larzconf

Layered, typed application configuration. Pure Python, zero dependencies.

Read configuration from defaults, files, .env, and environment variables — layered so later sources win — and access it with the right types. No more int(os.environ.get("PORT", "8000")) scattered through your code, and no dependency to add.

from larzconf import Config

config = (Config()
          .from_dict({"port": 8000, "debug": False})   # defaults
          .from_file("config.json", optional=True)      # committed config
          .from_dotenv(".env")                           # local dev
          .from_env(prefix="APP_"))                      # production wins

config.int("port")                 # int, not "8000"
config.bool("debug")               # True / False from "true"/"1"/"yes"...
config.list("allowed_hosts")       # comma-split -> list
config.get("db.host")              # dotted access into nested config
config.get("secret_key", required=True)   # raises if missing

Why

• Zero dependencies. No python-dotenv, no pydantic-settings, no framework. Just the standard library.
• The 12-factor layering you actually want. Defaults, then a config file, then .env for local development, then real environment variables in production — each layer deep-merges over the last.
• Typed, not stringly. config.int(...), .bool(...), .float(...), .list(...) coerce values and raise a clear ConfigError on bad input; required=True fails fast at startup instead of at 3am.
• Nested & dotted. JSON config nests naturally; access it with config.get("db.host"). Env vars map to nesting via __ (APP_DB__HOST → db.host).
• Validates end-to-end. Hand the whole config to a schema (larzvalidate, or anything with .load(dict)) to type-check it once at boot.

Install

pip install larzconf

Loading

Config().from_dict({...})              # defaults / literals
       .from_json("config.json")       # a JSON file
       .from_dotenv(".env")            # KEY=VALUE (export/comments/quotes ok)
       .from_env(prefix="APP_")        # APP_PORT -> port, APP_DB__HOST -> db.host
       .from_file(path, optional=True) # dispatch by extension

Each call returns the Config, so they chain, and each layer overrides the ones before it.

Reading

config.get("key", default=None, cast=int, required=False)
config.int("port")            config.float("ratio")
config.bool("debug")          config.list("hosts")     config.str("name")
config.get("db.host")         # nested, dotted
"port" in config             config["port"]            config.as_dict()

Validate at startup (optional)

from larzvalidate import Schema, Integer, Boolean, String

class Settings(Schema):
    port  = Integer(min=1, max=65535, default=8000)
    debug = Boolean(default=False)
    secret_key = String(required=True)

settings = config.validate(Settings())   # cleaned, typed dict — or raises

larzconf doesn't depend on larzvalidate — validate() accepts anything with a .load(dict) method.

Tests

python -m unittest discover -s tests -v      # 25 tests, zero deps

The Larz stack

Pure-Python, zero-dependency building blocks: larz · larzchain · larzmoney · larzcrypt · larzdb · larzagent · larzchart · larzmark · larztask · larzvault · larzvm · larzcache · larzvalidate · larzid · larzrpc · larzstate · larzhttp · larzconf

License

MIT © larz-scripter