zspec 1.0.5

Composable specification pattern for Python

zspec

Composable specification pattern for Python 3.14+.

The problem

Business rules tend to spread across your codebase. A check like "is this order eligible for free shipping?" might live in a service method, duplicated in a view, slightly different in a validation layer. When requirements change, you hunt down every copy and hope you found them all.

The Specification pattern solves this by turning each rule into a single, testable object. Combine them with &, |, ^, ~ to express complex logic without writing new classes.

Installation

pip install zspec

Quick start

from dataclasses import dataclass
from zspec import Specification


@dataclass
class Product:
    name: str
    price: int
    in_stock: bool


# Define specifications as simple subclasses
class InStock(Specification[Product]):
    def is_satisfied_by(self, p: Product) -> bool:
        return p.in_stock


class Affordable(Specification[Product]):
    def __init__(self, max_price: int) -> None:
        self.max_price = max_price

    def is_satisfied_by(self, p: Product) -> bool:
        return p.price <= self.max_price


# Compose with &, |, ~
in_stock = InStock()
reasonable = Affordable(max_price=1000)
eligible = in_stock & reasonable

product = Product(name="Laptop", price=800, in_stock=True)
assert eligible(product)  # True -- callable directly
assert eligible.is_satisfied_by(product)  # same thing

Features

• Composable --- combine specs with & (and), | (or), ^ (xor), ~ (not)
• Type-safe --- generic Specification[T] preserves the candidate type
• Bulk combinators --- Specification.all_of(...) and Specification.any_of(...)
• Zero dependencies --- standard library only

API overview

• spec & other — Both must be satisfied (AND)
• spec | other — At least one must be satisfied (OR)
• spec ^ other — Exactly one must be satisfied (XOR)
• ~spec — Negation (NOT)
• spec(candidate) — Shorthand for is_satisfied_by
• Specification.of(fn) — Create a spec from a callable
• Specification.true() / Specification.false() — Always / never satisfied
• spec.filter(iterable) — Lazy filter over a collection
• spec.reject(iterable) — Inverse of filter
• spec.partition(iterable) — Split into (passed, failed) lists
• Specification.all_of(specs) — Reduce with AND, returns None for empty input
• Specification.any_of(specs) — Reduce with OR, returns None for empty input
• explain(spec, candidate) — Debug tree showing what passed/failed

Translators

Convert specification trees to SQL, MongoDB filters, Django Q objects, or custom formats:

from dataclasses import dataclass
from zspec import SqlTranslator, SqlFragment, Specification


@dataclass
class Product:
    price: int
    in_stock: bool


class MinPrice(Specification[Product]):
    def __init__(self, price: int) -> None:
        self.price = price

    def is_satisfied_by(self, candidate: Product) -> bool:
        return candidate.price >= self.price


class MySql(SqlTranslator):
    def _translate(self, spec: Specification[Product]) -> SqlFragment:
        match spec:
            case MinPrice(price=price):
                return SqlFragment("price >= %s", (price,))
            case _:
                return super()._translate(spec)


spec = MinPrice(100) & MinPrice(200)
fragment = MySql().translate(spec)
assert fragment.sql == "(price >= %s AND price >= %s)"
assert fragment.params == (100, 200)

License

MIT