Allow to read only specific columns through CLI read command (#1787)

Allow to read only specific columns throug CLI read command

Author: norberttech

src/cli/src/Flow/CLI/Command/FileReadCommand.php

@@ -4,7 +4,7 @@
 
 namespace Flow\CLI\Command;
 
-use function Flow\CLI\{option_bool, option_int, option_int_nullable};
+use function Flow\CLI\{option_bool, option_int, option_int_nullable, option_list_of_strings};
 use function Flow\ETL\DSL\{df};
 use Flow\CLI\Arguments\{FilePathArgument};
 use Flow\CLI\Command\Traits\{CSVOptions, ConfigOptions, ExcelOptions, JSONOptions, ParquetOptions, XMLOptions};
@@ -45,6 +45,7 @@ public function configure() : void
             ->addOption('input-file-batch-size', null, InputOption::VALUE_REQUIRED, 'Number of rows that are going to be read and displayed in one batch, when set to -1 whole dataset will be displayed at once', self::DEFAULT_BATCH_SIZE)
             ->addOption('input-file-limit', null, InputOption::VALUE_REQUIRED, 'Limit number of rows that are going to be used to infer file schema, when not set whole file is analyzed', null)
             ->addOption('output-truncate', null, InputOption::VALUE_REQUIRED, 'Truncate output to given number of characters, when set to -1 output is not truncated at all', 20)
+            ->addOption('output-columns', null, InputOption::VALUE_REQUIRED | InputOption::VALUE_IS_ARRAY, 'Columns to include in output, when not set all columns are displayed', [])
             ->addOption('schema-auto-cast', null, InputOption::VALUE_OPTIONAL, 'When set Flow will try to automatically cast values to more precise data types, for example datetime strings will be casted to datetime type', false);
 
         $this->addConfigOptions($this);
@@ -82,6 +83,12 @@ protected function execute(InputInterface $input, OutputInterface $output) : int
             $df->limit($limit);
         }
 
+        $outputColumns = option_list_of_strings('output-columns', $input);
+
+        if (\count($outputColumns)) {
+            $df->select(...$outputColumns);
+        }
+
         $formatter = new AsciiTableFormatter();
 
         $df->run(function (Rows $rows) use ($style, $formatter, $outputTruncate) : void {

src/cli/tests/Flow/CLI/Tests/Integration/FileReadCommandTest.php

@@ -137,6 +137,109 @@ public function test_read_rows_text() : void
         );
     }
 
+    public function test_read_rows_with_output_columns_empty_maintains_all_columns() : void
+    {
+        $tester = new CommandTester(new FileReadCommand('read'));
+
+        $tester->execute([
+            'input-file' => __DIR__ . '/Fixtures/orders.csv',
+            '--input-file-limit' => 2,
+        ]);
+
+        $tester->assertCommandIsSuccessful();
+
+        // Should contain all original columns (order_id, created_at, updated_at, discount, address, notes, items)
+        $output = $tester->getDisplay();
+        self::assertStringContainsString('order_id', $output);
+        self::assertStringContainsString('created_at', $output);
+        self::assertStringContainsString('updated_at', $output);
+        self::assertStringContainsString('discount', $output);
+        self::assertStringContainsString('address', $output);
+        self::assertStringContainsString('notes', $output);
+        self::assertStringContainsString('items', $output);
+        self::assertStringContainsString('2 rows', $output);
+    }
+
+    public function test_read_rows_with_output_columns_multiple_columns() : void
+    {
+        $tester = new CommandTester(new FileReadCommand('read'));
+
+        $tester->execute([
+            'input-file' => __DIR__ . '/Fixtures/orders.csv',
+            '--input-file-limit' => 3,
+            '--output-columns' => ['order_id', 'discount', 'created_at'],
+        ]);
+
+        $tester->assertCommandIsSuccessful();
+
+        self::assertStringContainsString(
+            <<<'OUTPUT'
++----------------------+----------+----------------------+
+|             order_id | discount |           created_at |
++----------------------+----------+----------------------+
+| e13d7098-5a78-3389-9 |    12.45 | 2024-06-17T19:24:49+ |
+| 947df050-3abb-3f5a-9 |          | 2024-02-23T19:18:53+ |
+| 6315f9e2-86bf-3321-a |     47.1 | 2024-04-02T11:30:25+ |
++----------------------+----------+----------------------+
+3 rows
+OUTPUT,
+            $tester->getDisplay()
+        );
+    }
+
+    public function test_read_rows_with_output_columns_nonexistent_column() : void
+    {
+        $tester = new CommandTester(new FileReadCommand('read'));
+
+        $tester->execute([
+            'input-file' => __DIR__ . '/Fixtures/orders.csv',
+            '--input-file-limit' => 2,
+            '--output-columns' => ['nonexistent_column'],
+        ]);
+
+        $tester->assertCommandIsSuccessful();
+
+        self::assertStringContainsString(
+            <<<'OUTPUT'
++--------------------+
+| nonexistent_column |
++--------------------+
+|                    |
+|                    |
++--------------------+
+2 rows
+OUTPUT,
+            $tester->getDisplay()
+        );
+    }
+
+    public function test_read_rows_with_output_columns_single_column() : void
+    {
+        $tester = new CommandTester(new FileReadCommand('read'));
+
+        $tester->execute([
+            'input-file' => __DIR__ . '/Fixtures/orders.csv',
+            '--input-file-limit' => 3,
+            '--output-columns' => ['order_id'],
+        ]);
+
+        $tester->assertCommandIsSuccessful();
+
+        self::assertStringContainsString(
+            <<<'OUTPUT'
++----------------------+
+|             order_id |
++----------------------+
+| e13d7098-5a78-3389-9 |
+| 947df050-3abb-3f5a-9 |
+| 6315f9e2-86bf-3321-a |
++----------------------+
+3 rows
+OUTPUT,
+            $tester->getDisplay()
+        );
+    }
+
     public function test_read_rows_xml() : void
     {
         $tester = new CommandTester(new FileReadCommand('read'));