edgedb-dto 1.0.2

Edgedb DTO is a library that generates DTO pydantic dataclasses from edgeql files.

Edgedb DTO

  •   Example : Simple   •   Example : Linked DTOS   •  

Introduction

EdgeDB DTO is a CLI tool for Python that generates pydantic dataclasses for edgeql queries (EdgeDB query language).

Table of Contents

• Python Code
• Features
• Installation
• Example pipeline
• CLI Options
• Dependencies
• Limitations
• Contributors
• License

Python Code

# main.py
from simple.queries.dto import InsertPerson
from edgedb import create_client

CLIENT = create_client()

def insert_three_people():
    rob = InsertPerson(name="Rob")
    alice = InsertPerson(name="Alice", best_friend=rob)
    tom = InsertPerson(name="Tom", best_friend=alice)
    tom.run(executor=CLIENT, transaction=True)

insert_three_people()

Features

• Mix and Match multiple Classes: Missed ORMs while using EdgeDB in Python? Look no further! edgedb-dto supports embedding DTOs in each other.
• Easy, intuitive and in 3 steps:
  • Write .edgeql queries.
  • run edgedb-py && edgedb-dto in the terminal
  • Import generated classes
• Validation using pydantic: edgedb-dto generates pydantic dataclasses that you can modify as you wish to validate your input data.
• Asynchronous and Synchronous Support: Supports both asynchronous and synchronous database calls.

Installation

To install EdgeDB DTO from PyPI, run:

pip install edgedb-dto

Example Pipeline

Define your database

Project Tree

Schema

We have the following simple schema:

# default.esdl
module default {
    type Person{
        required name: str{
            constraint exclusive;
        }
        best_friend: Person;
    }
}

EdgeQL Query

We also have the following simple .edgeql query:

# insert_person.edgeql
insert Person{
    name := $name,
    best_friend := $best_friend
} unless conflict on .name else (
    select Person filter .name = $name
)

Generate DTO Classes

Let's do some Magic ✨

edgedb-py --no-skip-pydantic --target=blocking && edgedb-dto

PS : To do async calls just remove --target option (run edgedb-py --help for more context)

New Project Tree

Generated DTO Class

# insert_person_edgeql_dto.py
from ..insert_person_edgeql import InsertPersonResult, insert_person

@dataclass
class InsertPerson(DTO):
    name: str
    best_friend: DTO | uuid.UUID | None = None

    def _query(self, **kwargs):
        return insert_person(**kwargs)

    def run(self, executor: Client, transaction: bool = False
) -> InsertPersonResult | None:
        return self._run(executor, transaction)

Using our DTO Classes

Now let's populate our database with ease:

# main.py
from simple.queries.dto import InsertPerson
from edgedb import create_client

CLIENT = create_client()

def insert_three_people():
    rob = InsertPerson(name="Rob")
    alice = InsertPerson(name="Alice", best_friend=rob)
    tom = InsertPerson(name="Tom", best_friend=alice)
    tom.run(executor=CLIENT, transaction=True)

insert_three_people()

By calling run method on the tom DTO class, we managed to insert all of our data into the database in a single transaction.

CLI Options

• --source-directory (-s) [Optional argument]: Source directory containing EdgeQL files.
  Default: Edgedb-DTO will scan for the fiels generated by edgedb-py.
• --output-directory (-o) [Optional argument]: Output directory for generated DTO classes.
  Default: the files will be generated in a folder named dto in the parent folder of the EdgeQL Python files.
  Please note that it's better for type safety to not specify the output-directory

Dependencies

• python: version=^3.11 is required to run this library.
• jinja2: For templating the DTO classes.
• pydantic: For data validation and settings management.
• edgedb: For interacting with the EdgeDB database.

Limitations

N + 1 Problem

edgedb-dto does not act like a query builder. So the more DTOs you link, the more back and forth calls are made to the database. In the last example, 3 queries to the database are made to complete the transaction.

Contributors

• Mohamed SAHNOUN
• Jean-Gabriel BARTHELEMY

License

This project is licensed under the MIT License - see the LICENSE file for details.