trigger_is(): allow testing function arguments

[decibel]

It'd be good if we could verify the arguments (or at least number of arguments) being passed to a trigger function by a specific trigger. According to [1], The arguments are concatinated together as NUL terminated strings in pg_trigger.tgargs.

1: http://www.postgresql.org/docs/devel/static/catalog-pg-trigger.html

[rodo]

Actually trigger_is is defined as is :

SELECT trigger_is( :schema, :table, :trigger, :func_schema, :function, :description );
SELECT trigger_is( :schema, :table, :trigger, :func_schema, :function );
SELECT trigger_is( :table, :trigger, :function, :description );
SELECT trigger_is( :table, :trigger, :function );

What about adding these functions to detect the number of argument ?

SELECT trigger_is( :schema, :table, :trigger, :func_schema, :function, nbargs, :description );
SELECT trigger_is( :schema, :table, :trigger, :func_schema, :function, nbargs );
SELECT trigger_is( :table, :trigger, :function, :description, nbargs );
SELECT trigger_is( :table, :trigger, :function, nbargs );

[decibel]

That would be better than nothing.

Another option would be to use the trigger definition parsing code I wrote for cat_tools.

[rodo]

I'll have a look at your function, as your code is under MIT License too it's possible to use it in pgtap, it's better to keep pgtap with less dependencies as possible.

[theory]

Jesus, trigger parsing code? In PL/pgSQL? o_O

[decibel]

Well, I wanted it to be portable... It's not quite as bad as you might think.

If we didn't care about portable then I think the best way to handle this would be a C function that calls (or duplicates) the appropriate bits of backend code. I don't think that some other PL language buys us much over the plpgsql, and the C code means nothing's actually being parsed.

[theory]

I'm sure it's no worse than some of the crazy shit PL/pgSQL already does, like trying to distinguish a string that executes a prepared statement from any other query. Fine with me to integrate that code if you're willing to contribute it, @decibel.

[decibel]

It's MIT license, so have at it.