refactor(i18n): 코드포인트 기반 문자열 분할을 공통화한다

src/main/kotlin/kr/co/vividnext/sodalive/i18n/translation/CodePointTextSplitter.kt

@@ -0,0 +1,41 @@
+package kr.co.vividnext.sodalive.i18n.translation
+
+internal fun splitTextByCodePoint(text: String, maxSize: Int, codePointSize: (Int) -> Int): List<String> {
+    val chunks = mutableListOf<String>()
+    var start = 0
+    do {
+        val end = findChunkEnd(text, start, maxSize, codePointSize)
+        chunks.add(text.substring(start, end))
+        start = end
+    } while (start < text.length)
+    return chunks
+}
+
+internal fun takeTextPrefixByCodePoint(text: String, maxSize: Int, codePointSize: (Int) -> Int): String {
+    return text.substring(0, findChunkEnd(text, 0, maxSize, codePointSize))
+}
+
+internal fun utf8CodePointSize(codePoint: Int): Int {
+    return when {
+        codePoint <= 0x7F -> 1
+        codePoint <= 0x7FF -> 2
+        codePoint in 0xD800..0xDFFF -> 1
+        codePoint <= 0xFFFF -> 3
+        else -> 4
+    }
+}
+
+private fun findChunkEnd(text: String, start: Int, maxSize: Int, codePointSize: (Int) -> Int): Int {
+    var end = start
+    var size = 0
+    while (end < text.length) {
+        val codePoint = text.codePointAt(end)
+        val nextSize = codePointSize(codePoint)
+        if (size + nextSize > maxSize) {
+            break
+        }
+        size += nextSize
+        end += Character.charCount(codePoint)
+    }
+    return end
+}

src/main/kotlin/kr/co/vividnext/sodalive/i18n/translation/PapagoTranslationService.kt

@@ -78,17 +78,7 @@ class PapagoTranslationService(
     }
 
     private fun splitText(text: String): List<String> {
-        val chunks = mutableListOf<String>()
-        var start = 0
-        do {
-            var end = minOf(start + MAX_REQUEST_TEXT_LENGTH, text.length)
-            if (end < text.length && text[end - 1].isHighSurrogate() && text[end].isLowSurrogate()) {
-                end -= 1
-            }
-            chunks.add(text.substring(start, end))
-            start = end
-        } while (start < text.length)
-        return chunks
+        return splitTextByCodePoint(text, MAX_REQUEST_TEXT_LENGTH, Character::charCount)
     }
 
     private fun validateLanguages(sourceLanguage: String, targetLanguage: String): Boolean {