问题
I need to split dates between two and ignore saturday and sunday from it. Built in function on 0.11.1 will help to get day of the week but how to find out whether that is saturday or Sunday? Anyone has any idea of it? My expected output described below.
Input:
User Fromdate Todate
Raj 10/3/2013 10/8/2013
James 10/4/2013 10/7/2013
etc..
Expected Output:
Raj 10/3/2013
Raj 10/4/2013
Raj 10/7/2013
Raj 10/8/2013
James 10/4/2013
James 10/7/2013
回答1:
Since the Pig DateTime objects are really Unix epoch time in milliseconds, this can be easily done with out-of-the-box Pig operators.
(DaysBetween(ToDate('10/3/2013','MM/dd/yyyy'),ToDate(0L)) + 4L) % 7
- Yields a long in range 0...6, where 0 = Sun, 1 = Mon, ... etc
- 0L represents 1/1/1970, a Thursday
- Adding 4L days makes 0 = Sunday
Satisfy yourself that this is so from Unix command line:
$> date -d '1/1/1970' +%w-%a
4-Thu
$> date -d '10/3/2013' +%w-%a
4-Thu
Of course, if you are comfortable with UDF, and this is a commonly occurring requirement, that's the best solution.
Carter Shore
回答2:
You'll need to write a UDF. You can use Java's Calendar class to do this.
来源：https://stackoverflow.com/questions/19152907/apache-pig-get-the-day-of-the-week-and-split-accordingly