Upload 5 files
Browse files- tests/test_drift.py +196 -85
- tests/test_intent.py +291 -0
- tests/test_orchestrator.py +286 -0
- tests/test_scheduler.py +128 -0
tests/test_drift.py
CHANGED
|
@@ -19,6 +19,7 @@ from __future__ import annotations
|
|
| 19 |
|
| 20 |
import copy
|
| 21 |
from dataclasses import asdict
|
|
|
|
| 22 |
|
| 23 |
import pytest
|
| 24 |
|
|
@@ -36,8 +37,8 @@ from osint_core.drift import (
|
|
| 36 |
)
|
| 37 |
|
| 38 |
|
| 39 |
-
def make_telemetry(**overrides):
|
| 40 |
-
data = {
|
| 41 |
"run_id": "run_test_001",
|
| 42 |
"manifest_hash": "manifest_good",
|
| 43 |
"dependency_hash": "deps_good",
|
|
@@ -61,8 +62,8 @@ def make_telemetry(**overrides):
|
|
| 61 |
return TelemetrySnapshot(**data)
|
| 62 |
|
| 63 |
|
| 64 |
-
def make_baseline(**overrides):
|
| 65 |
-
data = {
|
| 66 |
"runtime_p95_ms": 500,
|
| 67 |
"error_rate_threshold": 2,
|
| 68 |
"timeout_threshold": 1,
|
|
@@ -85,8 +86,8 @@ def make_baseline(**overrides):
|
|
| 85 |
return data
|
| 86 |
|
| 87 |
|
| 88 |
-
def make_policy_result(**overrides):
|
| 89 |
-
data = {
|
| 90 |
"decision": "allow",
|
| 91 |
"allowed_modules": ["resource_links"],
|
| 92 |
"blocked_modules": [],
|
|
@@ -96,7 +97,22 @@ def make_policy_result(**overrides):
|
|
| 96 |
return data
|
| 97 |
|
| 98 |
|
| 99 |
-
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 100 |
vector = DriftVector()
|
| 101 |
|
| 102 |
assert vector.statistical == 0.0
|
|
@@ -107,7 +123,11 @@ def test_drift_vector_defaults_to_zero():
|
|
| 107 |
assert vector.policy == 0.0
|
| 108 |
|
| 109 |
|
| 110 |
-
def
|
|
|
|
|
|
|
|
|
|
|
|
|
| 111 |
signals = [
|
| 112 |
DriftSignal(
|
| 113 |
name="weak_adversarial_signal",
|
|
@@ -142,7 +162,8 @@ def test_aggregate_signals_uses_max_score_per_type():
|
|
| 142 |
assert vector.policy == 0.0
|
| 143 |
|
| 144 |
|
| 145 |
-
def
|
|
|
|
| 146 |
vector = DriftVector(
|
| 147 |
statistical=0.9,
|
| 148 |
adversarial=0.4,
|
|
@@ -151,6 +172,8 @@ def test_dominant_type_respects_priority_not_raw_score():
|
|
| 151 |
|
| 152 |
assert choose_dominant_drift_type(vector) == DriftType.ADVERSARIAL
|
| 153 |
|
|
|
|
|
|
|
| 154 |
vector = DriftVector(
|
| 155 |
statistical=0.9,
|
| 156 |
adversarial=0.4,
|
|
@@ -160,45 +183,41 @@ def test_dominant_type_respects_priority_not_raw_score():
|
|
| 160 |
assert choose_dominant_drift_type(vector) == DriftType.POLICY
|
| 161 |
|
| 162 |
|
| 163 |
-
def
|
| 164 |
-
vector = DriftVector(
|
| 165 |
-
|
| 166 |
-
|
| 167 |
-
|
| 168 |
-
|
| 169 |
-
|
| 170 |
-
|
| 171 |
-
|
| 172 |
-
|
| 173 |
-
|
| 174 |
-
|
| 175 |
-
|
| 176 |
-
|
| 177 |
-
|
| 178 |
-
|
| 179 |
-
|
| 180 |
-
|
| 181 |
-
|
| 182 |
-
|
| 183 |
-
|
| 184 |
-
|
| 185 |
-
|
| 186 |
-
|
| 187 |
-
|
| 188 |
-
|
| 189 |
-
|
| 190 |
-
|
| 191 |
-
|
| 192 |
-
|
| 193 |
-
def test_recommend_correction_defaults_to_observe():
|
| 194 |
-
vector = DriftVector(statistical=0.1, operational=0.1)
|
| 195 |
-
|
| 196 |
-
assert recommend_correction(vector) == "OBSERVE"
|
| 197 |
|
| 198 |
|
| 199 |
-
def test_policy_violation_creates_policy_signal_and_revert_recommendation(
|
| 200 |
-
telemetry
|
| 201 |
-
baseline
|
|
|
|
| 202 |
policy_result = make_policy_result(
|
| 203 |
decision="constrain",
|
| 204 |
blocked_modules=["port_scan"],
|
|
@@ -224,13 +243,14 @@ def test_policy_violation_creates_policy_signal_and_revert_recommendation():
|
|
| 224 |
assert any(signal.drift_type == DriftType.POLICY for signal in assessment.signals)
|
| 225 |
|
| 226 |
|
| 227 |
-
def test_authorization_gate_trigger_creates_policy_signal(
|
|
|
|
|
|
|
| 228 |
telemetry = make_telemetry(
|
| 229 |
modules_requested=["http_headers"],
|
| 230 |
modules_blocked=["http_headers"],
|
| 231 |
authorized_target=False,
|
| 232 |
)
|
| 233 |
-
baseline = make_baseline()
|
| 234 |
policy_result = make_policy_result(
|
| 235 |
decision="constrain",
|
| 236 |
blocked_modules=["http_headers"],
|
|
@@ -253,14 +273,15 @@ def test_authorization_gate_trigger_creates_policy_signal():
|
|
| 253 |
assert assessment.recommended_correction == "REVERT"
|
| 254 |
|
| 255 |
|
| 256 |
-
def test_adversarial_patterns_create_constrain_recommendation(
|
|
|
|
|
|
|
|
|
|
| 257 |
telemetry = make_telemetry(
|
| 258 |
input_rejected=True,
|
| 259 |
rejection_reason="Input contains a blocked pattern.",
|
| 260 |
sanitized_input_trace="https://example.com/?next=http://169.254.169.254/latest",
|
| 261 |
)
|
| 262 |
-
baseline = make_baseline()
|
| 263 |
-
policy_result = make_policy_result()
|
| 264 |
|
| 265 |
assessment = assess_drift(
|
| 266 |
telemetry=telemetry,
|
|
@@ -273,10 +294,31 @@ def test_adversarial_patterns_create_constrain_recommendation():
|
|
| 273 |
assert assessment.recommended_correction == "CONSTRAIN"
|
| 274 |
|
| 275 |
|
| 276 |
-
def
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 277 |
telemetry = make_telemetry(duration_ms=1200)
|
| 278 |
-
baseline = make_baseline(runtime_p95_ms=500)
|
| 279 |
-
policy_result = make_policy_result()
|
| 280 |
|
| 281 |
assessment = assess_drift(
|
| 282 |
telemetry=telemetry,
|
|
@@ -288,10 +330,11 @@ def test_operational_runtime_drift_detected():
|
|
| 288 |
assert any(signal.name == "runtime_boundary_exceeded" for signal in assessment.signals)
|
| 289 |
|
| 290 |
|
| 291 |
-
def test_operational_error_drift_detected(
|
|
|
|
|
|
|
|
|
|
| 292 |
telemetry = make_telemetry(error_count=3)
|
| 293 |
-
baseline = make_baseline(error_rate_threshold=2)
|
| 294 |
-
policy_result = make_policy_result()
|
| 295 |
|
| 296 |
assessment = assess_drift(
|
| 297 |
telemetry=telemetry,
|
|
@@ -303,10 +346,28 @@ def test_operational_error_drift_detected():
|
|
| 303 |
assert any(signal.name == "error_threshold_exceeded" for signal in assessment.signals)
|
| 304 |
|
| 305 |
|
| 306 |
-
def
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 307 |
telemetry = make_telemetry(manifest_hash="manifest_changed")
|
| 308 |
-
baseline = make_baseline(expected_manifest_hash="manifest_good")
|
| 309 |
-
policy_result = make_policy_result()
|
| 310 |
|
| 311 |
assessment = assess_drift(
|
| 312 |
telemetry=telemetry,
|
|
@@ -319,10 +380,11 @@ def test_structural_manifest_mismatch_reverts():
|
|
| 319 |
assert assessment.recommended_correction == "REVERT"
|
| 320 |
|
| 321 |
|
| 322 |
-
def test_structural_dependency_mismatch_reverts(
|
|
|
|
|
|
|
|
|
|
| 323 |
telemetry = make_telemetry(dependency_hash="deps_changed")
|
| 324 |
-
baseline = make_baseline(expected_dependency_hash="deps_good")
|
| 325 |
-
policy_result = make_policy_result()
|
| 326 |
|
| 327 |
assessment = assess_drift(
|
| 328 |
telemetry=telemetry,
|
|
@@ -334,7 +396,27 @@ def test_structural_dependency_mismatch_reverts():
|
|
| 334 |
assert assessment.recommended_correction == "REVERT"
|
| 335 |
|
| 336 |
|
| 337 |
-
def
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 338 |
telemetry = make_telemetry(
|
| 339 |
indicator_hash="hmac_abc123",
|
| 340 |
output_hash="output_changed",
|
|
@@ -342,7 +424,6 @@ def test_behavioral_same_input_different_output_reverts():
|
|
| 342 |
baseline = make_baseline(
|
| 343 |
known_output_hashes={"hmac_abc123": "output_good"},
|
| 344 |
)
|
| 345 |
-
policy_result = make_policy_result()
|
| 346 |
|
| 347 |
assessment = assess_drift(
|
| 348 |
telemetry=telemetry,
|
|
@@ -355,10 +436,11 @@ def test_behavioral_same_input_different_output_reverts():
|
|
| 355 |
assert assessment.recommended_correction == "REVERT"
|
| 356 |
|
| 357 |
|
| 358 |
-
def test_behavioral_invalid_schema_reverts(
|
|
|
|
|
|
|
|
|
|
| 359 |
telemetry = make_telemetry(output_schema_valid=False)
|
| 360 |
-
baseline = make_baseline()
|
| 361 |
-
policy_result = make_policy_result()
|
| 362 |
|
| 363 |
assessment = assess_drift(
|
| 364 |
telemetry=telemetry,
|
|
@@ -370,12 +452,14 @@ def test_behavioral_invalid_schema_reverts():
|
|
| 370 |
assert assessment.recommended_correction == "REVERT"
|
| 371 |
|
| 372 |
|
| 373 |
-
def test_statistical_shift_can_adapt_when_no_higher_priority_signal(
|
|
|
|
|
|
|
|
|
|
| 374 |
telemetry = make_telemetry(indicator_type="ip")
|
| 375 |
baseline = make_baseline(
|
| 376 |
input_type_distribution={"domain": 0.9, "username": 0.1},
|
| 377 |
)
|
| 378 |
-
policy_result = make_policy_result()
|
| 379 |
|
| 380 |
assessment = assess_drift(
|
| 381 |
telemetry=telemetry,
|
|
@@ -388,7 +472,30 @@ def test_statistical_shift_can_adapt_when_no_higher_priority_signal():
|
|
| 388 |
assert assessment.recommended_correction == "ADAPT"
|
| 389 |
|
| 390 |
|
| 391 |
-
def
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 392 |
telemetry = make_telemetry(indicator_type="ip")
|
| 393 |
baseline = make_baseline(
|
| 394 |
input_type_distribution={"domain": 0.9, "username": 0.1},
|
|
@@ -417,7 +524,10 @@ def test_policy_drift_overrides_statistical_adaptation():
|
|
| 417 |
assert assessment.recommended_correction == "REVERT"
|
| 418 |
|
| 419 |
|
| 420 |
-
def test_adversarial_drift_overrides_statistical_adaptation(
|
|
|
|
|
|
|
|
|
|
| 421 |
telemetry = make_telemetry(
|
| 422 |
indicator_type="ip",
|
| 423 |
sanitized_input_trace="http://169.254.169.254/latest",
|
|
@@ -425,7 +535,6 @@ def test_adversarial_drift_overrides_statistical_adaptation():
|
|
| 425 |
baseline = make_baseline(
|
| 426 |
input_type_distribution={"domain": 0.9, "username": 0.1},
|
| 427 |
)
|
| 428 |
-
policy_result = make_policy_result()
|
| 429 |
|
| 430 |
assessment = assess_drift(
|
| 431 |
telemetry=telemetry,
|
|
@@ -439,7 +548,7 @@ def test_adversarial_drift_overrides_statistical_adaptation():
|
|
| 439 |
assert assessment.recommended_correction == "CONSTRAIN"
|
| 440 |
|
| 441 |
|
| 442 |
-
def test_estimate_confidence_increases_with_signal_count_and_tier():
|
| 443 |
low_signal = DriftSignal(
|
| 444 |
name="weak",
|
| 445 |
drift_type=DriftType.STATISTICAL,
|
|
@@ -459,14 +568,16 @@ def test_estimate_confidence_increases_with_signal_count_and_tier():
|
|
| 459 |
|
| 460 |
assert estimate_confidence([]) == 0.0
|
| 461 |
assert estimate_confidence([high_signal]) > estimate_confidence([low_signal])
|
| 462 |
-
assert estimate_confidence([low_signal, high_signal]) >= estimate_confidence([high_signal])
|
| 463 |
|
|
|
|
|
|
|
| 464 |
|
| 465 |
-
def test_assess_drift_is_pure_and_does_not_mutate_inputs():
|
| 466 |
-
telemetry = make_telemetry()
|
| 467 |
-
baseline = make_baseline()
|
| 468 |
-
policy_result = make_policy_result()
|
| 469 |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 470 |
telemetry_before = copy.deepcopy(asdict(telemetry))
|
| 471 |
baseline_before = copy.deepcopy(baseline)
|
| 472 |
policy_before = copy.deepcopy(policy_result)
|
|
@@ -482,11 +593,11 @@ def test_assess_drift_is_pure_and_does_not_mutate_inputs():
|
|
| 482 |
assert policy_result == policy_before
|
| 483 |
|
| 484 |
|
| 485 |
-
def test_clean_execution_observes_without_significant_drift(
|
| 486 |
-
telemetry
|
| 487 |
-
baseline
|
| 488 |
-
policy_result
|
| 489 |
-
|
| 490 |
assessment = assess_drift(
|
| 491 |
telemetry=telemetry,
|
| 492 |
baseline=baseline,
|
|
@@ -497,4 +608,4 @@ def test_clean_execution_observes_without_significant_drift():
|
|
| 497 |
assert assessment.signals == []
|
| 498 |
assert assessment.dominant_type is None
|
| 499 |
assert assessment.recommended_correction == "OBSERVE"
|
| 500 |
-
assert assessment.confidence == 0.0
|
|
|
|
| 19 |
|
| 20 |
import copy
|
| 21 |
from dataclasses import asdict
|
| 22 |
+
from typing import Any
|
| 23 |
|
| 24 |
import pytest
|
| 25 |
|
|
|
|
| 37 |
)
|
| 38 |
|
| 39 |
|
| 40 |
+
def make_telemetry(**overrides: Any) -> TelemetrySnapshot:
|
| 41 |
+
data: dict[str, Any] = {
|
| 42 |
"run_id": "run_test_001",
|
| 43 |
"manifest_hash": "manifest_good",
|
| 44 |
"dependency_hash": "deps_good",
|
|
|
|
| 62 |
return TelemetrySnapshot(**data)
|
| 63 |
|
| 64 |
|
| 65 |
+
def make_baseline(**overrides: Any) -> dict[str, Any]:
|
| 66 |
+
data: dict[str, Any] = {
|
| 67 |
"runtime_p95_ms": 500,
|
| 68 |
"error_rate_threshold": 2,
|
| 69 |
"timeout_threshold": 1,
|
|
|
|
| 86 |
return data
|
| 87 |
|
| 88 |
|
| 89 |
+
def make_policy_result(**overrides: Any) -> dict[str, Any]:
|
| 90 |
+
data: dict[str, Any] = {
|
| 91 |
"decision": "allow",
|
| 92 |
"allowed_modules": ["resource_links"],
|
| 93 |
"blocked_modules": [],
|
|
|
|
| 97 |
return data
|
| 98 |
|
| 99 |
|
| 100 |
+
@pytest.fixture
|
| 101 |
+
def telemetry() -> TelemetrySnapshot:
|
| 102 |
+
return make_telemetry()
|
| 103 |
+
|
| 104 |
+
|
| 105 |
+
@pytest.fixture
|
| 106 |
+
def baseline() -> dict[str, Any]:
|
| 107 |
+
return make_baseline()
|
| 108 |
+
|
| 109 |
+
|
| 110 |
+
@pytest.fixture
|
| 111 |
+
def policy_result() -> dict[str, Any]:
|
| 112 |
+
return make_policy_result()
|
| 113 |
+
|
| 114 |
+
|
| 115 |
+
def test_drift_vector_defaults_to_zero() -> None:
|
| 116 |
vector = DriftVector()
|
| 117 |
|
| 118 |
assert vector.statistical == 0.0
|
|
|
|
| 123 |
assert vector.policy == 0.0
|
| 124 |
|
| 125 |
|
| 126 |
+
def test_aggregate_signals_empty_returns_zero_vector() -> None:
|
| 127 |
+
assert aggregate_signals([]) == DriftVector()
|
| 128 |
+
|
| 129 |
+
|
| 130 |
+
def test_aggregate_signals_uses_max_score_per_type() -> None:
|
| 131 |
signals = [
|
| 132 |
DriftSignal(
|
| 133 |
name="weak_adversarial_signal",
|
|
|
|
| 162 |
assert vector.policy == 0.0
|
| 163 |
|
| 164 |
|
| 165 |
+
def test_dominant_type_prefers_adversarial_over_statistical() -> None:
|
| 166 |
+
# Adversarial outranks statistical even if statistical has a higher raw score.
|
| 167 |
vector = DriftVector(
|
| 168 |
statistical=0.9,
|
| 169 |
adversarial=0.4,
|
|
|
|
| 172 |
|
| 173 |
assert choose_dominant_drift_type(vector) == DriftType.ADVERSARIAL
|
| 174 |
|
| 175 |
+
|
| 176 |
+
def test_dominant_type_prefers_policy_over_all() -> None:
|
| 177 |
vector = DriftVector(
|
| 178 |
statistical=0.9,
|
| 179 |
adversarial=0.4,
|
|
|
|
| 183 |
assert choose_dominant_drift_type(vector) == DriftType.POLICY
|
| 184 |
|
| 185 |
|
| 186 |
+
def test_dominant_type_respects_structural_over_behavioral_over_operational() -> None:
|
| 187 |
+
vector = DriftVector(structural=0.1, behavioral=0.9, operational=1.0)
|
| 188 |
+
assert choose_dominant_drift_type(vector) == DriftType.STRUCTURAL
|
| 189 |
+
|
| 190 |
+
vector = DriftVector(behavioral=0.2, adversarial=0.9, operational=1.0)
|
| 191 |
+
assert choose_dominant_drift_type(vector) == DriftType.BEHAVIORAL
|
| 192 |
+
|
| 193 |
+
|
| 194 |
+
@pytest.mark.parametrize(
|
| 195 |
+
("vector", "expected"),
|
| 196 |
+
[
|
| 197 |
+
(DriftVector(policy=0.6, statistical=1.0, adversarial=0.2), "REVERT"),
|
| 198 |
+
(DriftVector(structural=0.5), "REVERT"),
|
| 199 |
+
(DriftVector(behavioral=0.7), "REVERT"),
|
| 200 |
+
(DriftVector(adversarial=0.3, statistical=0.9), "CONSTRAIN"),
|
| 201 |
+
(DriftVector(statistical=0.5), "ADAPT"),
|
| 202 |
+
(DriftVector(statistical=0.1, operational=0.1), "OBSERVE"),
|
| 203 |
+
],
|
| 204 |
+
ids=[
|
| 205 |
+
"policy_revert",
|
| 206 |
+
"structural_revert",
|
| 207 |
+
"behavioral_revert",
|
| 208 |
+
"adversarial_constrain",
|
| 209 |
+
"statistical_adapt",
|
| 210 |
+
"default_observe",
|
| 211 |
+
],
|
| 212 |
+
)
|
| 213 |
+
def test_recommend_correction(vector: DriftVector, expected: str) -> None:
|
| 214 |
+
assert recommend_correction(vector) == expected
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 215 |
|
| 216 |
|
| 217 |
+
def test_policy_violation_creates_policy_signal_and_revert_recommendation(
|
| 218 |
+
telemetry: TelemetrySnapshot,
|
| 219 |
+
baseline: dict[str, Any],
|
| 220 |
+
) -> None:
|
| 221 |
policy_result = make_policy_result(
|
| 222 |
decision="constrain",
|
| 223 |
blocked_modules=["port_scan"],
|
|
|
|
| 243 |
assert any(signal.drift_type == DriftType.POLICY for signal in assessment.signals)
|
| 244 |
|
| 245 |
|
| 246 |
+
def test_authorization_gate_trigger_creates_policy_signal(
|
| 247 |
+
baseline: dict[str, Any],
|
| 248 |
+
) -> None:
|
| 249 |
telemetry = make_telemetry(
|
| 250 |
modules_requested=["http_headers"],
|
| 251 |
modules_blocked=["http_headers"],
|
| 252 |
authorized_target=False,
|
| 253 |
)
|
|
|
|
| 254 |
policy_result = make_policy_result(
|
| 255 |
decision="constrain",
|
| 256 |
blocked_modules=["http_headers"],
|
|
|
|
| 273 |
assert assessment.recommended_correction == "REVERT"
|
| 274 |
|
| 275 |
|
| 276 |
+
def test_adversarial_patterns_create_constrain_recommendation(
|
| 277 |
+
baseline: dict[str, Any],
|
| 278 |
+
policy_result: dict[str, Any],
|
| 279 |
+
) -> None:
|
| 280 |
telemetry = make_telemetry(
|
| 281 |
input_rejected=True,
|
| 282 |
rejection_reason="Input contains a blocked pattern.",
|
| 283 |
sanitized_input_trace="https://example.com/?next=http://169.254.169.254/latest",
|
| 284 |
)
|
|
|
|
|
|
|
| 285 |
|
| 286 |
assessment = assess_drift(
|
| 287 |
telemetry=telemetry,
|
|
|
|
| 294 |
assert assessment.recommended_correction == "CONSTRAIN"
|
| 295 |
|
| 296 |
|
| 297 |
+
def test_input_rejected_without_trace_does_not_trigger_adversarial_drift(
|
| 298 |
+
baseline: dict[str, Any],
|
| 299 |
+
policy_result: dict[str, Any],
|
| 300 |
+
) -> None:
|
| 301 |
+
telemetry = make_telemetry(
|
| 302 |
+
input_rejected=True,
|
| 303 |
+
rejection_reason="",
|
| 304 |
+
sanitized_input_trace="",
|
| 305 |
+
)
|
| 306 |
+
|
| 307 |
+
assessment = assess_drift(
|
| 308 |
+
telemetry=telemetry,
|
| 309 |
+
baseline=baseline,
|
| 310 |
+
policy_result=policy_result,
|
| 311 |
+
)
|
| 312 |
+
|
| 313 |
+
assert assessment.drift_vector.adversarial == 0.0
|
| 314 |
+
assert not any(s.drift_type == DriftType.ADVERSARIAL for s in assessment.signals)
|
| 315 |
+
|
| 316 |
+
|
| 317 |
+
def test_operational_runtime_drift_detected(
|
| 318 |
+
baseline: dict[str, Any],
|
| 319 |
+
policy_result: dict[str, Any],
|
| 320 |
+
) -> None:
|
| 321 |
telemetry = make_telemetry(duration_ms=1200)
|
|
|
|
|
|
|
| 322 |
|
| 323 |
assessment = assess_drift(
|
| 324 |
telemetry=telemetry,
|
|
|
|
| 330 |
assert any(signal.name == "runtime_boundary_exceeded" for signal in assessment.signals)
|
| 331 |
|
| 332 |
|
| 333 |
+
def test_operational_error_drift_detected(
|
| 334 |
+
baseline: dict[str, Any],
|
| 335 |
+
policy_result: dict[str, Any],
|
| 336 |
+
) -> None:
|
| 337 |
telemetry = make_telemetry(error_count=3)
|
|
|
|
|
|
|
| 338 |
|
| 339 |
assessment = assess_drift(
|
| 340 |
telemetry=telemetry,
|
|
|
|
| 346 |
assert any(signal.name == "error_threshold_exceeded" for signal in assessment.signals)
|
| 347 |
|
| 348 |
|
| 349 |
+
def test_operational_timeout_drift_detected(
|
| 350 |
+
baseline: dict[str, Any],
|
| 351 |
+
policy_result: dict[str, Any],
|
| 352 |
+
) -> None:
|
| 353 |
+
telemetry = make_telemetry(timeout_count=2)
|
| 354 |
+
baseline = make_baseline(timeout_threshold=1)
|
| 355 |
+
|
| 356 |
+
assessment = assess_drift(
|
| 357 |
+
telemetry=telemetry,
|
| 358 |
+
baseline=baseline,
|
| 359 |
+
policy_result=policy_result,
|
| 360 |
+
)
|
| 361 |
+
|
| 362 |
+
assert assessment.drift_vector.operational > 0.0
|
| 363 |
+
assert any(signal.name == "timeout_threshold_exceeded" for signal in assessment.signals)
|
| 364 |
+
|
| 365 |
+
|
| 366 |
+
def test_structural_manifest_mismatch_reverts(
|
| 367 |
+
baseline: dict[str, Any],
|
| 368 |
+
policy_result: dict[str, Any],
|
| 369 |
+
) -> None:
|
| 370 |
telemetry = make_telemetry(manifest_hash="manifest_changed")
|
|
|
|
|
|
|
| 371 |
|
| 372 |
assessment = assess_drift(
|
| 373 |
telemetry=telemetry,
|
|
|
|
| 380 |
assert assessment.recommended_correction == "REVERT"
|
| 381 |
|
| 382 |
|
| 383 |
+
def test_structural_dependency_mismatch_reverts(
|
| 384 |
+
baseline: dict[str, Any],
|
| 385 |
+
policy_result: dict[str, Any],
|
| 386 |
+
) -> None:
|
| 387 |
telemetry = make_telemetry(dependency_hash="deps_changed")
|
|
|
|
|
|
|
| 388 |
|
| 389 |
assessment = assess_drift(
|
| 390 |
telemetry=telemetry,
|
|
|
|
| 396 |
assert assessment.recommended_correction == "REVERT"
|
| 397 |
|
| 398 |
|
| 399 |
+
def test_structural_runtime_python_version_mismatch_reverts(
|
| 400 |
+
baseline: dict[str, Any],
|
| 401 |
+
policy_result: dict[str, Any],
|
| 402 |
+
) -> None:
|
| 403 |
+
telemetry = make_telemetry(runtime_python_version="3.13.1")
|
| 404 |
+
|
| 405 |
+
assessment = assess_drift(
|
| 406 |
+
telemetry=telemetry,
|
| 407 |
+
baseline=baseline,
|
| 408 |
+
policy_result=policy_result,
|
| 409 |
+
)
|
| 410 |
+
|
| 411 |
+
assert assessment.drift_vector.structural > 0.0
|
| 412 |
+
assert assessment.recommended_correction == "REVERT"
|
| 413 |
+
assert any(signal.name == "runtime_python_version_changed" for signal in assessment.signals)
|
| 414 |
+
|
| 415 |
+
|
| 416 |
+
def test_behavioral_same_input_different_output_reverts(
|
| 417 |
+
baseline: dict[str, Any],
|
| 418 |
+
policy_result: dict[str, Any],
|
| 419 |
+
) -> None:
|
| 420 |
telemetry = make_telemetry(
|
| 421 |
indicator_hash="hmac_abc123",
|
| 422 |
output_hash="output_changed",
|
|
|
|
| 424 |
baseline = make_baseline(
|
| 425 |
known_output_hashes={"hmac_abc123": "output_good"},
|
| 426 |
)
|
|
|
|
| 427 |
|
| 428 |
assessment = assess_drift(
|
| 429 |
telemetry=telemetry,
|
|
|
|
| 436 |
assert assessment.recommended_correction == "REVERT"
|
| 437 |
|
| 438 |
|
| 439 |
+
def test_behavioral_invalid_schema_reverts(
|
| 440 |
+
baseline: dict[str, Any],
|
| 441 |
+
policy_result: dict[str, Any],
|
| 442 |
+
) -> None:
|
| 443 |
telemetry = make_telemetry(output_schema_valid=False)
|
|
|
|
|
|
|
| 444 |
|
| 445 |
assessment = assess_drift(
|
| 446 |
telemetry=telemetry,
|
|
|
|
| 452 |
assert assessment.recommended_correction == "REVERT"
|
| 453 |
|
| 454 |
|
| 455 |
+
def test_statistical_shift_can_adapt_when_no_higher_priority_signal(
|
| 456 |
+
baseline: dict[str, Any],
|
| 457 |
+
policy_result: dict[str, Any],
|
| 458 |
+
) -> None:
|
| 459 |
telemetry = make_telemetry(indicator_type="ip")
|
| 460 |
baseline = make_baseline(
|
| 461 |
input_type_distribution={"domain": 0.9, "username": 0.1},
|
| 462 |
)
|
|
|
|
| 463 |
|
| 464 |
assessment = assess_drift(
|
| 465 |
telemetry=telemetry,
|
|
|
|
| 472 |
assert assessment.recommended_correction == "ADAPT"
|
| 473 |
|
| 474 |
|
| 475 |
+
def test_statistical_module_usage_shift_detected(
|
| 476 |
+
baseline: dict[str, Any],
|
| 477 |
+
policy_result: dict[str, Any],
|
| 478 |
+
) -> None:
|
| 479 |
+
telemetry = make_telemetry(
|
| 480 |
+
modules_executed=["resource_links", "dns_lookup"],
|
| 481 |
+
)
|
| 482 |
+
baseline = make_baseline(
|
| 483 |
+
module_usage_distribution={"resource_links": 1.0},
|
| 484 |
+
)
|
| 485 |
+
|
| 486 |
+
assessment = assess_drift(
|
| 487 |
+
telemetry=telemetry,
|
| 488 |
+
baseline=baseline,
|
| 489 |
+
policy_result=policy_result,
|
| 490 |
+
)
|
| 491 |
+
|
| 492 |
+
assert assessment.drift_vector.statistical > 0.0
|
| 493 |
+
assert any(signal.name == "module_usage_distribution_shifted" for signal in assessment.signals)
|
| 494 |
+
|
| 495 |
+
|
| 496 |
+
def test_policy_drift_overrides_statistical_adaptation(
|
| 497 |
+
baseline: dict[str, Any],
|
| 498 |
+
) -> None:
|
| 499 |
telemetry = make_telemetry(indicator_type="ip")
|
| 500 |
baseline = make_baseline(
|
| 501 |
input_type_distribution={"domain": 0.9, "username": 0.1},
|
|
|
|
| 524 |
assert assessment.recommended_correction == "REVERT"
|
| 525 |
|
| 526 |
|
| 527 |
+
def test_adversarial_drift_overrides_statistical_adaptation(
|
| 528 |
+
baseline: dict[str, Any],
|
| 529 |
+
policy_result: dict[str, Any],
|
| 530 |
+
) -> None:
|
| 531 |
telemetry = make_telemetry(
|
| 532 |
indicator_type="ip",
|
| 533 |
sanitized_input_trace="http://169.254.169.254/latest",
|
|
|
|
| 535 |
baseline = make_baseline(
|
| 536 |
input_type_distribution={"domain": 0.9, "username": 0.1},
|
| 537 |
)
|
|
|
|
| 538 |
|
| 539 |
assessment = assess_drift(
|
| 540 |
telemetry=telemetry,
|
|
|
|
| 548 |
assert assessment.recommended_correction == "CONSTRAIN"
|
| 549 |
|
| 550 |
|
| 551 |
+
def test_estimate_confidence_increases_with_signal_count_and_tier() -> None:
|
| 552 |
low_signal = DriftSignal(
|
| 553 |
name="weak",
|
| 554 |
drift_type=DriftType.STATISTICAL,
|
|
|
|
| 568 |
|
| 569 |
assert estimate_confidence([]) == 0.0
|
| 570 |
assert estimate_confidence([high_signal]) > estimate_confidence([low_signal])
|
|
|
|
| 571 |
|
| 572 |
+
# Contract: adding a signal should strictly increase confidence.
|
| 573 |
+
assert estimate_confidence([low_signal, high_signal]) > estimate_confidence([high_signal])
|
| 574 |
|
|
|
|
|
|
|
|
|
|
|
|
|
| 575 |
|
| 576 |
+
def test_assess_drift_is_pure_and_does_not_mutate_inputs(
|
| 577 |
+
telemetry: TelemetrySnapshot,
|
| 578 |
+
baseline: dict[str, Any],
|
| 579 |
+
policy_result: dict[str, Any],
|
| 580 |
+
) -> None:
|
| 581 |
telemetry_before = copy.deepcopy(asdict(telemetry))
|
| 582 |
baseline_before = copy.deepcopy(baseline)
|
| 583 |
policy_before = copy.deepcopy(policy_result)
|
|
|
|
| 593 |
assert policy_result == policy_before
|
| 594 |
|
| 595 |
|
| 596 |
+
def test_clean_execution_observes_without_significant_drift(
|
| 597 |
+
telemetry: TelemetrySnapshot,
|
| 598 |
+
baseline: dict[str, Any],
|
| 599 |
+
policy_result: dict[str, Any],
|
| 600 |
+
) -> None:
|
| 601 |
assessment = assess_drift(
|
| 602 |
telemetry=telemetry,
|
| 603 |
baseline=baseline,
|
|
|
|
| 608 |
assert assessment.signals == []
|
| 609 |
assert assessment.dominant_type is None
|
| 610 |
assert assessment.recommended_correction == "OBSERVE"
|
| 611 |
+
assert assessment.confidence == pytest.approx(0.0)
|
tests/test_intent.py
ADDED
|
@@ -0,0 +1,291 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
"""
|
| 2 |
+
tests/test_intent.py
|
| 3 |
+
====================
|
| 4 |
+
|
| 5 |
+
Contract tests for osint_core.intent.
|
| 6 |
+
|
| 7 |
+
Core invariants:
|
| 8 |
+
- Intent packets are immutable.
|
| 9 |
+
- Intent packets do not store raw indicators.
|
| 10 |
+
- Scope boundaries are explicit and validated.
|
| 11 |
+
- Forbidden operations cannot appear in allowed operations.
|
| 12 |
+
- Packets can be signed and verified.
|
| 13 |
+
- Signature tampering is detected.
|
| 14 |
+
- Risk and rollback helpers are deterministic.
|
| 15 |
+
"""
|
| 16 |
+
|
| 17 |
+
from __future__ import annotations
|
| 18 |
+
|
| 19 |
+
from dataclasses import FrozenInstanceError, replace
|
| 20 |
+
|
| 21 |
+
import pytest
|
| 22 |
+
|
| 23 |
+
from osint_core.intent import (
|
| 24 |
+
DEFAULT_FORBIDDEN_OPERATIONS,
|
| 25 |
+
IntentErrorCode,
|
| 26 |
+
IntentPacket,
|
| 27 |
+
IntentValidationError,
|
| 28 |
+
canonical_json,
|
| 29 |
+
create_intent_packet,
|
| 30 |
+
default_rollback_for_risk,
|
| 31 |
+
derive_risk_label,
|
| 32 |
+
find_raw_indicator_fields,
|
| 33 |
+
hash_manifest_payload,
|
| 34 |
+
intent_fingerprint,
|
| 35 |
+
make_scope,
|
| 36 |
+
risk_score,
|
| 37 |
+
sign_payload,
|
| 38 |
+
unsigned_intent_fingerprint,
|
| 39 |
+
validate_intent,
|
| 40 |
+
validate_scope,
|
| 41 |
+
verify_intent_signature,
|
| 42 |
+
)
|
| 43 |
+
|
| 44 |
+
|
| 45 |
+
TEST_SECRET = "test-intent-signing-secret"
|
| 46 |
+
TARGET_HASH = "a" * 64
|
| 47 |
+
MANIFEST_HASH = "b" * 64
|
| 48 |
+
|
| 49 |
+
|
| 50 |
+
def make_valid_scope(**overrides):
|
| 51 |
+
data = {
|
| 52 |
+
"target_hash": TARGET_HASH,
|
| 53 |
+
"indicator_type": "domain",
|
| 54 |
+
"allowed_operations": ["resource_links"],
|
| 55 |
+
"success_criteria": ["links_generated"],
|
| 56 |
+
}
|
| 57 |
+
data.update(overrides)
|
| 58 |
+
return make_scope(**data)
|
| 59 |
+
|
| 60 |
+
|
| 61 |
+
def make_valid_packet(**overrides):
|
| 62 |
+
scope = overrides.pop("scope", make_valid_scope())
|
| 63 |
+
data = {
|
| 64 |
+
"action": "enrich_indicator",
|
| 65 |
+
"purpose": "Generate passive OSINT source links for a validated indicator.",
|
| 66 |
+
"scope": scope,
|
| 67 |
+
"requested_modules": ["resource_links"],
|
| 68 |
+
"expected_side_effects": ["report_created", "audit_event_created"],
|
| 69 |
+
"rollback_strategy": "observe_only",
|
| 70 |
+
"risk_label": "low",
|
| 71 |
+
"manifest_hash": MANIFEST_HASH,
|
| 72 |
+
"signing_secret": TEST_SECRET,
|
| 73 |
+
}
|
| 74 |
+
data.update(overrides)
|
| 75 |
+
return create_intent_packet(**data)
|
| 76 |
+
|
| 77 |
+
|
| 78 |
+
def test_make_scope_adds_default_forbidden_operations():
|
| 79 |
+
scope = make_valid_scope()
|
| 80 |
+
for operation in DEFAULT_FORBIDDEN_OPERATIONS:
|
| 81 |
+
assert operation in scope.forbidden_operations
|
| 82 |
+
assert scope.target_hash == TARGET_HASH
|
| 83 |
+
assert scope.indicator_type == "domain"
|
| 84 |
+
assert scope.allowed_operations == ("resource_links",)
|
| 85 |
+
|
| 86 |
+
|
| 87 |
+
def test_scope_rejects_missing_target_hash():
|
| 88 |
+
result = validate_scope(make_valid_scope(target_hash="c" * 64))
|
| 89 |
+
assert result.ok is True
|
| 90 |
+
with pytest.raises(IntentValidationError) as exc:
|
| 91 |
+
make_valid_scope(target_hash="")
|
| 92 |
+
assert exc.value.code == IntentErrorCode.MISSING_FIELD
|
| 93 |
+
|
| 94 |
+
|
| 95 |
+
def test_scope_rejects_non_hash_target_identity():
|
| 96 |
+
with pytest.raises(IntentValidationError) as exc:
|
| 97 |
+
make_valid_scope(target_hash="example.com")
|
| 98 |
+
assert exc.value.code == IntentErrorCode.INVALID_SCOPE
|
| 99 |
+
|
| 100 |
+
|
| 101 |
+
def test_scope_rejects_empty_allowed_operations():
|
| 102 |
+
with pytest.raises(IntentValidationError) as exc:
|
| 103 |
+
make_valid_scope(allowed_operations=[])
|
| 104 |
+
assert exc.value.code == IntentErrorCode.MISSING_FIELD
|
| 105 |
+
|
| 106 |
+
|
| 107 |
+
def test_scope_rejects_forbidden_operation_overlap():
|
| 108 |
+
with pytest.raises(IntentValidationError) as exc:
|
| 109 |
+
make_valid_scope(allowed_operations=["resource_links", "port_scan"])
|
| 110 |
+
assert exc.value.code == IntentErrorCode.FORBIDDEN_OPERATION_REQUESTED
|
| 111 |
+
|
| 112 |
+
|
| 113 |
+
def test_scope_rejects_invalid_time_horizon():
|
| 114 |
+
with pytest.raises(IntentValidationError) as exc:
|
| 115 |
+
make_valid_scope(time_horizon_seconds=0)
|
| 116 |
+
assert exc.value.code == IntentErrorCode.INVALID_SCOPE
|
| 117 |
+
|
| 118 |
+
with pytest.raises(IntentValidationError) as exc:
|
| 119 |
+
make_valid_scope(time_horizon_seconds=90_000)
|
| 120 |
+
assert exc.value.code == IntentErrorCode.INVALID_SCOPE
|
| 121 |
+
|
| 122 |
+
|
| 123 |
+
def test_create_intent_packet_signs_and_verifies():
|
| 124 |
+
packet = make_valid_packet()
|
| 125 |
+
assert isinstance(packet, IntentPacket)
|
| 126 |
+
assert packet.signature is not None
|
| 127 |
+
assert verify_intent_signature(packet, secret=TEST_SECRET) is True
|
| 128 |
+
|
| 129 |
+
|
| 130 |
+
def test_intent_packet_is_immutable():
|
| 131 |
+
packet = make_valid_packet()
|
| 132 |
+
with pytest.raises(FrozenInstanceError):
|
| 133 |
+
packet.purpose = "mutated" # type: ignore[misc]
|
| 134 |
+
|
| 135 |
+
|
| 136 |
+
def test_unsigned_payload_excludes_signature():
|
| 137 |
+
packet = make_valid_packet()
|
| 138 |
+
payload = packet.unsigned_payload()
|
| 139 |
+
assert "signature" not in payload
|
| 140 |
+
assert packet.signature is not None
|
| 141 |
+
|
| 142 |
+
|
| 143 |
+
def test_signature_tampering_is_detected():
|
| 144 |
+
packet = make_valid_packet()
|
| 145 |
+
tampered = replace(packet, purpose="Changed purpose after signing.")
|
| 146 |
+
with pytest.raises(IntentValidationError) as exc:
|
| 147 |
+
verify_intent_signature(tampered, secret=TEST_SECRET)
|
| 148 |
+
assert exc.value.code == IntentErrorCode.SIGNATURE_MISMATCH
|
| 149 |
+
|
| 150 |
+
|
| 151 |
+
def test_unsigned_packet_fails_verification():
|
| 152 |
+
packet = create_intent_packet(
|
| 153 |
+
action="enrich_indicator",
|
| 154 |
+
purpose="Generate passive links.",
|
| 155 |
+
scope=make_valid_scope(),
|
| 156 |
+
requested_modules=["resource_links"],
|
| 157 |
+
expected_side_effects=["report_created"],
|
| 158 |
+
rollback_strategy="observe_only",
|
| 159 |
+
risk_label="low",
|
| 160 |
+
manifest_hash=MANIFEST_HASH,
|
| 161 |
+
sign=False,
|
| 162 |
+
)
|
| 163 |
+
assert packet.signature is None
|
| 164 |
+
with pytest.raises(IntentValidationError) as exc:
|
| 165 |
+
verify_intent_signature(packet, secret=TEST_SECRET)
|
| 166 |
+
assert exc.value.code == IntentErrorCode.UNSIGNED_PACKET
|
| 167 |
+
|
| 168 |
+
|
| 169 |
+
def test_packet_rejects_invalid_action():
|
| 170 |
+
with pytest.raises(IntentValidationError) as exc:
|
| 171 |
+
make_valid_packet(action="delete_everything") # type: ignore[arg-type]
|
| 172 |
+
assert exc.value.code == IntentErrorCode.INVALID_ACTION
|
| 173 |
+
|
| 174 |
+
|
| 175 |
+
def test_packet_rejects_invalid_risk_label():
|
| 176 |
+
with pytest.raises(IntentValidationError) as exc:
|
| 177 |
+
make_valid_packet(risk_label="extreme") # type: ignore[arg-type]
|
| 178 |
+
assert exc.value.code == IntentErrorCode.INVALID_RISK
|
| 179 |
+
|
| 180 |
+
|
| 181 |
+
def test_packet_rejects_invalid_rollback_strategy():
|
| 182 |
+
with pytest.raises(IntentValidationError) as exc:
|
| 183 |
+
make_valid_packet(rollback_strategy="YOLO") # type: ignore[arg-type]
|
| 184 |
+
assert exc.value.code == IntentErrorCode.INVALID_ROLLBACK
|
| 185 |
+
|
| 186 |
+
|
| 187 |
+
def test_packet_rejects_invalid_manifest_hash():
|
| 188 |
+
with pytest.raises(IntentValidationError) as exc:
|
| 189 |
+
make_valid_packet(manifest_hash="not-a-hash")
|
| 190 |
+
assert exc.value.code == IntentErrorCode.MISSING_FIELD
|
| 191 |
+
|
| 192 |
+
|
| 193 |
+
def test_packet_rejects_empty_purpose():
|
| 194 |
+
with pytest.raises(IntentValidationError) as exc:
|
| 195 |
+
make_valid_packet(purpose=" ")
|
| 196 |
+
assert exc.value.code == IntentErrorCode.MISSING_FIELD
|
| 197 |
+
|
| 198 |
+
|
| 199 |
+
def test_raw_indicator_field_detection():
|
| 200 |
+
payload = {
|
| 201 |
+
"safe": {"target_hash": TARGET_HASH},
|
| 202 |
+
"unsafe": {
|
| 203 |
+
"raw_indicator": "example.com",
|
| 204 |
+
"nested": {"email": "user@example.com"},
|
| 205 |
+
},
|
| 206 |
+
}
|
| 207 |
+
findings = find_raw_indicator_fields(payload)
|
| 208 |
+
assert "unsafe.raw_indicator" in findings
|
| 209 |
+
assert "unsafe.nested.email" in findings
|
| 210 |
+
|
| 211 |
+
|
| 212 |
+
def test_validate_intent_rejects_raw_indicator_like_fields():
|
| 213 |
+
packet = make_valid_packet()
|
| 214 |
+
unsafe_dict = packet.to_dict()
|
| 215 |
+
unsafe_dict["raw_indicator"] = "example.com"
|
| 216 |
+
findings = find_raw_indicator_fields(unsafe_dict)
|
| 217 |
+
assert "raw_indicator" in findings
|
| 218 |
+
|
| 219 |
+
|
| 220 |
+
def test_canonical_json_is_deterministic():
|
| 221 |
+
assert canonical_json({"b": 2, "a": 1}) == canonical_json({"a": 1, "b": 2})
|
| 222 |
+
|
| 223 |
+
|
| 224 |
+
def test_sign_payload_is_deterministic_for_same_payload_and_secret():
|
| 225 |
+
payload = {"a": 1, "b": 2}
|
| 226 |
+
assert sign_payload(payload, TEST_SECRET) == sign_payload(payload, TEST_SECRET)
|
| 227 |
+
assert sign_payload(payload, TEST_SECRET) != sign_payload(payload, "different-secret")
|
| 228 |
+
|
| 229 |
+
|
| 230 |
+
def test_hash_manifest_payload_is_stable():
|
| 231 |
+
payload = {"artifact": "test", "version": "1.0.0"}
|
| 232 |
+
assert hash_manifest_payload(payload) == hash_manifest_payload(payload)
|
| 233 |
+
assert len(hash_manifest_payload(payload)) == 64
|
| 234 |
+
|
| 235 |
+
|
| 236 |
+
def test_intent_fingerprints_are_stable_and_distinct():
|
| 237 |
+
packet = make_valid_packet()
|
| 238 |
+
signed_fp = intent_fingerprint(packet)
|
| 239 |
+
unsigned_fp = unsigned_intent_fingerprint(packet)
|
| 240 |
+
assert len(signed_fp) == 64
|
| 241 |
+
assert len(unsigned_fp) == 64
|
| 242 |
+
assert signed_fp != unsigned_fp
|
| 243 |
+
|
| 244 |
+
|
| 245 |
+
def test_validate_intent_accepts_valid_packet():
|
| 246 |
+
result = validate_intent(make_valid_packet())
|
| 247 |
+
assert result.ok is True
|
| 248 |
+
assert result.errors == ()
|
| 249 |
+
assert result.error_codes == ()
|
| 250 |
+
|
| 251 |
+
|
| 252 |
+
def test_risk_score_mapping():
|
| 253 |
+
assert risk_score("low") == 0.25
|
| 254 |
+
assert risk_score("medium") == 0.5
|
| 255 |
+
assert risk_score("high") == 0.75
|
| 256 |
+
assert risk_score("critical") == 1.0
|
| 257 |
+
|
| 258 |
+
|
| 259 |
+
def test_default_rollback_for_risk():
|
| 260 |
+
assert default_rollback_for_risk("low") == "observe_only"
|
| 261 |
+
assert default_rollback_for_risk("medium") == "disable_module"
|
| 262 |
+
assert default_rollback_for_risk("high") == "sandbox"
|
| 263 |
+
assert default_rollback_for_risk("critical") == "revert"
|
| 264 |
+
|
| 265 |
+
|
| 266 |
+
def test_derive_risk_label_for_low_risk_passive_modules():
|
| 267 |
+
assert derive_risk_label(
|
| 268 |
+
requested_modules=["resource_links"],
|
| 269 |
+
authorized_target=False,
|
| 270 |
+
) == "low"
|
| 271 |
+
|
| 272 |
+
|
| 273 |
+
def test_derive_risk_label_for_conditional_authorized_modules():
|
| 274 |
+
assert derive_risk_label(
|
| 275 |
+
requested_modules=["http_headers"],
|
| 276 |
+
authorized_target=True,
|
| 277 |
+
) == "medium"
|
| 278 |
+
|
| 279 |
+
|
| 280 |
+
def test_derive_risk_label_for_conditional_unauthorized_modules():
|
| 281 |
+
assert derive_risk_label(
|
| 282 |
+
requested_modules=["http_headers"],
|
| 283 |
+
authorized_target=False,
|
| 284 |
+
) == "high"
|
| 285 |
+
|
| 286 |
+
|
| 287 |
+
def test_derive_risk_label_for_forbidden_modules():
|
| 288 |
+
assert derive_risk_label(
|
| 289 |
+
requested_modules=["nmap"],
|
| 290 |
+
authorized_target=True,
|
| 291 |
+
) == "critical"
|
tests/test_orchestrator.py
ADDED
|
@@ -0,0 +1,286 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
"""
|
| 2 |
+
Tests for osint_core.orchestrator module
|
| 3 |
+
"""
|
| 4 |
+
|
| 5 |
+
import pytest
|
| 6 |
+
|
| 7 |
+
from osint_core.orchestrator import (
|
| 8 |
+
OrchestratorAgent,
|
| 9 |
+
ExecutionStatus,
|
| 10 |
+
create_orchestrator,
|
| 11 |
+
list_skills,
|
| 12 |
+
get_skill,
|
| 13 |
+
SKILLS_REGISTRY,
|
| 14 |
+
)
|
| 15 |
+
from osint_core.policy import PolicyDecision
|
| 16 |
+
|
| 17 |
+
|
| 18 |
+
def test_create_orchestrator():
|
| 19 |
+
"""Test orchestrator agent creation"""
|
| 20 |
+
agent = create_orchestrator()
|
| 21 |
+
assert isinstance(agent, OrchestratorAgent)
|
| 22 |
+
assert agent.role == "orchestrator"
|
| 23 |
+
assert len(agent.skills) > 0
|
| 24 |
+
|
| 25 |
+
|
| 26 |
+
def test_list_skills():
|
| 27 |
+
"""Test skills registry listing"""
|
| 28 |
+
skills = list_skills()
|
| 29 |
+
assert isinstance(skills, dict)
|
| 30 |
+
assert "resource_links" in skills
|
| 31 |
+
assert "dns_records" in skills
|
| 32 |
+
assert "http_headers" in skills
|
| 33 |
+
|
| 34 |
+
|
| 35 |
+
def test_get_skill():
|
| 36 |
+
"""Test individual skill retrieval"""
|
| 37 |
+
skill = get_skill("resource_links")
|
| 38 |
+
assert skill is not None
|
| 39 |
+
assert skill.name == "Resource Links"
|
| 40 |
+
assert skill.canonical_name == "resource_links"
|
| 41 |
+
assert skill.requires_authorization is False
|
| 42 |
+
|
| 43 |
+
# Test conditional skill
|
| 44 |
+
http_skill = get_skill("http_headers")
|
| 45 |
+
assert http_skill is not None
|
| 46 |
+
assert http_skill.requires_authorization is True
|
| 47 |
+
|
| 48 |
+
|
| 49 |
+
def test_get_nonexistent_skill():
|
| 50 |
+
"""Test retrieval of non-existent skill"""
|
| 51 |
+
skill = get_skill("nonexistent_skill")
|
| 52 |
+
assert skill is None
|
| 53 |
+
|
| 54 |
+
|
| 55 |
+
def test_create_context_valid_input():
|
| 56 |
+
"""Test execution context creation with valid input"""
|
| 57 |
+
agent = create_orchestrator()
|
| 58 |
+
context = agent.create_context(
|
| 59 |
+
raw_indicator="example.com",
|
| 60 |
+
indicator_type_hint="Domain",
|
| 61 |
+
requested_modules=["resource_links"],
|
| 62 |
+
authorized_target=False,
|
| 63 |
+
passive_only=True,
|
| 64 |
+
)
|
| 65 |
+
|
| 66 |
+
assert context.run_id.startswith("run_")
|
| 67 |
+
assert context.indicator_type == "domain"
|
| 68 |
+
assert context.normalized_indicator == "example.com"
|
| 69 |
+
assert len(context.indicator_hash) == 64 # SHA256 hex
|
| 70 |
+
assert context.requested_modules == ["resource_links"]
|
| 71 |
+
assert context.authorized_target is False
|
| 72 |
+
assert context.passive_only is True
|
| 73 |
+
assert len(context.errors) == 0
|
| 74 |
+
|
| 75 |
+
|
| 76 |
+
def test_create_context_invalid_input():
|
| 77 |
+
"""Test execution context creation with invalid input"""
|
| 78 |
+
agent = create_orchestrator()
|
| 79 |
+
context = agent.create_context(
|
| 80 |
+
raw_indicator="<script>alert('xss')</script>",
|
| 81 |
+
indicator_type_hint="Auto",
|
| 82 |
+
requested_modules=["resource_links"],
|
| 83 |
+
authorized_target=False,
|
| 84 |
+
passive_only=True,
|
| 85 |
+
)
|
| 86 |
+
|
| 87 |
+
assert context.indicator_type == "unknown"
|
| 88 |
+
assert context.normalized_indicator == ""
|
| 89 |
+
assert len(context.errors) > 0
|
| 90 |
+
|
| 91 |
+
|
| 92 |
+
def test_execute_workflow_with_valid_domain():
|
| 93 |
+
"""Test full workflow execution with valid domain"""
|
| 94 |
+
agent = create_orchestrator()
|
| 95 |
+
workflow = agent.execute_workflow(
|
| 96 |
+
raw_indicator="example.com",
|
| 97 |
+
indicator_type_hint="Domain",
|
| 98 |
+
requested_modules=["resource_links", "dns_records"],
|
| 99 |
+
authorized_target=False,
|
| 100 |
+
passive_only=True,
|
| 101 |
+
)
|
| 102 |
+
|
| 103 |
+
assert workflow.validation_result.ok is True
|
| 104 |
+
assert workflow.context.indicator_type == "domain"
|
| 105 |
+
assert workflow.policy_evaluation.decision == PolicyDecision.ALLOW
|
| 106 |
+
assert len(workflow.policy_evaluation.allowed_modules) == 2
|
| 107 |
+
assert "resource_links" in workflow.policy_evaluation.allowed_modules
|
| 108 |
+
assert "dns_records" in workflow.policy_evaluation.allowed_modules
|
| 109 |
+
assert len(workflow.skill_results) == 2
|
| 110 |
+
assert workflow.duration_ms > 0
|
| 111 |
+
|
| 112 |
+
|
| 113 |
+
def test_execute_workflow_blocks_unauthorized_modules():
|
| 114 |
+
"""Test that unauthorized modules are blocked"""
|
| 115 |
+
agent = create_orchestrator()
|
| 116 |
+
workflow = agent.execute_workflow(
|
| 117 |
+
raw_indicator="example.com",
|
| 118 |
+
indicator_type_hint="Domain",
|
| 119 |
+
requested_modules=["resource_links", "http_headers"],
|
| 120 |
+
authorized_target=False, # Not authorized
|
| 121 |
+
passive_only=True,
|
| 122 |
+
)
|
| 123 |
+
|
| 124 |
+
assert workflow.validation_result.ok is True
|
| 125 |
+
assert workflow.policy_evaluation.decision == PolicyDecision.CONSTRAIN
|
| 126 |
+
assert "resource_links" in workflow.policy_evaluation.allowed_modules
|
| 127 |
+
assert "http_headers" in workflow.policy_evaluation.blocked_modules
|
| 128 |
+
# Only resource_links should be executed
|
| 129 |
+
assert len([r for r in workflow.skill_results if r.status == ExecutionStatus.COMPLETED]) == 1
|
| 130 |
+
|
| 131 |
+
|
| 132 |
+
def test_execute_workflow_allows_authorized_modules():
|
| 133 |
+
"""Test that authorized modules are allowed when authorized"""
|
| 134 |
+
agent = create_orchestrator()
|
| 135 |
+
workflow = agent.execute_workflow(
|
| 136 |
+
raw_indicator="example.com",
|
| 137 |
+
indicator_type_hint="Domain",
|
| 138 |
+
requested_modules=["http_headers"],
|
| 139 |
+
authorized_target=True, # Authorized
|
| 140 |
+
passive_only=False, # Not passive-only mode
|
| 141 |
+
)
|
| 142 |
+
|
| 143 |
+
assert workflow.validation_result.ok is True
|
| 144 |
+
assert "http_headers" in workflow.policy_evaluation.allowed_modules
|
| 145 |
+
assert len(workflow.policy_evaluation.blocked_modules) == 0
|
| 146 |
+
|
| 147 |
+
|
| 148 |
+
def test_execute_workflow_with_invalid_input():
|
| 149 |
+
"""Test workflow execution with invalid input"""
|
| 150 |
+
agent = create_orchestrator()
|
| 151 |
+
workflow = agent.execute_workflow(
|
| 152 |
+
raw_indicator="!!!invalid!!!",
|
| 153 |
+
indicator_type_hint="Auto",
|
| 154 |
+
requested_modules=["resource_links"],
|
| 155 |
+
authorized_target=False,
|
| 156 |
+
passive_only=True,
|
| 157 |
+
)
|
| 158 |
+
|
| 159 |
+
assert workflow.validation_result.ok is False
|
| 160 |
+
assert len(workflow.skill_results) == 0
|
| 161 |
+
assert workflow.correction_verb == "REVERT"
|
| 162 |
+
|
| 163 |
+
|
| 164 |
+
def test_execute_workflow_blocks_wrong_indicator_type():
|
| 165 |
+
"""Test that skills requiring specific indicator types are blocked"""
|
| 166 |
+
agent = create_orchestrator()
|
| 167 |
+
workflow = agent.execute_workflow(
|
| 168 |
+
raw_indicator="username123",
|
| 169 |
+
indicator_type_hint="Username",
|
| 170 |
+
requested_modules=["dns_records"], # Requires domain
|
| 171 |
+
authorized_target=False,
|
| 172 |
+
passive_only=True,
|
| 173 |
+
)
|
| 174 |
+
|
| 175 |
+
assert workflow.validation_result.ok is True
|
| 176 |
+
assert workflow.context.indicator_type == "username"
|
| 177 |
+
assert "dns_records" in workflow.policy_evaluation.allowed_modules
|
| 178 |
+
# DNS skill should be blocked because username is not compatible
|
| 179 |
+
dns_result = next((r for r in workflow.skill_results if r.skill_name == "DNS Records"), None)
|
| 180 |
+
assert dns_result is not None
|
| 181 |
+
assert dns_result.status == ExecutionStatus.BLOCKED
|
| 182 |
+
|
| 183 |
+
|
| 184 |
+
def test_drift_detection_with_policy_violations():
|
| 185 |
+
"""Test drift detection when policy violations occur"""
|
| 186 |
+
agent = create_orchestrator()
|
| 187 |
+
workflow = agent.execute_workflow(
|
| 188 |
+
raw_indicator="example.com",
|
| 189 |
+
indicator_type_hint="Domain",
|
| 190 |
+
requested_modules=["http_headers"], # Requires auth
|
| 191 |
+
authorized_target=False, # No auth
|
| 192 |
+
passive_only=True,
|
| 193 |
+
)
|
| 194 |
+
|
| 195 |
+
# Should detect policy drift
|
| 196 |
+
assert workflow.drift_vector["policy"] > 0
|
| 197 |
+
assert workflow.correction_verb in ["CONSTRAIN", "REVERT"]
|
| 198 |
+
|
| 199 |
+
|
| 200 |
+
def test_correction_verb_choices():
|
| 201 |
+
"""Test that correction verbs follow the priority rules"""
|
| 202 |
+
agent = create_orchestrator()
|
| 203 |
+
|
| 204 |
+
# Low drift should result in OBSERVE
|
| 205 |
+
workflow1 = agent.execute_workflow(
|
| 206 |
+
raw_indicator="example.com",
|
| 207 |
+
indicator_type_hint="Domain",
|
| 208 |
+
requested_modules=["resource_links"],
|
| 209 |
+
authorized_target=False,
|
| 210 |
+
passive_only=True,
|
| 211 |
+
)
|
| 212 |
+
assert workflow1.correction_verb == "OBSERVE"
|
| 213 |
+
|
| 214 |
+
# Policy violation should result in CONSTRAIN or REVERT
|
| 215 |
+
workflow2 = agent.execute_workflow(
|
| 216 |
+
raw_indicator="example.com",
|
| 217 |
+
indicator_type_hint="Domain",
|
| 218 |
+
requested_modules=["http_headers"],
|
| 219 |
+
authorized_target=False,
|
| 220 |
+
passive_only=True,
|
| 221 |
+
)
|
| 222 |
+
assert workflow2.correction_verb in ["CONSTRAIN", "REVERT"]
|
| 223 |
+
|
| 224 |
+
|
| 225 |
+
def test_skill_execution_timing():
|
| 226 |
+
"""Test that skill execution tracks duration"""
|
| 227 |
+
agent = create_orchestrator()
|
| 228 |
+
workflow = agent.execute_workflow(
|
| 229 |
+
raw_indicator="example.com",
|
| 230 |
+
indicator_type_hint="Domain",
|
| 231 |
+
requested_modules=["resource_links"],
|
| 232 |
+
authorized_target=False,
|
| 233 |
+
passive_only=True,
|
| 234 |
+
)
|
| 235 |
+
|
| 236 |
+
assert workflow.duration_ms > 0
|
| 237 |
+
for result in workflow.skill_results:
|
| 238 |
+
if result.status == ExecutionStatus.COMPLETED:
|
| 239 |
+
assert result.duration_ms >= 0
|
| 240 |
+
|
| 241 |
+
|
| 242 |
+
def test_skills_registry_structure():
|
| 243 |
+
"""Test that skills registry has correct structure"""
|
| 244 |
+
for skill_name, skill in SKILLS_REGISTRY.items():
|
| 245 |
+
assert skill.canonical_name == skill_name
|
| 246 |
+
assert isinstance(skill.name, str)
|
| 247 |
+
assert isinstance(skill.description, str)
|
| 248 |
+
assert isinstance(skill.required_indicator_types, list)
|
| 249 |
+
assert isinstance(skill.tools, list)
|
| 250 |
+
assert isinstance(skill.requires_authorization, bool)
|
| 251 |
+
assert skill.category in ["validation", "passive_lookup", "conditional_fetch", "analysis"]
|
| 252 |
+
|
| 253 |
+
|
| 254 |
+
def test_url_parsing_skill():
|
| 255 |
+
"""Test URL parsing skill with URL indicator"""
|
| 256 |
+
agent = create_orchestrator()
|
| 257 |
+
workflow = agent.execute_workflow(
|
| 258 |
+
raw_indicator="https://example.com/path",
|
| 259 |
+
indicator_type_hint="URL",
|
| 260 |
+
requested_modules=["local_url_parse"],
|
| 261 |
+
authorized_target=False,
|
| 262 |
+
passive_only=True,
|
| 263 |
+
)
|
| 264 |
+
|
| 265 |
+
assert workflow.validation_result.ok is True
|
| 266 |
+
assert workflow.context.indicator_type == "url"
|
| 267 |
+
assert len(workflow.skill_results) == 1
|
| 268 |
+
result = workflow.skill_results[0]
|
| 269 |
+
assert result.status == ExecutionStatus.COMPLETED
|
| 270 |
+
assert "scheme" in result.data
|
| 271 |
+
|
| 272 |
+
|
| 273 |
+
def test_multiple_modules_execution():
|
| 274 |
+
"""Test execution of multiple modules in parallel"""
|
| 275 |
+
agent = create_orchestrator()
|
| 276 |
+
workflow = agent.execute_workflow(
|
| 277 |
+
raw_indicator="example.com",
|
| 278 |
+
indicator_type_hint="Domain",
|
| 279 |
+
requested_modules=["resource_links", "dns_records"],
|
| 280 |
+
authorized_target=False,
|
| 281 |
+
passive_only=True,
|
| 282 |
+
)
|
| 283 |
+
|
| 284 |
+
assert len(workflow.skill_results) == 2
|
| 285 |
+
completed = [r for r in workflow.skill_results if r.status == ExecutionStatus.COMPLETED]
|
| 286 |
+
assert len(completed) == 2
|
tests/test_scheduler.py
ADDED
|
@@ -0,0 +1,128 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
from osint_core.scheduler import (
|
| 2 |
+
INVARIANT_CHECKS,
|
| 3 |
+
DecisionPacket,
|
| 4 |
+
ScheduleReason,
|
| 5 |
+
ShortcutDebt,
|
| 6 |
+
SystemState,
|
| 7 |
+
fits_deadline,
|
| 8 |
+
safe_utility,
|
| 9 |
+
schedule_decision,
|
| 10 |
+
total_required_time_ms,
|
| 11 |
+
)
|
| 12 |
+
|
| 13 |
+
|
| 14 |
+
def make_packet(**overrides):
|
| 15 |
+
data = {
|
| 16 |
+
"intent_id": "intent_123",
|
| 17 |
+
"action": "enrich_indicator",
|
| 18 |
+
"risk_label": "low",
|
| 19 |
+
"confidence": 0.90,
|
| 20 |
+
"reversibility": 0.90,
|
| 21 |
+
"deadline_ms": 1000,
|
| 22 |
+
"verification_cost_ms": 100,
|
| 23 |
+
"execution_cost_ms": 100,
|
| 24 |
+
"rollback_cost_ms": 100,
|
| 25 |
+
"expected_utility_decay": 0.10,
|
| 26 |
+
"required_checks": INVARIANT_CHECKS,
|
| 27 |
+
"rollback_plan": "observe_only",
|
| 28 |
+
"uncertainty_notes": (),
|
| 29 |
+
}
|
| 30 |
+
data.update(overrides)
|
| 31 |
+
return DecisionPacket(**data)
|
| 32 |
+
|
| 33 |
+
|
| 34 |
+
def test_total_required_time_and_deadline_fit():
|
| 35 |
+
packet = make_packet()
|
| 36 |
+
assert total_required_time_ms(packet) == 300
|
| 37 |
+
assert fits_deadline(packet) is True
|
| 38 |
+
|
| 39 |
+
|
| 40 |
+
def test_fast_path_for_low_risk_reversible_action():
|
| 41 |
+
decision = schedule_decision(make_packet(risk_label="low"))
|
| 42 |
+
assert decision.route == "FAST"
|
| 43 |
+
assert decision.allowed is True
|
| 44 |
+
assert decision.reason == ScheduleReason.FAST_PATH_AVAILABLE
|
| 45 |
+
|
| 46 |
+
|
| 47 |
+
def test_deliberative_path_for_high_risk_action_that_fits_deadline():
|
| 48 |
+
packet = make_packet(
|
| 49 |
+
risk_label="high",
|
| 50 |
+
confidence=0.90,
|
| 51 |
+
reversibility=0.70,
|
| 52 |
+
rollback_plan="sandbox",
|
| 53 |
+
rollback_cost_ms=200,
|
| 54 |
+
deadline_ms=1000,
|
| 55 |
+
)
|
| 56 |
+
decision = schedule_decision(packet)
|
| 57 |
+
assert decision.route == "DELIBERATIVE"
|
| 58 |
+
assert decision.allowed is True
|
| 59 |
+
|
| 60 |
+
|
| 61 |
+
def test_invariant_checks_cannot_be_skipped():
|
| 62 |
+
packet = make_packet(required_checks=("scope_validated",))
|
| 63 |
+
decision = schedule_decision(packet)
|
| 64 |
+
assert decision.route == "FAIL_CLOSED"
|
| 65 |
+
assert decision.allowed is False
|
| 66 |
+
assert decision.reason == ScheduleReason.INVARIANT_VIOLATION
|
| 67 |
+
assert "hash_salt_present" in decision.skipped_checks
|
| 68 |
+
|
| 69 |
+
|
| 70 |
+
def test_high_risk_without_rollback_fails_closed():
|
| 71 |
+
packet = make_packet(
|
| 72 |
+
risk_label="critical",
|
| 73 |
+
rollback_plan="",
|
| 74 |
+
rollback_cost_ms=0,
|
| 75 |
+
)
|
| 76 |
+
decision = schedule_decision(packet)
|
| 77 |
+
assert decision.route == "FAIL_CLOSED"
|
| 78 |
+
assert decision.reason == ScheduleReason.MISSING_ROLLBACK
|
| 79 |
+
|
| 80 |
+
|
| 81 |
+
def test_deadline_too_tight_routes_to_containment_when_reversible():
|
| 82 |
+
packet = make_packet(
|
| 83 |
+
deadline_ms=100,
|
| 84 |
+
verification_cost_ms=100,
|
| 85 |
+
execution_cost_ms=100,
|
| 86 |
+
rollback_cost_ms=100,
|
| 87 |
+
reversibility=0.90,
|
| 88 |
+
)
|
| 89 |
+
decision = schedule_decision(packet)
|
| 90 |
+
assert decision.route == "CONTAINMENT"
|
| 91 |
+
assert decision.reason == ScheduleReason.DEADLINE_TOO_TIGHT
|
| 92 |
+
assert decision.authority_scale == 0.25
|
| 93 |
+
|
| 94 |
+
|
| 95 |
+
def test_deadline_too_tight_and_not_reversible_fails_closed():
|
| 96 |
+
packet = make_packet(
|
| 97 |
+
deadline_ms=100,
|
| 98 |
+
reversibility=0.20,
|
| 99 |
+
)
|
| 100 |
+
decision = schedule_decision(packet)
|
| 101 |
+
assert decision.route == "FAIL_CLOSED"
|
| 102 |
+
assert decision.reason == ScheduleReason.NO_SAFE_ACTION_FITS
|
| 103 |
+
|
| 104 |
+
|
| 105 |
+
def test_shortcut_debt_forces_containment():
|
| 106 |
+
state = SystemState(shortcut_debt=ShortcutDebt(emergency_overrides=2), shortcut_debt_limit=0.70)
|
| 107 |
+
decision = schedule_decision(make_packet(), state)
|
| 108 |
+
assert decision.route == "CONTAINMENT"
|
| 109 |
+
assert decision.reason == ScheduleReason.SHORTCUT_DEBT_TOO_HIGH
|
| 110 |
+
|
| 111 |
+
|
| 112 |
+
def test_contested_trust_state_forces_containment():
|
| 113 |
+
state = SystemState(trust_state="contested")
|
| 114 |
+
decision = schedule_decision(make_packet(), state)
|
| 115 |
+
assert decision.route == "CONTAINMENT"
|
| 116 |
+
assert decision.reason == ScheduleReason.TRUST_STATE_DEGRADED
|
| 117 |
+
|
| 118 |
+
|
| 119 |
+
def test_low_confidence_high_risk_forces_containment():
|
| 120 |
+
packet = make_packet(risk_label="high", confidence=0.20, rollback_plan="sandbox")
|
| 121 |
+
decision = schedule_decision(packet)
|
| 122 |
+
assert decision.route == "CONTAINMENT"
|
| 123 |
+
assert decision.reason == ScheduleReason.LOW_CONFIDENCE
|
| 124 |
+
|
| 125 |
+
|
| 126 |
+
def test_safe_utility_is_bounded():
|
| 127 |
+
score = safe_utility(make_packet())
|
| 128 |
+
assert 0.0 <= score <= 1.0
|