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 --- codegen/emoji.zig | 146 ++++++++++++++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 146 insertions(+) create mode 100644 codegen/emoji.zig (limited to 'codegen/emoji.zig') 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(); +} -- cgit v1.2.3