diff options
| author | 2025-04-30 12:02:17 -0400 | |
|---|---|---|
| committer | 2025-04-30 12:02:17 -0400 | |
| commit | 7a212f5ec5aabf016d17d3ed28649e7982b810ef (patch) | |
| tree | c6b06b0a0afb0ed2ba18f147d9ee200e5eee09a1 /src/WidthData.zig | |
| parent | Factor out 'Data' for grapheme and DisplayWidth (diff) | |
| download | zg-7a212f5ec5aabf016d17d3ed28649e7982b810ef.tar.gz zg-7a212f5ec5aabf016d17d3ed28649e7982b810ef.tar.xz zg-7a212f5ec5aabf016d17d3ed28649e7982b810ef.zip | |
grapheme now Graphemes, Data files gone
Diffstat (limited to 'src/WidthData.zig')
| -rw-r--r-- | src/WidthData.zig | 102 |
1 files changed, 0 insertions, 102 deletions
diff --git a/src/WidthData.zig b/src/WidthData.zig deleted file mode 100644 index ca7eaf0..0000000 --- a/src/WidthData.zig +++ /dev/null | |||
| @@ -1,102 +0,0 @@ | |||
| 1 | const std = @import("std"); | ||
| 2 | const builtin = @import("builtin"); | ||
| 3 | const compress = std.compress; | ||
| 4 | const mem = std.mem; | ||
| 5 | const testing = std.testing; | ||
| 6 | |||
| 7 | const Graphemes = @import("Graphemes"); | ||
| 8 | |||
| 9 | g_data: Graphemes, | ||
| 10 | s1: []u16 = undefined, | ||
| 11 | s2: []i4 = undefined, | ||
| 12 | owns_gdata: bool, | ||
| 13 | |||
| 14 | const Self = @This(); | ||
| 15 | |||
| 16 | pub fn init(allocator: mem.Allocator) mem.Allocator.Error!Self { | ||
| 17 | var self: Self = try Self.setup(allocator); | ||
| 18 | errdefer { | ||
| 19 | allocator.free(self.s1); | ||
| 20 | allocator.free(self.s2); | ||
| 21 | } | ||
| 22 | self.owns_gdata = true; | ||
| 23 | self.g_data = try Graphemes.init(allocator); | ||
| 24 | errdefer self.g_data.deinit(allocator); | ||
| 25 | return self; | ||
| 26 | } | ||
| 27 | |||
| 28 | pub fn initWithGraphemeData(allocator: mem.Allocator, g_data: Graphemes) mem.Allocator.Error!Self { | ||
| 29 | var self = try Self.setup(allocator); | ||
| 30 | self.g_data = g_data; | ||
| 31 | self.owns_gdata = false; | ||
| 32 | return self; | ||
| 33 | } | ||
| 34 | |||
| 35 | // Sets up the DisplayWidthData, leaving the GraphemeData undefined. | ||
| 36 | fn setup(allocator: mem.Allocator) mem.Allocator.Error!Self { | ||
| 37 | const decompressor = compress.flate.inflate.decompressor; | ||
| 38 | const in_bytes = @embedFile("dwp"); | ||
| 39 | var in_fbs = std.io.fixedBufferStream(in_bytes); | ||
| 40 | var in_decomp = decompressor(.raw, in_fbs.reader()); | ||
| 41 | var reader = in_decomp.reader(); | ||
| 42 | |||
| 43 | const endian = builtin.cpu.arch.endian(); | ||
| 44 | |||
| 45 | var self: Self = undefined; | ||
| 46 | |||
| 47 | const stage_1_len: u16 = reader.readInt(u16, endian) catch unreachable; | ||
| 48 | self.s1 = try allocator.alloc(u16, stage_1_len); | ||
| 49 | errdefer allocator.free(self.s1); | ||
| 50 | for (0..stage_1_len) |i| self.s1[i] = reader.readInt(u16, endian) catch unreachable; | ||
| 51 | |||
| 52 | const stage_2_len: u16 = reader.readInt(u16, endian) catch unreachable; | ||
| 53 | self.s2 = try allocator.alloc(i4, stage_2_len); | ||
| 54 | errdefer allocator.free(self.s2); | ||
| 55 | for (0..stage_2_len) |i| self.s2[i] = @intCast(reader.readInt(i8, endian) catch unreachable); | ||
| 56 | |||
| 57 | return self; | ||
| 58 | } | ||
| 59 | |||
| 60 | pub fn deinit(self: *const Self, allocator: mem.Allocator) void { | ||
| 61 | allocator.free(self.s1); | ||
| 62 | allocator.free(self.s2); | ||
| 63 | if (self.owns_gdata) self.g_data.deinit(allocator); | ||
| 64 | } | ||
| 65 | |||
| 66 | /// codePointWidth returns the number of cells `cp` requires when rendered | ||
| 67 | /// in a fixed-pitch font (i.e. a terminal screen). This can range from -1 to | ||
| 68 | /// 3, where BACKSPACE and DELETE return -1 and 3-em-dash returns 3. C0/C1 | ||
| 69 | /// control codes return 0. If `cjk` is true, ambiguous code points return 2, | ||
| 70 | /// otherwise they return 1. | ||
| 71 | pub fn codePointWidth(self: Self, cp: u21) i4 { | ||
| 72 | return self.s2[self.s1[cp >> 8] + (cp & 0xff)]; | ||
| 73 | } | ||
| 74 | |||
| 75 | test "codePointWidth" { | ||
| 76 | const wd = try Self.init(std.testing.allocator); | ||
| 77 | defer wd.deinit(std.testing.allocator); | ||
| 78 | try testing.expectEqual(@as(i4, 0), wd.codePointWidth(0x0000)); // null | ||
| 79 | try testing.expectEqual(@as(i4, -1), wd.codePointWidth(0x8)); // \b | ||
| 80 | try testing.expectEqual(@as(i4, -1), wd.codePointWidth(0x7f)); // DEL | ||
| 81 | try testing.expectEqual(@as(i4, 0), wd.codePointWidth(0x0005)); // Cf | ||
| 82 | try testing.expectEqual(@as(i4, 0), wd.codePointWidth(0x0007)); // \a BEL | ||
| 83 | try testing.expectEqual(@as(i4, 0), wd.codePointWidth(0x000A)); // \n LF | ||
| 84 | try testing.expectEqual(@as(i4, 0), wd.codePointWidth(0x000B)); // \v VT | ||
| 85 | try testing.expectEqual(@as(i4, 0), wd.codePointWidth(0x000C)); // \f FF | ||
| 86 | try testing.expectEqual(@as(i4, 0), wd.codePointWidth(0x000D)); // \r CR | ||
| 87 | try testing.expectEqual(@as(i4, 0), wd.codePointWidth(0x000E)); // SQ | ||
| 88 | try testing.expectEqual(@as(i4, 0), wd.codePointWidth(0x000F)); // SI | ||
| 89 | |||
| 90 | try testing.expectEqual(@as(i4, 0), wd.codePointWidth(0x070F)); // Cf | ||
| 91 | try testing.expectEqual(@as(i4, 1), wd.codePointWidth(0x0603)); // Cf Arabic | ||
| 92 | |||
| 93 | try testing.expectEqual(@as(i4, 1), wd.codePointWidth(0x00AD)); // soft-hyphen | ||
| 94 | try testing.expectEqual(@as(i4, 2), wd.codePointWidth(0x2E3A)); // two-em dash | ||
| 95 | try testing.expectEqual(@as(i4, 3), wd.codePointWidth(0x2E3B)); // three-em dash | ||
| 96 | |||
| 97 | try testing.expectEqual(@as(i4, 1), wd.codePointWidth(0x00BD)); // ambiguous halfwidth | ||
| 98 | |||
| 99 | try testing.expectEqual(@as(i4, 1), wd.codePointWidth('é')); | ||
| 100 | try testing.expectEqual(@as(i4, 2), wd.codePointWidth('😊')); | ||
| 101 | try testing.expectEqual(@as(i4, 2), wd.codePointWidth('统')); | ||
| 102 | } | ||