Here is an example of a no arg that will return a different value for each
row:
https://code.google.com/p/gbif-occurrencestore/source/browse/trunk/occurrence-store/src/main/java/org/gbif/occurrencestore/hive/udf/UuidUDF.java
Hope this helps,
Tim
On Mon, Sep 30, 2013 at 10:59 PM, Yang <teddyyyy...@gmail.com> wrote:
> thanks! at first I did have a no-arg evaluate(), but somehow
>
> select myfunction(), field1, field2 from mytable ;
>
> spits out the same value for myfunction() for each row. so I was wondering
> whether the UDF got called only 1 time, because the hive compiler sees that
> the argument is void, so that
> all the invocations would be "having the same value", then I tried to pass
> in a param to prevent this possibility.
>
>
> On Mon, Sep 30, 2013 at 1:55 PM, Tim Robertson
> <timrobertson...@gmail.com>wrote:
>
>> It's been ages since I wrote one, but the differences to mine:
>>
>> a) I use LongWritable: public LongWritable evaluate(LongWritable
>> startAt) {
>> b) I have annotations on the class (but I think they are just for docs)
>> @Description(name = "row_sequence",
>> value = "_FUNC_() - Returns a generated row sequence number starting
>> from 1")
>> @UDFType(deterministic = false)
>> public class UDFRowSequence extends UDF {
>>
>> Hope this helps!
>> Tim
>>
>>
>>
>> On Mon, Sep 30, 2013 at 10:47 PM, Yang <teddyyyy...@gmail.com> wrote:
>>
>>> I wrote a super simple UDF, but got some errors:
>>>
>>> UDF:
>>>
>>> package yy;
>>> import org.apache.hadoop.hive.ql.exec.UDF;
>>> import java.util.Random;
>>> import java.util.UUID;
>>> import java.lang.management.*;
>>>
>>> public class MyUdf extends UDF {
>>> static Random rand = new Random(System.currentTimeMillis() +
>>> Thread.currentThread().getId()* 1000000);
>>> String name = ManagementFactory.getRuntimeMXBean().getName();
>>> long startValue = Long.valueOf(name.replaceAll("[^\\d]+", "")) *
>>> 10000 + Thread.currentThread().getId() * 1000;
>>> public long evaluate(long x ) {
>>> //return (long)UUID.randomUUID().hashCode();
>>> //return rand.nextLong();
>>> return startValue++;
>>> }
>>> }
>>>
>>>
>>>
>>>
>>>
>>> sql script:
>>>
>>> CREATE TEMPORARY FUNCTION gen_uniq2 AS 'yy.MyUdf';
>>> select gen_uniq2(field1), field2
>>> from yy_mapping limit 10;
>>>
>>> field1 is bigint, field2 is int
>>>
>>>
>>>
>>>
>>>
>>> error:
>>>
>>> hive> source aa.sql;
>>> Added ./MyUdf.jar to class path
>>> Added resource: ./MyUdf.jar
>>> OK
>>> Time taken: 0.0070 seconds
>>> FAILED: SemanticException [Error 10014]: Line 2:7 Wrong arguments
>>> 'field1': No matching method for class yy.MyUdf with (bigint). Possible
>>> choices: _FUNC_()
>>>
>>>
>>>
>>>
>>>
>>> so I'm declaring a UDF with arg of long, so that should work for a
>>> bigint (more importantly it's complaining not long vs bigint, but bigint vs
>>> void ). I tried changing both to int, same failure
>>>
>>>
>>> thanks!
>>> yang
>>>
>>>
>>>
>>
>
