diff options
Diffstat (limited to 'src')
| -rw-r--r-- | src/NumericData.zig | 75 | ||||
| -rw-r--r-- | src/main.zig | 32 |
2 files changed, 95 insertions, 12 deletions
diff --git a/src/NumericData.zig b/src/NumericData.zig new file mode 100644 index 0000000..baf8f11 --- /dev/null +++ b/src/NumericData.zig | |||
| @@ -0,0 +1,75 @@ | |||
| 1 | const std = @import("std"); | ||
| 2 | const builtin = @import("builtin"); | ||
| 3 | const compress = std.compress; | ||
| 4 | const mem = std.mem; | ||
| 5 | const testing = std.testing; | ||
| 6 | |||
| 7 | allocator: mem.Allocator, | ||
| 8 | s1: []u16 = undefined, | ||
| 9 | s2: []u8 = undefined, | ||
| 10 | |||
| 11 | const Self = @This(); | ||
| 12 | |||
| 13 | pub fn init(allocator: mem.Allocator) !Self { | ||
| 14 | const decompressor = compress.deflate.decompressor; | ||
| 15 | const in_bytes = @embedFile("numeric"); | ||
| 16 | var in_fbs = std.io.fixedBufferStream(in_bytes); | ||
| 17 | var in_decomp = try decompressor(allocator, in_fbs.reader(), null); | ||
| 18 | defer in_decomp.deinit(); | ||
| 19 | var reader = in_decomp.reader(); | ||
| 20 | |||
| 21 | const endian = builtin.cpu.arch.endian(); | ||
| 22 | |||
| 23 | var self = Self{ .allocator = allocator }; | ||
| 24 | |||
| 25 | const stage_1_len: u16 = try reader.readInt(u16, endian); | ||
| 26 | self.s1 = try allocator.alloc(u16, stage_1_len); | ||
| 27 | for (0..stage_1_len) |i| self.s1[i] = try reader.readInt(u16, endian); | ||
| 28 | |||
| 29 | const stage_2_len: u16 = try reader.readInt(u16, endian); | ||
| 30 | self.s2 = try allocator.alloc(u8, stage_2_len); | ||
| 31 | _ = try reader.readAll(self.s2); | ||
| 32 | |||
| 33 | return self; | ||
| 34 | } | ||
| 35 | |||
| 36 | pub fn deinit(self: *Self) void { | ||
| 37 | self.allocator.free(self.s1); | ||
| 38 | self.allocator.free(self.s2); | ||
| 39 | } | ||
| 40 | |||
| 41 | /// True if `cp` is any numeric type. | ||
| 42 | pub fn isNumber(self: Self, cp: u21) bool { | ||
| 43 | return self.isNumeric(cp) or self.isDigit(cp) or self.isDecimal(cp); | ||
| 44 | } | ||
| 45 | |||
| 46 | /// True if `cp` is numeric. | ||
| 47 | pub inline fn isNumeric(self: Self, cp: u21) bool { | ||
| 48 | return self.s2[self.s1[cp >> 8] + (cp & 0xff)] & 1 == 1; | ||
| 49 | } | ||
| 50 | |||
| 51 | /// True if `cp` is a digit. | ||
| 52 | pub inline fn isDigit(self: Self, cp: u21) bool { | ||
| 53 | return self.s2[self.s1[cp >> 8] + (cp & 0xff)] & 2 == 2; | ||
| 54 | } | ||
| 55 | |||
| 56 | /// True if `cp` is decimal. | ||
| 57 | pub inline fn isDecimal(self: Self, cp: u21) bool { | ||
| 58 | return self.s2[self.s1[cp >> 8] + (cp & 0xff)] & 4 == 4; | ||
| 59 | } | ||
| 60 | |||
| 61 | test "isDecimal" { | ||
| 62 | var self = try init(testing.allocator); | ||
| 63 | defer self.deinit(); | ||
| 64 | |||
| 65 | try testing.expect(self.isNumber('\u{277f}')); | ||
| 66 | try testing.expect(self.isNumber('3')); | ||
| 67 | try testing.expect(self.isNumeric('\u{277f}')); | ||
| 68 | try testing.expect(self.isDigit('\u{2070}')); | ||
| 69 | try testing.expect(self.isDecimal('3')); | ||
| 70 | |||
| 71 | try testing.expect(!self.isNumber('z')); | ||
| 72 | try testing.expect(!self.isNumeric('1')); | ||
| 73 | try testing.expect(!self.isDigit('2')); | ||
| 74 | try testing.expect(!self.isDecimal('g')); | ||
| 75 | } | ||
diff --git a/src/main.zig b/src/main.zig index 0b0d550..52d823c 100644 --- a/src/main.zig +++ b/src/main.zig | |||
| @@ -11,18 +11,20 @@ const std = @import("std"); | |||
| 11 | // const strWidth = @import("display_width").strWidth; | 11 | // const strWidth = @import("display_width").strWidth; |
| 12 | 12 | ||
| 13 | // const CodePointIterator = @import("ziglyph").CodePointIterator; | 13 | // const CodePointIterator = @import("ziglyph").CodePointIterator; |
| 14 | // const CodePointIterator = @import("code_point").Iterator; | 14 | const CodePointIterator = @import("code_point").Iterator; |
| 15 | 15 | ||
| 16 | // const ascii = @import("ascii"); | 16 | // const ascii = @import("ascii"); |
| 17 | // const ascii = std.ascii; | 17 | // const ascii = std.ascii; |
| 18 | 18 | ||
| 19 | // const Normalize = @import("ziglyph").Normalizer; | 19 | // const Normalize = @import("ziglyph").Normalizer; |
| 20 | const Normalize = @import("Normalize"); | 20 | // const Normalize = @import("Normalize"); |
| 21 | 21 | ||
| 22 | // const CaseFold = @import("CaseFold"); | 22 | // const CaseFold = @import("CaseFold"); |
| 23 | 23 | ||
| 24 | // const GenCatData = @import("GenCatData"); | 24 | // const GenCatData = @import("GenCatData"); |
| 25 | 25 | ||
| 26 | const NumericData = @import("NumericData"); | ||
| 27 | |||
| 26 | pub fn main() !void { | 28 | pub fn main() !void { |
| 27 | var args_iter = std.process.args(); | 29 | var args_iter = std.process.args(); |
| 28 | _ = args_iter.skip(); | 30 | _ = args_iter.skip(); |
| @@ -39,9 +41,9 @@ pub fn main() !void { | |||
| 39 | ); | 41 | ); |
| 40 | defer allocator.free(input); | 42 | defer allocator.free(input); |
| 41 | 43 | ||
| 42 | var norm_data = try Normalize.NormData.init(allocator); | 44 | // var norm_data = try Normalize.NormData.init(allocator); |
| 43 | defer norm_data.deinit(); | 45 | // defer norm_data.deinit(); |
| 44 | var norm = Normalize{ .norm_data = &norm_data }; | 46 | // var norm = Normalize{ .norm_data = &norm_data }; |
| 45 | // var norm = try Normalize.init(allocator); | 47 | // var norm = try Normalize.init(allocator); |
| 46 | // defer norm.deinit(); | 48 | // defer norm.deinit(); |
| 47 | 49 | ||
| @@ -52,10 +54,13 @@ pub fn main() !void { | |||
| 52 | // defer fold_data.deinit(); | 54 | // defer fold_data.deinit(); |
| 53 | // var caser = CaseFold{ .fold_data = &fold_data }; | 55 | // var caser = CaseFold{ .fold_data = &fold_data }; |
| 54 | 56 | ||
| 57 | var num_data = try NumericData.init(allocator); | ||
| 58 | defer num_data.deinit(); | ||
| 59 | |||
| 55 | // var iter = GraphemeIterator.init(input, &data); | 60 | // var iter = GraphemeIterator.init(input, &data); |
| 56 | // defer iter.deinit(); | 61 | // defer iter.deinit(); |
| 57 | // var iter = CodePointIterator{ .bytes = input }; | 62 | var iter = CodePointIterator{ .bytes = input }; |
| 58 | var iter = std.mem.splitScalar(u8, input, '\n'); | 63 | // var iter = std.mem.splitScalar(u8, input, '\n'); |
| 59 | 64 | ||
| 60 | var result: usize = 0; | 65 | var result: usize = 0; |
| 61 | // var prev_line: []const u8 = ""; | 66 | // var prev_line: []const u8 = ""; |
| @@ -65,11 +70,11 @@ pub fn main() !void { | |||
| 65 | // while (iter.next()) |cp| result += codePointWidth(@intCast(cp.code)); | 70 | // while (iter.next()) |cp| result += codePointWidth(@intCast(cp.code)); |
| 66 | // while (iter.next()) |_| result += 1; | 71 | // while (iter.next()) |_| result += 1; |
| 67 | // while (iter.next()) |line| result += strWidth(line, &data); | 72 | // while (iter.next()) |line| result += strWidth(line, &data); |
| 68 | while (iter.next()) |line| { | 73 | // while (iter.next()) |line| { |
| 69 | const nfc = try norm.nfc(allocator, line); | 74 | // const nfc = try norm.nfc(allocator, line); |
| 70 | result += nfc.slice.len; | 75 | // result += nfc.slice.len; |
| 71 | // nfc.deinit(); | 76 | // // nfc.deinit(); |
| 72 | } | 77 | // } |
| 73 | // while (iter.next()) |cp| { | 78 | // while (iter.next()) |cp| { |
| 74 | // if (cp.code == 'É') std.debug.print("`{u}` Gc: {s}\n", .{ cp.code, @tagName(gencat_data.gc(cp.code)) }); | 79 | // if (cp.code == 'É') std.debug.print("`{u}` Gc: {s}\n", .{ cp.code, @tagName(gencat_data.gc(cp.code)) }); |
| 75 | // result += 1; | 80 | // result += 1; |
| @@ -80,6 +85,9 @@ pub fn main() !void { | |||
| 80 | // } | 85 | // } |
| 81 | // prev_line = line; | 86 | // prev_line = line; |
| 82 | // } | 87 | // } |
| 88 | while (iter.next()) |cp| { | ||
| 89 | if (num_data.isNumberic(cp)) result += 1; | ||
| 90 | } | ||
| 83 | 91 | ||
| 84 | std.debug.print("result: {}, took: {}\n", .{ result, timer.lap() / std.time.ns_per_ms }); | 92 | std.debug.print("result: {}, took: {}\n", .{ result, timer.lap() / std.time.ns_per_ms }); |
| 85 | } | 93 | } |