refactor: move selector parsing to Node (#12543)

packages/puppeteer-core/Herebyfile.mjs

@@ -48,9 +48,10 @@ export const generateInjectedTask = task({
       entryPoints: ['src/injected/injected.ts'],
       bundle: true,
       format: 'cjs',
-      target: ['chrome117', 'firefox118'],
+      target: ['chrome125', 'firefox125'],
       minify: true,
       write: false,
+      legalComments: 'none',
     });
     const template = await readFile('src/templates/injected.ts.tmpl', 'utf8');
     await mkdir('src/generated', {recursive: true});
@@ -136,6 +137,16 @@ export const buildTask = task({
               'utf-8'
             );
             break;
+          case 'parsel-js':
+            license = await readFile(
+              path.join(
+                path.dirname(require.resolve('parsel-js')),
+                '..',
+                'LICENSE'
+              ),
+              'utf-8'
+            );
+            break;
           default:
             throw new Error(`Add license handling for ${path}`);
         }

packages/puppeteer-core/src/common/CSSQueryHandler.ts

@@ -0,0 +1,29 @@
+/**
+ * @license
+ * Copyright 2023 Google Inc.
+ * SPDX-License-Identifier: Apache-2.0
+ */
+
+import type PuppeteerUtil from '../injected/injected.js';
+
+import {QueryHandler} from './QueryHandler.js';
+
+/**
+ * @internal
+ */
+export class CSSQueryHandler extends QueryHandler {
+  static override querySelector = (
+    element: Node,
+    selector: string,
+    {cssQuerySelector}: PuppeteerUtil
+  ): Node | null => {
+    return cssQuerySelector(element, selector);
+  };
+  static override querySelectorAll = (
+    element: Node,
+    selector: string,
+    {cssQuerySelectorAll}: PuppeteerUtil
+  ): Iterable<Node> => {
+    return cssQuerySelectorAll(element, selector);
+  };
+}

packages/puppeteer-core/src/common/GetQueryHandler.ts

@@ -6,9 +6,11 @@
 
 import {ARIAQueryHandler} from '../cdp/AriaQueryHandler.js';
 
+import {CSSQueryHandler} from './CSSQueryHandler.js';
 import {customQueryHandlers} from './CustomQueryHandler.js';
 import {PierceQueryHandler} from './PierceQueryHandler.js';
 import {PQueryHandler} from './PQueryHandler.js';
+import {parsePSelectors} from './PSelectorParser.js';
 import type {QueryHandler} from './QueryHandler.js';
 import {TextQueryHandler} from './TextQueryHandler.js';
 import {XPathQueryHandler} from './XPathQueryHandler.js';
@@ -45,5 +47,15 @@ export function getQueryHandlerAndSelector(selector: string): {
       }
     }
   }
-  return {updatedSelector: selector, QueryHandler: PQueryHandler};
+  const [pSelector, isPureCSS] = parsePSelectors(selector);
+  if (isPureCSS) {
+    return {
+      updatedSelector: selector,
+      QueryHandler: CSSQueryHandler,
+    };
+  }
+  return {
+    updatedSelector: JSON.stringify(pSelector),
+    QueryHandler: PQueryHandler,
+  };
 }

packages/puppeteer-core/src/common/PSelectorParser.ts

@@ -4,20 +4,18 @@
  * SPDX-License-Identifier: Apache-2.0
  */
 
-import {type Token, tokenize, TOKENS, stringify} from 'parsel-js';
+import {
-
+  type Token,
-export type CSSSelector = string;
+  tokenize,
-export interface PPseudoSelector {
+  TOKENS,
-  name: string;
+  stringify,
-  value: string;
+} from '../../third_party/parsel-js/parsel-js.js';
-}
+import type {
-export const enum PCombinator {
+  ComplexPSelector,
-  Descendent = '>>>',
+  ComplexPSelectorList,
-  Child = '>>>>',
+  CompoundPSelector,
-}
+} from '../injected/PQuerySelector.js';
-export type CompoundPSelector = Array<CSSSelector | PPseudoSelector>;
+import {PCombinator} from '../injected/PQuerySelector.js';
-export type ComplexPSelector = Array<CompoundPSelector | PCombinator>;
-export type ComplexPSelectorList = ComplexPSelector[];
 
 TOKENS['combinator'] = /\s*(>>>>?|[\s>+~])\s*/g;
 
@@ -34,6 +32,9 @@ const unquote = (text: string): string => {
   });
 };
 
+/**
+ * @internal
+ */
 export function parsePSelectors(
   selector: string
 ): [selector: ComplexPSelectorList, isPureCSS: boolean] {

packages/puppeteer-core/src/common/common.ts

@@ -25,6 +25,7 @@ export * from './PDFOptions.js';
 export * from './PierceQueryHandler.js';
 export * from './PQueryHandler.js';
 export * from './Product.js';
+export * from './PSelectorParser.js';
 export * from './Puppeteer.js';
 export * from './QueryHandler.js';
 export * from './ScriptInjector.js';

packages/puppeteer-core/src/injected/CSSSelector.ts

@@ -0,0 +1,20 @@
+/**
+ * @license
+ * Copyright 2024 Google Inc.
+ * SPDX-License-Identifier: Apache-2.0
+ */
+
+export const cssQuerySelector = (
+  root: Node,
+  selector: string
+): Element | null => {
+  // @ts-expect-error assume element root
+  return root.querySelector(selector);
+};
+export const cssQuerySelectorAll = function (
+  root: Node,
+  selector: string
+): Iterable<Element> {
+  // @ts-expect-error assume element root
+  return root.querySelectorAll(selector);
+};

packages/puppeteer-core/src/injected/PQuerySelector.ts

@@ -9,21 +9,43 @@ import {AsyncIterableUtil} from '../util/AsyncIterableUtil.js';
 
 import {ariaQuerySelectorAll} from './ARIAQuerySelector.js';
 import {customQuerySelectors} from './CustomQuerySelector.js';
-import {
-  type ComplexPSelector,
-  type ComplexPSelectorList,
-  type CompoundPSelector,
-  type CSSSelector,
-  parsePSelectors,
-  PCombinator,
-  type PPseudoSelector,
-} from './PSelectorParser.js';
 import {textQuerySelectorAll} from './TextQuerySelector.js';
 import {pierce, pierceAll} from './util.js';
 import {xpathQuerySelectorAll} from './XPathQuerySelector.js';
 
 const IDENT_TOKEN_START = /[-\w\P{ASCII}*]/;
 
+/**
+ * @internal
+ */
+export type CSSSelector = string;
+/**
+ * @internal
+ */
+export interface PPseudoSelector {
+  name: string;
+  value: string;
+}
+/**
+ * @internal
+ */
+export const enum PCombinator {
+  Descendent = '>>>',
+  Child = '>>>>',
+}
+/**
+ * @internal
+ */
+export type CompoundPSelector = Array<CSSSelector | PPseudoSelector>;
+/**
+ * @internal
+ */
+export type ComplexPSelector = Array<CompoundPSelector | PCombinator>;
+/**
+ * @internal
+ */
+export type ComplexPSelectorList = ComplexPSelector[];
+
 interface QueryableNode extends Node {
   querySelectorAll: typeof Document.prototype.querySelectorAll;
 }
@@ -32,24 +54,15 @@ const isQueryableNode = (node: Node): node is QueryableNode => {
   return 'querySelectorAll' in node;
 };
 
-class SelectorError extends Error {
-  constructor(selector: string, message: string) {
-    super(`${selector} is not a valid selector: ${message}`);
-  }
-}
-
 class PQueryEngine {
-  #input: string;
-
   #complexSelector: ComplexPSelector;
   #compoundSelector: CompoundPSelector = [];
   #selector: CSSSelector | PPseudoSelector | undefined = undefined;
 
   elements: AwaitableIterable<Node>;
 
-  constructor(element: Node, input: string, complexSelector: ComplexPSelector) {
+  constructor(element: Node, complexSelector: ComplexPSelector) {
     this.elements = [element];
-    this.#input = input;
     this.#complexSelector = complexSelector;
     this.#next();
   }
@@ -71,7 +84,6 @@ class PQueryEngine {
 
     for (; this.#selector !== undefined; this.#next()) {
       const selector = this.#selector;
-      const input = this.#input;
       if (typeof selector === 'string') {
         // The regular expression tests if the selector is a type/universal
         // selector. Any other case means we want to apply the selector onto
@@ -128,10 +140,7 @@ class PQueryEngine {
               default:
                 const querySelector = customQuerySelectors.get(selector.name);
                 if (!querySelector) {
-                  throw new SelectorError(
+                  throw new Error(`Unknown selector type: ${selector.name}`);
-                    input,
-                    `Unknown selector type: ${selector.name}`
-                  );
                 }
                 yield* querySelector.querySelectorAll(element, selector.value);
             }
@@ -240,17 +249,7 @@ export const pQuerySelectorAll = function (
   root: Node,
   selector: string
 ): AwaitableIterable<Node> {
-  let selectors: ComplexPSelectorList;
+  const selectors = JSON.parse(selector) as ComplexPSelectorList;
-  let isPureCSS: boolean;
-  try {
-    [selectors, isPureCSS] = parsePSelectors(selector);
-  } catch (error) {
-    return (root as unknown as QueryableNode).querySelectorAll(selector);
-  }
-
-  if (isPureCSS) {
-    return (root as unknown as QueryableNode).querySelectorAll(selector);
-  }
   // If there are any empty elements, then this implies the selector has
   // contiguous combinators (e.g. `>>> >>>>`) or starts/ends with one which we
   // treat as illegal, similar to existing behavior.
@@ -267,15 +266,12 @@ export const pQuerySelectorAll = function (
       });
     })
   ) {
-    throw new SelectorError(
+    throw new Error('Multiple deep combinators found in sequence.');
-      selector,
-      'Multiple deep combinators found in sequence.'
-    );
   }
 
   return domSort(
     AsyncIterableUtil.flatMap(selectors, selectorParts => {
-      const query = new PQueryEngine(root, selector, selectorParts);
+      const query = new PQueryEngine(root, selectorParts);
       void query.run();
       return query.elements;
     })

packages/puppeteer-core/src/injected/injected.ts

@@ -8,6 +8,7 @@ import {Deferred} from '../util/Deferred.js';
 import {createFunction} from '../util/Function.js';
 
 import * as ARIAQuerySelector from './ARIAQuerySelector.js';
+import * as CSSSelector from './CSSSelector.js';
 import * as CustomQuerySelectors from './CustomQuerySelector.js';
 import * as PierceQuerySelector from './PierceQuerySelector.js';
 import {IntervalPoller, MutationPoller, RAFPoller} from './Poller.js';
@@ -31,6 +32,7 @@ const PuppeteerUtil = Object.freeze({
   ...TextQuerySelector,
   ...util,
   ...XPathQuerySelector,
+  ...CSSSelector,
   Deferred,
   createFunction,
   createTextContent,

packages/puppeteer-core/third_party/parsel-js/package.json

@@ -0,0 +1,3 @@
+{
+  "type": "module"
+}

packages/puppeteer-core/third_party/parsel-js/parsel-js.ts

@@ -0,0 +1,4 @@
+// esline-disable rulesdir/check-license
+export {tokenize, TOKENS, stringify} from 'parsel-js';
+
+export type * from 'parsel-js';