diff options
| author | 2024-02-11 19:08:34 -0400 | |
|---|---|---|
| committer | 2024-02-11 19:08:34 -0400 | |
| commit | ed5ce42ba8fc67db8a8bb385490ccec7a218a7e3 (patch) | |
| tree | 807136431dd297ccc273895f09bc7152e5347bb2 /src/gbp_gen.zig | |
| download | zg-ed5ce42ba8fc67db8a8bb385490ccec7a218a7e3.tar.gz zg-ed5ce42ba8fc67db8a8bb385490ccec7a218a7e3.tar.xz zg-ed5ce42ba8fc67db8a8bb385490ccec7a218a7e3.zip | |
init
Diffstat (limited to 'src/gbp_gen.zig')
| -rw-r--r-- | src/gbp_gen.zig | 178 |
1 files changed, 178 insertions, 0 deletions
diff --git a/src/gbp_gen.zig b/src/gbp_gen.zig new file mode 100644 index 0000000..578543d --- /dev/null +++ b/src/gbp_gen.zig | |||
| @@ -0,0 +1,178 @@ | |||
| 1 | const std = @import("std"); | ||
| 2 | |||
| 3 | const gbp = @import("ziglyph").grapheme_break; | ||
| 4 | |||
| 5 | const Map = struct { | ||
| 6 | store: [12]Prop = [_]Prop{.none} ** 12, | ||
| 7 | len: u8 = 0, | ||
| 8 | |||
| 9 | fn getOrPut(self: *Map, prop: Prop) usize { | ||
| 10 | var index: ?usize = null; | ||
| 11 | for (0..self.store.len) |i| { | ||
| 12 | if (self.store[i] == prop) index = i; | ||
| 13 | } | ||
| 14 | |||
| 15 | if (index) |idx| { | ||
| 16 | return idx; | ||
| 17 | } else { | ||
| 18 | self.store[self.len] = prop; | ||
| 19 | self.len += 1; | ||
| 20 | return self.len - 1; | ||
| 21 | } | ||
| 22 | } | ||
| 23 | }; | ||
| 24 | |||
| 25 | const Prop = enum { | ||
| 26 | none, | ||
| 27 | |||
| 28 | control, | ||
| 29 | extend, | ||
| 30 | hangul_l, | ||
| 31 | hangul_lv, | ||
| 32 | hangul_lvt, | ||
| 33 | hangul_v, | ||
| 34 | hangul_t, | ||
| 35 | prepend, | ||
| 36 | regional, | ||
| 37 | spacing, | ||
| 38 | zwj, | ||
| 39 | |||
| 40 | fn forCodePoint(cp: u21) Prop { | ||
| 41 | if (gbp.isControl(cp)) return .control; | ||
| 42 | if (gbp.isExtend(cp)) return .extend; | ||
| 43 | if (gbp.isL(cp)) return .hangul_l; | ||
| 44 | if (gbp.isLv(cp)) return .hangul_lv; | ||
| 45 | if (gbp.isLvt(cp)) return .hangul_lvt; | ||
| 46 | if (gbp.isT(cp)) return .hangul_t; | ||
| 47 | if (gbp.isV(cp)) return .hangul_v; | ||
| 48 | if (gbp.isPrepend(cp)) return .prepend; | ||
| 49 | if (gbp.isRegionalIndicator(cp)) return .regional; | ||
| 50 | if (gbp.isSpacingmark(cp)) return .spacing; | ||
| 51 | if (gbp.isZwj(cp)) return .zwj; | ||
| 52 | |||
| 53 | return .none; | ||
| 54 | } | ||
| 55 | }; | ||
| 56 | |||
| 57 | pub fn main() !void { | ||
| 58 | var stage_1: [4352]u21 = undefined; | ||
| 59 | var stage_2: [1_114_112]u4 = undefined; | ||
| 60 | var stage_3 = Map{}; | ||
| 61 | |||
| 62 | var current_block_offset: u21 = 0; | ||
| 63 | |||
| 64 | for (0..0x10ffff + 1) |i| { | ||
| 65 | const cp: u21 = @intCast(i); | ||
| 66 | const stage_1_index = cp >> 8; | ||
| 67 | const stage_2_index = current_block_offset + (cp & 0xff); | ||
| 68 | const stage_3_index = stage_3.getOrPut(Prop.forCodePoint(cp)); | ||
| 69 | stage_1[stage_1_index] = current_block_offset; | ||
| 70 | stage_2[stage_2_index] = @intCast(stage_3_index); | ||
| 71 | if (cp & 0xff == 255) current_block_offset += 256; | ||
| 72 | } | ||
| 73 | |||
| 74 | var args_iter = std.process.args(); | ||
| 75 | _ = args_iter.skip(); | ||
| 76 | const output_path = args_iter.next() orelse @panic("No output file arg!"); | ||
| 77 | |||
| 78 | var out_file = try std.fs.cwd().createFile(output_path, .{}); | ||
| 79 | defer out_file.close(); | ||
| 80 | var out_buf = std.io.bufferedWriter(out_file.writer()); | ||
| 81 | const writer = out_buf.writer(); | ||
| 82 | |||
| 83 | const prop_code = | ||
| 84 | \\const Prop = enum { | ||
| 85 | \\ none, | ||
| 86 | \\ | ||
| 87 | \\ control, | ||
| 88 | \\ extend, | ||
| 89 | \\ hangul_l, | ||
| 90 | \\ hangul_lv, | ||
| 91 | \\ hangul_lvt, | ||
| 92 | \\ hangul_v, | ||
| 93 | \\ hangul_t, | ||
| 94 | \\ prepend, | ||
| 95 | \\ regional, | ||
| 96 | \\ spacing, | ||
| 97 | \\ zwj, | ||
| 98 | \\}; | ||
| 99 | \\ | ||
| 100 | ; | ||
| 101 | |||
| 102 | try writer.writeAll(prop_code); | ||
| 103 | |||
| 104 | try writer.writeAll("const stage_1 = [_]u21{"); | ||
| 105 | for (stage_1, 0..) |v, i| { | ||
| 106 | if (i != 0) try writer.writeByte(','); | ||
| 107 | _ = try writer.print("{}", .{v}); | ||
| 108 | } | ||
| 109 | try writer.writeAll("};\n"); | ||
| 110 | |||
| 111 | try writer.writeAll("const stage_2 = [_]u4{"); | ||
| 112 | for (stage_2, 0..) |v, i| { | ||
| 113 | if (i != 0) try writer.writeByte(','); | ||
| 114 | _ = try writer.print("{}", .{v}); | ||
| 115 | } | ||
| 116 | try writer.writeAll("};\n"); | ||
| 117 | |||
| 118 | try writer.writeAll("const stage_3 = [_]Prop{"); | ||
| 119 | for (stage_3.store, 0..) |v, i| { | ||
| 120 | if (i != 0) try writer.writeByte(','); | ||
| 121 | _ = try writer.print(".{s}", .{@tagName(v)}); | ||
| 122 | } | ||
| 123 | try writer.writeAll("};\n"); | ||
| 124 | |||
| 125 | const code = | ||
| 126 | \\inline fn getProp(cp: u21) Prop { | ||
| 127 | \\ const stage_1_index = cp >> 8; | ||
| 128 | \\ const stage_2_index = stage_1[stage_1_index] + (cp & 0xff); | ||
| 129 | \\ const stage_3_index = stage_2[stage_2_index]; | ||
| 130 | \\ return stage_3[stage_3_index]; | ||
| 131 | \\} | ||
| 132 | \\ | ||
| 133 | \\pub inline fn isControl(cp: u21) bool { | ||
| 134 | \\ return getProp(cp) == .control; | ||
| 135 | \\} | ||
| 136 | \\ | ||
| 137 | \\pub inline fn isExtend(cp: u21) bool { | ||
| 138 | \\ return getProp(cp) == .extend; | ||
| 139 | \\} | ||
| 140 | \\ | ||
| 141 | \\pub inline fn isL(cp: u21) bool { | ||
| 142 | \\ return getProp(cp) == .hangul_l; | ||
| 143 | \\} | ||
| 144 | \\pub inline fn isLv(cp: u21) bool { | ||
| 145 | \\ return getProp(cp) == .hangul_lv; | ||
| 146 | \\} | ||
| 147 | \\pub inline fn isLvt(cp: u21) bool { | ||
| 148 | \\ return getProp(cp) == .hangul_lvt; | ||
| 149 | \\} | ||
| 150 | \\pub inline fn isV(cp: u21) bool { | ||
| 151 | \\ return getProp(cp) == .hangul_v; | ||
| 152 | \\} | ||
| 153 | \\pub inline fn isT(cp: u21) bool { | ||
| 154 | \\ return getProp(cp) == .hangul_t; | ||
| 155 | \\} | ||
| 156 | \\ | ||
| 157 | \\pub inline fn isPrepend(cp: u21) bool { | ||
| 158 | \\ return getProp(cp) == .prepend; | ||
| 159 | \\} | ||
| 160 | \\ | ||
| 161 | \\pub inline fn isRegionalIndicator(cp: u21) bool { | ||
| 162 | \\ return getProp(cp) == .regional; | ||
| 163 | \\} | ||
| 164 | \\ | ||
| 165 | \\pub inline fn isSpacingmark(cp: u21) bool { | ||
| 166 | \\ return getProp(cp) == .spacing; | ||
| 167 | \\} | ||
| 168 | \\ | ||
| 169 | \\pub inline fn isZwj(cp: u21) bool { | ||
| 170 | \\ return getProp(cp) == .zwj; | ||
| 171 | \\} | ||
| 172 | \\ | ||
| 173 | ; | ||
| 174 | |||
| 175 | try writer.writeAll(code); | ||
| 176 | |||
| 177 | try out_buf.flush(); | ||
| 178 | } | ||