gel-pydantic-codegen

This tool generates Python typesafe async code for EdgeQL queries using Pydantic V2.

The generated models can be directly used with other libraries such as FastAPI.

This is an alternative to the built-in code generator of the official gel-python library.

For legacy EdgeDB <v6 support, check the last edgedb-pydantic-codegen release and its PyPI page.

Usage

In an Gel initialized project, simply run

uvx gel-pydantic-codegen <directory>

All *.edgeql files in <directory> and its subdirectories will be processed and the generated code saved next to them.

Generated code example

from enum import StrEnum

from gel import AsyncIOExecutor
from pydantic import BaseModel, TypeAdapter

EDGEQL_QUERY = r"""
with
  discord_id := <int64>$discord_id,
  moecoins := <optional int32>$moecoins,
  blood_shards := <optional int32>$blood_shards,
  updated := (
    update waicolle::Player
    filter .client = global client and .user.discord_id = discord_id
    set {
      moecoins := .moecoins + (moecoins ?? 0),
      blood_shards := .blood_shards + (blood_shards ?? 0),
    }
  )
select updated {
  game_mode,
  moecoins,
  blood_shards,
  user: {
    discord_id,
    discord_id_str,
  },
}
"""


class WaicolleGameMode(StrEnum):
    WAIFU = "WAIFU"
    HUSBANDO = "HUSBANDO"
    ALL = "ALL"


class PlayerAddCoinsResultUser(BaseModel):
    discord_id: int
    discord_id_str: str


class PlayerAddCoinsResult(BaseModel):
    game_mode: WaicolleGameMode
    moecoins: int
    blood_shards: int
    user: PlayerAddCoinsResultUser


adapter = TypeAdapter(PlayerAddCoinsResult | None)


async def player_add_coins(
    executor: AsyncIOExecutor,
    *,
    discord_id: int,
    moecoins: int | None = None,
    blood_shards: int | None = None,
) -> PlayerAddCoinsResult | None:
    resp = await executor.query_single_json(
        EDGEQL_QUERY,
        discord_id=discord_id,
        moecoins=moecoins,
        blood_shards=blood_shards,
    )
    return adapter.validate_json(resp, strict=False)

Caveats

Currently this tool does not support:

• TupleType, RangeType and MultiRangeType collections
• std::duration, cal::relative_duration, cal::date_duration, cfg::memory and ext::pgvector::vector objects