Calculate string encoding size more precisely

compiler/backend/src/org/jetbrains/kotlin/codegen/StringConcatGenerator.kt

@@ -33,6 +33,12 @@ class StringConcatGenerator(val mode: JvmStringConcat, val mv: InstructionAdapte
             fun constant(value: String) = Item(JAVA_STRING_TYPE, ItemType.CONSTANT, value)
             fun parameter(type: Type) = Item(type, ItemType.PARAMETER, "\u0001")
         }
+
+        val encodedUTF8Size by lazy {
+            value.encodedUTF8Size()
+        }
+
+        fun fitEncodingLimit() = encodedUTF8Size <= STRING_UTF8_ENCODING_BYTE_LIMIT
     }
 
     private val items = arrayListOf<Item>()
@@ -183,28 +189,31 @@ class StringConcatGenerator(val mode: JvmStringConcat, val mv: InstructionAdapte
 
     private fun fitRestrictions(items: List<Item>): ArrayList<Item> {
         val result = arrayListOf<Item>()
+        //Split long CONSTANT and INLINED_CONSTANT into smaller strings and convert them into CONSTANT
         items.forEach { item ->
             when (item.itemType) {
-                //split INLINED_CONSTANT becomes split CONSTANT one
+                //split INLINED_CONSTANT becomes split CONSTANT
-                ItemType.CONSTANT, ItemType.INLINED_CONSTANT -> splitStringConstant(item.value).forEach { part ->
+                ItemType.CONSTANT, ItemType.INLINED_CONSTANT ->
-                    result.add(
+                    if (item.fitEncodingLimit()) result.add(item) else splitStringConstant(item.value).forEach { part ->
-                        Item(
+                        result.add(
-                            item.type,
+                            Item(
-                            ItemType.CONSTANT,
+                                item.type,
-                            part
+                                ItemType.CONSTANT,
+                                part
+                            )
                         )
-                    )
+                    }
-                }
                 else -> result.add(item)
             }
         }
 
-        var recipe = buildrecipe(result)
+        //Check restriction for recipe string
-        while (recipe.length >= MAX_CONST_STRING_PART_LIMIT) {
+        var recipe = buildRecipe(result)
-            val item = items.filter { it.itemType == ItemType.INLINED_CONSTANT }.maxByOrNull { it.value.length } ?: break
+        while (recipe.toString().encodedUTF8Size() > STRING_UTF8_ENCODING_BYTE_LIMIT) {
-            //move longest INLINED_CONSTANT to CONSTANT
+            val item = items.filter { it.itemType == ItemType.INLINED_CONSTANT }.maxByOrNull { it.encodedUTF8Size } ?: break
+            //move largest INLINED_CONSTANT to CONSTANT
             item.itemType = ItemType.CONSTANT
-            recipe = buildrecipe(result)
+            recipe = buildRecipe(result)
         }
 
         return result

compiler/backend/src/org/jetbrains/kotlin/codegen/codegenUtil.kt

@@ -680,24 +680,52 @@ internal fun LabelNode.linkWithLabel(): LabelNode {
 fun linkedLabel(): Label = LabelNode().linkWithLabel().label
 
 // Strings in constant pool contain at most 2^16-1 = 65535 bytes.
-// Non-BMP characters in UTF8 require at most 4 bytes.
+const val STRING_UTF8_ENCODING_BYTE_LIMIT: Int = 65535
-const val MAX_CONST_STRING_PART_LIMIT: Int = 65535 / 4
 
 fun splitStringConstant(value: String): List<String> {
     val length = value.length
 
-    return if (length <= MAX_CONST_STRING_PART_LIMIT) {
+    //Each CHAR could be encoded maximum in 3 bytes
+    return if (length <= STRING_UTF8_ENCODING_BYTE_LIMIT / 3) {
         listOf(value)
     } else {
+        val result = arrayListOf<String>()
+
         // Split strings into parts, each of which satisfies JVM class file constant pool constraints.
         // Note that even if we split surrogate pairs between parts, they will be joined on concatenation.
-        var i = 0
+        var accumulatedSize = 0
-        val result = arrayListOf<String>()
+        var charOffsetInString = 0
-        while (i < length) {
+        var lastStringBeginning = 0
-            val j = minOf(i + MAX_CONST_STRING_PART_LIMIT, length)
+        while (charOffsetInString < length) {
-            result.add(value.substring(i, j))
+            val charCode = value[charOffsetInString].code
-            i += MAX_CONST_STRING_PART_LIMIT
+            val encodedCharSize = when {
+                charCode in 1..127 -> 1
+                charCode <= 2047 -> 2
+                else -> 3
+            }
+            if (accumulatedSize + encodedCharSize > STRING_UTF8_ENCODING_BYTE_LIMIT) {
+                result.add(value.substring(lastStringBeginning, charOffsetInString))
+                lastStringBeginning = charOffsetInString
+                accumulatedSize = 0
+            }
+            accumulatedSize += encodedCharSize
+            ++charOffsetInString
         }
+        result.add(value.substring(lastStringBeginning, charOffsetInString))
+
         result
     }
 }
+
+fun String.encodedUTF8Size(): Int {
+    var result = 0
+    for (char in this) {
+        val charCode = char.code
+        when {
+            charCode in 1..127 -> result++
+            charCode <= 2047 -> result += 2
+            else -> result += 3
+        }
+    }
+    return result
+}