scribe.js-ocr 0.7.3 → 0.7.4

package/js/export/writeText.js

@@ -54,17 +54,21 @@ export function writeText(ocrCurrent, minpage = 0, maxpage = -1, reflowText = fa
 
         if (docxMode) {
           let fontStyle = '';
-          if (wordObj.style === 'italic') {
+          if (wordObj.style.italic) {
             fontStyle += '<w:i/>';
-          } else if (wordObj.style === 'bold') {
+          } else if (wordObj.style.bold) {
             fontStyle += '<w:b/>';
           }
 
-          if (wordObj.smallCaps) {
+          if (wordObj.style.smallCaps) {
             fontStyle += '<w:smallCaps/>';
           }
 
-          if (wordObj.sup) {
+          if (wordObj.style.underline) {
+            fontStyle += '<w:u w:val="single"/>';
+          }
+
+          if (wordObj.style.sup) {
             fontStyle += '<w:vertAlign w:val="superscript"/>';
           }
 
@@ -79,7 +83,7 @@ export function writeText(ocrCurrent, minpage = 0, maxpage = -1, reflowText = fa
             } else if (supPrev) {
               textStr = `${textStr}</w:t></w:r><w:r>${styleStr}<w:t xml:space="preserve"> `;
             // If this word is a superscript, no space is added between words.
-            } else if (wordObj.sup && i > 0) {
+            } else if (wordObj.style.sup && i > 0) {
               textStr = `${textStr}</w:t></w:r><w:r>${styleStr}<w:t xml:space="preserve">`;
             } else {
               textStr = `${textStr} </w:t></w:r><w:r>${styleStr}<w:t xml:space="preserve">`;
@@ -89,7 +93,7 @@ export function writeText(ocrCurrent, minpage = 0, maxpage = -1, reflowText = fa
           }
 
           fontStylePrev = fontStyle;
-          supPrev = wordObj.sup;
+          supPrev = wordObj.style.sup;
         } else if (newLine) {
           textStr = `${textStr}\n`;
         } else if (h > 0 || g > 0 || i > 0) {

package/js/extractTables.js

@@ -22,11 +22,11 @@ export function extractTableContent(pageObj, layoutObj) {
 
 // TODO: This currently creates junk rows with only punctuation, as those bounding boxes are so small they often do not overlap with other lines.
 /**
-   * Extracts words from a page that are within the bounding boxes of the table, organized into arrays of rows and columns.
-   * The output is in the form of a 3D array, where the first dimension is the row, the second dimension is the column, and the third dimension is the word.
-   * @param {OcrPage} pageObj
-   * @param {Array<import('./objects/layoutObjects.js').LayoutBoxBase>} boxes
-   */
+ * Extracts words from a page that are within the bounding boxes of the table, organized into arrays of rows and columns.
+ * The output is in the form of a 3D array, where the first dimension is the row, the second dimension is the column, and the third dimension is the word.
+ * @param {OcrPage} pageObj
+ * @param {Array<import('./objects/layoutObjects.js').LayoutBoxBase>} boxes
+ */
 export function extractSingleTableContent(pageObj, boxes) {
   /** @type {Array<OcrWord>} */
   const wordArr = [];

package/js/fontContainerMain.js

@@ -24,103 +24,145 @@ export async function loadBuiltInFontsRaw(glyphSet = 'latin') {
   let /** @type {Promise<ArrayBuffer>} */carlitoNormal;
   let /** @type {Promise<ArrayBuffer>} */carlitoItalic;
   let /** @type {Promise<ArrayBuffer>} */carlitoBold;
+  let /** @type {Promise<ArrayBuffer>} */carlitoBoldItalic;
   let /** @type {Promise<ArrayBuffer>} */centuryNormal;
   let /** @type {Promise<ArrayBuffer>} */centuryItalic;
   let /** @type {Promise<ArrayBuffer>} */centuryBold;
+  let /** @type {Promise<ArrayBuffer>} */centuryBoldItalic;
   let /** @type {Promise<ArrayBuffer>} */garamondNormal;
   let /** @type {Promise<ArrayBuffer>} */garamondItalic;
   let /** @type {Promise<ArrayBuffer>} */garamondBold;
+  let /** @type {Promise<ArrayBuffer>} */garamondBoldItalic;
   let /** @type {Promise<ArrayBuffer>} */palatinoNormal;
   let /** @type {Promise<ArrayBuffer>} */palatinoItalic;
   let /** @type {Promise<ArrayBuffer>} */palatinoBold;
+  let /** @type {Promise<ArrayBuffer>} */palatinoBoldItalic;
   let /** @type {Promise<ArrayBuffer>} */nimbusRomanNormal;
   let /** @type {Promise<ArrayBuffer>} */nimbusRomanItalic;
   let /** @type {Promise<ArrayBuffer>} */nimbusRomanBold;
+  let /** @type {Promise<ArrayBuffer>} */nimbusRomanBoldItalic;
   let /** @type {Promise<ArrayBuffer>} */nimbusSansNormal;
   let /** @type {Promise<ArrayBuffer>} */nimbusSansItalic;
   let /** @type {Promise<ArrayBuffer>} */nimbusSansBold;
+  let /** @type {Promise<ArrayBuffer>} */nimbusSansBoldItalic;
   let /** @type {Promise<ArrayBuffer>} */nimbusMonoNormal;
   let /** @type {Promise<ArrayBuffer>} */nimbusMonoItalic;
   let /** @type {Promise<ArrayBuffer>} */nimbusMonoBold;
+  let /** @type {Promise<ArrayBuffer>} */nimbusMonoBoldItalic;
   if (typeof process === 'undefined') {
     if (glyphSet === 'latin') {
       carlitoNormal = fetch(new URL('../fonts/latin/Carlito-Regular.woff', import.meta.url)).then((res) => res.arrayBuffer());
       carlitoItalic = fetch(new URL('../fonts/latin/Carlito-Italic.woff', import.meta.url)).then((res) => res.arrayBuffer());
       carlitoBold = fetch(new URL('../fonts/latin/Carlito-Bold.woff', import.meta.url)).then((res) => res.arrayBuffer());
+      carlitoBoldItalic = fetch(new URL('../fonts/latin/Carlito-BoldItalic.woff', import.meta.url)).then((res) => res.arrayBuffer());
       centuryNormal = fetch(new URL('../fonts/latin/Century-Regular.woff', import.meta.url)).then((res) => res.arrayBuffer());
       centuryItalic = fetch(new URL('../fonts/latin/Century-Italic.woff', import.meta.url)).then((res) => res.arrayBuffer());
       centuryBold = fetch(new URL('../fonts/latin/Century-Bold.woff', import.meta.url)).then((res) => res.arrayBuffer());
+      centuryBoldItalic = fetch(new URL('../fonts/latin/Century-BoldItalic.woff', import.meta.url)).then((res) => res.arrayBuffer());
       garamondNormal = fetch(new URL('../fonts/latin/Garamond-Regular.woff', import.meta.url)).then((res) => res.arrayBuffer());
       garamondItalic = fetch(new URL('../fonts/latin/Garamond-Italic.woff', import.meta.url)).then((res) => res.arrayBuffer());
       garamondBold = fetch(new URL('../fonts/latin/Garamond-Bold.woff', import.meta.url)).then((res) => res.arrayBuffer());
+      garamondBoldItalic = fetch(new URL('../fonts/latin/Garamond-BoldItalic.woff', import.meta.url)).then((res) => res.arrayBuffer());
       palatinoNormal = fetch(new URL('../fonts/latin/Palatino-Regular.woff', import.meta.url)).then((res) => res.arrayBuffer());
       palatinoItalic = fetch(new URL('../fonts/latin/Palatino-Italic.woff', import.meta.url)).then((res) => res.arrayBuffer());
       palatinoBold = fetch(new URL('../fonts/latin/Palatino-Bold.woff', import.meta.url)).then((res) => res.arrayBuffer());
+      palatinoBoldItalic = fetch(new URL('../fonts/latin/Palatino-BoldItalic.woff', import.meta.url)).then((res) => res.arrayBuffer());
       nimbusRomanNormal = fetch(new URL('../fonts/latin/NimbusRoman-Regular.woff', import.meta.url)).then((res) => res.arrayBuffer());
       nimbusRomanItalic = fetch(new URL('../fonts/latin/NimbusRoman-Italic.woff', import.meta.url)).then((res) => res.arrayBuffer());
       nimbusRomanBold = fetch(new URL('../fonts/latin/NimbusRoman-Bold.woff', import.meta.url)).then((res) => res.arrayBuffer());
+      nimbusRomanBoldItalic = fetch(new URL('../fonts/latin/NimbusRoman-BoldItalic.woff', import.meta.url)).then((res) => res.arrayBuffer());
       nimbusSansNormal = fetch(new URL('../fonts/latin/NimbusSans-Regular.woff', import.meta.url)).then((res) => res.arrayBuffer());
       nimbusSansItalic = fetch(new URL('../fonts/latin/NimbusSans-Italic.woff', import.meta.url)).then((res) => res.arrayBuffer());
       nimbusSansBold = fetch(new URL('../fonts/latin/NimbusSans-Bold.woff', import.meta.url)).then((res) => res.arrayBuffer());
+      nimbusSansBoldItalic = fetch(new URL('../fonts/latin/NimbusSans-BoldItalic.woff', import.meta.url)).then((res) => res.arrayBuffer());
       nimbusMonoNormal = fetch(new URL('../fonts/latin/NimbusMono-Regular.woff', import.meta.url)).then((res) => res.arrayBuffer());
       nimbusMonoItalic = fetch(new URL('../fonts/latin/NimbusMono-Italic.woff', import.meta.url)).then((res) => res.arrayBuffer());
       nimbusMonoBold = fetch(new URL('../fonts/latin/NimbusMono-Bold.woff', import.meta.url)).then((res) => res.arrayBuffer());
+      nimbusMonoBoldItalic = fetch(new URL('../fonts/latin/NimbusMono-BoldItalic.woff', import.meta.url)).then((res) => res.arrayBuffer());
     } else {
       carlitoNormal = fetch(new URL('../fonts/all/Carlito-Regular.woff', import.meta.url)).then((res) => res.arrayBuffer());
       carlitoItalic = fetch(new URL('../fonts/all/Carlito-Italic.woff', import.meta.url)).then((res) => res.arrayBuffer());
       carlitoBold = fetch(new URL('../fonts/all/Carlito-Bold.woff', import.meta.url)).then((res) => res.arrayBuffer());
+      carlitoBoldItalic = fetch(new URL('../fonts/all/Carlito-BoldItalic.woff', import.meta.url)).then((res) => res.arrayBuffer());
       centuryNormal = fetch(new URL('../fonts/all/Century-Regular.woff', import.meta.url)).then((res) => res.arrayBuffer());
       centuryItalic = fetch(new URL('../fonts/all/Century-Italic.woff', import.meta.url)).then((res) => res.arrayBuffer());
       centuryBold = fetch(new URL('../fonts/all/Century-Bold.woff', import.meta.url)).then((res) => res.arrayBuffer());
+      centuryBoldItalic = fetch(new URL('../fonts/all/Century-BoldItalic.woff', import.meta.url)).then((res) => res.arrayBuffer());
       garamondNormal = fetch(new URL('../fonts/all/Garamond-Regular.woff', import.meta.url)).then((res) => res.arrayBuffer());
       garamondItalic = fetch(new URL('../fonts/all/Garamond-Italic.woff', import.meta.url)).then((res) => res.arrayBuffer());
       garamondBold = fetch(new URL('../fonts/all/Garamond-Bold.woff', import.meta.url)).then((res) => res.arrayBuffer());
+      garamondBoldItalic = fetch(new URL('../fonts/all/Garamond-BoldItalic.woff', import.meta.url)).then((res) => res.arrayBuffer());
       palatinoNormal = fetch(new URL('../fonts/all/Palatino-Regular.woff', import.meta.url)).then((res) => res.arrayBuffer());
       palatinoItalic = fetch(new URL('../fonts/all/Palatino-Italic.woff', import.meta.url)).then((res) => res.arrayBuffer());
       palatinoBold = fetch(new URL('../fonts/all/Palatino-Bold.woff', import.meta.url)).then((res) => res.arrayBuffer());
+      palatinoBoldItalic = fetch(new URL('../fonts/all/Palatino-BoldItalic.woff', import.meta.url)).then((res) => res.arrayBuffer());
       nimbusRomanNormal = fetch(new URL('../fonts/all/NimbusRoman-Regular.woff', import.meta.url)).then((res) => res.arrayBuffer());
       nimbusRomanItalic = fetch(new URL('../fonts/all/NimbusRoman-Italic.woff', import.meta.url)).then((res) => res.arrayBuffer());
       nimbusRomanBold = fetch(new URL('../fonts/all/NimbusRoman-Bold.woff', import.meta.url)).then((res) => res.arrayBuffer());
+      nimbusRomanBoldItalic = fetch(new URL('../fonts/all/NimbusRoman-BoldItalic.woff', import.meta.url)).then((res) => res.arrayBuffer());
       nimbusSansNormal = fetch(new URL('../fonts/all/NimbusSans-Regular.woff', import.meta.url)).then((res) => res.arrayBuffer());
       nimbusSansItalic = fetch(new URL('../fonts/all/NimbusSans-Italic.woff', import.meta.url)).then((res) => res.arrayBuffer());
       nimbusSansBold = fetch(new URL('../fonts/all/NimbusSans-Bold.woff', import.meta.url)).then((res) => res.arrayBuffer());
+      nimbusSansBoldItalic = fetch(new URL('../fonts/all/NimbusSans-BoldItalic.woff', import.meta.url)).then((res) => res.arrayBuffer());
       nimbusMonoNormal = fetch(new URL('../fonts/all/NimbusMono-Regular.woff', import.meta.url)).then((res) => res.arrayBuffer());
       nimbusMonoItalic = fetch(new URL('../fonts/all/NimbusMono-Italic.woff', import.meta.url)).then((res) => res.arrayBuffer());
       nimbusMonoBold = fetch(new URL('../fonts/all/NimbusMono-Bold.woff', import.meta.url)).then((res) => res.arrayBuffer());
+      nimbusMonoBoldItalic = fetch(new URL('../fonts/all/NimbusMono-BoldItalic.woff', import.meta.url)).then((res) => res.arrayBuffer());
     }
   } else {
     const { readFile } = await import('fs/promises');
     carlitoNormal = readFile(new URL('../fonts/all/Carlito-Regular.woff', import.meta.url)).then((res) => res.buffer);
     carlitoItalic = readFile(new URL('../fonts/all/Carlito-Italic.woff', import.meta.url)).then((res) => res.buffer);
     carlitoBold = readFile(new URL('../fonts/all/Carlito-Bold.woff', import.meta.url)).then((res) => res.buffer);
+    carlitoBoldItalic = readFile(new URL('../fonts/all/Carlito-BoldItalic.woff', import.meta.url)).then((res) => res.buffer);
     centuryNormal = readFile(new URL('../fonts/all/Century-Regular.woff', import.meta.url)).then((res) => res.buffer);
     centuryItalic = readFile(new URL('../fonts/all/Century-Italic.woff', import.meta.url)).then((res) => res.buffer);
     centuryBold = readFile(new URL('../fonts/all/Century-Bold.woff', import.meta.url)).then((res) => res.buffer);
+    centuryBoldItalic = readFile(new URL('../fonts/all/Century-BoldItalic.woff', import.meta.url)).then((res) => res.buffer);
     garamondNormal = readFile(new URL('../fonts/all/Garamond-Regular.woff', import.meta.url)).then((res) => res.buffer);
     garamondItalic = readFile(new URL('../fonts/all/Garamond-Italic.woff', import.meta.url)).then((res) => res.buffer);
     garamondBold = readFile(new URL('../fonts/all/Garamond-Bold.woff', import.meta.url)).then((res) => res.buffer);
+    garamondBoldItalic = readFile(new URL('../fonts/all/Garamond-BoldItalic.woff', import.meta.url)).then((res) => res.buffer);
     palatinoNormal = readFile(new URL('../fonts/all/Palatino-Regular.woff', import.meta.url)).then((res) => res.buffer);
     palatinoItalic = readFile(new URL('../fonts/all/Palatino-Italic.woff', import.meta.url)).then((res) => res.buffer);
     palatinoBold = readFile(new URL('../fonts/all/Palatino-Bold.woff', import.meta.url)).then((res) => res.buffer);
+    palatinoBoldItalic = readFile(new URL('../fonts/all/Palatino-BoldItalic.woff', import.meta.url)).then((res) => res.buffer);
     nimbusRomanNormal = readFile(new URL('../fonts/all/NimbusRoman-Regular.woff', import.meta.url)).then((res) => res.buffer);
     nimbusRomanItalic = readFile(new URL('../fonts/all/NimbusRoman-Italic.woff', import.meta.url)).then((res) => res.buffer);
     nimbusRomanBold = readFile(new URL('../fonts/all/NimbusRoman-Bold.woff', import.meta.url)).then((res) => res.buffer);
+    nimbusRomanBoldItalic = readFile(new URL('../fonts/all/NimbusRoman-BoldItalic.woff', import.meta.url)).then((res) => res.buffer);
     nimbusSansNormal = readFile(new URL('../fonts/all/NimbusSans-Regular.woff', import.meta.url)).then((res) => res.buffer);
     nimbusSansItalic = readFile(new URL('../fonts/all/NimbusSans-Italic.woff', import.meta.url)).then((res) => res.buffer);
     nimbusSansBold = readFile(new URL('../fonts/all/NimbusSans-Bold.woff', import.meta.url)).then((res) => res.buffer);
+    nimbusSansBoldItalic = readFile(new URL('../fonts/all/NimbusSans-BoldItalic.woff', import.meta.url)).then((res) => res.buffer);
     nimbusMonoNormal = readFile(new URL('../fonts/all/NimbusMono-Regular.woff', import.meta.url)).then((res) => res.buffer);
     nimbusMonoItalic = readFile(new URL('../fonts/all/NimbusMono-Italic.woff', import.meta.url)).then((res) => res.buffer);
     nimbusMonoBold = readFile(new URL('../fonts/all/NimbusMono-Bold.woff', import.meta.url)).then((res) => res.buffer);
+    nimbusMonoBoldItalic = readFile(new URL('../fonts/all/NimbusMono-BoldItalic.woff', import.meta.url)).then((res) => res.buffer);
   }
 
   const srcObj = {
-    Carlito: { normal: await carlitoNormal, italic: await carlitoItalic, bold: await carlitoBold },
-    Century: { normal: await centuryNormal, italic: await centuryItalic, bold: await centuryBold },
-    Garamond: { normal: await garamondNormal, italic: await garamondItalic, bold: await garamondBold },
-    Palatino: { normal: await palatinoNormal, italic: await palatinoItalic, bold: await palatinoBold },
-    NimbusRoman: { normal: await nimbusRomanNormal, italic: await nimbusRomanItalic, bold: await nimbusRomanBold },
-    NimbusSans: { normal: await nimbusSansNormal, italic: await nimbusSansItalic, bold: await nimbusSansBold },
-    NimbusMono: { normal: await nimbusMonoNormal, italic: await nimbusMonoItalic, bold: await nimbusMonoBold },
+    Carlito: {
+      normal: await carlitoNormal, italic: await carlitoItalic, bold: await carlitoBold, boldItalic: await carlitoBoldItalic,
+    },
+    Century: {
+      normal: await centuryNormal, italic: await centuryItalic, bold: await centuryBold, boldItalic: await centuryBoldItalic,
+    },
+    Garamond: {
+      normal: await garamondNormal, italic: await garamondItalic, bold: await garamondBold, boldItalic: await garamondBoldItalic,
+    },
+    Palatino: {
+      normal: await palatinoNormal, italic: await palatinoItalic, bold: await palatinoBold, boldItalic: await palatinoBoldItalic,
+    },
+    NimbusRoman: {
+      normal: await nimbusRomanNormal, italic: await nimbusRomanItalic, bold: await nimbusRomanBold, boldItalic: await nimbusRomanBoldItalic,
+    },
+    NimbusSans: {
+      normal: await nimbusSansNormal, italic: await nimbusSansItalic, bold: await nimbusSansBold, boldItalic: await nimbusSansBoldItalic,
+    },
+    NimbusMono: {
+      normal: await nimbusMonoNormal, italic: await nimbusMonoItalic, bold: await nimbusMonoBold, boldItalic: await nimbusMonoBoldItalic,
+    },
   };
 
   FontCont.raw = await /** @type {FontContainer} */(/** @type {any} */(loadFontsFromSource(srcObj)));
@@ -217,6 +259,7 @@ export async function updateFontContWorkerMain(params = {}) {
       };
       if (value.italic) input.src[key].italic = value.italic.src;
       if (value.bold) input.src[key].bold = value.bold.src;
+      if (value.boldItalic) input.src[key].boldItalic = value.boldItalic.src;
     }
 
     for (let i = 0; i < gs.schedulerInner.workers.length; i++) {

package/js/fontStatistics.js

@@ -3,6 +3,7 @@
 
 import {
   determineSansSerif,
+  getStyleLookup,
   quantile,
   replaceObjectProperties,
   round6,
@@ -243,13 +244,13 @@ function calcFontMetricsPage(pageObj) {
 
   for (const lineObj of pageObj.lines) {
     for (const wordObj of lineObj.words) {
-      const wordFontFamily = determineSansSerif(wordObj.font) || 'Default';
+      const wordFontFamily = determineSansSerif(wordObj.style.font) || 'Default';
 
       // This condition should not occur, however has in the past due to parsing bugs.  Skipping to avoid entire program crashing if this occurs.
       if (wordObj.chars && wordObj.chars.length !== wordObj.text.length) continue;
 
       // Do not include superscripts, dropcaps, and low-confidence words in statistics for font optimization.
-      if (wordObj.conf < 80 || wordObj.lang === 'chi_sim' || wordObj.sup || wordObj.smallCaps) continue;
+      if (wordObj.conf < 80 || wordObj.lang === 'chi_sim' || wordObj.style.sup || wordObj.style.smallCaps) continue;
       /** @type {Object.<string, FontMetricsRawFamily>} */
       const fontMetricsRawLine = {};
 
@@ -275,14 +276,18 @@ function calcFontMetricsPage(pageObj) {
             fontMetricsRawLine[wordFontFamily] = new FontMetricsRawFamily();
           }
 
-          if (!fontMetricsRawLine[wordFontFamily][wordObj.style].width[charUnicode]) {
-            fontMetricsRawLine[wordFontFamily][wordObj.style].width[charUnicode] = [];
-            fontMetricsRawLine[wordFontFamily][wordObj.style].height[charUnicode] = [];
+          const styleLookup = getStyleLookup(wordObj.style);
+
+          if (!['normal', 'italic', 'bold'].includes(styleLookup)) continue;
+
+          if (!fontMetricsRawLine[wordFontFamily][styleLookup].width[charUnicode]) {
+            fontMetricsRawLine[wordFontFamily][styleLookup].width[charUnicode] = [];
+            fontMetricsRawLine[wordFontFamily][styleLookup].height[charUnicode] = [];
           }
 
-          fontMetricsRawLine[wordFontFamily][wordObj.style].width[charUnicode].push(charWidth / charNorm);
-          fontMetricsRawLine[wordFontFamily][wordObj.style].height[charUnicode].push(charHeight / charNorm);
-          fontMetricsRawLine[wordFontFamily][wordObj.style].obs += 1;
+          fontMetricsRawLine[wordFontFamily][styleLookup].width[charUnicode].push(charWidth / charNorm);
+          fontMetricsRawLine[wordFontFamily][styleLookup].height[charUnicode].push(charHeight / charNorm);
+          fontMetricsRawLine[wordFontFamily][styleLookup].obs += 1;
 
           if (k + 1 < wordObj.chars.length) {
             const charObjNext = wordObj.chars[k + 1];
@@ -295,12 +300,12 @@ function calcFontMetricsPage(pageObj) {
             if (trailingSpace + charWidthNext > 0) {
               const bigramUnicode = `${charUnicode},${wordObj.chars[k + 1].text.charCodeAt(0)}`;
 
-              if (!fontMetricsRawLine[wordFontFamily][wordObj.style].kerning[bigramUnicode]) {
-                fontMetricsRawLine[wordFontFamily][wordObj.style].kerning[bigramUnicode] = [];
-                fontMetricsRawLine[wordFontFamily][wordObj.style].kerning2[bigramUnicode] = [];
+              if (!fontMetricsRawLine[wordFontFamily][styleLookup].kerning[bigramUnicode]) {
+                fontMetricsRawLine[wordFontFamily][styleLookup].kerning[bigramUnicode] = [];
+                fontMetricsRawLine[wordFontFamily][styleLookup].kerning2[bigramUnicode] = [];
               }
-              fontMetricsRawLine[wordFontFamily][wordObj.style].kerning[bigramUnicode].push(trailingSpace / charNorm);
-              fontMetricsRawLine[wordFontFamily][wordObj.style].kerning2[bigramUnicode].push((trailingSpace + charWidthNext) / charNorm);
+              fontMetricsRawLine[wordFontFamily][styleLookup].kerning[bigramUnicode].push(trailingSpace / charNorm);
+              fontMetricsRawLine[wordFontFamily][styleLookup].kerning2[bigramUnicode].push((trailingSpace + charWidthNext) / charNorm);
             }
           }
         }

package/js/fontSupp.js

@@ -44,7 +44,7 @@ const calcSuppFontInfoForWords = async (words) => {
   const fontSizeArr = [];
   for (const word of wordsRes) {
     fontSizeArr.push(calcWordFontSize(word));
-    const sansSerif = determineSansSerif(word.font);
+    const sansSerif = determineSansSerif(word.style.font);
     if (sansSerif !== 'Default') {
       if (sansSerif === 'SansDefault') {
         sansVotes++;
@@ -53,9 +53,9 @@ const calcSuppFontInfoForWords = async (words) => {
       }
     }
   }
-  if (words[0].size) {
-    // @ts-ignore
-    fontSizeMult = quantile(fontSizeArr, 0.5) / words[0].size;
+
+  if (words[0].style.size) {
+    fontSizeMult = quantile(fontSizeArr, 0.5) / words[0].style.size;
   }
 
   return { sansVotes, serifVotes, fontSizeMult };
@@ -83,36 +83,36 @@ export const calcSuppFontInfo = async (ocrArr) => {
       let wordFontLast;
       let wordFontSizeLast;
       for (const word of line.words) {
-        if (word.font) {
-          if (skipFonts.has(word.font)) {
+        if (word.style.font) {
+          if (skipFonts.has(word.style.font)) {
             continue;
           // Printing words off screen is a common method of hiding text in PDFs.
           } else if (word.bbox.left < 0 || word.bbox.top < 0 || word.bbox.right > page.dims.width || word.bbox.bottom > page.dims.height) {
             continue;
-          } else if (!calcFonts.has(word.font)) {
-            const sansSerifUnknown = determineSansSerif(word.font) === 'Default';
+          } else if (!calcFonts.has(word.style.font)) {
+            const sansSerifUnknown = determineSansSerif(word.style.font) === 'Default';
             if (sansSerifUnknown || !word.visualCoords) {
-              calcFonts.add(word.font);
+              calcFonts.add(word.style.font);
             } else {
-              skipFonts.add(word.font);
+              skipFonts.add(word.style.font);
               continue;
             }
           }
 
-          if (!fontExamples[word.font]) {
-            fontExamples[word.font] = [];
-          } else if (fontExamples[word.font].length > 3) {
+          if (!fontExamples[word.style.font]) {
+            fontExamples[word.style.font] = [];
+          } else if (fontExamples[word.style.font].length > 3) {
             continue;
           }
 
-          if (word.font !== wordFontLast || word.size !== wordFontSizeLast) {
-            fontExamples[word.font].push([word]);
+          if (word.style.font !== wordFontLast || word.style.size !== wordFontSizeLast) {
+            fontExamples[word.style.font].push([word]);
           } else {
-            fontExamples[word.font][fontExamples[word.font].length - 1].push(word);
+            fontExamples[word.style.font][fontExamples[word.style.font].length - 1].push(word);
           }
 
-          wordFontLast = word.font;
-          wordFontSizeLast = word.size;
+          wordFontLast = word.style.font;
+          wordFontSizeLast = word.style.size;
         }
       }
     }
@@ -158,8 +158,8 @@ export const calcSuppFontInfo = async (ocrArr) => {
   for (const page of ocrArr) {
     for (const line of page.lines) {
       for (const word of line.words) {
-        if (word.font && word.size && FontProps.sizeMult[word.font]) {
-          word.size = Math.round(word.size * FontProps.sizeMult[word.font] * 1000) / 1000;
+        if (word.style.font && word.style.size && FontProps.sizeMult[word.style.font]) {
+          word.style.size = Math.round(word.style.size * FontProps.sizeMult[word.style.font] * 1000) / 1000;
         }
       }
     }

package/js/global.d.ts

@@ -1,5 +1,18 @@
 declare global {
 
+    type Style = {
+        font: ?string;
+        size: ?number;
+        bold: boolean;
+        italic: boolean;
+        underline: boolean;
+        smallCaps: boolean;
+        sup: boolean;
+        dropcap: boolean;
+    };
+
+    type StyleLookup = ('normal'|'bold'|'italic'|'boldItalic');
+      
     // OCR objects
     type OcrPage = import("./objects/ocrObjects.js").OcrPage;
     type OcrLine = import("./objects/ocrObjects.js").OcrLine;
@@ -17,12 +30,14 @@ declare global {
         normal: FontContainerFont;
         italic: FontContainerFont;
         bold: FontContainerFont;
+        boldItalic: FontContainerFont;
     };
 
     type FontContainerFamilyUpload = {
         normal: FontContainerFont | null;
         italic: FontContainerFont | null;
         bold: FontContainerFont | null;
+        boldItalic: FontContainerFont | null;
     };
 
     type FontContainerFamily = FontContainerFamilyBuiltIn | FontContainerFamilyUpload;
@@ -42,12 +57,14 @@ declare global {
         normal: ArrayBuffer;
         italic: ArrayBuffer;
         bold: ArrayBuffer;
+        boldItalic: ArrayBuffer;
     };
 
     type fontSrcUpload = {
         normal: ArrayBuffer | null;
         italic: ArrayBuffer | null;
         bold: ArrayBuffer | null;
+        boldItalic: ArrayBuffer | null;
     };
 
     type opentypeFont = import("../lib/opentype.module.js").Font;

package/js/import/convertPageAbbyy.js

@@ -171,8 +171,18 @@ export async function convertPageAbbyy({ ocrStr, n }) {
         /** @type {Array<Array<OcrChar>>} */
         const charObjArrLine = Array(wordStrArr.length);
         text = text.fill('');
-        let styleArr = Array(wordStrArr.length);
-        styleArr = styleArr.fill('normal');
+
+        /** @type {Array<boolean>} */
+        const italicArr = Array(wordStrArr.length).fill(false);
+        /** @type {Array<boolean>} */
+        const boldArr = Array(wordStrArr.length).fill(false);
+        /** @type {Array<boolean>} */
+        const underlineArr = Array(wordStrArr.length).fill(false);
+        /** @type {Array<boolean>} */
+        const supArr = Array(wordStrArr.length).fill(false);
+        /** @type {Array<boolean>} */
+        const dropcapArr = Array(wordStrArr.length).fill(false);
+
         /** @type {Array<boolean>} */
         const smallCapsArr = Array(wordStrArr.length).fill(false);
         /** @type {Array<boolean>} */
@@ -184,28 +194,32 @@ export async function convertPageAbbyy({ ocrStr, n }) {
 
           if (typeof (letterArr[0][1]) !== 'undefined') {
             if (dropCap && i === 0) {
-              styleArr[i] = 'dropcap';
+              dropcapArr[i] = true;
             } else if (/superscript=['"](1|true)/i.test(letterArr[0][1])) {
-              styleArr[i] = 'sup';
-            } else if (/italic=['"](1|true)/i.test(letterArr[0][1])) {
-              styleArr[i] = 'italic';
-              stylesLine.italic = true;
+              supArr[i] = true;
             } else {
-              styleArr[i] = 'normal';
-              stylesLine.normal = true;
+              if (/italic=['"](1|true)/i.test(letterArr[0][1])) {
+                italicArr[i] = true;
+              }
+
+              if (/bold=['"](1|true)/i.test(letterArr[0][1])) {
+                boldArr[i] = true;
+              }
+
+              if (/underline=['"](1|true)/i.test(letterArr[0][1])) {
+                underlineArr[i] = true;
+              }
             }
 
             if (/smallcaps=['"](1|true)/i.test(letterArr[0][1])) {
               smallCapsArr[i] = true;
             }
-          } else if (i > 0) {
-            if (styleArr[i - 1] === 'dropcap') {
-              styleArr[i] = 'normal';
-              smallCapsArr[i] = false;
-            } else {
-              styleArr[i] = styleArr[i - 1];
-              smallCapsArr[i] = smallCapsArr[i - 1];
-            }
+          } else if (i > 0 && !dropcapArr[i - 1]) {
+            italicArr[i] = italicArr[i - 1];
+            boldArr[i] = boldArr[i - 1];
+            underlineArr[i] = underlineArr[i - 1];
+            supArr[i] = supArr[i - 1];
+            smallCapsArr[i] = smallCapsArr[i - 1];
           }
 
           // Abbyy will sometimes misidentify capital letters immediately following drop caps as small caps,
@@ -372,18 +386,26 @@ export async function convertPageAbbyy({ ocrStr, n }) {
 
           console.assert(wordObj.chars.length === text[i].length, `Likely parsing error for word: ${id}. Number of letters in text does not match number of \`ocrChar\` objects.`);
 
-          if (styleArr[i] === 'italic') {
-            wordObj.style = 'italic';
+          if (italicArr[i]) {
+            wordObj.style.italic = true;
+          }
+
+          if (boldArr[i]) {
+            wordObj.style.bold = true;
+          }
+
+          if (underlineArr[i]) {
+            wordObj.style.underline = true;
           }
 
-          wordObj.smallCaps = smallCapsArr[i];
+          wordObj.style.smallCaps = smallCapsArr[i];
 
-          if (fontName) wordObj.font = fontName;
+          if (fontName) wordObj.style.font = fontName;
 
-          if (styleArr[i] === 'sup') {
-            wordObj.sup = true;
-          } else if (styleArr[i] === 'dropcap') {
-            wordObj.dropcap = true;
+          if (supArr[i]) {
+            wordObj.style.sup = true;
+          } else if (dropcapArr[i]) {
+            wordObj.style.dropcap = true;
           }
 
           lineObj.words.push(wordObj);

package/js/import/convertPageBlocks.js

@@ -140,11 +140,11 @@ export async function convertPageBlocks({
           // The `word` object has a `is_italic` property, but it is always false.
           // Therefore, the font name is checked to determine if the word is italic.
           // See: https://github.com/naptha/tesseract.js/issues/907
-          if (keepItalic && /italic/i.test(word.font_name)) wordObj.style = 'italic';
+          if (keepItalic && /italic/i.test(word.font_name)) wordObj.style.italic = true;
 
           // Our fork of Tesseract Legacy should be able to recognize fonts, so this information is included.
           // The generic HOCR importer does not include font information, as this is assumed to be unreliable.
-          wordObj.font = word.font_name;
+          wordObj.style.font = word.font_name;
 
           wordObj.chars = [];
           for (let m = 0; m < word.symbols.length; m++) {

package/js/import/convertPageHocr.js

@@ -247,8 +247,8 @@ export async function convertPageHocr({
 
       if (debugMode) wordObj.raw = match;
 
-      if (italic) wordObj.style = 'italic';
-      if (fontName) wordObj.font = fontName;
+      if (italic) wordObj.style.italic = true;
+      if (fontName) wordObj.style.font = fontName;
 
       wordObj.conf = wordConf;
 
@@ -302,19 +302,6 @@ export async function convertPageHocr({
 
       const styleStr = match.match(/style=['"]([^'"]+)/)?.[1];
 
-      let smallCaps = false;
-      /** @type {('normal'|'italic'|'bold')} */
-      let fontStyle = 'normal';
-      if (styleStr && /italic/i.test(styleStr)) {
-        fontStyle = 'italic';
-      } else if (styleStr && /bold/i.test(styleStr)) {
-        fontStyle = 'bold';
-      }
-
-      if (styleStr && /small-caps/i.test(styleStr)) {
-        smallCaps = true;
-      }
-
       const confMatch = titleStrWord.match(/(?:;|\s)x_wconf\s+(\d+)/)?.[1] || '0';
       const wordConf = parseInt(confMatch) || 0;
 
@@ -327,16 +314,19 @@ export async function convertPageHocr({
         const wordFontSizeStr = titleStrWord.match(/(?:;|\s)x_fsize\s+(\d+)/)?.[1];
         if (wordFontSizeStr) {
           const wordFontSize = parseInt(wordFontSizeStr);
-          if (wordFontSize) wordObj.size = wordFontSize;
+          if (wordFontSize) wordObj.style.size = wordFontSize;
         }
       }
 
-      wordObj.style = fontStyle;
-      if (fontName) wordObj.font = fontName;
+      if (styleStr) {
+        if (/italic/i.test(styleStr)) wordObj.style.italic = true;
+        if (/bold/i.test(styleStr)) wordObj.style.bold = true;
+        if (/small-caps/i.test(styleStr)) wordObj.style.smallCaps = true;
+      }
 
-      wordObj.sup = wordSup;
+      if (wordSup) wordObj.style.sup = true;
 
-      wordObj.smallCaps = smallCaps;
+      if (fontName) wordObj.style.font = fontName;
 
       wordObj.conf = wordConf;