spark unix_timestamp data type mismatch

问题:

Could someone help guide me in what data type or format I need to submit from_unixtime for the spark from_unixtime() function to work?

When I try the following it works, but responds not with current_timestamp.

from_unixtime(current_timestamp())

The response is below:

fromunixtime(currenttimestamp(),yyyy-MM-dd HH:mm:ss)

When I try to input

from_unixtime(1392394861,"yyyy-MM-dd HH:mm:ss.SSSS")

The above simply fails with a type mismatch:

error: type mismatch; found : Int(1392394861) required: org.apache.spark.sql.Column from_unixtime(1392394861,"yyyy-MM-dd HH:mm:ss.SSSS")

What am I missing? I've tried a number of different things and tried reading documentation on using date/time in spark and every example I've tried fails with type mismatches.

回答1:

Use lit() to create a column of literal value, like this:

from_unixtime(lit(1392394861), "yyyy-MM-dd HH:mm:ss.SSSS")

or, as zero323 mentioned:

from_unixtime(current_timestamp().cast("long"))