diff options
| author | 2026-02-05 07:30:22 -0500 | |
|---|---|---|
| committer | 2026-02-05 07:30:22 -0500 | |
| commit | 80d679d84701a63c0fd3eccd7d3a71a32fab8943 (patch) | |
| tree | 0f50ddce3770bd4d95dd4b045d023f84c4cd5fa8 /codegen/emoji.zig | |
| parent | Delete upper and lower.zig (diff) | |
| parent | Merge branch 'master' into emoji (diff) | |
| download | zg-80d679d84701a63c0fd3eccd7d3a71a32fab8943.tar.gz zg-80d679d84701a63c0fd3eccd7d3a71a32fab8943.tar.xz zg-80d679d84701a63c0fd3eccd7d3a71a32fab8943.zip | |
Merge remote-tracking branch 'jacob/emoji' into no-allocation
Thanks again, Jacob!
Diffstat (limited to 'codegen/emoji.zig')
| -rw-r--r-- | codegen/emoji.zig | 146 |
1 files changed, 146 insertions, 0 deletions
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 @@ | |||
| 1 | const std = @import("std"); | ||
| 2 | const builtin = @import("builtin"); | ||
| 3 | |||
| 4 | pub const Emoji = packed struct { | ||
| 5 | Emoji: bool = false, | ||
| 6 | Emoji_Presentation: bool = false, | ||
| 7 | Emoji_Modifier: bool = false, | ||
| 8 | Emoji_Modifier_Base: bool = false, | ||
| 9 | Emoji_Component: bool = false, | ||
| 10 | Extended_Pictographic: bool = false, | ||
| 11 | }; | ||
| 12 | |||
| 13 | const block_size = 256; | ||
| 14 | const Block = [block_size]u6; | ||
| 15 | |||
| 16 | comptime { | ||
| 17 | if (@bitSizeOf(u6) != @bitSizeOf(Emoji)) { | ||
| 18 | @compileError("Emoji doesn't have expected bit size."); | ||
| 19 | } | ||
| 20 | } | ||
| 21 | |||
| 22 | const BlockMap = std.HashMap( | ||
| 23 | Block, | ||
| 24 | u16, | ||
| 25 | struct { | ||
| 26 | pub fn hash(_: @This(), k: Block) u64 { | ||
| 27 | var hasher = std.hash.Wyhash.init(0); | ||
| 28 | std.hash.autoHashStrat(&hasher, k, .DeepRecursive); | ||
| 29 | return hasher.final(); | ||
| 30 | } | ||
| 31 | |||
| 32 | pub fn eql(_: @This(), a: Block, b: Block) bool { | ||
| 33 | return std.mem.eql(u6, &a, &b); | ||
| 34 | } | ||
| 35 | }, | ||
| 36 | std.hash_map.default_max_load_percentage, | ||
| 37 | ); | ||
| 38 | |||
| 39 | pub fn main() !void { | ||
| 40 | var arena = std.heap.ArenaAllocator.init(std.heap.page_allocator); | ||
| 41 | defer arena.deinit(); | ||
| 42 | const allocator = arena.allocator(); | ||
| 43 | |||
| 44 | var emoji_map = std.AutoHashMap(u21, Emoji).init(allocator); | ||
| 45 | defer emoji_map.deinit(); | ||
| 46 | |||
| 47 | var line_buf: [4096]u8 = undefined; | ||
| 48 | |||
| 49 | // Process Emoji | ||
| 50 | var in_file = try std.fs.cwd().openFile("data/unicode/emoji/emoji-data.txt", .{}); | ||
| 51 | defer in_file.close(); | ||
| 52 | var in_buf = std.io.bufferedReader(in_file.reader()); | ||
| 53 | const in_reader = in_buf.reader(); | ||
| 54 | |||
| 55 | while (try in_reader.readUntilDelimiterOrEof(&line_buf, '\n')) |line| { | ||
| 56 | if (line.len == 0 or line[0] == '#') continue; | ||
| 57 | const no_comment = if (std.mem.indexOfScalar(u8, line, '#')) |octo| line[0..octo] else line; | ||
| 58 | |||
| 59 | var field_iter = std.mem.tokenizeAny(u8, no_comment, "; "); | ||
| 60 | var current_code: [2]u21 = undefined; | ||
| 61 | |||
| 62 | var i: usize = 0; | ||
| 63 | while (field_iter.next()) |field| : (i += 1) { | ||
| 64 | switch (i) { | ||
| 65 | 0 => { | ||
| 66 | // Code point(s) | ||
| 67 | if (std.mem.indexOf(u8, field, "..")) |dots| { | ||
| 68 | current_code = .{ | ||
| 69 | try std.fmt.parseInt(u21, field[0..dots], 16), | ||
| 70 | try std.fmt.parseInt(u21, field[dots + 2 ..], 16), | ||
| 71 | }; | ||
| 72 | } else { | ||
| 73 | const code = try std.fmt.parseInt(u21, field, 16); | ||
| 74 | current_code = .{ code, code }; | ||
| 75 | } | ||
| 76 | }, | ||
| 77 | 1 => { | ||
| 78 | // Prop | ||
| 79 | const prop = std.meta.stringToEnum(std.meta.FieldEnum(Emoji), field) orelse return error.InvalidProp; | ||
| 80 | for (current_code[0]..current_code[1] + 1) |code| { | ||
| 81 | const cp: u21 = @intCast(code); | ||
| 82 | const gop = try emoji_map.getOrPut(cp); | ||
| 83 | if (!gop.found_existing) gop.value_ptr.* = .{}; | ||
| 84 | switch (prop) { | ||
| 85 | inline else => |tag| { | ||
| 86 | @field(gop.value_ptr.*, @tagName(tag)) = true; | ||
| 87 | }, | ||
| 88 | } | ||
| 89 | } | ||
| 90 | }, | ||
| 91 | else => {}, | ||
| 92 | } | ||
| 93 | } | ||
| 94 | } | ||
| 95 | |||
| 96 | var blocks_map = BlockMap.init(allocator); | ||
| 97 | defer blocks_map.deinit(); | ||
| 98 | |||
| 99 | var stage1 = std.ArrayList(u16).init(allocator); | ||
| 100 | defer stage1.deinit(); | ||
| 101 | |||
| 102 | var stage2 = std.ArrayList(u6).init(allocator); | ||
| 103 | defer stage2.deinit(); | ||
| 104 | |||
| 105 | var block: Block = [_]u6{0} ** block_size; | ||
| 106 | var block_len: u16 = 0; | ||
| 107 | |||
| 108 | for (0..0x110000) |i| { | ||
| 109 | const cp: u21 = @intCast(i); | ||
| 110 | const emoji = emoji_map.get(cp) orelse Emoji{}; | ||
| 111 | |||
| 112 | block[block_len] = @bitCast(emoji); | ||
| 113 | block_len += 1; | ||
| 114 | |||
| 115 | if (block_len < block_size and cp != 0x10ffff) continue; | ||
| 116 | |||
| 117 | const gop = try blocks_map.getOrPut(block); | ||
| 118 | if (!gop.found_existing) { | ||
| 119 | gop.value_ptr.* = @intCast(stage2.items.len); | ||
| 120 | try stage2.appendSlice(&block); | ||
| 121 | } | ||
| 122 | |||
| 123 | try stage1.append(gop.value_ptr.*); | ||
| 124 | block_len = 0; | ||
| 125 | } | ||
| 126 | |||
| 127 | var args_iter = try std.process.argsWithAllocator(allocator); | ||
| 128 | defer args_iter.deinit(); | ||
| 129 | _ = args_iter.skip(); | ||
| 130 | const output_path = args_iter.next() orelse @panic("No output file arg!"); | ||
| 131 | |||
| 132 | const compressor = std.compress.flate.deflate.compressor; | ||
| 133 | var out_file = try std.fs.cwd().createFile(output_path, .{}); | ||
| 134 | defer out_file.close(); | ||
| 135 | var out_comp = try compressor(.raw, out_file.writer(), .{ .level = .best }); | ||
| 136 | const writer = out_comp.writer(); | ||
| 137 | |||
| 138 | const endian = builtin.cpu.arch.endian(); | ||
| 139 | try writer.writeInt(u16, @intCast(stage1.items.len), endian); | ||
| 140 | for (stage1.items) |i| try writer.writeInt(u16, i, endian); | ||
| 141 | |||
| 142 | try writer.writeInt(u16, @intCast(stage2.items.len), endian); | ||
| 143 | for (stage2.items) |i| try writer.writeInt(u8, i, endian); | ||
| 144 | |||
| 145 | try out_comp.flush(); | ||
| 146 | } | ||