Configuration

Store Location

By default dfstore saves everything to ~/.dfstore. There are three ways to change this.

Environment Variable

export DFSTORE_PATH=/data/my_store

Any Python or CLI call will use this path automatically.

Per-Call Argument

Every API function and CLI command accepts --store-path / store_path=:

dfstore.save(df, name="data", store_path="/data/my_store")
dfstore.get("data", store_path="/data/my_store")
dfstore.list(store_path="/data/my_store")

dfstore save data.csv --name data --store-path /data/my_store
dfstore list --store-path /data/my_store

DFStore Class

For scripts that make many calls to the same store, instantiate DFStore directly:

from dfstore import DFStore
from pathlib import Path

store = DFStore(Path("/data/my_store"))

store.save(df, name="employees")
store.get("employees")
store.list()

---

Priority Order

When resolving the store path, dfstore checks in this order:

1. store_path argument passed directly to the function / CLI flag
2. DFSTORE_PATH environment variable
3. Default: ~/.dfstore

---

Store Layout

<store_path>/
  index.json          ← all metadata (one JSON object per DataFrame)
  data/
    <name>/
      v1.parquet
      v2.parquet
      …

The index.json file is updated atomically on every save. Parquet files are never modified after creation — each save writes a new file.