{
  "context": "This criteria evaluates how well the engineer uses ConnectorX's type system and data type conversion capabilities. The focus is on proper usage of read_sql() with different return_type parameters, understanding type mappings between database and Python/Arrow types, and correctly handling nullable types.",
  "type": "weighted_checklist",
  "checklist": [
    {
      "name": "Use read_sql function",
      "description": "Uses connectorx.read_sql() or cx.read_sql() to load data from the database",
      "max_score": 20
    },
    {
      "name": "Pandas format loading",
      "description": "Uses return_type='pandas' parameter or default behavior to load data into pandas DataFrame format",
      "max_score": 15
    },
    {
      "name": "Arrow format loading",
      "description": "Uses return_type='arrow' or return_type='arrow2' parameter to load data into arrow format",
      "max_score": 15
    },
    {
      "name": "Pandas type extraction",
      "description": "Correctly extracts pandas DataFrame column types using df.dtypes and converts them to strings",
      "max_score": 12
    },
    {
      "name": "Arrow type extraction",
      "description": "Correctly extracts arrow table/RecordBatch column types using schema inspection (e.g., table.schema)",
      "max_score": 13
    },
    {
      "name": "Nullable type handling",
      "description": "Demonstrates understanding that ConnectorX returns nullable types (Int64, boolean) for pandas format instead of converting NULL integers to float64",
      "max_score": 15
    },
    {
      "name": "Format type differences",
      "description": "Demonstrates that type representations differ between pandas and arrow formats (e.g., pandas 'Int64' vs arrow 'int64')",
      "max_score": 10
    }
  ]
}