I'm constructing a snowflake stored procedure and I'm facing difficulty in using the passed argument in the snowflake procedure.
create or replace procedure dumper(n float)
returns float
language javascript
execute as caller
as
$$
var text = "select file_name from table(information_schema.COPY_HISTORY(TABLE_NAME=> 'records', start_time=> dateadd(hours, ?, current_timestamp())));";
var statement = snowflake.createStatement({sqlText: text, binds: [n]});
var result = statement.execute();
return statement.getRowCount();
$$
;
attempting to call the above procedure
call dumper(-2);
result in the following error
JavaScript execution error: Uncaught ReferenceError: n is not defined in DUMPER at ' var statement = snowflake.createStatement({sqlText: text, binds: [n]});' position 70 stackstrace: DUMPER line: 4
I tried using the interpolation one discussed over here but that too had no success.
Any clue on how to work with passed argument.
You have to capitalize "N" in your JavaScript code:
var statement = snowflake.createStatement({sqlText: text, binds: [N]});
Variables passed into Snowflake stored procedures behave like other object names until they're inside the JavaScript. If they're not double quoted, then Snowflake implicitly capitalizes them. Remember to uppercase all parameters passed into SPs and UDFs in Snowflake. Variables defined inside the SP or UDF using JavaScript follow the normal rules for the language.
Since the regular rules apply to Snowflake identifiers as they do to variables passed into procedures and functions, you can double quote parameters if you want to use lower or mixed case variable names:
create or replace function echo_string("n" string)
returns string
language javascript
as
$$
return n; // This works because "n" is double quoted in the signature
$$;
select echo_string('Hello world.');
Related
I tried substring method but it is not working and giving me this error.
create or replace procedure
dsa ()
returns varchar
language javascript
$$
var b="{asdfasdf}"
var c=b.substring(1);
return c;
$$;
call dsa();
SQL compilation error: syntax error line 5 at position 0 unexpected '$$ var b="{asdfasdf}" var c=b.substring(1); return c; $$'.
You're missing the keyword AS in your stored procedure definition. You also need to define two parameters on the overloaded JavaScript function for substring. Finally, are you sure you want to use a stored procedure for this instead of a UDF or just SQL?
create or replace procedure
dsa ()
returns varchar
language javascript
as
$$
var b="{asdfasdf}"
var c=b.substring(0,1);
return c;
$$;
call dsa();
My code obtains the column list of a table I have created
the attributes of this table are the contextual values of a session in snowflake
such as USER,DATABASE,WAREHOUSE...ETC
Afterwards it places those attribute names into an array
I then try to call on these names in making an insert query and this is where I am struggling with the syntax
Because each value in my array is USER, DATABASE,WAREHOUSE
I am trying to call on the context functions
like CURRENT_WAREHOUSE()
Can someone please help me with the syntax
for(i=0;i<arr.length;i++){
v_sqlCode = `INSERT INTO SESSION_ATTRIBUTES( arr[i] )
"VALUES ("CALL CURRENT_"+arr[i]+"()")';
}
You can't directly use the output of a Snowflake stored procedure programmatically. If you need to use the output value, you have to collect it using the RESULT_SCAN table function. That can get a bit complex to handle, directly in code, so it's far better to place it into a helper function. See this example of one SP calling another one and using its output value. It does this by calling the SP using the callSP helper function. Use the entire SQL statement including the call command into the SQL parameter for that function. It will run the SP, grab the result from the result_scan table function and return it.
create or replace procedure HELLO_WORLD()
returns string
language javascript
as
$$
return "Hello, world.";
$$;
create or replace procedure CALL_HELLO_WORLD()
returns string
language javascript
execute as caller
as
$$
return callSP(`call HELLO_WORLD()`);
// ---- Main function above, helper functions below.
function callSP(sql){
let cmd = {sqlText: sql};
let stmt = snowflake.createStatement(cmd);
stmt.execute();
let result_scan = `select $1 from table(result_scan(last_query_id()));`;
let result_cmd = {sqlText: result_scan};
let result_stmt = snowflake.createStatement(result_cmd);
let rs = stmt.execute();
if(rs.next()) {
return rs.getColumnValue(1);
} else {
return null;
}
}
$$;
call call_hello_world();
I have to add a variable MaxDate in my SQL Stored Proc (shown below). The code gets errored out since MaxDate is not represented by its value. Any idea on how I can pass a variable in a stored proc?
create or replace procedure Load_Employee()
returns varchar not null
language javascript
EXECUTE AS CALLER
as
$$
//Variable Initialization
var IntegrationTable ='EMPLOYEE';
var TypeID=0;
var MaxDate=' ';
var cmd = "Select max(COMPLETED_DATE) from SCHEMA.TABLE where TARGET_TABLE_NAME= " + "'" + IntegrationTable + "'" ;
var sql = snowflake.createStatement({sqlText: cmd});
var result = sql.execute();
result.next();
MaxDate=result.getColumnValue(1);
var cmd=` Insert into PersonTable
select SHA1(concat(Person_id,'|','Person')) ,12345678,SHA1(concat('Payroll','|','Pay','|', Load_Date)) ,current_timestamp() , Tenant
from Schema.PERSONTABLE where Date_Added >= MaxDate
where TYPE='ABC' ;`;
$$
;
If your query to get MaxDate works right, then the value should be in the variable. The problem is it's not being replaced in the sql variable defining the insert statement.
Since you're using backticks to open and close the string, you can use a special JavaScript notation to replace the variable with its value, ${MaxDate}.
Your definition of the insert statement would look like this:
var cmd=` Insert into PersonTable
select SHA1(concat(Person_id,'|','Person')) ,12345678,SHA1(concat('Payroll','|','Pay','|', Load_Date)) ,current_timestamp() , Tenant
from Schema.PERSONTABLE where Date_Added >= ${MaxDate}
where TYPE='ABC' ;`;
If that doesn't work, try cutting the SP short with return MaxDate; to see what got assigned to that variable. Also it's very helpful to check the query history view to see what SQL actually ran inside a stored procedure.
Also, I think this is the same SP that was having an issue with a null return. You'll need to return a string value using something like return 'Success'; or something to avoid getting an error for the null return. That's because of the returns varchar not null in the definition.
I am copying data from a Snowflake table into an S3 external stage:
COPY INTO '#my_stage/my_folder/my_file.csv.gz' FROM (
SELECT *
FROM my_table
)
However this code runs daily and I don't want to overwrite my_file.csv.gz but rather keep all the historical versions. However I haven't found a way to create dynamic paths:
SET stage_name=CONCAT('#my_stage/my_folder/my_file', '_date.csv.gz');
COPY INTO $stage_name FROM (
SELECT *
FROM my_table
);
COPY INTO IDENTIFIER($stage_name) FROM (
SELECT *
FROM my_table
);
None of the later 2 queries work!
My question: How can I create dynamic Stage paths in Snowflake? Thanks
Here's a stored procedure you can use and modify. Note that the line with the comment to modify your copy into statement uses backticks instead of single or double quotes. In JavaScript, that allows use of single or double quotes in the string, multi-line constants, and replacement tokens in the form ${variable_name}
create or replace procedure COPY_TO_STAGE(PATH string)
returns variant
language javascript
as
$$
class Query{
constructor(statement){
this.statement = statement;
}
}
// Start of main function
var out = {};
// Change your copy into statement here.
var q = getQuery(`copy into '${PATH}' from (select * from my_table);`);
if (q.resultSet.next()) {
out["rows_unloaded"] = q.resultSet.getColumnValue("rows_unloaded");
out["input_bytes"] = q.resultSet.getColumnValue("input_bytes");
out["output_bytes"] = q.resultSet.getColumnValue("output_bytes");
} else {
out["Error"] = "Unknown error";
}
return out;
// End of main function
function getQuery(sql){
cmd1 = {sqlText: sql};
var query = new Query(snowflake.createStatement(cmd1));
query.resultSet = query.statement.execute();
return query;
}
$$;
Once you define it, you can use SQL variables as the input if you want:
SET stage_name=CONCAT('#my_stage/my_folder/my_file', '_date.csv.gz');
call copy_to_stage($stage_name);
This won't work. Unfortunately using variables for identifiers does not work for stages. You might need to create a Stored procedure with Dynamic SQL:
https://docs.snowflake.com/en/sql-reference/stored-procedures-usage.html#label-example-of-dynamic-sql-in-stored-procedure
So you can just call this procedure every day or generating a SP with several parameters for the path (Stage), the query which will be executed and the target filename.
Is there 'concat' function in GreenPlum? I can use concat function in postgresql and it works well, but when i use it in Greenplum, I got an error.
select concat('a', 'b');
ERROR: function concat(unknown, unknown) does not exist at character 8
HINT: No function matches the given name and argument types. You may need to add explicit type casts.
LINE 1: select concat('a', 'b');
^
Is there some other functions can instead of 'concat' function in GreenPlum? And I have tried to create a function to instead of it, but got some syntax errors also.
CREATE OR REPLACE FUNCTION my_concat(VARIADIC arr VARCHAR[] ) RETURNS VARCHAR AS $$ SELECT array_to_string(arr, ''); $$ LANGUAGE SQL;
ERROR: syntax error at or near "VARCHAR" at character 51
LINE 1: CREATE OR REPLACE FUNCTION my_concat(VARIADIC arr VARCHAR[] ...
^
Anyone can help? Thanks very much!
Like most databases, Greenplum uses "||" to concatenate two strings together.
SELECT 'Green' || 'plum';
Result:
Greenplum
its a versional issue , you have use || in place where ever u using contact function.
Greenplum doesn't have the concat function yet. May be you can modify your code to use "||" instead of concat.
Well,
First I agree that you should replace your code to use the correct SQL syntax '||' for concatenation.
If you really want to create a function to emulate the concat, you could do something like:
create or replace function myschema.concat(arg1 text, arg2 text)
returns text as
$body$
declare
v_arg1 text;
v_arg2 text;
begin
v_arg1 := arg1;
v_arg2 := arg2;
return v_arg1 || v_arg2;
end
$body$
language plpgsql volatile;
Then, the query will work:
select myschema.concat('test1', 'test2');
>>test1test2
Hope you are looking for the below query.
gpadmin=# CREATE OR REPLACE FUNCTION my_concat( character varying[] ) RETURNS VARCHAR AS $$ SELECT array_to_string($1, ''); $$ LANGUAGE SQL;
gpadmin=# select my_concat(ARRAY['Green','plum']);
my_concat
Greenplum