Unify passing options to Loaders/Extractors (#1207)

flow-php · Sep 5, 2024 · adf86f1 · adf86f1
1 parent b31d6d7
commit adf86f1
Show file tree

Hide file tree

Showing 10 changed files with 20 additions and 163 deletions.
diff --git a/examples/topics/data_source/csv/description.md b/examples/topics/data_source/csv/description.md
@@ -2,7 +2,7 @@ Read data from a csv file.
 
 ```php
 function from_csv(
-    string|Path|array $path,
+    string|Path $path,
     bool $with_header = true,
     bool $empty_to_null = true,
     ?string $delimiter = null,

diff --git a/examples/topics/data_source/json/description.md b/examples/topics/data_source/json/description.md
@@ -2,7 +2,7 @@ Read data from a json file.
 
 ```php
 function from_json(
-    string|Path|array $path,
+    string|Path $path,
     ?string $pointer = null,
     ?Schema $schema = null,
 );

diff --git a/examples/topics/data_source/xml/description.md b/examples/topics/data_source/xml/description.md
@@ -2,7 +2,7 @@ Read data from a json file.
 
 ```php
 function from_xml(
-    string|Path|array $path,
+    string|Path $path,
     string $xml_node_path = ''
 );
 ```

diff --git a/src/adapter/etl-adapter-csv/src/Flow/ETL/Adapter/CSV/functions.php b/src/adapter/etl-adapter-csv/src/Flow/ETL/Adapter/CSV/functions.php
@@ -4,45 +4,25 @@
 
 namespace Flow\ETL\Adapter\CSV;
 
-use function Flow\ETL\DSL\from_all;
 use Flow\ETL\Adapter\CSV\Detector\{Option, Options};
 use Flow\ETL\Row\Schema;
-use Flow\ETL\{Attribute\DocumentationDSL, Attribute\Module, Attribute\Type as DSLType, Extractor, Loader};
+use Flow\ETL\{Attribute\DocumentationDSL, Attribute\Module, Attribute\Type as DSLType, Loader};
 use Flow\Filesystem\{Path, SourceStream};
 
 /**
  * @param int<1, max> $characters_read_in_line
  */
 #[DocumentationDSL(module: Module::CSV, type: DSLType::EXTRACTOR)]
 function from_csv(
-    string|Path|array $path,
+    string|Path $path,
     bool $with_header = true,
     bool $empty_to_null = true,
     ?string $delimiter = null,
     ?string $enclosure = null,
     ?string $escape = null,
     int $characters_read_in_line = 1000,
     ?Schema $schema = null
-) : Extractor {
-    if (\is_array($path)) {
-        $extractors = [];
-
-        foreach ($path as $file_path) {
-            $extractors[] = new CSVExtractor(
-                \is_string($file_path) ? Path::realpath($file_path) : $file_path,
-                $with_header,
-                $empty_to_null,
-                $delimiter,
-                $enclosure,
-                $escape,
-                $characters_read_in_line,
-                $schema
-            );
-        }
-
-        return from_all(...$extractors);
-    }
-
+) : CSVExtractor {
     return new CSVExtractor(
         \is_string($path) ? Path::realpath($path) : $path,
         $with_header,

diff --git a/...adapter/etl-adapter-csv/tests/Flow/ETL/Adapter/CSV/Tests/Integration/CSVExtractorTest.php b/...adapter/etl-adapter-csv/tests/Flow/ETL/Adapter/CSV/Tests/Integration/CSVExtractorTest.php
@@ -99,32 +99,6 @@ public function test_extracting_csv_empty_headers() : void
         );
     }
 
-    public function test_extracting_csv_files_from_directory_recursively() : void
-    {
-        $extractor = from_csv(
-            [
-                Path::realpath(__DIR__ . '/../Fixtures/annual-enterprise-survey-2019-financial-year-provisional-csv.csv'),
-                Path::realpath(__DIR__ . '/../Fixtures/nested/annual-enterprise-survey-2019-financial-year-provisional-csv.csv'),
-            ],
-            false
-        );
-
-        $total = 0;
-
-        /** @var Rows $rows */
-        foreach ($extractor->extract(new FlowContext(Config::default())) as $rows) {
-            $rows->each(function (Row $row) : void {
-                $this->assertSame(
-                    ['e00', 'e01', 'e02', 'e03', 'e04', 'e05', 'e06', 'e07', 'e08', 'e09'],
-                    \array_keys($row->toArray())
-                );
-            });
-            $total += $rows->count();
-        }
-
-        self::assertSame(1998, $total);
-    }
-
     public function test_extracting_csv_files_with_header() : void
     {
         $path = __DIR__ . '/../Fixtures/annual-enterprise-survey-2019-financial-year-provisional-csv.csv';

diff --git a/src/adapter/etl-adapter-json/src/Flow/ETL/Adapter/JSON/functions.php b/src/adapter/etl-adapter-json/src/Flow/ETL/Adapter/JSON/functions.php
@@ -4,36 +4,21 @@
 
 namespace Flow\ETL\Adapter\JSON;
 
-use function Flow\ETL\DSL\from_all;
 use Flow\ETL\Adapter\JSON\JSONMachine\JsonExtractor;
 use Flow\ETL\Row\Schema;
-use Flow\ETL\{Attribute\DocumentationDSL, Attribute\Module, Attribute\Type, Extractor, Loader};
+use Flow\ETL\{Attribute\DocumentationDSL, Attribute\Module, Attribute\Type, Loader};
 use Flow\Filesystem\Path;
 
 /**
- * @param array<Path|string>|Path|string $path - string is internally turned into stream
+ * @param Path|string $path - string is internally turned into stream
  * @param ?string $pointer - if you want to iterate only results of a subtree, use a pointer, read more at https://github.com/halaxa/json-machine#parsing-a-subtree
  */
 #[DocumentationDSL(module: Module::JSON, type: Type::EXTRACTOR)]
 function from_json(
-    string|Path|array $path,
+    string|Path $path,
     ?string $pointer = null,
     ?Schema $schema = null,
-) : Extractor {
-    if (\is_array($path)) {
-        $extractors = [];
-
-        foreach ($path as $file) {
-            $extractors[] = new JsonExtractor(
-                \is_string($file) ? Path::realpath($file) : $file,
-                $pointer,
-                $schema
-            );
-        }
-
-        return from_all(...$extractors);
-    }
-
+) : JsonExtractor {
     return new JsonExtractor(
         \is_string($path) ? Path::realpath($path) : $path,
         $pointer,

diff --git a/src/adapter/etl-adapter-parquet/src/Flow/ETL/Adapter/Parquet/functions.php b/src/adapter/etl-adapter-parquet/src/Flow/ETL/Adapter/Parquet/functions.php
@@ -4,47 +4,23 @@
 
 namespace Flow\ETL\Adapter\Parquet;
 
-use function Flow\ETL\DSL\from_all;
-use Flow\ETL\Exception\InvalidArgumentException;
 use Flow\ETL\Row\Schema;
-use Flow\ETL\{Attribute\DocumentationDSL, Attribute\Module, Attribute\Type as DSLType, Extractor, Loader};
+use Flow\ETL\{Attribute\DocumentationDSL, Attribute\Module, Attribute\Type as DSLType, Loader};
 use Flow\Filesystem\Path;
 use Flow\Parquet\ParquetFile\Compressions;
 use Flow\Parquet\{ByteOrder, Options};
 
 /**
- * @param array<Path>|Path|string $path
  * @param array<string> $columns
- *
- * @return Extractor
  */
 #[DocumentationDSL(module: Module::PARQUET, type: DSLType::EXTRACTOR)]
 function from_parquet(
-    string|Path|array $path,
+    string|Path $path,
     array $columns = [],
     Options $options = new Options(),
     ByteOrder $byte_order = ByteOrder::LITTLE_ENDIAN,
     ?int $offset = null,
-) : Extractor {
-    if (\is_array($path)) {
-        $extractors = [];
-
-        if ($offset !== null) {
-            throw new InvalidArgumentException('Offset can be used only with single file path, not with pattern');
-        }
-
-        foreach ($path as $filePath) {
-            $extractors[] = new ParquetExtractor(
-                $filePath,
-                $options,
-                $byte_order,
-                $columns
-            );
-        }
-
-        return from_all(...$extractors);
-    }
-
+) : ParquetExtractor {
     return new ParquetExtractor(
         \is_string($path) ? Path::realpath($path) : $path,
         $options,

diff --git a/src/adapter/etl-adapter-text/src/Flow/ETL/Adapter/Text/functions.php b/src/adapter/etl-adapter-text/src/Flow/ETL/Adapter/Text/functions.php
@@ -4,30 +4,13 @@
 
 namespace Flow\ETL\Adapter\Text;
 
-use Flow\ETL\{Attribute\DocumentationDSL, Attribute\Module, Attribute\Type, Extractor, Loader};
+use Flow\ETL\{Attribute\DocumentationDSL, Attribute\Module, Attribute\Type, Loader};
 use Flow\Filesystem\Path;
 
-/**
- * @param array<Path|string>|Path|string $path
- *
- * @return Extractor
- */
 #[DocumentationDSL(module: Module::TEXT, type: Type::EXTRACTOR)]
 function from_text(
-    string|Path|array $path,
-) : Extractor {
-    if (\is_array($path)) {
-        $extractors = [];
-
-        foreach ($path as $file_path) {
-            $extractors[] = new TextExtractor(
-                \is_string($file_path) ? Path::realpath($file_path) : $file_path,
-            );
-        }
-
-        return new Extractor\ChainExtractor(...$extractors);
-    }
-
+    string|Path $path,
+) : TextExtractor {
     return new TextExtractor(
         \is_string($path) ? Path::realpath($path) : $path,
     );

diff --git a/...pter/etl-adapter-text/tests/Flow/ETL/Adapter/Text/Tests/Integration/TextExtractorTest.php b/...pter/etl-adapter-text/tests/Flow/ETL/Adapter/Text/Tests/Integration/TextExtractorTest.php
@@ -7,7 +7,7 @@
 use function Flow\ETL\Adapter\Text\{from_text};
 use Flow\ETL\Adapter\Text\TextExtractor;
 use Flow\ETL\Extractor\Signal;
-use Flow\ETL\{Config, Flow, FlowContext, Row, Rows};
+use Flow\ETL\{Config, Flow, FlowContext, Row};
 use Flow\Filesystem\Path;
 use PHPUnit\Framework\TestCase;
 
@@ -28,28 +28,6 @@ public function test_extracting_text_file() : void
         self::assertSame(1024, $rows->count());
     }
 
-    public function test_extracting_text_files_from_directory() : void
-    {
-        $extractor = from_text(
-            [
-                __DIR__ . '/../Fixtures/annual-enterprise-survey-2019-financial-year-provisional-csv.csv',
-                __DIR__ . '/../Fixtures/nested/annual-enterprise-survey-2019-financial-year-provisional-csv.csv',
-            ],
-        );
-
-        $total = 0;
-
-        /** @var Rows $rows */
-        foreach ($extractor->extract(new FlowContext(Config::default())) as $rows) {
-            $rows->each(function (Row $row) : void {
-                $this->assertInstanceOf(Row\Entry\StringEntry::class, $row->get('text'));
-            });
-            $total += $rows->count();
-        }
-
-        self::assertSame(2048, $total);
-    }
-
     public function test_limit() : void
     {
         $extractor = new TextExtractor(Path::realpath(__DIR__ . '/../Fixtures/orders_flow.csv'));

diff --git a/src/adapter/etl-adapter-xml/src/Flow/ETL/Adapter/XML/functions.php b/src/adapter/etl-adapter-xml/src/Flow/ETL/Adapter/XML/functions.php
@@ -4,37 +4,18 @@
 
 namespace Flow\ETL\Adapter\XML;
 
-use function Flow\ETL\DSL\from_all;
 use Flow\ETL\{Adapter\XML\Loader\XMLLoader,
     Adapter\XML\XMLWriter\DOMDocumentWriter,
     Attribute\DocumentationDSL,
     Attribute\Module,
-    Attribute\Type as DSLType,
-    Extractor};
+    Attribute\Type as DSLType};
 use Flow\Filesystem\Path;
 
-/**
- * @param array<Path|string>|Path|string $path
- */
 #[DocumentationDSL(module: Module::XML, type: DSLType::EXTRACTOR)]
 function from_xml(
-    string|Path|array $path,
+    string|Path $path,
     string $xml_node_path = ''
-) : Extractor {
-    if (\is_array($path)) {
-        /** @var array<Extractor> $extractors */
-        $extractors = [];
-
-        foreach ($path as $next_path) {
-            $extractors[] = new XMLParserExtractor(
-                \is_string($next_path) ? Path::realpath($next_path) : $next_path,
-                $xml_node_path
-            );
-        }
-
-        return from_all(...$extractors);
-    }
-
+) : XMLParserExtractor {
     return new XMLParserExtractor(
         \is_string($path) ? Path::realpath($path) : $path,
         $xml_node_path