mirror of
https://github.com/norandom/OpenUltraSAST.git
synced 2026-11-13 17:58:18 +00:00
feat(detection-core): add closed evidence validation context
This commit is contained in:
@@ -0,0 +1,390 @@
|
||||
"""Closed validation context for raw detection-core evidence records."""
|
||||
|
||||
from __future__ import annotations
|
||||
|
||||
from dataclasses import dataclass, fields
|
||||
|
||||
from openultrasast.core import canonical
|
||||
from openultrasast.core.custody import (
|
||||
PopulationCustodyPolicy,
|
||||
PopulationCustodyPort,
|
||||
derive_population_custody_policy_digest,
|
||||
validate_population_custody_log,
|
||||
)
|
||||
from openultrasast.core.evidence import EvidenceBasis, EvidenceRecord, derive_record_id
|
||||
from openultrasast.core.factpack import (
|
||||
FactPack,
|
||||
ReviewIndependencePolicy,
|
||||
derive_fact_pack_content_digest,
|
||||
derive_review_evidence_digest,
|
||||
derive_review_independence_policy_digest,
|
||||
validate_fact_pack,
|
||||
)
|
||||
from openultrasast.core.factvalidate import validate_fact_relations
|
||||
from openultrasast.core.family import validate_record
|
||||
from openultrasast.core.family_catalog import FamilyCatalog, resolve_contract, validate_catalog
|
||||
from openultrasast.core.lineage import LineageManifest, derive_lineage_manifest_digest
|
||||
from openultrasast.core.model import (
|
||||
ModelProgramDeclaration,
|
||||
ModelRequest,
|
||||
ModelResult,
|
||||
derive_model_program_declaration_digest,
|
||||
)
|
||||
from openultrasast.core.provenance import (
|
||||
NamedInputs,
|
||||
ProvenanceTrustPolicy,
|
||||
derive_trust_policy_digest,
|
||||
validate_provenance,
|
||||
)
|
||||
|
||||
EVIDENCE_VALIDATION_CONTEXT_SCHEMA = "openultrasast.detection-core.evidence-validation-context.v2"
|
||||
EVIDENCE_VALIDATION_INPUT_SCHEMA = "openultrasast.detection-core.evidence-validation-input.v2"
|
||||
FACT_SET_SCHEMA = "openultrasast.detection-core.fact-set.v2"
|
||||
|
||||
|
||||
class EvidenceInputValidationError(ValueError):
|
||||
"""A raw evidence input cannot be reproduced in its closed context."""
|
||||
|
||||
|
||||
@dataclass(frozen=True)
|
||||
class EvidenceValidationContext:
|
||||
family_catalog: FamilyCatalog
|
||||
fact_packs: tuple[FactPack, ...]
|
||||
model_programs: tuple[ModelProgramDeclaration, ...]
|
||||
lineages: tuple[LineageManifest, ...]
|
||||
fact_review_policy: ReviewIndependencePolicy
|
||||
provenance_trust_policy: ProvenanceTrustPolicy
|
||||
custody_policy: PopulationCustodyPolicy
|
||||
context_digest: str
|
||||
|
||||
|
||||
@dataclass(frozen=True)
|
||||
class EvidenceValidationInput:
|
||||
record: EvidenceRecord
|
||||
named_inputs: NamedInputs
|
||||
model_request: ModelRequest | None
|
||||
model_result: ModelResult | None
|
||||
input_digest: str
|
||||
|
||||
|
||||
def fact_set_bytes(packs: tuple[FactPack, ...]) -> bytes:
|
||||
if not isinstance(packs, tuple) or not all(isinstance(pack, FactPack) for pack in packs):
|
||||
raise EvidenceInputValidationError("fact packs must be a tuple of FactPack values")
|
||||
return canonical.canonical_bytes(
|
||||
{
|
||||
"schema_version": FACT_SET_SCHEMA,
|
||||
"fact_pack_digests": tuple(sorted(pack.normalized_content_digest for pack in packs)),
|
||||
}
|
||||
)
|
||||
|
||||
|
||||
def derive_fact_set_digest(packs: tuple[FactPack, ...]) -> str:
|
||||
return canonical.bytes_digest(fact_set_bytes(packs))
|
||||
|
||||
|
||||
def derive_evidence_validation_context_digest(context: EvidenceValidationContext) -> str:
|
||||
packs = tuple(
|
||||
sorted(
|
||||
context.fact_packs,
|
||||
key=lambda pack: (
|
||||
pack.normalized_content_digest,
|
||||
pack.review_evidence_digest,
|
||||
pack.review_policy_digest,
|
||||
pack.review_identity,
|
||||
),
|
||||
)
|
||||
)
|
||||
return canonical.content_digest(
|
||||
{
|
||||
"schema_version": EVIDENCE_VALIDATION_CONTEXT_SCHEMA,
|
||||
"family_catalog_digest": context.family_catalog.catalog_digest,
|
||||
"fact_pack_review_bindings": tuple(
|
||||
{
|
||||
"normalized_content_digest": pack.normalized_content_digest,
|
||||
"review_evidence_digest": pack.review_evidence_digest,
|
||||
"review_policy_digest": pack.review_policy_digest,
|
||||
"review_identity": pack.review_identity,
|
||||
}
|
||||
for pack in packs
|
||||
),
|
||||
"model_program_digests": tuple(sorted(program.declaration_digest for program in context.model_programs)),
|
||||
"lineage_manifest_digests": tuple(sorted(lineage.manifest_digest for lineage in context.lineages)),
|
||||
"fact_review_policy_digest": context.fact_review_policy.policy_digest,
|
||||
"provenance_trust_policy_digest": context.provenance_trust_policy.policy_digest,
|
||||
"custody_policy_digest": context.custody_policy.policy_digest,
|
||||
}
|
||||
)
|
||||
|
||||
|
||||
def make_evidence_validation_context(
|
||||
*,
|
||||
family_catalog: FamilyCatalog,
|
||||
fact_packs: tuple[FactPack, ...],
|
||||
model_programs: tuple[ModelProgramDeclaration, ...],
|
||||
lineages: tuple[LineageManifest, ...],
|
||||
fact_review_policy: ReviewIndependencePolicy,
|
||||
provenance_trust_policy: ProvenanceTrustPolicy,
|
||||
custody_policy: PopulationCustodyPolicy,
|
||||
) -> EvidenceValidationContext:
|
||||
provisional = EvidenceValidationContext(
|
||||
family_catalog,
|
||||
fact_packs,
|
||||
model_programs,
|
||||
lineages,
|
||||
fact_review_policy,
|
||||
provenance_trust_policy,
|
||||
custody_policy,
|
||||
"0" * 64,
|
||||
)
|
||||
return EvidenceValidationContext(
|
||||
family_catalog,
|
||||
fact_packs,
|
||||
model_programs,
|
||||
lineages,
|
||||
fact_review_policy,
|
||||
provenance_trust_policy,
|
||||
custody_policy,
|
||||
derive_evidence_validation_context_digest(provisional),
|
||||
)
|
||||
|
||||
|
||||
def derive_evidence_validation_input_digest(value: EvidenceValidationInput) -> str:
|
||||
return canonical.content_digest(
|
||||
{
|
||||
"schema_version": EVIDENCE_VALIDATION_INPUT_SCHEMA,
|
||||
"record_id": value.record.record_id,
|
||||
"named_inputs_digest": value.named_inputs.manifest_digest,
|
||||
"model_request_digest": None if value.model_request is None else value.model_request.request_id,
|
||||
"model_result_digest": None if value.model_result is None else value.model_result.result_digest,
|
||||
}
|
||||
)
|
||||
|
||||
|
||||
def make_evidence_validation_input(
|
||||
*,
|
||||
record: EvidenceRecord,
|
||||
named_inputs: NamedInputs,
|
||||
model_request: ModelRequest | None = None,
|
||||
model_result: ModelResult | None = None,
|
||||
) -> EvidenceValidationInput:
|
||||
provisional = EvidenceValidationInput(record, named_inputs, model_request, model_result, "0" * 64)
|
||||
return EvidenceValidationInput(
|
||||
record,
|
||||
named_inputs,
|
||||
model_request,
|
||||
model_result,
|
||||
derive_evidence_validation_input_digest(provisional),
|
||||
)
|
||||
|
||||
|
||||
def _unique(values: tuple[str, ...], field: str) -> None:
|
||||
if len(set(values)) != len(values):
|
||||
raise EvidenceInputValidationError(f"duplicate {field} identity")
|
||||
|
||||
|
||||
def _validate_context_collections(context: EvidenceValidationContext) -> None:
|
||||
collections = (
|
||||
(context.fact_packs, FactPack, "fact_packs"),
|
||||
(context.model_programs, ModelProgramDeclaration, "model_programs"),
|
||||
(context.lineages, LineageManifest, "lineages"),
|
||||
)
|
||||
for values, expected, field in collections:
|
||||
if not isinstance(values, tuple) or not all(isinstance(item, expected) for item in values):
|
||||
raise EvidenceInputValidationError(f"{field} must be a tuple of {expected.__name__} values")
|
||||
_unique(tuple(pack.normalized_content_digest for pack in context.fact_packs), "fact pack")
|
||||
_unique(tuple(program.declaration_digest for program in context.model_programs), "model program")
|
||||
_unique(tuple(lineage.manifest_digest for lineage in context.lineages), "lineage manifest")
|
||||
|
||||
|
||||
def _validate_context_types(context: EvidenceValidationContext) -> None:
|
||||
fields = (
|
||||
(context.family_catalog, FamilyCatalog, "family_catalog"),
|
||||
(context.fact_review_policy, ReviewIndependencePolicy, "fact_review_policy"),
|
||||
(context.provenance_trust_policy, ProvenanceTrustPolicy, "provenance_trust_policy"),
|
||||
(context.custody_policy, PopulationCustodyPolicy, "custody_policy"),
|
||||
)
|
||||
for value, expected, field in fields:
|
||||
if not isinstance(value, expected):
|
||||
raise EvidenceInputValidationError(f"{field} must be a {expected.__name__}")
|
||||
_validate_context_collections(context)
|
||||
|
||||
|
||||
def _validate_context_policies(context: EvidenceValidationContext) -> None:
|
||||
review = context.fact_review_policy
|
||||
review.__post_init__()
|
||||
expected_review = derive_review_independence_policy_digest(
|
||||
review.reviewed_reviewer_digests,
|
||||
review.forbidden_identity_pairs,
|
||||
review.trusted_store_identities,
|
||||
review.maximum_evidence_bytes,
|
||||
)
|
||||
if review.policy_digest != expected_review:
|
||||
raise EvidenceInputValidationError("fact review policy digest cannot be re-derived")
|
||||
trust = context.provenance_trust_policy
|
||||
trust.__post_init__()
|
||||
if trust.policy_digest != derive_trust_policy_digest(trust.trusted_store_identities, trust.maximum_input_bytes):
|
||||
raise EvidenceInputValidationError("provenance trust policy digest cannot be re-derived")
|
||||
custody = context.custody_policy
|
||||
custody.__post_init__()
|
||||
if custody.policy_digest != derive_population_custody_policy_digest(custody.reviewed_custodian_digests):
|
||||
raise EvidenceInputValidationError("custody policy digest cannot be re-derived")
|
||||
|
||||
|
||||
def _validate_context_members(context: EvidenceValidationContext) -> None:
|
||||
catalog_digests = {entry.contract_digest for entry in context.family_catalog.entries}
|
||||
lineage_digests = {lineage.manifest_digest for lineage in context.lineages}
|
||||
for lineage in context.lineages:
|
||||
expected = derive_lineage_manifest_digest(lineage.arm_id, lineage.subject_digest, lineage.uses)
|
||||
if lineage.manifest_digest != expected:
|
||||
raise EvidenceInputValidationError("lineage manifest digest cannot be re-derived")
|
||||
for pack in context.fact_packs:
|
||||
pack.__post_init__()
|
||||
if pack.normalized_content_digest != derive_fact_pack_content_digest(pack):
|
||||
raise EvidenceInputValidationError("fact pack content digest cannot be re-derived")
|
||||
if pack.review_evidence_digest != derive_review_evidence_digest(pack.review_evidence):
|
||||
raise EvidenceInputValidationError("fact pack review evidence digest cannot be re-derived")
|
||||
for program in context.model_programs:
|
||||
program.__post_init__()
|
||||
if program.declaration_digest != derive_model_program_declaration_digest(program):
|
||||
raise EvidenceInputValidationError("model declaration digest cannot be re-derived")
|
||||
if program.family_contract_digest not in catalog_digests:
|
||||
raise EvidenceInputValidationError("model declaration family contract is absent from the catalog")
|
||||
if program.lineage_manifest_digest not in lineage_digests:
|
||||
raise EvidenceInputValidationError("model declaration lineage is absent from the context")
|
||||
|
||||
|
||||
def _validate_context(context: EvidenceValidationContext) -> None:
|
||||
if not isinstance(context, EvidenceValidationContext):
|
||||
raise EvidenceInputValidationError("context must be an EvidenceValidationContext")
|
||||
_validate_context_types(context)
|
||||
expected = derive_evidence_validation_context_digest(context)
|
||||
if context.context_digest != expected:
|
||||
raise EvidenceInputValidationError("context digest cannot be re-derived")
|
||||
validate_catalog(context.family_catalog)
|
||||
_validate_context_policies(context)
|
||||
_validate_context_members(context)
|
||||
|
||||
|
||||
def _validate_record_identity(record: EvidenceRecord) -> None:
|
||||
if not isinstance(record, EvidenceRecord):
|
||||
raise EvidenceInputValidationError("record must be an EvidenceRecord")
|
||||
record.packet.__post_init__()
|
||||
record.decision.__post_init__()
|
||||
expected = derive_record_id(record.packet, record.decision)
|
||||
if record.record_id != expected:
|
||||
raise EvidenceInputValidationError("record identity cannot be re-derived")
|
||||
record.__post_init__()
|
||||
|
||||
|
||||
def _validate_raw_deterministic_input(value: EvidenceValidationInput) -> None:
|
||||
if value.model_request is not None or value.model_result is not None:
|
||||
raise EvidenceInputValidationError("model handoff replay belongs to task 7.2")
|
||||
decision = value.record.decision
|
||||
if decision.basis is not EvidenceBasis.DETERMINISTIC:
|
||||
raise EvidenceInputValidationError("raw validation accepts deterministic records only before model or proof replay")
|
||||
provenance_values = tuple(getattr(decision.provenance, item.name) for item in fields(decision.provenance))
|
||||
if any(item is not None for item in provenance_values):
|
||||
raise EvidenceInputValidationError("deterministic record contains model, proof, or lineage provenance")
|
||||
|
||||
|
||||
def _validate_input_types(value: EvidenceValidationInput) -> None:
|
||||
if not isinstance(value.record, EvidenceRecord):
|
||||
raise EvidenceInputValidationError("record must be an EvidenceRecord")
|
||||
if not isinstance(value.named_inputs, NamedInputs):
|
||||
raise EvidenceInputValidationError("named_inputs must be NamedInputs")
|
||||
if value.model_request is not None and not isinstance(value.model_request, ModelRequest):
|
||||
raise EvidenceInputValidationError("model_request must be a ModelRequest or None")
|
||||
if value.model_result is not None and not isinstance(value.model_result, ModelResult):
|
||||
raise EvidenceInputValidationError("model_result must be a ModelResult or None")
|
||||
|
||||
|
||||
def _validate_packs(
|
||||
context: EvidenceValidationContext,
|
||||
population_digest: str,
|
||||
custody: PopulationCustodyPort,
|
||||
resolver: canonical.ContentResolver,
|
||||
) -> None:
|
||||
for pack in context.fact_packs:
|
||||
contract = resolve_contract(
|
||||
context.family_catalog,
|
||||
pack.family_id,
|
||||
pack.family_contract_digest,
|
||||
pack.ecosystem,
|
||||
)
|
||||
validate_fact_pack(
|
||||
pack,
|
||||
contract,
|
||||
population_digest,
|
||||
context.lineages,
|
||||
custody,
|
||||
context.custody_policy,
|
||||
context.fact_review_policy,
|
||||
resolver,
|
||||
)
|
||||
|
||||
|
||||
def validate_evidence_input(
|
||||
value: EvidenceValidationInput,
|
||||
context: EvidenceValidationContext,
|
||||
custody: PopulationCustodyPort,
|
||||
resolver: canonical.ContentResolver,
|
||||
) -> EvidenceRecord:
|
||||
try:
|
||||
if not isinstance(value, EvidenceValidationInput):
|
||||
raise EvidenceInputValidationError("value must be an EvidenceValidationInput")
|
||||
_validate_context(context)
|
||||
_validate_input_types(value)
|
||||
if value.input_digest != derive_evidence_validation_input_digest(value):
|
||||
raise EvidenceInputValidationError("input digest cannot be re-derived")
|
||||
_validate_record_identity(value.record)
|
||||
_validate_raw_deterministic_input(value)
|
||||
record = value.record
|
||||
contract = resolve_contract(
|
||||
context.family_catalog,
|
||||
record.packet.family_id,
|
||||
record.packet.family_contract_digest,
|
||||
record.packet.ecosystem,
|
||||
)
|
||||
validate_record(contract, record)
|
||||
validate_provenance(record.packet.provenance, value.named_inputs, resolver, context.provenance_trust_policy)
|
||||
if record.packet.provenance.fact_set_digest != derive_fact_set_digest(context.fact_packs):
|
||||
raise EvidenceInputValidationError("record fact set does not match the validation context")
|
||||
population = record.packet.provenance.population_digest
|
||||
validate_population_custody_log(custody.current_log(population), context.custody_policy, current=custody)
|
||||
_validate_packs(context, population, custody, resolver)
|
||||
validate_fact_relations(
|
||||
record,
|
||||
contract,
|
||||
context.fact_packs,
|
||||
population,
|
||||
context.lineages,
|
||||
custody,
|
||||
context.custody_policy,
|
||||
context.fact_review_policy,
|
||||
resolver,
|
||||
)
|
||||
return record
|
||||
except EvidenceInputValidationError:
|
||||
raise
|
||||
except (AttributeError, LookupError, TypeError, ValueError) as exc:
|
||||
message = str(exc)
|
||||
if "provenance" not in message and exc.__class__.__module__.endswith("provenance"):
|
||||
message = f"provenance validation failed: {message}"
|
||||
raise EvidenceInputValidationError(f"evidence input validation failed: {message}") from exc
|
||||
|
||||
|
||||
__all__ = [
|
||||
"EVIDENCE_VALIDATION_CONTEXT_SCHEMA",
|
||||
"EVIDENCE_VALIDATION_INPUT_SCHEMA",
|
||||
"FACT_SET_SCHEMA",
|
||||
"EvidenceInputValidationError",
|
||||
"EvidenceValidationContext",
|
||||
"EvidenceValidationInput",
|
||||
"derive_evidence_validation_context_digest",
|
||||
"derive_evidence_validation_input_digest",
|
||||
"derive_fact_set_digest",
|
||||
"fact_set_bytes",
|
||||
"make_evidence_validation_context",
|
||||
"make_evidence_validation_input",
|
||||
"validate_evidence_input",
|
||||
]
|
||||
Reference in New Issue
Block a user