From 338ce3a365842abe4fe80dc45292ed1109bddc0f Mon Sep 17 00:00:00 2001
From: Mario Zechner <badlogicgames@gmail.com>
Date: Wed, 29 Apr 2026 23:01:04 +0200
Subject: [PATCH] fix(tui): normalize Thai AM terminal output

closes #3904
---
 packages/tui/src/tui.ts                     |  4 ++--
 packages/tui/src/utils.ts                   | 20 ++++++++++----------
 packages/tui/test/truncate-to-width.test.ts | 15 +++++++++++----
 3 files changed, 23 insertions(+), 16 deletions(-)

diff --git a/packages/tui/src/tui.ts b/packages/tui/src/tui.ts
index c5fb8701..fda30282 100644
--- a/packages/tui/src/tui.ts
+++ b/packages/tui/src/tui.ts
@@ -9,7 +9,7 @@ import { performance } from "node:perf_hooks";
 import { isKeyRelease, matchesKey } from "./keys.js";
 import type { Terminal } from "./terminal.js";
 import { getCapabilities, isImageLine, setCellDimensions } from "./terminal-image.js";
-import { extractSegments, sliceByColumn, sliceWithWidth, visibleWidth } from "./utils.js";
+import { extractSegments, normalizeTerminalOutput, sliceByColumn, sliceWithWidth, visibleWidth } from "./utils.js";
 
 /**
  * Component interface - all components must implement this
@@ -800,7 +800,7 @@ export class TUI extends Container {
 		for (let i = 0; i < lines.length; i++) {
 			const line = lines[i];
 			if (!isImageLine(line)) {
-				lines[i] = line + reset;
+				lines[i] = normalizeTerminalOutput(line) + reset;
 			}
 		}
 		return lines;
diff --git a/packages/tui/src/utils.ts b/packages/tui/src/utils.ts
index b8be6461..05d1ce50 100644
--- a/packages/tui/src/utils.ts
+++ b/packages/tui/src/utils.ts
@@ -178,16 +178,6 @@ function graphemeWidth(segment: string): number {
 		return 2;
 	}
 
-	// Thai SARA AM (U+0E33) and Lao VOWEL SIGN AM (U+0EB3) are encoded as
-	// spacing letters, but terminals commonly render isolated occurrences as
-	// mark-like clusters (compat decomposition: nonspacing niggahita + vowel).
-	// Count isolated clusters conservatively to avoid cursor drift and stale
-	// cells during differential rendering. When attached to a base consonant,
-	// the base code point is first and normal per-codepoint width below applies.
-	if (cp === 0x0e33 || cp === 0x0eb3) {
-		return 2;
-	}
-
 	let width = eastAsianWidth(cp);
 
 	// Trailing halfwidth/fullwidth forms and AM vowels that segment with a base.
@@ -265,6 +255,16 @@ export function visibleWidth(str: string): number {
 	return width;
 }
 
+/**
+ * Normalize text for terminal output without changing logical editor content.
+ * Some terminals render precomposed Thai/Lao AM vowels inconsistently during
+ * differential repaint. Their compatibility decompositions have the same cell
+ * width but avoid stale-cell artifacts in terminal renderers.
+ */
+export function normalizeTerminalOutput(str: string): string {
+	return str.replace(/\u0e33/g, "\u0e4d\u0e32").replace(/\u0eb3/g, "\u0ecd\u0eb2");
+}
+
 /**
  * Extract ANSI escape sequences from a string at the given position.
  */
diff --git a/packages/tui/test/truncate-to-width.test.ts b/packages/tui/test/truncate-to-width.test.ts
index 256a8f29..42d93ffa 100644
--- a/packages/tui/test/truncate-to-width.test.ts
+++ b/packages/tui/test/truncate-to-width.test.ts
@@ -1,6 +1,6 @@
 import assert from "node:assert";
 import { describe, it } from "node:test";
-import { truncateToWidth, visibleWidth } from "../src/utils.js";
+import { normalizeTerminalOutput, truncateToWidth, visibleWidth } from "../src/utils.js";
 
 describe("truncateToWidth", () => {
 	it("keeps output within width for very large unicode input", () => {
@@ -60,10 +60,17 @@ describe("visibleWidth", () => {
 		assert.strictEqual(visibleWidth("\t\x1b[31m界\x1b[0m"), 5);
 	});
 
-	it("counts isolated Thai and Lao AM clusters conservatively", () => {
-		assert.strictEqual(visibleWidth("ำ"), 2);
-		assert.strictEqual(visibleWidth("ຳ"), 2);
+	it("keeps Thai and Lao AM clusters at their normal cell width", () => {
+		assert.strictEqual(visibleWidth("ำ"), 1);
+		assert.strictEqual(visibleWidth("ຳ"), 1);
 		assert.strictEqual(visibleWidth("กำ"), 2);
 		assert.strictEqual(visibleWidth("ກຳ"), 2);
 	});
+
+	it("normalizes Thai and Lao AM vowels only for terminal output", () => {
+		assert.strictEqual(normalizeTerminalOutput("ำ"), "ํา");
+		assert.strictEqual(normalizeTerminalOutput("ຳ"), "ໍາ");
+		assert.strictEqual(visibleWidth(normalizeTerminalOutput("ำabc")), visibleWidth("ำabc"));
+		assert.strictEqual(visibleWidth(normalizeTerminalOutput("ຳabc")), visibleWidth("ຳabc"));
+	});
 });