Pystorex

A lightweight Python state management library inspired by NgRx/Redux patterns and ReactiveX for Python (reactivex). Manage application state with reducers, handle side effects with effects, compose middleware, and select state slices efficiently.

Features

• Typed State: Define your root state using typing_extensions.TypedDict and store it in immutables.Map, fully generic.
• Reducers: Pure functions to update state in response to actions.
• Effects: Handle side effects by listening to action streams and optionally dispatching actions.
• Middleware: Insert custom logic (logging, thunks, error handling) into the dispatch pipeline.
• Selectors: Memoized and configurable (deep compare, TTL) state accessors.
• Immutable Updates: Use shallow copy at the feature level or integrate with immutables.Map and utility functions (to_immutable, to_dict, to_pydantic).
• Hot Module Management: Register/unregister feature reducers and effects at runtime.

Installation

pip install pystorex

Requires Python 3.9+ support.

Quick Start

This example demonstrates how to define state using TypedDict and handle state updates with immutables.Map for better performance and clarity.

import time
from typing import Optional
from typing_extensions import TypedDict
from reactivex import operators as ops
from immutables import Map

from pystorex.actions import create_action
from pystorex import create_store, create_reducer, on, create_effect
from pystorex.store_selectors import create_selector
from pystorex.middleware import LoggerMiddleware
from pystorex.map_utils import batch_update

# 1. Define state model (TypedDict)
class CounterState(TypedDict):
    count: int
    loading: bool
    error: Optional[str]
    last_updated: Optional[float]

counter_initial_state = CounterState(
    count=0, loading=False, error=None, last_updated=None
)

# 2. Define Actions
increment = create_action("increment")
decrement = create_action("decrement")
reset = create_action("reset", lambda value: value)
increment_by = create_action("incrementBy", lambda amount: amount)
load_count_request = create_action("loadCountRequest")
load_count_success = create_action("loadCountSuccess", lambda value: value)
load_count_failure = create_action("loadCountFailure", lambda error: error)

# 3. Define Reducer

def counter_handler(state: Map, action) -> Map:
    now = time.time()
    if action.type == increment.type:
        return state.set("count", state["count"] + 1).set("last_updated", now)
    elif action.type == decrement.type:
        return batch_update(state, {"count": state["count"] - 1, "last_updated": now})
    elif action.type == reset.type:
        return batch_update(state, {"count": action.payload, "last_updated": now})
    elif action.type == increment_by.type:
        return batch_update(state, {"count": state["count"] + action.payload, "last_updated": now})
    elif action.type == load_count_request.type:
        return batch_update(state, {"loading": True, "error": None})
    elif action.type == load_count_success.type:
        return batch_update(state, {"loading": False, "count": action.payload, "last_updated": now})
    elif action.type == load_count_failure.type:
        return batch_update(state, {"loading": False, "error": action.payload})
    return state

counter_reducer = create_reducer(
    counter_initial_state,
    on(increment, counter_handler),
    on(decrement, counter_handler),
    on(reset, counter_handler),
    on(increment_by, counter_handler),
    on(load_count_request, counter_handler),
    on(load_count_success, counter_handler),
    on(load_count_failure, counter_handler),
)

# 4. Define Effects
class CounterEffects:
    @create_effect
    def load_count(self, action_stream):
        return action_stream.pipe(
            ops.filter(lambda action: action.type == load_count_request.type),
            ops.do_action(lambda _: print("Effect: Loading counter...")),
            ops.delay(1.0),
            ops.map(lambda _: load_count_success(42))
        )

# 5. Create Store and Register Modules
store = create_store()
store.apply_middleware(LoggerMiddleware)
store.register_root({"counter": counter_reducer})
store.register_effects(CounterEffects)

# 6. Use Selector to Subscribe to State
get_counter_state = lambda state: state["counter"]
get_count = create_selector(
    get_counter_state,
    result_fn=lambda counter: counter.get("count", 0)
)
store.select(get_count).subscribe(
    lambda c: print(f"Count: {c[1]}")
)

# 7. Execute Example Operations
if __name__ == "__main__":
    store.dispatch(increment())
    store.dispatch(increment_by(5))
    store.dispatch(decrement())
    store.dispatch(reset(10))
    store.dispatch(load_count_request())
    # Give Effects some time
    time.sleep(2)

Notes

• State management has been updated to use TypedDict and immutables.Map, avoiding the performance overhead of Pydantic models during frequent state updates.
• Use batch_update and built-in methods of immutables.Map to ensure immutability of state updates.
• Pydantic models can be dynamically converted as needed using utility functions. See the example source code for details.

Examples

This project includes example scripts demonstrating monolithic and modular usage patterns:

• examples/counter_example/counter_example_monolithic.py: Monolithic Counter example using TypedDict + Map.
• examples/counter_example/main.py: Modular Counter example.
• examples/detection_example/...: Detection examples.

Run them with:

python examples/counter_example/counter_example_monolithic.py
python examples/counter_example/main.py

Publishing to PyPI

1. Ensure pyproject.toml & setup.cfg are configured.

2. Install build tools:

   pip install --upgrade build twine
   

3. Build distributions:

   python -m build
   

4. Upload:

   twine upload dist/*
   

Contributing

• Fork the repo
• Create a feature branch
• Write tests (pytest) and update docs
• Submit a Pull Request

License

MIT