toHashSet is suboptimal for inputs with a lot of duplicates #KT-23142

Author: Abduqodiri Qurbonzoda

libraries/stdlib/common/src/generated/_Arrays.kt

@@ -8686,7 +8686,7 @@ public inline fun <K, V, M : MutableMap<in K, in V>> CharArray.associateTo(desti
 @SinceKotlin("1.4")
 @ExperimentalStdlibApi
 public inline fun <K, V> Array<out K>.associateWith(valueSelector: (K) -> V): Map<K, V> {
-    val result = LinkedHashMap<K, V>()
+    val result = LinkedHashMap<K, V>(mapCapacity(size).coerceAtLeast(16))
     return associateWithTo(result, valueSelector)
 }
 
@@ -8704,7 +8704,7 @@ public inline fun <K, V> Array<out K>.associateWith(valueSelector: (K) -> V): Ma
 @ExperimentalStdlibApi
 @kotlin.internal.InlineOnly
 public inline fun <V> ByteArray.associateWith(valueSelector: (Byte) -> V): Map<Byte, V> {
-    val result = LinkedHashMap<Byte, V>()
+    val result = LinkedHashMap<Byte, V>(mapCapacity(size).coerceAtLeast(16))
     return associateWithTo(result, valueSelector)
 }
 
@@ -8722,7 +8722,7 @@ public inline fun <V> ByteArray.associateWith(valueSelector: (Byte) -> V): Map<B
 @ExperimentalStdlibApi
 @kotlin.internal.InlineOnly
 public inline fun <V> ShortArray.associateWith(valueSelector: (Short) -> V): Map<Short, V> {
-    val result = LinkedHashMap<Short, V>()
+    val result = LinkedHashMap<Short, V>(mapCapacity(size).coerceAtLeast(16))
     return associateWithTo(result, valueSelector)
 }
 
@@ -8740,7 +8740,7 @@ public inline fun <V> ShortArray.associateWith(valueSelector: (Short) -> V): Map
 @ExperimentalStdlibApi
 @kotlin.internal.InlineOnly
 public inline fun <V> IntArray.associateWith(valueSelector: (Int) -> V): Map<Int, V> {
-    val result = LinkedHashMap<Int, V>()
+    val result = LinkedHashMap<Int, V>(mapCapacity(size).coerceAtLeast(16))
     return associateWithTo(result, valueSelector)
 }
 
@@ -8758,7 +8758,7 @@ public inline fun <V> IntArray.associateWith(valueSelector: (Int) -> V): Map<Int
 @ExperimentalStdlibApi
 @kotlin.internal.InlineOnly
 public inline fun <V> LongArray.associateWith(valueSelector: (Long) -> V): Map<Long, V> {
-    val result = LinkedHashMap<Long, V>()
+    val result = LinkedHashMap<Long, V>(mapCapacity(size).coerceAtLeast(16))
     return associateWithTo(result, valueSelector)
 }
 
@@ -8776,7 +8776,7 @@ public inline fun <V> LongArray.associateWith(valueSelector: (Long) -> V): Map<L
 @ExperimentalStdlibApi
 @kotlin.internal.InlineOnly
 public inline fun <V> FloatArray.associateWith(valueSelector: (Float) -> V): Map<Float, V> {
-    val result = LinkedHashMap<Float, V>()
+    val result = LinkedHashMap<Float, V>(mapCapacity(size).coerceAtLeast(16))
     return associateWithTo(result, valueSelector)
 }
 
@@ -8794,7 +8794,7 @@ public inline fun <V> FloatArray.associateWith(valueSelector: (Float) -> V): Map
 @ExperimentalStdlibApi
 @kotlin.internal.InlineOnly
 public inline fun <V> DoubleArray.associateWith(valueSelector: (Double) -> V): Map<Double, V> {
-    val result = LinkedHashMap<Double, V>()
+    val result = LinkedHashMap<Double, V>(mapCapacity(size).coerceAtLeast(16))
     return associateWithTo(result, valueSelector)
 }
 
@@ -8812,7 +8812,7 @@ public inline fun <V> DoubleArray.associateWith(valueSelector: (Double) -> V): M
 @ExperimentalStdlibApi
 @kotlin.internal.InlineOnly
 public inline fun <V> BooleanArray.associateWith(valueSelector: (Boolean) -> V): Map<Boolean, V> {
-    val result = LinkedHashMap<Boolean, V>()
+    val result = LinkedHashMap<Boolean, V>(mapCapacity(size).coerceAtLeast(16))
     return associateWithTo(result, valueSelector)
 }
 
@@ -8830,7 +8830,7 @@ public inline fun <V> BooleanArray.associateWith(valueSelector: (Boolean) -> V):
 @ExperimentalStdlibApi
 @kotlin.internal.InlineOnly
 public inline fun <V> CharArray.associateWith(valueSelector: (Char) -> V): Map<Char, V> {
-    val result = LinkedHashMap<Char, V>()
+    val result = LinkedHashMap<Char, V>(mapCapacity(size.coerceAtMost(128)).coerceAtLeast(16))
     return associateWithTo(result, valueSelector)
 }
 
@@ -9145,7 +9145,7 @@ public fun BooleanArray.toHashSet(): HashSet<Boolean> {
  * Returns a new [HashSet] of all elements.
  */
 public fun CharArray.toHashSet(): HashSet<Char> {
-    return toCollection(HashSet<Char>(mapCapacity(size)))
+    return toCollection(HashSet<Char>(mapCapacity(size.coerceAtMost(128))))
 }
 
 /**
@@ -9439,7 +9439,7 @@ public fun CharArray.toSet(): Set<Char> {
     return when (size) {
         0 -> emptySet()
         1 -> setOf(this[0])
-        else -> toCollection(LinkedHashSet<Char>(mapCapacity(size)))
+        else -> toCollection(LinkedHashSet<Char>(mapCapacity(size.coerceAtMost(128))))
     }
 }
 
@@ -11168,9 +11168,7 @@ public infix fun CharArray.subtract(other: Iterable<Char>): Set<Char> {
  * The returned set preserves the element iteration order of the original array.
  */
 public fun <T> Array<out T>.toMutableSet(): MutableSet<T> {
-    val set = LinkedHashSet<T>(mapCapacity(size))
-    for (item in this) set.add(item)
-    return set
+    return toCollection(LinkedHashSet<T>(mapCapacity(size)))
 }
 
 /**
@@ -11179,9 +11177,7 @@ public fun <T> Array<out T>.toMutableSet(): MutableSet<T> {
  * The returned set preserves the element iteration order of the original array.
  */
 public fun ByteArray.toMutableSet(): MutableSet<Byte> {
-    val set = LinkedHashSet<Byte>(mapCapacity(size))
-    for (item in this) set.add(item)
-    return set
+    return toCollection(LinkedHashSet<Byte>(mapCapacity(size)))
 }
 
 /**
@@ -11190,9 +11186,7 @@ public fun ByteArray.toMutableSet(): MutableSet<Byte> {
  * The returned set preserves the element iteration order of the original array.
  */
 public fun ShortArray.toMutableSet(): MutableSet<Short> {
-    val set = LinkedHashSet<Short>(mapCapacity(size))
-    for (item in this) set.add(item)
-    return set
+    return toCollection(LinkedHashSet<Short>(mapCapacity(size)))
 }
 
 /**
@@ -11201,9 +11195,7 @@ public fun ShortArray.toMutableSet(): MutableSet<Short> {
  * The returned set preserves the element iteration order of the original array.
  */
 public fun IntArray.toMutableSet(): MutableSet<Int> {
-    val set = LinkedHashSet<Int>(mapCapacity(size))
-    for (item in this) set.add(item)
-    return set
+    return toCollection(LinkedHashSet<Int>(mapCapacity(size)))
 }
 
 /**
@@ -11212,9 +11204,7 @@ public fun IntArray.toMutableSet(): MutableSet<Int> {
  * The returned set preserves the element iteration order of the original array.
  */
 public fun LongArray.toMutableSet(): MutableSet<Long> {
-    val set = LinkedHashSet<Long>(mapCapacity(size))
-    for (item in this) set.add(item)
-    return set
+    return toCollection(LinkedHashSet<Long>(mapCapacity(size)))
 }
 
 /**
@@ -11223,9 +11213,7 @@ public fun LongArray.toMutableSet(): MutableSet<Long> {
  * The returned set preserves the element iteration order of the original array.
  */
 public fun FloatArray.toMutableSet(): MutableSet<Float> {
-    val set = LinkedHashSet<Float>(mapCapacity(size))
-    for (item in this) set.add(item)
-    return set
+    return toCollection(LinkedHashSet<Float>(mapCapacity(size)))
 }
 
 /**
@@ -11234,9 +11222,7 @@ public fun FloatArray.toMutableSet(): MutableSet<Float> {
  * The returned set preserves the element iteration order of the original array.
  */
 public fun DoubleArray.toMutableSet(): MutableSet<Double> {
-    val set = LinkedHashSet<Double>(mapCapacity(size))
-    for (item in this) set.add(item)
-    return set
+    return toCollection(LinkedHashSet<Double>(mapCapacity(size)))
 }
 
 /**
@@ -11245,9 +11231,7 @@ public fun DoubleArray.toMutableSet(): MutableSet<Double> {
  * The returned set preserves the element iteration order of the original array.
  */
 public fun BooleanArray.toMutableSet(): MutableSet<Boolean> {
-    val set = LinkedHashSet<Boolean>(mapCapacity(size))
-    for (item in this) set.add(item)
-    return set
+    return toCollection(LinkedHashSet<Boolean>(mapCapacity(size)))
 }
 
 /**
@@ -11256,9 +11240,7 @@ public fun BooleanArray.toMutableSet(): MutableSet<Boolean> {
  * The returned set preserves the element iteration order of the original array.
  */
 public fun CharArray.toMutableSet(): MutableSet<Char> {
-    val set = LinkedHashSet<Char>(mapCapacity(size))
-    for (item in this) set.add(item)
-    return set
+    return toCollection(LinkedHashSet<Char>(mapCapacity(size.coerceAtMost(128))))
 }
 
 /**

libraries/stdlib/common/src/generated/_Strings.kt

@@ -708,7 +708,7 @@ public inline fun <K, V, M : MutableMap<in K, in V>> CharSequence.associateTo(de
  */
 @SinceKotlin("1.3")
 public inline fun <V> CharSequence.associateWith(valueSelector: (Char) -> V): Map<Char, V> {
-    val result = LinkedHashMap<Char, V>(mapCapacity(length).coerceAtLeast(16))
+    val result = LinkedHashMap<Char, V>(mapCapacity(length.coerceAtMost(128)).coerceAtLeast(16))
     return associateWithTo(result, valueSelector)
 }
 
@@ -742,7 +742,7 @@ public fun <C : MutableCollection<in Char>> CharSequence.toCollection(destinatio
  * Returns a new [HashSet] of all characters.
  */
 public fun CharSequence.toHashSet(): HashSet<Char> {
-    return toCollection(HashSet<Char>(mapCapacity(length)))
+    return toCollection(HashSet<Char>(mapCapacity(length.coerceAtMost(128))))
 }
 
 /**
@@ -772,7 +772,7 @@ public fun CharSequence.toSet(): Set<Char> {
     return when (length) {
         0 -> emptySet()
         1 -> setOf(this[0])
-        else -> toCollection(LinkedHashSet<Char>(mapCapacity(length)))
+        else -> toCollection(LinkedHashSet<Char>(mapCapacity(length.coerceAtMost(128))))
     }
 }
 

libraries/stdlib/common/src/generated/_UArrays.kt

@@ -4101,7 +4101,7 @@ public inline fun ShortArray.toUShortArray(): UShortArray {
 @ExperimentalUnsignedTypes
 @kotlin.internal.InlineOnly
 public inline fun <V> UIntArray.associateWith(valueSelector: (UInt) -> V): Map<UInt, V> {
-    val result = LinkedHashMap<UInt, V>()
+    val result = LinkedHashMap<UInt, V>(mapCapacity(size).coerceAtLeast(16))
     return associateWithTo(result, valueSelector)
 }
 
@@ -4120,7 +4120,7 @@ public inline fun <V> UIntArray.associateWith(valueSelector: (UInt) -> V): Map<U
 @ExperimentalUnsignedTypes
 @kotlin.internal.InlineOnly
 public inline fun <V> ULongArray.associateWith(valueSelector: (ULong) -> V): Map<ULong, V> {
-    val result = LinkedHashMap<ULong, V>()
+    val result = LinkedHashMap<ULong, V>(mapCapacity(size).coerceAtLeast(16))
     return associateWithTo(result, valueSelector)
 }
 
@@ -4139,7 +4139,7 @@ public inline fun <V> ULongArray.associateWith(valueSelector: (ULong) -> V): Map
 @ExperimentalUnsignedTypes
 @kotlin.internal.InlineOnly
 public inline fun <V> UByteArray.associateWith(valueSelector: (UByte) -> V): Map<UByte, V> {
-    val result = LinkedHashMap<UByte, V>()
+    val result = LinkedHashMap<UByte, V>(mapCapacity(size).coerceAtLeast(16))
     return associateWithTo(result, valueSelector)
 }
 
@@ -4158,7 +4158,7 @@ public inline fun <V> UByteArray.associateWith(valueSelector: (UByte) -> V): Map
 @ExperimentalUnsignedTypes
 @kotlin.internal.InlineOnly
 public inline fun <V> UShortArray.associateWith(valueSelector: (UShort) -> V): Map<UShort, V> {
-    val result = LinkedHashMap<UShort, V>()
+    val result = LinkedHashMap<UShort, V>(mapCapacity(size).coerceAtLeast(16))
     return associateWithTo(result, valueSelector)
 }
 

libraries/stdlib/test/text/StringTest.kt

@@ -8,6 +8,7 @@ package test.text
 import kotlin.test.*
 import test.*
 import test.collections.behaviors.iteratorBehavior
+import test.collections.behaviors.setBehavior
 import test.collections.compare
 import kotlin.math.sign
 import kotlin.random.Random
@@ -1651,4 +1652,19 @@ ${"    "}
         assertFailsWith<IndexOutOfBoundsException> { "".elementAt(0) }
         assertFailsWith<IndexOutOfBoundsException> { "a c".elementAt(-1) }
     }
+
+    @Test
+    fun toHashSet() {
+        compare(hashSetOf('A', 'B', 'C'), "ACAABBAC".toHashSet()) { setBehavior() }
+
+        buildString {
+            repeat(100) { append('1') }
+            append('2')
+            repeat(100) { append('3') }
+            append('4')
+            repeat(100) { append('5') }
+        }.let {
+            compare(hashSetOf('1', '2', '3', '4', '5'), it.toHashSet()) { setBehavior() }
+        }
+    }
 }

libraries/tools/kotlin-stdlib-gen/src/templates/Sets.kt

@@ -5,10 +5,6 @@
 
 package templates
 
-import templates.DocExtensions.collection
-import templates.DocExtensions.element
-import templates.DocExtensions.mapResult
-import templates.DocExtensions.pluralize
 import templates.Family.*
 import templates.SequenceClass.*
 
@@ -35,11 +31,8 @@ object SetOps : TemplateGroupBase() {
             """
         }
         body(ArraysOfObjects, ArraysOfPrimitives) {
-            """
-            val set = LinkedHashSet<T>(mapCapacity(size))
-            for (item in this) set.add(item)
-            return set
-            """
+            val capacity = "size" + if (primitive == PrimitiveType.Char) ".coerceAtMost(128)" else ""
+            "return toCollection(LinkedHashSet<T>(mapCapacity($capacity)))"
         }
         body(Sequences) {
             """

libraries/tools/kotlin-stdlib-gen/src/templates/Snapshots.kt

@@ -55,20 +55,20 @@ object Snapshots : TemplateGroupBase() {
                     1 -> setOf(if (this is List) this[0] else iterator().next())
                     else -> toCollection(LinkedHashSet<T>(mapCapacity(size)))
                 }
-
             }
             return toCollection(LinkedHashSet<T>()).optimizeReadOnlySet()
             """
         }
         body(Sequences) { "return toCollection(LinkedHashSet<T>()).optimizeReadOnlySet()" }
 
         body(CharSequences, ArraysOfObjects, ArraysOfPrimitives) {
-            val size = if (f == CharSequences) "length" else "size"
+            val size = f.code.size
+            val capacity = if (f == CharSequences || primitive == PrimitiveType.Char) "$size.coerceAtMost(128)" else size
             """
             return when ($size) {
                 0 -> emptySet()
                 1 -> setOf(this[0])
-                else -> toCollection(LinkedHashSet<T>(mapCapacity($size)))
+                else -> toCollection(LinkedHashSet<T>(mapCapacity($capacity)))
             }
             """
         }
@@ -82,8 +82,11 @@ object Snapshots : TemplateGroupBase() {
         returns("HashSet<T>")
         body { "return toCollection(HashSet<T>(mapCapacity(collectionSizeOrDefault(12))))" }
         body(Sequences) { "return toCollection(HashSet<T>())" }
-        body(CharSequences) { "return toCollection(HashSet<T>(mapCapacity(length)))" }
-        body(ArraysOfObjects, ArraysOfPrimitives) { "return toCollection(HashSet<T>(mapCapacity(size)))" }
+        body(CharSequences, ArraysOfObjects, ArraysOfPrimitives) {
+            val size = f.code.size
+            val capacity = if (f == CharSequences || primitive == PrimitiveType.Char) "$size.coerceAtMost(128)" else size
+            "return toCollection(HashSet<T>(mapCapacity($capacity)))"
+        }
     }
 
     val f_toSortedSet = fn("toSortedSet()") {
@@ -467,13 +470,18 @@ object Snapshots : TemplateGroupBase() {
             else -> "samples.collections.Collections.Transformations.associateWith"
         })
         body {
-            val resultMap = when (family) {
-                Iterables -> "LinkedHashMap<K, V>(mapCapacity(collectionSizeOrDefault(10)).coerceAtLeast(16))"
-                CharSequences -> "LinkedHashMap<K, V>(mapCapacity(length).coerceAtLeast(16))"
-                else -> "LinkedHashMap<K, V>()"
+            val capacity = when (family) {
+                Iterables -> "mapCapacity(collectionSizeOrDefault(10)).coerceAtLeast(16)"
+                CharSequences -> "mapCapacity(length.coerceAtMost(128)).coerceAtLeast(16)"
+                ArraysOfObjects, ArraysOfPrimitives, ArraysOfUnsigned -> if (primitive == PrimitiveType.Char) {
+                    "mapCapacity(size.coerceAtMost(128)).coerceAtLeast(16)"
+                } else {
+                    "mapCapacity(size).coerceAtLeast(16)"
+                }
+                else -> ""
             }
             """
-            val result = $resultMap
+            val result = LinkedHashMap<K, V>($capacity)
             return associateWithTo(result, valueSelector)
             """
         }