diff options
| author | 2024-02-27 18:04:32 -0400 | |
|---|---|---|
| committer | 2024-02-27 18:04:32 -0400 | |
| commit | ecd9c2277de17e24fa26aefee955caa10b5b990c (patch) | |
| tree | b76f3e2b9886456f61b0001f4dcae15d00a76856 /codegen | |
| parent | Normalizer 2x faster than Ziglyph; Uses 2x memory (diff) | |
| download | zg-ecd9c2277de17e24fa26aefee955caa10b5b990c.tar.gz zg-ecd9c2277de17e24fa26aefee955caa10b5b990c.tar.xz zg-ecd9c2277de17e24fa26aefee955caa10b5b990c.zip | |
General Category with GenCatData
Diffstat (limited to 'codegen')
| -rw-r--r-- | codegen/gencat.zig | 172 |
1 files changed, 172 insertions, 0 deletions
diff --git a/codegen/gencat.zig b/codegen/gencat.zig new file mode 100644 index 0000000..5407040 --- /dev/null +++ b/codegen/gencat.zig | |||
| @@ -0,0 +1,172 @@ | |||
| 1 | const std = @import("std"); | ||
| 2 | const builtin = @import("builtin"); | ||
| 3 | |||
| 4 | const Gc = enum { | ||
| 5 | Cc, | ||
| 6 | Cf, | ||
| 7 | Cn, | ||
| 8 | Co, | ||
| 9 | Cs, | ||
| 10 | Ll, | ||
| 11 | Lm, | ||
| 12 | Lo, | ||
| 13 | Lt, | ||
| 14 | Lu, | ||
| 15 | Mc, | ||
| 16 | Me, | ||
| 17 | Mn, | ||
| 18 | Nd, | ||
| 19 | Nl, | ||
| 20 | No, | ||
| 21 | Pc, | ||
| 22 | Pd, | ||
| 23 | Pe, | ||
| 24 | Pf, | ||
| 25 | Pi, | ||
| 26 | Po, | ||
| 27 | Ps, | ||
| 28 | Sc, | ||
| 29 | Sk, | ||
| 30 | Sm, | ||
| 31 | So, | ||
| 32 | Zl, | ||
| 33 | Zp, | ||
| 34 | Zs, | ||
| 35 | }; | ||
| 36 | |||
| 37 | const block_size = 256; | ||
| 38 | const Block = [block_size]u5; | ||
| 39 | |||
| 40 | const BlockMap = std.HashMap( | ||
| 41 | Block, | ||
| 42 | u16, | ||
| 43 | struct { | ||
| 44 | pub fn hash(_: @This(), k: Block) u64 { | ||
| 45 | var hasher = std.hash.Wyhash.init(0); | ||
| 46 | std.hash.autoHashStrat(&hasher, k, .DeepRecursive); | ||
| 47 | return hasher.final(); | ||
| 48 | } | ||
| 49 | |||
| 50 | pub fn eql(_: @This(), a: Block, b: Block) bool { | ||
| 51 | return std.mem.eql(u5, &a, &b); | ||
| 52 | } | ||
| 53 | }, | ||
| 54 | std.hash_map.default_max_load_percentage, | ||
| 55 | ); | ||
| 56 | |||
| 57 | pub fn main() !void { | ||
| 58 | var arena = std.heap.ArenaAllocator.init(std.heap.page_allocator); | ||
| 59 | defer arena.deinit(); | ||
| 60 | const allocator = arena.allocator(); | ||
| 61 | |||
| 62 | var flat_map = std.AutoHashMap(u21, u5).init(allocator); | ||
| 63 | defer flat_map.deinit(); | ||
| 64 | |||
| 65 | var line_buf: [4096]u8 = undefined; | ||
| 66 | |||
| 67 | // Process DerivedEastAsianWidth.txt | ||
| 68 | var in_file = try std.fs.cwd().openFile("data/unicode/extracted/DerivedGeneralCategory.txt", .{}); | ||
| 69 | defer in_file.close(); | ||
| 70 | var in_buf = std.io.bufferedReader(in_file.reader()); | ||
| 71 | const in_reader = in_buf.reader(); | ||
| 72 | |||
| 73 | while (try in_reader.readUntilDelimiterOrEof(&line_buf, '\n')) |line| { | ||
| 74 | if (line.len == 0 or line[0] == '#') continue; | ||
| 75 | |||
| 76 | const no_comment = if (std.mem.indexOfScalar(u8, line, '#')) |octo| line[0..octo] else line; | ||
| 77 | |||
| 78 | var field_iter = std.mem.tokenizeAny(u8, no_comment, "; "); | ||
| 79 | var current_code: [2]u21 = undefined; | ||
| 80 | |||
| 81 | var i: usize = 0; | ||
| 82 | while (field_iter.next()) |field| : (i += 1) { | ||
| 83 | switch (i) { | ||
| 84 | 0 => { | ||
| 85 | // Code point(s) | ||
| 86 | if (std.mem.indexOf(u8, field, "..")) |dots| { | ||
| 87 | current_code = .{ | ||
| 88 | try std.fmt.parseInt(u21, field[0..dots], 16), | ||
| 89 | try std.fmt.parseInt(u21, field[dots + 2 ..], 16), | ||
| 90 | }; | ||
| 91 | } else { | ||
| 92 | const code = try std.fmt.parseInt(u21, field, 16); | ||
| 93 | current_code = .{ code, code }; | ||
| 94 | } | ||
| 95 | }, | ||
| 96 | 1 => { | ||
| 97 | // General category | ||
| 98 | const gc = std.meta.stringToEnum(Gc, field) orelse return error.UnknownGenCat; | ||
| 99 | for (current_code[0]..current_code[1] + 1) |cp| try flat_map.put(@intCast(cp), @intFromEnum(gc)); | ||
| 100 | }, | ||
| 101 | else => {}, | ||
| 102 | } | ||
| 103 | } | ||
| 104 | } | ||
| 105 | |||
| 106 | var blocks_map = BlockMap.init(allocator); | ||
| 107 | defer blocks_map.deinit(); | ||
| 108 | |||
| 109 | var stage1 = std.ArrayList(u16).init(allocator); | ||
| 110 | defer stage1.deinit(); | ||
| 111 | |||
| 112 | var stage2 = std.ArrayList(u5).init(allocator); | ||
| 113 | defer stage2.deinit(); | ||
| 114 | |||
| 115 | var stage3 = std.ArrayList(u5).init(allocator); | ||
| 116 | defer stage3.deinit(); | ||
| 117 | |||
| 118 | var block: Block = [_]u5{0} ** block_size; | ||
| 119 | var block_len: u16 = 0; | ||
| 120 | |||
| 121 | for (0..0x110000) |i| { | ||
| 122 | const cp: u21 = @intCast(i); | ||
| 123 | const gc = flat_map.get(cp).?; | ||
| 124 | |||
| 125 | const stage3_idx = blk: { | ||
| 126 | for (stage3.items, 0..) |gci, j| { | ||
| 127 | if (gc == gci) break :blk j; | ||
| 128 | } | ||
| 129 | try stage3.append(gc); | ||
| 130 | break :blk stage3.items.len - 1; | ||
| 131 | }; | ||
| 132 | |||
| 133 | // Process block | ||
| 134 | block[block_len] = @intCast(stage3_idx); | ||
| 135 | block_len += 1; | ||
| 136 | |||
| 137 | if (block_len < block_size and cp != 0x10ffff) continue; | ||
| 138 | |||
| 139 | const gop = try blocks_map.getOrPut(block); | ||
| 140 | if (!gop.found_existing) { | ||
| 141 | gop.value_ptr.* = @intCast(stage2.items.len); | ||
| 142 | try stage2.appendSlice(&block); | ||
| 143 | } | ||
| 144 | |||
| 145 | try stage1.append(gop.value_ptr.*); | ||
| 146 | block_len = 0; | ||
| 147 | } | ||
| 148 | |||
| 149 | var args_iter = try std.process.argsWithAllocator(allocator); | ||
| 150 | defer args_iter.deinit(); | ||
| 151 | _ = args_iter.skip(); | ||
| 152 | const output_path = args_iter.next() orelse @panic("No output file arg!"); | ||
| 153 | |||
| 154 | const compressor = std.compress.deflate.compressor; | ||
| 155 | var out_file = try std.fs.cwd().createFile(output_path, .{}); | ||
| 156 | defer out_file.close(); | ||
| 157 | var out_comp = try compressor(allocator, out_file.writer(), .{ .level = .best_compression }); | ||
| 158 | defer out_comp.deinit(); | ||
| 159 | const writer = out_comp.writer(); | ||
| 160 | |||
| 161 | const endian = builtin.cpu.arch.endian(); | ||
| 162 | try writer.writeInt(u16, @intCast(stage1.items.len), endian); | ||
| 163 | for (stage1.items) |i| try writer.writeInt(u16, i, endian); | ||
| 164 | |||
| 165 | try writer.writeInt(u16, @intCast(stage2.items.len), endian); | ||
| 166 | for (stage2.items) |i| try writer.writeInt(u8, i, endian); | ||
| 167 | |||
| 168 | try writer.writeInt(u8, @intCast(stage3.items.len), endian); | ||
| 169 | for (stage3.items) |i| try writer.writeInt(u8, i, endian); | ||
| 170 | |||
| 171 | try out_comp.flush(); | ||
| 172 | } | ||