Collection Methods

This is a translation reference to convert the Oracle Collection Methods to Snowflake

This section is a work in progress, information may change in the future

Description

A collection method is a PL/SQL subprogram—either a function that returns information about a collection or a procedure that operates on a collection. Collection methods make collections easier to use and your applications easier to maintain.

(Oracle PL/SQL Language Reference COLLECTION METHODS)

Some of these methods can be mapped to native Snowflake semi-structured operations. The ones that can't or have differences will be mapped to a UDF implementation.

Current SnowConvert Support

The next table shows a summary of the current support provided by the SnowConvert tool. Please keep into account that translations may still not be final, and more work may be needed.

Method	Current recognition status	Current translation status	Mapped to
DELETE	Not Recognized.	Not Translated.	UDF
TRIM	Not Recognized.	Not Translated.	UDF (To be defined)
EXTEND	Not Recognized.	Not Translated.	UDF
EXISTS	Not Recognized.	Not Translated.	ARRAY_CONTAINS
FIRST	Not Recognized.	Not Translated.	UDF
LAST	Not Recognized.	Not Translated.	UDF
COUNT	Not Recognized.	Not Translated.	ARRAY_SIZE
LIMIT	Not Recognized.	Not Translated.	Not Supported.
PRIOR	Not Recognized.	Not Translated.	UDF (To be defined)
NEXT	Not Recognized.	Not Translated.	UDF (To be defined)

Sample Source Patterns

COUNT

This method returns the count of "non-undefined" (not to be confused with null) elements within a collection (nested tables can become sparse leaving these elements in between). In associative arrays, it returns the number of keys in the array.

Oracle

Query

CREATE OR REPLACE PROCEDURE collection_count
IS
    TYPE varray_typ IS VARRAY(5) OF INTEGER;
    TYPE nt_typ IS TABLE OF INTEGER;
    TYPE aa_typ IS TABLE OF INTEGER INDEX BY VARCHAR2(20);
    
    associative_array aa_typ := aa_typ('abc'=>1, 'bca'=>1);
    varray_variable varray_typ := varray_typ(1, 2, 3);
    nt_variable nt_typ := nt_typ(1, 2, 3, 4);
BEGIN
    DBMS_OUTPUT.PUT_LINE(associative_array.COUNT);
    DBMS_OUTPUT.PUT_LINE(varray_variable.COUNT);
    DBMS_OUTPUT.PUT_LINE(nt_variable.COUNT);
END;

CALL collection_count();

Result

DBMS OUTPUT
-----------
2
3
4

Snowflake

The snowflake equivalent is the ARRAY_SIZE method.

Query

CREATE OR REPLACE PROCEDURE PUBLIC.collection_count()
RETURNS VARCHAR
LANGUAGE SQL
EXECUTE AS CALLER
AS
$$
DECLARE
    associative_array OBJECT := OBJECT_CONSTRUCT('abc', 1, 'bca', 1);
    varray_variable ARRAY := ARRAY_CONSTRUCT(1, 2, 3);
    nt_variable ARRAY := ARRAY_CONSTRUCT(1, 2, 3, 4);
BEGIN
    CALL DBMS_OUTPUT.PUT_LINE(ARRAY_SIZE(OBJECT_KEYS(:associative_array)));
    CALL DBMS_OUTPUT.PUT_LINE(ARRAY_SIZE(:varray_variable));
    CALL DBMS_OUTPUT.PUT_LINE(ARRAY_SIZE(:nt_variable));
END;
$$;

CALL PUBLIC.collection_count();
SELECT * FROM DBMS_OUTPUT.DBMS_OUTPUT_LOG;

Result

DBMS OUTPUT
-----------
2
3
4

EXISTS

This method returns true if the given element is contained within the collection. In associative arrays, it tests if the key is contained.

Oracle

Query

CREATE OR REPLACE PROCEDURE collection_exists
IS
    TYPE nt_typ IS TABLE OF INTEGER;
    TYPE aa_typ IS TABLE OF INTEGER INDEX BY VARCHAR2(20);
    
    associative_array aa_typ := aa_typ('abc'=>1, 'bca'=>1);
    nt_variable nt_typ := nt_typ(1, 2, 3, 4);
BEGIN
    IF associative_array.EXISTS('abc')
    THEN DBMS_OUTPUT.PUT_LINE('Found');
    END IF;
    
    IF NOT associative_array.EXISTS('not found')
    THEN DBMS_OUTPUT.PUT_LINE('Not found');
    END IF;
    
    IF nt_variable.EXISTS(1)
    THEN DBMS_OUTPUT.PUT_LINE('Found');
    END IF;
    
    IF NOT nt_variable.EXISTS(5)
    THEN DBMS_OUTPUT.PUT_LINE('Not found');
    END IF;
END;
/

CALL collection_exists();

Result

DBMS OUTPUT
-----------
2
3
4

Snowflake

The snowflake equivalent is the ARRAY_CONTAINS method. Note that, when using Varchar elements, casting to Variant is necessary.

Query

CREATE OR REPLACE PROCEDURE PUBLIC.collection_exists()
RETURNS VARCHAR
LANGUAGE SQL
EXECUTE AS CALLER
AS
$$
DECLARE
    associative_array OBJECT := OBJECT_CONSTRUCT('abc', 1, 'bca', 1);
    nt_variable ARRAY := ARRAY_CONSTRUCT(1, 2, 3, 4);
BEGIN
    IF (ARRAY_CONTAINS('abc'::VARIANT, OBJECT_KEYS(associative_array)))
    THEN CALL DBMS_OUTPUT.PUT_LINE('Found');
    END IF;
    
    IF (NOT ARRAY_CONTAINS('not found'::VARIANT, OBJECT_KEYS(associative_array)))
    THEN CALL DBMS_OUTPUT.PUT_LINE('Not found');
    END IF;
    
    IF (ARRAY_CONTAINS(1, nt_variable))
    THEN CALL DBMS_OUTPUT.PUT_LINE('Found');
    END IF;
    
    IF (NOT ARRAY_CONTAINS(5, nt_variable))
    THEN CALL DBMS_OUTPUT.PUT_LINE('Not found');
    END IF;
END;
$$;

CALL PUBLIC.collection_exists();
SELECT * FROM DBMS_OUTPUT.DBMS_OUTPUT_LOG;

Result

DBMS OUTPUT
-----------
2
3
4

FIRST/LAST

These two methods return the First/Last element of the collection, respectively. If the collection is empty it returns null. This operation is mapped to a UDF, which will be added in further revisions.

Oracle

Query

CREATE OR REPLACE PROCEDURE collection_first_last
IS
    TYPE nt_typ IS TABLE OF INTEGER;
    TYPE aa_typ IS TABLE OF INTEGER INDEX BY VARCHAR2(20);
    
    associative_array aa_typ := aa_typ('abc'=>1, 'bca'=>1);
    nt_variable nt_typ := nt_typ();
BEGIN
    DBMS_OUTPUT.PUT_LINE(associative_array.FIRST);
    DBMS_OUTPUT.PUT_LINE(associative_array.LAST);
    
    DBMS_OUTPUT.PUT_LINE(nt_variable.FIRST);
    DBMS_OUTPUT.PUT_LINE(nt_variable.LAST);
    nt_variable := nt_typ(1, 2, 3, 4);
    DBMS_OUTPUT.PUT_LINE(nt_variable.FIRST);
    DBMS_OUTPUT.PUT_LINE(nt_variable.LAST);
END;
/

CALL collection_first_last();

Result

DBMS OUTPUT
-----------
abc
bca
             --These empty spaces are due to it evaluating to null

1
4

Snowflake

Query

CREATE OR REPLACE PROCEDURE PUBLIC.collection_first_last()
RETURNS VARCHAR
LANGUAGE SQL
EXECUTE AS CALLER
AS
$$
DECLARE
    associative_array OBJECT := OBJECT_CONSTRUCT('abc', 1, 'bca', 1);
    nt_variable ARRAY := ARRAY_CONSTRUCT();
BEGIN
    CALL DBMS_OUTPUT.PUT_LINE(ARRAY_FIRST(:associative_array));
    CALL DBMS_OUTPUT.PUT_LINE(ARRAY_LAST(:associative_array));
    
    CALL DBMS_OUTPUT.PUT_LINE(ARRAY_FIRST(:nt_variable));
    CALL DBMS_OUTPUT.PUT_LINE(ARRAY_LAST(:nt_variable));
    nt_variable := ARRAY_CONSTRUCT(1, 2, 3, 4);
    CALL DBMS_OUTPUT.PUT_LINE(ARRAY_FIRST(:nt_variable));
    CALL DBMS_OUTPUT.PUT_LINE(ARRAY_LAST(:nt_variable));
END;
$$;

CALL PUBLIC.collection_first_last();
SELECT * FROM DBMS_OUTPUT.DBMS_OUTPUT_LOG;

UDFs

CREATE OR REPLACE FUNCTION ARRAY_FIRST(array_variable VARIANT)
RETURNS VARIANT
LANGUAGE SQL
AS
$$
    IFF (IS_OBJECT(array_variable),
        ARRAY_FIRST(OBJECT_KEYS(array_variable)),
        IFF (ARRAY_SIZE(array_variable) = 0, null, array_variable[0]))
$$;

CREATE OR REPLACE FUNCTION ARRAY_LAST(array_variable VARIANT)
RETURNS VARIANT
LANGUAGE SQL
AS
$$
    IFF (IS_OBJECT(array_variable),
        ARRAY_LAST(OBJECT_KEYS(array_variable)),
        IFF (ARRAY_SIZE(array_variable) = 0, null, array_variable[ARRAY_SIZE(array_variable)-1]))
$$;

Result

DBMS OUTPUT
-----------
abc
bca
             --These empty spaces are due to it evaluating to null

1
4

DELETE

This method is used to remove elements from a Collection. It has three possible variants:

• .DELETE removes all elements.

• .DELETE(n) removes the element whose index matches 'n'.

• .DELETE(n, m) removes in the indexes from 'n' through 'm'.

In Oracle, using this operation on Nested Tables causes it to have "undefined" elements within it due to them being sparse.

Please note that the second and third versions do not apply to Varrays.

Oracle

For the sake of simplicity, this sample only checks on the number of elements but may be modified to display the contents of each collection.

Query

CREATE OR REPLACE PROCEDURE collection_delete
IS
    TYPE varray_typ IS VARRAY(5) OF INTEGER;
    TYPE nt_typ IS TABLE OF INTEGER;
    TYPE aa_typ IS TABLE OF INTEGER INDEX BY VARCHAR2(20);
    
    associative_array1 aa_typ := aa_typ('abc'=>1, 'def'=>2, 'ghi'=>3, 'jkl'=>4);
    associative_array2 aa_typ := aa_typ('abc'=>1, 'def'=>2, 'ghi'=>3, 'jkl'=>4);
    associative_array3 aa_typ := aa_typ('abc'=>1, 'def'=>2, 'ghi'=>3, 'jkl'=>4);
    
    varray_variable1 varray_typ := varray_typ(1, 2, 3, 4);
    
    nt_variable1 nt_typ := nt_typ(1, 2, 3, 4);
    nt_variable2 nt_typ := nt_typ(1, 2, 3, 4);
    nt_variable3 nt_typ := nt_typ(1, 2, 3, 4);
BEGIN
    varray_variable1.DELETE;--delete everything
    
    nt_variable1.DELETE;--delete everything
    nt_variable2.DELETE(2);--delete second position
    nt_variable3.DELETE(2, 3);--delete range
    
    associative_array1.DELETE;--delete everything
    associative_array2.DELETE('def');--delete second position
    associative_array3.DELETE('def', 'jkl');--delete range
    
    DBMS_OUTPUT.PUT_LINE(varray_variable1.COUNT);
    DBMS_OUTPUT.PUT_LINE(nt_variable1.COUNT);
    DBMS_OUTPUT.PUT_LINE(nt_variable2.COUNT);
    DBMS_OUTPUT.PUT_LINE(nt_variable3.COUNT);
    
    DBMS_OUTPUT.PUT_LINE(associative_array1.COUNT);
    DBMS_OUTPUT.PUT_LINE(associative_array2.COUNT);
    DBMS_OUTPUT.PUT_LINE(associative_array3.COUNT);
END;
/

CALL collection_delete();

Result

DBMS OUTPUT
-----------
0
0
3
2
0
3
1

Snowflake

Snowflake does not support deletions from an existing ARRAY and for this reason, the only offered workaround is to rebuild a new ARRAY depending on the original parameters of the DELETE.

Note that a UDF was added to implement the functionality for the update of the element.

This UDF will be added in later revisions.

Query

CREATE OR REPLACE PROCEDURE PUBLIC.collection_delete()
RETURNS VARCHAR
LANGUAGE SQL
EXECUTE AS CALLER
AS
$$
DECLARE
    associative_array1 OBJECT := OBJECT_CONSTRUCT('abc'=>1, 'def'=>2, 'ghi'=>3, 'jkl'=>4);
    associative_array2 OBJECT := OBJECT_CONSTRUCT('abc'=>1, 'def'=>2, 'ghi'=>3, 'jkl'=>4);
    associative_array3 OBJECT := OBJECT_CONSTRUCT('abc'=>1, 'def'=>2, 'ghi'=>3, 'jkl'=>4);
    
    varray_variable1 ARRAY := ARRAY_CONSTRUCT(1, 2, 3, 4);
    
    nt_variable1 ARRAY := ARRAY_CONSTRUCT(1, 2, 3, 4);
    nt_variable2 ARRAY := ARRAY_CONSTRUCT(1, 2, 3, 4);
    nt_variable3 ARRAY := ARRAY_CONSTRUCT(1, 2, 3, 4);
BEGIN
    varray_variable1 := ARRAY_CONSTRUCT();--delete everything
    
    nt_variable1 := ARRAY_CONSTRUCT();--delete everything
    nt_variable2 := ARRAY_DELETE_UDF(nt_variable2, 2);--delete second position
    nt_variable3 := ARRAY_DELETE_UDF(nt_variable3, 2, 3);--delete range
    
    associative_array1 := OBJECT_CONSTRUCT();--delete everything
    associative_array2 := ASSOCIATIVE_ARRAY_DELETE_UDF('def');--delete second position
    associative_array3 := ASSOCIATIVE_ARRAY_DELETE_UDF('def', 'jkl');--delete range
    
    CALL DBMS_OUTPUT.PUT_LINE(ARRAY_SIZE(varray_variable1));
    CALL DBMS_OUTPUT.PUT_LINE(ARRAY_SIZE(nt_variable1);
    CALL DBMS_OUTPUT.PUT_LINE(ARRAY_SIZE(nt_variable2);
    CALL DBMS_OUTPUT.PUT_LINE(ARRAY_SIZE(nt_variable3);
    
    CALL DBMS_OUTPUT.PUT_LINE(ARRAY_SIZE(associative_array1));
    CALL DBMS_OUTPUT.PUT_LINE(ARRAY_SIZE(associative_array2));
    CALL DBMS_OUTPUT.PUT_LINE(ARRAY_SIZE(associative_array3));
END;
$$;

CALL PUBLIC.collection_first_last();
SELECT * FROM DBMS_OUTPUT.DBMS_OUTPUT_LOG;

Result

DBMS OUTPUT
-----------
0
0
3
2
0
3
1

EXTEND

This method is used to append new elements to a Nested Table or a Varray. It has three possible variants:

• .EXTEND inserts a null element.

• .EXTEND(n) inserts 'n' null elements.

• .EXTEND(n, i) inserts 'n' copies of the element at 'i'.

Oracle

Query

CREATE OR REPLACE PROCEDURE collection_extend
IS 
    TYPE varray_typ IS VARRAY(5) OF INTEGER;
    TYPE nt_typ IS TABLE OF INTEGER;
    
    nt_variable1 nt_typ := nt_typ(1, 2, 3, 4);
    varray_variable1 varray_typ := varray_typ(1, 2, 3);
    varray_variable2 varray_typ := varray_typ(1, 2, 3);
BEGIN
    nt_variable1.EXTEND;
    varray_variable1.EXTEND(2);
    varray_variable2.EXTEND(2, 1);
    
    DBMS_OUTPUT.PUT_LINE(nt_variable1.COUNT);
    DBMS_OUTPUT.PUT_LINE(varray_variable1.COUNT);
    DBMS_OUTPUT.PUT_LINE(varray_variable2.COUNT);
END;
/

CALL collection_extend();

Result

DBMS OUTPUT
-----------
5
5
5

Snowflake

Note that a UDF was added to implement the functionality for the update of the element.

This UDF will be added in later revisions.

Query

CREATE OR REPLACE PROCEDURE PUBLIC.collection_first_last()
RETURNS VARCHAR
LANGUAGE SQL
EXECUTE AS CALLER
AS
$$
DECLARE
    nt_variable1 ARRAY := ARRAY_CONSTRUCT(1, 2, 3, 4);
    varray_variable1 ARRAY := ARRAY_CONSTRUCT(1, 2, 3);
    varray_variable2 ARRAY := ARRAY_CONSTRUCT(1, 2, 3);
BEGIN
    nt_variable1 := ARRAY_EXTEND_UDF(nt_variable);
    varray_variable1 := ARRAY_EXTEND_UDF(varray_variable1, 2);
    varray_variable2 := ARRAY_EXTEND_UDF(varray_variable2, 2, 1);
    
    CALL DBMS_OUTPUT.PUT_LINE(ARRAY_SIZE(nt_variable1);
    CALL DBMS_OUTPUT.PUT_LINE(ARRAY_SIZE(varray_variable1));
    CALL DBMS_OUTPUT.PUT_LINE(ARRAY_SIZE(varray_variable2));
END;
$$;

CALL PUBLIC.collection_first_last();
SELECT * FROM DBMS_OUTPUT.DBMS_OUTPUT_LOG;

Result

DBMS OUTPUT
-----------
5
5
5

TRIM

This method is used to remove the last elements from a Nested Table or a Varray. It has two possible variants:

• .TRIM removes the last element.

• .TRIM(n) removes the last 'n' elements.

This functionality may be implemented using ARRAY_SLICE

Oracle

Query

CREATE OR REPLACE PROCEDURE collection_trim
IS 
    TYPE varray_typ IS VARRAY(5) OF INTEGER;
    TYPE nt_typ IS TABLE OF INTEGER;
    
    varray_variable1 varray_typ := varray_typ(1, 2, 3);
    nt_variable1 nt_typ := nt_typ(1, 2, 3, 4);
BEGIN
    varray_variable1.TRIM;
    nt_variable1.TRIM(2);
    
    DBMS_OUTPUT.PUT_LINE(nt_variable1.COUNT);
    DBMS_OUTPUT.PUT_LINE(varray_variable1.COUNT);
END;
/

CALL collection_trim();

Result

DBMS OUTPUT
-----------
2
2

LIMIT

This method returns the maximum limit of a Varray.

This method is not supported in Snowflake.

Oracle

Query

CREATE OR REPLACE PROCEDURE collection_limit
IS 
    TYPE varray_typ1 IS VARRAY(5) OF INTEGER;
    TYPE varray_typ2 IS VARRAY(6) OF INTEGER;
    
    varray_variable1 varray_typ1 := varray_typ1(1, 2, 3);
    varray_variable2 varray_typ2 := varray_typ2(1, 2, 3, 4);
BEGIN
    DBMS_OUTPUT.PUT_LINE(varray_variable1.LIMIT);
    DBMS_OUTPUT.PUT_LINE(varray_variable2.LIMIT);
END;
/

CALL collection_limit();

Result

DBMS OUTPUT
-----------
5
6

PRIOR/NEXT

This method returns the prior/next index, given an index. If there is not a prior/next then it returns null. It is most frequently used to traverse a collection.

Oracle

Query

CREATE OR REPLACE PROCEDURE collection_prior_next
IS 
    TYPE varray_typ1 IS VARRAY(5) OF INTEGER;
    TYPE aa_typ IS TABLE OF INTEGER INDEX BY VARCHAR2(20);
    
    varray_variable1 varray_typ1 := varray_typ1(-1, -2, -3);
    associative_array1 aa_typ := aa_typ('abc'=>1, 'def'=>2, 'ghi'=>3, 'jkl'=>4);
BEGIN
    DBMS_OUTPUT.PUT_LINE(varray_variable1.PRIOR(1));
    DBMS_OUTPUT.PUT_LINE(varray_variable1.PRIOR(2));
    DBMS_OUTPUT.PUT_LINE(varray_variable1.NEXT(2));
    DBMS_OUTPUT.PUT_LINE(varray_variable1.NEXT(3));
    
    DBMS_OUTPUT.PUT_LINE(associative_array1.PRIOR('abc'));
    DBMS_OUTPUT.PUT_LINE(associative_array1.PRIOR('def'));
    DBMS_OUTPUT.PUT_LINE(associative_array1.NEXT('ghi'));
    DBMS_OUTPUT.PUT_LINE(associative_array1.NEXT('jkl'));
    DBMS_OUTPUT.PUT_LINE(associative_array1.PRIOR('not found'));
END;
/

CALL collection_prior_next();

Result

DBMS OUTPUT
-----------
        -- Empty spaces are due to null results
1
3


abc
jkl

jkl

Known Issues

1. Limit method is not supported in Snowflake

Snowflake does not have support for limited-space varrays. For this reason, this method is not supported.

Related EWIs

1. MSCEWI1058: Functionality is not currently supported by Snowflake Scripting.