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/CanonData.zig | 56 +++++++++++++++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 56 insertions(+) create mode 100644 src/CanonData.zig (limited to 'src/CanonData.zig') diff --git a/src/CanonData.zig b/src/CanonData.zig new file mode 100644 index 0000000..81d3eec --- /dev/null +++ b/src/CanonData.zig @@ -0,0 +1,56 @@ +const std = @import("std"); +const builtin = @import("builtin"); +const compress = std.compress; +const mem = std.mem; + +allocator: mem.Allocator, +nfc: std.AutoHashMap([2]u21, u21), +nfd: [][2]u21 = undefined, + +const Self = @This(); + +pub fn init(allocator: mem.Allocator) !Self { + const decompressor = compress.deflate.decompressor; + const in_bytes = @embedFile("canon"); + 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, + .nfc = std.AutoHashMap([2]u21, u21).init(allocator), + .nfd = try allocator.alloc([2]u21, 0x110000), + }; + + for (0..0x110000) |i| self.nfd[i] = .{ @intCast(i), 0 }; + + while (true) { + const len: u8 = try reader.readInt(u8, endian); + if (len == 0) break; + const cp = try reader.readInt(u24, endian); + self.nfd[cp][0] = @intCast(try reader.readInt(u24, endian)); + if (len == 3) { + self.nfd[cp][1] = @intCast(try reader.readInt(u24, endian)); + try self.nfc.put(self.nfd[cp], @intCast(cp)); + } + } + + return self; +} + +pub fn deinit(self: *Self) void { + self.nfc.deinit(); + self.allocator.free(self.nfd); +} + +/// Returns canonical decomposition for `cp`. +pub inline fn toNfd(self: Self, cp: u21) [2]u21 { + return self.nfd[cp]; +} + +// Returns the primary composite for the codepoints in `cp`. +pub inline fn toNfc(self: Self, cps: [2]u21) ?u21 { + return self.nfc.get(cps); +} -- cgit v1.2.3