Make serialized JSON deterministic for hashes

Эта статья ещё не доступна на языке «Русский»; показан оригинал.

methodology · en · актуально на 2026-09-21 · изменено 2026-09-21, ревизия 2 · unreviewed

Define a narrow serialization contract before hashing JSON, and distinguish repeatable Python output from cross-language canonical JSON.

Define the accepted data domain

For an internal Python-only fingerprint, restrict input to string-keyed objects, lists, strings, booleans, null and integers if floating-point equivalence is unnecessary. Reject unsupported values rather than silently converting them.

Example serializer

import hashlib
import json

def fingerprint(value):
    encoded = json.dumps(value, sort_keys=True, separators=(",", ":"),
                         ensure_ascii=False, allow_nan=False).encode("utf-8")
    return hashlib.sha256(encoded).hexdigest()

Python documents sorted keys and JSON serialization options. The function assumes its input has already passed the chosen domain validation; it is not a complete canonicalization validator.

Acceptance fixtures

Objects with the same string keys inserted in different orders should hash equally. Lists with different element order should not. Reject NaN and infinity. Decide explicitly whether 1 and 1.0 or differently normalized Unicode strings represent the same application value.

Limits

Do not use this function as an interoperable signature format without a specified cross-language canonicalization scheme and shared test vectors. Keep the serialization version with stored hashes so a later encoder or domain-policy change is not mistaken for content tampering.

Область и основание

Original worked method and proposed acceptance fixtures; no empirical performance result is claimed. The cited primary documentation was read for the specific technical behavior described.

Актуально на: 2026-09-21. Статус: unreviewed (задокументированной рецензии нет) — правки сбрасывают статус рецензии. Считайте текст непроверенным справочным материалом и сверяйтесь с источниками.

Источники

Атрибуция и лицензия

Последнее изменение: Replaced generic draft with a specific procedure, example, failure cases and correctly scoped sources; removed unrelated product applicability.

Оригинальный материал: CC BY 4.0. Материалы по ссылкам сохраняют собственные права.