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/CombiningData.zig | 48 ++++++++++++++++++++++++++++++++++++++++++++++++
 1 file changed, 48 insertions(+)
 create mode 100644 src/CombiningData.zig

(limited to 'src/CombiningData.zig')

diff --git a/src/CombiningData.zig b/src/CombiningData.zig
new file mode 100644
index 0000000..95c947d
--- /dev/null
+++ b/src/CombiningData.zig
@@ -0,0 +1,48 @@
+const std = @import("std");
+const builtin = @import("builtin");
+const compress = std.compress;
+const mem = std.mem;
+
+allocator: mem.Allocator,
+s1: []u16 = undefined,
+s2: []u8 = undefined,
+
+const Self = @This();
+
+pub fn init(allocator: mem.Allocator) !Self {
+    const decompressor = compress.deflate.decompressor;
+    const in_bytes = @embedFile("ccc");
+    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 };
+
+    const stage_1_len: u16 = try reader.readInt(u16, endian);
+    self.s1 = try allocator.alloc(u16, stage_1_len);
+    for (0..stage_1_len) |i| self.s1[i] = try reader.readInt(u16, endian);
+
+    const stage_2_len: u16 = try reader.readInt(u16, endian);
+    self.s2 = try allocator.alloc(u8, stage_2_len);
+    _ = try reader.readAll(self.s2);
+
+    return self;
+}
+
+pub fn deinit(self: *Self) void {
+    self.allocator.free(self.s1);
+    self.allocator.free(self.s2);
+}
+
+/// Returns the canonical combining class for a code point.
+pub inline fn ccc(self: Self, cp: u21) u8 {
+    return self.s2[self.s1[cp >> 8] + (cp & 0xff)];
+}
+
+/// True if `cp` is a starter code point, not a combining character.
+pub inline fn isStarter(self: Self, cp: u21) bool {
+    return self.s2[self.s1[cp >> 8] + (cp & 0xff)] == 0;
+}
-- 
cgit v1.2.3