Add extensions functions to iterate over codepoints (#38)

Co-authored-by: cketti <[email protected]>

Author: OptimumCode

kotlin-codepoints-deluxe/src/commonMain/kotlin/CharSequenceExtensions.kt

@@ -6,6 +6,8 @@ package de.cketti.codepoints.deluxe
 
 import de.cketti.codepoints.codePointAt as intCodePointAt
 import de.cketti.codepoints.codePointBefore as intCodePointBefore
+import de.cketti.codepoints.forEachCodePoint as intForEachCodePoint
+import de.cketti.codepoints.forEachCodePointIndexed as intForEachCodePointIndexed
 
 /**
  * Returns the Unicode code point at the specified index.
@@ -51,3 +53,24 @@ fun CharSequence.codePointSequence(): CodePointSequence {
 fun CharSequence.codePointIterator(startIndex: Int = 0, endIndex: Int = length): CodePointIterator {
     return CodePointIterator(this, startIndex, endIndex)
 }
+
+/**
+ * Performs the given [action] for each code point in this character sequence.
+ */
+inline fun CharSequence.forEachCodePoint(action: (codePoint: CodePoint) -> Unit) {
+    intForEachCodePoint { codePoint ->
+        action(codePoint.toCodePoint())
+    }
+}
+
+/**
+ * Performs the given [action] for each code point in this character sequence.
+ *
+ * @param action The start index of the current code point is provided as the first argument to this function. The
+ *   code point as [CodePoint] instance as the second argument.
+ */
+inline fun CharSequence.forEachCodePointIndexed(action: (index: Int, codePoint: CodePoint) -> Unit) {
+    intForEachCodePointIndexed { index, codePoint ->
+        action(index, codePoint.toCodePoint())
+    }
+}

kotlin-codepoints-deluxe/src/commonTest/kotlin/CharSequenceExtensionsTest.kt

@@ -57,4 +57,53 @@ class CharSequenceExtensionsTest {
         assertEquals(0xD83E.toCodePoint(), "\uD83E\uDD95\uD83E\uDD96".codePointBefore(1))
         assertEquals(0xD83E.toCodePoint(), "\uD83E\uDD95\uD83E\uDD96".codePointBefore(3))
     }
+
+    @Test
+    fun forEachCodepoint() {
+        fun CharSequence.collectCodepoints(): List<CodePoint> = buildList { forEachCodePoint { add(it) } }
+
+        assertEquals(
+            emptyList(),
+            "".collectCodepoints(),
+        )
+        assertEquals(
+            listOf('a'.toCodePoint()),
+            "a".collectCodepoints(),
+        )
+        assertEquals(
+            listOf('a'.toCodePoint(), 0xFFFF.toCodePoint()),
+            "a\uFFFF".collectCodepoints(),
+        )
+        assertEquals(
+            listOf(0x1F995.toCodePoint(), 'a'.toCodePoint(), 0x1F996.toCodePoint()),
+            "\uD83E\uDD95a\uD83E\uDD96".collectCodepoints(),
+        )
+    }
+
+    @Test
+    fun forEachCodepointIndexed() {
+        fun CharSequence.collectCodepoints(): List<Pair<Int, CodePoint>> =
+            buildList { forEachCodePointIndexed { index, codepoint -> add(index to codepoint) } }
+
+        assertEquals(
+            emptyList(),
+            "".collectCodepoints(),
+        )
+        assertEquals(
+            listOf(0 to 'a'.toCodePoint()),
+            "a".collectCodepoints(),
+        )
+        assertEquals(
+            listOf(0 to 'a'.toCodePoint(), 1 to 0x1F995.toCodePoint()),
+            "a\uD83E\uDD95".collectCodepoints(),
+        )
+        assertEquals(
+            listOf(
+                0 to 0x1F995.toCodePoint(),
+                2 to 'a'.toCodePoint(),
+                3 to 0x1F996.toCodePoint(),
+            ),
+            "\uD83E\uDD95a\uD83E\uDD96".collectCodepoints(),
+        )
+    }
 }

kotlin-codepoints/src/commonMain/kotlin/CharSequenceExtensions.kt

@@ -160,3 +160,43 @@ fun CharSequence.offsetByCodePoints(index: Int, codePointOffset: Int): Int {
         return currentIndex + 1
     }
 }
+
+/**
+ * Performs the given [action] for each code point in this character sequence.
+ */
+inline fun CharSequence.forEachCodePoint(action: (codePoint: Int) -> Unit) {
+    forEachCodePointIndexed { _, codePoint ->
+        action(codePoint)
+    }
+}
+
+/**
+ * Performs the given [action] for each code point in this character sequence.
+ *
+ * @param action The start index of the current code point is provided as the first argument to this function. The code
+ *   point value as the second argument.
+ */
+inline fun CharSequence.forEachCodePointIndexed(action: (index: Int, codePoint: Int) -> Unit) {
+    var index = 0
+    val endIndex = length
+    while (index < endIndex) {
+        val codePointStartIndex = index
+
+        val firstChar = this[index]
+        index++
+
+        if (firstChar.isHighSurrogate() && index < endIndex) {
+            val nextChar = this[index]
+            if (nextChar.isLowSurrogate()) {
+                index++
+
+                val codePoint = CodePoints.toCodePoint(firstChar, nextChar)
+                action(codePointStartIndex, codePoint)
+
+                continue
+            }
+        }
+
+        action(codePointStartIndex, firstChar.code)
+    }
+}

kotlin-codepoints/src/commonTest/kotlin/CharSequenceExtensionsTest.kt

@@ -161,4 +161,53 @@ class CharSequenceExtensionsTest {
             "\uD83E\uDD95".offsetByCodePoints(index = 2, codePointOffset = -2)
         }
     }
+
+    @Test
+    fun forEachCodepoint() {
+        fun CharSequence.collectCodepoints(): List<Int> = buildList { forEachCodePoint { add(it) } }
+
+        assertEquals(
+            emptyList(),
+            "".collectCodepoints(),
+        )
+        assertEquals(
+            listOf('a'.code),
+            "a".collectCodepoints(),
+        )
+        assertEquals(
+            listOf('a'.code, 0xFFFF),
+            "a\uFFFF".collectCodepoints(),
+        )
+        assertEquals(
+            listOf(0x1F995, 'a'.code, 0x1F996),
+            "\uD83E\uDD95a\uD83E\uDD96".collectCodepoints(),
+        )
+    }
+
+    @Test
+    fun forEachCodepointIndexed() {
+        fun CharSequence.collectCodepoints(): List<Pair<Int, Int>> =
+            buildList { forEachCodePointIndexed { index, codepoint -> add(index to codepoint) } }
+
+        assertEquals(
+            emptyList(),
+            "".collectCodepoints(),
+        )
+        assertEquals(
+            listOf(0 to 'a'.code),
+            "a".collectCodepoints(),
+        )
+        assertEquals(
+            listOf(0 to 'a'.code, 1 to 0x1F995),
+            "a\uD83E\uDD95".collectCodepoints(),
+        )
+        assertEquals(
+            listOf(
+                0 to 0x1F995,
+                2 to 'a'.code,
+                3 to 0x1F996,
+            ),
+            "\uD83E\uDD95a\uD83E\uDD96".collectCodepoints(),
+        )
+    }
 }