2 files changed, 120 insertions, 5 deletions

diff --git a/src/display_width.zig b/src/display_width.zig
new file mode 100644
index 0000000..e06aa8f
--- /dev/null
+++ b/src/display_width.zig
@@ -0,0 +1,111 @@
+const std = @import("std");
+const testing = std.testing;
+
+const CodePointIterator = @import("CodePoint").CodePointIterator;
+const GraphemeIterator = @import("Grapheme").GraphemeIterator;
+const dwp = @import("dwp");
+
+/// codePointWidth returns the number of cells `cp` requires when rendered
+/// in a fixed-pitch font (i.e. a terminal screen). This can range from -1 to
+/// 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(cp: u21) i3 {
+    return dwp.stage_2[dwp.stage_1[cp >> 8] + (cp & 0xff)];
+}
+
+fn strWidth(str: []const u8) usize {
+    var total: isize = 0;
+    var giter = GraphemeIterator.init(str);
+
+    while (giter.next()) |gc| {
+        var cp_iter = CodePointIterator{ .bytes = str[gc.offset..][0..gc.len] };
+        var gc_total: isize = 0;
+
+        while (cp_iter.next()) |cp| {
+            var w = codePointWidth(cp.code);
+
+            if (w != 0) {
+                // Handle text emoji sequence.
+                if (cp_iter.next()) |ncp| {
+                    // emoji text sequence.
+                    if (ncp.code == 0xFE0E) w = 1;
+                }
+
+                // Only adding width of first non-zero-width code point.
+                if (gc_total == 0) gc_total = w;
+            }
+        }
+
+        total += gc_total;
+    }
+
+    return if (total > 0) @intCast(total) else 0;
+}
+
+test "display_width Width" {
+    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(usize, 5), strWidth("Hello\r\n"));
+    try testing.expectEqual(@as(usize, 1), strWidth("\u{0065}\u{0301}"));
+    try testing.expectEqual(@as(usize, 2), strWidth("\u{1F476}\u{1F3FF}\u{0308}\u{200D}\u{1F476}\u{1F3FF}"));
+    try testing.expectEqual(@as(usize, 8), strWidth("Hello 😊"));
+    try testing.expectEqual(@as(usize, 8), strWidth("Héllo 😊"));
+    try testing.expectEqual(@as(usize, 8), strWidth("Héllo :)"));
+    try testing.expectEqual(@as(usize, 8), strWidth("Héllo 🇪🇸"));
+    try testing.expectEqual(@as(usize, 2), strWidth("\u{26A1}")); // Lone emoji
+    try testing.expectEqual(@as(usize, 1), strWidth("\u{26A1}\u{FE0E}")); // Text sequence
+    try testing.expectEqual(@as(usize, 2), strWidth("\u{26A1}\u{FE0F}")); // Presentation sequence
+    try testing.expectEqual(@as(usize, 0), strWidth("A\x08")); // Backspace
+    try testing.expectEqual(@as(usize, 0), strWidth("\x7FA")); // DEL
+    try testing.expectEqual(@as(usize, 0), strWidth("\x7FA\x08\x08")); // never less than o
+
+    // wcwidth Python lib tests. See: https://github.com/jquast/wcwidth/blob/master/tests/test_core.py
+    const empty = "";
+    try testing.expectEqual(@as(usize, 0), strWidth(empty));
+    const with_null = "hello\x00world";
+    try testing.expectEqual(@as(usize, 10), strWidth(with_null));
+    const hello_jp = "コンニチハ, セカイ!";
+    try testing.expectEqual(@as(usize, 19), strWidth(hello_jp));
+    const control = "\x1b[0m";
+    try testing.expectEqual(@as(usize, 3), strWidth(control));
+    const balinese = "\u{1B13}\u{1B28}\u{1B2E}\u{1B44}";
+    try testing.expectEqual(@as(usize, 3), strWidth(balinese));
+
+    // These commented out tests require a new specification for complex scripts.
+    // See: https://www.unicode.org/L2/L2023/23107-terminal-suppt.pdf
+    // const jamo = "\u{1100}\u{1160}";
+    // try testing.expectEqual(@as(usize, 3), strWidth(jamo));
+    // const devengari = "\u{0915}\u{094D}\u{0937}\u{093F}";
+    // try testing.expectEqual(@as(usize, 3), strWidth(devengari));
+    // const tamal = "\u{0b95}\u{0bcd}\u{0bb7}\u{0bcc}";
+    // try testing.expectEqual(@as(usize, 5), strWidth(tamal));
+    // const kannada_1 = "\u{0cb0}\u{0ccd}\u{0c9d}\u{0cc8}";
+    // try testing.expectEqual(@as(usize, 3), strWidth(kannada_1));
+    // The following passes but as a mere coincidence.
+    const kannada_2 = "\u{0cb0}\u{0cbc}\u{0ccd}\u{0c9a}";
+    try testing.expectEqual(@as(usize, 2), strWidth(kannada_2));
+}
diff --git a/src/main.zig b/src/main.zig
index fe49300..3e65c7b 100644
--- a/src/main.zig
+++ b/src/main.zig
@@ -1,7 +1,10 @@
 const std = @import("std");
 
 // const GraphemeIterator = @import("ziglyph").GraphemeIterator;
-const GraphemeIterator = @import("Grapheme").GraphemeIterator;
+// const GraphemeIterator = @import("Grapheme").GraphemeIterator;
+// const codePointWidth = @import("ziglyph").display_width.codePointWidth;
+const codePointWidth = @import("display_width").codePointWidth;
+const CodePointIterator = @import("CodePoint").CodePointIterator;
 
 pub fn main() !void {
     var gpa = std.heap.GeneralPurposeAllocator(.{}){};
@@ -11,14 +14,15 @@ pub fn main() !void {
     const input = try std.fs.cwd().readFileAlloc(allocator, "lang_mix.txt", std.math.maxInt(u32));
     defer allocator.free(input);
 
-    var result: usize = 0;
-    var iter = GraphemeIterator.init(input);
+    var result: isize = 0;
+    // var iter = GraphemeIterator.init(input);
+    var iter = CodePointIterator{ .bytes = input };
 
     var timer = try std.time.Timer.start();
 
     // for (0..50) |_| {
-    while (iter.next()) |_| result += 1;
-    iter.cp_iter.i = 0;
+    while (iter.next()) |cp| result += codePointWidth(@intCast(cp.code));
+    // iter.cp_iter.i = 0;
     // }
 
     std.debug.print("result: {}, took: {}\n", .{ result, timer.lap() / std.time.ns_per_ms });