summaryrefslogtreecommitdiff
path: root/codegen/wbp.zig
diff options
context:
space:
mode:
authorGravatar Sam Atman2025-12-23 09:34:19 -0500
committerGravatar Sam Atman2025-12-23 09:34:19 -0500
commit79b133e5d88fe6cfce337dd401fc09999db08852 (patch)
tree8b3f9062edde82724c73147abf42143a885640fc /codegen/wbp.zig
parentMerge branch 'develop-next' (diff)
parentUse takeDelimiterInclusive to support Zig 0.15.2 (diff)
downloadzg-79b133e5d88fe6cfce337dd401fc09999db08852.tar.gz
zg-79b133e5d88fe6cfce337dd401fc09999db08852.tar.xz
zg-79b133e5d88fe6cfce337dd401fc09999db08852.zip
Merge branch 'fifteen-two'
Close #90 Close #87 Close #83 Thanks everyone.
Diffstat (limited to 'codegen/wbp.zig')
-rw-r--r--codegen/wbp.zig39
1 files changed, 19 insertions, 20 deletions
diff --git a/codegen/wbp.zig b/codegen/wbp.zig
index 741103e..33eeea5 100644
--- a/codegen/wbp.zig
+++ b/codegen/wbp.zig
@@ -43,7 +43,7 @@ const BlockMap = std.HashMap(
43 std.hash_map.default_max_load_percentage, 43 std.hash_map.default_max_load_percentage,
44); 44);
45 45
46pub fn main() !void { 46pub fn main() anyerror!void {
47 var arena = std.heap.ArenaAllocator.init(std.heap.page_allocator); 47 var arena = std.heap.ArenaAllocator.init(std.heap.page_allocator);
48 defer arena.deinit(); 48 defer arena.deinit();
49 const allocator = arena.allocator(); 49 const allocator = arena.allocator();
@@ -51,15 +51,11 @@ pub fn main() !void {
51 var flat_map = std.AutoHashMap(u21, u5).init(allocator); 51 var flat_map = std.AutoHashMap(u21, u5).init(allocator);
52 defer flat_map.deinit(); 52 defer flat_map.deinit();
53 53
54 var line_buf: [4096]u8 = undefined;
55
56 // Process HangulSyllableType.txt 54 // Process HangulSyllableType.txt
57 var in_file = try std.fs.cwd().openFile("data/unicode/auxiliary/WordBreakProperty.txt", .{}); 55 const in_file = @embedFile("WordBreakProperty.txt");
58 defer in_file.close(); 56 var in_reader = std.io.Reader.fixed(in_file);
59 var in_buf = std.io.bufferedReader(in_file.reader()); 57 while (in_reader.takeDelimiterInclusive('\n')) |took| {
60 const in_reader = in_buf.reader(); 58 const line = std.mem.trimRight(u8, took, "\n");
61
62 while (try in_reader.readUntilDelimiterOrEof(&line_buf, '\n')) |line| {
63 if (line.len == 0 or line[0] == '#') continue; 59 if (line.len == 0 or line[0] == '#') continue;
64 60
65 const no_comment = if (std.mem.indexOfScalar(u8, line, '#')) |octo| line[0..octo] else line; 61 const no_comment = if (std.mem.indexOfScalar(u8, line, '#')) |octo| line[0..octo] else line;
@@ -90,15 +86,19 @@ pub fn main() !void {
90 else => {}, 86 else => {},
91 } 87 }
92 } 88 }
89 } else |err| switch (err) {
90 error.EndOfStream => {},
91 else => {
92 return err;
93 },
93 } 94 }
94
95 var blocks_map = BlockMap.init(allocator); 95 var blocks_map = BlockMap.init(allocator);
96 defer blocks_map.deinit(); 96 defer blocks_map.deinit();
97 97
98 var stage1 = std.ArrayList(u16).init(allocator); 98 var stage1 = std.array_list.Managed(u16).init(allocator);
99 defer stage1.deinit(); 99 defer stage1.deinit();
100 100
101 var stage2 = std.ArrayList(u5).init(allocator); 101 var stage2 = std.array_list.Managed(u5).init(allocator);
102 defer stage2.deinit(); 102 defer stage2.deinit();
103 103
104 var block: Block = [_]u5{0} ** block_size; 104 var block: Block = [_]u5{0} ** block_size;
@@ -129,18 +129,17 @@ pub fn main() !void {
129 _ = args_iter.skip(); 129 _ = args_iter.skip();
130 const output_path = args_iter.next() orelse @panic("No output file arg!"); 130 const output_path = args_iter.next() orelse @panic("No output file arg!");
131 131
132 const compressor = std.compress.flate.deflate.compressor; 132 var write_buf: [4096]u8 = undefined;
133 var out_file = try std.fs.cwd().createFile(output_path, .{}); 133 var out_file = try std.fs.cwd().createFile(output_path, .{});
134 defer out_file.close(); 134 defer out_file.close();
135 var out_comp = try compressor(.raw, out_file.writer(), .{ .level = .best }); 135 var writer = out_file.writer(&write_buf);
136 const writer = out_comp.writer();
137 136
138 const endian = builtin.cpu.arch.endian(); 137 const endian = builtin.cpu.arch.endian();
139 try writer.writeInt(u16, @intCast(stage1.items.len), endian); 138 try writer.interface.writeInt(u16, @intCast(stage1.items.len), endian);
140 for (stage1.items) |i| try writer.writeInt(u16, i, endian); 139 for (stage1.items) |i| try writer.interface.writeInt(u16, i, endian);
141 140
142 try writer.writeInt(u16, @intCast(stage2.items.len), endian); 141 try writer.interface.writeInt(u16, @intCast(stage2.items.len), endian);
143 for (stage2.items) |i| try writer.writeInt(u8, i, endian); 142 for (stage2.items) |i| try writer.interface.writeInt(u8, i, endian);
144 143
145 try out_comp.flush(); 144 try writer.interface.flush();
146} 145}