From b216ed99b4d13cffa00c040f0c096d014ffe8997 Mon Sep 17 00:00:00 2001
From: Alexander Udalov <Alexander.Udalov@jetbrains.com>
Date: Tue, 19 Jan 2016 03:34:08 +0300
Subject: [PATCH] Simplify UTF-8 decoding, just trivially convert bytes to
 16-bit Unicode

ASM will do the rest (see ByteVector#encodeUTF8)
---
 .../kotlin/serialization/jvm/utfEncoding.kt   | 44 +++----------------
 1 file changed, 5 insertions(+), 39 deletions(-)

diff --git a/core/descriptor.loader.java/src/org/jetbrains/kotlin/serialization/jvm/utfEncoding.kt b/core/descriptor.loader.java/src/org/jetbrains/kotlin/serialization/jvm/utfEncoding.kt
index 90c7b4f3496..1df2759d292 100644
--- a/core/descriptor.loader.java/src/org/jetbrains/kotlin/serialization/jvm/utfEncoding.kt
+++ b/core/descriptor.loader.java/src/org/jetbrains/kotlin/serialization/jvm/utfEncoding.kt
@@ -23,14 +23,6 @@ const val MAX_UTF8_INFO_LENGTH = 65535
 
 const val UTF8_MODE_MARKER = 0.toChar()
 
-// Leading bytes are prefixed with 110 in UTF-8
-private val LEADING_BYTE_MASK = 0b11000000
-// Continuation bytes are prefixed with 10 in UTF-8
-private val CONTINUATION_BYTE_MASK = 0b10000000
-
-private val TWO_LOWER_BITS_MASK = 0b00000011
-private val SIX_LOWER_BITS_MASK = 0b00111111
-
 fun bytesToStrings(bytes: ByteArray): Array<String> {
     val result = ArrayList<String>(1)
     val buffer = StringBuilder()
@@ -41,30 +33,16 @@ fun bytesToStrings(bytes: ByteArray): Array<String> {
     bytesInBuffer += 2
 
     for (b in bytes) {
-        if (b >= 0) {
-            buffer.append(b.toChar())
+        val c = b.toInt() and 0xFF // 0 <= c <= 255
+        buffer.append(c.toChar())
+        if (0 < b && b <= 127) {
             bytesInBuffer++
-            // Zeros occupy two bytes
-            if (b == 0.toByte()) bytesInBuffer++
         }
         else {
-            val int = b.toInt() and 0xFF
-            val leadingByte = LEADING_BYTE_MASK or (int shr 6)
-            val continuationByte = CONTINUATION_BYTE_MASK or (int and SIX_LOWER_BITS_MASK)
-            val encodedByte = (leadingByte shl 8) or continuationByte
-
-            buffer.append(encodedByte.toChar())
             bytesInBuffer += 2
-
-            if (bytesInBuffer > MAX_UTF8_INFO_LENGTH) {
-                result.add(buffer.substring(0, buffer.length - 1))
-                buffer.setLength(0)
-                buffer.append(encodedByte.toChar())
-                bytesInBuffer = 2
-            }
         }
 
-        if (bytesInBuffer == MAX_UTF8_INFO_LENGTH) {
+        if (bytesInBuffer >= MAX_UTF8_INFO_LENGTH - 1) {
             result.add(buffer.toString())
             buffer.setLength(0)
             bytesInBuffer = 0
@@ -85,19 +63,7 @@ fun stringsToBytes(strings: Array<String>): ByteArray {
     var i = 0
     for (s in strings) {
         for (si in 0..s.length - 1) {
-            val c = s[si]
-
-            val int = c.toInt()
-            if (int <= 127) {
-                result[i++] = c.toByte()
-            }
-            else {
-                val leadingByte = (int and 0xFFFF) shr 8
-                val continuationByte = int and 0xFF
-                val higherBits = (leadingByte and TWO_LOWER_BITS_MASK) shl 6
-                val lowerBits = continuationByte and SIX_LOWER_BITS_MASK
-                result[i++] = (higherBits or lowerBits).toByte()
-            }
+            result[i++] = s[si].toByte()
         }
     }