Merge branch '1.x' into integration/symfony-indexof

CHANGELOG.md

@@ -1,6 +1,7 @@
-## [Unreleased] - 2025-02-11
+## [Unreleased] - 2025-02-12
 
 ### Added
+- [#1470](https://github.com/flow-php/flow/pull/1470) - **ByteString IsUTF8 function with Test** - [@f-lapinski](https://github.com/f-lapinski)
 - [#1465](https://github.com/flow-php/flow/pull/1465) - **jsonl writing example** - [@norberttech](https://github.com/norberttech)
 - [#1465](https://github.com/flow-php/flow/pull/1465) - **csv writing example** - [@norberttech](https://github.com/norberttech)
 - [#1465](https://github.com/flow-php/flow/pull/1465) - **xml writing example** - [@norberttech](https://github.com/norberttech)
@@ -155,6 +156,8 @@
 - [#1240](https://github.com/flow-php/flow/pull/1240) - **Update Homebrew TAP formula: flow-php to version: 0.10.0** - [@norberttech](https://github.com/norberttech)
 
 ### Fixed
+- [#1470](https://github.com/flow-php/flow/pull/1470) - **Add Missing Test for return type of function** - [@f-lapinski](https://github.com/f-lapinski)
+- [#1469](https://github.com/flow-php/flow/pull/1469) - **JSONLines Loader would occasionally write a newline to the start of the file.** - [@jmortlock](https://github.com/jmortlock)
 - [#1457](https://github.com/flow-php/flow/pull/1457) - **missing entry types to JSON/CSV entry normalizers** - [@norberttech](https://github.com/norberttech)
 - [d82381](https://github.com/flow-php/flow/commit/d823813af9be12d5cc21d4b8e2a01171ce0e0c90) - **failing example test** - [@norberttech](https://github.com/norberttech)
 - [#1448](https://github.com/flow-php/flow/pull/1448) - **ParquetOutput didn't implement Output interface** - [@radozato](https://github.com/radozato)

src/adapter/etl-adapter-json/src/Flow/ETL/Adapter/JSON/JsonLinesLoader.php

@@ -15,11 +15,6 @@ final class JsonLinesLoader implements Closure, Loader, Loader\FileLoader
 
     private int $flags = JSON_THROW_ON_ERROR;
 
-    /**
-     * @var array<string, int>
-     */
-    private array $writes = [];
-
     public function __construct(private readonly Path $path)
     {
     }
@@ -65,16 +60,7 @@ public function write(Rows $nextRows, array $partitions, FlowContext $context) :
         $streams = $context->streams();
         $normalizer = new RowsNormalizer(new EntryNormalizer($this->dateTimeFormat));
 
-        if (!$streams->isOpen($this->path, $partitions)) {
-            $stream = $streams->writeTo($this->path, $partitions);
-
-            if (!\array_key_exists($stream->path()->path(), $this->writes)) {
-                $this->writes[$stream->path()->path()] = 0;
-            }
-
-        } else {
-            $stream = $streams->writeTo($this->path, $partitions);
-        }
+        $stream = $streams->writeTo($this->path, $partitions);
 
         $this->writeJSON($nextRows, $stream, $normalizer);
     }
@@ -103,11 +89,7 @@ private function writeJSON(Rows $rows, DestinationStream $stream, RowsNormalizer
                 throw new RuntimeException('Failed to encode JSON: ' . $e->getMessage(), 0, $e);
             }
 
-            $json = ($this->writes[$stream->path()->path()] > 0) ? ("\n" . $json) : $json;
-
-            $stream->append($json);
-
-            $this->writes[$stream->path()->path()]++;
+            $stream->append($json . "\n");
         }
     }
 }

src/core/etl/src/Flow/ETL/Function/IsUtf8.php

@@ -0,0 +1,34 @@
+<?php
+
+declare(strict_types=1);
+
+namespace Flow\ETL\Function;
+
+use function Flow\ETL\DSL\{type_boolean};
+use function Symfony\Component\String\{b};
+use Flow\ETL\Function\ScalarFunction\TypedScalarFunction;
+use Flow\ETL\PHP\Type\Type;
+use Flow\ETL\Row;
+
+final class IsUtf8 extends ScalarFunctionChain implements TypedScalarFunction
+{
+    public function __construct(private readonly ScalarFunction|string $string)
+    {
+    }
+
+    public function eval(Row $row) : mixed
+    {
+        $string = (new Parameter($this->string))->asString($row);
+
+        if ($string === null) {
+            return null;
+        }
+
+        return b($string)->isUtf8();
+    }
+
+    public function returns() : Type
+    {
+        return type_boolean();
+    }
+}

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

@@ -321,6 +321,11 @@ public function isType(string|Type ...$types) : self
         return new IsType($this, ...$types);
     }
 
+    public function isUtf8() : IsUtf8
+    {
+        return new IsUtf8($this);
+    }
+
     public function jsonDecode(ScalarFunction|int $flags = JSON_THROW_ON_ERROR) : self
     {
         return new JsonDecode($this, $flags);

src/core/etl/tests/Flow/ETL/Tests/Unit/Function/IsUtf8Test.php

@@ -0,0 +1,39 @@
+<?php
+
+declare(strict_types=1);
+
+namespace Flow\ETL\Tests\Unit\Function;
+
+use function Flow\ETL\DSL\row;
+use function Flow\ETL\DSL\{ref, str_entry, type_boolean};
+use Flow\ETL\Function\IsUtf8;
+use Flow\ETL\PHP\Type\Type;
+use Flow\ETL\Tests\FlowTestCase;
+
+final class IsUtf8Test extends FlowTestCase
+{
+    public function test_is_utf_8() : void
+    {
+        self::assertTrue(
+            ref('str')->isUtf8()->eval(
+                row(str_entry('str', 'Lorem Ipsum'))
+            )
+        );
+
+        self::assertFalse(
+            ref('str')->isUtf8()->eval(
+                row(str_entry('str', "\xc3\x28"))
+            )
+        );
+    }
+
+    public function test_returns_method_returns_string_boolean() : void
+    {
+        $isUtf8Function = new IsUtf8('Lorem Ipsum');
+        $returnType = $isUtf8Function->returns();
+
+        self::assertInstanceOf(Type::class, $returnType);
+
+        self::assertTrue($returnType->isEqual(type_boolean()));
+    }
+}