Is there a way to easily/elegantly update multiple columns in a table with record values from a query in Oracle 10g?
I have a table (call it Source) which has for simplicities sake, 2 columns, ID and KEY.
ID Key ---- ---- 1 1000 2 1000 3 5000 4 1000 .. 101 8000 102 9000 103 7000 104 9000 ... 201 5 202 5 ...
I have another table (call it KeyMap) that takes trunc(ID/100) and uses it as a batchID
with the columns being a key map for the IDs within the batch:
trunc(ID/100) key1 key2 key3 key4 ..... key99 ------------- ---- ---- ---- ---- 0 1000 1000 5000 1000 1 8000 9000 7000 9000 2 5 5
The ID's are created and processed in batches, so at the conclusion of the batch processing I would like to call a stored procedure to update the record in the KeyMap table with the new Key values with 1 update statement using a sub-select or collection providing those key values.
Is this possible and what is the best/most efficient way of doing this?
I'll limit my criticism to say that your table design is not normalized, and isn't very pretty, but I'll assume you have your reasons. I typically do these "rotation" queries by using DECODE combined with a aggregate column, grouping by my key - in this case, your pseudo-key, trunc(ID/100). Combine that with the update syntax that uses tuples:
UPDATE Foo
SET (a, b, c, d)
= (w, x, y, z);
and you get:
UPDATE KeyMap
SET
( key1
, key2
, key3
, key4
...
, key99
)
= ( SELECT MAX(decode(mod(ID, 100), 1, Key, NULL))
, MAX(decode(mod(ID, 100), 2, Key, NULL))
, MAX(decode(mod(ID, 100), 3, Key, NULL))
, MAX(decode(mod(ID, 100), 4, Key, NULL))
...
, MAX(decode(mod(ID, 100), 99, Key, NULL))
FROM Source
WHERE Trunc(Source.ID / 100) = KeyMap.batchId
GROUP BY Trunc(Source.ID / 100)
)
WHERE BatchId = <x>;
you could generate an Oracle VARRAY and pass in your keys in a batch. your procedure could iterate over the VARRAY and update the table.
来源:https://stackoverflow.com/questions/1003011/oracle-10g-pl-sql-select-results-as-update-column-values