UnaryMinus¶

Overview¶

The UnaryMinus expression represents the unary minus operator that negates numeric values and intervals in Spark SQL. It supports both overflow-safe ANSI mode and standard negation behavior, handling various numeric types including decimals and calendar intervals.

Syntax¶

-expression
NEGATIVE(expression)

// DataFrame API
col("column").unary_-

Arguments¶

Argument	Type	Description
child	Expression	The expression to negate
failOnError	Boolean	Whether to fail on arithmetic overflow (defaults to ANSI mode setting)

Return Type¶

Returns the same data type as the input expression. The negated value maintains the original precision and scale for decimal types.

Supported Data Types¶

Numeric types: ByteType, ShortType, IntegerType, LongType, FloatType, DoubleType
DecimalType (arbitrary precision)
CalendarIntervalType
DayTimeIntervalType
YearMonthIntervalType

Algorithm¶

For decimal types, uses the built-in unary_$minus() method
For integer types in ANSI mode, applies overflow checking via MathUtils.negateExact()
For standard numeric types, performs basic negation with careful code generation to avoid compilation issues
For calendar intervals, uses IntervalUtils.negate() or IntervalUtils.negateExact() based on error handling mode
For ANSI interval types, applies IntervalMathUtils.negateExact() for overflow protection

Partitioning Behavior¶

This expression preserves partitioning characteristics:

Does not require shuffle operations
Maintains existing partitioning schemes
Can be pushed down to individual partitions for parallel execution

Edge Cases¶

Null handling: Returns null for null input (null intolerant behavior)
Integer overflow: In ANSI mode, throws exception on overflow (e.g., negating Long.MIN_VALUE)
Decimal precision: Maintains original precision and scale without loss
Interval overflow: Calendar and ANSI intervals check for overflow when failOnError is enabled
Code generation edge case: Uses temporary variables to avoid compiler issues with direct negation of literals

Code Generation¶

Supports full code generation (Tungsten) with optimized paths:

Uses nullSafeCodeGen for most numeric types
Special handling for decimal types via defineCodeGen
Generates overflow-safe code when ANSI mode is enabled
Falls back to interpreted evaluation only for unsupported data types

Examples¶

-- Basic negation
SELECT -42;
-- Result: -42

-- With column reference
SELECT -price FROM products;

-- Using function alias
SELECT NEGATIVE(quantity) FROM inventory;

-- With intervals
SELECT -INTERVAL '2' DAY;
-- Result: -2 days

// DataFrame API usage
import org.apache.spark.sql.functions._

df.select(-col("amount"))
df.select(col("value").unary_-)

// With expressions
df.select(expr("-price"))