fix(javascript): trace wrapped Electron IPC paths in large bundles (#1874)

Co-authored-by: David <17119984+fengdwx@users.noreply.github.com>
This commit is contained in:
DavidandDavid authored and GitHub committed 2026-10-12 09:28:59 +08:00
1 parent d93fbda390
commit 87b9793c53
46 files changed
+3213 -355

No files matched your search

@@ -41,7 +41,11 @@ import {
type ApplicationCoverage,
} from "../../domain/javascript/javascriptApplicationEvidenceSchemas.js";
import type { JsonValue } from "../../domain/jsonValue.js";
import { javaScriptAnalysisCoverageLimits } from "../../domain/javascript/javascriptAnalysisResourceControls.js";
import {
JAVASCRIPT_ANALYSIS_RESOURCE_DEFAULTS,
JAVASCRIPT_SEMANTIC_NODE_MAXIMUM,
javaScriptAnalysisCoverageLimits,
} from "../../domain/javascript/javascriptAnalysisResourceControls.js";
import { projectSemanticFunctionFingerprints } from "./JavaScriptSemanticGraphFingerprintProjection.js";
import {
projectSemanticClosureCaptures,
@@ -65,6 +69,8 @@ import {
type SemanticGraphProjectionState as BuilderState,
} from "./JavaScriptSemanticGraphConstruction.js";
import type { JavaScriptSemanticFileProjection } from "../../domain/javascript/javascriptSemanticFileProjection.js";
import type { ElectronCallPath } from "../../domain/javascript/electronStaticAnalysisTypes.js";
import { projectElectronSemanticCallPaths } from "./JavaScriptSemanticGraphElectronProjection.js";
interface BuilderInput {
readonly rootArtifactSha256: string;
@@ -105,13 +111,13 @@ interface FileContext extends SemanticFlowProjectionContext {
* (`truncated`, `omitted_nodes`, `omitted_relations`, `limits`) rather than
* silently dropping data.
*
* Sizing: a retained semantic node costs roughly 3 KB once serialized, and V8
* additionally caps a single string at 512 MB. The result is serialized after the
* object graph is built and validated, so both phases must fit together. A single
* minified vendor bundle can account for most of a tree's nodes, so a whole-tree
* ceiling without a per-file ceiling does not bound the work. Both are applied.
* Defaults bound work even when ample heap is available. Callers may raise the
* capacities explicitly; isolated worker heap, bounded record transport and
* parent result admission independently constrain actual retained memory. The
* result uses incremental serialization rather than a single JSON string.
*/
export const SEMANTIC_GRAPH_NODE_CEILING = 100_000;
export const SEMANTIC_GRAPH_NODE_CEILING =
JAVASCRIPT_ANALYSIS_RESOURCE_DEFAULTS.semanticNodes;
/**
* Per-file share of the semantic node budget.
@@ -119,7 +125,8 @@ export const SEMANTIC_GRAPH_NODE_CEILING = 100_000;
* Bound each source file independently so one large bundled library cannot
* consume the whole tree's budget and starve the application's own modules.
*/
export const SEMANTIC_GRAPH_FILE_NODE_CEILING = 20_000;
export const SEMANTIC_GRAPH_FILE_NODE_CEILING =
JAVASCRIPT_ANALYSIS_RESOURCE_DEFAULTS.semanticNodesPerFile;
export const buildJavaScriptSemanticGraph = ({
rootArtifactSha256,
@@ -129,7 +136,11 @@ export const buildJavaScriptSemanticGraph = ({
const projection = createJavaScriptSemanticGraphProjection();
for (const analyzed of analysis.files) {
if (analyzed.semantic !== null)
projection.projectFile(analyzed.file, analyzed.semantic.ir);
projection.projectFile(
analyzed.file,
analyzed.semantic.ir,
analyzed.javascript?.electron.call_paths,
);
else if (
!analyzed.file.text.included &&
analyzed.file.text.reason === "outside-analysis-scope"
@@ -162,11 +173,13 @@ export interface JavaScriptSemanticGraphProjection {
readonly projectFile: (
file: JavaScriptArtifactFile,
ir: JavaScriptSemanticIr,
paths?: readonly ElectronCallPath[],
) => void;
/** Project one file in steps between its projection passes. */
readonly projectFileSteps: (
file: JavaScriptArtifactFile,
ir: JavaScriptSemanticIr,
paths?: readonly ElectronCallPath[],
) => Generator<void, void>;
readonly finish: (
rootArtifactSha256: string,
@@ -179,234 +192,241 @@ export interface JavaScriptSemanticGraphProjection {
}
/** Accumulate semantic nodes while each source IR is still file-local. */
export const createJavaScriptSemanticGraphProjection =
(): JavaScriptSemanticGraphProjection => {
const state = emptyState();
const fingerprints: JavaScriptSemanticFingerprint[] = [];
let truncatedFiles = 0;
let unavailableFiles = 0;
let resourceTruncatedFiles = 0;
const executionLimits = new Map<
string,
ApplicationCoverage["limits"][number]
>();
const projectSourceSteps = function* (
file: JavaScriptArtifactFile,
ir: JavaScriptSemanticIr,
): Generator<void, void> {
if (state.nodes.size >= SEMANTIC_GRAPH_NODE_CEILING) {
truncatedFiles += 1;
return;
}
const remainingTreeBudget =
SEMANTIC_GRAPH_NODE_CEILING - state.nodes.size;
state.fileNodeBudget = Math.min(
SEMANTIC_GRAPH_FILE_NODE_CEILING,
remainingTreeBudget,
);
state.fileNodesDropped = false;
fingerprints.push(...(yield* projectFileSteps(file, ir, state)));
// Nodes were dropped only if the budget actually blocked creation. A file
// that exactly fills its share ends with a zero budget without dropping
// anything, so the remaining budget alone cannot decide truncation.
if (state.fileNodesDropped) truncatedFiles += 1;
state.fileNodeBudget = null;
};
const finishWith = <Value>(
factory: (input: unknown) => Value,
rootArtifactSha256: string,
applicationGraph: BuilderInput["applicationGraph"],
): Value => {
bindSemanticGraphApplicationNodes(state, applicationGraph);
if (state.roots.size === 0) addFallbackRoot(rootArtifactSha256, state);
const relations = [...state.relations.values()];
const unknowns = [...state.unknowns.values()];
type Family = (typeof JAVASCRIPT_SEMANTIC_RELATION_FAMILIES)[number];
const retainedByFamily = new Map<Family, number>();
const unknownIdsByFamily = new Map<Family, string[]>();
for (const relation of relations) {
const family = JAVASCRIPT_SEMANTIC_RELATION_FAMILY[relation.relation];
retainedByFamily.set(family, (retainedByFamily.get(family) ?? 0) + 1);
}
for (const unknown of unknowns) {
const identifiers = unknownIdsByFamily.get(unknown.family);
if (identifiers === undefined)
unknownIdsByFamily.set(unknown.family, [unknown.unknown_id]);
else identifiers.push(unknown.unknown_id);
}
const graph = factory({
schema: "JavaScriptSemanticRelationGraph",
root_artifact_sha256: rootArtifactSha256,
application_graph_id: applicationGraph.graph_id,
root_node_ids: [...state.roots],
evidence_contexts: state.evidenceContexts.contexts,
nodes: [...state.nodes.values()],
relations,
// The input is transferred to the factory; `fingerprints` is cleared below.
fingerprints: [...fingerprints],
unknowns,
coverage: {
status:
truncatedFiles > 0 || unavailableFiles > 0 ? "partial" : "unknown",
truncated: truncatedFiles > 0 || resourceTruncatedFiles > 0,
omitted_nodes: truncatedFiles > 0 || unavailableFiles > 0 ? null : 0,
export const createJavaScriptSemanticGraphProjection = (
limits: { readonly totalNodes?: number; readonly fileNodes?: number } = {},
): JavaScriptSemanticGraphProjection => {
const nodeCeiling = limits.totalNodes ?? SEMANTIC_GRAPH_NODE_CEILING;
const fileNodeCeiling = limits.fileNodes ?? SEMANTIC_GRAPH_FILE_NODE_CEILING;
for (const value of [nodeCeiling, fileNodeCeiling])
if (
!Number.isSafeInteger(value) ||
value < 1 ||
value > JAVASCRIPT_SEMANTIC_NODE_MAXIMUM
)
throw new RangeError("Semantic graph requires valid node capacities");
const state = emptyState();
const fingerprints: JavaScriptSemanticFingerprint[] = [];
let truncatedFiles = 0;
let unavailableFiles = 0;
let resourceTruncatedFiles = 0;
const executionLimits = new Map<
string,
ApplicationCoverage["limits"][number]
>();
const projectSourceSteps = function* (
file: JavaScriptArtifactFile,
ir: JavaScriptSemanticIr,
paths: readonly ElectronCallPath[] = [],
): Generator<void, void> {
if (state.nodes.size >= nodeCeiling) {
truncatedFiles += 1;
return;
}
const remainingTreeBudget = nodeCeiling - state.nodes.size;
state.fileNodeBudget = Math.min(fileNodeCeiling, remainingTreeBudget);
state.fileNodesDropped = false;
fingerprints.push(...(yield* projectFileSteps(file, ir, state, paths)));
// Nodes were dropped only if the budget actually blocked creation. A file
// that exactly fills its share ends with a zero budget without dropping
// anything, so the remaining budget alone cannot decide truncation.
if (state.fileNodesDropped) truncatedFiles += 1;
state.fileNodeBudget = null;
};
const finishWith = <Value>(
factory: (input: unknown) => Value,
rootArtifactSha256: string,
applicationGraph: BuilderInput["applicationGraph"],
): Value => {
bindSemanticGraphApplicationNodes(state, applicationGraph);
if (state.roots.size === 0) addFallbackRoot(rootArtifactSha256, state);
const relations = [...state.relations.values()];
const unknowns = [...state.unknowns.values()];
type Family = (typeof JAVASCRIPT_SEMANTIC_RELATION_FAMILIES)[number];
const retainedByFamily = new Map<Family, number>();
const unknownIdsByFamily = new Map<Family, string[]>();
for (const relation of relations) {
const family = JAVASCRIPT_SEMANTIC_RELATION_FAMILY[relation.relation];
retainedByFamily.set(family, (retainedByFamily.get(family) ?? 0) + 1);
}
for (const unknown of unknowns) {
const identifiers = unknownIdsByFamily.get(unknown.family);
if (identifiers === undefined)
unknownIdsByFamily.set(unknown.family, [unknown.unknown_id]);
else identifiers.push(unknown.unknown_id);
}
const graph = factory({
schema: "JavaScriptSemanticRelationGraph",
root_artifact_sha256: rootArtifactSha256,
application_graph_id: applicationGraph.graph_id,
root_node_ids: [...state.roots],
evidence_contexts: state.evidenceContexts.contexts,
nodes: [...state.nodes.values()],
relations,
// The input is transferred to the factory; `fingerprints` is cleared below.
fingerprints: [...fingerprints],
unknowns,
coverage: {
status:
truncatedFiles > 0 || unavailableFiles > 0 ? "partial" : "unknown",
truncated: truncatedFiles > 0 || resourceTruncatedFiles > 0,
omitted_nodes: truncatedFiles > 0 || unavailableFiles > 0 ? null : 0,
omitted_relations:
truncatedFiles > 0 || unavailableFiles > 0 ? null : 0,
limits: [
...executionLimits.values(),
...(truncatedFiles > 0
? [
{
name: "semantic_graph_node_ceiling",
value: nodeCeiling,
unit: "items" as const,
},
{
name: "semantic_graph_file_node_ceiling",
value: fileNodeCeiling,
unit: "items" as const,
},
]
: []),
],
families: JAVASCRIPT_SEMANTIC_RELATION_FAMILIES.map((family) => ({
family,
status: semanticFamilyStatus(family),
retained_relations: retainedByFamily.get(family) ?? 0,
omitted_relations:
truncatedFiles > 0 || unavailableFiles > 0 ? null : 0,
limits: [
...executionLimits.values(),
...(truncatedFiles > 0
? [
{
name: "semantic_graph_node_ceiling",
value: SEMANTIC_GRAPH_NODE_CEILING,
unit: "items" as const,
},
]
: []),
],
families: JAVASCRIPT_SEMANTIC_RELATION_FAMILIES.map((family) => ({
family,
status: semanticFamilyStatus(family),
retained_relations: retainedByFamily.get(family) ?? 0,
omitted_relations:
truncatedFiles > 0 || unavailableFiles > 0 ? null : 0,
unknown_ids: unknownIdsByFamily.get(family) ?? [],
})),
},
limitations: [
"The semantic graph contains static syntax observations and conservative relationship candidates; it does not claim runtime execution.",
"Local data flow does not claim control-flow-sensitive reaching definitions or arbitrary dynamic property resolution.",
"Promise ownership covers explicit unshadowed Promise construction, static factories, aggregation, chaining, and await syntax only.",
"Function fingerprints are static candidates; equal digests can remain ambiguous and do not prove behavioral equivalence.",
"Event extraction covers EventEmitter-style literal registrations, removals, and dispatch candidates; dynamic names remain unknown.",
"Timer extraction covers global or node:timers scheduling and exact local-handle cancellation.",
"Child-process extraction covers asynchronous node:child_process creation, literal argv/env/stdio options, exit/error listeners, and kill signals.",
"Configuration extraction covers process.env, process.argv, node:fs reads, and direct logical defaults.",
"Request extraction covers fetch, WebSocket, node:http/node:https construction, direct option fields, and exact local response consumers.",
"Boundary extraction covers unshadowed JSON/global coercions plus parse and validation method candidates.",
"Resource extraction covers built-in filesystem/network acquisition and exact local close/destroy/end handles.",
],
});
state.nodes.clear();
state.relations.clear();
state.unknowns.clear();
state.roots.clear();
state.evidenceContexts.clear();
fingerprints.length = 0;
return graph;
};
return {
remainingFileNodeBudget: () =>
Math.min(
SEMANTIC_GRAPH_FILE_NODE_CEILING,
Math.max(0, SEMANTIC_GRAPH_NODE_CEILING - state.nodes.size),
),
adoptFileProjection: (projection) => {
const newNodes = projection.nodes.filter(
({ node_id }) => !state.nodes.has(node_id),
);
if (
newNodes.length >
Math.min(
SEMANTIC_GRAPH_FILE_NODE_CEILING,
SEMANTIC_GRAPH_NODE_CEILING - state.nodes.size,
)
)
throw new RangeError(
"Transferred semantic nodes exceed the offered tree capacity",
);
for (const context of projection.evidenceContexts)
state.evidenceContexts.adopt(context);
for (const node of newNodes) state.nodes.set(node.node_id, node);
for (const relation of projection.relations)
state.relations.set(relation.relation_id, relation);
for (const unknown of projection.unknowns)
state.unknowns.set(unknown.unknown_id, unknown);
for (const root of projection.roots) state.roots.add(root);
fingerprints.push(...projection.fingerprints);
if (projection.truncated) truncatedFiles += 1;
unknown_ids: unknownIdsByFamily.get(family) ?? [],
})),
},
recordUnavailableFile: (file, reason, limits, unknownReason) => {
const retainedLimits = javaScriptAnalysisCoverageLimits(limits);
for (const limit of retainedLimits) {
executionLimits.set(
JSON.stringify([limit.name, limit.value, limit.unit]),
limit,
);
}
state.fileNodeBudget = Math.min(
SEMANTIC_GRAPH_FILE_NODE_CEILING,
Math.max(0, SEMANTIC_GRAPH_NODE_CEILING - state.nodes.size),
);
const module = retainNode(state, file, {
kind: "module",
roleKey: "module",
location: null,
label: file.path,
functionNodeId: null,
});
state.fileNodeBudget = null;
if (module !== null) state.roots.add(module.node_id);
const evidence = {
...observedSemanticEvidence(file, null),
authority: "unknown" as const,
state: "unknown" as const,
confidence: "unknown" as const,
coverage: partialApplicationCoverage(retainedLimits, null),
limitations: [reason],
};
for (const family of JAVASCRIPT_SEMANTIC_RELATION_FAMILIES) {
const unknown = createJavaScriptSemanticGraphUnknown(
{
node_id: module?.node_id ?? null,
family,
relation_kinds: JAVASCRIPT_SEMANTIC_RELATIONS.filter(
(relation) =>
JAVASCRIPT_SEMANTIC_RELATION_FAMILY[relation] === family,
),
reason: unknownReason,
detail: reason,
candidate_node_ids: [],
evidence,
},
state.evidenceContexts,
);
state.unknowns.set(unknown.unknown_id, unknown);
}
unavailableFiles += 1;
if (unknownReason === "resource-limit" && retainedLimits.length > 0)
resourceTruncatedFiles += 1;
},
projectFile: (file, ir) => {
const steps = projectSourceSteps(file, ir);
while (steps.next().done !== true);
},
projectFileSteps: projectSourceSteps,
finish: (rootArtifactSha256, applicationGraph) =>
finishWith(
createJavaScriptSemanticGraph,
rootArtifactSha256,
applicationGraph,
),
finishImmutableSteps: (rootArtifactSha256, applicationGraph) =>
finishWith(
sealTransferredJavaScriptSemanticGraphSteps,
rootArtifactSha256,
applicationGraph,
),
};
limitations: [
"The semantic graph contains static syntax observations and conservative relationship candidates; it does not claim runtime execution.",
"Local data flow does not claim control-flow-sensitive reaching definitions or arbitrary dynamic property resolution.",
"Promise ownership covers explicit unshadowed Promise construction, static factories, aggregation, chaining, and await syntax only.",
"Function fingerprints are static candidates; equal digests can remain ambiguous and do not prove behavioral equivalence.",
"Event extraction covers EventEmitter-style literal registrations, removals, and dispatch candidates; dynamic names remain unknown.",
"Timer extraction covers global or node:timers scheduling and exact local-handle cancellation.",
"Child-process extraction covers asynchronous node:child_process creation, literal argv/env/stdio options, exit/error listeners, and kill signals.",
"Configuration extraction covers process.env, process.argv, node:fs reads, and direct logical defaults.",
"Request extraction covers fetch, WebSocket, node:http/node:https construction, direct option fields, and exact local response consumers.",
"Boundary extraction covers unshadowed JSON/global coercions plus parse and validation method candidates.",
"Resource extraction covers built-in filesystem/network acquisition and exact local close/destroy/end handles.",
],
});
state.nodes.clear();
state.relations.clear();
state.unknowns.clear();
state.roots.clear();
state.evidenceContexts.clear();
fingerprints.length = 0;
return graph;
};
return {
remainingFileNodeBudget: () =>
Math.min(fileNodeCeiling, Math.max(0, nodeCeiling - state.nodes.size)),
adoptFileProjection: (projection) => {
const newNodes = projection.nodes.filter(
({ node_id }) => !state.nodes.has(node_id),
);
if (
newNodes.length >
Math.min(fileNodeCeiling, nodeCeiling - state.nodes.size)
)
throw new RangeError(
"Transferred semantic nodes exceed the offered tree capacity",
);
for (const context of projection.evidenceContexts)
state.evidenceContexts.adopt(context);
for (const node of newNodes) state.nodes.set(node.node_id, node);
for (const relation of projection.relations)
state.relations.set(relation.relation_id, relation);
for (const unknown of projection.unknowns)
state.unknowns.set(unknown.unknown_id, unknown);
for (const root of projection.roots) state.roots.add(root);
fingerprints.push(...projection.fingerprints);
if (projection.truncated) truncatedFiles += 1;
},
recordUnavailableFile: (file, reason, limits, unknownReason) => {
const retainedLimits = javaScriptAnalysisCoverageLimits(limits);
for (const limit of retainedLimits) {
executionLimits.set(
JSON.stringify([limit.name, limit.value, limit.unit]),
limit,
);
}
state.fileNodeBudget = Math.min(
fileNodeCeiling,
Math.max(0, nodeCeiling - state.nodes.size),
);
const module = retainNode(state, file, {
kind: "module",
roleKey: "module",
location: null,
label: file.path,
functionNodeId: null,
});
state.fileNodeBudget = null;
if (module !== null) state.roots.add(module.node_id);
const evidence = {
...observedSemanticEvidence(file, null),
authority: "unknown" as const,
state: "unknown" as const,
confidence: "unknown" as const,
coverage: partialApplicationCoverage(retainedLimits, null),
limitations: [reason],
};
for (const family of JAVASCRIPT_SEMANTIC_RELATION_FAMILIES) {
const unknown = createJavaScriptSemanticGraphUnknown(
{
node_id: module?.node_id ?? null,
family,
relation_kinds: JAVASCRIPT_SEMANTIC_RELATIONS.filter(
(relation) =>
JAVASCRIPT_SEMANTIC_RELATION_FAMILY[relation] === family,
),
reason: unknownReason,
detail: reason,
candidate_node_ids: [],
evidence,
},
state.evidenceContexts,
);
state.unknowns.set(unknown.unknown_id, unknown);
}
unavailableFiles += 1;
if (unknownReason === "resource-limit" && retainedLimits.length > 0)
resourceTruncatedFiles += 1;
},
projectFile: (file, ir, paths) => {
const steps = projectSourceSteps(file, ir, paths);
while (steps.next().done !== true);
},
projectFileSteps: projectSourceSteps,
finish: (rootArtifactSha256, applicationGraph) =>
finishWith(
createJavaScriptSemanticGraph,
rootArtifactSha256,
applicationGraph,
),
finishImmutableSteps: (rootArtifactSha256, applicationGraph) =>
finishWith(
sealTransferredJavaScriptSemanticGraphSteps,
rootArtifactSha256,
applicationGraph,
),
};
};
/** Project one source with the parent's remaining node capacity, retaining no full IR. */
export function* projectJavaScriptSemanticFileSteps(
file: JavaScriptArtifactFile,
ir: JavaScriptSemanticIr,
nodeBudget: number,
paths: readonly ElectronCallPath[] = [],
): Generator<void, JavaScriptSemanticFileProjection> {
if (
!Number.isSafeInteger(nodeBudget) ||
nodeBudget < 0 ||
nodeBudget > SEMANTIC_GRAPH_FILE_NODE_CEILING
nodeBudget > JAVASCRIPT_SEMANTIC_NODE_MAXIMUM
)
throw new RangeError(
"File semantic projection requires a valid offered node capacity",
@@ -414,7 +434,7 @@ export function* projectJavaScriptSemanticFileSteps(
const state = emptyState();
state.fileNodeBudget = nodeBudget;
state.fileNodesDropped = false;
const fingerprints = yield* projectFileSteps(file, ir, state);
const fingerprints = yield* projectFileSteps(file, ir, state, paths);
return {
roots: [...state.roots],
evidenceContexts: state.evidenceContexts.contexts,
@@ -430,6 +450,7 @@ function* projectFileSteps(
file: JavaScriptArtifactFile,
ir: JavaScriptSemanticIr,
state: BuilderState,
paths: readonly ElectronCallPath[] = [],
): Generator<void, JavaScriptSemanticFingerprint[]> {
yield;
const moduleNode = retainNode(state, file, {
@@ -441,6 +462,7 @@ function* projectFileSteps(
});
if (moduleNode === null) return [];
state.roots.add(moduleNode.node_id);
yield* projectElectronSemanticCallPaths(file, paths, state);
// Node creation keeps source order, which the per-file node budget relies
// on; yields between nodes do not change it.
const callableNodes = new Map<string, JavaScriptSemanticGraphNode>();