Updated to fix support for Windows Node.js

js/generalWorkerMain.js

@@ -6,37 +6,50 @@ export async function initGeneralWorker() {
   // This method of creating workers works natively in the browser, Node.js, and Webpack 5.
   // Do not change without confirming compatibility with all three.
   const obj = {};
-  let worker;
-  if (typeof process === 'undefined') {
-    worker = new Worker(new URL('./worker/generalWorker.js', import.meta.url), { type: 'module' });
-  } else {
-    const WorkerNode = typeof process === 'undefined' ? Worker : (await import('web-worker')).default;
-    worker = new WorkerNode(new URL('./worker/generalWorker.js', import.meta.url), { type: 'module' });
-  }
+  const Worker = typeof process === 'undefined' ? globalThis.Worker : (await import('worker_threads')).Worker;
+  const worker = new Worker(new URL('./worker/generalWorker.js', import.meta.url), { type: 'module' });
 
   return new Promise((resolve, reject) => {
-    worker.onerror = (err) => {
+
+    const errorHandler = (err) => {
       console.error(err);
     };
 
+    if (typeof process === 'undefined') {
+      // @ts-ignore
+      worker.onerror = errorHandler;
+    } else {
+      // @ts-ignore
+      worker.on('error', errorHandler);
+    }
+
     const workerPromises = {};
     let promiseId = 0;
 
     const ready = new Promise((innerResolve, innerReject) => {
       workerPromises['0'] = { resolve: innerResolve, reject: innerReject, func: 'ready' };
     });
 
-    worker.onmessage = async (event) => {
-      if (workerPromises[event.data.id]) {
-        if (event.data.status === 'reject') {
-          console.log(event.data.data);
-          workerPromises[event.data.id].reject(event.data.data);
+
+    const messageHandler = async (data) => {
+      if (workerPromises[data.id]) {
+        if (data.status === 'reject') {
+          console.log(data.data);
+          workerPromises[data.id].reject(data.data);
         } else {
-          workerPromises[event.data.id].resolve(event.data.data);
+          workerPromises[data.id].resolve(data.data);
         }
       }
     };
 
+    if (typeof process === 'undefined') {
+      // @ts-ignore
+      worker.onmessage = (event) => messageHandler(event.data);
+    } else {
+      // @ts-ignore
+      worker.on('message', messageHandler);
+    }
+
     /**
        * Wraps a function to be called via worker messages.
        * @param {string} func The function name to call.

js/worker/generalWorker.js

@@ -15,7 +15,8 @@ import {
 } from './compareOCRModule.js';
 import { optimizeFont } from './optimizeFontModule.js';
 
-// import Tesseract from "../../tess/tesseract.esm.min.js";
+const parentPort = typeof process === 'undefined' ? globalThis : (await import('worker_threads')).parentPort;
+if (!parentPort) throw new Error('This file must be run in a worker');
 
 const Tesseract = typeof process === 'undefined' ? (await import('../../tess/tesseract.esm.min.js')).default : await import('@scribe.js/tesseract.js');
 
@@ -296,7 +297,7 @@ export const recognizeAndConvert2 = async ({
 
       const xB = { recognize: res1.data, convert: { legacy: null, lstm: resLSTM } };
 
-      postMessage({ data: xB, id: `${id}b`, status: 'resolve' });
+      parentPort.postMessage({ data: xB, id: `${id}b`, status: 'resolve' });
     })();
   } else if (!options.lstm && options.legacy) {
     const legacyBlocks = /** @type {Array<import('@scribe.js/tesseract.js').Block>} */(res0.data.blocks);
@@ -312,10 +313,10 @@ export const recognizeAndConvert2 = async ({
 
   const x = { recognize: res0.data, convert: { legacy: resLegacy, lstm: resLSTM } };
 
-  postMessage({ data: x, id, status: 'resolve' });
+  parentPort.postMessage({ data: x, id, status: 'resolve' });
 
   // Both promises must resolve for the scheduler to move on, even if only one OCR engine is being run.
-  if (!options.legacy || !options.lstm) postMessage({ data: null, id: `${id}b` });
+  if (!options.legacy || !options.lstm) parentPort.postMessage({ data: null, id: `${id}b` });
 };
 
 /**
@@ -373,13 +374,11 @@ async function compareOCRPageImpWrap(args) {
   return await compareOCRPageImp(args);
 }
 
-postMessage({ data: 'ready', id: 0, status: 'resolve' });
+const handleMessage = async (data) => {
 
-// eslint-disable-next-line no-restricted-globals
-addEventListener('message', async (e) => {
-  const func = e.data[0];
-  const args = e.data[1];
-  const id = e.data[2];
+  const func = data[0];
+  const args = data[1];
+  const id = data[2];
 
   if (func === 'recognizeAndConvert2') {
     recognizeAndConvert2(args, id);
@@ -415,6 +414,15 @@ addEventListener('message', async (e) => {
     loadFontsWorker,
     updateFontContWorker,
   })[func](args)
-    .then((x) => postMessage({ data: x, id, status: 'resolve' }))
-    .catch((err) => postMessage({ data: err, id, status: 'reject' }));
-});
+    .then((x) => parentPort.postMessage({ data: x, id, status: 'resolve' }))
+    .catch((err) => parentPort.postMessage({ data: err, id, status: 'reject' }));
+};
+
+if (typeof process === 'undefined') {
+  onmessage = (event) => handleMessage(event.data);
+} else {
+  parentPort.on('message', handleMessage);
+}
+
+
+parentPort.postMessage({ data: 'ready', id: 0, status: 'resolve' });

mupdf/mupdf-async.js

@@ -27,31 +27,35 @@ export async function initMuPDFWorker() {
   // This method of creating workers works natively in the browser, Node.js, and Webpack 5.
   // Do not change without confirming compatibility with all three.
   const mupdf = {};
-  let worker;
+  const Worker = typeof process === 'undefined' ? globalThis.Worker : (await import('worker_threads')).Worker;
+  const worker = new Worker(new URL('./mupdf-worker.js', import.meta.url), { type: 'module' });
+
+  const errorHandler = (err) => {
+    console.error(err);
+  };
+
   if (typeof process === 'undefined') {
-    worker = new Worker(new URL('./mupdf-worker.js', import.meta.url), { type: 'module' });
+    // @ts-ignore
+    worker.onerror = errorHandler;
   } else {
-    const WorkerNode = typeof process === 'undefined' ? Worker : (await import('web-worker')).default;
-    worker = new WorkerNode(new URL('./mupdf-worker.js', import.meta.url), { type: 'module' });
+    // @ts-ignore
+    worker.on('error', errorHandler);
   }
 
-  worker.onerror = function (error) {
-    throw error;
-  };
-
   let readyResolve;
   const readyPromise = new Promise((resolve, reject) => {
     readyResolve = resolve;
   });
 
   worker.promises = {};
   worker.promiseId = 0;
-  worker.onmessage = async function (event) {
-    if (typeof event.data === 'string' && event.data === 'READY') {
+
+  const messageHandler = async (data) => {
+    if (typeof data === 'string' && data === 'READY') {
       readyResolve();
       return;
     }
-    const [type, id, result] = event.data;
+    const [type, id, result] = data;
     if (type === 'RESULT') {
       // worker.promises[id].resolve(result);
       if (['drawPageAsPNG'].includes(worker.promises[id].func)) {
@@ -69,6 +73,14 @@ export async function initMuPDFWorker() {
     }
   };
 
+  if (typeof process === 'undefined') {
+    // @ts-ignore
+    worker.onmessage = (event) => messageHandler(event.data);
+  } else {
+    // @ts-ignore
+    worker.on('message', messageHandler);
+  }
+
   function wrap(func) {
     return function (...args) {
       return new Promise((resolve, reject) => {

mupdf/mupdf-worker.js

@@ -20,6 +20,9 @@
 // Artifex Software, Inc., 1305 Grant Avenue - Suite 200, Novato,
 // CA 94945, U.S.A., +1(415)492-9861, for further information.
 
+const parentPort = typeof process === 'undefined' ? globalThis : (await import('worker_threads')).parentPort;
+if (!parentPort) throw new Error('This file must be run in a worker');
+
 // Copied from https://gist.github.com/jonleighton/958841
 function arrayBufferToBase64(arrayBuffer) {
   let base64 = '';
@@ -127,7 +130,7 @@ Module.onRuntimeInitialized = function () {
   mupdf.outlineNext = Module.cwrap('outlineNext', 'number', ['number']);
   wasm_checkNativeText = Module.cwrap('checkNativeText', 'number', ['number', 'number']);
   mupdf.writeDocument = Module.cwrap('writeDocument', 'null', []);
-  postMessage('READY');
+  parentPort.postMessage('READY');
   ready = true;
 };
 
@@ -446,19 +449,25 @@ mupdf.search = function (doc, page, dpi, needle) {
   return JSON.parse(mupdf.searchJSON(doc, page, dpi, needle));
 };
 
-addEventListener('message', (event) => {
-  const [func, args, id] = event.data;
+const handleMessage = (data) => {
+  const [func, args, id] = data;
   if (!ready) {
-    postMessage(['ERROR', id, { name: 'NotReadyError', message: 'WASM module is not ready yet' }]);
+    parentPort.postMessage(['ERROR', id, { name: 'NotReadyError', message: 'WASM module is not ready yet' }]);
     return;
   }
   try {
     const result = mupdf[func](...args);
-    if (result instanceof ArrayBuffer) postMessage(['RESULT', id, result], [result]);
+    if (result instanceof ArrayBuffer) parentPort.postMessage(['RESULT', id, result], [result]);
     else if (result?.buffer instanceof ArrayBuffer) {
-      postMessage(['RESULT', id, result], [result.buffer]);
-    } else postMessage(['RESULT', id, result]);
+      parentPort.postMessage(['RESULT', id, result], [result.buffer]);
+    } else parentPort.postMessage(['RESULT', id, result]);
   } catch (error) {
-    postMessage(['ERROR', id, { name: error.name, message: error.message }]);
+    parentPort.postMessage(['ERROR', id, { name: error.name, message: error.message }]);
   }
-});
+}
+
+if (typeof process === 'undefined') {
+  onmessage = (event) => handleMessage(event.data);
+} else {
+  parentPort.on('message', handleMessage);
+}

package.json

@@ -52,7 +52,6 @@
     "canvas": "^2.11.2",
     "commander": "^11.1.0",
     "puppeteer": "^22.13.0",
-    "@scribe.js/tesseract.js": "^5.1.1",
-    "web-worker": "~1.2.0"
+    "@scribe.js/tesseract.js": "^5.1.1"
   }
 }