Settings and Configuration¶
Sayer provides a built-in way to configure your CLI application using centralized settings, powered by the SAYER_SETTINGS_MODULE environment variable.
This mechanism lets you:
- Keep config logic out of command code
- Load settings dynamically
- Easily override settings in dev, test, and prod environments
โ๏ธ Default Settings Class¶
Sayer includes a built-in Settings dataclass that defines the default configuration used when no SAYER_SETTINGS_MODULE is set,
or when internal tools need immediate settings access.
This class covers:
- Debug mode and logging
- Output color systems
- Version tagging
- Serialization utilities
๐ฆ Location¶
from sayer.conf.settings import Settings
This is not to be confused with your application-level settings โ it is used internally and as a base for custom configurations.
๐งพ Default Fields¶
| Field | Type | Description |
|---|---|---|
debug |
bool |
Enables debug mode (more logging, errors) |
logging_level |
str |
Logging threshold (DEBUG, INFO, etc.) |
version |
str |
Version of the Sayer library (auto-filled) |
is_logging_setup |
bool |
Tracks if logging has already been configured |
force_terminal |
bool | None |
Force terminal output regardless of environment |
color_system |
"auto" | "standard" | "256" | ... |
Controls terminal color profile |
display_full_help |
bool |
Flag indicating the the display of each command must be diplayed in full |
display_help_length |
int |
The length of the help if display_full_help is set to False. Defaults to 99 |
formatter_class |
RichHelpFormatter |
The formatter used to render the Sayer Rich UI. This can be overridden by any custom renderer. |
๐๏ธ Field Examples¶
settings.debug = True
settings.logging_level = "DEBUG"
settings.force_terminal = True
settings.color_system = "256"
๐ง logging_config Property¶
This is a dynamic property that returns a LoggingConfig object based on the current logging level.
config = settings.logging_config
Sayer uses this internally to configure rich/standard logging output. You can override it:
from sayer.logging import CustomLoggingConfig
settings.logging_config = CustomLoggingConfig(level="DEBUG")
๐ Converting to Dict or Tuples¶
The Settings class provides two helpers to introspect or export the config:
dict(...)¶
settings.dict(exclude_none=True, upper=True)
Args:
exclude_none: Omit fields withNonevaluesupper: Transform keys to uppercase (useful for env-like exports)
Returns:
{
"DEBUG": True,
"LOGGING_LEVEL": "INFO",
...
}
tuple(...)¶
settings.tuple(exclude_none=True, upper=False)
Returns:
[("debug", True), ("logging_level", "INFO"), ...]
Useful for exporting to .env files or logging config introspection.
๐งฐ Internal Use¶
The Settings class is primarily used:
- As the fallback configuration if
SAYER_SETTINGS_MODULEis not defined - By components like
sayer.loggingandrichterminal integration - To bootstrap internal state when settings aren't user-defined
๐งช Customizing in Tests¶
You can use it in test environments or override fields dynamically:
from sayer.conf.global_settings import Settings
s = Settings(debug=True, logging_level="DEBUG")
print(s.dict())
๐งฌ Summary¶
| Feature | Supported |
|---|---|
| Built-in debug mode | โ |
| Color and terminal control | โ |
| Logging abstraction | โ |
| Dict export | โ |
| Tuple export | โ |
| Used in fallback mode | โ |
This class is the core config contract Sayer uses internally โ you can use it as a reference to build more advanced or environment-specific config systems.
๐ฆ How It Works¶
At runtime, Sayer looks for the environment variable:
SAYER_SETTINGS_MODULE=your_project.settings.Dev
It will:
- Import the specified module
- Access the specified class (e.g.,
Dev) - Instantiate and expose it as
settingsanywhere you need
๐ง Anatomy of a Settings Module¶
# your_project/settings.py
from sayer.conf.global_settings import BaseSettings
class Base(BaseSettings):
debug = False
retries = 3
class Dev(Base):
debug = True
retries = 10
๐ Setting the Module¶
You must set the environment variable before launching the CLI:
export SAYER_SETTINGS_MODULE=your_project.settings.Dev
python app.py my-command
You can also set it inline:
SAYER_SETTINGS_MODULE=your_project.settings.Dev python app.py my-command
๐งช Accessing Settings in Commands¶
from sayer.conf import settings
@command()
def show():
print("Debug mode:", settings.debug)
No need to pass it explicitly โ it's global, and lazily loaded on first access.
๐งฐ Settings Best Practices¶
1. Split environments:¶
# settings.py
from dataclasses import dataclass
from sayer.conf.global_settings import BaseSettings
class Base(BaseSettings):
retries = 3
class Dev(Base):
debug = True
class Prod(Base):
debug = False
retries = 1
Use different envs like:
SAYER_SETTINGS_MODULE=myapp.settings.Dev
2. Type Your Settings¶
Sayer doesn't enforce typing but it's recommended:
class Dev:
debug: bool = True
retries: int = 3
You can combine this with Pydantic, attrs, or msgspec for stricter validation if desired.
๐งช Overriding Settings in Tests¶
Tests that rely on settings should override the environment variable or monkey-patch sayer.conf.settings.
Example:
import os
from sayer.conf import settings
def test_behavior(monkeypatch):
monkeypatch.setenv("SAYER_SETTINGS_MODULE", "tests.fake_settings.Mocked")
assert settings.debug is True
Or if settings is already imported:
settings.debug = True # override directly for testing
๐ง Under the Hood¶
If SAYER_SETTINGS_MODULE is unset, it will use the internal by default or if invalid, Sayer will raise a helpful exception.
๐งฐ Recap¶
| Feature | Description |
|---|---|
SAYER_SETTINGS_MODULE |
Points to your settings class |
settings |
Auto-loaded global settings object |
| Lazy loading | Only instantiated when accessed |
| Nested inheritance | Use Base, Dev, Prod hierarchies |
| Test override | Patch env var or mutate settings in test |
โ Example Project Structure¶
myapp/
โโโ settings.py
โ โโโ class Base
โ โโโ class Dev(Base)
โ โโโ class Prod(Base)
โโโ commands.py
โโโ main.py
Set it:
export SAYER_SETTINGS_MODULE=myapp.settings.Dev
Access it:
from sayer.conf import settings
print(settings.retries)
๐ Next: Commands