diff options
Diffstat (limited to '')
| -rw-r--r-- | codegen/case_prop.zig | 136 |
1 files changed, 136 insertions, 0 deletions
diff --git a/codegen/case_prop.zig b/codegen/case_prop.zig new file mode 100644 index 0000000..ce7ee0d --- /dev/null +++ b/codegen/case_prop.zig | |||
| @@ -0,0 +1,136 @@ | |||
| 1 | const std = @import("std"); | ||
| 2 | const builtin = @import("builtin"); | ||
| 3 | const mem = std.mem; | ||
| 4 | |||
| 5 | const block_size = 256; | ||
| 6 | const Block = [block_size]u8; | ||
| 7 | |||
| 8 | const BlockMap = std.HashMap( | ||
| 9 | Block, | ||
| 10 | u16, | ||
| 11 | struct { | ||
| 12 | pub fn hash(_: @This(), k: Block) u64 { | ||
| 13 | var hasher = std.hash.Wyhash.init(0); | ||
| 14 | std.hash.autoHashStrat(&hasher, k, .DeepRecursive); | ||
| 15 | return hasher.final(); | ||
| 16 | } | ||
| 17 | |||
| 18 | pub fn eql(_: @This(), a: Block, b: Block) bool { | ||
| 19 | return mem.eql(u8, &a, &b); | ||
| 20 | } | ||
| 21 | }, | ||
| 22 | std.hash_map.default_max_load_percentage, | ||
| 23 | ); | ||
| 24 | |||
| 25 | pub fn main() !void { | ||
| 26 | var arena = std.heap.ArenaAllocator.init(std.heap.page_allocator); | ||
| 27 | defer arena.deinit(); | ||
| 28 | const allocator = arena.allocator(); | ||
| 29 | |||
| 30 | var flat_map = std.AutoHashMap(u21, u8).init(allocator); | ||
| 31 | defer flat_map.deinit(); | ||
| 32 | |||
| 33 | var line_buf: [4096]u8 = undefined; | ||
| 34 | |||
| 35 | // Process DerivedCoreProperties.txt | ||
| 36 | var in_file = try std.fs.cwd().openFile("data/unicode/DerivedCoreProperties.txt", .{}); | ||
| 37 | defer in_file.close(); | ||
| 38 | var in_buf = std.io.bufferedReader(in_file.reader()); | ||
| 39 | const in_reader = in_buf.reader(); | ||
| 40 | |||
| 41 | while (try in_reader.readUntilDelimiterOrEof(&line_buf, '\n')) |line| { | ||
| 42 | if (line.len == 0 or line[0] == '#') continue; | ||
| 43 | const no_comment = if (mem.indexOfScalar(u8, line, '#')) |octo| line[0..octo] else line; | ||
| 44 | |||
| 45 | var field_iter = mem.tokenizeAny(u8, no_comment, "; "); | ||
| 46 | var current_code: [2]u21 = undefined; | ||
| 47 | |||
| 48 | var i: usize = 0; | ||
| 49 | while (field_iter.next()) |field| : (i += 1) { | ||
| 50 | switch (i) { | ||
| 51 | 0 => { | ||
| 52 | // Code point(s) | ||
| 53 | if (mem.indexOf(u8, field, "..")) |dots| { | ||
| 54 | current_code = .{ | ||
| 55 | try std.fmt.parseInt(u21, field[0..dots], 16), | ||
| 56 | try std.fmt.parseInt(u21, field[dots + 2 ..], 16), | ||
| 57 | }; | ||
| 58 | } else { | ||
| 59 | const code = try std.fmt.parseInt(u21, field, 16); | ||
| 60 | current_code = .{ code, code }; | ||
| 61 | } | ||
| 62 | }, | ||
| 63 | 1 => { | ||
| 64 | // Props | ||
| 65 | var bit: u8 = 0; | ||
| 66 | |||
| 67 | if (mem.eql(u8, field, "Lowercase")) bit = 1; | ||
| 68 | if (mem.eql(u8, field, "Uppercase")) bit = 2; | ||
| 69 | if (mem.eql(u8, field, "Cased")) bit = 4; | ||
| 70 | |||
| 71 | if (bit != 0) { | ||
| 72 | for (current_code[0]..current_code[1] + 1) |cp| { | ||
| 73 | const gop = try flat_map.getOrPut(@intCast(cp)); | ||
| 74 | if (!gop.found_existing) gop.value_ptr.* = 0; | ||
| 75 | gop.value_ptr.* |= bit; | ||
| 76 | } | ||
| 77 | } | ||
| 78 | }, | ||
| 79 | else => {}, | ||
| 80 | } | ||
| 81 | } | ||
| 82 | } | ||
| 83 | |||
| 84 | var blocks_map = BlockMap.init(allocator); | ||
| 85 | defer blocks_map.deinit(); | ||
| 86 | |||
| 87 | var stage1 = std.ArrayList(u16).init(allocator); | ||
| 88 | defer stage1.deinit(); | ||
| 89 | |||
| 90 | var stage2 = std.ArrayList(u8).init(allocator); | ||
| 91 | defer stage2.deinit(); | ||
| 92 | |||
| 93 | var block: Block = [_]u8{0} ** block_size; | ||
| 94 | var block_len: u16 = 0; | ||
| 95 | |||
| 96 | for (0..0x110000) |i| { | ||
| 97 | const cp: u21 = @intCast(i); | ||
| 98 | const prop = flat_map.get(cp) orelse 0; | ||
| 99 | |||
| 100 | // Process block | ||
| 101 | block[block_len] = 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 | const compressor = std.compress.deflate.compressor; | ||
| 122 | var out_file = try std.fs.cwd().createFile(output_path, .{}); | ||
| 123 | defer out_file.close(); | ||
| 124 | var out_comp = try compressor(allocator, out_file.writer(), .{ .level = .best_compression }); | ||
| 125 | defer out_comp.deinit(); | ||
| 126 | const writer = out_comp.writer(); | ||
| 127 | |||
| 128 | const endian = builtin.cpu.arch.endian(); | ||
| 129 | try writer.writeInt(u16, @intCast(stage1.items.len), endian); | ||
| 130 | for (stage1.items) |i| try writer.writeInt(u16, i, endian); | ||
| 131 | |||
| 132 | try writer.writeInt(u16, @intCast(stage2.items.len), endian); | ||
| 133 | try writer.writeAll(stage2.items); | ||
| 134 | |||
| 135 | try out_comp.flush(); | ||
| 136 | } | ||