Configuration

The store backends can be configured via environment vars and/or overridden during runtime initialization.

Initializing a configurable store

Initialize a store with the configured defaults from environment:

from anystore import get_store

store = get_store()

Initialize a store with override options:

from anystore import get_store

store = get_store(
    uri="redis://localhost",
    default_ttl=3600,
    backend_config={"redis_prefix": "my-prefix"},
)

Environment vars

All the keyword arguments to get_store can be configured as default values via environment vars, using uppercase names and ANYSTORE_ prefix.

Nested config dicts (such as backend_config) can be configured via environment vars as well, using __ as a nesting separator:

ANYSTORE_BACKEND_CONFIG__REDIS_PREFIX=my-prefix

Environment vars values will be casted based on their type (e.g. "0" for "False", "yes" for "True").

Configuration

The actual backend is inferred from the uri scheme. If the uri is a relative path, the scheme will be file and therefore the store a local file storage backend.

Backend-specific settings can be passed through via backend_config or ANYSTORE_BACKEND_CONFIG__*

Setting	init keyword	env var	Default
Base store uri	uri	ANYSTORE_URI	./anystore
Serialization mode	serialization_mode	ANYSTORE_SERIALIZATION_MODE	auto
Error handling for non-existing keys	raise_on_nonexist	ANYSTORE_RAISE_ON_NONEXIST	True
Key ttl for backends that support it (e.g. redis, sql)	default_ttl	ANYSTORE_DEFAULT_TTL	0

File-like backends

anystore is built on top of fsspec, and therefore any supported backend is configurable. For some backends, additional libraries need to be installed. Refer to their documentation.

The backend is inferred by the scheme of the uri.

Commonly used

ANYSTORE_URI=./local-data                          # relative local path
ANYSTORE_URI=file:///var/data                      # absolute local path
ANYSTORE_URI=s3://my_bucket                        # AWS or S3 like
ANYSTORE_URI=s3://my_bucket/prefix/foo             # Sub path in S3
ANYSTORE_URI=gcs://my_bucket                       # Google cloud storage
ANYSTORE_URI=sftp://my_server.net:/var/lib/data    # Connect via ssh
ANYSTORE_URI=https://example.org/files             # Read-only via file listing

Backend specific config

Redis

uri / ANYSTORE_URI can be a full redis connection string, e.g.:

redis://user:password@hostname:port

A key prefix for all stored data can be set via redis_prefix in backend_config or ANYSTORE_BACKEND_CONFIG__REDIS_PREFIX

debug mode (fakeredis)

For development purposes, fake redis via REDIS_DEBUG=1 in environment.

S3 like

Configure parameters via client_kwargs in backend_config, e.g.:

from anystore import get_store

store = get_store(uri="s3://mybucket/data", backend_config={"client_kwargs": {
    "aws_access_key_id": "my-key",
    "aws_secret_access_key": "***",
    "endpoint_url": "https://s3.local"
}})

Or via environment vars:

AWS_ACCESS_KEY_ID=***
AWS_SECRET_ACCESS_KEY=***
FSSPEC_S3_ENDPOINT_URL=https://s3.local

SQL

uri / ANYSTORE_URI can be a full sql connection string:

# absolute
sqlite:////var/data.db
# relative to current wd
sqlite:///data.db

# postgres with credentials and ssl params
postgresql://user:password@hostname:port/database?sslmode=verify-full

Configure parameters via sql in backend_config or env ANYSTORE_BACKEND_CONFIG__SQL__*

sql settings and their defaults

from anystore import get_store

store = get_store(uri="postgresql:///db", backend_config={"sql": {
    "table": "anystore",
    "pool_size": 5,
    "engine_kwargs": {}  # sqlalchemy kwargs
}})

ANYSTORE_BACKEND_CONFIG__SQL__TABLE=anystore
ANYSTORE_BACKEND_CONFIG__SQL__POOL_SIZE=5