From 9a0fb96c0c28540493a205b85d1b89d2c9b50f2b Mon Sep 17 00:00:00 2001 From: Jose Colon Rodriguez Date: Fri, 1 Mar 2024 18:51:43 -0400 Subject: Normalizer.eqlIgnoreCase compatibility caseless matching --- src/FoldData.zig | 48 ++++++++++++++++++++++++++++++++++++++++++++++++ 1 file changed, 48 insertions(+) create mode 100644 src/FoldData.zig (limited to 'src/FoldData.zig') diff --git a/src/FoldData.zig b/src/FoldData.zig new file mode 100644 index 0000000..139c677 --- /dev/null +++ b/src/FoldData.zig @@ -0,0 +1,48 @@ +const std = @import("std"); +const builtin = @import("builtin"); +const compress = std.compress; +const mem = std.mem; + +allocator: mem.Allocator, +fold: [][]u21 = undefined, + +const Self = @This(); + +pub fn init(allocator: mem.Allocator) !Self { + const decompressor = compress.deflate.decompressor; + const in_bytes = @embedFile("fold"); + 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, + .fold = try allocator.alloc([]u21, 0x110000), + }; + + @memset(self.fold, &.{}); + + while (true) { + const len: u8 = try reader.readInt(u8, endian); + if (len == 0) break; + const cp = try reader.readInt(u24, endian); + self.fold[cp] = try allocator.alloc(u21, len - 1); + for (0..len - 1) |i| { + self.fold[cp][i] = @intCast(try reader.readInt(u24, endian)); + } + } + + return self; +} + +pub fn deinit(self: *Self) void { + for (self.fold) |slice| self.allocator.free(slice); + self.allocator.free(self.fold); +} + +/// Returns the case fold for `cp`. +pub inline fn caseFold(self: Self, cp: u21) []const u21 { + return self.fold[cp]; +} -- cgit v1.2.3