Differences from psycopg2

psycopg3 uses the common DBAPI structure of many other database adapter and tries to behave as close as possible to psycopg2. There are however a few differences to be aware of.

Server-side binding

psycopg3 sends the query and the parameters to the server separately, instead of merging them client-side. PostgreSQL may behave slightly differently in this case, usually throwing an error and suggesting to use an explicit cast.

cur.execute("select '[10,20,30]'::jsonb -> 1").fetchone()
# returns (20,)

cur.execute("select '[10,20,30]'::jsonb -> %s", [1]).fetchone()
# raises an exception:
# UndefinedFunction: operator does not exist: jsonb -> numeric

cur.execute("select '[10,20,30]'::jsonb -> %s::int", [1]).fetchone()
# returns (20,)

PostgreSQL will also reject the execution of several queries at once (separated by semicolon), if they contain parameters. If parameters are used you should use distinct execute() calls; otherwise you may consider merging the query client-side, using psycopg3.sql module.

Certain commands cannot be used with server-side binding, for instance SET or NOTIFY:

>>> cur.execute("SET timezone TO %s", ["utc"])
psycopg3.errors.SyntaxError: syntax error at or near "$1"

Sometimes PostgreSQL offers an alternative (e.g. SELECT set_config(), SELECT pg_notify()). If no alternative exist you can use psycopg3.sql to compose the query client-side.

You cannot use IN %s and pass a tuple, because IN () is an SQL construct. You must use = any(%s) and pass a list. Note that this also works for an empty list, whereas an empty tuple would have resulted in an error.

Different adaptation system

The adaptation system has been completely rewritten, in order to address server-side parameters adaptation, but also to consider performance, flexibility, ease of customization.

Builtin data types should work as expected; if you have wrapped a custom data type you should check the Data adaptation configuration topic.

Copy is no more file-based

psycopg2 exposes a few copy methods to interact with PostgreSQL COPY. The interface doesn’t make easy to load dynamically-generated data to the database.

There is now a single copy() method, which is similar to psycopg2 copy_expert() in accepting a free-form COPY command and returns an object to read/write data, block-wise or record-wise. The different usage pattern also enables COPY to be used in async interactions.

See Using COPY TO and COPY FROM for the details.

with connection

When the connection is used as context manager, at the end of the context the connection will be closed. In psycopg2 only the transaction is closed, so a connection can be used in several contexts, but the behaviour is surprising for people used to several other Python classes wrapping resources, such as files.

callproc() is gone

cursor.callproc() is not implemented. The method has a simplistic semantic which doesn’t account for PostgreSQL positional parameters, procedures, set-returning functions. Use a normal execute() with SELECT function_name(...) or CALL procedure_name(...) instead.

What’s new in psycopg3


to be completed

  • asyncio support.

  • Several data types are adapted out-of-the-box: uuid, network, range, bytea, array of any supported type are dealt with automatically.

  • Access to the low-level libpq functions.