From ecd9c2277de17e24fa26aefee955caa10b5b990c Mon Sep 17 00:00:00 2001 From: Jose Colon Rodriguez Date: Tue, 27 Feb 2024 18:04:32 -0400 Subject: General Category with GenCatData --- src/main.zig | 33 +++++++++++++++++++++------------ 1 file changed, 21 insertions(+), 12 deletions(-) (limited to 'src/main.zig') diff --git a/src/main.zig b/src/main.zig index 0f1aab5..c521c4f 100644 --- a/src/main.zig +++ b/src/main.zig @@ -11,14 +11,16 @@ 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 Normalizer = @import("ziglyph").Normalizer; -const NormData = @import("Normalizer").NormData; -const Normalizer = @import("Normalizer"); +// const NormData = @import("Normalizer").NormData; +// const Normalizer = @import("Normalizer"); + +const GenCatData = @import("GenCatData"); pub fn main() !void { var args_iter = std.process.args(); @@ -32,16 +34,19 @@ pub fn main() !void { const input = try std.fs.cwd().readFileAlloc(allocator, in_path, std.math.maxInt(u32)); defer allocator.free(input); - var data = try NormData.init(allocator); - defer data.deinit(); - var n = Normalizer{ .norm_data = &data }; + // var data = try NormData.init(allocator); + // defer data.deinit(); + // var n = Normalizer{ .norm_data = &data }; // var n = try Normalizer.init(allocator); // defer n.deinit(); + var gencat_data = try GenCatData.init(allocator); + defer gencat_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 result: isize = 0; @@ -50,10 +55,14 @@ 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 n.nfc(allocator, line); - result += nfc.slice.len; - // nfc.deinit(); + // while (iter.next()) |line| { + // const nfc = try n.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; } std.debug.print("result: {}, took: {}\n", .{ result, timer.lap() / std.time.ns_per_ms }); -- cgit v1.2.3