fix(formats): prevent CLI execution when bundled

Mearman · Mearman · commit eda2e5c70d53 · 2026-01-21T14:12:39.000Z
Add filename check to CLI guard condition in format handlers to prevent
execution when code is bundled (e.g., via Vite). The previous check only
verified if the module URL matched the process argument, which could
match incorrectly in bundled contexts.

build(vite): simplify external module pattern

Replace array-based externals with function pattern for node: prefix
matching. This is more concise and handles all node: modules uniformly.

refactor(framework): add sync case registry variant

Add registerCasesSync() function for backward compatibility with code
that doesn't require async seed pre-loading. The async registerCases()
now pre-loads benchmark data to determine seed nodes upfront.

Update entry point to use registerCasesSync for consistency.
diff --git a/src/experiments/framework/entry/run-framework-experiments.ts b/src/experiments/framework/entry/run-framework-experiments.ts
@@ -19,7 +19,7 @@ import { fileURLToPath } from "node:url";
 
 import { getCoreClaims } from "../claims/registry.js";
 import { CaseRegistry } from "../registry/case-registry.js";
-import { registerCases } from "../registry/register-cases.js";
+import { registerCasesSync } from "../registry/register-cases.js";
 import { registerExpansionSuts } from "../registry/register-suts.js";
 import { SUTRegistry } from "../registry/sut-registry.js";
 import { TABLE_SPECS } from "../renderers/table-specs.js";
@@ -50,7 +50,7 @@ const main = async (): Promise<void> => {
 	// Initialize registries
 	console.log("1. Initializing registries...");
 	const sutRegistry = registerExpansionSuts(new SUTRegistry());
-	const caseRegistry = registerCases(new CaseRegistry());
+	const caseRegistry = registerCasesSync(new CaseRegistry());
 
 	console.log(`   - ${sutRegistry.size} SUTs registered`);
 	console.log(`   - ${caseRegistry.size} cases registered`);
diff --git a/src/experiments/framework/registry/register-cases.ts b/src/experiments/framework/registry/register-cases.ts
@@ -52,44 +52,65 @@ export const BENCHMARK_CASES: Array<{
 
 /**
  * Create case definitions for benchmark graphs.
+ * Pre-loads benchmark data to determine seed nodes.
  */
-const createBenchmarkCaseDefinitions = (): CaseDefinition<BenchmarkGraphExpander>[] => BENCHMARK_CASES.map((benchmark) => {
-	const inputs: CaseInputs = {
-		summary: {
-			datasetId: benchmark.id,
-			expectedNodes: benchmark.expectedNodes,
-		},
-		artefacts: [
-			{
-				type: "graph",
-				uri: `benchmark://${benchmark.id}`,
+const createBenchmarkCaseDefinitions = async (): Promise<CaseDefinition<BenchmarkGraphExpander>[]> => {
+	const cases: CaseDefinition<BenchmarkGraphExpander>[] = [];
+
+	for (const benchmark of BENCHMARK_CASES) {
+		// Load benchmark data to get node IDs
+		const benchmarkData = await loadBenchmarkByIdFromUrl(benchmark.id);
+		const nodes = benchmarkData.graph.getAllNodes();
+
+		// Determine seed nodes (first and last)
+		let seeds: string[];
+		if (nodes.length >= 2) {
+			const lastNode = nodes.at(-1);
+			seeds = lastNode ? [nodes[0].id, lastNode.id] : [nodes[0].id];
+		} else if (nodes.length === 1) {
+			seeds = [nodes[0].id];
+		} else {
+			seeds = [];
+		}
+
+		const inputs: CaseInputs = {
+			summary: {
+				datasetId: benchmark.id,
+				expectedNodes: benchmark.expectedNodes,
+				seeds, // Store seeds in inputs
 			},
-		],
-	};
+			artefacts: [
+				{
+					type: "graph",
+					uri: `benchmark://${benchmark.id}`,
+				},
+			],
+		};
 
-	const caseSpec: EvaluationCase = {
-		caseId: generateCaseId(benchmark.name, inputs),
-		name: benchmark.name,
-		caseClass: benchmark.caseClass,
-		inputs,
-		version: "1.0.0",
-		tags: benchmark.tags,
-	};
+		const caseSpec: EvaluationCase = {
+			caseId: generateCaseId(benchmark.name, inputs),
+			name: benchmark.name,
+			caseClass: benchmark.caseClass,
+			inputs,
+			version: "1.0.0",
+			tags: benchmark.tags,
+		};
 
-	return {
-		case: caseSpec,
-		createExpander: async (inputsForExpander: CaseInputs) => {
-			const datasetId = inputsForExpander.summary?.datasetId as string;
-			const benchmarkData = await loadBenchmarkByIdFromUrl(datasetId);
-			return new BenchmarkGraphExpander(benchmarkData.graph, benchmarkData.meta.directed);
-		},
-		getSeeds: (_inputsForSeeds: CaseInputs) => {
-			// Default: return first and last nodes (will be populated after loading)
-			// In practice, seeds are determined after the expander is created
-			return [];
-		},
-	};
-});
+		cases.push({
+			case: caseSpec,
+			createExpander: async (_inputsForExpander: CaseInputs) => {
+				// Use the pre-loaded data
+				return new BenchmarkGraphExpander(benchmarkData.graph, benchmarkData.meta.directed);
+			},
+			getSeeds: (_inputsForSeeds: CaseInputs) => {
+				// Return the pre-computed seeds
+				return seeds;
+			},
+		});
+	}
+
+	return cases;
+};
 
 /**
  * Synthetic graph case class types.
@@ -132,23 +153,76 @@ export const createSyntheticCaseDefinition = (type: SyntheticGraphType, nodes: n
 };
 
 /**
- * Register all cases with a registry.
+ * Register all cases with a registry (sync version, doesn't pre-load seeds).
+ * Use this for backward compatibility when seeds aren't needed upfront.
+ *
+ * @param registry - Registry to populate (defaults to new instance)
+ * @returns The populated registry
+ */
+export const registerCasesSync = (registry: GraphCaseRegistry = new CaseRegistry()): GraphCaseRegistry => {
+	// Register benchmark cases (without seed pre-loading)
+	const cases: CaseDefinition<BenchmarkGraphExpander>[] = BENCHMARK_CASES.map((benchmark) => {
+		const inputs: CaseInputs = {
+			summary: {
+				datasetId: benchmark.id,
+				expectedNodes: benchmark.expectedNodes,
+			},
+			artefacts: [
+				{
+					type: "graph",
+					uri: `benchmark://${benchmark.id}`,
+				},
+			],
+		};
+
+		const caseSpec: EvaluationCase = {
+			caseId: generateCaseId(benchmark.name, inputs),
+			name: benchmark.name,
+			caseClass: benchmark.caseClass,
+			inputs,
+			version: "1.0.0",
+			tags: benchmark.tags,
+		};
+
+		return {
+			case: caseSpec,
+			createExpander: async (inputsForExpander: CaseInputs) => {
+				const datasetId = inputsForExpander.summary?.datasetId as string;
+				const benchmarkData = await loadBenchmarkByIdFromUrl(datasetId);
+				return new BenchmarkGraphExpander(benchmarkData.graph, benchmarkData.meta.directed);
+			},
+			getSeeds: (_inputsForSeeds: CaseInputs) => {
+				// Seeds will be determined from the expander after creation
+				// The SUT factory should handle empty seeds
+				return [];
+			},
+		};
+	});
+
+	registry.registerAll(cases);
+	return registry;
+};
+
+/**
+ * Register all cases with a registry (async version, pre-loads seeds).
+ * This version loads benchmark data during registration to determine seed nodes.
  *
  * @param registry - Registry to populate (defaults to new instance)
  * @returns The populated registry
  */
-export const registerCases = (registry: GraphCaseRegistry = new CaseRegistry()): GraphCaseRegistry => {
-	// Register benchmark cases
-	const benchmarkCases = createBenchmarkCaseDefinitions();
+export const registerCases = async (registry: GraphCaseRegistry = new CaseRegistry()): Promise<GraphCaseRegistry> => {
+	// Register benchmark cases (async now, pre-loads seeds)
+	const benchmarkCases = await createBenchmarkCaseDefinitions();
 	registry.registerAll(benchmarkCases);
 
 	return registry;
 };
 
 /**
- * Global case registry with all cases registered.
+ * Global case registry with all cases registered (sync version).
+ * For async version with pre-loaded seeds, use `await registerCases()` instead.
  */
-export const graphCaseRegistry = registerCases(new CaseRegistry());
+export const graphCaseRegistry = registerCasesSync(new CaseRegistry());
 
 /**
  * Get seeds for a benchmark graph.
diff --git a/src/formats/gml/fetch.ts b/src/formats/gml/fetch.ts
@@ -191,8 +191,8 @@ const main = async (): Promise<void> => {
 	}
 };
 
-// Run CLI if executed directly
-if (import.meta.url === `file://${process.argv[1]}`) {
+// Run CLI if executed directly (not when bundled)
+if (import.meta.url === `file://${process.argv[1]}` && process.argv[1].endsWith("fetch.ts")) {
 	main().catch((error: unknown) => {
 		console.error("Error:", error);
 		process.exit(1);
diff --git a/src/formats/gml/parse.ts b/src/formats/gml/parse.ts
@@ -405,8 +405,8 @@ const main = async (): Promise<void> => {
 	}
 };
 
-// Run CLI if executed directly
-if (import.meta.url === `file://${process.argv[1]}`) {
+// Run CLI if executed directly (not when bundled)
+if (import.meta.url === `file://${process.argv[1]}` && process.argv[1].endsWith("parse.ts")) {
 	main().catch((error: unknown) => {
 		console.error("Error:", error);
 		process.exit(1);
diff --git a/src/formats/gml/serialize.ts b/src/formats/gml/serialize.ts
@@ -181,8 +181,8 @@ const main = async (): Promise<void> => {
 	}
 };
 
-// Run CLI if executed directly
-if (import.meta.url === `file://${process.argv[1]}`) {
+// Run CLI if executed directly (not when bundled)
+if (import.meta.url === `file://${process.argv[1]}` && process.argv[1].endsWith("serialize.ts")) {
 	main().catch((error: unknown) => {
 		console.error("Error:", error);
 		process.exit(1);
diff --git a/src/formats/pajek/fetch.ts b/src/formats/pajek/fetch.ts
@@ -174,8 +174,8 @@ const main = async (): Promise<void> => {
 	}
 };
 
-// Run CLI if executed directly
-if (import.meta.url === `file://${process.argv[1]}`) {
+// Run CLI if executed directly (not when bundled)
+if (import.meta.url === `file://${process.argv[1]}` && process.argv[1].endsWith("fetch.ts")) {
 	main().catch((error: unknown) => {
 		console.error("Error:", error);
 		process.exit(1);
diff --git a/src/formats/snap/fetch.ts b/src/formats/snap/fetch.ts
@@ -140,8 +140,8 @@ const main = async (): Promise<void> => {
 	}
 };
 
-// Run CLI if executed directly
-if (import.meta.url === `file://${process.argv[1]}`) {
+// Run CLI if executed directly (not when bundled)
+if (import.meta.url === `file://${process.argv[1]}` && process.argv[1].endsWith("fetch.ts")) {
 	main().catch((error: unknown) => {
 		console.error("Error:", error);
 		process.exit(1);
diff --git a/src/formats/snap/parse.ts b/src/formats/snap/parse.ts
@@ -192,8 +192,8 @@ const main = async (): Promise<void> => {
 	console.error(`Nodes: ${json.nodes.length}, Edges: ${json.edges.length}, Directed: ${json.meta.directed}`);
 };
 
-// Run CLI if executed directly
-if (import.meta.url === `file://${process.argv[1]}`) {
+// Run CLI if executed directly (not when bundled)
+if (import.meta.url === `file://${process.argv[1]}` && process.argv[1].endsWith("parse.ts")) {
 	main().catch((error: unknown) => {
 		console.error("Error:", error);
 		process.exit(1);
diff --git a/src/formats/ucinet/fetch.ts b/src/formats/ucinet/fetch.ts
@@ -174,8 +174,8 @@ const main = async (): Promise<void> => {
 	}
 };
 
-// Run CLI if executed directly
-if (import.meta.url === `file://${process.argv[1]}`) {
+// Run CLI if executed directly (not when bundled)
+if (import.meta.url === `file://${process.argv[1]}` && process.argv[1].endsWith("fetch.ts")) {
 	main().catch((error: unknown) => {
 		console.error("Error:", error);
 		process.exit(1);
diff --git a/vite.config.ts b/vite.config.ts
@@ -34,12 +34,7 @@ export default defineConfig({
 				target: "node18",
 				modulePreload: false,
 				rollupOptions: {
-					external: [
-						"node:fs",
-						"node:path",
-						"node:process",
-						"node:module",
-					],
+					external: (id: string) => id.startsWith("node:"),
 				},
 			}
 		: {
