diff options
Diffstat (limited to 'src/GeneralCategories.zig')
| -rw-r--r-- | src/GeneralCategories.zig | 59 |
1 files changed, 32 insertions, 27 deletions
diff --git a/src/GeneralCategories.zig b/src/GeneralCategories.zig index a69f7a2..b7c82c0 100644 --- a/src/GeneralCategories.zig +++ b/src/GeneralCategories.zig | |||
| @@ -1,7 +1,8 @@ | |||
| 1 | const std = @import("std"); | 1 | //! General Categories |
| 2 | const builtin = @import("builtin"); | 2 | |
| 3 | const compress = std.compress; | 3 | s1: []u16 = undefined, |
| 4 | const mem = std.mem; | 4 | s2: []u5 = undefined, |
| 5 | s3: []u5 = undefined, | ||
| 5 | 6 | ||
| 6 | /// General Category | 7 | /// General Category |
| 7 | pub const Gc = enum { | 8 | pub const Gc = enum { |
| @@ -37,13 +38,15 @@ pub const Gc = enum { | |||
| 37 | Zs, // Separator, Space | 38 | Zs, // Separator, Space |
| 38 | }; | 39 | }; |
| 39 | 40 | ||
| 40 | s1: []u16 = undefined, | 41 | const GeneralCategories = @This(); |
| 41 | s2: []u5 = undefined, | ||
| 42 | s3: []u5 = undefined, | ||
| 43 | 42 | ||
| 44 | const Self = @This(); | 43 | pub fn init(allocator: Allocator) Allocator.Error!GeneralCategories { |
| 44 | var gencat = GeneralCategories{}; | ||
| 45 | try gencat.setup(allocator); | ||
| 46 | return gencat; | ||
| 47 | } | ||
| 45 | 48 | ||
| 46 | pub fn init(allocator: mem.Allocator) !Self { | 49 | pub fn setup(self: *GeneralCategories, allocator: Allocator) Allocator.Error!void { |
| 47 | const decompressor = compress.flate.inflate.decompressor; | 50 | const decompressor = compress.flate.inflate.decompressor; |
| 48 | const in_bytes = @embedFile("gencat"); | 51 | const in_bytes = @embedFile("gencat"); |
| 49 | var in_fbs = std.io.fixedBufferStream(in_bytes); | 52 | var in_fbs = std.io.fixedBufferStream(in_bytes); |
| @@ -52,39 +55,35 @@ pub fn init(allocator: mem.Allocator) !Self { | |||
| 52 | 55 | ||
| 53 | const endian = builtin.cpu.arch.endian(); | 56 | const endian = builtin.cpu.arch.endian(); |
| 54 | 57 | ||
| 55 | var self = Self{}; | 58 | const s1_len: u16 = reader.readInt(u16, endian) catch unreachable; |
| 56 | |||
| 57 | const s1_len: u16 = try reader.readInt(u16, endian); | ||
| 58 | self.s1 = try allocator.alloc(u16, s1_len); | 59 | self.s1 = try allocator.alloc(u16, s1_len); |
| 59 | errdefer allocator.free(self.s1); | 60 | errdefer allocator.free(self.s1); |
| 60 | for (0..s1_len) |i| self.s1[i] = try reader.readInt(u16, endian); | 61 | for (0..s1_len) |i| self.s1[i] = try reader.readInt(u16, endian); |
| 61 | 62 | ||
| 62 | const s2_len: u16 = try reader.readInt(u16, endian); | 63 | const s2_len: u16 = reader.readInt(u16, endian) catch unreachable; |
| 63 | self.s2 = try allocator.alloc(u5, s2_len); | 64 | self.s2 = try allocator.alloc(u5, s2_len); |
| 64 | errdefer allocator.free(self.s2); | 65 | errdefer allocator.free(self.s2); |
| 65 | for (0..s2_len) |i| self.s2[i] = @intCast(try reader.readInt(u8, endian)); | 66 | for (0..s2_len) |i| self.s2[i] = @intCast(reader.readInt(u8, endian) catch unreachable); |
| 66 | 67 | ||
| 67 | const s3_len: u16 = try reader.readInt(u8, endian); | 68 | const s3_len: u16 = reader.readInt(u8, endian) catch unreachable; |
| 68 | self.s3 = try allocator.alloc(u5, s3_len); | 69 | self.s3 = try allocator.alloc(u5, s3_len); |
| 69 | errdefer allocator.free(self.s3); | 70 | errdefer allocator.free(self.s3); |
| 70 | for (0..s3_len) |i| self.s3[i] = @intCast(try reader.readInt(u8, endian)); | 71 | for (0..s3_len) |i| self.s3[i] = @intCast(reader.readInt(u8, endian) catch unreachable); |
| 71 | |||
| 72 | return self; | ||
| 73 | } | 72 | } |
| 74 | 73 | ||
| 75 | pub fn deinit(self: *const Self, allocator: mem.Allocator) void { | 74 | pub fn deinit(self: *const GeneralCategories, allocator: mem.Allocator) void { |
| 76 | allocator.free(self.s1); | 75 | allocator.free(self.s1); |
| 77 | allocator.free(self.s2); | 76 | allocator.free(self.s2); |
| 78 | allocator.free(self.s3); | 77 | allocator.free(self.s3); |
| 79 | } | 78 | } |
| 80 | 79 | ||
| 81 | /// Lookup the General Category for `cp`. | 80 | /// Lookup the General Category for `cp`. |
| 82 | pub fn gc(self: Self, cp: u21) Gc { | 81 | pub fn gc(self: GeneralCategories, cp: u21) Gc { |
| 83 | return @enumFromInt(self.s3[self.s2[self.s1[cp >> 8] + (cp & 0xff)]]); | 82 | return @enumFromInt(self.s3[self.s2[self.s1[cp >> 8] + (cp & 0xff)]]); |
| 84 | } | 83 | } |
| 85 | 84 | ||
| 86 | /// True if `cp` has an C general category. | 85 | /// True if `cp` has an C general category. |
| 87 | pub fn isControl(self: Self, cp: u21) bool { | 86 | pub fn isControl(self: GeneralCategories, cp: u21) bool { |
| 88 | return switch (self.gc(cp)) { | 87 | return switch (self.gc(cp)) { |
| 89 | .Cc, | 88 | .Cc, |
| 90 | .Cf, | 89 | .Cf, |
| @@ -97,7 +96,7 @@ pub fn isControl(self: Self, cp: u21) bool { | |||
| 97 | } | 96 | } |
| 98 | 97 | ||
| 99 | /// True if `cp` has an L general category. | 98 | /// True if `cp` has an L general category. |
| 100 | pub fn isLetter(self: Self, cp: u21) bool { | 99 | pub fn isLetter(self: GeneralCategories, cp: u21) bool { |
| 101 | return switch (self.gc(cp)) { | 100 | return switch (self.gc(cp)) { |
| 102 | .Ll, | 101 | .Ll, |
| 103 | .Lm, | 102 | .Lm, |
| @@ -110,7 +109,7 @@ pub fn isLetter(self: Self, cp: u21) bool { | |||
| 110 | } | 109 | } |
| 111 | 110 | ||
| 112 | /// True if `cp` has an M general category. | 111 | /// True if `cp` has an M general category. |
| 113 | pub fn isMark(self: Self, cp: u21) bool { | 112 | pub fn isMark(self: GeneralCategories, cp: u21) bool { |
| 114 | return switch (self.gc(cp)) { | 113 | return switch (self.gc(cp)) { |
| 115 | .Mc, | 114 | .Mc, |
| 116 | .Me, | 115 | .Me, |
| @@ -121,7 +120,7 @@ pub fn isMark(self: Self, cp: u21) bool { | |||
| 121 | } | 120 | } |
| 122 | 121 | ||
| 123 | /// True if `cp` has an N general category. | 122 | /// True if `cp` has an N general category. |
| 124 | pub fn isNumber(self: Self, cp: u21) bool { | 123 | pub fn isNumber(self: GeneralCategories, cp: u21) bool { |
| 125 | return switch (self.gc(cp)) { | 124 | return switch (self.gc(cp)) { |
| 126 | .Nd, | 125 | .Nd, |
| 127 | .Nl, | 126 | .Nl, |
| @@ -132,7 +131,7 @@ pub fn isNumber(self: Self, cp: u21) bool { | |||
| 132 | } | 131 | } |
| 133 | 132 | ||
| 134 | /// True if `cp` has an P general category. | 133 | /// True if `cp` has an P general category. |
| 135 | pub fn isPunctuation(self: Self, cp: u21) bool { | 134 | pub fn isPunctuation(self: GeneralCategories, cp: u21) bool { |
| 136 | return switch (self.gc(cp)) { | 135 | return switch (self.gc(cp)) { |
| 137 | .Pc, | 136 | .Pc, |
| 138 | .Pd, | 137 | .Pd, |
| @@ -147,7 +146,7 @@ pub fn isPunctuation(self: Self, cp: u21) bool { | |||
| 147 | } | 146 | } |
| 148 | 147 | ||
| 149 | /// True if `cp` has an S general category. | 148 | /// True if `cp` has an S general category. |
| 150 | pub fn isSymbol(self: Self, cp: u21) bool { | 149 | pub fn isSymbol(self: GeneralCategories, cp: u21) bool { |
| 151 | return switch (self.gc(cp)) { | 150 | return switch (self.gc(cp)) { |
| 152 | .Sc, | 151 | .Sc, |
| 153 | .Sk, | 152 | .Sk, |
| @@ -159,7 +158,7 @@ pub fn isSymbol(self: Self, cp: u21) bool { | |||
| 159 | } | 158 | } |
| 160 | 159 | ||
| 161 | /// True if `cp` has an Z general category. | 160 | /// True if `cp` has an Z general category. |
| 162 | pub fn isSeparator(self: Self, cp: u21) bool { | 161 | pub fn isSeparator(self: GeneralCategories, cp: u21) bool { |
| 163 | return switch (self.gc(cp)) { | 162 | return switch (self.gc(cp)) { |
| 164 | .Zl, | 163 | .Zl, |
| 165 | .Zp, | 164 | .Zp, |
| @@ -168,3 +167,9 @@ pub fn isSeparator(self: Self, cp: u21) bool { | |||
| 168 | else => false, | 167 | else => false, |
| 169 | }; | 168 | }; |
| 170 | } | 169 | } |
| 170 | |||
| 171 | const std = @import("std"); | ||
| 172 | const builtin = @import("builtin"); | ||
| 173 | const compress = std.compress; | ||
| 174 | const mem = std.mem; | ||
| 175 | const Allocator = mem.Allocator; | ||