Easy way to use python dict as postgresql composite

[Piatachock]

Is there a way to do subj?
In Postgre, I have a table with composite type, like this:

CREATE TYPE user_info AS (
    name   text,
    surname   text,
    alias text
);

CREATE TABLE users AS (
    user_id bigserial,
    info user_info
);

In Python, I have dict reflecting composite type, like this:

data = {
    'name': 'John',
    'surname': 'Doe',
    'alias': 'Mr. Example',
}

I can't use this dict in query directly:

conn = asyncpg.connect(...)
conn.execute('INSERT INTO users (user_info) VALUES ($1)', data)  # WRONG

So I have to convert it to tuple, like this:

tuple_data = data['name'], data['surname'], data['alias']
conn.execute('INSERT INTO users (user_info) VALUES ($1)', tuple_data)

But that conversion requires me to mimic field order of postgre composite in python code, or else inserted data will be messed up.
Is there a way to insert composite column represented as key-value pairs in your code without using explicit knowledge of composite's field order?

As far as I understand, asyncpg has introspection of postgre types anyway, so maybe field order of composite types is presented inside asyncpg meta information already - so that asyncpg clients can use it?

[elprans]

This seems useful. I implemented support for mappings as composite input in #351.