[sql] Is there a workaround for ORA-01795: maximum number of expressions in a list is 1000 error?

Is there a workaround for

'ORA-01795: maximum number of expressions in a list is 1000 error'

I have a query and it is selecting fields based on the value of one field. I am using the in clause and there are 10000+ values

example:

select field1, field2, field3 
from table1 
where name in 
(
'value1',
'value2',
...
'value10000+'
);

Every time I execute the query I get the ORA-01795: maximum number of expressions in a list is 1000 error. I am trying to execute the query in TOAD, no difference, the same error. How would I modify the query to get it to work?

Thanks in advance

I realize this is an old question and referring to TOAD but if you need to code around this using c# you can split up the list through a for loop. You can essentially do the same with Java using subList();

    List<Address> allAddresses = GetAllAddresses();
    List<Employee> employees = GetAllEmployees(); // count > 1000

    List<Address> addresses = new List<Address>();

    for (int i = 0; i < employees.Count; i += 1000)
    {
        int count = ((employees.Count - i) < 1000) ? (employees.Count - i) - 1 : 1000;
        var query = (from address in allAddresses
                     where employees.GetRange(i, count).Contains(address.EmployeeId)
                     && address.State == "UT"
                     select address).ToList();

        addresses.AddRange(query);
    }

Hope this helps someone.

I ran into this issue recently and figured out a cheeky way of doing it without stringing together additional IN clauses

You could make use of Tuples

SELECT field1, field2, field3
FROM table1
WHERE (1, name) IN ((1, value1), (1, value2), (1, value3),.....(1, value5000));

Oracle does allow >1000 Tuples but not simple values. More on this here,

https://community.oracle.com/message/3515498#3515498
and
https://community.oracle.com/thread/958612

This is of course if you don't have the option of using a subquery inside IN to get the values you need from a temp table.

Some workaround solutions are:

1. Split up IN clause

Split IN clause to multiple IN clauses where literals are less than 1000 and combine them using OR clauses:

Split the original "WHERE" clause from one "IN" condition to several "IN" condition:

Select id from x where id in (1, 2, ..., 1000,…,1500);

To:

Select id from x where id in (1, 2, ..., 999) OR id in (1000,...,1500);

2. Use tuples

The limit of 1000 applies to sets of single items: (x) IN ((1), (2), (3), ...). There is no limit if the sets contain two or more items: (x, 0) IN ((1,0), (2,0), (3,0), ...):

Select id from x where (x.id, 0) IN ((1, 0), (2, 0), (3, 0),.....(n, 0));

3. Use temporary table

Select id from x where id in (select id from <temporary-table>);

    **Divide a list to lists of n size**

    import java.util.AbstractList;
    import java.util.ArrayList;
    import java.util.List;

    public final class PartitionUtil<T> extends AbstractList<List<T>> {

        private final List<T> list;
        private final int chunkSize;

        private PartitionUtil(List<T> list, int chunkSize) {
            this.list = new ArrayList<>(list);
            this.chunkSize = chunkSize;
        }

        public static <T> PartitionUtil<T> ofSize(List<T> list, int chunkSize) {
            return new PartitionUtil<>(list, chunkSize);
        }

        @Override
        public List<T> get(int index) {
            int start = index * chunkSize;
            int end = Math.min(start + chunkSize, list.size());

            if (start > end) {
                throw new IndexOutOfBoundsException("Index " + index + " is out of the list range <0," + (size() - 1) + ">");
            }

            return new ArrayList<>(list.subList(start, end));
        }

        @Override
        public int size() {
            return (int) Math.ceil((double) list.size() / (double) chunkSize);
        }
    }





Function call : 
              List<List<String>> containerNumChunks = PartitionUtil.ofSize(list, 999)

more details: https://e.printstacktrace.blog/divide-a-list-to-lists-of-n-size-in-Java-8/

There's also workaround doing disjunction of your array, worked for me as other solutions were hard to implement using some old framework.

select * from tableA where id = 1 or id = 2 or id = 3 ...

But for better perfo, I would use Nikolai Nechai's solution with unions, if possible.

there is also another way to resolve this issue. lets say you have two tables Table1 and Table2. and it is required to fetch all entries of Table1 not referred/present in Table2 using Criteria query. So go ahead like this...

List list=new ArrayList(); 
Criteria cr=session.createCriteria(Table1.class);
cr.add(Restrictions.sqlRestriction("this_.id not in (select t2.t1_id from Table2 t2 )"));
.
.

. . . It will perform all the subquery function directly in SQL without including 1000 or more parameters in SQL converted by Hibernate framework. It worked for me. Note: You may need to change SQL portion as per your requirement.

Please use an inner query inside of the in-clause:

select col1, col2, col3... from table1
 where id in (select id from table2 where conditions...)

Operato union

select * from tableA where tableA.Field1 in (1,2,...999)
union
select * from tableA where tableA.Field1 in (1000,1001,...1999)
union
select * from tableA where tableA.Field1 in (2000,2001,...2999)

Just use multiple in-clauses to get around this:

select field1, field2, field3 from table1 
where  name in ('value1', 'value2', ..., 'value999') 
    or name in ('value1000', ..., 'value1999') 
    or ...;

One more way:

CREATE OR REPLACE TYPE TYPE_TABLE_OF_VARCHAR2 AS TABLE OF VARCHAR(100);
-- ...
SELECT field1, field2, field3
  FROM table1
  WHERE name IN (
    SELECT * FROM table (SELECT CAST(? AS TYPE_TABLE_OF_VARCHAR2) FROM dual)
  );

I don't consider it's optimal, but it works. The hint /*+ CARDINALITY(...) */ would be very useful because Oracle does not understand cardinality of the array passed and can't estimate optimal execution plan.

As another alternative - batch insert into temporary table and using the last in subquery for IN predicate.

There is another option: with syntax. To use the OPs example, this would look like:

with data as (
  select 'value1' name from dual
  union all
  select 'value2' name from dual
  union all
...
  select 'value10000+' name from dual)
select field1, field2, field3 
from table1 t1
inner join data on t1.name = data.name;

I ran into this problem. In my case I had a list of data in Java where each item had an item_id and a customer_id. I have two tables in the DB with subscriptions to items respective customers. I want to get a list of all subscriptions to the items or to the customer for that item, together with the item id.

I tried three variants:

1. Multiple selects from Java (using tuples to get around the limit)
2. With-syntax
3. Temporary table

Option 1: Multiple Selects from Java

Basically, I first

select item_id, token 
from item_subs 
where (item_id, 0) in ((:item_id_0, 0)...(:item_id_n, 0))

Then

select cus_id, token 
from cus_subs 
where (cus_id, 0) in ((:cus_id_0, 0)...(:cus_id_n, 0))

Then I build a Map in Java with the cus_id as the key and a list of items as value, and for each found customer subscription I add (to the list returned from the first select) an entry for all relevant items with that item_id. It's much messier code

Option 2: With-syntax

Get everything at once with an SQL like

with data as (
  select :item_id_0 item_id, :cus_id_0 cus_id
  union all
  ...
  select :item_id_n item_id, :cus_id_n cus_id )
select I.item_id item_id, I.token token
from item_subs I
inner join data D on I.item_id = D.item_id
union all
select D.item_id item_id, C.token token
from cus_subs C
inner join data D on C.cus_id = D.cus_id

Option 3: Temporary table

Create a global temporary table with three fields: rownr (primary key), item_id and cus_id. Insert all the data there then run a very similar select to option 2, but linking in the temporary table instead of the with data

Performance

This is not a fully-scientific performance analysis.

• I'm running against a development database, with slightly over 1000 rows in my data set that I want to find subscriptions for.
• I've only tried one data set.
• I'm not in the same physical location as my DB server. It's not that far away, but I do notice if I try from home over the VPN then it's all much slower, even though it's the same distance (and it's not my home internet that's the problem).
• I was testing the full call, so my API calls another (also running in the same instance in dev) which also connects to to the DB to get the initial data set. But that is the same in all three cases.

YMMV.

That said, the temporary table option was much slower. As in double so slow. I was getting 14-15 seconds for option 1, 15-16 for option 2 and 30 for option 3.

I'll try them again from the same network as the DB server and check if that changes things when I get the chance.