From dcb13b9586a147f89bfd20159a140141473c8488 Mon Sep 17 00:00:00 2001
From: pencil-dog <60233442+pencil-dog@users.noreply.github.com>
Date: Thu, 25 Mar 2021 11:36:08 +0100
Subject: [PATCH] [Dict] add unique_scalar() function to avoid performance
 penalty hit when using Dict\unique() on large scalar arrays (#168)

Co-authored-by: yivi <ivan@yivoff.com>
---
 docs/component/dict.md              |  1 +
 src/Psl/Dict/unique_scalar.php      | 38 ++++++++++++++++++++++++
 src/Psl/Internal/Loader.php         |  1 +
 tests/Psl/Dict/UniqueScalarTest.php | 46 +++++++++++++++++++++++++++++
 4 files changed, 86 insertions(+)
 create mode 100644 src/Psl/Dict/unique_scalar.php
 create mode 100644 tests/Psl/Dict/UniqueScalarTest.php

diff --git a/docs/component/dict.md b/docs/component/dict.md
index de92a8b..9d785ac 100644
--- a/docs/component/dict.md
+++ b/docs/component/dict.md
@@ -49,5 +49,6 @@
 - [take_while](./../../src/Psl/Dict/take_while.php#L26)
 - [unique](./../../src/Psl/Dict/unique.php#L17)
 - [unique_by](./../../src/Psl/Dict/unique_by.php#L23)
+- [unique_scalar](./../../src/Psl/Dict/unique_scalar.php#L21)
 
 
diff --git a/src/Psl/Dict/unique_scalar.php b/src/Psl/Dict/unique_scalar.php
new file mode 100644
index 0000000..12d286e
--- /dev/null
+++ b/src/Psl/Dict/unique_scalar.php
@@ -0,0 +1,38 @@
+<?php
+
+declare(strict_types=1);
+
+namespace Psl\Dict;
+
+use function array_unique;
+use function is_array;
+
+/**
+ * Returns a new dict in which each value appears exactly once. Better performant than `Dict\unique()` when the values
+ * are only scalars.
+ *
+ * @template Tk of array-key
+ * @template Tv of scalar
+ *
+ * @param iterable<Tk, Tv> $iterable
+ *
+ * @return array<Tk, Tv>
+ */
+function unique_scalar(iterable $iterable): array
+{
+    if (is_array($iterable)) {
+        return array_unique($iterable);
+    }
+
+    return unique_by(
+        $iterable,
+        /**
+         * @param scalar $v
+         *
+         * @return scalar
+         *
+         * @pure
+         */
+        static fn($v) => $v
+    );
+}
diff --git a/src/Psl/Internal/Loader.php b/src/Psl/Internal/Loader.php
index f1ac854..4195898 100644
--- a/src/Psl/Internal/Loader.php
+++ b/src/Psl/Internal/Loader.php
@@ -136,6 +136,7 @@ final class Loader
         'Psl\Dict\take_while',
         'Psl\Dict\unique',
         'Psl\Dict\unique_by',
+        'Psl\Dict\unique_scalar',
         'Psl\Dict\diff',
         'Psl\Dict\diff_by_key',
         'Psl\Dict\intersect',
diff --git a/tests/Psl/Dict/UniqueScalarTest.php b/tests/Psl/Dict/UniqueScalarTest.php
new file mode 100644
index 0000000..d019619
--- /dev/null
+++ b/tests/Psl/Dict/UniqueScalarTest.php
@@ -0,0 +1,46 @@
+<?php
+
+declare(strict_types=1);
+
+namespace Psl\Tests\Dict;
+
+use PHPUnit\Framework\TestCase;
+use Psl\Collection;
+use Psl\Dict;
+use Psl\Iter;
+use Psl\Vec;
+
+final class UniqueScalarTest extends TestCase
+{
+
+    public function testUniqueScalars(): void
+    {
+        $array   = Vec\fill(10, 'foo');
+        $array[] = 'bar';
+
+        $unique = Dict\unique_scalar($array);
+
+        static::assertCount(2, $unique);
+        static::assertSame(['foo', 'bar'], Vec\values($unique));
+    }
+
+    public function testUniqueIterator()
+    {
+        $array = Iter\Iterator::create(['foo', 'foo', 'bar', 'bar', 'baz']);
+
+        $unique = Dict\unique_scalar($array);
+
+        static::assertCount(3, $unique);
+        static::assertSame(['foo', 'bar', 'baz'], Vec\values($unique));
+    }
+
+    public function testUniqueIteratorAgggregate()
+    {
+        $array = Collection\Map::fromArray(['foo', 'foo', 'bar', 'bar', 'baz']);
+
+        $unique = Dict\unique_scalar($array);
+
+        static::assertCount(3, $unique);
+        static::assertSame(['foo', 'bar', 'baz'], Vec\values($unique));
+    }
+}