data-cascade 0.2.0

Composable directory-based data cascade loader/saver with per-file origins and a convenient path/proxy API.

data-cascade

A composable directory-based data cascade loader/saver for Python (>=3.10). It merges files in a tree (YAML/JSON/TOML), keeps a per-key origin map for round-tripping, and offers both a path API and an ergonomic proxy API with dirty-file saves.

Features

• Merge __main__.* into the current node and per-file trees by stem name.
• Directory-level __config__.* with data.merge options (dict/list modes, per-key overrides, excludes).
• Supports YAML (ruamel.yaml preferred, PyYAML fallback), JSON (stdlib), TOML (tomllib or tomli; writers via tomli-w or toml).
• Returns a CascadeMap mapping each merged key path to the source file and local path.
• save_data_cascade writes values back to their original files, or sensible defaults for new keys.
• Cascade object with:
  • Path API: get("a.b[1]"), set("a.b[1]", value), delete("a.b[1]")
  • Proxy API: c.node("a").b[1].set(value) and c.node().a.b[1].get()
  • Dirty-file saves: c.save() only rewrites files you touched.

Install (with Poetry)

poetry add data-cascade
# Optionally enable YAML and TOML writers/readers
poetry add data-cascade -E yaml -E toml

Quick start

from pathlib import Path
from data_cascade import load_data_cascade, save_data_cascade

data, cmap = load_data_cascade(Path("data"))
data["project"] = {"name": "Alpha"}
save_data_cascade(Path("data"), data, cmap)

The Cascade object

from data_cascade import make_cascade

c = make_cascade("data")

# Path API
members = c.get('team.members')
c.set('team.members[1]', 'Carol')
c.save()  # only touches the file that owns team.members

# Proxy API
node = c.node('a').b[1]
node.set(42)
print(node.get())
c.save()

Configuring merge

Put a __config__.yaml in any directory. Example:

__config__:
  data:
    merge:
      dict: deep            # deep | override | first_wins
      list:
        mode: merge_by_key  # replace | extend | unique | merge_by_key
        key: id
      per_key:
        tasks:
          list:
            mode: unique
      exclude: ["scratch"]  # skip stems or keys

Handling missing libraries

YAML and TOML are optional. If a file is encountered and no handler exists for its extension, a warning is logged and the file is skipped/raises on save for that type. Install extras (-E yaml, -E toml) for full support.

Tasks (poethepoet)

poe test   # run pytest
poe pack   # packaging info
poe fmt    # placeholder for formatters

License

MIT