@chialab/pdfjs-lib 1.0.0-alpha.44 → 1.0.0-alpha.46

package/dist/browser/index.js

@@ -29287,6 +29287,25 @@ var flattenNodes = (node) => {
 var isNameObject = (obj) => {
   return obj !== null && typeof obj === "object" && "name" in obj;
 };
+var isLigature = (char) => {
+  if (!char || [...char].length !== 1) {
+    return false;
+  }
+  const code = char.codePointAt(0);
+  if (code == null) {
+    return false;
+  }
+  return (
+    // Alphabetic Presentation Forms
+    code >= 64256 && code <= 64335 || // Arabic Presentation Forms-A
+    code >= 64336 && code <= 65023 || // Arabic Presentation Forms-B
+    code >= 65136 && code <= 65279 || // Latin ligatures outside FB00 block
+    code === 338 || // Œ
+    code === 339 || // œ
+    code === 306 || // IJ
+    code === 307
+  );
+};
 var renderTextLayer = (root, options = {}) => {
   const { classes = false, styles = false, font } = options;
   const serializeAttributes = (attrs) => {
@@ -29321,7 +29340,7 @@ var renderTextLayer = (root, options = {}) => {
           "--tl-margin": `${margin}px`,
           "--tl-transform": scale !== 1 ? `scaleX(${scale})` : null
         }) : "";
-        return `<${tag3}${serializedAttrs3 ? ` ${serializedAttrs3}` : ""}${serializedStyle2 ? ` style="${serializedStyle2}` : ""}">${chunk.text}</${tag3}>`;
+        return `<${tag3}${serializedAttrs3 ? ` ${serializedAttrs3}` : ""}${serializedStyle2 ? ` style="${serializedStyle2}"` : ""}>${chunk.text}</${tag3}>`;
       }).join("");
       if (!classes && !styles) {
         return contents;
@@ -29830,42 +29849,16 @@ async function createTextLayer(page, {
     }
     if (oldTextItem?.glyphs.length) {
       if (newTextItem) {
-        const text = oldTextItem.glyphs.map((g) => g[0].unicode).join("") || "";
-        const dir = bidi(text).dir;
-        const firstGlyph = oldTextItem.glyphs[0];
-        const lastGlyph = oldTextItem.glyphs.at(-1);
-        const scale = oldTextItem.scaleX * oldTextItem.fontSize;
         const top = oldTextItem.top;
         const bottom = oldTextItem.top + oldTextItem.fontSize;
-        const left = oldTextItem.left;
-        const right = oldTextItem.left + (lastGlyph[1] + lastGlyph[2]) * scale;
         const nextTop = newTextItem.top;
         const nextBottom = newTextItem.top + newTextItem.fontSize;
-        switch (dir) {
-          case "ltr": {
-            if (newTextItem.left - right < lastGlyph[2] * scale) {
-              if (nextTop > top && nextTop < bottom || nextBottom > top && nextBottom < bottom) {
-                return;
-              }
-              if (checkNewLine(oldTextItem, newTextItem, true)) {
-                appendNewLineAsNeeded();
-                return;
-              }
-            }
-            break;
-          }
-          case "rtl": {
-            if (left - newTextItem.left < firstGlyph[2] * scale) {
-              if (nextTop > top && nextTop < bottom || nextBottom > top && nextBottom < bottom) {
-                return;
-              }
-              if (checkNewLine(oldTextItem, newTextItem, true)) {
-                appendNewLineAsNeeded();
-                return;
-              }
-            }
-            break;
-          }
+        if (nextTop > top && nextTop < bottom || nextBottom > top && nextBottom < bottom || nextTop <= top && nextBottom >= bottom) {
+          return;
+        }
+        if (checkNewLine(oldTextItem, newTextItem, true)) {
+          appendNewLineAsNeeded();
+          return;
         }
       }
       closeMarkedContent();
@@ -29906,6 +29899,44 @@ async function createTextLayer(page, {
       });
     }
   };
+  const addGlyph = (textItem, char) => {
+    const width = (char.width ?? 0) * fontSize / 1e3 || 0;
+    const charUnicode = char.unicode ?? "";
+    if (isLigature(charUnicode)) {
+      const normalized = charUnicode.normalize("NFKD");
+      const expansion = Array.from(normalized);
+      const widthPerChar = width / expansion.length;
+      for (const expandedChar of expansion) {
+        const syntheticGlyph = {
+          unicode: expandedChar,
+          isSpace: false
+        };
+        textItem.glyphs.push([syntheticGlyph, textPosition, widthPerChar]);
+        textPosition += widthPerChar + charSpacing * hScale;
+      }
+      return expansion.length;
+    }
+    textItem.glyphs.push([char, textPosition, width]);
+    textPosition += width + charSpacing * hScale;
+    if (char.isSpace) {
+      textPosition += wordSpacing * hScale;
+    }
+    return 1;
+  };
+  const processChars = (chars, textItem) => {
+    let skip = 0;
+    for (const char of chars) {
+      if (typeof char === "number") {
+        if (!skip) {
+          textPosition += -char * fontSize * hScale / 1e3;
+        }
+      } else if (skip) {
+        skip--;
+      } else {
+        skip = addGlyph(textItem, char) - 1;
+      }
+    }
+  };
   for (let i = 0; i < operatorsList.fnArray.length; i++) {
     const fnId = operatorsList.fnArray[i];
     const args = operatorsList.argsArray[i];
@@ -29927,19 +29958,7 @@ async function createTextLayer(page, {
         currentTextItem ?? (currentTextItem = createTextItem(fontFamily));
         createParagraphIfNeeded(lastTextItem, currentTextItem);
         const [chars] = args;
-        for (const char of chars) {
-          if (typeof char === "number") {
-            const tx = -char * fontSize * hScale / 1e3;
-            textPosition += tx;
-          } else if (char.unicode) {
-            const width = char.width * fontSize / 1e3 || 0;
-            currentTextItem.glyphs.push([char, textPosition, width]);
-            textPosition += width + charSpacing * hScale;
-            if (char.isSpace) {
-              textPosition += wordSpacing * hScale;
-            }
-          }
-        }
+        processChars(chars, currentTextItem);
         break;
       }
       case OPS.setTextRise:
@@ -29987,23 +30006,9 @@ async function createTextLayer(page, {
         const [text] = args;
         if (typeof text === "string") {
           const glyphs = currentTextItem.font.charsToGlyphs(text);
-          for (const char of glyphs) {
-            const width = char.width * fontSize / 1e3 || 0;
-            currentTextItem.glyphs.push([char, textPosition, width]);
-            textPosition += width + charSpacing * hScale;
-            if (char.isSpace) {
-              textPosition += wordSpacing * hScale;
-            }
-          }
+          processChars(glyphs, currentTextItem);
         } else if (Array.isArray(text)) {
-          for (const char of text) {
-            const width = char.width * fontSize / 1e3 || 0;
-            currentTextItem.glyphs.push([char, textPosition, width]);
-            textPosition += width + charSpacing * hScale;
-            if (char.isSpace) {
-              textPosition += wordSpacing * hScale;
-            }
-          }
+          processChars(text, currentTextItem);
         }
         break;
       }
@@ -30019,23 +30024,9 @@ async function createTextLayer(page, {
         lineMatrix = [...textMatrix];
         if (typeof text === "string") {
           const glyphs = currentTextItem.font.charsToGlyphs(text);
-          for (const char of glyphs) {
-            const width = char.width * fontSize / 1e3 || 0;
-            currentTextItem.glyphs.push([char, textPosition, width]);
-            textPosition += width + charSpacing * hScale;
-            if (char.isSpace) {
-              textPosition += wordSpacing * hScale;
-            }
-          }
+          processChars(glyphs, currentTextItem);
         } else if (Array.isArray(text)) {
-          for (const char of text) {
-            const width = char.width * fontSize / 1e3 || 0;
-            currentTextItem.glyphs.push([char, textPosition, width]);
-            textPosition += width + charSpacing * hScale;
-            if (char.isSpace) {
-              textPosition += wordSpacing * hScale;
-            }
-          }
+          processChars(text, currentTextItem);
         }
         break;
       }

package/dist/lib/TextLayer.d.ts

@@ -52,6 +52,8 @@ export interface TextLayerFigure extends TextLayerNode {
 export interface Glyph {
     unicode: string;
     isSpace: boolean;
+    fontChar?: string;
+    width?: number;
 }
 export type TextItem = {
     glyphs: [Glyph, number, number][];
@@ -78,6 +80,11 @@ export declare const renderTextLayer: (root: TextLayerRoot, options?: {
 }) => string;
 export declare const decorateStructTree: (node: StructTreeNodeWithAttrs | StructTreeContent, rootContainer: TextLayerRoot, graphics?: SvgRoot, annotations?: AnnotationData[], parents?: TextLayerElement[]) => void;
 export declare const measure: (text: string, font: string, fontSize: number, lang?: string) => TextMetrics;
+/**
+ * Expands ligatures using Unicode normalization (NFKD - Compatibility Decomposition).
+ * This automatically handles Latin ligatures (ff fi fl ffi ffl ſt st), IJ ligatures (IJ ij),
+ * and AE/OE letters (Æ æ Œ œ) by decomposing them into their component characters.
+ */
 export declare function createTextLayer(page: PDFPageProxy, { graphics, annotations: _annotations, }?: {
     graphics?: SvgRoot;
     annotations?: AnnotationData[];

package/dist/node/index.js

@@ -27888,6 +27888,25 @@ var flattenNodes = (node) => {
 var isNameObject = (obj) => {
   return obj !== null && typeof obj === "object" && "name" in obj;
 };
+var isLigature = (char) => {
+  if (!char || [...char].length !== 1) {
+    return false;
+  }
+  const code = char.codePointAt(0);
+  if (code == null) {
+    return false;
+  }
+  return (
+    // Alphabetic Presentation Forms
+    code >= 64256 && code <= 64335 || // Arabic Presentation Forms-A
+    code >= 64336 && code <= 65023 || // Arabic Presentation Forms-B
+    code >= 65136 && code <= 65279 || // Latin ligatures outside FB00 block
+    code === 338 || // Œ
+    code === 339 || // œ
+    code === 306 || // IJ
+    code === 307
+  );
+};
 var renderTextLayer = (root, options = {}) => {
   const { classes = false, styles = false, font } = options;
   const serializeAttributes = (attrs) => {
@@ -27922,7 +27941,7 @@ var renderTextLayer = (root, options = {}) => {
           "--tl-margin": `${margin}px`,
           "--tl-transform": scale !== 1 ? `scaleX(${scale})` : null
         }) : "";
-        return `<${tag3}${serializedAttrs3 ? ` ${serializedAttrs3}` : ""}${serializedStyle2 ? ` style="${serializedStyle2}` : ""}">${chunk.text}</${tag3}>`;
+        return `<${tag3}${serializedAttrs3 ? ` ${serializedAttrs3}` : ""}${serializedStyle2 ? ` style="${serializedStyle2}"` : ""}>${chunk.text}</${tag3}>`;
       }).join("");
       if (!classes && !styles) {
         return contents;
@@ -28431,42 +28450,16 @@ async function createTextLayer(page, {
     }
     if (oldTextItem?.glyphs.length) {
       if (newTextItem) {
-        const text = oldTextItem.glyphs.map((g) => g[0].unicode).join("") || "";
-        const dir = bidi(text).dir;
-        const firstGlyph = oldTextItem.glyphs[0];
-        const lastGlyph = oldTextItem.glyphs.at(-1);
-        const scale = oldTextItem.scaleX * oldTextItem.fontSize;
         const top = oldTextItem.top;
         const bottom = oldTextItem.top + oldTextItem.fontSize;
-        const left = oldTextItem.left;
-        const right = oldTextItem.left + (lastGlyph[1] + lastGlyph[2]) * scale;
         const nextTop = newTextItem.top;
         const nextBottom = newTextItem.top + newTextItem.fontSize;
-        switch (dir) {
-          case "ltr": {
-            if (newTextItem.left - right < lastGlyph[2] * scale) {
-              if (nextTop > top && nextTop < bottom || nextBottom > top && nextBottom < bottom) {
-                return;
-              }
-              if (checkNewLine(oldTextItem, newTextItem, true)) {
-                appendNewLineAsNeeded();
-                return;
-              }
-            }
-            break;
-          }
-          case "rtl": {
-            if (left - newTextItem.left < firstGlyph[2] * scale) {
-              if (nextTop > top && nextTop < bottom || nextBottom > top && nextBottom < bottom) {
-                return;
-              }
-              if (checkNewLine(oldTextItem, newTextItem, true)) {
-                appendNewLineAsNeeded();
-                return;
-              }
-            }
-            break;
-          }
+        if (nextTop > top && nextTop < bottom || nextBottom > top && nextBottom < bottom || nextTop <= top && nextBottom >= bottom) {
+          return;
+        }
+        if (checkNewLine(oldTextItem, newTextItem, true)) {
+          appendNewLineAsNeeded();
+          return;
         }
       }
       closeMarkedContent();
@@ -28507,6 +28500,44 @@ async function createTextLayer(page, {
       });
     }
   };
+  const addGlyph = (textItem, char) => {
+    const width = (char.width ?? 0) * fontSize / 1e3 || 0;
+    const charUnicode = char.unicode ?? "";
+    if (isLigature(charUnicode)) {
+      const normalized = charUnicode.normalize("NFKD");
+      const expansion = Array.from(normalized);
+      const widthPerChar = width / expansion.length;
+      for (const expandedChar of expansion) {
+        const syntheticGlyph = {
+          unicode: expandedChar,
+          isSpace: false
+        };
+        textItem.glyphs.push([syntheticGlyph, textPosition, widthPerChar]);
+        textPosition += widthPerChar + charSpacing * hScale;
+      }
+      return expansion.length;
+    }
+    textItem.glyphs.push([char, textPosition, width]);
+    textPosition += width + charSpacing * hScale;
+    if (char.isSpace) {
+      textPosition += wordSpacing * hScale;
+    }
+    return 1;
+  };
+  const processChars = (chars, textItem) => {
+    let skip = 0;
+    for (const char of chars) {
+      if (typeof char === "number") {
+        if (!skip) {
+          textPosition += -char * fontSize * hScale / 1e3;
+        }
+      } else if (skip) {
+        skip--;
+      } else {
+        skip = addGlyph(textItem, char) - 1;
+      }
+    }
+  };
   for (let i = 0; i < operatorsList.fnArray.length; i++) {
     const fnId = operatorsList.fnArray[i];
     const args = operatorsList.argsArray[i];
@@ -28528,19 +28559,7 @@ async function createTextLayer(page, {
         currentTextItem ?? (currentTextItem = createTextItem(fontFamily));
         createParagraphIfNeeded(lastTextItem, currentTextItem);
         const [chars] = args;
-        for (const char of chars) {
-          if (typeof char === "number") {
-            const tx = -char * fontSize * hScale / 1e3;
-            textPosition += tx;
-          } else if (char.unicode) {
-            const width = char.width * fontSize / 1e3 || 0;
-            currentTextItem.glyphs.push([char, textPosition, width]);
-            textPosition += width + charSpacing * hScale;
-            if (char.isSpace) {
-              textPosition += wordSpacing * hScale;
-            }
-          }
-        }
+        processChars(chars, currentTextItem);
         break;
       }
       case OPS.setTextRise:
@@ -28588,23 +28607,9 @@ async function createTextLayer(page, {
         const [text] = args;
         if (typeof text === "string") {
           const glyphs = currentTextItem.font.charsToGlyphs(text);
-          for (const char of glyphs) {
-            const width = char.width * fontSize / 1e3 || 0;
-            currentTextItem.glyphs.push([char, textPosition, width]);
-            textPosition += width + charSpacing * hScale;
-            if (char.isSpace) {
-              textPosition += wordSpacing * hScale;
-            }
-          }
+          processChars(glyphs, currentTextItem);
         } else if (Array.isArray(text)) {
-          for (const char of text) {
-            const width = char.width * fontSize / 1e3 || 0;
-            currentTextItem.glyphs.push([char, textPosition, width]);
-            textPosition += width + charSpacing * hScale;
-            if (char.isSpace) {
-              textPosition += wordSpacing * hScale;
-            }
-          }
+          processChars(text, currentTextItem);
         }
         break;
       }
@@ -28620,23 +28625,9 @@ async function createTextLayer(page, {
         lineMatrix = [...textMatrix];
         if (typeof text === "string") {
           const glyphs = currentTextItem.font.charsToGlyphs(text);
-          for (const char of glyphs) {
-            const width = char.width * fontSize / 1e3 || 0;
-            currentTextItem.glyphs.push([char, textPosition, width]);
-            textPosition += width + charSpacing * hScale;
-            if (char.isSpace) {
-              textPosition += wordSpacing * hScale;
-            }
-          }
+          processChars(glyphs, currentTextItem);
         } else if (Array.isArray(text)) {
-          for (const char of text) {
-            const width = char.width * fontSize / 1e3 || 0;
-            currentTextItem.glyphs.push([char, textPosition, width]);
-            textPosition += width + charSpacing * hScale;
-            if (char.isSpace) {
-              textPosition += wordSpacing * hScale;
-            }
-          }
+          processChars(text, currentTextItem);
         }
         break;
       }

package/package.json

@@ -1,7 +1,7 @@
 {
 	"name": "@chialab/pdfjs-lib",
 	"description": "A custom Mozilla's PDF.js build with better Node support and extras.",
-	"version": "1.0.0-alpha.44",
+	"version": "1.0.0-alpha.46",
 	"type": "module",
 	"author": "Chialab <dev@chialab.it>",
 	"license": "MIT",