MakeTimestampNTZ¶

Overview¶

MakeTimestampNTZ creates a timestamp without timezone (NTZ) by combining a date with a time value. This expression is a runtime replaceable binary expression that delegates its implementation to the DateTimeUtils.makeTimestampNTZ method.

Syntax¶

make_timestamp_ntz(date_expr, time_expr)

// DataFrame API
col("date_col").make_timestamp_ntz(col("time_col"))

Arguments¶

Argument	Type	Description
left	DateType	The date component for the timestamp
right	AnyTimeType	The time component (can be time or timestamp type)

Return Type¶

TimestampNTZType - A timestamp without timezone information.

Supported Data Types¶

Left operand: DateType only
Right operand: Any time-related type (AnyTimeType which includes time and timestamp types)

Algorithm¶

Takes a date value as the first argument to provide the date component
Takes a time value as the second argument to provide the time component
Delegates evaluation to DateTimeUtils.makeTimestampNTZ static method via StaticInvoke
Combines the date and time components into a single timestamp without timezone
Returns the resulting timestamp in NTZ format

Partitioning Behavior¶

This expression does not affect partitioning behavior:

Preserves existing partitioning as it's a row-level transformation
Does not require data shuffle
Can be applied within partition boundaries

Edge Cases¶

Null handling: If either input is null, the result will be null
Invalid dates: Behavior depends on the underlying DateTimeUtils.makeTimestampNTZ implementation
Time overflow: Time values that exceed 24-hour boundaries may wrap or cause errors depending on implementation
Timezone considerations: Output explicitly excludes timezone information (NTZ = No Time Zone)

Code Generation¶

This expression uses RuntimeReplaceable with StaticInvoke, which supports Tungsten code generation. The actual code generation is handled by the replacement StaticInvoke expression calling the static method.

Examples¶

-- Create timestamp NTZ from date and time
SELECT make_timestamp_ntz(DATE '2023-12-25', TIME '14:30:00') as christmas_afternoon;

-- Using with table columns
SELECT make_timestamp_ntz(date_col, time_col) as full_timestamp 
FROM events_table;

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

df.select(expr("make_timestamp_ntz(event_date, event_time)"))

// With column references  
df.withColumn("full_timestamp", expr("make_timestamp_ntz(date_col, time_col)"))