scribe.js-ocr 0.7.2 → 0.7.3

package/cli/cli.js

@@ -1,3 +1,4 @@
+import { detectPDFType } from './detectPDFType.js';
 import { extract } from './extract.js';
 import {
   check,
@@ -36,7 +37,7 @@ export const evalInternalCLI = async (pdfFile, ocrFile, options) => {
  * @param {string} pdfFile - Path to PDF file.
  * @param {?string} [outputDir='.'] - Output directory.
  * @param {Object} [options]
- * @param {'txt'} [options.format]
+ * @param {"pdf" | "hocr" | "docx" | "xlsx" | "txt" | "text" | "html"} [options.format]
  * @param {boolean} [options.reflow]
  */
 export const extractCLI = async (pdfFile, outputDir, options) => {
@@ -44,6 +45,16 @@ export const extractCLI = async (pdfFile, outputDir, options) => {
   process.exitCode = 0;
 };
 
+/**
+ *
+ * @param {string} pdfFile - Path to PDF file.
+ * @param {string} [outputPath] - Output file path.
+ */
+export const detectPDFTypeCLI = async (pdfFile, outputPath) => {
+  await detectPDFType(pdfFile, outputPath);
+  process.exitCode = 0;
+};
+
 /**
  *
  * @param {string} pdfFile - Path to PDF file.

package/cli/detectPDFType.js

@@ -1,20 +1,17 @@
-// Code for adding visualization to OCR output
-
 import fs from 'fs';
-import { createRequire } from 'module';
-import Worker from 'web-worker';
-import { initMuPDFWorker } from '../mupdf/mupdf-async.js';
-
-globalThis.Worker = Worker;
-globalThis.require = createRequire(import.meta.url);
+import scribe from '../scribe.js';
 
-const args = process.argv.slice(2);
+/**
+ *
+ * @param {string} pdfFile - Path to PDF file.
+ * @param {string} [outputPath] - Output file path.
+ *    If provided, the text will be extracted and saved to this path.
+ */
+export const detectPDFType = async (pdfFile, outputPath) => {
+  const mupdfScheduler = await scribe.data.image.getMuPDFScheduler(1);
+  const w = mupdfScheduler.workers[0];
 
-async function main() {
-  const w = await initMuPDFWorker();
-  const fileData = await fs.readFileSync(args[0]);
-
-  const outputPath = args[1];
+  const fileData = await fs.readFileSync(pdfFile);
 
   const pdfDoc = await w.openDocument(fileData, 'file.pdf');
   w.pdfDoc = pdfDoc;
@@ -32,10 +29,6 @@ async function main() {
 
   console.log('PDF Type:', type);
 
-  // Terminate all workers
-  w.terminate();
-
-  process.exitCode = 0;
-}
+  mupdfScheduler.scheduler.terminate();
 
-main();
+};

package/cli/extract.js

@@ -7,7 +7,7 @@ import scribe from '../scribe.js';
  * @param {string} pdfFile - Path to PDF file.
  * @param {?string} [output='.'] - Output file or directory.
  * @param {Object} [options]
- * @param {'txt'} [options.format]
+ * @param {Parameters<typeof scribe.download>[0]} [options.format]
  * @param {boolean} [options.reflow]
  */
 export const extract = async (pdfFile, output, options) => {
@@ -18,7 +18,9 @@ export const extract = async (pdfFile, output, options) => {
   const outputFile = outputDir === output ? `${path.basename(pdfFile).replace(/\.\w{1,5}$/i, `.${format}`)}` : path.basename(output);
   const outputPath = `${outputDir}/${outputFile}`;
 
-  scribe.setOptions({ reflow: true, extractText: true });
+  scribe.opt.reflow = true;
+  scribe.opt.extractText = true;
+
   await scribe.init();
   await scribe.importFiles([pdfFile]);
 

package/cli/scribe.js

@@ -4,6 +4,7 @@ import {
   checkCLI,
   confCLI,
   debugCLI,
+  detectPDFTypeCLI,
   evalInternalCLI, extractCLI, overlayCLI, recognizeCLI,
 } from './cli.js';
 
@@ -35,7 +36,7 @@ program
   .command('extract')
   .argument('<pdf_file>', 'Input PDF file.')
   .argument('[output]', 'Output directory or file to save results.', '.')
-  .addOption(new Option('-f, --format <ext>', 'Output format.').choices(['txt']).default('txt'))
+  .addOption(new Option('-f, --format <ext>', 'Output format.').choices(['pdf', 'hocr', 'docx', 'xlsx', 'txt', 'text', 'html']).default('txt'))
   .option('-r, --reflow', 'Reflow text by combining lines into paragraphs.')
   .description('Extract text from PDF file and save in requested format.')
   .action(extractCLI);
@@ -61,6 +62,13 @@ program
   .option('-w, --workers <number>', 'Number of workers to use. Default is up to 8.')
   .action(recognizeCLI);
 
+program
+  .command('type')
+  .argument('<pdf_file>', 'Input PDF file.')
+  .argument('[output]', 'Output file path to save text.')
+  .description('Detect PDF file type (\'Text native\', \'Image + OCR text\', or \'Image native\').')
+  .action(detectPDFTypeCLI);
+
 program
   .command('debug')
   .argument('<pdf_file>', 'Input PDF file.')

package/js/export/export.js

@@ -207,7 +207,7 @@ export async function exportData(format = 'txt', minPage = 0, maxPage = -1) {
 /**
  * Runs `exportData` and saves the result as a download (browser) or local file (Node.js).
  * @public
- * @param {'pdf'|'hocr'|'docx'|'xlsx'|'txt'|'text'} format
+ * @param {'pdf'|'hocr'|'docx'|'xlsx'|'txt'|'text'|'html'} format
  * @param {string} fileName
  * @param {number} [minPage=0] - First page to export.
  * @param {number} [maxPage=-1] - Last page to export (inclusive). -1 exports through the last page.

package/js/export/writeHtml.js

@@ -5,6 +5,31 @@ import { assignParagraphs } from '../utils/reflowPars.js';
 import { pageMetricsArr } from '../containers/dataContainer.js';
 import ocr from '../objects/ocrObjects.js';
 
+/**
+ * Calculate the font metrics for a given font and font size.
+ * This is used to get metrics that match `ctx.measureText`, but without requiring a canvas.
+ * @param {FontContainerFont} fontI
+ * @param {number} fontSize
+ */
+const calcFontMetrics = (fontI, fontSize) => {
+  const os2 = fontI.opentype.tables.os2;
+  const unitsPerEm = fontI.opentype.unitsPerEm;
+
+  // Bit 7: Use_Typo_Metrics (1 = Yes)
+  // eslint-disable-next-line no-bitwise
+  if (os2.fsSelection >> 7 & 1) {
+    return {
+      fontBoundingBoxAscent: Math.round(os2.sTypoAscender * (fontSize / unitsPerEm)),
+      fontBoundingBoxDescent: Math.round(os2.sTypoDescender * (fontSize / unitsPerEm)),
+    };
+  }
+
+  return {
+    fontBoundingBoxAscent: Math.round(os2.usWinAscent * (fontSize / unitsPerEm)),
+    fontBoundingBoxDescent: Math.round(os2.usWinDescent * (fontSize / unitsPerEm)),
+  };
+};
+
 /**
  *
  * @param {string} text
@@ -33,18 +58,11 @@ const makeSmallCapsDivs = (text, fontSizeHTMLSmallCaps) => {
  *    If omitted, all words are included.
  */
 export function writeHtml(ocrCurrent, minpage = 0, maxpage = -1, reflowText = false, removeMargins = false, wordIds = null) {
-  if (!(typeof process === 'undefined')) {
-    throw new Error('HTML exports are not supported in Node.js');
-  }
-
-  const canvas = new OffscreenCanvas(1, 1);
-  const ctx = /** @type {OffscreenCanvasRenderingContext2D} */ (canvas.getContext('2d'));
-
   const fontsUsed = new Set();
 
   const pad = 5;
 
-  let bodyStr = '<body>';
+  let bodyStr = '<body>\n';
 
   if (maxpage === -1) maxpage = ocrCurrent.length - 1;
 
@@ -71,7 +89,7 @@ export function writeHtml(ocrCurrent, minpage = 0, maxpage = -1, reflowText = fa
       }
     }
 
-    bodyStr += `<div class="scribe-page" id="page${g}" style="position:absolute;top:${top}px;">`;
+    bodyStr += `  <div class="scribe-page" id="page${g}" style="position:absolute;top:${top}px;">\n`;
     if (removeMargins) {
       top += Math.min((maxBottom - minTop) + 200, pageMetricsArr[g].dims.height + 10);
     } else {
@@ -130,9 +148,7 @@ export function writeHtml(ocrCurrent, minpage = 0, maxpage = -1, reflowText = fa
 
         const fontSizeHTML = fontSize * scale;
 
-        ctx.font = `${fontI.fontFaceStyle} ${fontI.fontFaceWeight} ${fontSizeHTML}px ${fontI.fontFaceName}`;
-
-        const metrics = ctx.measureText(wordStr);
+        const metrics = calcFontMetrics(fontI, fontSizeHTML);
 
         const fontSizeHTMLSmallCaps = fontSize * scale * fontI.smallCapsMult;
 
@@ -174,29 +190,29 @@ export function writeHtml(ocrCurrent, minpage = 0, maxpage = -1, reflowText = fa
         // Line height must match the height of the font bounding box for the font metrics to be accurate.
         styleStr += `line-height:${metrics.fontBoundingBoxAscent + metrics.fontBoundingBoxDescent}px;`;
 
-        bodyStr += `<span class="scribe-word" id="${wordObj.id}" style="${styleStr}">${innerHTML}</span>`;
+        bodyStr += `    <span class="scribe-word" id="${wordObj.id}" style="${styleStr}">${innerHTML}</span>`;
       }
     }
 
-    bodyStr += '</div>';
+    bodyStr += '\n  </div>\n';
 
     opt.progressHandler({ n: g, type: 'export', info: { } });
   }
 
-  let styleStr = '<style>.scribe-word {';
+  let styleStr = '<style>\n  .scribe-word {\n';
 
-  styleStr += 'position:absolute;';
-  styleStr += `padding-left:${pad}px;`;
-  styleStr += `padding-right:${pad}px;`;
-  styleStr += 'z-index:1;';
-  styleStr += 'white-space:nowrap;';
+  styleStr += '    position:absolute;\n';
+  styleStr += `    padding-left:${pad}px;\n`;
+  styleStr += `    padding-right:${pad}px;\n`;
+  styleStr += '    z-index:1;\n';
+  styleStr += '    white-space:nowrap;\n';
   if (opt.kerning) {
-    styleStr += 'font-kerning:normal;';
+    styleStr += '    font-kerning:normal;\n';
   } else {
-    styleStr += 'font-kerning:none;';
+    styleStr += '    font-kerning:none;\n';
   }
 
-  styleStr += '}';
+  styleStr += '  }\n';
 
   for (const fontI of fontsUsed) {
     const cdnPath = 'https://cdn.jsdelivr.net/npm/scribe.js-ocr@0.7.1/fonts/all/';
@@ -205,19 +221,19 @@ export function writeHtml(ocrCurrent, minpage = 0, maxpage = -1, reflowText = fa
     const fontName = `${fontI.family}-${styleTitleCase}.woff`;
     const fontPath = cdnPath + fontName;
 
-    styleStr += `@font-face {
+    styleStr += `  @font-face {
     font-family: '${fontI.fontFaceName}';
     font-style: ${fontI.fontFaceStyle};
     font-weight: ${fontI.fontFaceWeight};
     src: url('${fontPath}');
-}\n`;
+  }\n`;
   }
 
-  styleStr += '</style>';
+  styleStr += '</style>\n';
 
-  bodyStr += '</body>';
+  bodyStr += '</body>\n';
 
-  const htmlStr = `<html><head>${styleStr}</head>${bodyStr}</html>`;
+  const htmlStr = `<html>\n<head>\n${styleStr}</head>\n${bodyStr}</html>`;
 
   return htmlStr;
 }

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "scribe.js-ocr",
-  "version": "0.7.2",
+  "version": "0.7.3",
   "description": "High-quality OCR and text extraction for images and PDFs.",
   "main": "scribe.js",
   "directories": {