summaryrefslogtreecommitdiff
path: root/codegen/emoji.zig
diff options
context:
space:
mode:
authorGravatar Jose Colon Rodriguez2024-02-14 12:02:02 -0400
committerGravatar Jose Colon Rodriguez2024-02-14 12:02:02 -0400
commit703a824c1bb7fc41535c9515c5a2209d04899d19 (patch)
tree0572d99ce590a644bf54444a41e20a9688fecb19 /codegen/emoji.zig
parentRemoved unreachables from CodePointIterator (diff)
downloadzg-703a824c1bb7fc41535c9515c5a2209d04899d19.tar.gz
zg-703a824c1bb7fc41535c9515c5a2209d04899d19.tar.xz
zg-703a824c1bb7fc41535c9515c5a2209d04899d19.zip
Code reorg; Added UCD
Diffstat (limited to 'codegen/emoji.zig')
-rw-r--r--codegen/emoji.zig93
1 files changed, 93 insertions, 0 deletions
diff --git a/codegen/emoji.zig b/codegen/emoji.zig
new file mode 100644
index 0000000..acad0ca
--- /dev/null
+++ b/codegen/emoji.zig
@@ -0,0 +1,93 @@
1const std = @import("std");
2
3const emoji = @import("ziglyph").emoji;
4
5const block_size = 256;
6const Block = [block_size]bool;
7
8const 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 std.mem.eql(bool, &a, &b);
20 }
21 },
22 std.hash_map.default_max_load_percentage,
23);
24
25pub 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 blocks_map = BlockMap.init(allocator);
31 defer blocks_map.deinit();
32
33 var stage1 = std.ArrayList(u16).init(allocator);
34 defer stage1.deinit();
35
36 var stage2 = std.ArrayList(bool).init(allocator);
37 defer stage2.deinit();
38
39 var block: Block = [_]bool{false} ** block_size;
40 var block_len: u16 = 0;
41
42 for (0..0x10ffff + 1) |cp| {
43 const isEmoji = emoji.isExtendedPictographic(@intCast(cp));
44
45 block[block_len] = isEmoji;
46 block_len += 1;
47
48 if (block_len < block_size and cp != 0x10ffff) continue;
49
50 const gop = try blocks_map.getOrPut(block);
51 if (!gop.found_existing) {
52 gop.value_ptr.* = @intCast(stage2.items.len);
53 try stage2.appendSlice(&block);
54 }
55
56 try stage1.append(gop.value_ptr.*);
57 block_len = 0;
58 }
59
60 var args_iter = std.process.args();
61 _ = args_iter.skip();
62 const output_path = args_iter.next() orelse @panic("No output file arg!");
63
64 var out_file = try std.fs.cwd().createFile(output_path, .{});
65 defer out_file.close();
66 var out_buf = std.io.bufferedWriter(out_file.writer());
67 const writer = out_buf.writer();
68
69 try writer.print("const stage_1 = [{}]u16{{", .{stage1.items.len});
70 for (stage1.items) |v| {
71 _ = try writer.print("{},", .{v});
72 }
73 try writer.writeAll("};\n");
74
75 try writer.print("const stage_2 = [{}]bool{{", .{stage2.items.len});
76 for (stage2.items) |v| {
77 _ = try writer.print("{},", .{v});
78 }
79 try writer.writeAll("};\n");
80
81 const code =
82 \\pub inline fn isExtendedPictographic(cp: u21) bool {
83 \\ const stage_1_index = cp >> 8;
84 \\ const stage_2_index = stage_1[stage_1_index] + (cp & 0xff);
85 \\ return stage_2[stage_2_index];
86 \\}
87 \\
88 ;
89
90 try writer.writeAll(code);
91
92 try out_buf.flush();
93}