Search
ctrl+/
Ask AI
ctrl+.
Light
Dark
System
Sign in

Code Generation​

The gel-python package exposes a command-line tool to generate various kinds of type-safe code from EdgeQL queries and your schema:

  • queries: typesafe functions from *.edgeql files, using dataclasses for objects primarily.

  • models: a programmatic query-builder and Pydantic-based models generator.

Copy
$ 
uvx gel generate py/queries
Copy
$ 
uvx gel generate py/models


The gel generate commands supports the same set of connection options as the gel CLI.


-I, --instance <instance>
--dsn <dsn>
--credentials-file <path/to/credentials.json>
-H, --host <host>
-P, --port <port>
-b, --branch <branch>
-u, --user <user>
--password
--password-from-stdin
--tls-ca-file <path/to/certificate>
--tls-security <insecure | no_host_verification | strict | default>




Queries​


Consider a simple query that lives in a file called get_number.edgeql:


Copy
select <int64>$arg;


Running the code generator will generate a new file called get_number_async_edgeql.py containing the following code (roughly):


Copy
from __future__ import annotations
import gel


async def get_number(
    client: gel.AsyncIOClient,
    *,
    arg: int,
) -> int:
    return await client.query_single(
        """\
        select <int64>$arg\
        """,
        arg=arg,
    )




Target​


By default, the generated code uses an async API. The generator supports additional targets via the --target flag.


Copy
$ 
gel generate py/queries --target async        # generate async function (default)
Copy
$ 
gel generate py/queries --target blocking     # generate blocking code


The names of the generated files will differ accordingly: {query_filename}_{target}_edgeql.py.






Single-file mode​


It may be preferable to generate a single file containing all the generated functions. This can be done by passing the --file flag.


Copy
$ 
gel generate py/queries --file


This generates a single file called generated_{target}_edgeql.py in the root of your project.








Models​


The models generator will generate Pydantic classes and a programmatic query builder. It reflects your full schema, as well as our standard library into functions and Pydantic classes which we've enhanced to make a truly powerful type-safe programmatic data layer.


Copy
import datetime
from models import User, std
from gel import create_client

def main():
    client = create_client()

    # Create a new User instance and save it to the database
    bob = User(name='Bob', dob=datetime.date(1984, 3, 1))
    client.save(bob)

    # Select all Users
    users = client.query(User)

    # Select all users with names like "Bob"
    bob_like = client.query(User.filter(lambda u: std.ilike(u.name, '%bob%')))

    # Update an object
    bob.name = 'Robert'
    client.save(bob)

    # Delete an object
    client.execute(User.filter(id=bob.id).delete())

    client.close()

if __name__ == '__main__':
    main()