Add c0 and c1 control width options

This allows a build of DisplayWidth to give characters in those classes
a width, for cases where they'll be printed with a substitute in the
final display.  It also raises the size of possible characters from an
i3 to an i4, to accommodate printing C1s as e.g. <80> or \u{80}.

2 files changed, 36 insertions, 32 deletions

diff --git a/src/DisplayWidth.zig b/src/DisplayWidth.zig
index 621b8c1..04e6b0c 100644
--- a/src/DisplayWidth.zig
+++ b/src/DisplayWidth.zig
@@ -1,5 +1,6 @@
 const std = @import("std");
 const builtin = @import("builtin");
+const options = @import("options");
 const ArrayList = std.ArrayList;
 const mem = std.mem;
 const simd = std.simd;
@@ -60,6 +61,7 @@ test "strWidth" {
     const data = try DisplayWidthData.init(testing.allocator);
     defer data.deinit();
     const self = Self{ .data = &data };
+    const c0 = options.c0_width orelse 0;
 
     try testing.expectEqual(@as(usize, 5), self.strWidth("Hello\r\n"));
     try testing.expectEqual(@as(usize, 1), self.strWidth("\u{0065}\u{0301}"));
@@ -74,19 +76,21 @@ test "strWidth" {
     try testing.expectEqual(@as(usize, 1), self.strWidth("\u{2764}")); // Default text presentation
     try testing.expectEqual(@as(usize, 1), self.strWidth("\u{2764}\u{FE0E}")); // Default text presentation with VS15 selector
     try testing.expectEqual(@as(usize, 2), self.strWidth("\u{2764}\u{FE0F}")); // Default text presentation with VS16 selector
-    try testing.expectEqual(@as(usize, 0), self.strWidth("A\x08")); // Backspace
-    try testing.expectEqual(@as(usize, 0), self.strWidth("\x7FA")); // DEL
-    try testing.expectEqual(@as(usize, 0), self.strWidth("\x7FA\x08\x08")); // never less than o
+    const expect_bs: usize = if (c0 == 0) 0 else 1 + c0;
+    try testing.expectEqual(expect_bs, self.strWidth("A\x08")); // Backspace
+    try testing.expectEqual(expect_bs, self.strWidth("\x7FA")); // DEL
+    const expect_long_del: usize = if (c0 == 0) 0 else 1 + (c0 * 3);
+    try testing.expectEqual(expect_long_del, self.strWidth("\x7FA\x08\x08")); // never less than 0
 
     // wcwidth Python lib tests. See: https://github.com/jquast/wcwidth/blob/master/tests/test_core.py
     const empty = "";
     try testing.expectEqual(@as(usize, 0), self.strWidth(empty));
     const with_null = "hello\x00world";
-    try testing.expectEqual(@as(usize, 10), self.strWidth(with_null));
+    try testing.expectEqual(@as(usize, 10 + c0), self.strWidth(with_null));
     const hello_jp = "コンニチハ, セカイ!";
     try testing.expectEqual(@as(usize, 19), self.strWidth(hello_jp));
     const control = "\x1b[0m";
-    try testing.expectEqual(@as(usize, 3), self.strWidth(control));
+    try testing.expectEqual(@as(usize, 3 + c0), self.strWidth(control));
     const balinese = "\u{1B13}\u{1B28}\u{1B2E}\u{1B44}";
     try testing.expectEqual(@as(usize, 3), self.strWidth(balinese));
 
diff --git a/src/WidthData.zig b/src/WidthData.zig
index 1b7fb2e..d77879e 100644
--- a/src/WidthData.zig
+++ b/src/WidthData.zig
@@ -9,7 +9,7 @@ const GraphemeData = @import("GraphemeData");
 allocator: mem.Allocator,
 g_data: GraphemeData,
 s1: []u16 = undefined,
-s2: []i3 = undefined,
+s2: []i4 = undefined,
 
 const Self = @This();
 
@@ -34,7 +34,7 @@ pub fn init(allocator: mem.Allocator) mem.Allocator.Error!Self {
     for (0..stage_1_len) |i| self.s1[i] = reader.readInt(u16, endian) catch unreachable;
 
     const stage_2_len: u16 = reader.readInt(u16, endian) catch unreachable;
-    self.s2 = try allocator.alloc(i3, stage_2_len);
+    self.s2 = try allocator.alloc(i4, stage_2_len);
     errdefer allocator.free(self.s2);
     for (0..stage_2_len) |i| self.s2[i] = @intCast(reader.readInt(i8, endian) catch unreachable);
 
@@ -52,33 +52,33 @@ pub fn deinit(self: *const Self) void {
 /// 3, where BACKSPACE and DELETE return -1 and 3-em-dash returns 3. C0/C1
 /// control codes return 0. If `cjk` is true, ambiguous code points return 2,
 /// otherwise they return 1.
-pub fn codePointWidth(self: Self, cp: u21) i3 {
+pub fn codePointWidth(self: Self, cp: u21) i4 {
     return self.s2[self.s1[cp >> 8] + (cp & 0xff)];
 }
 
 test "codePointWidth" {
-    try testing.expectEqual(@as(i3, 0), codePointWidth(0x0000)); // null
-    try testing.expectEqual(@as(i3, -1), codePointWidth(0x8)); // \b
-    try testing.expectEqual(@as(i3, -1), codePointWidth(0x7f)); // DEL
-    try testing.expectEqual(@as(i3, 0), codePointWidth(0x0005)); // Cf
-    try testing.expectEqual(@as(i3, 0), codePointWidth(0x0007)); // \a BEL
-    try testing.expectEqual(@as(i3, 0), codePointWidth(0x000A)); // \n LF
-    try testing.expectEqual(@as(i3, 0), codePointWidth(0x000B)); // \v VT
-    try testing.expectEqual(@as(i3, 0), codePointWidth(0x000C)); // \f FF
-    try testing.expectEqual(@as(i3, 0), codePointWidth(0x000D)); // \r CR
-    try testing.expectEqual(@as(i3, 0), codePointWidth(0x000E)); // SQ
-    try testing.expectEqual(@as(i3, 0), codePointWidth(0x000F)); // SI
-
-    try testing.expectEqual(@as(i3, 0), codePointWidth(0x070F)); // Cf
-    try testing.expectEqual(@as(i3, 1), codePointWidth(0x0603)); // Cf Arabic
-
-    try testing.expectEqual(@as(i3, 1), codePointWidth(0x00AD)); // soft-hyphen
-    try testing.expectEqual(@as(i3, 2), codePointWidth(0x2E3A)); // two-em dash
-    try testing.expectEqual(@as(i3, 3), codePointWidth(0x2E3B)); // three-em dash
-
-    try testing.expectEqual(@as(i3, 1), codePointWidth(0x00BD)); // ambiguous halfwidth
-
-    try testing.expectEqual(@as(i3, 1), codePointWidth('é'));
-    try testing.expectEqual(@as(i3, 2), codePointWidth('😊'));
-    try testing.expectEqual(@as(i3, 2), codePointWidth('统'));
+    try testing.expectEqual(@as(i4, 0), codePointWidth(0x0000)); // null
+    try testing.expectEqual(@as(i4, -1), codePointWidth(0x8)); // \b
+    try testing.expectEqual(@as(i4, -1), codePointWidth(0x7f)); // DEL
+    try testing.expectEqual(@as(i4, 0), codePointWidth(0x0005)); // Cf
+    try testing.expectEqual(@as(i4, 0), codePointWidth(0x0007)); // \a BEL
+    try testing.expectEqual(@as(i4, 0), codePointWidth(0x000A)); // \n LF
+    try testing.expectEqual(@as(i4, 0), codePointWidth(0x000B)); // \v VT
+    try testing.expectEqual(@as(i4, 0), codePointWidth(0x000C)); // \f FF
+    try testing.expectEqual(@as(i4, 0), codePointWidth(0x000D)); // \r CR
+    try testing.expectEqual(@as(i4, 0), codePointWidth(0x000E)); // SQ
+    try testing.expectEqual(@as(i4, 0), codePointWidth(0x000F)); // SI
+
+    try testing.expectEqual(@as(i4, 0), codePointWidth(0x070F)); // Cf
+    try testing.expectEqual(@as(i4, 1), codePointWidth(0x0603)); // Cf Arabic
+
+    try testing.expectEqual(@as(i4, 1), codePointWidth(0x00AD)); // soft-hyphen
+    try testing.expectEqual(@as(i4, 2), codePointWidth(0x2E3A)); // two-em dash
+    try testing.expectEqual(@as(i4, 3), codePointWidth(0x2E3B)); // three-em dash
+
+    try testing.expectEqual(@as(i4, 1), codePointWidth(0x00BD)); // ambiguous halfwidth
+
+    try testing.expectEqual(@as(i4, 1), codePointWidth('é'));
+    try testing.expectEqual(@as(i4, 2), codePointWidth('😊'));
+    try testing.expectEqual(@as(i4, 2), codePointWidth('统'));
 }