From 4dd8121ebb60c023465199d565557699cf70fea6 Mon Sep 17 00:00:00 2001
From: Shibo Lyu <hi@lao.sb>
Date: Tue, 28 Jan 2025 14:31:14 +0800
Subject: [PATCH] feat: support any unicode grapheme clusters.

---
 deno.json               |  5 +++--
 deno.lock               |  7 ++++++-
 logic/character.ts      | 13 +++----------
 tests/character.test.ts |  6 ++----
 4 files changed, 14 insertions(+), 17 deletions(-)

diff --git a/deno.json b/deno.json
index 948c947..2ac2e2b 100644
--- a/deno.json
+++ b/deno.json
@@ -1,8 +1,9 @@
 {
   "name": "@textplace/core",
-  "version": "0.3.2",
+  "version": "0.4.0",
   "exports": "./mod.ts",
   "imports": {
-    "@deno/dnt": "jsr:@deno/dnt@^0.41.3"
+    "@deno/dnt": "jsr:@deno/dnt@^0.41.3",
+    "@std/cli": "jsr:@std/cli@^1.0.11"
   }
 }
diff --git a/deno.lock b/deno.lock
index 491b9f0..afae4c4 100644
--- a/deno.lock
+++ b/deno.lock
@@ -7,6 +7,7 @@
     "jsr:@std/assert@0.223": "0.223.0",
     "jsr:@std/assert@0.226": "0.226.0",
     "jsr:@std/bytes@0.223": "0.223.0",
+    "jsr:@std/cli@^1.0.11": "1.0.11",
     "jsr:@std/fmt@0.223": "0.223.0",
     "jsr:@std/fmt@1": "1.0.3",
     "jsr:@std/fs@0.223": "0.223.0",
@@ -54,6 +55,9 @@
     "@std/bytes@0.223.0": {
       "integrity": "84b75052cd8680942c397c2631318772b295019098f40aac5c36cead4cba51a8"
     },
+    "@std/cli@1.0.11": {
+      "integrity": "ec219619fdcd31bcf0d8e53bee1e2706ec9a02f70255365a094f69755dadd340"
+    },
     "@std/fmt@0.223.0": {
       "integrity": "6deb37794127dfc7d7bded2586b9fc6f5d50e62a8134846608baf71ffc1a5208"
     },
@@ -151,7 +155,8 @@
   },
   "workspace": {
     "dependencies": [
-      "jsr:@deno/dnt@~0.41.3"
+      "jsr:@deno/dnt@~0.41.3",
+      "jsr:@std/cli@^1.0.11"
     ]
   }
 }
diff --git a/logic/character.ts b/logic/character.ts
index d1b9f69..074001d 100644
--- a/logic/character.ts
+++ b/logic/character.ts
@@ -1,7 +1,6 @@
+import { unicodeWidth } from "@std/cli/unicode-width";
+
 const segmenter = new Intl.Segmenter("en", { granularity: "grapheme" });
-const cjkRegex =
-  /[\p{Unified_Ideograph}\u30A0-\u30FF\u3040-\u309F\u31F0-\u31FF]/u;
-const printableASCIIRegex = /^[\x20-\x7E]$/;
 
 export function getCharacterWidth(ch: string): number {
   const segments = [...segmenter.segment(ch)];
@@ -11,11 +10,5 @@ export function getCharacterWidth(ch: string): number {
     );
   }
 
-  const matchesASCII = ch.match(printableASCIIRegex);
-  const matchesCJK = ch.match(cjkRegex);
-
-  if (!matchesASCII && !matchesCJK) throw new Error(`Invalid character: ${ch}`);
-
-  // TODO: Support Emojis.
-  return matchesCJK ? 2 : 1;
+  return unicodeWidth(ch);
 }
diff --git a/tests/character.test.ts b/tests/character.test.ts
index 886d7ad..d1330f9 100644
--- a/tests/character.test.ts
+++ b/tests/character.test.ts
@@ -26,14 +26,12 @@ Deno.test("getCharacterWidth CJK", () => {
   assertEquals(getCharacterWidth("グ"), 2);
   assertEquals(getCharacterWidth("ソ"), 2);
 
-  assertThrows(() => getCharacterWidth("？"));
-  assertThrows(() => getCharacterWidth("！"));
+  assertEquals(getCharacterWidth("？"), 2);
+  assertEquals(getCharacterWidth("！"), 2);
   assertThrows(() => getCharacterWidth("你好"));
   assertThrows(() => getCharacterWidth("ヨスガノ"));
 });
 
 Deno.test("getCharacterWidth previously faulty cases", () => {
   assertEquals(getCharacterWidth("𤲶"), 2);
-
-  assertThrows(() => getCharacterWidth("𤲶"[0]));
 });