Merge pull request #7 from cketti/refactor

Reshuffle code

Author: cketti

src/commonImplementation/kotlin/CodePoints.kt

@@ -2,44 +2,54 @@
 
 package de.cketti.codepoints
 
+import de.cketti.codepoints.internal.charCount as commonCharCount
+import de.cketti.codepoints.internal.highSurrogate as commonHighSurrogate
+import de.cketti.codepoints.internal.isBmpCodePoint as commonIsBmpCodePoint
+import de.cketti.codepoints.internal.isSupplementaryCodePoint as commonIsSupplementaryCodePoint
+import de.cketti.codepoints.internal.isSurrogatePair as commonIsSurrogatePair
+import de.cketti.codepoints.internal.isValidCodePoint as commonIsValidCodePoint
+import de.cketti.codepoints.internal.lowSurrogate as commonLowSurrogate
+import de.cketti.codepoints.internal.toChars as commonToChars
+import de.cketti.codepoints.internal.toCodePoint as commonToCodePoint
+
 actual object CodePoints {
-    actual inline fun isValidCodePoint(codePoint: Int): Boolean {
-        return CommonCodePoints.isValidCodePoint(codePoint)
+    actual fun isValidCodePoint(codePoint: Int): Boolean {
+        return commonIsValidCodePoint(codePoint)
     }
 
-    actual inline fun isBmpCodePoint(codePoint: Int): Boolean {
-        return CommonCodePoints.isBmpCodePoint(codePoint)
+    actual fun isBmpCodePoint(codePoint: Int): Boolean {
+        return commonIsBmpCodePoint(codePoint)
     }
 
-    actual inline fun isSupplementaryCodePoint(codePoint: Int): Boolean {
-        return CommonCodePoints.isSupplementaryCodePoint(codePoint)
+    actual fun isSupplementaryCodePoint(codePoint: Int): Boolean {
+        return commonIsSupplementaryCodePoint(codePoint)
     }
 
-    actual inline fun charCount(codePoint: Int): Int {
-        return CommonCodePoints.charCount(codePoint)
+    actual fun charCount(codePoint: Int): Int {
+        return commonCharCount(codePoint)
     }
 
-    actual inline fun isSurrogatePair(highSurrogate: Char, lowSurrogate: Char): Boolean {
-        return CommonCodePoints.isSurrogatePair(highSurrogate, lowSurrogate)
+    actual fun isSurrogatePair(highSurrogate: Char, lowSurrogate: Char): Boolean {
+        return commonIsSurrogatePair(highSurrogate, lowSurrogate)
     }
 
     actual fun highSurrogate(codePoint: Int): Char {
-        return CommonCodePoints.highSurrogate(codePoint)
+        return commonHighSurrogate(codePoint)
     }
 
     actual fun lowSurrogate(codePoint: Int): Char {
-        return CommonCodePoints.lowSurrogate(codePoint)
+        return commonLowSurrogate(codePoint)
     }
 
-    actual inline fun toCodePoint(highSurrogate: Char, lowSurrogate: Char): Int {
-        return CommonCodePoints.toCodePoint(highSurrogate, lowSurrogate)
+    actual fun toCodePoint(highSurrogate: Char, lowSurrogate: Char): Int {
+        return commonToCodePoint(highSurrogate, lowSurrogate)
     }
 
-    actual inline fun toChars(codePoint: Int): CharArray {
-        return CommonCodePoints.toChars(codePoint)
+    actual fun toChars(codePoint: Int): CharArray {
+        return commonToChars(codePoint)
     }
 
-    actual inline fun toChars(codePoint: Int, destination: CharArray, offset: Int): Int {
-        return CommonCodePoints.toChars(codePoint, destination, offset)
+    actual fun toChars(codePoint: Int, destination: CharArray, offset: Int): Int {
+        return commonToChars(codePoint, destination, offset)
     }
 }

src/commonImplementation/kotlin/CommonCodePoints.kt

@@ -1,7 +1,8 @@
 package de.cketti.codepoints
 
 import kotlin.text.StringBuilder
+import de.cketti.codepoints.internal.appendCodePoint as commonAppendCodePoint
 
 actual fun StringBuilder.appendCodePoint(codePoint: Int): StringBuilder = apply {
-    CommonStringBuilderFunctions.appendCodePoint(this, codePoint)
+    commonAppendCodePoint(this, codePoint)
 }

src/commonImplementation/kotlin/CommonStringBuilderFunctions.kt

@@ -2,18 +2,23 @@
 
 package de.cketti.codepoints
 
-actual inline fun String.codePointAt(index: Int): Int {
-    return CommonStringFunctions.codePointAt(this, index)
+import de.cketti.codepoints.internal.codePointAt as commonCodePointAt
+import de.cketti.codepoints.internal.codePointBefore as commonCodePointBefore
+import de.cketti.codepoints.internal.codePointCount as commonCodePointCount
+import de.cketti.codepoints.internal.offsetByCodePoints as commonOffsetByCodePoints
+
+actual fun String.codePointAt(index: Int): Int {
+    return commonCodePointAt(this, index)
 }
 
-actual inline fun String.codePointBefore(index: Int): Int {
-    return CommonStringFunctions.codePointBefore(this, index)
+actual fun String.codePointBefore(index: Int): Int {
+    return commonCodePointBefore(this, index)
 }
 
-actual inline fun String.codePointCount(beginIndex: Int, endIndex: Int): Int {
-    return CommonStringFunctions.codePointCount(this, beginIndex, endIndex)
+actual fun String.codePointCount(beginIndex: Int, endIndex: Int): Int {
+    return commonCodePointCount(this, beginIndex, endIndex)
 }
 
-actual inline fun String.offsetByCodePoints(index: Int, codePointOffset: Int): Int {
-    return CommonStringFunctions.offsetByCodePoints(this, index, codePointOffset)
+actual fun String.offsetByCodePoints(index: Int, codePointOffset: Int): Int {
+    return commonOffsetByCodePoints(this, index, codePointOffset)
 }

src/commonImplementation/kotlin/CommonStringFunctions.kt

@@ -0,0 +1,75 @@
+package de.cketti.codepoints.internal
+
+private const val MIN_SUPPLEMENTARY_CODE_POINT = 0x10000
+private const val MAX_CODE_POINT = 0x10FFFF
+
+private const val MIN_HIGH_SURROGATE = 0xD800
+private const val MIN_LOW_SURROGATE = 0xDC00
+
+private const val SURROGATE_DECODE_OFFSET =
+    MIN_SUPPLEMENTARY_CODE_POINT - (MIN_HIGH_SURROGATE shl 10) - MIN_LOW_SURROGATE
+
+private const val HIGH_SURROGATE_ENCODE_OFFSET =
+    (MIN_HIGH_SURROGATE - (MIN_SUPPLEMENTARY_CODE_POINT ushr 10))
+
+internal fun isValidCodePoint(codePoint: Int): Boolean {
+    return codePoint in 0..MAX_CODE_POINT
+}
+
+internal fun isBmpCodePoint(codePoint: Int): Boolean {
+    return codePoint ushr 16 == 0
+}
+
+internal fun isSupplementaryCodePoint(codePoint: Int): Boolean {
+    return codePoint in MIN_SUPPLEMENTARY_CODE_POINT..MAX_CODE_POINT
+}
+
+internal fun charCount(codePoint: Int): Int {
+    return if (codePoint < MIN_SUPPLEMENTARY_CODE_POINT) 1 else 2
+}
+
+internal fun isSurrogatePair(highSurrogate: Char, lowSurrogate: Char): Boolean {
+    return highSurrogate.isHighSurrogate() && lowSurrogate.isLowSurrogate()
+}
+
+internal fun highSurrogate(codePoint: Int): Char {
+    return ((codePoint ushr 10) + HIGH_SURROGATE_ENCODE_OFFSET).toChar() 
+}
+
+internal fun lowSurrogate(codePoint: Int): Char {
+    return ((codePoint and 0x3FF) + MIN_LOW_SURROGATE).toChar()
+}
+
+internal fun toCodePoint(highSurrogate: Char, lowSurrogate: Char): Int {
+    return (highSurrogate.code shl 10) + lowSurrogate.code + SURROGATE_DECODE_OFFSET
+}
+
+internal fun toChars(codePoint: Int): CharArray {
+    return if (isBmpCodePoint(codePoint)) {
+        charArrayOf(codePoint.toChar())
+    } else {
+        charArrayOf(highSurrogate(codePoint), lowSurrogate(codePoint))
+    }
+}
+
+internal fun toChars(codePoint: Int, destination: CharArray, offset: Int): Int {
+    if (isBmpCodePoint(codePoint)) {
+        destination.setSafe(offset, codePoint.toChar())
+        return 1
+    } else {
+        // When writing the low surrogate succeeds but writing the high surrogate fails (offset = -1), the 
+        // destination will be modified even though the method throws. This feels wrong, but matches the behavior
+        // of the Java stdlib implementation.
+        destination.setSafe(offset + 1, lowSurrogate(codePoint))
+        destination.setSafe(offset, highSurrogate(codePoint))
+        return 2
+    }
+}
+
+private fun CharArray.setSafe(index: Int, value: Char) {
+    if (index !in this.indices) {
+        throw IndexOutOfBoundsException("Size: $size, offset: $index")
+    }
+
+    this[index] = value
+}

src/commonImplementation/kotlin/StringBuilderExtensions.kt

@@ -0,0 +1,10 @@
+package de.cketti.codepoints.internal
+
+internal fun appendCodePoint(builder: StringBuilder, codePoint: Int) {
+    if (isBmpCodePoint(codePoint)) {
+        builder.append(codePoint.toChar())
+    } else {
+        builder.append(highSurrogate(codePoint))
+        builder.append(lowSurrogate(codePoint))
+    }
+}