diff options
Diffstat (limited to 'src/gbp_gen.zig')
| -rw-r--r-- | src/gbp_gen.zig | 71 |
1 files changed, 18 insertions, 53 deletions
diff --git a/src/gbp_gen.zig b/src/gbp_gen.zig index 578543d..7673931 100644 --- a/src/gbp_gen.zig +++ b/src/gbp_gen.zig | |||
| @@ -2,26 +2,6 @@ const std = @import("std"); | |||
| 2 | 2 | ||
| 3 | const gbp = @import("ziglyph").grapheme_break; | 3 | const gbp = @import("ziglyph").grapheme_break; |
| 4 | 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 { | 5 | const Prop = enum { |
| 26 | none, | 6 | none, |
| 27 | 7 | ||
| @@ -55,22 +35,20 @@ const Prop = enum { | |||
| 55 | }; | 35 | }; |
| 56 | 36 | ||
| 57 | pub fn main() !void { | 37 | pub fn main() !void { |
| 58 | var stage_1: [4352]u21 = undefined; | 38 | var a = [_]?Prop{null} ** 1_114_112; |
| 59 | var stage_2: [1_114_112]u4 = undefined; | ||
| 60 | var stage_3 = Map{}; | ||
| 61 | |||
| 62 | var current_block_offset: u21 = 0; | ||
| 63 | 39 | ||
| 64 | for (0..0x10ffff + 1) |i| { | 40 | // for ('\u{0}'..'\u{10ffff}') |i| { |
| 41 | for ('\u{0}'..'\u{10}') |i| { | ||
| 65 | const cp: u21 = @intCast(i); | 42 | const cp: u21 = @intCast(i); |
| 66 | const stage_1_index = cp >> 8; | 43 | const prop = Prop.forCodePoint(cp); |
| 67 | const stage_2_index = current_block_offset + (cp & 0xff); | 44 | if (prop == .none) continue; |
| 68 | const stage_3_index = stage_3.getOrPut(Prop.forCodePoint(cp)); | 45 | a[cp] = prop; |
| 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 | } | 46 | } |
| 73 | 47 | ||
| 48 | const cp = '\u{10ffff}'; | ||
| 49 | const prop = Prop.forCodePoint(cp); | ||
| 50 | if (prop != .none) a[cp] = prop; | ||
| 51 | |||
| 74 | var args_iter = std.process.args(); | 52 | var args_iter = std.process.args(); |
| 75 | _ = args_iter.skip(); | 53 | _ = args_iter.skip(); |
| 76 | const output_path = args_iter.next() orelse @panic("No output file arg!"); | 54 | const output_path = args_iter.next() orelse @panic("No output file arg!"); |
| @@ -101,33 +79,20 @@ pub fn main() !void { | |||
| 101 | 79 | ||
| 102 | try writer.writeAll(prop_code); | 80 | try writer.writeAll(prop_code); |
| 103 | 81 | ||
| 104 | try writer.writeAll("const stage_1 = [_]u21{"); | 82 | try writer.writeAll("const array = [_]?Prop{"); |
| 105 | for (stage_1, 0..) |v, i| { | 83 | for (&a, 0..) |v, i| { |
| 106 | if (i != 0) try writer.writeByte(','); | 84 | if (i != 0) try writer.writeByte(','); |
| 107 | _ = try writer.print("{}", .{v}); | 85 | if (v) |p| { |
| 108 | } | 86 | _ = try writer.print(".{s}", .{@tagName(p)}); |
| 109 | try writer.writeAll("};\n"); | 87 | } else { |
| 110 | 88 | try writer.writeAll("null"); | |
| 111 | try writer.writeAll("const stage_2 = [_]u4{"); | 89 | } |
| 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 | } | 90 | } |
| 123 | try writer.writeAll("};\n"); | 91 | try writer.writeAll("};\n"); |
| 124 | 92 | ||
| 125 | const code = | 93 | const code = |
| 126 | \\inline fn getProp(cp: u21) Prop { | 94 | \\inline fn getProp(cp: u21) Prop { |
| 127 | \\ const stage_1_index = cp >> 8; | 95 | \\ return if (array[cp]) |prop| prop else .none; |
| 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 | \\} | 96 | \\} |
| 132 | \\ | 97 | \\ |
| 133 | \\pub inline fn isControl(cp: u21) bool { | 98 | \\pub inline fn isControl(cp: u21) bool { |