Merge pull request #20 from cketti/substring_CodePointIterator

Add support for iterating over parts of a string

Author: cketti

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

@@ -8,23 +8,38 @@ import kotlin.jvm.JvmInline
 @JvmInline
 value class CodePointSequence(private val text: String) : Sequence<CodePoint> {
     override fun iterator(): CodePointIterator {
-        return CodePointIterator(text)
+        return text.codePointIterator()
     }
 }
 
 /**
  * Iterator for [CodePoint]s in the given [String].
+ * 
+ * The `startIndex` and `endIndex` parameters are the regular `String` indices, i.e. the number of `Char`s from the 
+ * start of the string.
  */
-class CodePointIterator(private val text: String) : Iterator<CodePoint> {
-    private var index = 0
+class CodePointIterator(
+    private val text: String,
+    startIndex: Int,
+    private val endIndex: Int
+) : Iterator<CodePoint> {
+    private var index = startIndex
 
     override fun hasNext(): Boolean {
-        return index < text.length
+        return index < endIndex
     }
 
     override fun next(): CodePoint {
-        return text.codePointAt(index).also { codePoint ->
-            index += codePoint.charCount
+        return if (index + 1 == endIndex) {
+            text[index].toCodePoint().also { 
+                index++ 
+            }
+        } else if (hasNext()) {
+            text.codePointAt(index).also { codePoint ->
+                index += codePoint.charCount
+            }
+        } else {
+            throw IndexOutOfBoundsException()
         }
     }
 }

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

@@ -45,6 +45,6 @@ fun String.codePointSequence(): CodePointSequence {
 /**
  * Iterator for [CodePoint]s in this string.
  */
-fun String.codePointIterator(): CodePointIterator {
-    return CodePointIterator(this)
+fun String.codePointIterator(startIndex: Int = 0, endIndex: Int = length): CodePointIterator {
+    return CodePointIterator(this, startIndex, endIndex)
 }

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

@@ -29,4 +29,74 @@ class CodePointSequenceTest {
             iterator.next()
         }
     }
+
+    @Test
+    fun codePointIterator_with_start_index() {
+        val iterator = "a\uD83E\uDD95b".codePointIterator(startIndex = 1)
+
+        assertTrue(iterator.hasNext())
+        assertEquals("\uD83E\uDD95".codePointAt(0), iterator.next())
+        assertTrue(iterator.hasNext())
+        assertEquals('b'.toCodePoint(), iterator.next())
+        assertFalse(iterator.hasNext())
+        assertFailsWith<IndexOutOfBoundsException> {
+            iterator.next()
+        }
+    }
+
+    @Test
+    fun codePointIterator_with_end_index() {
+        val iterator = "a\uD83E\uDD95b".codePointIterator(endIndex = 3)
+
+        assertTrue(iterator.hasNext())
+        assertEquals('a'.toCodePoint(), iterator.next())
+        assertTrue(iterator.hasNext())
+        assertEquals("\uD83E\uDD95".codePointAt(0), iterator.next())
+        assertFalse(iterator.hasNext())
+        assertFailsWith<IndexOutOfBoundsException> {
+            iterator.next()
+        }
+    }
+
+    @Test
+    fun codePointIterator_with_start_and_end_index() {
+        val iterator = "ab\uD83E\uDD95c".codePointIterator(startIndex = 1, endIndex = 4)
+
+        assertTrue(iterator.hasNext())
+        assertEquals('b'.toCodePoint(), iterator.next())
+        assertTrue(iterator.hasNext())
+        assertEquals("\uD83E\uDD95".codePointAt(0), iterator.next())
+        assertFalse(iterator.hasNext())
+        assertFailsWith<IndexOutOfBoundsException> {
+            iterator.next()
+        }
+    }
+
+    @Test
+    fun codePointIterator_with_start_index_inside_surrogate_pair() {
+        val iterator = "a\uD83E\uDD95b".codePointIterator(startIndex = 2)
+
+        assertTrue(iterator.hasNext())
+        assertEquals('\uDD95'.toCodePoint(), iterator.next())
+        assertTrue(iterator.hasNext())
+        assertEquals('b'.toCodePoint(), iterator.next())
+        assertFalse(iterator.hasNext())
+        assertFailsWith<IndexOutOfBoundsException> {
+            iterator.next()
+        }
+    }
+    
+    @Test
+    fun codePointIterator_with_end_index_inside_surrogate_pair() {
+        val iterator = "a\uD83E\uDD95b".codePointIterator(endIndex = 2)
+
+        assertTrue(iterator.hasNext())
+        assertEquals('a'.toCodePoint(), iterator.next())
+        assertTrue(iterator.hasNext())
+        assertEquals('\uD83E'.toCodePoint(), iterator.next())
+        assertFalse(iterator.hasNext())
+        assertFailsWith<IndexOutOfBoundsException> {
+            iterator.next()
+        }
+    }
 }