From d7df2b4b92e198fbdbe5cfc29095d48980675004 Mon Sep 17 00:00:00 2001 From: Jacob Sandlund Date: Tue, 24 Jun 2025 07:55:16 -0400 Subject: Add Emoji module and codegen/emoji --- README.md | 28 +++++++++++ build.zig | 27 ++++++++++ codegen/emoji.zig | 146 ++++++++++++++++++++++++++++++++++++++++++++++++++++++ src/Emoji.zig | 132 ++++++++++++++++++++++++++++++++++++++++++++++++ 4 files changed, 333 insertions(+) create mode 100644 codegen/emoji.zig create mode 100644 src/Emoji.zig diff --git a/README.md b/README.md index bfa8d5e..5912ce4 100644 --- a/README.md +++ b/README.md @@ -520,6 +520,34 @@ test "Scripts" { } ``` +## Emoji + +To get information about emoji and emoji-like characters, use the `Emoji` module. + +In your `build.zig`: + +```zig +exe.root_module.addImport("Emoji", zg.module("Emoji")); +``` + +In your code: + +```zig +const Emoji = @import("Emoji"); + +test "Emoji" { + const emoji = try Emoji.init(allocator); + defer emoji.deinit(allocator); + + try expect(emoji.isEmoji(0x1F415)); // πŸ• + try expect(emoji.isEmojiPresentation(0x1F408)); // 🐈 + try expect(emoji.isEmojiModifier(0x1F3FF)); // 🏿 + try expect(emoji.isEmojiModifierBase(0x1F977)); // πŸ₯· + try expect(emoji.isEmojiComponent(0x1F9B0)); // 🦰 + try expect(emoji.isExtendedPictographic(0x1F005)); // πŸ€… +} +``` + ## Relation to Ziglyph zg is a total re-write of some of the components of Ziglyph. The idea was to diff --git a/build.zig b/build.zig index 58fd3e7..839cf13 100644 --- a/build.zig +++ b/build.zig @@ -22,6 +22,16 @@ pub fn build(b: *std.Build) void { const run_gbp_gen_exe = b.addRunArtifact(gbp_gen_exe); const gbp_gen_out = run_gbp_gen_exe.addOutputFileArg("gbp.bin.z"); + // Emoji + const emoji_gen_exe = b.addExecutable(.{ + .name = "emoji", + .root_source_file = b.path("codegen/emoji.zig"), + .target = b.graph.host, + .optimize = .Debug, + }); + const run_emoji_gen_exe = b.addRunArtifact(emoji_gen_exe); + const emoji_gen_out = run_emoji_gen_exe.addOutputFileArg("emoji.bin.z"); + // Display width const cjk = b.option(bool, "cjk", "Ambiguous code points are wide (display width: 2).") orelse false; const options = b.addOptions(); @@ -215,6 +225,22 @@ pub fn build(b: *std.Build) void { }); const grapheme_tr = b.addRunArtifact(grapheme_t); + // Emoji module + const emoji = b.addModule("Emoji", .{ + .root_source_file = b.path("src/Emoji.zig"), + .target = target, + .optimize = optimize, + }); + emoji.addAnonymousImport("emoji", .{ .root_source_file = emoji_gen_out }); + + const emoji_t = b.addTest(.{ + .name = "Emoji", + .root_module = emoji, + .target = target, + .optimize = optimize, + }); + const emoji_tr = b.addRunArtifact(emoji_t); + // ASCII utilities const ascii = b.addModule("ascii", .{ .root_source_file = b.path("src/ascii.zig"), @@ -452,6 +478,7 @@ pub fn build(b: *std.Build) void { test_step.dependOn(&code_point_tr.step); test_step.dependOn(&display_width_tr.step); test_step.dependOn(&grapheme_tr.step); + test_step.dependOn(&emoji_tr.step); test_step.dependOn(&ascii_tr.step); test_step.dependOn(&ccc_data_tr.step); test_step.dependOn(&canon_data_tr.step); diff --git a/codegen/emoji.zig b/codegen/emoji.zig new file mode 100644 index 0000000..0a4dbe6 --- /dev/null +++ b/codegen/emoji.zig @@ -0,0 +1,146 @@ +const std = @import("std"); +const builtin = @import("builtin"); + +pub const Emoji = packed struct { + Emoji: bool = false, + Emoji_Presentation: bool = false, + Emoji_Modifier: bool = false, + Emoji_Modifier_Base: bool = false, + Emoji_Component: bool = false, + Extended_Pictographic: bool = false, +}; + +const block_size = 256; +const Block = [block_size]u6; + +comptime { + if (@bitSizeOf(u6) != @bitSizeOf(Emoji)) { + @compileError("Emoji doesn't have expected bit size."); + } +} + +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(u6, &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 emoji_map = std.AutoHashMap(u21, Emoji).init(allocator); + defer emoji_map.deinit(); + + var line_buf: [4096]u8 = undefined; + + // Process Emoji + var in_file = try std.fs.cwd().openFile("data/unicode/emoji/emoji-data.txt", .{}); + defer in_file.close(); + var in_buf = std.io.bufferedReader(in_file.reader()); + const in_reader = in_buf.reader(); + + while (try in_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(std.meta.FieldEnum(Emoji), field) orelse return error.InvalidProp; + for (current_code[0]..current_code[1] + 1) |code| { + const cp: u21 = @intCast(code); + const gop = try emoji_map.getOrPut(cp); + if (!gop.found_existing) gop.value_ptr.* = .{}; + switch (prop) { + inline else => |tag| { + @field(gop.value_ptr.*, @tagName(tag)) = true; + }, + } + } + }, + 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(u6).init(allocator); + defer stage2.deinit(); + + var block: Block = [_]u6{0} ** block_size; + var block_len: u16 = 0; + + for (0..0x110000) |i| { + const cp: u21 = @intCast(i); + const emoji = emoji_map.get(cp) orelse Emoji{}; + + block[block_len] = @bitCast(emoji); + 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 = try std.process.argsWithAllocator(allocator); + defer args_iter.deinit(); + _ = args_iter.skip(); + const output_path = args_iter.next() orelse @panic("No output file arg!"); + + const compressor = std.compress.flate.deflate.compressor; + var out_file = try std.fs.cwd().createFile(output_path, .{}); + defer out_file.close(); + var out_comp = try compressor(.raw, out_file.writer(), .{ .level = .best }); + const writer = out_comp.writer(); + + const endian = builtin.cpu.arch.endian(); + try writer.writeInt(u16, @intCast(stage1.items.len), endian); + for (stage1.items) |i| try writer.writeInt(u16, i, endian); + + try writer.writeInt(u16, @intCast(stage2.items.len), endian); + for (stage2.items) |i| try writer.writeInt(u8, i, endian); + + try out_comp.flush(); +} diff --git a/src/Emoji.zig b/src/Emoji.zig new file mode 100644 index 0000000..bf7014d --- /dev/null +++ b/src/Emoji.zig @@ -0,0 +1,132 @@ +const std = @import("std"); +const builtin = @import("builtin"); +const mem = std.mem; +const Allocator = mem.Allocator; +const compress = std.compress; +const unicode = std.unicode; + +const CodePoint = @import("code_point").CodePoint; +const CodePointIterator = @import("code_point").Iterator; + +s1: []u16 = undefined, +s2: []u6 = undefined, + +const Emoji = @This(); + +// This must be an exact match of `Emoji` from `codegen/emoji.zig`. +pub const Properties = packed struct { + Emoji: bool = false, + Emoji_Presentation: bool = false, + Emoji_Modifier: bool = false, + Emoji_Modifier_Base: bool = false, + Emoji_Component: bool = false, + Extended_Pictographic: bool = false, +}; + +pub fn init(allocator: Allocator) Allocator.Error!Emoji { + var emoji = Emoji{}; + try emoji.setup(allocator); + return emoji; +} + +pub fn setup(emoji: *Emoji, allocator: Allocator) Allocator.Error!void { + const decompressor = compress.flate.inflate.decompressor; + const in_bytes = @embedFile("emoji"); + var in_fbs = std.io.fixedBufferStream(in_bytes); + var in_decomp = decompressor(.raw, in_fbs.reader()); + var reader = in_decomp.reader(); + + const endian = builtin.cpu.arch.endian(); + + const s1_len: u16 = reader.readInt(u16, endian) catch unreachable; + emoji.s1 = try allocator.alloc(u16, s1_len); + errdefer allocator.free(emoji.s1); + for (0..s1_len) |i| emoji.s1[i] = reader.readInt(u16, endian) catch unreachable; + + const s2_len: u16 = reader.readInt(u16, endian) catch unreachable; + emoji.s2 = try allocator.alloc(u6, s2_len); + errdefer allocator.free(emoji.s2); + for (0..s2_len) |i| emoji.s2[i] = @intCast(reader.readInt(u8, endian) catch unreachable); +} + +pub fn deinit(emoji: *const Emoji, allocator: Allocator) void { + allocator.free(emoji.s1); + allocator.free(emoji.s2); +} + +/// Lookup the emoji properties for a code point. +fn properties(emoji: Emoji, cp: u21) Properties { + return @bitCast(emoji.s2[emoji.s1[cp >> 8] + (cp & 0xff)]); +} + +pub fn isEmoji(emoji: Emoji, cp: u21) bool { + return properties(emoji, cp).Emoji; +} + +pub fn isEmojiPresentation(emoji: Emoji, cp: u21) bool { + return properties(emoji, cp).Emoji_Presentation; +} + +pub fn isEmojiModifier(emoji: Emoji, cp: u21) bool { + return properties(emoji, cp).Emoji_Modifier; +} + +pub fn isEmojiModifierBase(emoji: Emoji, cp: u21) bool { + return properties(emoji, cp).Emoji_Modifier_Base; +} + +pub fn isEmojiComponent(emoji: Emoji, cp: u21) bool { + return properties(emoji, cp).Emoji_Component; +} + +pub fn isExtendedPictographic(emoji: Emoji, cp: u21) bool { + return properties(emoji, cp).Extended_Pictographic; +} + +test "isEmoji" { + const emoji = try Emoji.init(std.testing.allocator); + defer emoji.deinit(std.testing.allocator); + + try std.testing.expect(emoji.isEmoji(0x1F415)); // πŸ• + try std.testing.expect(!emoji.isEmoji(0x3042)); // あ +} + +test "isEmojiPresentation" { + const emoji = try Emoji.init(std.testing.allocator); + defer emoji.deinit(std.testing.allocator); + + try std.testing.expect(emoji.isEmojiPresentation(0x1F408)); // 🐈 + try std.testing.expect(!emoji.isEmojiPresentation(0x267E)); // ♾️ +} + +test "isEmojiModifier" { + const emoji = try Emoji.init(std.testing.allocator); + defer emoji.deinit(std.testing.allocator); + + try std.testing.expect(emoji.isEmojiModifier(0x1F3FF)); // 🏿 + try std.testing.expect(!emoji.isEmojiModifier(0x1F385)); // πŸŽ… +} + +test "isEmojiModifierBase" { + const emoji = try Emoji.init(std.testing.allocator); + defer emoji.deinit(std.testing.allocator); + + try std.testing.expect(emoji.isEmojiModifierBase(0x1F977)); // πŸ₯· + try std.testing.expect(!emoji.isEmojiModifierBase(0x1F4F8)); // πŸ“Έ +} + +test "isEmojiComponent" { + const emoji = try Emoji.init(std.testing.allocator); + defer emoji.deinit(std.testing.allocator); + + try std.testing.expect(emoji.isEmojiComponent(0x1F9B0)); // 🦰 + try std.testing.expect(!emoji.isEmojiComponent(0x1F9B5)); // 🦡 +} + +test "isExtendedPictographic" { + const emoji = try Emoji.init(std.testing.allocator); + defer emoji.deinit(std.testing.allocator); + + try std.testing.expect(emoji.isExtendedPictographic(0x1F005)); // πŸ€… + try std.testing.expect(!emoji.isExtendedPictographic(0x2A)); // * +} -- cgit v1.2.3