From ad306724ae574b1a22abbcb6de37e65a69db82e4 Mon Sep 17 00:00:00 2001 From: Jose Colon Rodriguez Date: Mon, 26 Feb 2024 20:22:41 -0400 Subject: Using NormData nfkd --- src/CompatData.zig | 50 ++++++++++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 50 insertions(+) create mode 100644 src/CompatData.zig (limited to 'src/CompatData.zig') diff --git a/src/CompatData.zig b/src/CompatData.zig new file mode 100644 index 0000000..a1f5de6 --- /dev/null +++ b/src/CompatData.zig @@ -0,0 +1,50 @@ +const std = @import("std"); +const builtin = @import("builtin"); +const compress = std.compress; +const mem = std.mem; + +allocator: mem.Allocator, +nfkd: [][]u21 = undefined, + +const Self = @This(); + +pub fn init(allocator: mem.Allocator) !Self { + const decompressor = compress.deflate.decompressor; + const in_bytes = @embedFile("compat"); + var in_fbs = std.io.fixedBufferStream(in_bytes); + var in_decomp = try decompressor(allocator, in_fbs.reader(), null); + defer in_decomp.deinit(); + var reader = in_decomp.reader(); + + const endian = builtin.cpu.arch.endian(); + var self = Self{ + .allocator = allocator, + .nfkd = try allocator.alloc([]u21, 0x110000), + }; + + for (0..0x110000) |i| self.nfkd[i] = &.{}; + + while (true) { + const len: u8 = try reader.readInt(u8, endian); + if (len == 0) break; + const cp = try reader.readInt(u24, endian); + self.nfkd[cp] = try allocator.alloc(u21, len - 1); + for (0..len - 1) |i| { + self.nfkd[cp][i] = @intCast(try reader.readInt(u24, endian)); + } + } + + return self; +} + +pub fn deinit(self: *Self) void { + for (self.nfkd) |slice| { + if (slice.len != 0) self.allocator.free(slice); + } + self.allocator.free(self.nfkd); +} + +/// Returns compatibility decomposition for `cp`. +pub inline fn toNfkd(self: Self, cp: u21) []u21 { + return self.nfkd[cp]; +} -- cgit v1.2.3