Upgrade PDF.js to v5 (#24948)

* Reapply "Upgrade PDF.js to v5 (#24646)" (#24946)
* Upgrade core-js, caniuse-lite and babel

GitOrigin-RevId: 63398189301b5f5adc8a17b332d92dccfc26d612

patches/pdfjs-dist+5.1.91.patch

@@ -0,0 +1,22 @@
+diff --git a/node_modules/pdfjs-dist/build/pdf.worker.mjs b/node_modules/pdfjs-dist/build/pdf.worker.mjs
+index 6c5c6f1..bb6b7d1 100644
+--- a/node_modules/pdfjs-dist/build/pdf.worker.mjs
++++ b/node_modules/pdfjs-dist/build/pdf.worker.mjs
+@@ -1830,7 +1830,7 @@ async function __wbg_init(module_or_path) {
+     }
+   }
+   if (typeof module_or_path === 'undefined') {
+-    module_or_path = new URL('qcms_bg.wasm', import.meta.url);
++    module_or_path = new URL(/* webpackIgnore: true */ 'qcms_bg.wasm', import.meta.url);
+   }
+   const imports = __wbg_get_imports();
+   if (typeof module_or_path === 'string' || typeof Request === 'function' && module_or_path instanceof Request || typeof URL === 'function' && module_or_path instanceof URL) {
+@@ -5358,7 +5358,7 @@ var OpenJPEG = (() => {
+       if (Module["locateFile"]) {
+         return locateFile("openjpeg.wasm");
+       }
+-      return new URL("openjpeg.wasm", import.meta.url).href;
++      return new URL(/* webpackIgnore: true */ "openjpeg.wasm", import.meta.url).href;
+     }
+     function getBinarySync(file) {
+       if (file == wasmBinaryFile && wasmBinary) {

services/web/babel.config.json

@@ -6,7 +6,7 @@
       {
         "useBuiltIns": "usage",
         // This version must be aligned with the `core-js` version in `package.json`
-        "corejs": { "version": "3.38" },
+        "corejs": { "version": "3.41" },
         "exclude": [
           // Exclude Array.prototype.push polyfill, as it's not needed and affects performance in Chrome
           "es.array.push",

services/web/frontend/js/features/pdf-preview/components/pdf-js-viewer.tsx

@@ -168,10 +168,10 @@ function PdfJsViewer({ url, pdfFile }: PdfJsViewerProps) {
       setStartFetch(performance.now())
 
       const abortController = new AbortController()
-      const handleFetchError = (err: Error) => {
+      const handleFetchError = (err: any) => {
         if (abortController.signal.aborted) return
         // The error is already logged at the call-site with additional context.
-        if (err instanceof PDFJS.MissingPDFException) {
+        if (err instanceof PDFJS.ResponseException && err.missing) {
           setError('rendering-error-expected')
         } else {
           setError('rendering-error')

services/web/frontend/js/features/pdf-preview/util/pdf-caching-transport.js

@@ -155,14 +155,11 @@ export function generatePdfCachingTransportFactory() {
             return blob
           })
           .catch(err => {
+            const { statusCode, url } = OError.getFullInfo(err)
             throw OError.tag(
-              new PDFJS.MissingPDFException(),
+              new PDFJS.ResponseException(undefined, statusCode, true),
               'cache-fallback',
-              {
-                statusCode: OError.getFullInfo(err).statusCode,
-                url: OError.getFullInfo(err).url,
-                err,
-              }
+              { statusCode, url, err }
             )
           })
       }
@@ -198,11 +195,12 @@ export function generatePdfCachingTransportFactory() {
               metrics.failedCount++
               metrics.failedOnce = true
             }
-            throw OError.tag(new PDFJS.MissingPDFException(), 'caching', {
-              statusCode: OError.getFullInfo(err).statusCode,
-              url: OError.getFullInfo(err).url,
-              err,
-            })
+            const { statusCode, url } = OError.getFullInfo(err)
+            throw OError.tag(
+              new PDFJS.ResponseException(undefined, statusCode, true),
+              'caching',
+              { statusCode, url, err }
+            )
           }
           metrics.failedCount++
           metrics.failedOnce = true
@@ -226,11 +224,12 @@ export function generatePdfCachingTransportFactory() {
           }).catch(err => {
             if (canTryFromCache(err)) return fetchFromCache()
             if (isExpectedError(err)) {
-              throw OError.tag(new PDFJS.MissingPDFException(), 'fallback', {
-                statusCode: OError.getFullInfo(err).statusCode,
-                url: OError.getFullInfo(err).url,
-                err,
-              })
+              const { statusCode, url } = OError.getFullInfo(err)
+              throw OError.tag(
+                new PDFJS.ResponseException(undefined, statusCode, true),
+                'fallback',
+                { statusCode, url, err }
+              )
             }
             throw err
           })
@@ -251,7 +250,7 @@ export function generatePdfCachingTransportFactory() {
           if (abortSignal.aborted) return
           err = OError.tag(err, 'fatal pdf download error', getDebugInfo())
           debugConsole.error(err)
-          if (!(err instanceof PDFJS.MissingPDFException)) {
+          if (!(err instanceof PDFJS.ResponseException && err.missing)) {
             captureException(err, {
               tags: {
                 fromPdfCaching: true,

services/web/frontend/js/features/pdf-preview/util/pdf-js-wrapper.ts

@@ -57,7 +57,7 @@ export default class PDFJSWrapper {
     url: string
     pdfFile: Record<string, any>
     abortController: AbortController
-    handleFetchError: (error: Error) => void
+    handleFetchError: (error: any) => void
   }) {
     this.url = url
 
@@ -90,7 +90,10 @@ export default class PDFJSWrapper {
 
       return doc
     } catch (error: any) {
-      if (!error || error.name !== 'MissingPDFException') {
+      if (
+        !error ||
+        !(error instanceof PDFJS.ResponseException && error.missing === true)
+      ) {
         captureException(error, {
           tags: { handler: 'pdf-preview' },
         })

services/web/frontend/js/features/pdf-preview/util/pdf-js.ts

@@ -11,6 +11,8 @@ PDFJS.GlobalWorkerOptions.workerPort = new Worker(
 
 export const imageResourcesPath = '/images/pdfjs-dist/'
 const cMapUrl = '/js/pdfjs-dist/cmaps/'
+const wasmUrl = '/js/pdfjs-dist/wasm/'
+const iccUrl = '/js/pdfjs-dist/iccs/'
 const standardFontDataUrl = '/fonts/pdfjs-dist/'
 
 const params = new URLSearchParams(window.location.search)
@@ -24,6 +26,8 @@ export const loadPdfDocumentFromUrl = (
   PDFJS.getDocument({
     url,
     cMapUrl,
+    wasmUrl,
+    iccUrl,
     standardFontDataUrl,
     disableFontFace,
     disableAutoFetch: true, // only fetch the data needed for the displayed pages

services/web/package.json

@@ -178,12 +178,12 @@
     "yauzl": "^2.10.0"
   },
   "devDependencies": {
-    "@babel/cli": "^7.24.8",
-    "@babel/core": "^7.25.2",
-    "@babel/preset-env": "^7.25.3",
-    "@babel/preset-react": "^7.24.7",
-    "@babel/preset-typescript": "^7.24.7",
-    "@babel/register": "^7.24.6",
+    "@babel/cli": "^7.27.0",
+    "@babel/core": "^7.26.10",
+    "@babel/preset-env": "^7.26.9",
+    "@babel/preset-react": "^7.26.3",
+    "@babel/preset-typescript": "^7.27.0",
+    "@babel/register": "^7.25.9",
     "@codemirror/autocomplete": "github:overleaf/codemirror-autocomplete#6445cd056671c98d12d1c597ba705e11327ec4c5",
     "@codemirror/commands": "^6.8.0",
     "@codemirror/lang-markdown": "^6.3.2",
@@ -274,7 +274,7 @@
     "classnames": "^2.2.6",
     "cookie-signature": "^1.2.1",
     "copy-webpack-plugin": "^11.0.0",
-    "core-js": "^3.38.1",
+    "core-js": "^3.41.0",
     "css-loader": "^6.8.1",
     "css-minimizer-webpack-plugin": "^5.0.1",
     "cypress": "13.13.2",
@@ -320,7 +320,7 @@
     "nock": "^13.5.6",
     "nvd3": "^1.8.6",
     "overleaf-editor-core": "*",
-    "pdfjs-dist": "4.10.38",
+    "pdfjs-dist": "5.1.91",
     "pirates": "^4.0.1",
     "postcss": "^8.4.31",
     "postcss-loader": "^7.3.3",

services/web/webpack.config.js

@@ -380,12 +380,22 @@ module.exports = {
           context: `${dictionariesDir}/dictionaries`,
         },
         // Copy CMap files (used to provide support for non-Latin characters),
-        // fonts and images from pdfjs-dist package to build output.
+        // wasm, ICC profiles, fonts and images from pdfjs-dist package to build output.
         {
           from: 'cmaps',
           to: 'js/pdfjs-dist/cmaps',
           context: pdfjsDir,
         },
+        {
+          from: 'iccs',
+          to: 'js/pdfjs-dist/iccs',
+          context: pdfjsDir,
+        },
+        {
+          from: 'wasm',
+          to: 'js/pdfjs-dist/wasm',
+          context: pdfjsDir,
+        },
         {
           from: 'standard_fonts',
           to: 'fonts/pdfjs-dist',