diff options
Diffstat (limited to 'codegen/numeric.zig')
| -rw-r--r-- | codegen/numeric.zig | 45 |
1 files changed, 20 insertions, 25 deletions
diff --git a/codegen/numeric.zig b/codegen/numeric.zig index 80fc7c2..ad8490c 100644 --- a/codegen/numeric.zig +++ b/codegen/numeric.zig | |||
| @@ -1,5 +1,6 @@ | |||
| 1 | const std = @import("std"); | 1 | const std = @import("std"); |
| 2 | const builtin = @import("builtin"); | 2 | const builtin = @import("builtin"); |
| 3 | const mem = std.mem; | ||
| 3 | 4 | ||
| 4 | const block_size = 256; | 5 | const block_size = 256; |
| 5 | const Block = [block_size]u8; | 6 | const Block = [block_size]u8; |
| @@ -15,7 +16,7 @@ const BlockMap = std.HashMap( | |||
| 15 | } | 16 | } |
| 16 | 17 | ||
| 17 | pub fn eql(_: @This(), a: Block, b: Block) bool { | 18 | pub fn eql(_: @This(), a: Block, b: Block) bool { |
| 18 | return std.mem.eql(u8, &a, &b); | 19 | return mem.eql(u8, &a, &b); |
| 19 | } | 20 | } |
| 20 | }, | 21 | }, |
| 21 | std.hash_map.default_max_load_percentage, | 22 | std.hash_map.default_max_load_percentage, |
| @@ -31,17 +32,17 @@ pub fn main() !void { | |||
| 31 | 32 | ||
| 32 | var line_buf: [4096]u8 = undefined; | 33 | var line_buf: [4096]u8 = undefined; |
| 33 | 34 | ||
| 34 | // Process DerivedCombiningClass.txt | 35 | // Process DerivedNumericType.txt |
| 35 | var num_file = try std.fs.cwd().openFile("data/unicode/extracted/DerivedNumericType.txt", .{}); | 36 | var in_file = try std.fs.cwd().openFile("data/unicode/extracted/DerivedNumericType.txt", .{}); |
| 36 | defer num_file.close(); | 37 | defer in_file.close(); |
| 37 | var num_buf = std.io.bufferedReader(num_file.reader()); | 38 | var in_buf = std.io.bufferedReader(in_file.reader()); |
| 38 | const num_reader = num_buf.reader(); | 39 | const in_reader = in_buf.reader(); |
| 39 | 40 | ||
| 40 | while (try num_reader.readUntilDelimiterOrEof(&line_buf, '\n')) |line| { | 41 | while (try in_reader.readUntilDelimiterOrEof(&line_buf, '\n')) |line| { |
| 41 | if (line.len == 0 or line[0] == '#') continue; | 42 | if (line.len == 0 or line[0] == '#') continue; |
| 42 | const no_comment = if (std.mem.indexOfScalar(u8, line, '#')) |octo| line[0..octo] else line; | 43 | const no_comment = if (mem.indexOfScalar(u8, line, '#')) |octo| line[0..octo] else line; |
| 43 | 44 | ||
| 44 | var field_iter = std.mem.tokenizeAny(u8, no_comment, "; "); | 45 | var field_iter = mem.tokenizeAny(u8, no_comment, "; "); |
| 45 | var current_code: [2]u21 = undefined; | 46 | var current_code: [2]u21 = undefined; |
| 46 | 47 | ||
| 47 | var i: usize = 0; | 48 | var i: usize = 0; |
| @@ -49,7 +50,7 @@ pub fn main() !void { | |||
| 49 | switch (i) { | 50 | switch (i) { |
| 50 | 0 => { | 51 | 0 => { |
| 51 | // Code point(s) | 52 | // Code point(s) |
| 52 | if (std.mem.indexOf(u8, field, "..")) |dots| { | 53 | if (mem.indexOf(u8, field, "..")) |dots| { |
| 53 | current_code = .{ | 54 | current_code = .{ |
| 54 | try std.fmt.parseInt(u21, field[0..dots], 16), | 55 | try std.fmt.parseInt(u21, field[0..dots], 16), |
| 55 | try std.fmt.parseInt(u21, field[dots + 2 ..], 16), | 56 | try std.fmt.parseInt(u21, field[dots + 2 ..], 16), |
| @@ -61,25 +62,19 @@ pub fn main() !void { | |||
| 61 | }, | 62 | }, |
| 62 | 1 => { | 63 | 1 => { |
| 63 | // Numeric type | 64 | // Numeric type |
| 64 | if (std.mem.eql(u8, field, "Numeric")) { | 65 | var bit: u8 = 0; |
| 65 | for (current_code[0]..current_code[1] + 1) |cp| { | 66 | |
| 66 | const gop = try flat_map.getOrPut(@intCast(cp)); | 67 | if (mem.eql(u8, field, "Numeric")) bit = 1; |
| 67 | if (!gop.found_existing) gop.value_ptr.* = 0; | 68 | if (mem.eql(u8, field, "Digit")) bit = 2; |
| 68 | gop.value_ptr.* |= 1; | 69 | if (mem.eql(u8, field, "Decimal")) bit = 4; |
| 69 | } | 70 | |
| 70 | } else if (std.mem.eql(u8, field, "Digit")) { | 71 | if (bit != 0) { |
| 71 | for (current_code[0]..current_code[1] + 1) |cp| { | ||
| 72 | const gop = try flat_map.getOrPut(@intCast(cp)); | ||
| 73 | if (!gop.found_existing) gop.value_ptr.* = 0; | ||
| 74 | gop.value_ptr.* |= 2; | ||
| 75 | } | ||
| 76 | } else if (std.mem.eql(u8, field, "Decimal")) { | ||
| 77 | for (current_code[0]..current_code[1] + 1) |cp| { | 72 | for (current_code[0]..current_code[1] + 1) |cp| { |
| 78 | const gop = try flat_map.getOrPut(@intCast(cp)); | 73 | const gop = try flat_map.getOrPut(@intCast(cp)); |
| 79 | if (!gop.found_existing) gop.value_ptr.* = 0; | 74 | if (!gop.found_existing) gop.value_ptr.* = 0; |
| 80 | gop.value_ptr.* |= 4; | 75 | gop.value_ptr.* |= bit; |
| 81 | } | 76 | } |
| 82 | } else continue; | 77 | } |
| 83 | }, | 78 | }, |
| 84 | else => {}, | 79 | else => {}, |
| 85 | } | 80 | } |