From 9b435e69cb9f1572728b38457fabc9636bc47143 Mon Sep 17 00:00:00 2001 From: Jose Colon Rodriguez Date: Fri, 1 Mar 2024 19:28:41 -0400 Subject: Changes when case folded check; 20ms faster --- codegen/fold.zig | 54 +++++++++++++++++++++++++++++++++++++++++++++++++++--- 1 file changed, 51 insertions(+), 3 deletions(-) (limited to 'codegen') diff --git a/codegen/fold.zig b/codegen/fold.zig index 7977e61..b3192e7 100644 --- a/codegen/fold.zig +++ b/codegen/fold.zig @@ -8,7 +8,51 @@ pub fn main() !void { defer arena.deinit(); const allocator = arena.allocator(); - // Process DerivedEastAsianWidth.txt + // Process DerivedCoreProperties.txt + var cp_file = try std.fs.cwd().openFile("data/unicode/DerivedCoreProperties.txt", .{}); + defer cp_file.close(); + var cp_buf = std.io.bufferedReader(cp_file.reader()); + const cp_reader = cp_buf.reader(); + + var cp_map = std.AutoHashMap(u21, void).init(allocator); + defer cp_map.deinit(); + + var line_buf: [4096]u8 = undefined; + + cp_lines: while (try cp_reader.readUntilDelimiterOrEof(&line_buf, '\n')) |line| { + if (line.len == 0 or line[0] == '#') continue; + + const no_comment = if (std.mem.indexOfScalar(u8, line, '#')) |octo| line[0..octo] else line; + + var field_iter = std.mem.tokenizeAny(u8, no_comment, "; "); + var current_code: [2]u21 = undefined; + + var i: usize = 0; + while (field_iter.next()) |field| : (i += 1) { + switch (i) { + 0 => { + // Code point(s) + if (std.mem.indexOf(u8, field, "..")) |dots| { + current_code = .{ + try std.fmt.parseInt(u21, field[0..dots], 16), + try std.fmt.parseInt(u21, field[dots + 2 ..], 16), + }; + } else { + const code = try std.fmt.parseInt(u21, field, 16); + current_code = .{ code, code }; + } + }, + 1 => { + // Core property + if (!mem.eql(u8, field, "Changes_When_Casefolded")) continue :cp_lines; + for (current_code[0]..current_code[1] + 1) |cp| try cp_map.put(@intCast(cp), {}); + }, + else => {}, + } + } + } + + // Process CaseFolding.txt var in_file = try std.fs.cwd().openFile("data/unicode/CaseFolding.txt", .{}); defer in_file.close(); var in_buf = std.io.bufferedReader(in_file.reader()); @@ -27,7 +71,6 @@ pub fn main() !void { const writer = out_comp.writer(); const endian = builtin.cpu.arch.endian(); - var line_buf: [4096]u8 = undefined; lines: while (try in_reader.readUntilDelimiterOrEof(&line_buf, '\n')) |line| { if (line.len == 0 or line[0] == '#') continue; @@ -41,7 +84,12 @@ pub fn main() !void { var i: usize = 0; while (field_iter.next()) |field| : (i += 1) { switch (i) { - 0 => cps[0] = try fmt.parseInt(u24, field, 16), + 0 => { + var cp = try fmt.parseInt(u21, field, 16); + cp <<= 1; + if (cp_map.contains(cp)) cp |= 1; + cps[0] = cp; + }, 1 => { if (!mem.eql(u8, field, "C") and !mem.eql(u8, field, "F")) continue :lines; -- cgit v1.2.3