1 files changed, 84 insertions, 38 deletions

diff --git a/src/file_pane.zig b/src/file_pane.zig
index cf2b9571..abf37448 100644
--- a/src/file_pane.zig
+++ b/src/file_pane.zig
@@ -4,6 +4,7 @@
 //! number gutter and the syntax recolor). The rest of a file pane's behaviour
 //! is the pane machinery in pardes.zig, which does not care what kind it is.
 const std = @import("std");
+const vaxis = @import("vaxis");
 const pardes = @import("pardes.zig");
 const config = @import("config.zig");
 const Pardes = pardes.Pardes;
@@ -82,13 +83,23 @@ pub fn dumpPane(
     };
 }
 
+pub fn graphemeDisplayWidth(grapheme: []const u8) usize {
+    if (std.mem.eql(u8, grapheme, "\t")) return config.tab_width;
+    return @max(1, @as(usize, vaxis.gwidth.gwidth(grapheme, .unicode)));
+}
+
 pub fn byteDisplayWidth(byte: u8) usize {
     return if (byte == '\t') config.tab_width else 1;
 }
 
 pub fn displayWidth(text: []const u8) usize {
     var width: usize = 0;
-    for (text) |byte| width +|= byteDisplayWidth(byte);
+    var at: usize = 0;
+    while (at < text.len) {
+        const end = modal.nextGrapheme(text, at);
+        width +|= graphemeDisplayWidth(text[at..end]);
+        at = end;
+    }
     return width;
 }
 
@@ -96,10 +107,13 @@ pub fn displayWidth(text: []const u8) usize {
 /// maps back to that one tab byte.
 pub fn byteAtDisplay(text: []const u8, display_col: usize) usize {
     var col: usize = 0;
-    for (text, 0..) |byte, i| {
-        const next = col +| byteDisplayWidth(byte);
-        if (display_col < next) return i;
+    var at: usize = 0;
+    while (at < text.len) {
+        const end = modal.nextGrapheme(text, at);
+        const next = col +| graphemeDisplayWidth(text[at..end]);
+        if (display_col < next) return at;
         col = next;
+        at = end;
     }
     return text.len;
 }
@@ -107,8 +121,8 @@ pub fn byteAtDisplay(text: []const u8, display_col: usize) usize {
 /// File cursor columns may live past EOL. Tabs expand before that boundary;
 /// every virtual column after it remains one screen cell.
 pub fn rawDisplayCol(line_text: []const u8, raw_col: usize) usize {
-    const bounded = @min(raw_col, line_text.len);
-    return displayWidth(line_text[0..bounded]) +| (raw_col - bounded);
+    const bounded = modal.graphemeStart(line_text, @min(raw_col, line_text.len));
+    return displayWidth(line_text[0..bounded]) +| (raw_col -| line_text.len);
 }
 
 pub fn rawAtDisplay(line_text: []const u8, display_col: usize) usize {
@@ -117,6 +131,27 @@ pub fn rawAtDisplay(line_text: []const u8, display_col: usize) usize {
     return byteAtDisplay(line_text, display_col);
 }
 
+pub fn byteAtDisplayFrom(line_text: []const u8, from_raw: usize, display_col: usize) usize {
+    if (from_raw >= line_text.len) return from_raw +| display_col;
+    const from = modal.graphemeStart(line_text, from_raw);
+    return from +| rawAtDisplay(line_text[from..], display_col);
+}
+
+pub fn lineDisplayOffset(line_text: []const u8, from_raw: usize, to_raw: usize) i32 {
+    const from_display = rawDisplayCol(line_text, from_raw);
+    const to_display = rawDisplayCol(line_text, to_raw);
+    if (to_display >= from_display) return @intCast(to_display - from_display);
+    return -@as(i32, @intCast(from_display - to_display));
+}
+
+pub fn lineDisplayEndOffset(line_text: []const u8, from_raw: usize, at_raw: usize) i32 {
+    const start = lineDisplayOffset(line_text, from_raw, at_raw);
+    if (at_raw >= line_text.len) return start;
+    const at = modal.graphemeStart(line_text, at_raw);
+    const end = modal.nextGrapheme(line_text, at);
+    return start + @as(i32, @intCast(graphemeDisplayWidth(line_text[at..end]))) - 1;
+}
+
 pub fn sourceLine(pane: *const Pane, row: i32) []const u8 {
     const f = pane.file orelse return "";
     if (row < 0) return "";
@@ -127,51 +162,60 @@ pub fn displayOffset(pane: *const Pane, row: i32, from_raw: i32, to_raw: i32) i3
     const line_text = sourceLine(pane, row);
     const from: usize = @intCast(@max(0, from_raw));
     const to: usize = @intCast(@max(0, to_raw));
-    const from_display = rawDisplayCol(line_text, from);
-    const to_display = rawDisplayCol(line_text, to);
-    if (to_display >= from_display) return @intCast(to_display - from_display);
-    return -@as(i32, @intCast(from_display - to_display));
+    return lineDisplayOffset(line_text, from, to);
 }
 
 pub fn displayEndOffset(pane: *const Pane, row: i32, from_raw: i32, at_raw: i32) i32 {
-    const start = displayOffset(pane, row, from_raw, at_raw);
     const line_text = sourceLine(pane, row);
-    const at: usize = @intCast(@max(0, at_raw));
-    if (at >= line_text.len) return start;
-    return start + @as(i32, @intCast(byteDisplayWidth(line_text[at]))) - 1;
+    return lineDisplayEndOffset(line_text, @intCast(@max(0, from_raw)), @intCast(@max(0, at_raw)));
 }
 
 pub fn byteAtRowDisplay(pane: *const Pane, row: i32, from_raw: i32, display_col: i32) i32 {
     const line_text = sourceLine(pane, row);
-    const from: usize = @intCast(@max(0, from_raw));
-    const display: usize = @intCast(@max(0, display_col));
-    if (from >= line_text.len) return @intCast(from +| display);
-    return @intCast(from +| rawAtDisplay(line_text[from..], display));
+    return @intCast(byteAtDisplayFrom(line_text, @intCast(@max(0, from_raw)), @intCast(@max(0, display_col))));
 }
 
-/// Convert between rendered pane rows (whose file body includes PREFIX_W)
-/// and source-byte columns. Non-file rows are already in byte coordinates.
+/// Convert between rendered cells and UTF-8 byte columns. Tag rows always
+/// need grapheme conversion; file body rows additionally skip PREFIX_W.
 pub fn renderedLineByteCol(pane: *const Pane, row: i32, line_text: []const u8, display_col: usize) usize {
-    if (pane.file == null or row < pardes.BOX_H) return display_col;
+    if (row < pardes.BOX_H) return rawAtDisplay(line_text, display_col);
+    if (pane.file == null) return rawAtDisplay(line_text, display_col);
     const prefix = @min(@as(usize, config.PREFIX_W), line_text.len);
     if (display_col <= prefix) return display_col;
     return prefix +| rawAtDisplay(line_text[prefix..], display_col - prefix);
 }
 
 pub fn renderedLineDisplayCol(pane: *const Pane, row: i32, line_text: []const u8, byte_col: usize) usize {
-    if (pane.file == null or row < pardes.BOX_H) return byte_col;
+    if (row < pardes.BOX_H) return rawDisplayCol(line_text, byte_col);
+    if (pane.file == null) return rawDisplayCol(line_text, byte_col);
     const prefix = @min(@as(usize, config.PREFIX_W), line_text.len);
     if (byte_col <= prefix) return byte_col;
     return prefix +| rawDisplayCol(line_text[prefix..], byte_col - prefix);
 }
 
+test "display columns map complete Unicode graphemes" {
+    const text = "é界e\u{301}x";
+    try std.testing.expectEqual(@as(usize, 5), displayWidth(text));
+    try std.testing.expectEqual(@as(usize, 0), byteAtDisplay(text, 0));
+    try std.testing.expectEqual(@as(usize, 2), byteAtDisplay(text, 1));
+    try std.testing.expectEqual(@as(usize, 2), byteAtDisplay(text, 2));
+    try std.testing.expectEqual(@as(usize, 5), byteAtDisplay(text, 3));
+    try std.testing.expectEqual(@as(usize, 8), byteAtDisplay(text, 4));
+    try std.testing.expectEqual(text.len, byteAtDisplay(text, 5));
+    try std.testing.expectEqual(@as(usize, 3), rawDisplayCol(text, 5));
+    try std.testing.expectEqual(@as(usize, 5), rawAtDisplay(text, 3));
+    try std.testing.expectEqual(@as(usize, 2), graphemeDisplayWidth("👩\u{200d}🚀"));
+}
+
 fn fitEnd(text: []const u8, start: usize, width: usize) usize {
     var end = start;
     var used: usize = 0;
-    while (end < text.len) : (end += 1) {
-        const next = used +| byteDisplayWidth(text[end]);
-        if (next > width) return if (end == start) start + 1 else end;
-        used = next;
+    while (end < text.len) {
+        const next_end = modal.nextGrapheme(text, end);
+        const next_used = used +| graphemeDisplayWidth(text[end..next_end]);
+        if (next_used > width) return if (end == start) next_end else end;
+        used = next_used;
+        end = next_end;
     }
     return end;
 }
@@ -253,7 +297,7 @@ pub fn textOffset(pane: *Pane, text: []const u8, cursor: modal.Cursor) usize {
     const row = @min(cursor.row, index.len - 1);
     const start = index[row];
     const end = if (row + 1 < index.len) index[row + 1] - 1 else text.len;
-    return @min(start + cursor.col, end);
+    return start + modal.graphemeStart(text[start..end], @min(cursor.col, end - start));
 }
 
 pub fn textLineStart(pane: *Pane, text: []const u8, row: usize) usize {
@@ -274,7 +318,9 @@ pub fn textPosition(pane: *Pane, text: []const u8, offset: usize) modal.Cursor {
             return std.math.order(key, item);
         }
     }.cmp) - 1;
-    return .{ .row = row, .col = bounded - index[row] };
+    const start = index[row];
+    const end = if (row + 1 < index.len) index[row + 1] - 1 else text.len;
+    return .{ .row = row, .col = modal.graphemeStart(text[start..end], @min(bounded - start, end - start)) };
 }
 
 pub fn open(p: *Pardes, id: usize, path: []const u8, line: usize) !*Pane {
@@ -577,17 +623,14 @@ fn fillBody(dst: ?[]u8, pane: *Pane, f: *State, width: usize, record_wrap: bool)
             if (dst) |out| @memcpy(out[written..][0..prefix.len], prefix);
             written += prefix.len;
 
-            // A byte cut, like the hscroll one below, and it can land inside a
-            // multi-byte glyph for the same reason: columns here are BYTES.
-            // Surface.print decodes by hand and emits U+FFFD per undecodable
-            // byte, so a split glyph renders as a replacement char rather than
-            // panicking — see test/snapshots/badutf.snap. It cannot overflow
-            // the pane either: a UTF-8 sequence is never fewer bytes than the
-            // cells it draws in.
+            // Wrap and horizontal-scroll cuts are always grapheme boundaries.
+            // Source columns remain byte offsets, while widths are terminal
+            // cells; keeping the conversion here prevents a view operation
+            // from manufacturing malformed UTF-8.
             const end = if (width == 0) text.len else fitEnd(text, at, width);
             const take = end - at;
             const cut = if (pane.hscroll > 0 and width == 0)
-                @min(@as(usize, @intCast(pane.hscroll)), take)
+                modal.graphemeStart(text[at..end], @min(@as(usize, @intCast(pane.hscroll)), take))
             else
                 0;
             const shown = text[at + cut .. end];
@@ -690,10 +733,12 @@ pub fn recolorSyntax(p: *Pardes, pane: *Pane, f: *State, r: pardes.Rect, tx: u16
             else
                 line.len;
         }
+        hs = modal.graphemeStart(line, @min(hs, line.len));
         var c: usize = 0;
         var screen_c: usize = 0;
-        while (hs + c < limit and config.PREFIX_W + screen_c < tw) : (c += 1) {
-            const cells = byteDisplayWidth(line[hs + c]);
+        while (hs + c < limit and config.PREFIX_W + screen_c < tw) {
+            const grapheme_end = @min(limit, modal.nextGrapheme(line, hs + c));
+            const cells = graphemeDisplayWidth(line[hs + c .. grapheme_end]);
             const idx = base + hs + c;
             if (idx >= f.highlight_start) {
                 const hidx = idx - f.highlight_start;
@@ -710,6 +755,7 @@ pub fn recolorSyntax(p: *Pardes, pane: *Pane, f: *State, r: pardes.Rect, tx: u16
                 }
             }
             screen_c += cells;
+            c = grapheme_end - hs;
         }
     }
 }