gentle-pi 2.1.2 → 2.3.0

package/tests/review-compact.test.ts

@@ -1,243 +0,0 @@
-import assert from "node:assert/strict";
-import test from "node:test";
-import {
-	CAUSAL_DISPOSITION,
-	COMPACT_EVIDENCE_CLASS,
-	COMPACT_REVIEW_STATE,
-	completeCompactCorrection,
-	completeCompactReview,
-	completeCompactVerification,
-	beginCompactCorrection,
-	createCompactRefuterRequest,
-	createCompactValidatorRequest,
-	createCompactReceipt,
-	freezeCompactValidatorRequest,
-	createCompactReviewState,
-} from "../lib/review-compact.ts";
-import { REVIEW_RISK_TIER } from "../lib/review-risk.ts";
-import {
-	REVIEW_MODE,
-	REVIEW_PROJECTION,
-	type SnapshotV1,
-} from "../lib/review-snapshot.ts";
-import { REVIEW_LENS, REVIEW_ROUTE } from "../lib/review-triggers.ts";
-
-function snapshot(): SnapshotV1 {
-	return {
-		schema: "gentle-ai.review-snapshot/v1",
-		mode: REVIEW_MODE.ORDINARY,
-		repository_root: "/repo",
-		base_tree: "1".repeat(40),
-		complete_snapshot_tree: "2".repeat(40),
-		review_projection: { kind: REVIEW_PROJECTION.COMPLETE },
-		initial_review_tree: "2".repeat(40),
-		genesis_paths: ["src/value.ts"],
-		intended_untracked: [],
-		diff_evidence: {
-			event: "ordinary-start",
-			changedLines: 10,
-			triviality: "unproven",
-			evidenceComplete: true,
-			executableChanged: true,
-			configurationChanged: false,
-			hotPathChanged: false,
-			riskSignal: false,
-			resilienceSignal: false,
-			reliabilitySignal: false,
-		},
-		route: REVIEW_ROUTE.STANDARD,
-		lenses: [REVIEW_LENS.READABILITY],
-		risk_tier: REVIEW_RISK_TIER.MEDIUM,
-		original_changed_lines: 10,
-		correction_budget: 5,
-		policy_hash: "a".repeat(64),
-		object_store: {
-			snapshot_directory: "/tmp/snapshot",
-			object_directory: "/tmp/snapshot/objects",
-			alternate_object_directory: "/repo/.git/objects",
-			metadata_path: "/tmp/snapshot/snapshot.json",
-			sensitivity: "workspace-content",
-			cleanup_trigger: "lineage-terminal",
-			cleanup_action: "delete-isolated-object-store",
-		},
-	};
-}
-
-test("causal review assigns missing IDs and admits only proven candidate-caused severe findings", () => {
-	const reviewing = createCompactReviewState({
-		lineageId: "compact-causal",
-		snapshot: snapshot(),
-		policyHash: "a".repeat(64),
-	});
-	const completed = completeCompactReview(reviewing, {
-		lens_results: [{
-			findings: [
-				{
-					location: "src/value.ts:2",
-					severity: "CRITICAL",
-					claim: "The candidate activates an invalid branch.",
-					evidence_class: COMPACT_EVIDENCE_CLASS.DETERMINISTIC,
-					causal_disposition: CAUSAL_DISPOSITION.BEHAVIOR_ACTIVATED,
-					proof_refs: ["changed-hunk:src/value.ts:2"],
-				},
-				{
-					location: "src/value.ts:1",
-					severity: "BLOCKER",
-					claim: "The base already contains a separate defect.",
-					evidence_class: COMPACT_EVIDENCE_CLASS.DETERMINISTIC,
-					causal_disposition: CAUSAL_DISPOSITION.PRE_EXISTING,
-					proof_refs: ["before-after:base-and-candidate"],
-				},
-			],
-			evidence: ["reviewed exact candidate tree"],
-		}],
-	});
-
-	assert.equal(completed.state, COMPACT_REVIEW_STATE.CORRECTION_REQUIRED);
-	assert.deepEqual(completed.correction_ids, ["READABILITY-002"]);
-	assert.equal(completed.follow_ups.length, 1);
-	assert.equal(completed.follow_ups[0]?.finding_id, "READABILITY-001");
-});
-
-test("unknown, malformed, insufficient, and incomplete inferential severe claims escalate", () => {
-	const reviewing = createCompactReviewState({
-		lineageId: "compact-escalation",
-		snapshot: snapshot(),
-		policyHash: "a".repeat(64),
-	});
-	const lensResults = [{
-			findings: [{
-				location: "src/value.ts:2",
-				severity: "CRITICAL",
-				claim: "The candidate may activate an invalid branch.",
-				evidence_class: COMPACT_EVIDENCE_CLASS.INFERENTIAL,
-				causal_disposition: CAUSAL_DISPOSITION.INTRODUCED,
-				proof_refs: ["differential-test:tests/value.test.ts"],
-			}],
-			evidence: [],
-		}];
-	const completed = completeCompactReview(reviewing, {
-		lens_results: lensResults,
-		refuter_request_hash: createCompactRefuterRequest(reviewing, lensResults)?.request_hash,
-	});
-	assert.equal(completed.state, COMPACT_REVIEW_STATE.ESCALATED);
-	assert.match(completed.escalation_reasons.join("\n"), /complete refuter batch/i);
-});
-
-test("refuter request assigns native IDs and rejects incomplete or foreign batches", () => {
-	const input = { lens_results: [{ findings: [{
-		location: "src/value.ts:2", severity: "CRITICAL", claim: "The branch may fail.",
-		evidence_class: "inferential", causal_disposition: "introduced",
-		proof_refs: ["differential-test:tests/value.test.ts"],
-	}], evidence: [] }] };
-	for (const rows of [[], [
-		{ finding_id: "READABILITY-001", outcome: "refuted", proof_refs: ["before-after:proof"] },
-		{ finding_id: "READABILITY-001", outcome: "refuted", proof_refs: ["before-after:proof"] },
-	], [{ finding_id: "FOREIGN-001", outcome: "refuted", proof_refs: ["before-after:proof"] }]]) {
-		const reviewing = createCompactReviewState({ lineageId: `refuter-${rows.length}-${rows[0]?.finding_id ?? "empty"}`, snapshot: snapshot(), policyHash: "a".repeat(64) });
-		const request = createCompactRefuterRequest(reviewing, input.lens_results);
-		assert.deepEqual(request?.findings.map(({ id }) => id), ["READABILITY-001"]);
-		const completed = completeCompactReview(reviewing, { ...input, refuter_request_hash: request?.request_hash, refuter_results: rows });
-		assert.equal(completed.state, COMPACT_REVIEW_STATE.ESCALATED);
-	}
-});
-
-test("informational malformed and duplicate IDs are replaced without escalation", () => {
-	const completed = completeCompactReview(createCompactReviewState({ lineageId: "informational-ids", snapshot: snapshot(), policyHash: "a".repeat(64) }), {
-		lens_results: [{ findings: [
-			{ id: "bad", location: "src/value.ts:1", severity: "WARNING", claim: "Info.", proof_refs: [] },
-			{ id: "READABILITY-010", location: "src/value.ts:2", severity: "WARNING", claim: "Info.", proof_refs: [] },
-			{ id: "READABILITY-010", location: "src/value.ts:3", severity: "SUGGESTION", claim: "Info.", proof_refs: [] },
-			{ id: "also-bad", location: "src/value.ts:4", severity: "CRITICAL", claim: "Severe.", evidence_class: "deterministic", causal_disposition: "introduced", proof_refs: ["changed-hunk:src/value.ts:4"] },
-		], evidence: [] }],
-	});
-	assert.equal(completed.state, COMPACT_REVIEW_STATE.ESCALATED);
-	assert.equal(completed.escalation_reasons.length, 1);
-	assert.match(completed.escalation_reasons[0] ?? "", /ALSO-BAD/);
-	for (const finding of completed.findings.filter(({ severity }) => severity === "WARNING" || severity === "SUGGESTION")) {
-		assert.equal(completed.outcomes[finding.id], "info");
-	}
-});
-
-test("one forecast, one repository-sized correction, one targeted validator, and final evidence close the lineage", () => {
-	const reviewed = completeCompactReview(createCompactReviewState({
-		lineageId: "compact-correction",
-		snapshot: snapshot(),
-		policyHash: "a".repeat(64),
-	}), {
-		lens_results: [{
-			findings: [{
-				id: "READABILITY-009",
-				location: "src/value.ts:2",
-				severity: "BLOCKER",
-				claim: "The candidate removes the required guard.",
-				evidence_class: "deterministic",
-				causal_disposition: "introduced",
-				proof_refs: ["changed-hunk:src/value.ts:2"],
-			}],
-			evidence: [],
-		}],
-	});
-	const forecast = beginCompactCorrection(reviewed, 3);
-	const correction = {
-		candidate_tree: "3".repeat(40),
-		changed_paths: ["src/value.ts"],
-		changed_lines: 3,
-		fix_diff: "diff --git a/src/value.ts b/src/value.ts",
-		fix_diff_hash: "b".repeat(64),
-	};
-	const proof = {
-		original_criteria: { passed: true, evidence: ["original suite passed"] },
-		correction_regression: { passed: true, evidence: ["guard regression passed"] },
-	};
-	const request = createCompactValidatorRequest(forecast, correction);
-	const frozen = freezeCompactValidatorRequest(forecast, request);
-	const validation = { request_hash: request.request_hash, correction_ids: ["READABILITY-009"], ...proof, fix_caused_findings: [], follow_ups: [] };
-	assert.throws(() => completeCompactCorrection(frozen, { ...correction, candidate_tree: "4".repeat(40) }, [], validation), /frozen validator request/i);
-	assert.equal(completeCompactCorrection(frozen, correction, [], { ...validation, original_criteria: { passed: false, evidence: ["independent acceptance failure"] } }).state, COMPACT_REVIEW_STATE.ESCALATED);
-	const corrected = completeCompactCorrection(frozen, correction, [], validation);
-	assert.equal(corrected.state, COMPACT_REVIEW_STATE.VALIDATING);
-	const terminal = completeCompactVerification(corrected, "full suite passed", true);
-	assert.equal(terminal.state, COMPACT_REVIEW_STATE.APPROVED);
-	const receipt = createCompactReceipt(terminal, "c".repeat(64));
-	assert.equal(receipt.body.original_changed_lines, 10);
-	assert.equal(receipt.body.correction_budget, 5);
-	assert.deepEqual(receipt.body.correction_ids, ["READABILITY-009"]);
-	assert.match(receipt.body.evidence_hash, /^[0-9a-f]{64}$/);
-});
-
-test("correction forecast and actual repository evidence cannot exceed the frozen budget", () => {
-	const reviewed = completeCompactReview(createCompactReviewState({
-		lineageId: "compact-budget",
-		snapshot: snapshot(),
-		policyHash: "a".repeat(64),
-	}), {
-		lens_results: [{
-			findings: [{
-				location: "src/value.ts:2",
-				severity: "BLOCKER",
-				claim: "The candidate removes the required guard.",
-				evidence_class: "deterministic",
-				causal_disposition: "introduced",
-				proof_refs: ["changed-hunk:src/value.ts:2"],
-			}],
-			evidence: [],
-		}],
-	});
-	assert.throws(() => beginCompactCorrection(reviewed, 0), /positive/);
-	const overBudget = beginCompactCorrection(reviewed, 6);
-	assert.equal(overBudget.state, COMPACT_REVIEW_STATE.ESCALATED);
-	const forecast = beginCompactCorrection(reviewed, 5);
-	assert.throws(() => completeCompactCorrection(forecast, {
-		candidate_tree: "3".repeat(40),
-		changed_paths: ["src/value.ts"],
-		changed_lines: 6,
-		fix_diff: "diff",
-		fix_diff_hash: "b".repeat(64),
-	}, [], {
-		correction_ids: forecast.correction_ids,
-		original_criteria: { passed: true, evidence: ["passed"] },
-		correction_regression: { passed: true, evidence: ["passed"] },
-		follow_ups: [],
-	}), /exceeding frozen budget/i);
-});

package/tests/review-refuter-adapter.test.ts

@@ -1,89 +0,0 @@
-import assert from "node:assert/strict";
-import test from "node:test";
-import { normalizeRefuterBatch } from "../lib/review-refuter-adapter.ts";
-import type { CompactRefuterRequest } from "../lib/review-compact.ts";
-
-const request: CompactRefuterRequest = {
-	request_hash: "a".repeat(64),
-	findings: [
-		{
-			id: "READABILITY-001",
-			lens: "review-readability",
-			location: "lib/value.ts:1",
-			severity: "CRITICAL",
-			claim: "The value may be invalid.",
-			evidence_class: "inferential",
-			causal_disposition: "introduced",
-			proof_refs: ["changed-hunk:lib/value.ts:1"],
-		},
-	],
-};
-
-test("refuter adapter normalizes only the exact complete frozen batch", () => {
-	const normalized = normalizeRefuterBatch(request, {
-		schema: "gentle-ai.refuter-result-batch/v1",
-		request_hash: request.request_hash,
-		results: [{
-			finding_id: "READABILITY-001",
-			outcome: "refuted",
-			proof_refs: ["changed-hunk:lib/value.ts:1"],
-		}],
-	});
-
-	assert.deepEqual(normalized, {
-		status: "normalized",
-		refuter_request_hash: request.request_hash,
-		refuter_results: [{
-			finding_id: "READABILITY-001",
-			outcome: "refuted",
-			proof_refs: ["changed-hunk:lib/value.ts:1"],
-		}],
-	});
-});
-
-test("refuter adapter preserves published inconclusive outcomes", () => {
-	const normalized = normalizeRefuterBatch(request, {
-		schema: "gentle-ai.refuter-result-batch/v1",
-		request_hash: request.request_hash,
-		results: [{ finding_id: "READABILITY-001", outcome: "inconclusive", proof_refs: ["changed-hunk:lib/value.ts:1"] }],
-	});
-	assert.equal(normalized.status, "normalized");
-	assert.equal(normalized.status === "normalized" && normalized.refuter_results[0]?.outcome, "inconclusive");
-});
-
-test("refuter adapter accepts independent concrete refuter proof", () => {
-	const normalized = normalizeRefuterBatch(request, {
-		schema: "gentle-ai.refuter-result-batch/v1",
-		request_hash: request.request_hash,
-		results: [{
-			finding_id: "READABILITY-001",
-			outcome: "refuted",
-			proof_refs: ["differential-test:independent reproduction passes"],
-		}],
-	});
-
-	assert.equal(normalized.status, "normalized");
-	assert.deepEqual(normalized.status === "normalized" && normalized.refuter_results[0]?.proof_refs, [
-		"differential-test:independent reproduction passes",
-	]);
-});
-
-test("refuter adapter rejects prose, aliases, invalid outcomes, and empty or malformed proof", () => {
-	for (const batch of [
-		`prose ${JSON.stringify({ schema: "gentle-ai.refuter-result-batch/v1", request_hash: request.request_hash, results: [] })}`,
-		{ schema: "gentle-ai.refuter-result-batch/v1", request_hash: request.request_hash, results: [{ id: "READABILITY-001", resolution: "refuted", proof_refs: ["changed-hunk:lib/value.ts:1"] }] },
-		{ schema: "gentle-ai.refuter-result-batch/v1", request_hash: request.request_hash, results: [{ finding_id: "READABILITY-001", outcome: "refuted", proof_refs: [] }] },
-		{ schema: "gentle-ai.refuter-result-batch/v1", request_hash: request.request_hash, results: [{ finding_id: "READABILITY-001", outcome: "refuted", proof_refs: [""] }] },
-		{ schema: "gentle-ai.refuter-result-batch/v1", request_hash: request.request_hash, results: [{ finding_id: "READABILITY-001", outcome: "refuted", proof_refs: [" untrimmed proof"] }] },
-	]) assert.equal(normalizeRefuterBatch(request, batch).status, "invalid");
-});
-
-test("refuter adapter rejects wrong hashes and incomplete, duplicate, or extra frozen IDs", () => {
-	const row = { finding_id: "READABILITY-001", outcome: "refuted", proof_refs: ["changed-hunk:lib/value.ts:1"] };
-	for (const batch of [
-		{ schema: "gentle-ai.refuter-result-batch/v1", request_hash: "b".repeat(64), results: [row] },
-		{ schema: "gentle-ai.refuter-result-batch/v1", request_hash: request.request_hash, results: [] },
-		{ schema: "gentle-ai.refuter-result-batch/v1", request_hash: request.request_hash, results: [row, row] },
-		{ schema: "gentle-ai.refuter-result-batch/v1", request_hash: request.request_hash, results: [{ ...row, finding_id: "READABILITY-002" }] },
-	]) assert.equal(normalizeRefuterBatch(request, batch).status, "invalid");
-});