{
"test_id": "year",
"pyspark_version": "3.5",
"generated_at": "2025-10-27T14:13:17.252710",
"input_data": [
{
"id": 1,
"name": "Alice",
"hire_date": "2020-01-15",
"birth_date": "1990-05-20"
},
{
"id": 2,
"name": "Bob",
"hire_date": "2019-03-10",
"birth_date": "1985-12-03"
},
{
"id": 3,
"name": "Charlie",
"hire_date": "2021-07-22",
"birth_date": "1992-08-14"
}
],
"operation": "DataFrame operation: year",
"expected_output": {
"schema": {
"field_count": 1,
"field_names": [
"year(hire_date)"
],
"field_types": [
"integer"
],
"fields": [
{
"name": "year(hire_date)",
"type": "integer",
"nullable": true
}
]
},
"data": [
{
"year(hire_date)": 2020
},
{
"year(hire_date)": 2019
},
{
"year(hire_date)": 2021
}
],
"row_count": 3
}
}