Introduction
Execution Mode
Flow DataFrame supports two execution modes that control how scalar functions handle validation errors:
- LENIENT (default): Validation errors are ignored, processing continues with fallback values (null, false, [], '')
- STRICT: Validation errors throw
InvalidArgumentExceptionimmediately, stopping execution
Changing Execution Mode
use Flow\ETL\Function\ExecutionMode;
(data_frame())
->read(from_array([
['name' => 'John', 'value' => '10'],
['name' => null, 'value' => '20'],
]))
->mode(ExecutionMode::STRICT)
->withEntry('upper', ref('name')->upper())
->write(to_stream(__DIR__ . '/output.csv', truncate: false))
->run();
Behavior Comparison
LENIENT Mode (Default)
When invalid data is encountered, functions return fallback values and processing continues:
(data_frame())
->read(from_array([
['text' => 'hello'],
['text' => null], // Invalid data
]))
->withEntry('upper', ref('text')->upper())
->fetch();
// Result:
// [
// ['text' => 'hello', 'upper' => 'HELLO'],
// ['text' => null, 'upper' => null], // Returns null, continues processing
// ]
STRICT Mode
When invalid data is encountered, an exception is thrown immediately:
(data_frame())
->mode(ExecutionMode::STRICT)
->read(from_array([
['text' => 'hello'],
['text' => null], // Invalid data
]))
->withEntry('upper', ref('text')->upper())
->fetch();
// Throws: InvalidArgumentException: Upper function requires non-null string
Adapters
Bridges
Contributors
Join us on GitHub
