How can I insert data using a plpgsql function?

Question

I'm trying to insert data into a table using a plpgsql function or stored procedure. However, I want to insert multiple records at once. I now these will all be VARCHAR so I thought I could use a function formed like function(tablename VARCHAR, records VARCHAR[][]). But then I found out multidimensional array support in plpgsql isn.t quite so fantastic.

This is what my function looks like at the moment. This doesn't produce the result I'm looking for. When I ask

SELECT insert_data('tennis', ARRAY[ARRAY['1','2'], ARRAY['3','4']])

I get the following error

ERROR:  syntax error at or near "{"
LINE 1: INSERT INTO tennis VALUES (null, {{1}}), (null, {{3}});
                                         ^
QUERY:  INSERT INTO tennis VALUES (null, {{1}}), (null, {{3}});
CONTEXT:  PL/pgSQL function "insert_data" line 26 at EXECUTE statement

However I am expecting a query like

INSERT INTO tennis VALUES (null, '1', '2'), (null, '3', '4');

which would work because table tennis has this structure.

CREATE OR REPLACE FUNCTION insert_data (dsetname_in VARCHAR, records VARCHAR[][])
RETURNS BOOLEAN AS $PROC$
DECLARE
    insertquery TEXT;
    val VARCHAR;
    i INT;
    j INT;
BEGIN
    insertquery = $$INSERT INTO $$ || dsetname_in || $$ VALUES $$;
    FOR i IN array_lower(records, 1)..array_upper(records, 1)
    LOOP
        insertquery = insertquery || $$(null, $$;
        FOR j IN array_lower(records[i:i], 1)..array_upper(records[i:i], 1)
        LOOP
            val = records[i:i][j:j];
            insertquery = insertquery || val;
            IF j <> array_upper(records[i:i], 1) THEN
                insertquery = insertquery || $$, $$;
            END IF;
        END LOOP;
        insertquery = insertquery || $$)$$;
        IF i <> array_upper(records, 1) THEN
            insertquery = insertquery || $$, $$;
        END IF;
    END LOOP;
    insertquery = insertquery || $$;$$;
    EXECUTE insertquery;
    RETURN TRUE;
END;$PROC$ LANGUAGE 'plpgsql';

Answer 1

I'm dubious about the value of this whole approach because I don't see that it adds any useful level of abstraction; but if you must do it, and all your values are character strings, I think the cleanest approach is this:

CREATE OR REPLACE FUNCTION insert_data(dsetname_in text, records text[])
RETURNS VOID LANGUAGE plpgsql AS $PROC$
DECLARE
  insertquery text;
  i int;
BEGIN
  insertquery := 'INSERT INTO ' || dsetname_in || ' VALUES ';
  FOR i IN array_lower(records, 1)..array_upper(records, 1)
  LOOP
    insertquery := insertquery || '(null, ' || records[i] || '),';
  END LOOP;
  insertquery := left(insertquery, char_length(insertquery) - 1);
  EXECUTE insertquery;
END;
$PROC$;

You can then call it like this, which seems cleaner than what you showed for the nested arrays:

SELECT insert_data('tennis',
                   ARRAY[$$'1','2'$$,
                         $$'3','4'$$]);