Skip to content

fix(deep-scan): preserve convergence across alias consolidation #762

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
Hughhhhcoder wants to merge 1 commit into
base: main
Choose a base branch
from
+217 −34
Open

fix(deep-scan): preserve convergence across alias consolidation #762

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
82 changes: 50 additions & 32 deletions plugins/codex-security/mcp-app/src/deep-scan/artifact-validation.ts
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -77,11 +77,9 @@ export async function validateReducerArtifacts(input: {
} else {
validateRetainedFindings(result, [], previous);
}
const previousFindingIds = new Set((previous?.findings ?? []).map(scanFindingIdentity));
const retainedFindings = previousFindingAssignments(result, previous);
return {
newFindings: result.findings.filter((finding) => (
!previousFindingIds.has(scanFindingIdentity(finding))
)).length
newFindings: result.findings.filter((finding) => !retainedFindings.has(finding)).length
};
}

Expand All @@ -97,23 +95,11 @@ export function reconcileDeepReduction(
if (source.scanId !== result.scanId) throw new Error("Deep reduction source belongs to a different scan.");
if (source.complete === false) throw new Error("Deep reduction source is only a checkpoint, not a complete result.");
}
validateRetainedFindings(result, discoveries.map((discovery) => discovery.result), previous ?? undefined);
retainSourceFindings(result, { discoveries, previous });
const unmatched = new Set(result.findings);
for (const finding of previous?.findings ?? []) {
const previousRefs = findingSourceIds(finding);
const retained = (
previousRefs.length > 0
? result.findings.find((current) => (
findingSourceIds(current).some((ref) => previousRefs.includes(ref))
))
: undefined
) ?? [...unmatched].find((current) => (
scanFindingIdentity(current) === scanFindingIdentity(finding)
));
if (retained) {
validateReductionHasFindings(result, discoveries.map((discovery) => discovery.result), previous ?? undefined);
retainSourceFindings(result, { discoveries, previous }, false);
for (const [retained, previousFindings] of previousFindingAssignments(result, previous)) {
for (const finding of previousFindings) {
preserveFindingDetails(retained, finding);
unmatched.delete(retained);
}
}
retainSourceFindings(result, { discoveries, previous });
Expand Down Expand Up @@ -167,7 +153,11 @@ function findingSourceIds(finding: Record<string, unknown>): string[] {
));
}

function retainSourceFindings(result: ScanDraftInput, inputs: DeepReductionSources): void {
function retainSourceFindings(
result: ScanDraftInput,
inputs: DeepReductionSources,
requireComplete = true,
): void {
type Finding = Record<string, unknown>;
const sources = new Map<string, Finding>();
for (const discovery of inputs.discoveries) {
Expand Down Expand Up @@ -209,12 +199,12 @@ function retainSourceFindings(result: ScanDraftInput, inputs: DeepReductionSourc
}
}
const missing = [...sources.keys()].filter((id) => !claimed.has(id));
if (missing.length) throw new Error(`Deep reduction left unaccounted source findings: ${missing.join(", ")}.`);
if (requireComplete && missing.length) {
throw new Error(`Deep reduction left unaccounted source findings: ${missing.join(", ")}.`);
}
}


/** Preserve previously accepted identities and never discard every reported finding. */
export function validateRetainedFindings(
function validateReductionHasFindings(
result: ScanDraftInput,
sources: ScanDraftInput[],
previous?: ScanDraftInput
Expand All @@ -226,16 +216,44 @@ export function validateRetainedFindings(
) {
throw new Error("Deep reduction discarded every accepted Standard scan finding.");
}
}

const currentFindingIds = new Set(result.findings.map(scanFindingIdentity));
function previousFindingAssignments(
result: ScanDraftInput,
previous?: ScanDraftInput | null,
): Map<ScanDraftInput["findings"][number], ScanDraftInput["findings"]> {
const assignments = new Map<ScanDraftInput["findings"][number], ScanDraftInput["findings"]>();
for (const finding of previous?.findings ?? []) {
if (currentFindingIds.has(scanFindingIdentity(finding))) continue;
throw Object.assign(new Error(
"Deep reduction discarded or changed a previously accepted finding identity."
), {
code: "merge_traceability_unstable_candidate_id"
});
const previousRefs = findingSourceIds(finding);
const matches = previousRefs.length > 0
? result.findings.filter((current) => {
const currentRefs = new Set(findingSourceIds(current));
return previousRefs.every((ref) => currentRefs.has(ref));
})
: result.findings.filter((current) => scanFindingIdentity(current) === scanFindingIdentity(finding));
if (matches.length !== 1) {
throw Object.assign(new Error(
"Deep reduction discarded, split, or ambiguously reassigned a previously accepted finding."
), {
code: "merge_traceability_unstable_candidate_id"
});
}
const retained = matches[0]!;
const assigned = assignments.get(retained) ?? [];
assigned.push(finding);
assignments.set(retained, assigned);
}
return assignments;
}

/** Preserve previous source lineage (or legacy identity) and never discard every finding. */
export function validateRetainedFindings(
result: ScanDraftInput,
sources: ScanDraftInput[],
previous?: ScanDraftInput
): void {
validateReductionHasFindings(result, sources, previous);
previousFindingAssignments(result, previous);
}

function parseStoredScanDraft(
Expand Down
169 changes: 167 additions & 2 deletions plugins/codex-security/mcp-app/tests/test_deep_scan_artifact_validation.mjs
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -33,6 +33,7 @@ const root = await realpath(await mkdtemp(path.join(tmpdir(), "deep-scan-artifac
try {
await testDiscoveryValidation(root);
await testReducerValidation(root);
await testReducerLineageConvergence(root);
await testEmptyDiscoveryAndReduction(root);
} finally {
await rm(root, { recursive: true, force: true });
Expand Down Expand Up @@ -384,13 +385,13 @@ async function testReducerValidation(root) {
provenance: { source: "local_plugin", sourceFindingIds: ["origin:b"] },
},
]));
await validateReducerArtifacts({
assert.deepEqual(await validateReducerArtifacts({
artifacts,
artifactDir,
resultPath,
reducerId: "dedup-colliding-previous",
sources: collidingSources,
}, scanId);
}, scanId), { newFindings: 0 });
const reconciledCollisions = JSON.parse(await readFile(resultPath, "utf8")).findings;
assert.deepEqual(
reconciledCollisions.map((item) => item.provenance.sourceFindingIds),
Expand Down Expand Up @@ -521,6 +522,170 @@ async function testReducerValidation(root) {
}
}

async function testReducerLineageConvergence(root) {
const artifacts = await createLayout(path.join(root, "lineage-convergence"));
const artifactDir = path.join(artifacts.dedupRoot, "dedup-lineage", "output");
const resultPath = path.join(artifactDir, "result.json");
await mkdir(artifactDir, { recursive: true });

const originalA = finding("alias-a", "src/alias-a.js");
const originalB = finding("alias-b", "src/alias-b.js");
const previousA = {
...originalA,
provenance: {
source: "local_plugin",
sourceFindingIds: ["alias:a"],
sourceFindings: [{ id: "alias:a", finding: originalA }]
}
};
const previousB = {
...originalB,
provenance: {
source: "local_plugin",
sourceFindingIds: ["alias:b"],
sourceFindings: [{ id: "alias:b", finding: originalB }]
}
};
const previousAliases = draft([previousA, previousB]);

await writeResult(resultPath, draft([{
...originalA,
provenance: {
source: "local_plugin",
sourceFindingIds: ["alias:a", "alias:b"]
}
}]));
assert.deepEqual(await validateReducerArtifacts({
artifacts,
artifactDir,
resultPath,
reducerId: "dedup-alias-collapse",
sources: { discoveries: [], previous: previousAliases }
}, scanId), { newFindings: 0 });
const reconciledAlias = JSON.parse(await readFile(resultPath, "utf8")).findings[0];
assert.deepEqual(reconciledAlias.provenance.sourceFindingIds, ["alias:a", "alias:b"]);
assert.equal(reconciledAlias.provenance.previousFindings[0].identity.anchor, "alias-b");

const previousResultPath = path.join(
artifacts.dedupRoot,
"dedup-lineage-previous",
"output",
"result.json"
);
await mkdir(path.dirname(previousResultPath), { recursive: true });
await writeResult(previousResultPath, previousAliases);
assert.deepEqual(await validateReducerArtifacts({
artifacts,
artifactDir,
resultPath,
reducerId: "dedup-alias-collapse-recovery",
previousReducerResultPath: previousResultPath
}, scanId), { newFindings: 0 });

const freshEvidence = finding("fresh-evidence", "src/fresh-evidence.js");
const freshSources = [{ workerId: "worker-fresh", result: draft([freshEvidence]) }];
const previousCombined = {
...originalA,
provenance: {
source: "local_plugin",
sourceFindingIds: ["alias:a", "alias:b"],
sourceFindings: [
{ id: "alias:a", finding: originalA },
{ id: "alias:b", finding: originalB }
]
}
};
await writeResult(resultPath, draft([
{
...originalA,
provenance: { source: "local_plugin", sourceFindingIds: ["alias:a"] }
},
{
...originalB,
provenance: {
source: "local_plugin",
sourceFindingIds: ["alias:b", "worker-fresh:0"]
}
}
]));
await assert.rejects(validateReducerArtifacts({
artifacts,
artifactDir,
resultPath,
reducerId: "dedup-lineage-split",
sources: { discoveries: freshSources, previous: draft([previousCombined]) }
}, scanId), (error) => (
error.code === "merge_traceability_unstable_candidate_id"
&& /split/.test(error.message)
));

await writeResult(resultPath, draft([{
...finding("merged-root", "src/merged-root.js"),
provenance: {
source: "local_plugin",
sourceFindingIds: ["alias:a", "alias:b", "worker-fresh:0"]
}
}]));
assert.deepEqual(await validateReducerArtifacts({
artifacts,
artifactDir,
resultPath,
reducerId: "dedup-existing-root-new-evidence",
sources: { discoveries: freshSources, previous: previousAliases }
}, scanId), { newFindings: 0 });

await writeResult(resultPath, draft([
{
...originalA,
provenance: {
source: "local_plugin",
sourceFindingIds: ["alias:a", "alias:b"]
}
},
{
...freshEvidence,
provenance: {
source: "local_plugin",
sourceFindingIds: ["worker-fresh:0"]
}
}
]));
assert.deepEqual(await validateReducerArtifacts({
artifacts,
artifactDir,
resultPath,
reducerId: "dedup-genuinely-new-root",
sources: { discoveries: freshSources, previous: previousAliases }
}, scanId), { newFindings: 1 });

const legacyPrevious = {
...originalA,
provenance: { source: "local_plugin" }
};
await writeResult(resultPath, draft([originalA]));
assert.deepEqual(await validateReducerArtifacts({
artifacts,
artifactDir,
resultPath,
reducerId: "dedup-legacy-identity",
sources: { discoveries: [], previous: draft([legacyPrevious]) }
}, scanId), { newFindings: 0 });
await writeResult(resultPath, draft([{
...finding("legacy-renamed", "src/alias-a.js"),
provenance: {
source: "local_plugin",
sourceFindingIds: ["previous:0"]
}
}]));
await assert.rejects(validateReducerArtifacts({
artifacts,
artifactDir,
resultPath,
reducerId: "dedup-legacy-renamed",
sources: { discoveries: [], previous: draft([legacyPrevious]) }
}, scanId), (error) => error.code === "merge_traceability_unstable_candidate_id");
}

async function testEmptyDiscoveryAndReduction(root) {
const artifacts = await createLayout(path.join(root, "empty"));
const worker = await createWorker(
Expand Down