Writing stored procedures in Snowflake Scripting¶
This topic provides an introduction to writing a stored procedure in SQL by using Snowflake Scripting. For more information about Snowflake Scripting, see the Snowflake Scripting Developer Guide.
Introduction¶
To write a stored procedure that uses Snowflake Scripting:
Use the CREATE PROCEDURE or WITH … CALL … command with LANGUAGE SQL.
In the body of the stored procedure (the AS clause), you use a Snowflake Scripting block.
Note
If you are creating a Snowflake Scripting procedure in SnowSQL or the Classic Console, you must use string literal delimiters (
'
or$$
) around the body of the stored procedure.For details, see Using Snowflake Scripting in SnowSQL, the Classic Console, and Python Connector.
You can capture log and trace data as your handler code executes. For more information, refer to Logging, tracing, and metrics.
Note the following:
The same rules around caller’s rights vs. owner’s rights apply to these stored procedures.
The same considerations and guidelines in Working with stored procedures apply to Snowflake Scripting stored procedures.
The following is an example of a simple stored procedure that returns the value of the argument that is passed in:
CREATE OR REPLACE PROCEDURE output_message(message VARCHAR)
RETURNS VARCHAR NOT NULL
LANGUAGE SQL
AS
BEGIN
RETURN message;
END;
Note: If you are using SnowSQL, the Classic Console, or the
execute_stream
or execute_string
method in Python Connector
code, use this example instead (see Using Snowflake Scripting in SnowSQL, the Classic Console, and Python Connector):
CREATE OR REPLACE PROCEDURE output_message(message VARCHAR)
RETURNS VARCHAR NOT NULL
LANGUAGE SQL
AS
$$
BEGIN
RETURN message;
END;
$$
;
The following is an example of calling the stored procedure:
CALL output_message('Hello World');
The following is an example of creating and calling an anonymous stored procedure by using the WITH … CALL … command:
WITH anonymous_output_message AS PROCEDURE (message VARCHAR)
RETURNS VARCHAR NOT NULL
LANGUAGE SQL
AS
$$
BEGIN
RETURN message;
END;
$$
CALL anonymous_output_message('Hello World');
Note that in an anonymous stored procedure, you must use string literal delimiters ('
or $$
) around the body of the procedure.
Using arguments passed to a stored procedure¶
If you pass in any arguments to your stored procedure, you can refer to those arguments by name in any Snowflake Scripting expression. See the next sections for more details:
Simple example of using arguments passed to a stored procedure
Using an argument when building a string for a SQL statement
Simple example of using arguments passed to a stored procedure¶
The following stored procedure uses the values of the arguments in IF and RETURN statements.
CREATE OR REPLACE PROCEDURE return_greater(number_1 INTEGER, number_2 INTEGER)
RETURNS INTEGER NOT NULL
LANGUAGE SQL
AS
BEGIN
IF (number_1 > number_2) THEN
RETURN number_1;
ELSE
RETURN number_2;
END IF;
END;
Note: If you are using SnowSQL, the Classic Console, or the
execute_stream
or execute_string
method in Python Connector
code, use this example instead (see Using Snowflake Scripting in SnowSQL, the Classic Console, and Python Connector):
CREATE OR REPLACE PROCEDURE return_greater(number_1 INTEGER, number_2 INTEGER)
RETURNS INTEGER NOT NULL
LANGUAGE SQL
AS
$$
BEGIN
IF (number_1 > number_2) THEN
RETURN number_1;
ELSE
RETURN number_2;
END IF;
END;
$$
;
The following is an example of calling the stored procedure:
CALL return_greater(2, 3);
Using an argument in a SQL statement (binding)¶
As is the case with Snowflake Scripting variables, if you need to use an argument in a SQL statement, put a colon (:
) in front
of the argument name. (See Using a variable in a SQL statement (binding).)
For example, the following stored procedure uses the id
argument in the WHERE clause of a SELECT statement. In the WHERE
clause, the argument is specified as :id
.
CREATE OR REPLACE PROCEDURE find_invoice_by_id(id VARCHAR)
RETURNS TABLE (id INTEGER, price NUMBER(12,2))
LANGUAGE SQL
AS
DECLARE
res RESULTSET DEFAULT (SELECT * FROM invoices WHERE id = :id);
BEGIN
RETURN TABLE(res);
END;
Note: If you are using SnowSQL, the Classic Console, or the
execute_stream
or execute_string
method in Python Connector
code, use this example instead (see Using Snowflake Scripting in SnowSQL, the Classic Console, and Python Connector):
CREATE OR REPLACE PROCEDURE find_invoice_by_id(id VARCHAR)
RETURNS TABLE (id INTEGER, price NUMBER(12,2))
LANGUAGE SQL
AS
$$
DECLARE
res RESULTSET DEFAULT (SELECT * FROM invoices WHERE id = :id);
BEGIN
RETURN TABLE(res);
END;
$$
;
The following is an example of calling the stored procedure:
CALL find_invoice_by_id('2');
In addition, the TO_QUERY function provides a simple syntax for accepting a SQL string directly in the FROM clause of a SELECT statement. For a comparison of the TO_QUERY function with dynamic SQL, see Constructing SQL at runtime.
Using an argument as an object identifier¶
If you need to use an argument to refer to an object (e.g. a table name in the FROM clause of a SELECT statement), use the IDENTIFIER keyword to indicate that the argument represents an object identifier. For example:
CREATE OR REPLACE PROCEDURE get_row_count(table_name VARCHAR)
RETURNS INTEGER NOT NULL
LANGUAGE SQL
AS
DECLARE
row_count INTEGER DEFAULT 0;
res RESULTSET DEFAULT (SELECT COUNT(*) AS COUNT FROM IDENTIFIER(:table_name));
c1 CURSOR FOR res;
BEGIN
FOR row_variable IN c1 DO
row_count := row_variable.count;
END FOR;
RETURN row_count;
END;
Note: If you are using SnowSQL, the Classic Console, or the
execute_stream
or execute_string
method in Python Connector
code, use this example instead (see Using Snowflake Scripting in SnowSQL, the Classic Console, and Python Connector):
CREATE OR REPLACE PROCEDURE get_row_count(table_name VARCHAR)
RETURNS INTEGER NOT NULL
LANGUAGE SQL
AS
$$
DECLARE
row_count INTEGER DEFAULT 0;
res RESULTSET DEFAULT (SELECT COUNT(*) AS COUNT FROM IDENTIFIER(:table_name));
c1 CURSOR FOR res;
BEGIN
FOR row_variable IN c1 DO
row_count := row_variable.count;
END FOR;
RETURN row_count;
END;
$$
;
The following is an example of calling the stored procedure:
CALL get_row_count('invoices');
Using an argument when building a string for a SQL statement¶
Note that if you are building a SQL statement as a string to be passed to EXECUTE IMMEDIATE (see Assigning a query to a declared RESULTSET), do not prefix the argument with a colon. For example:
CREATE OR REPLACE PROCEDURE find_invoice_by_id_via_execute_immediate(id VARCHAR)
RETURNS TABLE (id INTEGER, price NUMBER(12,2))
LANGUAGE SQL
AS
DECLARE
select_statement VARCHAR;
res RESULTSET;
BEGIN
select_statement := 'SELECT * FROM invoices WHERE id = ' || id;
res := (EXECUTE IMMEDIATE :select_statement);
RETURN TABLE(res);
END;
Returning tabular data¶
If you need to return tabular data (e.g. data from a RESULTSET) from your stored procedure, specify RETURNS TABLE(…) in your CREATE PROCEDURE statement.
If you know the Snowflake data types of the columns in the returned table, specify the column names and types in the RETURNS TABLE().
CREATE OR REPLACE PROCEDURE get_top_sales()
RETURNS TABLE (sales_date DATE, quantity NUMBER)
...
Otherwise (e.g. if you are determining the column types during run time), you can omit the column names and types:
CREATE OR REPLACE PROCEDURE get_top_sales()
RETURNS TABLE ()
...
Note
Currently, in the RETURNS TABLE(...)
clause, you cannot specify GEOGRAPHY as a column type. This
applies whether you are creating a stored or anonymous procedure.
CREATE OR REPLACE PROCEDURE test_return_geography_table_1()
RETURNS TABLE(g GEOGRAPHY)
...
WITH test_return_geography_table_1() AS PROCEDURE
RETURNS TABLE(g GEOGRAPHY)
...
CALL test_return_geography_table_1();
If you attempt to specify GEOGRAPHY as a column type, calling the stored procedure results in the error:
Stored procedure execution error: data type of returned table does not match expected returned table type
To work around this issue, you can omit the column arguments and types in RETURNS TABLE()
.
CREATE OR REPLACE PROCEDURE test_return_geography_table_1()
RETURNS TABLE()
...
WITH test_return_geography_table_1() AS PROCEDURE
RETURNS TABLE()
...
CALL test_return_geography_table_1();
If you need to return the data in a RESULTSET, use TABLE() in your RETURN statement.
For example:
CREATE OR REPLACE PROCEDURE get_top_sales()
RETURNS TABLE (sales_date DATE, quantity NUMBER)
LANGUAGE SQL
AS
DECLARE
res RESULTSET DEFAULT (SELECT sales_date, quantity FROM sales ORDER BY quantity DESC LIMIT 10);
BEGIN
RETURN TABLE(res);
END;
Note: If you are using SnowSQL, the Classic Console, or the
execute_stream
or execute_string
method in Python Connector
code, use this example instead (see Using Snowflake Scripting in SnowSQL, the Classic Console, and Python Connector):
CREATE OR REPLACE PROCEDURE get_top_sales()
RETURNS TABLE (sales_date DATE, quantity NUMBER)
LANGUAGE SQL
AS
$$
DECLARE
res RESULTSET DEFAULT (SELECT sales_date, quantity FROM sales ORDER BY quantity DESC LIMIT 10);
BEGIN
RETURN TABLE(res);
END;
$$
;
The following is an example of calling the stored procedure:
CALL get_top_sales();
Calling a stored procedure from another stored procedure¶
In a stored procedure, if you need to call another stored procedure, use one of the following approaches:
Calling a stored procedure without using the returned value¶
Use a CALL statement to call the stored procedure (as you normally would).
If you need to pass in any variables or arguments as input arguments in the CALL statement, remember to use a colon (:
) in
front of the variable name. (See Using a variable in a SQL statement (binding).)
The following is an example of a stored procedure that calls another stored procedure but does not depend on the return value.
First, create a table for use in the example:
-- Create a table for use in the example.
CREATE OR REPLACE TABLE int_table (value INTEGER);
Then, create the stored procedure that you will call from another stored procedure:
-- Create a stored procedure to be called from another stored procedure.
CREATE OR REPLACE PROCEDURE insert_value(value INTEGER)
RETURNS VARCHAR NOT NULL
LANGUAGE SQL
AS
BEGIN
INSERT INTO int_table VALUES (:value);
RETURN 'Rows inserted: ' || SQLROWCOUNT;
END;
Note: If you are using SnowSQL, the Classic Console, or the
execute_stream
or execute_string
method in Python Connector
code, use this example instead (see Using Snowflake Scripting in SnowSQL, the Classic Console, and Python Connector):
-- Create a stored procedure to be called from another stored procedure.
CREATE OR REPLACE PROCEDURE insert_value(value INTEGER)
RETURNS VARCHAR NOT NULL
LANGUAGE SQL
AS
$$
BEGIN
INSERT INTO int_table VALUES (:value);
RETURN 'Rows inserted: ' || SQLROWCOUNT;
END;
$$
;
Next, create a second stored procedure that calls the first stored procedure:
CREATE OR REPLACE PROCEDURE insert_two_values(value1 INTEGER, value2 INTEGER)
RETURNS VARCHAR NOT NULL
LANGUAGE SQL
AS
BEGIN
CALL insert_value(:value1);
CALL insert_value(:value2);
RETURN 'Finished calling stored procedures';
END;
Note: If you are using SnowSQL, the Classic Console, or the
execute_stream
or execute_string
method in Python Connector
code, use this example instead (see Using Snowflake Scripting in SnowSQL, the Classic Console, and Python Connector):
CREATE OR REPLACE PROCEDURE insert_two_values(value1 INTEGER, value2 INTEGER)
RETURNS VARCHAR NOT NULL
LANGUAGE SQL
AS
$$
BEGIN
CALL insert_value(:value1);
CALL insert_value(:value2);
RETURN 'Finished calling stored procedures';
END;
$$
;
Finally, call the second stored procedure:
CALL insert_two_values(4, 5);
Using the value returned from a stored procedure call¶
If are calling a stored procedure that returns a scalar value and you need to access that value, use the
INTO :snowflake_scripting_variable
clause in the CALL statement to capture the value in a
Snowflake Scripting variable.
The following example calls the get_row_count
stored procedure that was defined in
Using an argument as an object identifier.
CREATE OR REPLACE PROCEDURE count_greater_than(table_name VARCHAR, maximum_count INTEGER)
RETURNS BOOLEAN NOT NULL
LANGUAGE SQL
AS
DECLARE
count1 NUMBER;
BEGIN
CALL get_row_count(:table_name) INTO :count1;
IF (:count1 > maximum_count) THEN
RETURN TRUE;
ELSE
RETURN FALSE;
END IF;
END;
Note: If you are using SnowSQL, the Classic Console, or the
execute_stream
or execute_string
method in Python Connector
code, use this example instead (see Using Snowflake Scripting in SnowSQL, the Classic Console, and Python Connector):
CREATE OR REPLACE PROCEDURE count_greater_than(table_name VARCHAR, maximum_count INTEGER)
RETURNS BOOLEAN NOT NULL
LANGUAGE SQL
AS
$$
DECLARE
count1 NUMBER;
BEGIN
CALL get_row_count(:table_name) INTO :count1;
IF (:count1 > maximum_count) THEN
RETURN TRUE;
ELSE
RETURN FALSE;
END IF;
END;
$$
;
The following is an example of calling the stored procedure:
CALL count_greater_than('invoices', 3);
If the stored procedure returns a table, you can capture the return value by setting a RESULTSET to a string containing the CALL statement. (See Assigning a query to a declared RESULTSET.)
To retrieve the return value from the call, you can use a CURSOR for the RESULTSET. For example:
DECLARE
res1 RESULTSET;
BEGIN
res1 := (CALL my_procedure());
LET c1 CURSOR FOR res1;
FOR row_variable IN c1 DO
IF (row_variable.col1 > 0) THEN
...;
ELSE
...;
END IF;
END FOR;
...