Content-Type: text/plain;
	charset=utf-8
Mime-Version: 1.0 (Mac OS X Mail 16.0 \(3776.700.51\))
Subject: Re: re-novice coming back to pgsql: porting an SQLite update
 statement to postgres
From: Alban Hertroys <haramrae@gmail.com>
In-Reply-To: <5c86256a4e83b1af73da22faab54889564e93129.camel@kortschak.io>
Date: Sat, 14 Sep 2024 12:05:50 +0200
Cc: Dominique Devienne <ddevienne@gmail.com>,
 Adrian Klaver <adrian.klaver@aklaver.com>,
 Vincent Veyron <vv.lists@wanadoo.fr>,
 PostgreSQL General <pgsql-general@postgresql.org>
Content-Transfer-Encoding: quoted-printable
Message-Id: <7320B905-E950-4035-9A13-8C9447AB52EC@gmail.com>
References: <1aea530e4c96109b91fdde53a29e38beb625dc77.camel@kortschak.io>
 <20240723221121.51f2b3872d0ebfc36a6fa8ff@wanadoo.fr>
 <2f11b2ba-3182-492a-ab46-23cfa5ec913c@aklaver.com>
 <CAFCRh-8d-gsERdu-6W2fb0d6HLi8_bPkS7zVqm0J4ZakpvJQYw@mail.gmail.com>
 <735913e4c9601a2a7ad57d2253d74451c42ba899.camel@kortschak.io>
 <21078220628bcebca0395a52d42b558ed965bbc0.camel@kortschak.io>
 <5c86256a4e83b1af73da22faab54889564e93129.camel@kortschak.io>
To: Dan Kortschak <dan+pgsql@kortschak.io>
Archived-At: <https://www.postgresql.org/message-id/7320B905-E950-4035-9A13-8C9447AB52EC%40gmail.com>
Precedence: bulk


> On 14 Sep 2024, at 10:33, Dan Kortschak <dan+pgsql@kortschak.io> =
wrote:

(=E2=80=A6)

> I'm still having difficulties with the second part which is to update
> the contents of the amend array in the JSON.
>=20
> So far I'm able to append the relevant details to the append array, =
but
> I'm unable to correctly select the corrects elements from the $6
> argument, which is in the form
> [{"start":<RFC3339>,"end":<RFC3339>,"data":<object>}, ...]. The first
> update statement gives me broadly what I want, but includes elements =
of
> the array that it shouldn't.

(=E2=80=A6)

> If I filter on the start and end time, I end up with no element coming
> through at all and the "replace" field ends up null.
>=20
> update
> events
> set
> datastr =3D jsonb_set(
> datastr,
> '{amend}',
> datastr->'amend' || jsonb_build_object(
> 'time', $2::TEXT,
> 'msg', $3::TEXT,
> 'replace', (
> select *
> from
> jsonb($6::TEXT) as replacement
> where
> (replacement->>'start')::TIMESTAMP WITH TIME ZONE < endtime and
> (replacement->>'end')::TIMESTAMP WITH TIME ZONE > starttime
> )
> )
> )
> where
> starttime < $5 and
> endtime > $4 and
> bucketrow =3D (
> select rowid from buckets where id =3D $1
> );

That=E2=80=99s because the replacement data is an array of objects, not =
a single object.

You need to iterate through the array elements to build your replacement =
data, something like what I do here with a select (because that=E2=80=99s =
way easier to play around with):

with dollar6 as (
select jsonb($$[
                                                {
                                                        "data": { "foo": =
1, "bar": 2
                                                        },
                                                        "end": =
"2023-06-12T19:54:51Z",
                                                        "start": =
"2023-06-12T19:54:39Z"
                                                }
                                        ]$$::text) replacement
)
select *                               =20
from dollar6
cross join lateral jsonb_array_elements(replacement) r
where (r->>'start')::timestamptz <=3D current_timestamp;


There are probably other ways to attack this problem, this is the one I =
came up with.


Alban Hertroys
--
If you can't see the forest for the trees,
cut the trees and you'll find there is no forest.