Xquery string for taking out repeated data

Question

I receive an xml like this from a service:
<mov>
<id>123</id>
<timestamp>2009-06-15-13.38.50.148244</timestamp>
<data>some data for id 123</data>
</mov>
<mov>
<id>456</id>
<timestamp>2009-06-15-15.00.50.148244</timestamp>
<data>some data</data>
</mov>
<mov>
<id>123</id>
<timestamp>2009-06-15-21.30.50.123240</timestamp>
<data>updated data for id 123</data>
</mov>
(well, there are around 100 movs in a 300kb xml file)
I have to update a database table (whose primary key is id) with the received data.
They aren't suposed to send me repeated <id> values in one xml .. but it is happening once and again, as in the sample above with id=123 , so my merge obviously fails with an oracle error as it is unable to get a consistent set of records for the update.
Is there any XQuery expression to eliminate in the xmltable the <mov> with the repeated id value?
Specifically, I want to get rid of the first mov (the oldest one based on timestamp).
If the timestamp comparison is hard to achieve, getting rid of the prior movs for the same id would do.
Thanks.

Answer 1

Hi,
You could do it in SQL :
WITH t AS (
SELECT xmltype('<root>
<mov>
<id>123</id>
<timestamp>2009-06-15-13.38.50.148244</timestamp>
<data>some data for id 123</data>
</mov>
<mov>
<id>456</id>
<timestamp>2009-06-15-15.00.50.148244</timestamp>
<data>some data</data>
</mov>
<mov>
<id>123</id>
<timestamp>2009-06-15-21.30.50.123240</timestamp>
<data>updated data for id 123</data>
</mov>
</root>') doc
FROM dual
SELECT id, tstamp, data
FROM (
  SELECT id,
         to_timestamp(tstamp,'YYYY-MM-DD-HH24.MI.SS.FF') tstamp,
         data,
         row_number() over( partition by id
                            order by to_timestamp(tstamp,'YYYY-MM-DD-HH24.MI.SS.FF') desc ) rn
  FROM t, XMLTable(
   '/root/mov'
   passing t.doc
   columns
    id         number path 'id',
    tstamp     varchar2(30) path 'timestamp',
    data       varchar2(4000) path 'data'
  ) x
WHERE rn = 1;With XQuery only, as you've foreseen, it's not easy to deal with timestamps (at least in the format you get them).
Below, an example retrieving only the last occurrence of each id (your 2nd option) :
WITH t AS (
SELECT xmltype('<root>
<mov>
<id>123</id>
<timestamp>2009-06-15-13.38.50.148244</timestamp>
<data>some data for id 123</data>
</mov>
<mov>
<id>456</id>
<timestamp>2009-06-15-15.00.50.148244</timestamp>
<data>some data</data>
</mov>
<mov>
<id>123</id>
<timestamp>2009-06-15-21.30.50.123240</timestamp>
<data>updated data for id 123</data>
</mov>
</root>') doc
FROM dual
SELECT x.id,
       to_timestamp(x.tstamp,'YYYY-MM-DD-HH24.MI.SS.FF') tstamp,
       x.data
FROM t, XMLTable(
'for $i in distinct-values($d//mov/id)
  return $d//mov[id=$i][last()]'
passing t.doc as "d"
columns
  id      number         path 'id',
  tstamp  varchar2(30)   path 'timestamp',
  data    varchar2(4000) path 'data'
) x;Edited by: odie_63 on 27 avr. 2010 12:51