@govtechsg/oobee 0.10.39 → 0.10.43

package/src/npmIndex.ts

@@ -1,7 +1,7 @@
 import fs from 'fs';
 import path from 'path';
 import printMessage from 'print-message';
-import axe, { ImpactValue } from 'axe-core';
+import axe, { AxeResults, ImpactValue } from 'axe-core';
 import { fileURLToPath } from 'url';
 import { EnqueueStrategy } from 'crawlee';
 import constants, { BrowserTypes, RuleFlags, ScannerTypes } from './constants/constants.js';
@@ -16,7 +16,7 @@ import { createCrawleeSubFolders, filterAxeResults } from './crawlers/commonCraw
 import { createAndUpdateResultsFolders, createDetailsAndLogs } from './utils.js';
 import generateArtifacts from './mergeAxeResults.js';
 import { takeScreenshotForHTMLElements } from './screenshotFunc/htmlScreenshotFunc.js';
-import { silentLogger } from './logs.js';
+import { consoleLogger, silentLogger } from './logs.js';
 import { alertMessageOptions } from './constants/cliFunctions.js';
 import { evaluateAltText } from './crawlers/custom/evaluateAltText.js';
 import { escapeCssSelector } from './crawlers/custom/escapeCssSelector.js';
@@ -24,7 +24,7 @@ import { framesCheck } from './crawlers/custom/framesCheck.js';
 import { findElementByCssSelector } from './crawlers/custom/findElementByCssSelector.js';
 import { getAxeConfiguration } from './crawlers/custom/getAxeConfiguration.js';
 import { flagUnlabelledClickableElements } from './crawlers/custom/flagUnlabelledClickableElements.js';
-import { xPathToCss } from './crawlers/custom/xPathToCss.js';
+import xPathToCss from './crawlers/custom/xPathToCss.js';
 import { extractText } from './crawlers/custom/extractText.js';
 import { gradeReadability } from './crawlers/custom/gradeReadability.js';
 
@@ -65,7 +65,7 @@ export const init = async ({
   specifiedMaxConcurrency?: number;
   followRobots?: boolean;
 }) => {
-  console.log('Starting Oobee');
+  consoleLogger.info('Starting Oobee');
 
   const [date, time] = new Date().toLocaleString('sv').replaceAll(/-|:/g, '').split(' ');
   const domain = new URL(entryUrl).hostname;
@@ -126,7 +126,7 @@ export const init = async ({
             const cssSelector = xPathToCss(xpath);
             return cssSelector;
           } catch (e) {
-            console.error('Error converting XPath to CSS: ', xpath, e);
+            consoleLogger.error(`Error converting XPath to CSS: ${xpath} - ${e}`);
             return '';
           }
         })
@@ -197,7 +197,11 @@ export const init = async ({
     `;
   };
 
-  const pushScanResults = async (res, metadata, elementsToClick) => {
+  const pushScanResults = async (
+    res: { pageUrl: string; pageTitle: string; axeScanResults: AxeResults },
+    metadata: string,
+    elementsToClick: string[],
+  ) => {
     throwErrorIfTerminated();
     if (includeScreenshots) {
       // use chrome by default
@@ -211,7 +215,7 @@ export const init = async ({
       await page.waitForLoadState('networkidle');
 
       // click on elements to reveal hidden elements so screenshots can be taken
-      elementsToClick?.forEach(async elem => {
+      elementsToClick?.forEach(async (elem: string) => {
         try {
           await page.locator(elem).click();
         } catch (e) {
@@ -259,7 +263,7 @@ export const init = async ({
 
   const terminate = async () => {
     throwErrorIfTerminated();
-    console.log('Stopping Oobee');
+    consoleLogger.info('Stopping Oobee');
     isInstanceTerminated = true;
     scanDetails.endTime = new Date();
     scanDetails.urlsCrawled = urlsCrawled;

package/src/screenshotFunc/htmlScreenshotFunc.ts

@@ -7,14 +7,14 @@ import { Result } from 'axe-core';
 import { Page } from 'playwright';
 import { NodeResultWithScreenshot, ResultWithScreenshot } from '../crawlers/commonCrawlerFunc.js';
 
-const screenshotMap = {}; // Map of screenshot hashkey to its buffer value and screenshot path
+const screenshotMap: Record<string, string> = {}; // Map of screenshot hashkey to its buffer value and screenshot path
 
 export const takeScreenshotForHTMLElements = async (
   violations: Result[],
   page: Page,
   randomToken: string,
   locatorTimeout = 2000,
-  maxScreenshots = 50,
+  maxScreenshots = 100,
 ): Promise<ResultWithScreenshot[]> => {
   const newViolations: ResultWithScreenshot[] = [];
   let screenshotCount = 0;
@@ -75,30 +75,18 @@ export const takeScreenshotForHTMLElements = async (
   return newViolations;
 };
 
-const generateBufferHash = (buffer: Buffer) => {
+const generateBufferHash = (buffer: Buffer): string => {
   const hash = createHash('sha256');
   hash.update(buffer);
   return hash.digest('hex');
 };
 
-const isSameBufferHash = (buffer: Buffer, hash: string) => {
-  const bufferHash = generateBufferHash(buffer);
-  return hash === bufferHash;
-};
-
-const getIdenticalScreenshotKey = (buffer: Buffer) => {
-  for (const hashKey in screenshotMap) {
-    const isIdentical = isSameBufferHash(buffer, hashKey);
-    if (isIdentical) return hashKey;
-  }
-  return undefined;
-};
-
-const getScreenshotPath = (buffer: Buffer, randomToken: string) => {
-  let hashKey = getIdenticalScreenshotKey(buffer);
+const getScreenshotPath = (buffer: Buffer, randomToken: string): string => {
+  let hashKey = generateBufferHash(buffer);
+  const existingPath = screenshotMap[hashKey];
   // If exists identical entry in screenshot map, get its filepath
-  if (hashKey) {
-    return screenshotMap[hashKey];
+  if (existingPath) {
+    return existingPath;
   }
   // Create new entry in screenshot map
   hashKey = generateBufferHash(buffer);

package/src/screenshotFunc/pdfScreenshotFunc.ts

@@ -1,3 +1,10 @@
+// Monkey patch Path2D to avoid PDF.js crashing
+(globalThis as any).Path2D = class {
+  constructor(_path?: string) {}
+  rect(_x: number, _y: number, _width: number, _height: number) {}
+  addPath(_path: any, _transform?: any) {}
+};
+
 import _ from 'lodash';
 import  { getDocument, PDFPageProxy } from 'pdfjs-dist';
 import fs from 'fs';
@@ -25,11 +32,36 @@ interface pathObject {
   annot?: number;
 }
 
+// Use safe canvas to avoid Path2D issues
+function createSafeCanvas(width: number, height: number) {
+  const canvas = createCanvas(width, height);
+  const ctx = canvas.getContext('2d');
+
+  // Patch clip/stroke/fill/etc. to skip if Path2D is passed
+  const wrapIgnorePath2D = (fn: Function) =>
+    function (...args: any[]) {
+      if (args.length > 0 && args[0] instanceof (globalThis as any).Path2D) {
+        // Skip the operation
+        return;
+      }
+      return fn.apply(this, args);
+    };
+
+  ctx.clip = wrapIgnorePath2D(ctx.clip);
+  ctx.fill = wrapIgnorePath2D(ctx.fill);
+  ctx.stroke = wrapIgnorePath2D(ctx.stroke);
+  ctx.isPointInPath = wrapIgnorePath2D(ctx.isPointInPath);
+  ctx.isPointInStroke = wrapIgnorePath2D(ctx.isPointInStroke);
+
+  return canvas;
+}
+
+// CanvasFactory for Node.js
 function NodeCanvasFactory() {}
 NodeCanvasFactory.prototype = {
   create: function NodeCanvasFactory_create(width: number, height: number) {
     assert(width > 0 && height > 0, 'Invalid canvas size');
-    const canvas = createCanvas(width, height);
+    const canvas = createSafeCanvas(width, height);
     const context = canvas.getContext('2d');
     return {
       canvas,
@@ -69,6 +101,7 @@ export async function getPdfScreenshots(
   const newItems = _.cloneDeep(items);
   const loadingTask = getDocument({
     url: pdfFilePath,
+    canvasFactory,
     standardFontDataUrl: path.join(dirname, '../node_modules/pdfjs-dist/standard_fonts/'),
     disableFontFace: true,
     verbosity: 0,

package/src/types/text-readability.d.ts

@@ -0,0 +1,3 @@
+declare module 'text-readability' {
+  function fleschReadingEase(text: string): number;
+}

package/src/types/types.ts

@@ -21,7 +21,7 @@ export type StructureTree = {
   pageIndex?: number;
 };
 
-type DeviceDescriptor = {
+export type DeviceDescriptor = {
   viewport: ViewportSize;
   userAgent: string;
   deviceScaleFactor: number;