Right¶

Overview¶

The Right expression extracts a specified number of characters from the right side of a string. It is implemented as a RuntimeReplaceable expression that transforms into conditional logic using Substring operations during query planning.

Syntax¶

RIGHT(str, len)

// DataFrame API
import org.apache.spark.sql.functions._
df.select(expr("RIGHT(column_name, length)"))

Arguments¶

Argument	Type	Description
str	String	The input string from which to extract characters
len	Integer	The number of characters to extract from the right side

Return Type¶

Returns a StringType with the same collation as the input string.

Supported Data Types¶

Input string: StringTypeWithCollation (supports trim collation)
Length parameter: IntegerType

Algorithm¶

The expression is transformed at runtime into the following conditional logic:

If the input string is null, return null
If the length parameter is less than or equal to 0, return an empty string
Otherwise, use Substring with a negative offset (UnaryMinus(len)) to extract from the right
The Substring operation calculates the starting position from the end of the string
Implicit type casting is applied to ensure type compatibility

Partitioning Behavior¶

This expression preserves partitioning characteristics:

Does not require data shuffle as it operates on individual rows
Maintains the same partitioning scheme as the input DataFrame
Can be pushed down to individual partitions for parallel execution

Edge Cases¶

Null input: Returns null when the input string is null
Zero or negative length: Returns empty string (UTF8String.EMPTY_UTF8) when len ≤ 0
Length greater than string length: Returns the entire string when requested length exceeds string length
Empty string input: Returns empty string regardless of length parameter
Type coercion: Automatically casts compatible types through ImplicitCastInputTypes

Code Generation¶

This expression supports Tungsten code generation through its replacement mechanism:

Transforms into If, Substring, and UnaryMinus expressions at planning time
The replacement expressions support whole-stage code generation
Benefits from optimized UTF-8 string operations in generated code

Examples¶

-- Extract last 3 characters
SELECT RIGHT('Hello World', 3); -- Returns 'rld'

-- Handle edge cases
SELECT RIGHT('abc', 0);    -- Returns ''
SELECT RIGHT('abc', -1);   -- Returns ''
SELECT RIGHT('abc', 10);   -- Returns 'abc'
SELECT RIGHT(NULL, 3);     -- Returns NULL

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

df.select(expr("RIGHT(name, 3)"))
df.select(expr("RIGHT(description, 5)").as("suffix"))

// With column references
df.select(expr("RIGHT(text_column, length_column)"))