diff options
Diffstat (limited to 'src/CanonData.zig')
| -rw-r--r-- | src/CanonData.zig | 56 |
1 files changed, 56 insertions, 0 deletions
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 @@ | |||
| 1 | const std = @import("std"); | ||
| 2 | const builtin = @import("builtin"); | ||
| 3 | const compress = std.compress; | ||
| 4 | const mem = std.mem; | ||
| 5 | |||
| 6 | allocator: mem.Allocator, | ||
| 7 | nfc: std.AutoHashMap([2]u21, u21), | ||
| 8 | nfd: [][2]u21 = undefined, | ||
| 9 | |||
| 10 | const Self = @This(); | ||
| 11 | |||
| 12 | pub fn init(allocator: mem.Allocator) !Self { | ||
| 13 | const decompressor = compress.deflate.decompressor; | ||
| 14 | const in_bytes = @embedFile("canon"); | ||
| 15 | var in_fbs = std.io.fixedBufferStream(in_bytes); | ||
| 16 | var in_decomp = try decompressor(allocator, in_fbs.reader(), null); | ||
| 17 | defer in_decomp.deinit(); | ||
| 18 | var reader = in_decomp.reader(); | ||
| 19 | |||
| 20 | const endian = builtin.cpu.arch.endian(); | ||
| 21 | var self = Self{ | ||
| 22 | .allocator = allocator, | ||
| 23 | .nfc = std.AutoHashMap([2]u21, u21).init(allocator), | ||
| 24 | .nfd = try allocator.alloc([2]u21, 0x110000), | ||
| 25 | }; | ||
| 26 | |||
| 27 | for (0..0x110000) |i| self.nfd[i] = .{ @intCast(i), 0 }; | ||
| 28 | |||
| 29 | while (true) { | ||
| 30 | const len: u8 = try reader.readInt(u8, endian); | ||
| 31 | if (len == 0) break; | ||
| 32 | const cp = try reader.readInt(u24, endian); | ||
| 33 | self.nfd[cp][0] = @intCast(try reader.readInt(u24, endian)); | ||
| 34 | if (len == 3) { | ||
| 35 | self.nfd[cp][1] = @intCast(try reader.readInt(u24, endian)); | ||
| 36 | try self.nfc.put(self.nfd[cp], @intCast(cp)); | ||
| 37 | } | ||
| 38 | } | ||
| 39 | |||
| 40 | return self; | ||
| 41 | } | ||
| 42 | |||
| 43 | pub fn deinit(self: *Self) void { | ||
| 44 | self.nfc.deinit(); | ||
| 45 | self.allocator.free(self.nfd); | ||
| 46 | } | ||
| 47 | |||
| 48 | /// Returns canonical decomposition for `cp`. | ||
| 49 | pub inline fn toNfd(self: Self, cp: u21) [2]u21 { | ||
| 50 | return self.nfd[cp]; | ||
| 51 | } | ||
| 52 | |||
| 53 | // Returns the primary composite for the codepoints in `cp`. | ||
| 54 | pub inline fn toNfc(self: Self, cps: [2]u21) ?u21 { | ||
| 55 | return self.nfc.get(cps); | ||
| 56 | } | ||