diff options
Diffstat (limited to 'codegen/case.zig')
| -rw-r--r-- | codegen/case.zig | 145 |
1 files changed, 145 insertions, 0 deletions
diff --git a/codegen/case.zig b/codegen/case.zig new file mode 100644 index 0000000..9dffc7c --- /dev/null +++ b/codegen/case.zig | |||
| @@ -0,0 +1,145 @@ | |||
| 1 | const std = @import("std"); | ||
| 2 | const builtin = @import("builtin"); | ||
| 3 | |||
| 4 | const block_size = 256; | ||
| 5 | const Block = [block_size]u44; | ||
| 6 | |||
| 7 | comptime { | ||
| 8 | if (@bitSizeOf(u44) != 2 * @bitSizeOf(u21) + 2) { | ||
| 9 | @compileError("u44 doesn't have expected bit size."); | ||
| 10 | } | ||
| 11 | } | ||
| 12 | |||
| 13 | const BlockMap = std.HashMap( | ||
| 14 | Block, | ||
| 15 | u16, | ||
| 16 | struct { | ||
| 17 | pub fn hash(_: @This(), k: Block) u64 { | ||
| 18 | var hasher = std.hash.Wyhash.init(0); | ||
| 19 | std.hash.autoHashStrat(&hasher, k, .DeepRecursive); | ||
| 20 | return hasher.final(); | ||
| 21 | } | ||
| 22 | |||
| 23 | pub fn eql(_: @This(), a: Block, b: Block) bool { | ||
| 24 | return std.mem.eql(u44, &a, &b); | ||
| 25 | } | ||
| 26 | }, | ||
| 27 | std.hash_map.default_max_load_percentage, | ||
| 28 | ); | ||
| 29 | |||
| 30 | pub fn main() !void { | ||
| 31 | var arena = std.heap.ArenaAllocator.init(std.heap.page_allocator); | ||
| 32 | defer arena.deinit(); | ||
| 33 | const allocator = arena.allocator(); | ||
| 34 | |||
| 35 | var lower_map = std.AutoHashMap(u21, u21).init(allocator); | ||
| 36 | defer lower_map.deinit(); | ||
| 37 | |||
| 38 | var upper_map = std.AutoHashMap(u21, u21).init(allocator); | ||
| 39 | defer upper_map.deinit(); | ||
| 40 | |||
| 41 | // Process UnicodeData.txt | ||
| 42 | |||
| 43 | var in_reader = std.io.Reader.fixed(@embedFile("UnicodeData.txt")); | ||
| 44 | while (in_reader.takeDelimiterInclusive('\n')) |line| { | ||
| 45 | if (line.len == 0) continue; | ||
| 46 | |||
| 47 | var field_iter = std.mem.splitScalar(u8, line, ';'); | ||
| 48 | var cp: u21 = undefined; | ||
| 49 | |||
| 50 | var i: usize = 0; | ||
| 51 | while (field_iter.next()) |field| : (i += 1) { | ||
| 52 | if (field.len == 0) continue; | ||
| 53 | |||
| 54 | switch (i) { | ||
| 55 | 0 => cp = try std.fmt.parseInt(u21, field, 16), | ||
| 56 | |||
| 57 | 12 => { | ||
| 58 | // Uppercase mapping | ||
| 59 | try upper_map.put(cp, try std.fmt.parseInt(u21, field, 16)); | ||
| 60 | }, | ||
| 61 | |||
| 62 | 13 => { | ||
| 63 | // Lowercase mapping | ||
| 64 | try lower_map.put(cp, try std.fmt.parseInt(u21, field, 16)); | ||
| 65 | }, | ||
| 66 | |||
| 67 | else => {}, | ||
| 68 | } | ||
| 69 | } | ||
| 70 | } else |err| switch (err) { | ||
| 71 | error.EndOfStream => {}, | ||
| 72 | else => { | ||
| 73 | return err; | ||
| 74 | }, | ||
| 75 | } | ||
| 76 | |||
| 77 | var blocks_map = BlockMap.init(allocator); | ||
| 78 | defer blocks_map.deinit(); | ||
| 79 | |||
| 80 | var stage1 = std.array_list.Managed(u16).init(allocator); | ||
| 81 | defer stage1.deinit(); | ||
| 82 | |||
| 83 | var stage2 = std.array_list.Managed(u44).init(allocator); | ||
| 84 | defer stage2.deinit(); | ||
| 85 | |||
| 86 | var block: Block = [_]u44{0} ** block_size; | ||
| 87 | var block_len: u16 = 0; | ||
| 88 | |||
| 89 | for (0..0x110000) |i| { | ||
| 90 | const cp: u21 = @intCast(i); | ||
| 91 | var case_prop: u44 = 0; | ||
| 92 | |||
| 93 | if (lower_map.get(cp)) |lower| { | ||
| 94 | case_prop |= @as(u44, lower) << 2 | 1; | ||
| 95 | } | ||
| 96 | |||
| 97 | if (upper_map.get(cp)) |upper| { | ||
| 98 | case_prop |= @as(u44, upper) << (2 + 21) | 2; | ||
| 99 | } | ||
| 100 | |||
| 101 | block[block_len] = case_prop; | ||
| 102 | block_len += 1; | ||
| 103 | |||
| 104 | if (block_len < block_size and cp != 0x10ffff) continue; | ||
| 105 | |||
| 106 | const gop = try blocks_map.getOrPut(block); | ||
| 107 | if (!gop.found_existing) { | ||
| 108 | gop.value_ptr.* = @intCast(stage2.items.len); | ||
| 109 | try stage2.appendSlice(&block); | ||
| 110 | } | ||
| 111 | |||
| 112 | try stage1.append(gop.value_ptr.*); | ||
| 113 | block_len = 0; | ||
| 114 | } | ||
| 115 | |||
| 116 | var args_iter = try std.process.argsWithAllocator(allocator); | ||
| 117 | defer args_iter.deinit(); | ||
| 118 | _ = args_iter.skip(); | ||
| 119 | const output_path = args_iter.next() orelse @panic("No output file arg!"); | ||
| 120 | |||
| 121 | var write_buf: [4096]u8 = undefined; | ||
| 122 | var out_file = try std.fs.cwd().createFile(output_path, .{}); | ||
| 123 | defer out_file.close(); | ||
| 124 | var writer = out_file.writer(&write_buf); | ||
| 125 | |||
| 126 | try writer.interface.print( | ||
| 127 | \\//! This file is auto-generated. Do not edit. | ||
| 128 | \\ | ||
| 129 | \\pub const s1: [{}]u16 = .{{ | ||
| 130 | , .{stage1.items.len}); | ||
| 131 | for (stage1.items) |entry| try writer.interface.print("{}, ", .{entry}); | ||
| 132 | |||
| 133 | try writer.interface.print( | ||
| 134 | \\ | ||
| 135 | \\}}; | ||
| 136 | \\ | ||
| 137 | \\pub const s2: [{}]u44 = .{{ | ||
| 138 | , .{stage2.items.len}); | ||
| 139 | for (stage2.items) |entry| try writer.interface.print("{}, ", .{entry}); | ||
| 140 | |||
| 141 | try writer.interface.writeAll( | ||
| 142 | \\}; | ||
| 143 | ); | ||
| 144 | try writer.interface.flush(); | ||
| 145 | } | ||