Combined Pydantic settings class from multiple settings classes with individual environment files

Question

Let's say I have 2 simple BaseSettings classes that, each, load values from their individual environment files. Let's also say that there is a combined settings class that is based on the first 2 classes.

Settings = DatabaseSettigns + AuthSettings

How can I create the combined settings class with a flat namespace without having to do dynamic class creation or dynamic attribute value assignments, i.e., I want my IDE to be able to do auto-complete, etc by knowing what the attributes are.

When I use the example below, I ONLY get the values from the LAST class, i.e, AuthSettings, instead of BOTH of them.

How can I have it such that ALL component settings classes are properly read / initialized from their respective files?

import logging
log = logging.getLogger(__name__)
logging.basicConfig(level=logging.INFO,
                    format='%(message)s')

from pydantic import BaseModel
from pydantic_settings import BaseSettings, SettingsConfigDict

class DatabaseSettings(BaseSettings):
  model_config = SettingsConfigDict(
    env_prefix="APP_",
    env_file=".env.database",
    env_file_encoding="utf-8",
    extra="ignore",
  )
  db_host: str = "localhost"

class AuthSettings(BaseSettings):
  model_config = SettingsConfigDict(
    env_prefix="APP_",
    env_file=".env.auth",
    env_file_encoding="utf-8",
    extra="ignore",
  )
  auth_secret_key: str = "change-me"

class Settings(DatabaseSettings, AuthSettings):
  model_config = SettingsConfigDict(
    env_prefix='APP_',
    extra='ignore'
  )
  pass

settings = Settings()
log.info(f"Settings: {settings}")

# .env.database
APP_DB_HOST=db.example.com

# .env.auth
APP_AUTH_SECRET_KEY=secret-from-env-file

Current Result

Settings: auth_secret_key='secret-from-env-file' db_host='localhost'

Desired Result

Settings: auth_secret_key='secret-from-env-file' db_host='db.example.com'
# notice how the db_host value is the one read from the file instead of the default "localhost"

Mauricio Reisdoefer · Accepted Answer · 2025-11-24 20:33:31Z

3

You cannot automatically flatten multiple BaseSettings submodels into a single namespace in Pydantic while keeping composition pure (my previous answer). You can still use inheritance, but keep in mind that the model_config of the subclass (Settings) overwrites the configs of the parent classes. To fix your original problem, define all env_files in the final class using a list:

class Settings(DatabaseSettings, AuthSettings):
    model_config = SettingsConfigDict(
        env_prefix="APP_",
        env_file=[".env.database", ".env.auth"],
        env_file_encoding="utf-8",
        extra="ignore",
    )

answered Nov 24 at 20:33

Mauricio Reisdoefer

3587 bronze badges

New contributor

Sign up to request clarification or add additional context in comments.

Comments

Mauricio Reisdoefer · Accepted Answer · 2025-11-24 18:01:04Z

2

You can use composition instead of inheritance, which is more explicit, avoids conflicts between multiple BaseSettings classes, and preserves IDE auto-complete. It works like this:

class Settings(BaseModel):  # Doesn't inherit BaseSettings
    database: DatabaseSettings = DatabaseSettings()
    auth: AuthSettings = AuthSettings()

Why prefer composition?

Each settings class loads its own .env file correctly
No conflicts from multiple inheritance
Maintains clear structure and full IDE support
Easier to extend and maintain in the future

answered Nov 24 at 18:01

Mauricio Reisdoefer

3587 bronze badges

New contributor

5 Comments

HanSooloo Nov 24 at 18:02

I have seen that answer before, but I would like to keep a flat namespace. Is that possible?

Mauricio Reisdoefer Nov 24 at 18:16

You cannot automatically flatten multiple BaseSettings submodels into a single namespace in Pydantic while keeping composition pure. You can still use inheritance, but keep in mind that the model_config of the subclass (Settings) overwrites the configs of the parent classes. To fix your original problem, define all env_files in the final class using a list: ``` class Settings(DatabaseSettings, AuthSettings): model_config = SettingsConfigDict( env_prefix="APP_", env_file=[".env.database", ".env.auth"], env_file_encoding="utf-8", extra="ignore",

HanSooloo Nov 24 at 19:52

THIS should be the correct answer. Would you mind posting it as an answer and I can mark it as the solution?

Mauricio Reisdoefer Nov 24 at 18:19

I personally recommend using composition, as it tends to be easier to manage in larger projects. Alternatively, you could consolidate all settings into a single class if that fits your use case better.

HanSooloo Nov 26 at 21:58

Is there any way to initialize one of the sub-settings models with a custom environment file? i.e., trying to do something like settings = Settings(_env_file=f".env.{app_env}.database") so that I can load DEV or TEST or PROD versions of the config values.

Collectives™ on Stack Overflow

Combined Pydantic settings class from multiple settings classes with individual environment files

2 Answers 2

Comments

5 Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

2 Answers 2

Comments

5 Comments

Your Answer

Sign up or log in

Post as a guest

Related