Implemented fixes for supporting webpack; fixed bug in previous commit

js/fontContainerMain.js

@@ -14,16 +14,19 @@ import { gs } from './generalWorkerMain.js';
  * @param {string} fileName
  */
 async function fontPathToArrayBuffer(fileName) {
-  const browserMode = typeof process === 'undefined';
   const absPath = getFontAbsPath(fileName);
 
-  if (browserMode) {
+  if (typeof process === 'undefined') {
     const res = await fetch(absPath);
     return res.arrayBuffer();
+  // Important: Do not remove `else` statement.
+  // Some build tools (Webpack) need it to avoid trying to `node:fs` in the browser.
+  // eslint-disable-next-line no-else-return
+  } else {
+    const { readFileSync } = await import('node:fs');
+    const res = readFileSync(absPath);
+    return res.buffer;
   }
-  const { readFileSync } = await import('node:fs');
-  const res = readFileSync(absPath);
-  return res.buffer;
 }
 
 async function fontPathToArrayBufferAll(fileNameObj) {
@@ -117,7 +120,6 @@ export async function loadChiSimFont() {
  *    This should be used when switching from unvalidated to validated optimized fonts.
  */
 export async function enableFontOpt(enable, useInitial = false, forceWorkerUpdate = false) {
-
   // Enable/disable optimized font
   if (enable && useInitial && fontAll.optInitial) {
     fontAll.active = fontAll.optInitial;

js/fontEval.js

@@ -2,7 +2,9 @@ import { DebugData, fontMetricsObj, pageMetricsArr } from './containers/dataCont
 import { fontAll } from './containers/fontContainer.js';
 import { ImageCache } from './containers/imageContainer.js';
 import {
-  enableFontOpt, optimizeFontContainerAll, setDefaultFontAuto, loadBuiltInFontsRaw,
+  enableFontOpt,
+  loadBuiltInFontsRaw,
+  optimizeFontContainerAll, setDefaultFontAuto,
 } from './fontContainerMain.js';
 import { gs } from './generalWorkerMain.js';
 
@@ -15,8 +17,6 @@ import { gs } from './generalWorkerMain.js';
 export async function evalPageFonts(font, pageArr, n = 500) {
   if (!gs.scheduler) throw new Error('GeneralScheduler must be defined before this function can run.');
 
-  const browserMode = typeof process === 'undefined';
-
   let metricTotal = 0;
   let wordsTotal = 0;
 
@@ -28,7 +28,7 @@ export async function evalPageFonts(font, pageArr, n = 500) {
     // The Node.js canvas package does not currently support worke threads
     // https://github.com/Automattic/node-canvas/issues/1394
     let res;
-    if (!browserMode) {
+    if (!(typeof process === 'undefined')) {
       const { evalPageFont } = await import('./worker/compareOCRModule.js');
 
       res = await evalPageFont({
@@ -162,8 +162,6 @@ export async function evaluateFonts(pageArr) {
  * and (2) no images are provided to compare against.
  */
 export async function runFontOptimization(ocrArr) {
-  const browserMode = typeof process === 'undefined';
-
   await loadBuiltInFontsRaw();
 
   const fontRaw = fontAll.getContainer('raw');
@@ -195,7 +193,7 @@ export async function runFontOptimization(ocrArr) {
     await enableFontOpt(false);
 
     // This step needs to happen here as all fonts must be registered before initializing the canvas.
-    if (!browserMode) {
+    if (!(typeof process === 'undefined')) {
       const { initCanvasNode } = await import('./worker/compareOCRModule.js');
       await initCanvasNode();
     }

js/recognizeConvert.js

@@ -266,12 +266,10 @@ export function checkCharWarn(warnArr) {
   const charWarnCt = warnArr.filter((x) => x?.char === 'char_warning').length;
   const charGoodCt = warnArr.length - charErrorCt - charWarnCt;
 
-  const browserMode = typeof process === 'undefined';
-
   // The UI warning/error messages cannot be thrown within this function,
   // as that would make this file break when imported into contexts that do not have the main UI.
   if (charGoodCt === 0 && charErrorCt > 0) {
-    if (browserMode) {
+    if (typeof process === 'undefined') {
       const errorHTML = `No character-level OCR data detected. Abbyy XML is only supported with character-level data. 
         <a href="https://docs.scribeocr.com/faq.html#is-character-level-ocr-data-required--why" target="_blank" class="alert-link">Learn more.</a>`;
       opt.errorHandler(errorHTML);
@@ -281,7 +279,7 @@ export function checkCharWarn(warnArr) {
       opt.errorHandler(errorText);
     }
   } if (charGoodCt === 0 && charWarnCt > 0) {
-    if (browserMode) {
+    if (typeof process === 'undefined') {
       const warningHTML = `No character-level OCR data detected. Font optimization features will be disabled. 
         <a href="https://docs.scribeocr.com/faq.html#is-character-level-ocr-data-required--why" target="_blank" class="alert-link">Learn more.</a>`;
       opt.warningHandler(warningHTML);

js/utils/imageUtils.js

@@ -1,6 +1,4 @@
 /* eslint-disable no-bitwise */
-
-const browserMode = typeof process === 'undefined';
 /**
  * Handles various image formats, always returns a ImageBitmap.
  *
@@ -13,7 +11,7 @@ export async function getImageBitmap(img) {
   if (img === null) throw new Error('Input is null');
 
   if (typeof img === 'string') {
-    if (browserMode) {
+    if (typeof process === 'undefined') {
       const imgBlob = imageStrToBlob(img);
       const imgBit = await createImageBitmap(imgBlob);
       return imgBit;
@@ -26,7 +24,7 @@ export async function getImageBitmap(img) {
   // In Node.js the input is assumed to be already compatible with the `canvas.drawImage` method.
   // Additionally, `ImageBitmap` does not exist within the Node canvas package.
   // Second condition exists for type detection purposes.
-  if (!browserMode && (typeof img !== 'string') && (typeof img !== 'number')) return img;
+  if (!(typeof process === 'undefined') && (typeof img !== 'string') && (typeof img !== 'number')) return img;
 
   return img;
 }

js/worker/compareOCRModule.js

@@ -10,8 +10,6 @@ import { drawWordActual, drawWordRender } from './renderWordCanvas.js';
 import { fontAll } from '../containers/fontContainer.js';
 // import { CompDebug } from '../objects/imageObjects.js';
 
-const browserMode = typeof process === 'undefined';
-
 /** @type {OffscreenCanvasRenderingContext2D} */
 let calcCtx;
 /** @type {OffscreenCanvasRenderingContext2D} */
@@ -22,7 +20,7 @@ let viewCtx1;
 let viewCtx2;
 
 // Browser case
-if (browserMode) {
+if (typeof process === 'undefined') {
   // For whatever reason, this can fail silently in some browsers that do not support OffscreenCanvas, where the worker simply stops running.
   // Therefore, an explicit error message is added here to make the issue evident. Features will still fail, so this is not a fix.
   try {
@@ -327,7 +325,7 @@ export async function evalWords({
   /** @type {?CompDebugBrowser|CompDebugNode} */
   let debugImg = null;
   if (view) {
-    if (browserMode) {
+    if (typeof process === 'undefined') {
       const imageRaw = await viewCtx0.canvas.convertToBlob();
       const imageA = await viewCtx1.canvas.convertToBlob();
       const imageB = await viewCtx2.canvas.convertToBlob();
@@ -979,7 +977,7 @@ export async function checkWords(wordsA, binaryImage, imageRotated, pageMetricsO
     tessedit_pageseg_mode: '6', // "Single block"
   };
 
-  const inputImage = browserMode ? await calcCtx.canvas.convertToBlob() : await calcCtx.canvas.toBuffer('image/png');
+  const inputImage = typeof process === 'undefined' ? await calcCtx.canvas.convertToBlob() : await calcCtx.canvas.toBuffer('image/png');
 
   let res;
   if (options.tessScheduler) {

js/worker/generalWorker.js

@@ -15,9 +15,8 @@ import {
 import { optimizeFont } from './optimizeFontModule.js';
 
 // import Tesseract from "../../tess/tesseract.esm.min.js";
-const browserMode = typeof process === 'undefined';
 
-const Tesseract = browserMode ? (await import('../../tess/tesseract.esm.min.js')).default : await import('@scribe.js/tesseract.js');
+const Tesseract = typeof process === 'undefined' ? (await import('../../tess/tesseract.esm.min.js')).default : await import('@scribe.js/tesseract.js');
 
 const defaultConfigs = {
   // TODO: Add back support for multiple PSM modes.
@@ -64,7 +63,7 @@ const corePath = vanillaMode_ ? '../tess/core_vanilla/' : '../tess/core/';
 
 // Custom build is currently only used for browser version, while the Node.js version uses the published npm package.
 // If recognition capabilities are ever added for the Node.js version, then we should use the same build for consistency. .
-const tessConfig = browserMode ? {
+const tessConfig = typeof process === 'undefined' ? {
   corePath,
   workerPath: '../../tess/worker.min.js',
   // langPath: '/tess/tessdata_dist',

scribe.js

@@ -11,9 +11,10 @@ import { ImageCache } from './js/containers/imageContainer.js';
 import coords from './js/coordinates.js';
 import { drawDebugImages } from './js/debug.js';
 import { download, exportData } from './js/export/export.js';
-import { writeDebugCsv, convertToCSV } from './js/export/exportDebugCsv.js';
+import { convertToCSV, writeDebugCsv } from './js/export/exportDebugCsv.js';
 import { extractSingleTableContent } from './js/export/exportWriteTabular.js';
-import { loadBuiltInFontsRaw, enableFontOpt } from './js/fontContainerMain.js';
+import { extractInternalPDFText } from './js/extractPDFText.js';
+import { enableFontOpt, loadBuiltInFontsRaw } from './js/fontContainerMain.js';
 import { gs } from './js/generalWorkerMain.js';
 import { importFiles, importFilesSupp } from './js/import/import.js';
 import { calcBoxOverlap, combineOCRPage } from './js/modifyOCR.js';
@@ -30,7 +31,6 @@ import { imageStrToBlob } from './js/utils/imageUtils.js';
 import { countSubstringOccurrences, getRandomAlphanum, replaceSmartQuotes } from './js/utils/miscUtils.js';
 import { calcConf, mergeOcrWords, splitOcrWord } from './js/utils/ocrUtils.js';
 import { assignParagraphs } from './js/utils/reflowPars.js';
-import { extractInternalPDFText } from './js/extractPDFText.js';
 
 /**
  * Initialize the program and optionally pre-load resources.
@@ -83,7 +83,8 @@ const extractText = async (files, langs = ['eng'], outputFormat = 'txt', options
   const skipRecPDFTextOCR = options?.skipRecPDFTextOCR ?? false;
   init({ ocr: true, font: true });
   await importFiles(files, { extractPDFTextNative: skipRecPDFTextNative, extractPDFTextOCR: skipRecPDFTextOCR });
-  if (!(ImageCache.pdfType === 'text' && skipRecPDFTextNative || ImageCache.pdfType === 'ocr' && skipRecPDFTextOCR)) await recognize({ langs });
+  const skipRecPDF = inputData.pdfMode && (ImageCache.pdfType === 'text' && skipRecPDFTextNative || ImageCache.pdfType === 'ocr' && skipRecPDFTextOCR);
+  if (!skipRecPDF) await recognize({ langs });
   return exportData(outputFormat);
 };