scicanonicalhash 0.1.0

Deterministic hashing for arbitrary Python objects

Canonical Hash

Deterministic hashing for arbitrary Python objects.

Provides utilities for creating stable, deterministic hashes of Python objects, essential for cache key computation, data versioning, and reproducibility.

Usage

from scicanonicalhash import canonical_hash, generate_record_id

# Hash any supported Python object
h = canonical_hash(42)
h = canonical_hash([1, 2, 3])
h = canonical_hash({"key": "value"})

# Generates a 16-character hex string (first 64 bits of SHA-256)
assert len(h) == 16
assert canonical_hash(42) == canonical_hash(42)  # Deterministic

Supported Types

1. JSON-serializable primitives (None, bool, int, float, str)
2. numpy ndarrays (via shape + dtype + raw bytes)
3. pandas DataFrames (via columns + index + array serialization)
4. pandas Series (via name + array serialization)
5. Dicts (sorted keys, recursive serialization)
6. Lists/tuples (order-preserving, recursive serialization)

generate_record_id

Generate a unique record ID from type, schema version, content hash, and metadata:

rid = generate_record_id("MyData", 1, "abc123", {"subject": 1})