Added concat_ws (#1385)

Author: norberttech

src/core/etl/src/Flow/ETL/DSL/functions.php

@@ -35,6 +35,7 @@
     CollectUnique,
     Combine,
     Concat,
+    ConcatWithSeparator,
     Count,
     DateTimeFormat,
     DenseRank,
@@ -832,12 +833,24 @@ function combine(ScalarFunction|array $keys, ScalarFunction|array $values) : Com
     return new Combine($keys, $values);
 }
 
+/**
+ * Concat all values. If you want to concatenate values with separator use concat_ws function.
+ */
 #[DocumentationDSL(module: Module::CORE, type: DSLType::SCALAR_FUNCTION)]
 function concat(ScalarFunction|string ...$functions) : Concat
 {
     return new Concat(...$functions);
 }
 
+/**
+ * Concat all values with separator.
+ */
+#[DocumentationDSL(module: Module::CORE, type: DSLType::SCALAR_FUNCTION)]
+function concat_ws(ScalarFunction|string ...$functions) : ConcatWithSeparator
+{
+    return new ConcatWithSeparator(...$functions);
+}
+
 #[DocumentationDSL(module: Module::CORE, type: DSLType::SCALAR_FUNCTION)]
 function hash(mixed $value, Algorithm $algorithm = new NativePHPHash()) : Hash
 {

src/core/etl/src/Flow/ETL/Function/Concat.php

@@ -25,13 +25,19 @@ public function eval(Row $row) : mixed
     {
         $values = \array_map(fn (ScalarFunction|string $string) : mixed => \is_string($string) ? $string : Caster::default()->to(type_string(true))->value($string->eval($row)), $this->refs);
 
+        $concatValues = [];
+
         foreach ($values as $value) {
-            if (!\is_string($value)) {
-                return null;
+            if (\is_string($value)) {
+                $concatValues[] = $value;
             }
         }
 
+        if (\count($concatValues) === 0) {
+            return '';
+        }
+
         /** @var array<string> $values */
-        return \implode('', $values);
+        return \implode('', $concatValues);
     }
 }

src/core/etl/src/Flow/ETL/Function/ConcatWithSeparator.php

@@ -0,0 +1,50 @@
+<?php
+
+declare(strict_types=1);
+
+namespace Flow\ETL\Function;
+
+use function Flow\ETL\DSL\type_string;
+use Flow\ETL\PHP\Type\Caster;
+use Flow\ETL\Row;
+
+final class ConcatWithSeparator extends ScalarFunctionChain
+{
+    /**
+     * @var array<ScalarFunction|string>
+     */
+    private readonly array $refs;
+
+    public function __construct(
+        private readonly ScalarFunction|string $separator,
+        ScalarFunction|string ...$refs,
+    ) {
+        $this->refs = $refs;
+    }
+
+    public function eval(Row $row) : mixed
+    {
+        $separator = (new Parameter($this->separator))->asString($row);
+
+        if (!\is_string($separator)) {
+            return '';
+        }
+
+        $values = \array_map(fn (ScalarFunction|string $string) : mixed => \is_string($string) ? $string : Caster::default()->to(type_string(true))->value($string->eval($row)), $this->refs);
+
+        $concatValues = [];
+
+        foreach ($values as $value) {
+            if (\is_string($value)) {
+                $concatValues[] = $value;
+            }
+        }
+
+        if (\count($concatValues) === 0) {
+            return '';
+        }
+
+        /** @var array<string> $values */
+        return \implode($separator, $concatValues);
+    }
+}

src/core/etl/src/Flow/ETL/Function/ScalarFunctionChain.php

@@ -103,6 +103,11 @@ public function concat(ScalarFunction|string ...$params) : self
         return new Concat($this, ...$params);
     }
 
+    public function concatWithSeparator(ScalarFunction|string $separator, ScalarFunction|string ...$params) : self
+    {
+        return new ConcatWithSeparator($separator, ...$params);
+    }
+
     public function contains(ScalarFunction|string $needle) : self
     {
         return new Contains($this, $needle);

src/core/etl/tests/Flow/ETL/Tests/Integration/Function/ConcatTest.php

@@ -28,8 +28,33 @@ public function test_concat_on_non_string_value() : void
 
         self::assertSame(
             [
-                ['id' => 1, 'concat' => null],
-                ['id' => 2, 'concat' => null],
+                ['id' => 1, 'concat' => '1'],
+                ['id' => 2, 'concat' => '2'],
+            ],
+            $memory->dump()
+        );
+    }
+
+    public function test_concat_on_nulls() : void
+    {
+        (data_frame())
+            ->read(
+                from_array(
+                    [
+                        ['id' => 1, 'array' => ['field' => 'value']],
+                        ['id' => 2],
+                    ]
+                )
+            )
+            ->withEntry('concat', concat(lit(null), lit(null)))
+            ->drop('array')
+            ->write(to_memory($memory = new ArrayMemory()))
+            ->run();
+
+        self::assertSame(
+            [
+                ['id' => 1, 'concat' => ''],
+                ['id' => 2, 'concat' => ''],
             ],
             $memory->dump()
         );
@@ -54,7 +79,7 @@ public function test_concat_on_stringable_value() : void
         self::assertSame(
             [
                 ['id' => 1, 'concat' => '1-value'],
-                ['id' => 2, 'concat' => null],
+                ['id' => 2, 'concat' => '2-'],
             ],
             $memory->dump()
         );

src/core/etl/tests/Flow/ETL/Tests/Integration/Function/ConcatWithSeparatorTest.php

@@ -0,0 +1,87 @@
+<?php
+
+declare(strict_types=1);
+
+namespace Flow\ETL\Tests\Integration\Function;
+
+use function Flow\ETL\DSL\data_frame;
+use function Flow\ETL\DSL\{concat_ws, from_array, lit, ref, to_memory};
+use Flow\ETL\Memory\ArrayMemory;
+use Flow\ETL\Tests\FlowTestCase;
+
+final class ConcatWithSeparatorTest extends FlowTestCase
+{
+    public function test_concat_on_non_string_value() : void
+    {
+        (data_frame())
+            ->read(
+                from_array(
+                    [
+                        ['id' => 1],
+                        ['id' => 2],
+                    ]
+                )
+            )
+            ->withEntry('concat', concat_ws(lit(','), ref('id'), lit(null)))
+            ->write(to_memory($memory = new ArrayMemory()))
+            ->run();
+
+        self::assertSame(
+            [
+                ['id' => 1, 'concat' => '1'],
+                ['id' => 2, 'concat' => '2'],
+            ],
+            $memory->dump()
+        );
+    }
+
+    public function test_concat_on_nulls() : void
+    {
+        (data_frame())
+            ->read(
+                from_array(
+                    [
+                        ['id' => 1, 'array' => ['field' => 'value']],
+                        ['id' => 2],
+                    ]
+                )
+            )
+            ->withEntry('concat', concat_ws(lit(null), lit(null)))
+            ->drop('array')
+            ->write(to_memory($memory = new ArrayMemory()))
+            ->run();
+
+        self::assertSame(
+            [
+                ['id' => 1, 'concat' => ''],
+                ['id' => 2, 'concat' => ''],
+            ],
+            $memory->dump()
+        );
+    }
+
+    public function test_concat_with_separator() : void
+    {
+        (data_frame())
+            ->read(
+                from_array(
+                    [
+                        ['id' => 1],
+                        ['id' => 2],
+                    ]
+                )
+            )
+            ->withEntry('concat', concat_ws(lit('->'), lit('id'), ref('id')))
+            ->drop('array')
+            ->write(to_memory($memory = new ArrayMemory()))
+            ->run();
+
+        self::assertSame(
+            [
+                ['id' => 1, 'concat' => 'id->1'],
+                ['id' => 2, 'concat' => 'id->2'],
+            ],
+            $memory->dump()
+        );
+    }
+}