diff options
| author | 2026-02-04 18:01:36 -0500 | |
|---|---|---|
| committer | 2026-02-04 18:01:36 -0500 | |
| commit | ba5d9081b479e95ffa7f3baf751beedd370cec14 (patch) | |
| tree | c12041d8aab9f9ff68b25a2e2c9042073c3d5f61 /src/NormPropsData.zig | |
| parent | Convert Words module to no-allocation (diff) | |
| download | zg-ba5d9081b479e95ffa7f3baf751beedd370cec14.tar.gz zg-ba5d9081b479e95ffa7f3baf751beedd370cec14.tar.xz zg-ba5d9081b479e95ffa7f3baf751beedd370cec14.zip | |
Normalization and case folding
Both of which deserve some further attention.
Diffstat (limited to 'src/NormPropsData.zig')
| -rw-r--r-- | src/NormPropsData.zig | 46 |
1 files changed, 15 insertions, 31 deletions
diff --git a/src/NormPropsData.zig b/src/NormPropsData.zig index 7b53542..cca3556 100644 --- a/src/NormPropsData.zig +++ b/src/NormPropsData.zig | |||
| @@ -1,48 +1,32 @@ | |||
| 1 | //! Normalization Properties Data | 1 | //! Normalization Properties Data |
| 2 | 2 | ||
| 3 | s1: []u16 = undefined, | 3 | const Data = struct { |
| 4 | s2: []u4 = undefined, | 4 | s1: []const u16 = undefined, |
| 5 | s2: []const u3 = undefined, | ||
| 6 | }; | ||
| 7 | |||
| 8 | const norms = norm_props_data: { | ||
| 9 | const data = @import("normp"); | ||
| 10 | break :norm_props_data Data{ | ||
| 11 | .s1 = &data.s1, | ||
| 12 | .s2 = &data.s2, | ||
| 13 | }; | ||
| 14 | }; | ||
| 5 | 15 | ||
| 6 | const NormProps = @This(); | 16 | const NormProps = @This(); |
| 7 | 17 | ||
| 8 | pub fn init(allocator: mem.Allocator) !NormProps { | ||
| 9 | const in_bytes = @embedFile("normp"); | ||
| 10 | var in_fbs = std.io.fixedBufferStream(in_bytes); | ||
| 11 | var reader = in_fbs.reader(); | ||
| 12 | |||
| 13 | const endian = builtin.cpu.arch.endian(); | ||
| 14 | var norms = NormProps{}; | ||
| 15 | |||
| 16 | const stage_1_len: u16 = try reader.readInt(u16, endian); | ||
| 17 | norms.s1 = try allocator.alloc(u16, stage_1_len); | ||
| 18 | errdefer allocator.free(norms.s1); | ||
| 19 | for (0..stage_1_len) |i| norms.s1[i] = try reader.readInt(u16, endian); | ||
| 20 | |||
| 21 | const stage_2_len: u16 = try reader.readInt(u16, endian); | ||
| 22 | norms.s2 = try allocator.alloc(u4, stage_2_len); | ||
| 23 | errdefer allocator.free(norms.s2); | ||
| 24 | for (0..stage_2_len) |i| norms.s2[i] = @intCast(try reader.readInt(u8, endian)); | ||
| 25 | |||
| 26 | return norms; | ||
| 27 | } | ||
| 28 | |||
| 29 | pub fn deinit(norms: *const NormProps, allocator: mem.Allocator) void { | ||
| 30 | allocator.free(norms.s1); | ||
| 31 | allocator.free(norms.s2); | ||
| 32 | } | ||
| 33 | |||
| 34 | /// Returns true if `cp` is already in NFD form. | 18 | /// Returns true if `cp` is already in NFD form. |
| 35 | pub fn isNfd(norms: *const NormProps, cp: u21) bool { | 19 | pub fn isNfd(cp: u21) bool { |
| 36 | return norms.s2[norms.s1[cp >> 8] + (cp & 0xff)] & 1 == 0; | 20 | return norms.s2[norms.s1[cp >> 8] + (cp & 0xff)] & 1 == 0; |
| 37 | } | 21 | } |
| 38 | 22 | ||
| 39 | /// Returns true if `cp` is already in NFKD form. | 23 | /// Returns true if `cp` is already in NFKD form. |
| 40 | pub fn isNfkd(norms: *const NormProps, cp: u21) bool { | 24 | pub fn isNfkd(cp: u21) bool { |
| 41 | return norms.s2[norms.s1[cp >> 8] + (cp & 0xff)] & 2 == 0; | 25 | return norms.s2[norms.s1[cp >> 8] + (cp & 0xff)] & 2 == 0; |
| 42 | } | 26 | } |
| 43 | 27 | ||
| 44 | /// Returns true if `cp` is not allowed in any normalized form. | 28 | /// Returns true if `cp` is not allowed in any normalized form. |
| 45 | pub fn isFcx(norms: *const NormProps, cp: u21) bool { | 29 | pub fn isFcx(cp: u21) bool { |
| 46 | return norms.s2[norms.s1[cp >> 8] + (cp & 0xff)] & 4 == 4; | 30 | return norms.s2[norms.s1[cp >> 8] + (cp & 0xff)] & 4 == 4; |
| 47 | } | 31 | } |
| 48 | 32 | ||