@chialab/pdfjs-lib 1.0.0-alpha.43 → 1.0.0-alpha.45

package/dist/browser/index.js

@@ -27916,6 +27916,9 @@ function isCanvas(img) {
 function isImage(img) {
   return "src" in img;
 }
+function isVideoFrame(img) {
+  return "timestamp" in img;
+}
 var SvgPattern = class {
   constructor(src, width, height) {
     __publicField(this, "src");
@@ -28449,9 +28452,9 @@ var SvgCanvasContext = class {
     if (args.length === 2) {
       dx = args[0];
       dy = args[1];
-      if (isCanvas(image) || isImage(image)) {
-        sw = image.width;
-        sh = image.height;
+      if (isCanvas(image) || isImage(image) || isVideoFrame(image)) {
+        sw = image.width ?? image.displayWidth;
+        sh = image.height ?? image.displayHeight;
         dw = sw;
         dh = sh;
       } else {
@@ -28462,9 +28465,9 @@ var SvgCanvasContext = class {
       dy = args[1];
       dw = args[2];
       dh = args[3];
-      if (isCanvas(image) || isImage(image)) {
-        sw = image.width;
-        sh = image.height;
+      if (isCanvas(image) || isImage(image) || isVideoFrame(image)) {
+        sw = image.width ?? image.displayWidth;
+        sh = image.height ?? image.displayHeight;
       }
     } else if (args.length === 8) {
       sw = args[2];
@@ -28477,7 +28480,14 @@ var SvgCanvasContext = class {
       throw new Error("Invalid number of arguments");
     }
     const matrix = this._transformMatrix.translate(dx, dy);
-    const href = isCanvas(image) ? image.toDataURL() : isImage(image) ? image.src : "";
+    const canvas = createCanvas(dw, dh);
+    const ctx = canvas.getContext("2d");
+    if (sw != null && sh != null) {
+      ctx.drawImage(image, 0, 0, sw, sh, 0, 0, dw, dh);
+    } else {
+      ctx.drawImage(image, 0, 0, dw, dh);
+    }
+    const href = canvas.toDataURL();
     const svgImage = {
       tag: "image",
       attrs: {
@@ -29277,6 +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) => {
@@ -29311,7 +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;
@@ -29896,6 +29925,44 @@ 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];
@@ -29917,19 +29984,7 @@ 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:
@@ -29977,23 +30032,9 @@ 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;
       }
@@ -30009,23 +30050,9 @@ 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/index.d.ts

@@ -10,7 +10,7 @@ export * from './lib/StandardFontDataFactory';
 export * from './lib/AnnotationData';
 export * from './lib/CanvasGraphics';
 export * from './lib/SvgCanvasContext';
-export { type TextLayerRoot, type TextLayerNode, type TextLayerAnchor, type TextLayerFigure, type TextLayerElement, isTextNode, isAnchorNode, isFigureNode, isElementNode, measure, createTextLayer, renderTextLayer, } from './lib/TextLayer';
+export { type TextLayerRoot, type TextLayerNode, type TextLayerAnchor, type TextLayerFigure, type TextLayerElement, type TextLayerText, isTextNode, isAnchorNode, isFigureNode, isElementNode, measure, createTextLayer, renderTextLayer, } from './lib/TextLayer';
 export declare const textLayerUtils: {
     findNode: (root: import(".").TextLayerElement, callback: (node: import(".").TextLayerNode, parent: import(".").TextLayerElement) => boolean | null | undefined) => import(".").TextLayerNode | null;
     findNodes: (node: import(".").TextLayerElement, callback: (node: import(".").TextLayerNode, parent: import(".").TextLayerElement) => boolean | null | undefined) => import(".").TextLayerNode[];

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

@@ -26517,6 +26517,9 @@ function isCanvas(img) {
 function isImage(img) {
   return "src" in img;
 }
+function isVideoFrame(img) {
+  return "timestamp" in img;
+}
 var SvgPattern = class {
   constructor(src, width, height) {
     __publicField(this, "src");
@@ -27050,9 +27053,9 @@ var SvgCanvasContext = class {
     if (args.length === 2) {
       dx = args[0];
       dy = args[1];
-      if (isCanvas(image) || isImage(image)) {
-        sw = image.width;
-        sh = image.height;
+      if (isCanvas(image) || isImage(image) || isVideoFrame(image)) {
+        sw = image.width ?? image.displayWidth;
+        sh = image.height ?? image.displayHeight;
         dw = sw;
         dh = sh;
       } else {
@@ -27063,9 +27066,9 @@ var SvgCanvasContext = class {
       dy = args[1];
       dw = args[2];
       dh = args[3];
-      if (isCanvas(image) || isImage(image)) {
-        sw = image.width;
-        sh = image.height;
+      if (isCanvas(image) || isImage(image) || isVideoFrame(image)) {
+        sw = image.width ?? image.displayWidth;
+        sh = image.height ?? image.displayHeight;
       }
     } else if (args.length === 8) {
       sw = args[2];
@@ -27078,7 +27081,14 @@ var SvgCanvasContext = class {
       throw new Error("Invalid number of arguments");
     }
     const matrix = this._transformMatrix.translate(dx, dy);
-    const href = isCanvas(image) ? image.toDataURL() : isImage(image) ? image.src : "";
+    const canvas = createCanvas(dw, dh);
+    const ctx = canvas.getContext("2d");
+    if (sw != null && sh != null) {
+      ctx.drawImage(image, 0, 0, sw, sh, 0, 0, dw, dh);
+    } else {
+      ctx.drawImage(image, 0, 0, dw, dh);
+    }
+    const href = canvas.toDataURL();
     const svgImage = {
       tag: "image",
       attrs: {
@@ -27878,6 +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) => {
@@ -27912,7 +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;
@@ -28497,6 +28526,44 @@ 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];
@@ -28518,19 +28585,7 @@ 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:
@@ -28578,23 +28633,9 @@ 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;
       }
@@ -28610,23 +28651,9 @@ 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.43",
+	"version": "1.0.0-alpha.45",
 	"type": "module",
 	"author": "Chialab <dev@chialab.it>",
 	"license": "MIT",