mnemoreg 0.3.1

A tiny thread-safe registry mapping for simple plugin/registry use-cases.

mnemoreg

mnemoreg is a tiny, dependency-free, thread-safe registry mapping useful for registering callables and other values by string keys. It's intentionally small and suitable for embedding in other projects. It uses only the Python standard library and aims to provide a predictable, easy-to-use interface for shared, named objects.

Table of Contents

• Highlights
• Installation
• Quick start
• API summary
• Thread-safety and testing notes
• Serialization
• Development and testing
• Contributing
• Troubleshooting
• License

Highlights

• Small, single-file registry implementation (see mnemoreg/core.py).
• Thread-safe operations using threading.RLock and explicit bulk() context.
• Decorator-based registration for callables.
• Snapshot and (de)serialization helpers.
• Configurable overwrite behaviour (forbid / allow / warn).

Installation

Install from PyPI:

pip install mnemoreg

Or install from source:

git clone https://github.com/i3iorn/mnemoreg.git
cd mnemoreg
pip install .

Quick start

from mnemoreg import Registry

# Create a registry typed for string keys and int values
r = Registry[str, int]()
r["one"] = 1
print(r["one"])  # 1

# Register a callable under an explicit key
@r.register("plus")
def plus(x):
    return x + 1

print(r["plus"](4))  # 5

# Register using the function name as the key
@r.register()
def multiply(x, y):
    return x * y

print(r["multiply"](3, 4))  # 12

# Use the bulk context manager to perform multiple operations under the same lock
with r.bulk():
    r["a"] = 1
    r["b"] = 2

# Create a shallow snapshot
snap = r.snapshot()
print(snap)

See the tests/ directory for many additional examples and edge cases.

API summary

This is a concise summary — see mnemoreg/core.py docstrings for full details.

• Registry(*, lock: Optional[RLock]=None, log_level: int=logging.WARNING, overwrite_policy: int=OverwritePolicy.FORBID) — constructor.
• Mapping-like methods: __getitem__, __setitem__, __delitem__, __iter__, __len__, __contains__.
• register(key: Optional[str] = None) — decorator to register callables/objects.
• get(key, default=None), snapshot(), to_dict().
• from_dict(mapping), from_json(s) — classmethods to build from serialized data.
• to_json(**kwargs) — serialize to JSON string.
• bulk() — context manager that acquires the registry lock for batched operations.
• update(mapping), clear(), unregister(key), remove(key).

Exceptions raised:

• AlreadyRegisteredError — when a key must not already exist but does.
• NotRegisteredError — when accessing/deleting a key that does not exist.

Overwrite behaviour is controlled by OverwritePolicy enum (FORBID=0, ALLOW=1, WARN=2).

Thread-safety and testing notes

mnemoreg is guarded by a threading.RLock for mutating operations. Iteration and snapshot() return shallow copies to avoid exposing internal state to concurrent mutation.

If you write tests that intentionally start background threads which raise exceptions (for example, tests that exercise concurrency failure modes), pytest will surface a PytestUnhandledThreadExceptionWarning for uncaught exceptions in threads. To hide that specific warning only for the threaded test module, add this module-level filter to tests/test_registry_threaded.py:

# tests/test_registry_threaded.py
import pytest

# suppress only the thread-unhandled warning for this module
pytestmark = pytest.mark.filterwarnings(
    "ignore::pytest.PytestUnhandledThreadExceptionWarning"
)

For the pytest-asyncio deprecation warning shown by newer versions: configure the default fixture loop scope in your pytest configuration. For example, in pyproject.toml:

[tool.pytest.ini_options]
asyncio_mode = "strict"
asyncio_default_fixture_loop_scope = "function"

This sets the asyncio fixture loop scope explicitly and avoids the PytestDeprecationWarning about the unset asyncio_default_fixture_loop_scope.

Serialization

The registry supports basic JSON-friendly (de)serialization via to_dict, from_dict, to_json, and from_json. These operate on shallow copies of the internal store, so custom objects will need their own serialization logic before being stored if you need to persist them as JSON.

Example:

r = Registry[str, int]()
r["one"] = 1
s = r.to_json()
new_r = Registry.from_json(s)

Development and testing

Run tests with:

python -m pytest -vv

The test suite covers single-threaded and concurrent scenarios. If you see spurious warnings from async fixtures or thread exceptions while developing, use the options described above to configure pytest or narrow the warning filters to the affected test modules.

Contributing

Contributions are welcome. A suggested workflow:

1. Open an issue to discuss larger changes.
2. Branch from main (or master) for new work.
3. Add tests for new behaviour or bug fixes.
4. Run the test suite and make sure everything passes.
5. Create a pull request with a clear description of the changes.

Coding style: keep changes small and well-tested. Prefer plain stdlib implementations unless there is a clear productivity win from a dependency.

Troubleshooting

• AlreadyRegisteredError during concurrent writes: your test or production logic may be attempting to re-register a key; consider OverwritePolicy.ALLOW or adjust the test flow to avoid races.
• PytestUnhandledThreadExceptionWarning: see the module-level pytestmark example above to suppress the warning only in the threaded test module.
• pytest-asyncio deprecation warnings: set asyncio_default_fixture_loop_scope in pytest config as shown above.

If you hit something not covered here, please open an issue with a small reproduction.

License

mnemoreg is licensed under the MIT License — see the LICENSE file for details.

---

Maintainers: i3iorn