fix: validate all schemas in actor.json (#1071)

example output

```
Info: Validating input schema at .actor/input_schema.json
Success: Input schema is valid.
Info: Validating Dataset schema at .actor/dataset_schema.json
Success: Dataset schema is valid.
Info: Validating Output schema at actor/output_schema.json
Success: Output schema is valid.
Info: Validating Key-Value Store schema at .actor/key_value_store_schema.json
Success: Key-Value Store schema is valid.
```
closes #1036

---------

Co-authored-by: Vlad Frangu <me@vladfrangu.dev>

Author: l2ysho

package.json

@@ -69,6 +69,7 @@
         "@apify/actor-templates": "^0.1.5",
         "@apify/consts": "^2.36.0",
         "@apify/input_schema": "^3.17.0",
+        "@apify/json_schemas": "^0.13.0",
         "@apify/utilities": "^2.18.0",
         "@crawlee/memory-storage": "^3.12.0",
         "@inquirer/core": "^11.0.0",

src/commands/_register.ts

@@ -33,7 +33,7 @@ import { RunsIndexCommand } from './runs/_index.js';
 import { SecretsIndexCommand } from './secrets/_index.js';
 import { TasksIndexCommand } from './task/_index.js';
 import { TelemetryIndexCommand } from './telemetry/_index.js';
-import { ValidateInputSchemaCommand } from './validate-schema.js';
+import { ValidateSchemaCommand } from './validate-schema.js';
 
 export const apifyCommands = [
 	// namespaces
@@ -64,7 +64,7 @@ export const apifyCommands = [
 	TopLevelPullCommand,
 	ToplevelPushCommand,
 	RunCommand,
-	ValidateInputSchemaCommand,
+	ValidateSchemaCommand,
 	HelpCommand,
 
 	// test commands

src/commands/validate-schema.ts

@@ -1,21 +1,33 @@
 import process from 'node:process';
 
+import { validateInputSchema } from '@apify/input_schema';
+
 import { ApifyCommand } from '../lib/command-framework/apify-command.js';
 import { Args } from '../lib/command-framework/args.js';
-import { LOCAL_CONFIG_PATH } from '../lib/consts.js';
-import { readAndValidateInputSchema } from '../lib/input_schema.js';
-import { success } from '../lib/outputs.js';
+import { CommandExitCodes, LOCAL_CONFIG_PATH } from '../lib/consts.js';
+import {
+	readAndValidateInputSchema,
+	readInputSchema,
+	readStorageSchema,
+	validateDatasetSchema,
+	validateKvsSchema,
+	validateOutputSchema,
+} from '../lib/input_schema.js';
+import { error, info, success } from '../lib/outputs.js';
+import { Ajv2019 } from '../lib/utils.js';
 
-export class ValidateInputSchemaCommand extends ApifyCommand<typeof ValidateInputSchemaCommand> {
+export class ValidateSchemaCommand extends ApifyCommand<typeof ValidateSchemaCommand> {
 	static override name = 'validate-schema' as const;
 
-	static override description = `Validates Actor input schema from one of these locations (in priority order):
-  1. Object in '${LOCAL_CONFIG_PATH}' under "input" key
-  2. JSON file path in '${LOCAL_CONFIG_PATH}' "input" key
-  3. .actor/INPUT_SCHEMA.json
-  4. INPUT_SCHEMA.json
+	static override description = `Validates Actor schemas.
+
+When a path argument is provided, validates only the input schema at that path.
 
-Optionally specify custom schema path to validate.`;
+When no path is provided, validates all schemas found in '${LOCAL_CONFIG_PATH}':
+  - Input schema (from "input" key or default locations)
+  - Dataset schema (from "storages.dataset")
+  - Output schema (from "output")
+  - Key-Value Store schema (from "storages.keyValueStore")`;
 
 	static override group = 'Local Actor Development';
 
@@ -35,22 +47,115 @@ Optionally specify custom schema path to validate.`;
 	static override args = {
 		path: Args.string({
 			required: false,
-			description: 'Optional path to your INPUT_SCHEMA.json file. If not provided ./INPUT_SCHEMA.json is used.',
+			description: `Optional path to your INPUT_SCHEMA.json file. If not provided, validates all schemas in '${LOCAL_CONFIG_PATH}'.`,
 		}),
 	};
 
 	static override hiddenAliases = ['vis'];
 
 	async run() {
+		if (this.args.path) {
+			await this.validateInputSchemaAtPath(this.args.path);
+			return;
+		}
+
+		await this.validateAllSchemas();
+	}
+
+	private async validateInputSchemaAtPath(forcePath: string) {
 		await readAndValidateInputSchema({
-			forcePath: this.args.path,
+			forcePath,
 			cwd: process.cwd(),
-			getMessage: (path) =>
-				path
-					? `Validating input schema at ${path}`
-					: `Validating input schema embedded in '${LOCAL_CONFIG_PATH}'`,
+			getMessage: (path) => `Validating input schema at ${path ?? forcePath}`,
 		});
 
 		success({ message: 'Input schema is valid.' });
 	}
+
+	private async validateAllSchemas() {
+		const cwd = process.cwd();
+		let foundAny = false;
+		let hasErrors = false;
+
+		// Input schema — not using readAndValidateInputSchema here because it throws
+		// when no schema is found; in the all-schemas scan, a missing input schema
+		// should be silently skipped, not treated as an error.
+		try {
+			const { inputSchema, inputSchemaPath } = await readInputSchema({ cwd, throwOnMissing: true });
+
+			if (inputSchema) {
+				foundAny = true;
+
+				const location = inputSchemaPath ? `at ${inputSchemaPath}` : `embedded in '${LOCAL_CONFIG_PATH}'`;
+				info({ message: `Validating input schema ${location}` });
+
+				const validator = new Ajv2019({ strict: false });
+				validateInputSchema(validator, inputSchema);
+				success({ message: 'Input schema is valid.' });
+			}
+		} catch (err) {
+			foundAny = true;
+			hasErrors = true;
+			error({ message: (err as Error).message });
+		}
+
+		// Storage schemas (Dataset, Output, Key-Value Store)
+		const storageSchemas = [
+			{
+				label: 'Dataset',
+				read: () => readStorageSchema({ cwd, key: 'dataset', label: 'Dataset', throwOnMissing: true }),
+				validate: validateDatasetSchema,
+			},
+			{
+				label: 'Output',
+				read: () =>
+					readStorageSchema({
+						cwd,
+						key: 'output',
+						label: 'Output',
+						getRef: (config) => config?.output,
+						throwOnMissing: true,
+					}),
+				validate: validateOutputSchema,
+			},
+			{
+				label: 'Key-Value Store',
+				read: () =>
+					readStorageSchema({ cwd, key: 'keyValueStore', label: 'Key-Value Store', throwOnMissing: true }),
+				validate: validateKvsSchema,
+			},
+		];
+
+		for (const { label, read, validate } of storageSchemas) {
+			try {
+				const result = read();
+
+				if (result) {
+					foundAny = true;
+
+					const location = result.schemaPath
+						? `at ${result.schemaPath}`
+						: `embedded in '${LOCAL_CONFIG_PATH}'`;
+					info({ message: `Validating ${label} schema ${location}` });
+
+					validate(result.schema);
+					success({ message: `${label} schema is valid.` });
+				}
+			} catch (err) {
+				foundAny = true;
+				hasErrors = true;
+				error({ message: (err as Error).message });
+			}
+		}
+
+		if (!foundAny) {
+			throw new Error(
+				`No schemas found. Make sure '${LOCAL_CONFIG_PATH}' exists and defines at least one schema.`,
+			);
+		}
+
+		if (hasErrors) {
+			process.exitCode = CommandExitCodes.InvalidInput;
+		}
+	}
 }

src/lib/input_schema.ts

@@ -1,10 +1,16 @@
 import { existsSync, writeFileSync } from 'node:fs';
 import { join } from 'node:path';
 
+import type { Ajv, ErrorObject } from 'ajv';
 import { cloneDeep } from 'es-toolkit';
 
 import { KEY_VALUE_STORE_KEYS } from '@apify/consts';
 import { validateInputSchema } from '@apify/input_schema';
+import {
+	getDatasetSchemaValidator,
+	getKeyValueStoreSchemaValidator,
+	getOutputSchemaValidator,
+} from '@apify/json_schemas';
 
 import { ACTOR_SPECIFICATION_FOLDER, LOCAL_CONFIG_PATH } from './consts.js';
 import { info, warning } from './outputs.js';
@@ -24,7 +30,15 @@ const DEFAULT_INPUT_SCHEMA_PATHS = [
  * In such a case, path would be set to the location
  * where the input schema would be expected to be found (and e.g. can be created there).
  */
-export const readInputSchema = async ({ forcePath, cwd }: { forcePath?: string; cwd: string }) => {
+export const readInputSchema = async ({
+	forcePath,
+	cwd,
+	throwOnMissing = false,
+}: {
+	forcePath?: string;
+	cwd: string;
+	throwOnMissing?: boolean;
+}) => {
 	if (forcePath) {
 		return {
 			inputSchema: getJsonFileContent(forcePath),
@@ -34,7 +48,7 @@ export const readInputSchema = async ({ forcePath, cwd }: { forcePath?: string;
 
 	const localConfig = getLocalConfig(cwd);
 
-	if (typeof localConfig?.input === 'object') {
+	if (typeof localConfig?.input === 'object' && localConfig.input !== null) {
 		return {
 			inputSchema: localConfig.input as Record<string, unknown>,
 			inputSchemaPath: null,
@@ -43,8 +57,25 @@ export const readInputSchema = async ({ forcePath, cwd }: { forcePath?: string;
 
 	if (typeof localConfig?.input === 'string') {
 		const fullPath = join(cwd, ACTOR_SPECIFICATION_FOLDER, localConfig.input);
+		const schema = getJsonFileContent(fullPath);
+
+		if (!schema) {
+			if (throwOnMissing) {
+				throw new Error(`Input schema file not found at ${fullPath} (referenced in '${LOCAL_CONFIG_PATH}').`);
+			}
+
+			warning({
+				message: `Input schema file not found at ${fullPath} (referenced in '${LOCAL_CONFIG_PATH}').`,
+			});
+
+			return {
+				inputSchema: null,
+				inputSchemaPath: fullPath,
+			};
+		}
+
 		return {
-			inputSchema: getJsonFileContent(fullPath),
+			inputSchema: schema,
 			inputSchemaPath: fullPath,
 		};
 	}
@@ -115,11 +146,13 @@ export const readStorageSchema = ({
 	key,
 	label,
 	getRef,
+	throwOnMissing = false,
 }: {
 	cwd: string;
 	key: string;
 	label: string;
 	getRef?: (config: ReturnType<typeof getLocalConfig>) => unknown;
+	throwOnMissing?: boolean;
 }): { schema: Record<string, unknown>; schemaPath: string | null } | null => {
 	const localConfig = getLocalConfig(cwd);
 
@@ -137,6 +170,12 @@ export const readStorageSchema = ({
 		const schema = getJsonFileContent(fullPath);
 
 		if (!schema) {
+			if (throwOnMissing) {
+				throw new Error(
+					`${label} schema file not found at ${fullPath} (referenced in '${LOCAL_CONFIG_PATH}').`,
+				);
+			}
+
 			warning({
 				message: `${label} schema file not found at ${fullPath} (referenced in '${LOCAL_CONFIG_PATH}').`,
 			});
@@ -255,8 +294,40 @@ export const getDefaultsFromInputSchema = (inputSchema: any) => {
 	return defaults;
 };
 
+function formatSchemaValidationErrors(errors: ErrorObject[], schemaName: string): string {
+	const details = errors
+		.map((err) => {
+			const path = err.instancePath ? ` at ${err.instancePath}` : '';
+			return `  - ${err.message}${path}`;
+		})
+		.join('\n');
+
+	return `${schemaName} schema is not valid:\n${details}`;
+}
+
+export function validateDatasetSchema(schema: Record<string, unknown>): void {
+	const validate = getDatasetSchemaValidator();
+	if (!validate(schema)) {
+		throw new Error(formatSchemaValidationErrors(validate.errors!, 'Dataset'));
+	}
+}
+
+export function validateOutputSchema(schema: Record<string, unknown>): void {
+	const validate = getOutputSchemaValidator();
+	if (!validate(schema)) {
+		throw new Error(formatSchemaValidationErrors(validate.errors!, 'Output'));
+	}
+}
+
+export function validateKvsSchema(schema: Record<string, unknown>): void {
+	const validate = getKeyValueStoreSchemaValidator();
+	if (!validate(schema)) {
+		throw new Error(formatSchemaValidationErrors(validate.errors!, 'Key-Value Store'));
+	}
+}
+
 // Lots of code copied from @apify-packages/actor, this really should be moved to the shared input_schema package
-export const getAjvValidator = (inputSchema: any, ajvInstance: import('ajv').Ajv) => {
+export const getAjvValidator = (inputSchema: any, ajvInstance: Ajv) => {
 	const copyOfSchema = cloneDeep(inputSchema);
 	copyOfSchema.required = [];