"""Extrae hashes MD5/SHA1/SHA256/SHA512 de un texto, con offsets y algoritmo."""

import re

# Mas largo primero para evitar que un SHA256 quede como SHA1+resto.
_HASH_LENGTHS = (
    (128, "sha512"),
    (64, "sha256"),
    (40, "sha1"),
    (32, "md5"),
)

_HASH_CANDIDATE = re.compile(r"\b[A-Fa-f0-9]{32,128}\b")


def extract_file_hashes(text: str) -> list[dict]:
    """Extrae hashes hex con su algoritmo deducido por longitud.

    Reconoce MD5 (32), SHA1 (40), SHA256 (64) y SHA512 (128). Hashes
    de longitudes intermedias se ignoran. Devuelve `algorithm` ademas
    de los campos estandar.
    """
    results = []
    for m in _HASH_CANDIDATE.finditer(text):
        candidate = m.group(0)
        length = len(candidate)
        algorithm = next(
            (algo for size, algo in _HASH_LENGTHS if size == length),
            None,
        )
        if algorithm is None:
            continue
        results.append({
            "value": candidate,
            "start": m.start(),
            "end": m.end(),
            "type": "file_hash",
            "algorithm": algorithm,
        })
    return results