From 789625d2ce4b6e74f2372f87c20304639fa343ef Mon Sep 17 00:00:00 2001 From: Jose Colon Rodriguez Date: Sun, 24 Mar 2024 20:03:45 -0400 Subject: NumericData --- src/NumericData.zig | 75 +++++++++++++++++++++++++++++++++++++++++++++++++++++ src/main.zig | 32 ++++++++++++++--------- 2 files changed, 95 insertions(+), 12 deletions(-) create mode 100644 src/NumericData.zig (limited to 'src') 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 @@ +const std = @import("std"); +const builtin = @import("builtin"); +const compress = std.compress; +const mem = std.mem; +const testing = std.testing; + +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("numeric"); + 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); +} + +/// True if `cp` is any numeric type. +pub fn isNumber(self: Self, cp: u21) bool { + return self.isNumeric(cp) or self.isDigit(cp) or self.isDecimal(cp); +} + +/// True if `cp` is numeric. +pub inline fn isNumeric(self: Self, cp: u21) bool { + return self.s2[self.s1[cp >> 8] + (cp & 0xff)] & 1 == 1; +} + +/// True if `cp` is a digit. +pub inline fn isDigit(self: Self, cp: u21) bool { + return self.s2[self.s1[cp >> 8] + (cp & 0xff)] & 2 == 2; +} + +/// True if `cp` is decimal. +pub inline fn isDecimal(self: Self, cp: u21) bool { + return self.s2[self.s1[cp >> 8] + (cp & 0xff)] & 4 == 4; +} + +test "isDecimal" { + var self = try init(testing.allocator); + defer self.deinit(); + + try testing.expect(self.isNumber('\u{277f}')); + try testing.expect(self.isNumber('3')); + try testing.expect(self.isNumeric('\u{277f}')); + try testing.expect(self.isDigit('\u{2070}')); + try testing.expect(self.isDecimal('3')); + + try testing.expect(!self.isNumber('z')); + try testing.expect(!self.isNumeric('1')); + try testing.expect(!self.isDigit('2')); + try testing.expect(!self.isDecimal('g')); +} 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"); // const strWidth = @import("display_width").strWidth; // const CodePointIterator = @import("ziglyph").CodePointIterator; -// const CodePointIterator = @import("code_point").Iterator; +const CodePointIterator = @import("code_point").Iterator; // const ascii = @import("ascii"); // const ascii = std.ascii; // const Normalize = @import("ziglyph").Normalizer; -const Normalize = @import("Normalize"); +// const Normalize = @import("Normalize"); // const CaseFold = @import("CaseFold"); // const GenCatData = @import("GenCatData"); +const NumericData = @import("NumericData"); + pub fn main() !void { var args_iter = std.process.args(); _ = args_iter.skip(); @@ -39,9 +41,9 @@ pub fn main() !void { ); defer allocator.free(input); - var norm_data = try Normalize.NormData.init(allocator); - defer norm_data.deinit(); - var norm = Normalize{ .norm_data = &norm_data }; + // var norm_data = try Normalize.NormData.init(allocator); + // defer norm_data.deinit(); + // var norm = Normalize{ .norm_data = &norm_data }; // var norm = try Normalize.init(allocator); // defer norm.deinit(); @@ -52,10 +54,13 @@ pub fn main() !void { // defer fold_data.deinit(); // var caser = CaseFold{ .fold_data = &fold_data }; + var num_data = try NumericData.init(allocator); + defer num_data.deinit(); + // var iter = GraphemeIterator.init(input, &data); // defer iter.deinit(); - // var iter = CodePointIterator{ .bytes = input }; - var iter = std.mem.splitScalar(u8, input, '\n'); + var iter = CodePointIterator{ .bytes = input }; + // var iter = std.mem.splitScalar(u8, input, '\n'); var result: usize = 0; // var prev_line: []const u8 = ""; @@ -65,11 +70,11 @@ pub fn main() !void { // while (iter.next()) |cp| result += codePointWidth(@intCast(cp.code)); // while (iter.next()) |_| result += 1; // while (iter.next()) |line| result += strWidth(line, &data); - while (iter.next()) |line| { - const nfc = try norm.nfc(allocator, line); - result += nfc.slice.len; - // nfc.deinit(); - } + // while (iter.next()) |line| { + // const nfc = try norm.nfc(allocator, line); + // result += nfc.slice.len; + // // nfc.deinit(); + // } // while (iter.next()) |cp| { // if (cp.code == 'É') std.debug.print("`{u}` Gc: {s}\n", .{ cp.code, @tagName(gencat_data.gc(cp.code)) }); // result += 1; @@ -80,6 +85,9 @@ pub fn main() !void { // } // prev_line = line; // } + while (iter.next()) |cp| { + if (num_data.isNumberic(cp)) result += 1; + } std.debug.print("result: {}, took: {}\n", .{ result, timer.lap() / std.time.ns_per_ms }); } -- cgit v1.2.3