (!hasXpic(state) and cp1_is_emoji) setXpic(state);
     // GB9c: Indic Conjunct Break
-    const cp1_indic_prop = indic.stage_3[indic.stage_2[indic.stage_1[cp1 >> 8] + (cp1 & 0xff)]];
-    const cp2_indic_prop = indic.stage_3[indic.stage_2[indic.stage_1[cp2 >> 8] + (cp2 & 0xff)]];
     if (!hasIndic(state) and cp1_indic_prop == .Consonant) setIndic(state);
 
     // GB3: CR x LF
@@ -186,37 +192,35 @@ pub fn graphemeBreak(
     if (isBreaker(cp1)) return true;
 
     // GB6: Hangul L x (L|V|LV|VT)
-    const cp1_gbp_prop = gbp.stage_3[gbp.stage_2[gbp.stage_1[cp1 >> 8] + (cp1 & 0xff)]];
-    const cp2_gbp_prop = gbp.stage_3[gbp.stage_2[gbp.stage_1[cp2 >> 8] + (cp2 & 0xff)]];
-    if (cp1_gbp_prop == .hangul_l) {
-        if (cp2_gbp_prop == .hangul_l or
-            cp2_gbp_prop == .hangul_v or
-            cp2_gbp_prop == .hangul_lv or
-            cp2_gbp_prop == .hangul_lvt) return false;
+    if (cp1_gbp_prop == .L) {
+        if (cp2_gbp_prop == .L or
+            cp2_gbp_prop == .V or
+            cp2_gbp_prop == .LV or
+            cp2_gbp_prop == .LVT) return false;
     }
 
     // GB7: Hangul (LV | V) x (V | T)
-    if (cp1_gbp_prop == .hangul_lv or cp1_gbp_prop == .hangul_v) {
-        if (cp2_gbp_prop == .hangul_v or
-            cp2_gbp_prop == .hangul_t) return false;
+    if (cp1_gbp_prop == .LV or cp1_gbp_prop == .V) {
+        if (cp2_gbp_prop == .V or
+            cp2_gbp_prop == .T) return false;
     }
 
     // GB8: Hangul (LVT | T) x T
-    if (cp1_gbp_prop == .hangul_lvt or cp1_gbp_prop == .hangul_t) {
-        if (cp2_gbp_prop == .hangul_t) return false;
+    if (cp1_gbp_prop == .LVT or cp1_gbp_prop == .T) {
+        if (cp2_gbp_prop == .T) return false;
     }
 
     // GB9b: x (Extend | ZWJ)
-    if (cp2_gbp_prop == .extend or cp2_gbp_prop == .zwj) return false;
+    if (cp2_gbp_prop == .Extend or cp2_gbp_prop == .ZWJ) return false;
 
     // GB9a: x Spacing
-    if (cp2_gbp_prop == .spacing) return false;
+    if (cp2_gbp_prop == .SpacingMark) return false;
 
     // GB9b: Prepend x
-    if (cp1_gbp_prop == .prepend and !isBreaker(cp2)) return false;
+    if (cp1_gbp_prop == .Prepend and !isBreaker(cp2)) return false;
 
     // GB12, GB13: RI x RI
-    if (cp1_gbp_prop == .regional and cp2_gbp_prop == .regional) {
+    if (cp1_gbp_prop == .Regional_Indicator and cp2_gbp_prop == .Regional_Indicator) {
         if (hasRegional(state)) {
             unsetRegional(state);
             return true;
@@ -228,7 +232,7 @@ pub fn graphemeBreak(
 
     // GB11: Emoji Extend* ZWJ x Emoji
     if (hasXpic(state) and
-        cp1_gbp_prop == .zwj and
+        cp1_gbp_prop == .ZWJ and
         cp2_is_emoji)
     {
         unsetXpic(state);
@@ -251,7 +255,7 @@ pub fn graphemeBreak(
     }
 
     if (hasIndic(state) and
-        (cp1_indic_prop == .Linker or cp1_gbp_prop == .zwj) and
+        (cp1_indic_prop == .Linker or cp1_gbp_prop == .ZWJ) and
         cp2_indic_prop == .Consonant)
     {
         unsetIndic(state);