Re: Adaptation in psycopg3

On Wed, 25 Nov 2020 at 12:29, listas <listas(at)soft-com(dot)es> wrote:

> After reading the docs i have a question about the parameters in the
> 'in' clause.
>
> In psycopg2 i do:
>
> params = (1,2,3,4)
> cursor.execute("select * from mytable where field1 in %s", (params,))
>
> or
>
> params = ('black','red','green')
> cursor.execute("select * from mytable where field2 in %s", (params,))
>
> What will it be like in psycopg3, will it be the same?, will I have to
> create a special adapter?

Hollo Oswaldo,

"IN" cannot be used, because it's a SQL construct, so "(1, 2, 3)" is
not something that postgres will understand as a parameter.

You can use "= any (%s)" and pass a list. This is something you can do
in psycopg2 too, and it's actually a better choice, because it works
with empty lists too, unless `IN ()`, which is a syntax error for
Postgres.

What you can do is:

params = ['black','red','green']
cursor.execute("select * from mytable where field2 = any(%s)", (params,))

interesting fact: "= any" is what postgres really uses internally,
even if you use the "IN ()" syntax:

piro=# explain select * from mytable where myint in (1,2,3);
QUERY PLAN
---------------------------------------------------------
Seq Scan on mytable (cost=0.00..45.06 rows=38 width=4)
Filter: (myint = ANY ('{1,2,3}'::integer[]))
(2 rows)

-- Daniele