Spaces:

sitcod3
/

modul-x

Running

modul-x / app /validator /validation_cache.py

Upload folder using huggingface_hub

c94194a verified 11 days ago

4.7 kB

"""Result caching for the verification engine.
Two things are expensive and highly reusable in bulk verification:
  - **DNS/MX** for a domain (every address at that domain needs the same answer)
  - **the mailbox verdict** for an address (lists overlap; users re-run lists)
Both are cached here, with one hard correctness rule: **only definitive results
are cached**. Transient failures — a DNS timeout, a greylist 4xx, an unreachable
MX — are never stored, because remembering them would freeze a temporary
condition into a 24h "verdict" and suppress a later successful probe.
A cache outage degrades to "no cache"; it must never break validation.
"""
from __future__ import annotations

import json
from typing import Any, Callable, Coroutine, Optional

from ..config import settings
from ..models import DNSResult, SMTPResult
from ..utils.cache import get_cache as _default_get, set_cache as _default_set

# A DNS timeout/resolver error is inconclusive — never a cacheable fact.
_UNRESOLVED_DNS_CODES = frozenset({"timeout", "resolver_error"})

DNS_TTL = 24 * 3600        # MX/SPF/DMARC change slowly
VERDICT_TTL = 24 * 3600    # a mailbox verdict is stable within a day


class ValidationCache:
    """Thin typed layer over the shared Redis/memory cache."""

    def __init__(
        self,
        get: Optional[Callable[[str], Coroutine[Any, Any, Optional[str]]]] = None,
        set: Optional[Callable[..., Coroutine[Any, Any, None]]] = None,
        ttl_seconds: int = DNS_TTL,
    ):
        self._get = get or _default_get
        self._set = set or _default_set
        self.ttl_seconds = ttl_seconds

    # ---- keys ------------------------------------------------------
    @staticmethod
    def dns_key(domain: str) -> str:
        return f"vx:dns:{domain.strip().lower()}"

    @staticmethod
    def verdict_key(email: str, deep: bool) -> str:
        # `deep` is part of the identity: a fast-mode result carries no mailbox
        # proof and must never satisfy a deep request.
        return f"vx:v:{'d' if deep else 'f'}:{email.strip().lower()}"

    # ---- low level -------------------------------------------------
    async def _read_json(self, key: str) -> Optional[dict]:
        try:
            raw = await self._get(key)
        except Exception:
            return None  # cache outage -> treat as a miss
        if not raw:
            return None
        try:
            data = json.loads(raw)
        except Exception:
            return None  # corrupt payload -> miss, never raise
        return data if isinstance(data, dict) else None

    async def _write_json(self, key: str, payload: dict, ttl: int) -> None:
        try:
            await self._set(key, json.dumps(payload), ttl)
        except Exception:
            pass  # a cache write failure must not break validation

    # ---- DNS -------------------------------------------------------
    async def get_dns(self, domain: str) -> Optional[DNSResult]:
        data = await self._read_json(self.dns_key(domain))
        if not data:
            return None
        try:
            return DNSResult(**data)
        except Exception:
            return None

    async def put_dns(self, domain: str, result: DNSResult) -> None:
        if result.error_code in _UNRESOLVED_DNS_CODES:
            return  # transient -> never cache
        await self._write_json(
            self.dns_key(domain), result.model_dump(), self.ttl_seconds
        )

    # ---- mailbox verdict ------------------------------------------
    async def get_verdict(
        self, email: str, deep: bool
    ) -> Optional[tuple[SMTPResult, Optional[bool]]]:
        data = await self._read_json(self.verdict_key(email, deep))
        if not data:
            return None
        try:
            smtp = SMTPResult(**data["smtp"])
            catch_all = data.get("catch_all")
            return smtp, catch_all
        except Exception:
            return None

    async def put_verdict(
        self,
        email: str,
        deep: bool,
        smtp: SMTPResult,
        catch_all: Optional[bool],
    ) -> None:
        # Cacheable only when the mailbox question was actually answered, or the
        # domain is definitively catch-all (a domain fact, not a probe failure).
        definitive = smtp.verified is not None or catch_all is True
        if not definitive:
            return
        await self._write_json(
            self.verdict_key(email, deep),
            {"smtp": smtp.model_dump(), "catch_all": catch_all},
            VERDICT_TTL,
        )


# Shared instance used by the engine.
validation_cache = ValidationCache(
    ttl_seconds=getattr(settings, "dns_cache_ttl_seconds", DNS_TTL)
)

	"""Result caching for the verification engine.

	Two things are expensive and highly reusable in bulk verification:

	- DNS/MX for a domain (every address at that domain needs the same answer)
	- the mailbox verdict for an address (lists overlap; users re-run lists)

	Both are cached here, with one hard correctness rule: **only definitive results
	are cached**. Transient failures — a DNS timeout, a greylist 4xx, an unreachable
	MX — are never stored, because remembering them would freeze a temporary
	condition into a 24h "verdict" and suppress a later successful probe.

	A cache outage degrades to "no cache"; it must never break validation.
	"""
	from __future__ import annotations

	import json
	from typing import Any, Callable, Coroutine, Optional

	from ..config import settings
	from ..models import DNSResult, SMTPResult
	from ..utils.cache import get_cache as _default_get, set_cache as _default_set

	# A DNS timeout/resolver error is inconclusive — never a cacheable fact.
	_UNRESOLVED_DNS_CODES = frozenset({"timeout", "resolver_error"})

	DNS_TTL = 24 * 3600 # MX/SPF/DMARC change slowly
	VERDICT_TTL = 24 * 3600 # a mailbox verdict is stable within a day


	class ValidationCache:
	"""Thin typed layer over the shared Redis/memory cache."""

	def __init__(
	self,
	get: Optional[Callable[[str], Coroutine[Any, Any, Optional[str]]]] = None,
	set: Optional[Callable[..., Coroutine[Any, Any, None]]] = None,
	ttl_seconds: int = DNS_TTL,
	):
	self._get = get or _default_get
	self._set = set or _default_set
	self.ttl_seconds = ttl_seconds

	# ---- keys ------------------------------------------------------
	@staticmethod
	def dns_key(domain: str) -> str:
	return f"vx:dns:{domain.strip().lower()}"

	@staticmethod
	def verdict_key(email: str, deep: bool) -> str:
	# `deep` is part of the identity: a fast-mode result carries no mailbox
	# proof and must never satisfy a deep request.
	return f"vx:v:{'d' if deep else 'f'}:{email.strip().lower()}"

	# ---- low level -------------------------------------------------
	async def _read_json(self, key: str) -> Optional[dict]:
	try:
	raw = await self._get(key)
	except Exception:
	return None # cache outage -> treat as a miss
	if not raw:
	return None
	try:
	data = json.loads(raw)
	except Exception:
	return None # corrupt payload -> miss, never raise
	return data if isinstance(data, dict) else None

	async def _write_json(self, key: str, payload: dict, ttl: int) -> None:
	try:
	await self._set(key, json.dumps(payload), ttl)
	except Exception:
	pass # a cache write failure must not break validation

	# ---- DNS -------------------------------------------------------
	async def get_dns(self, domain: str) -> Optional[DNSResult]:
	data = await self._read_json(self.dns_key(domain))
	if not data:
	return None
	try:
	return DNSResult(**data)
	except Exception:
	return None

	async def put_dns(self, domain: str, result: DNSResult) -> None:
	if result.error_code in _UNRESOLVED_DNS_CODES:
	return # transient -> never cache
	await self._write_json(
	self.dns_key(domain), result.model_dump(), self.ttl_seconds
	)

	# ---- mailbox verdict ------------------------------------------
	async def get_verdict(
	self, email: str, deep: bool
	) -> Optional[tuple[SMTPResult, Optional[bool]]]:
	data = await self._read_json(self.verdict_key(email, deep))
	if not data:
	return None
	try:
	smtp = SMTPResult(**data["smtp"])
	catch_all = data.get("catch_all")
	return smtp, catch_all
	except Exception:
	return None

	async def put_verdict(
	self,
	email: str,
	deep: bool,
	smtp: SMTPResult,
	catch_all: Optional[bool],
	) -> None:
	# Cacheable only when the mailbox question was actually answered, or the
	# domain is definitively catch-all (a domain fact, not a probe failure).
	definitive = smtp.verified is not None or catch_all is True
	if not definitive:
	return
	await self._write_json(
	self.verdict_key(email, deep),
	{"smtp": smtp.model_dump(), "catch_all": catch_all},
	VERDICT_TTL,
	)


	# Shared instance used by the engine.
	validation_cache = ValidationCache(
	ttl_seconds=getattr(settings, "dns_cache_ttl_seconds", DNS_TTL)
	)