Avoid collapse leading and trailing emphatics (#1227)

* Avoid collapse leading and trailing emphatics

* Lint

* Fix typing

ext/js/language/ja/japanese.js

@@ -747,29 +747,40 @@ export function isEmphaticCodePoint(codePoint) {
  * @returns {string}
  */
 export function collapseEmphaticSequences(text, fullCollapse) {
-    let result = '';
-    let collapseCodePoint = -1;
-    for (let i = 0; i < text.length; ++i) {
-        const char = text[i];
-        const c = char.codePointAt(0) ?? -1;
-        if (isEmphaticCodePoint(c)) {
-            // Prevent match trailing emphatic
-            if (i === text.length - 1) {
-                result += char;
-                continue;
-            }
+    let left = 0;
+    while (left < text.length && isEmphaticCodePoint(/** @type {number} */ (text.codePointAt(left)))) {
+        ++left;
+    }
+    let right = text.length - 1;
+    while (right >= 0 && isEmphaticCodePoint(/** @type {number} */ (text.codePointAt(right)))) {
+        --right;
+    }
+    // Whole string is emphatic
+    if (left > right) {
+        return text;
+    }
+
+    const leadingEmphatics = text.substring(0, left);
+    const trailingEmphatics = text.substring(right + 1);
+    let middle = '';
+    let currentCollapsedCodePoint = -1;
 
-            if (collapseCodePoint !== c) {
-                collapseCodePoint = c;
+    for (let i = left; i <= right; ++i) {
+        const char = text[i];
+        const codePoint = /** @type {number} */ (char.codePointAt(0));
+        if (isEmphaticCodePoint(codePoint)) {
+            if (currentCollapsedCodePoint !== codePoint) {
+                currentCollapsedCodePoint = codePoint;
                 if (!fullCollapse) {
-                    result += char;
+                    middle += char;
                     continue;
                 }
             }
         } else {
-            collapseCodePoint = -1;
-            result += char;
+            currentCollapsedCodePoint = -1;
+            middle += char;
         }
     }
-    return result;
+
+    return leadingEmphatics + middle + trailingEmphatics;
 }

test/japanese-util.test.js

@@ -767,15 +767,6 @@ describe('Japanese utility functions', () => {
             [['かっっっこい', false], 'かっこい'],
             [['かっっっこい', true], 'かこい'],
 
-            [['こい', false], 'こい'],
-            [['こい', true], 'こい'],
-            [['っこい', false], 'っこい'],
-            [['っこい', true], 'こい'],
-            [['っっこい', false], 'っこい'],
-            [['っっこい', true], 'こい'],
-            [['っっっこい', false], 'っこい'],
-            [['っっっこい', true], 'こい'],
-
             [['すごい', false], 'すごい'],
             [['すごい', true], 'すごい'],
             [['すごーい', false], 'すごーい'],
@@ -787,21 +778,45 @@ describe('Japanese utility functions', () => {
             [['すっっごーーい', false], 'すっごーい'],
             [['すっっごーーい', true], 'すごい'],
 
+            [['こい', false], 'こい'],
+            [['こい', true], 'こい'],
+            [['っこい', false], 'っこい'],
+            [['っこい', true], 'っこい'],
+            [['っっこい', false], 'っっこい'],
+            [['っっこい', true], 'っっこい'],
+            [['っっっこい', false], 'っっっこい'],
+            [['っっっこい', true], 'っっっこい'],
+            [['こいっ', false], 'こいっ'],
+            [['こいっ', true], 'こいっ'],
+            [['こいっっ', false], 'こいっっ'],
+            [['こいっっ', true], 'こいっっ'],
+            [['こいっっっ', false], 'こいっっっ'],
+            [['こいっっっ', true], 'こいっっっ'],
+            [['っこいっ', false], 'っこいっ'],
+            [['っこいっ', true], 'っこいっ'],
+            [['っっこいっっ', false], 'っっこいっっ'],
+            [['っっこいっっ', true], 'っっこいっっ'],
+            [['っっっこいっっっ', false], 'っっっこいっっっ'],
+            [['っっっこいっっっ', true], 'っっっこいっっっ'],
+
             [['', false], ''],
             [['', true], ''],
             [['っ', false], 'っ'],
             [['っ', true], 'っ'],
             [['っっ', false], 'っっ'],
-            [['っっ', true], 'っ'],
-            [['っっっ', false], 'っっ'],
-            [['っっっ', true], 'っ'],
+            [['っっ', true], 'っっ'],
+            [['っっっ', false], 'っっっ'],
+            [['っっっ', true], 'っっっ'],
+
+            [['っーッかっこいいっーッ', false], 'っーッかっこいいっーッ'],
+            [['っーッかっこいいっーッ', true], 'っーッかこいいっーッ'],
+            [['っっーーッッかっこいいっっーーッッ', false], 'っっーーッッかっこいいっっーーッッ'],
+            [['っっーーッッかっこいいっっーーッッ', true], 'っっーーッッかこいいっっーーッッ'],
 
-            [['かっこいいっ', false], 'かっこいいっ'],
-            [['かっこいいっ', true], 'かこいいっ'],
-            [['かっこいいっっ', false], 'かっこいいっっ'],
-            [['かっこいいっっ', true], 'かこいいっ'],
-            [['かっこいいっっっ', false], 'かっこいいっっ'],
-            [['かっこいいっっっ', true], 'かこいいっ'],
+            [['っーッ', false], 'っーッ'],
+            [['っーッ', true], 'っーッ'],
+            [['っっーーッッ', false], 'っっーーッッ'],
+            [['っっーーッッ', true], 'っっーーッッ'],
         ];
 
         test.each(data)('%o -> %o', (input, output) => {