ata-validator 1.58.0 → 1.59.0

package/README.md

@@ -661,9 +661,11 @@ through a `setFormats()` export the module carries. `docs/API.md` has the
 details.
 
 **Fastify startup, 10 route schemas, from a cold process to the first validated request:
-ajv 18.3 ms, ata 2.67 ms, no build step required.** ata registers in 0.22 ms of that and
-compiles on the first request, so counting only registration would overstate the gap.
-Reproduce with `node benchmark/bench_fastify_boot.mjs`.
+ajv 30.3 ms, ata 9.7 ms, no build step required** (Node 25, ajv 8.20.0, loading each
+library included). ata loads in 3.7 ms, registers the routes in 0.3 ms and compiles on the
+first request, so counting only registration would overstate the gap. The first request
+here comes in the same tick as setup; a server that listens first has the code generator
+loaded by then. Reproduce with `node benchmark/bench_fastify_boot.mjs`.
 
 ### Standard Schema V1
 

package/codegen.js

@@ -4,7 +4,8 @@
 // loads it on demand, once a validator gets hot; importing this once at
 // startup has every schema compiled to generated code from its first call,
 // and makes parse() and the ahead-of-time bundle methods available in the
-// browser. On Node, ata-validator already includes it and this does nothing.
+// browser. On Node, ata-validator loads it when a validator first compiles or
+// once the event loop is free; importing this loads it now.
 const core = require('./lib/validator-core.js');
 if (!core._codegenRegistered()) core._registerCodegen(require('./lib/codegen-engine.js'));
 module.exports = {};

package/index.js

@@ -6,7 +6,19 @@
 // lib/codegen-engine.js, so that the edge entry (index.edge.js) and
 // ata-validator/lite can load the core without it.
 const core = require('./lib/validator-core.js');
-core._registerCodegen(require('./lib/codegen-engine.js'));
+// The code generator loads on first need, or once the event loop is free,
+// not with the package: it is most of what requiring the package parsed. The
+// timer is unref'd, so a script that has finished does not stay to load it.
+// The loader lives here rather than in the core, which the browser build
+// shares and which has no use for it there.
+let engine = null;
+const loadCodegen = () => {
+  if (engine === null) core._registerCodegen(engine = require('./lib/codegen-engine.js'));
+  return engine;
+};
+core._setCodegenLoader(loadCodegen);
+const warm = setTimeout(loadCodegen, 0);
+if (typeof warm === 'object' && typeof warm.unref === 'function') warm.unref();
 
 // The public names. lib/public-api.js holds the same for index.edge.js; they
 // are written out here rather than required so that loading the package in

package/lib/aot-impl.js

@@ -371,6 +371,30 @@ function defaultsDecl(validator) {
     `let _ataFill;\nfunction applyDefaults(d) { (_ataFill || (_ataFill = _ataDefaults(JSON.parse(${text}))))(d); }\n`;
 }
 
+// What a bundled validator runs on the data before it checks it, for the
+// options the caller of bundleStandalone named: the runtime's pass for
+// `coerceTypes`, `removeAdditional` and `useDefaults: true`, as one generated
+// function (preprocessSource), in the runtime's order. A bundle used to leave
+// coerceTypes and removeAdditional out without a word, and then rejected what
+// the runtime accepts. Where the generated pass declines and the runtime
+// would change the data with its closure pass, there is nothing to carry it
+// in, and the bundle refuses. `useDefaults` alone runs applyDefaults.
+// Returns { decl, call }.
+function bundlePreprocess(validator, opts, fillDecl, label) {
+  const coerceTypes = opts && opts.coerceTypes ? opts.coerceTypes : false;
+  const removeAdditional = opts && opts.removeAdditional ? opts.removeAdditional : false;
+  const useDefaults = !!(opts && opts.useDefaults === true);
+  if (!coerceTypes && !removeAdditional) return { decl: '', call: useDefaults && fillDecl ? 'applyDefaults(d);' : '' };
+  const internals = require('./validator-core')._internals;
+  const schema = internals.resolveSchemaForPreprocess(validator._schemaObj, validator._schemaMap);
+  let src = null;
+  try { src = require('./codegen-engine').preprocessSource(schema, { coerceTypes, removeAdditional, useDefaults }); } catch { src = null; }
+  if (src !== null) return { decl: `function _ataPre(d) {\n${src}\n}\n`, call: '_ataPre(d);' };
+  const changes = (coerceTypes && internals.buildCoercer(schema) !== null) || (removeAdditional && internals.buildRemover(schema) !== null);
+  if (changes) throw new Error(`${label}: ${coerceTypes ? 'coerceTypes' : 'removeAdditional'} cannot be compiled into a standalone module for this schema; validate with the runtime Validator`);
+  return { decl: '', call: useDefaults && fillDecl ? 'applyDefaults(d);' : '' };
+}
+
 function toStandaloneModule(validator, opts) {
   assertEmittable(validator, 'toStandaloneModule');
   validator._ensureCompiled();
@@ -912,6 +936,12 @@ function bundle(Validator, schemas, opts) {
 // Zero-dependency self-contained bundle — no require('ata-validator') needed at runtime.
 // opts.format: 'cjs' (default) or 'esm'.
 // opts.formats: { name: fn } — embedded in the output via Function#toString.
+// opts.order: true hands each validator's errors out in the runtime's order,
+// as toStandaloneModule does (lib/aot-order.js).
+// Each validator of a schema with a default carries `applyDefaults(data)`, the
+// runtime's defaults pass, as a property; validate does not run it unless
+// `useDefaults: true` is passed. `coerceTypes` and `removeAdditional` run
+// before the check, as at runtime (bundlePreprocess).
 function bundleStandalone(Validator, schemas, opts) {
   // A custom keyword is a function, and a standalone module imports nothing,
   // so there is no way to carry it. Refuse rather than emit a module that
@@ -930,11 +960,14 @@ function bundleStandalone(Validator, schemas, opts) {
   // Pass one compiles; pass two emits. The helpers that several schemas hoist
   // are the same text declaring the same name, so the bundle keeps one copy at
   // module scope, and pass two needs to know the set before it writes anything.
+  const ordered = !!(opts && opts.order === true);
+  const validatorOpts = ordered ? Object.fromEntries(Object.entries(bundleOpts).filter(([k]) => k !== 'order')) : bundleOpts;
   const compiledEntries = schemas.map((schema) => {
-    const v = new Validator(schema, bundleOpts);
+    const v = new Validator(schema, validatorOpts);
     v._ensureCompiled();
     const jsFn = v._jsFn;
     if (!jsFn || !jsFn._hybridSource) return null;
+    const noOrdinal = !ordered || !require(AOT_ORDER_MODULE).wantsOrdinals(v._schemaObj);
     // The schema the validator compiled, not the one the caller passed. They
     // differ whenever anything prepared the schema: draft-07 normalization,
     // `format` removed under assertFormat: false, keywords outside the
@@ -944,7 +977,7 @@ function bundleStandalone(Validator, schemas, opts) {
       v._schemaObj,
       v._schemaMap,
       v._userFormats,
-      { noOrdinal: true },
+      { noOrdinal },
     );
     if (jsFn._usesSafeRe || (jsErrFn && jsErrFn._usesSafeRe)) bundleUsesSafeRe = true;
     return { v, jsFn, jsErrFn };
@@ -990,12 +1023,25 @@ function bundleStandalone(Validator, schemas, opts) {
     if (preambleSrc) preamble = preamble ? `${preamble}\n${preambleSrc}` : preambleSrc;
     const closureSrc = closureDeclLines(jsFn, 'var').join('\n');
     if (closureSrc) preamble = preamble ? `${preamble}\n${closureSrc}\n` : closureSrc + '\n';
+    // The runtime's passes over the data, for the options asked for, and its
+    // defaults pass as the validator's applyDefaults.
+    const fillDecl = defaultsDecl(v);
+    const pre = bundlePreprocess(v, opts, fillDecl, 'bundleStandalone');
+    preamble = preamble ? `${preamble}\n${fillDecl}${pre.decl}` : fillDecl + pre.decl;
+    const ret = (body) => fillDecl
+      ? `var _f=function(d){${pre.call}${body}};_f.applyDefaults=applyDefaults;return _f`
+      : `return function(d){${pre.call}${body}}`;
+    // In order: the collector's errors sorted as the runtime sorts them.
+    const hasErr = !!(jsErrFn && jsErrFn._errSource);
+    const errDecl = ordered && hasErr
+      ? `${require(AOT_ORDER_MODULE).orderDecl(v._schemaObj)}var E=function(d){var r=(function(d){var _all=true;${errBody}})(d);return r&&r.errors?{valid:false,errors:_ataOrder(r.errors)}:r};`
+      : `var E=function(d){var _all=true;${errBody}};`;
     if (opts && opts.verbose) {
       // Embed the schema and a small resolver so errors carry parentSchema.
       const schemaLit = JSON.stringify(v._schemaObj);
-      return `(function(R){${preamble}var _S=${schemaLit};function _PS(p){if(!p||p[0]!=='#')return undefined;var s=p.slice(1);if(!s)return _S;var ps=s.split('/').filter(Boolean).map(function(x){return x.replace(/~1/g,'/').replace(/~0/g,'~')});var t=_S;for(var i=0;i<ps.length-1;i++){if(t==null||typeof t!=='object')return undefined;t=t[ps[i]]}return t}var E=function(d){var _all=true;${errBody}};var _v=function(d){${jsFn._hybridSource}};return function(d){var r=_v(d);if(r&&r.valid===false&&r.errors){var es=[];for(var i=0;i<r.errors.length;i++){var e=r.errors[i];es.push(Object.assign({},e,{parentSchema:_PS(e.schemaPath)}))}return{valid:false,errors:es}}return r}})(R)`;
+      return `(function(R){${preamble}var _S=${schemaLit};function _PS(p){if(!p||p[0]!=='#')return undefined;var s=p.slice(1);if(!s)return _S;var ps=s.split('/').filter(Boolean).map(function(x){return x.replace(/~1/g,'/').replace(/~0/g,'~')});var t=_S;for(var i=0;i<ps.length-1;i++){if(t==null||typeof t!=='object')return undefined;t=t[ps[i]]}return t}${errDecl}var _v=function(d){${jsFn._hybridSource}};${ret('var r=_v(d);if(r&&r.valid===false&&r.errors){var es=[];for(var i=0;i<r.errors.length;i++){var e=r.errors[i];es.push(Object.assign({},e,{parentSchema:_PS(e.schemaPath)}))}return{valid:false,errors:es}}return r')}})(R)`;
     }
-    return `(function(R){${preamble}var E=function(d){var _all=true;${errBody}};return function(d){${jsFn._hybridSource}}})(R)`;
+    return `(function(R){${preamble}${errDecl}${ret(jsFn._hybridSource)}})(R)`;
   });
   const arr = `[${fns.join(',')}]`;
   // The code-point counter the length checks call. toStandaloneModule embeds

package/lib/validator-core.js

@@ -25,6 +25,9 @@ function _registerCodegen(engine) {
   _codegen = engine;
   ({ compileToJS, compileToJSCodegen, compileToJSCodegenWithErrors, compileToJSCombined } = engine.jsCompiler);
 }
+// The code generator, loading it first where the entry loads it on need
+// (index.js); everywhere else, whatever is registered.
+let _loadCodegen = () => _codegen;
 // The ahead-of-time loaders without the code generator: the edge entry
 // (index.edge.js) registers lib/aot.browser.js here, so modules written by
 // `ata build` or Validator.bundle() load where generated code cannot run.
@@ -219,7 +222,7 @@ function collectRemovals(schema, actions, path) {
 // change within a realm and the probe itself is a code generation attempt.
 let _codegenAvailable = null;
 function codegenAvailable() {
-  if (_codegen === null) return false;
+  if (_loadCodegen() === null) return false;
   if (_codegenAvailable === null) {
     try {
       _codegenAvailable = compileFunction('return 1')() === 1;
@@ -951,7 +954,7 @@ class Validator {
   }
 
   _pos() {
-    return this._posCache || (this._posCache = require('./data-position-cache').createCache(_codegen !== null && _codegen.nativePositions ? _codegen.nativePositions() : null));
+    return this._posCache || (this._posCache = require('./data-position-cache').createCache(_loadCodegen() !== null && _codegen.nativePositions ? _codegen.nativePositions() : null));
   }
 
   _ensureVocabularies() {
@@ -1102,7 +1105,7 @@ class Validator {
     // pass looks for one. One reached through a reference into another
     // document is walked, since its defaults are not in this text.
     const mayHaveDefaults = preprocessSchema !== schemaObj || this._schemaStr.includes('"default"');
-    if (!this._interpretOnly && _codegen !== null) {
+    if (!this._interpretOnly && _loadCodegen() !== null) {
       const cacheable = !this._userFormats && !this._usesKeywords;
       const ppKey = cacheable ? mapKey + '\u0000' + String(options.coerceTypes) + '|' + (options.useDefaults !== false) + '|' + String(options.removeAdditional) : null;
       const hit = ppKey !== null ? _preprocessCache.get(ppKey) : undefined;
@@ -1726,7 +1729,7 @@ class Validator {
       return;
     }
     this._ensureVocabularies();
-    if (this._interpretOnly || _codegen === null || env('ATA_FORCE_NAPI')) return;
+    if (this._interpretOnly || env('ATA_FORCE_NAPI') || _loadCodegen() === null) return;
     _codegen.compileVerdict.call(this);
   }
 
@@ -1736,7 +1739,7 @@ class Validator {
   static fromStandalone(mod, schema, opts) {
     // It loads what Validator.bundleStandalone() writes, and compiles an error
     // function when the module carries none, so it goes with those methods.
-    if (_codegen === null && _aotLoader === null && _tiering === null) throw new TypeError('Validator.fromStandalone is not part of ata-validator/lite, like the bundle methods it loads for. Import it from ata-validator, or import a module from `ata build` directly.');
+    if (_loadCodegen() === null && _aotLoader === null && _tiering === null) throw new TypeError('Validator.fromStandalone is not part of ata-validator/lite, like the bundle methods it loads for. Import it from ata-validator, or import a module from `ata build` directly.');
     const options = opts || {};
     const schemaObj = typeof schema === "string" ? JSON.parse(schema) : schema;
 
@@ -1922,7 +1925,7 @@ function version() {
 // generated module shapes; browser bundles get `lib/aot.browser.js` (a stub
 // that throws) via the package.json `browser` field.
 function _aot() {
-  if (_codegen !== null) return _codegen.aot();
+  if (_loadCodegen() !== null) return _codegen.aot();
   if (_aotLoader !== null) return _aotLoader();
   // The browser build: emitters stay out of the page, as they did when the
   // generator was in it (the package.json browser field maps lib/aot.js to
@@ -2346,7 +2349,7 @@ function _buildParse(self) {
     later._untilCodegen = true;
     return later;
   }
-  if (_codegen === null) return decline('this entry has no code generator to build the copy with', 'Use validate(); a valid result carries the input as it is.');
+  if (_loadCodegen() === null) return decline('this entry has no code generator to build the copy with', 'Use validate(); a valid result carries the input as it is.');
   return _codegen.buildParse(self, decline, extended);
 }
 
@@ -2417,7 +2420,8 @@ for (const [name, pick] of [
 // Cold start: an interpreted twin answers a large schema's first calls where
 // the code generator is registered (see coldTwin in index.js). Lite has none.
 Validator.prototype._coldTwin = function () {
-  return _codegen === null || _codegen.coldTwin === undefined ? null : _codegen.coldTwin(this, Validator);
+  const cg = _loadCodegen();
+  return cg === null || cg.coldTwin === undefined ? null : cg.coldTwin(this, Validator);
 };
 
 _defineLazyMethod('validate', (self) => (data) => {
@@ -2529,9 +2533,10 @@ module.exports = {
 // How an entry registers the code generator. Not enumerable, so the exported
 // surface of `require('ata-validator')` is what it was.
 Object.defineProperty(module.exports, '_registerCodegen', { value: _registerCodegen, enumerable: false });
+Object.defineProperty(module.exports, '_setCodegenLoader', { value: (load) => { _loadCodegen = load; }, enumerable: false });
 Object.defineProperty(module.exports, '_registerTiering', { value: _registerTiering, enumerable: false });
 Object.defineProperty(module.exports, '_codegenRegistered', { value: () => _codegen !== null, enumerable: false });
 Object.defineProperty(module.exports, '_tiering', { get: () => _tiering, enumerable: false });
 Object.defineProperty(module.exports, '_registerAot', { value: _registerAot, enumerable: false });
 // What the compiled paths in index.js need from this module.
-Object.defineProperty(module.exports, '_internals', { value: { _interpArgs, _normalizeCallerSchemaUncached, resolveSchemaForPreprocess, _jsonSyntaxRejection, ABORT_EARLY_RESULT, HYBRID_TIER_CALLS, SIMDJSON_THRESHOLD, VALID_RESULT, _bindEntry, _bindVerdict, _compileCache, _compileCacheSet, _mustReject, _rememberInstance, _verboseWrap, compileCacheKey, getNative, isV1Dialect, resolveSchemaByPath }, enumerable: false });
+Object.defineProperty(module.exports, '_internals', { value: { _interpArgs, _normalizeCallerSchemaUncached, resolveSchemaForPreprocess, buildCoercer, buildRemover, _jsonSyntaxRejection, ABORT_EARLY_RESULT, HYBRID_TIER_CALLS, SIMDJSON_THRESHOLD, VALID_RESULT, _bindEntry, _bindVerdict, _compileCache, _compileCacheSet, _mustReject, _rememberInstance, _verboseWrap, compileCacheKey, getNative, isV1Dialect, resolveSchemaByPath }, enumerable: false });

package/lib/version.js

@@ -7,4 +7,4 @@
 //
 // Kept in lockstep with package.json by `tests/test_version_sync.js`.
 
-module.exports = '1.58.0';
+module.exports = '1.59.0';

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "ata-validator",
-  "version": "1.58.0",
+  "version": "1.59.0",
   "description": "JSON Schema validator that compiles to plain JavaScript, still runs where code generation is blocked, and can be compiled away at build time. Passes the full official test suite.",
   "type": "commonjs",
   "main": "index.js",
@@ -186,13 +186,13 @@
     "codegen.d.ts"
   ],
   "optionalDependencies": {
-    "@ata-validator/native-darwin-arm64": "1.58.0",
-    "@ata-validator/native-darwin-x64": "1.58.0",
-    "@ata-validator/native-linux-arm64-gnu": "1.58.0",
-    "@ata-validator/native-linux-arm64-musl": "1.58.0",
-    "@ata-validator/native-linux-x64-gnu": "1.58.0",
-    "@ata-validator/native-linux-x64-musl": "1.58.0",
-    "@ata-validator/native-win32-x64": "1.58.0"
+    "@ata-validator/native-darwin-arm64": "1.59.0",
+    "@ata-validator/native-darwin-x64": "1.59.0",
+    "@ata-validator/native-linux-arm64-gnu": "1.59.0",
+    "@ata-validator/native-linux-arm64-musl": "1.59.0",
+    "@ata-validator/native-linux-x64-gnu": "1.59.0",
+    "@ata-validator/native-linux-x64-musl": "1.59.0",
+    "@ata-validator/native-win32-x64": "1.59.0"
   },
   "peerDependencies": {
     "yaml": "^2.0.0"