Unified behavior of file analyze and file schema CLI Commands (#1785)

Author: norberttech

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

@@ -50,7 +50,8 @@ public function configure() : void
             ->addArgument('input-file', InputArgument::REQUIRED, 'Path to a file from which schema should be extracted.')
             ->addOption('input-file-format', null, InputArgument::OPTIONAL, 'File format. When not set file format is guessed from source file path extension', null)
             ->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('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('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);
         $this->addJSONInputOptions($this);
@@ -78,8 +79,11 @@ protected function execute(InputInterface $input, OutputInterface $output) : int
             return Command::FAILURE;
         }
 
-        $df->batchSize($batchSize)
-            ->autoCast();
+        $df->batchSize($batchSize);
+
+        if (option_bool('schema-auto-cast', $input)) {
+            $df->autoCast();
+        }
 
         $limit = option_int_nullable('input-file-limit', $input);
 
@@ -97,7 +101,7 @@ protected function execute(InputInterface $input, OutputInterface $output) : int
         }
 
         if (option_bool('stats-columns', $input)) {
-            $analyze->withColumnStatistics();
+            $analyze->withSchema()->withColumnStatistics();
         }
 
         $report = $df->run(

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

@@ -17,7 +17,7 @@ public function test_read_rows_csv() : void
         $application->add(new FileAnalyzeCommand());
         $tester = new CommandTester($application->get('file:analyze'));
 
-        $tester->execute(['input-file' => __DIR__ . '/Fixtures/orders.csv', '--input-file-limit' => 5, '--stats-schema' => true, '--stats-columns' => true]);
+        $tester->execute(['input-file' => __DIR__ . '/Fixtures/orders.csv', '--input-file-limit' => 5, '--stats-schema' => true, '--stats-columns' => true, '--schema-auto-cast' => true]);
 
         $tester->assertCommandIsSuccessful();
 
@@ -74,7 +74,7 @@ public function test_read_rows_csv_without_schema() : void
         $application->add(new FileAnalyzeCommand());
         $tester = new CommandTester($application->get('file:analyze'));
 
-        $tester->execute(['input-file' => __DIR__ . '/Fixtures/orders.csv', '--input-file-limit' => 5, '--stats-columns' => true]);
+        $tester->execute(['input-file' => __DIR__ . '/Fixtures/orders.csv', '--input-file-limit' => 5, '--stats-columns' => true, '--schema-auto-cast' => true]);
 
         $tester->assertCommandIsSuccessful();
 
@@ -93,17 +93,17 @@ public function test_read_rows_csv_without_schema() : void
 Columns
 -------
 
-┌────────────┬──────┬───────┬─────────────────┬───────────────────────────┬───────────────────────────┬────────────┬────────────┬────────────────────┬────────────────────┐
-│ Name       │ Type │ Nulls │ Distinct Values │ Min                       │ Max                       │ Min Length │ Max Length │ Min Elements Count │ Max Elements Count │
-├────────────┼──────┼───────┼─────────────────┼───────────────────────────┼───────────────────────────┼────────────┼────────────┼────────────────────┼────────────────────┤
-│ order_id   │ N/A  │ 0     │ 5               │ -                         │ -                         │ -          │ -          │ -                  │ -                  │
-│ created_at │ N/A  │ 0     │ 5               │ 2024-02-23T19:18:53+00:00 │ 2024-06-17T19:24:49+00:00 │ -          │ -          │ -                  │ -                  │
-│ updated_at │ N/A  │ 0     │ 5               │ 2024-02-23T19:18:53+00:00 │ 2024-06-17T19:24:49+00:00 │ -          │ -          │ -                  │ -                  │
-│ discount   │ N/A  │ 2     │ 3               │ 12.45                     │ 47.10                     │ -          │ -          │ -                  │ -                  │
-│ address    │ N/A  │ 0     │ 5               │ -                         │ -                         │ -          │ -          │ 4                  │ 4                  │
-│ notes      │ N/A  │ 0     │ 5               │ -                         │ -                         │ -          │ -          │ 1                  │ 5                  │
-│ items      │ N/A  │ 0     │ 5               │ -                         │ -                         │ -          │ -          │ 2                  │ 4                  │
-└────────────┴──────┴───────┴─────────────────┴───────────────────────────┴───────────────────────────┴────────────┴────────────┴────────────────────┴────────────────────┘
+┌────────────┬───────────────────────────────────────────────────────────────┬───────┬─────────────────┬───────────────────────────┬───────────────────────────┬────────────┬────────────┬────────────────────┬────────────────────┐
+│ Name       │ Type                                                          │ Nulls │ Distinct Values │ Min                       │ Max                       │ Min Length │ Max Length │ Min Elements Count │ Max Elements Count │
+├────────────┼───────────────────────────────────────────────────────────────┼───────┼─────────────────┼───────────────────────────┼───────────────────────────┼────────────┼────────────┼────────────────────┼────────────────────┤
+│ order_id   │ uuid                                                          │ 0     │ 5               │ -                         │ -                         │ -          │ -          │ -                  │ -                  │
+│ created_at │ datetime                                                      │ 0     │ 5               │ 2024-02-23T19:18:53+00:00 │ 2024-06-17T19:24:49+00:00 │ -          │ -          │ -                  │ -                  │
+│ updated_at │ datetime                                                      │ 0     │ 5               │ 2024-02-23T19:18:53+00:00 │ 2024-06-17T19:24:49+00:00 │ -          │ -          │ -                  │ -                  │
+│ discount   │ float                                                         │ 2     │ 3               │ 12.45                     │ 47.10                     │ -          │ -          │ -                  │ -                  │
+│ address    │ map<string, string>                                           │ 0     │ 5               │ -                         │ -                         │ -          │ -          │ 4                  │ 4                  │
+│ notes      │ list<string>                                                  │ 0     │ 5               │ -                         │ -                         │ -          │ -          │ 1                  │ 5                  │
+│ items      │ list<structure{sku: string, quantity: integer, price: float}> │ 0     │ 5               │ -                         │ -                         │ -          │ -          │ 2                  │ 4                  │
+└────────────┴───────────────────────────────────────────────────────────────┴───────┴─────────────────┴───────────────────────────┴───────────────────────────┴────────────┴────────────┴────────────────────┴────────────────────┘
 OUTPUT,
             $tester->getDisplay()
         );