fn extract_non_nullable_columns()

in datafusion/optimizer/src/eliminate_outer_join.rs [166:295]

• 118 lines of code
• 12 McCabe index (conditional complexity)

fn extract_non_nullable_columns(
    expr: &Expr,
    non_nullable_cols: &mut Vec<Column>,
    left_schema: &Arc<DFSchema>,
    right_schema: &Arc<DFSchema>,
    top_level: bool,
) -> Result<()> {
    match expr {
        Expr::Column(col) => {
            non_nullable_cols.push(col.clone());
            Ok(())
        }
        Expr::BinaryExpr(BinaryExpr { left, op, right }) => match op {
            // If one of the inputs are null for these operators, the results should be false.
            Operator::Eq
            | Operator::NotEq
            | Operator::Lt
            | Operator::LtEq
            | Operator::Gt
            | Operator::GtEq => {
                extract_non_nullable_columns(
                    left,
                    non_nullable_cols,
                    left_schema,
                    right_schema,
                    false,
                )?;
                extract_non_nullable_columns(
                    right,
                    non_nullable_cols,
                    left_schema,
                    right_schema,
                    false,
                )
            }
            Operator::And | Operator::Or => {
                // treat And as Or if does not from top level, such as
                // not (c1 < 10 and c2 > 100)
                if top_level && *op == Operator::And {
                    extract_non_nullable_columns(
                        left,
                        non_nullable_cols,
                        left_schema,
                        right_schema,
                        top_level,
                    )?;
                    extract_non_nullable_columns(
                        right,
                        non_nullable_cols,
                        left_schema,
                        right_schema,
                        top_level,
                    )?;
                    return Ok(());
                }

                let mut left_non_nullable_cols: Vec<Column> = vec![];
                let mut right_non_nullable_cols: Vec<Column> = vec![];

                extract_non_nullable_columns(
                    left,
                    &mut left_non_nullable_cols,
                    left_schema,
                    right_schema,
                    top_level,
                )?;
                extract_non_nullable_columns(
                    right,
                    &mut right_non_nullable_cols,
                    left_schema,
                    right_schema,
                    top_level,
                )?;

                // for query: select *** from a left join b where b.c1 ... or b.c2 ...
                // this can be eliminated to inner join.
                // for query: select *** from a left join b where a.c1 ... or b.c2 ...
                // this can not be eliminated.
                // If columns of relation exist in both sub exprs, any columns of this relation
                // can be added to non nullable columns.
                if !left_non_nullable_cols.is_empty()
                    && !right_non_nullable_cols.is_empty()
                {
                    for left_col in &left_non_nullable_cols {
                        for right_col in &right_non_nullable_cols {
                            if (left_schema.has_column(left_col)
                                && left_schema.has_column(right_col))
                                || (right_schema.has_column(left_col)
                                    && right_schema.has_column(right_col))
                            {
                                non_nullable_cols.push(left_col.clone());
                                break;
                            }
                        }
                    }
                }
                Ok(())
            }
            _ => Ok(()),
        },
        Expr::Not(arg) => extract_non_nullable_columns(
            arg,
            non_nullable_cols,
            left_schema,
            right_schema,
            false,
        ),
        Expr::IsNotNull(arg) => {
            if !top_level {
                return Ok(());
            }
            extract_non_nullable_columns(
                arg,
                non_nullable_cols,
                left_schema,
                right_schema,
                false,
            )
        }
        Expr::Cast(Cast { expr, data_type: _ })
        | Expr::TryCast(TryCast { expr, data_type: _ }) => extract_non_nullable_columns(
            expr,
            non_nullable_cols,
            left_schema,
            right_schema,
            false,
        ),
        _ => Ok(()),
    }
}