@chialab/pdfjs-lib 1.0.0-alpha.42 → 1.0.0-alpha.44

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");
@@ -28025,7 +28028,7 @@ var SvgCanvasContext = class {
       attrs: {},
       children: []
     };
-    this._currentGroup = this._root;
+    this._currentGroup = null;
     this._addNode(this._defs, this._root);
     this.resetTransform();
   }
@@ -28159,13 +28162,15 @@ var SvgCanvasContext = class {
     this._root.attrs.height = height;
   }
   save() {
-    this._groupStack.push(this._currentGroup ?? this._root);
+    if (this._currentGroup) {
+      this._groupStack.push(this._currentGroup);
+    }
     this._currentGroup = this._ensureTransformationGroup();
     this._styleStack.push(this._getStyleState());
     this._transformMatrixStack.push(this.getTransform());
   }
   restore() {
-    this._currentGroup = this._groupStack.pop() ?? this._root;
+    this._currentGroup = this._groupStack.pop() ?? null;
     if (this._styleStack.length) {
       this._applyStyleState(this._styleStack.pop());
     }
@@ -28447,9 +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 {
@@ -28460,9 +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];
@@ -28475,7 +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: {
@@ -28617,13 +28629,20 @@ var SvgCanvasContext = class {
       this._currentMarked.group,
       this._defs.children
     );
+    const x = Math.min(Math.max(bbox?.x ?? 0, 0), this._width);
+    const y = Math.min(Math.max(bbox?.y ?? 0, 0), this._height);
+    const width = Math.min(Math.max(bbox?.width ?? 0, 0), this._width - x);
+    const height = Math.min(Math.max(bbox?.height ?? 0, 0), this._height - y);
     this._currentMarked.node.attrs = {
       ...this._currentMarked.node.attrs,
-      x: bbox?.x ?? 0,
-      y: bbox?.y ?? 0,
-      width: bbox?.width ?? 0,
-      height: bbox?.height ?? 0
+      x,
+      y,
+      width,
+      height
     };
+    if (this._currentGroup && this._currentMarked.elements.includes(this._currentGroup)) {
+      this._currentGroup = this._groupStack.pop() ?? null;
+    }
     this._currentMarked = this._markedStack.pop() ?? null;
   }
   _ensureTransformationGroup() {
@@ -28660,7 +28679,7 @@ var SvgCanvasContext = class {
         markedStackToAdd.unshift(this._markedStack[i]);
       }
       for (const marked of markedStackToAdd) {
-        const parentMarked = this._markedStack[++i - 1];
+        const parentMarked = this._markedStack[i++];
         if (!parentMarked || parentMarked.elements.includes(group)) {
           this._addNode(marked.node, group);
         } else {
@@ -28895,7 +28914,7 @@ var SvgCanvasContext = class {
     this._defs.children = [];
     this._root.children = [];
     this._addNode(this._defs, this._root);
-    this._currentGroup = this._root;
+    this._currentGroup = null;
     this._groupStack = [];
     this._parents.clear();
   }
@@ -29680,11 +29699,10 @@ async function createTextLayer(page, {
     const scaleX = Math.sqrt(finalMatrix[0] ** 2 + finalMatrix[1] ** 2);
     const scaleY = Math.sqrt(finalMatrix[2] ** 2 + finalMatrix[3] ** 2);
     const finalFontSize = fontSize * scaleY;
-    const ascentOffset = finalFontSize * (font.ascent + font.descent);
+    const ascentOffset = font.ascent / (font.ascent - font.descent) * finalFontSize;
     const angle = -Math.atan2(finalMatrix[1], finalMatrix[0]);
     const top = y - ascentOffset * Math.cos(angle);
     const left = x + ascentOffset * Math.sin(angle);
-    appendNewLineAsNeeded(top);
     return {
       glyphs: [],
       font,
@@ -29790,14 +29808,66 @@ async function createTextLayer(page, {
     markedContent.children.push(lastTextLayer);
     resetTextItem();
   };
+  const checkNewLine = (oldTextItem, newTextItem, sameStyle = false) => {
+    if (sameStyle) {
+      if (oldTextItem.fontSize !== newTextItem.fontSize || oldTextItem.font.name !== newTextItem.font.name) {
+        return false;
+      }
+    }
+    if (newTextItem.top > oldTextItem.top) {
+      return oldTextItem.top + oldTextItem.fontSize < newTextItem.top;
+    }
+    return newTextItem.top + newTextItem.fontSize < oldTextItem.top;
+  };
   const createParagraphIfNeeded = (oldTextItem, newTextItem) => {
     if (structTree) {
+      if (oldTextItem && newTextItem) {
+        if (checkNewLine(oldTextItem, newTextItem)) {
+          appendNewLineAsNeeded();
+        }
+      }
       return;
     }
-    if (oldTextItem && newTextItem && oldTextItem.fontSize === newTextItem.fontSize) {
-      return;
-    }
-    if (oldTextItem) {
+    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;
+          }
+        }
+      }
       closeMarkedContent();
     }
     const currentMarkedContent = markedContent || rootContainer;
@@ -29823,15 +29893,11 @@ async function createTextLayer(page, {
       }
     }
   };
-  const appendNewLineAsNeeded = (top) => {
+  const appendNewLineAsNeeded = () => {
     if (!lastTextLayer) {
       return;
     }
     const computedText = getNodeContents(lastTextLayer);
-    const diff = Math.abs(lastTextLayer.top - top);
-    if (diff < lastTextLayer.fontSize) {
-      return;
-    }
     if (!computedText.match(/[\s\n]$/)) {
       lastTextLayer.text.push({
         text: "\n",

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/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");
@@ -26626,7 +26629,7 @@ var SvgCanvasContext = class {
       attrs: {},
       children: []
     };
-    this._currentGroup = this._root;
+    this._currentGroup = null;
     this._addNode(this._defs, this._root);
     this.resetTransform();
   }
@@ -26760,13 +26763,15 @@ var SvgCanvasContext = class {
     this._root.attrs.height = height;
   }
   save() {
-    this._groupStack.push(this._currentGroup ?? this._root);
+    if (this._currentGroup) {
+      this._groupStack.push(this._currentGroup);
+    }
     this._currentGroup = this._ensureTransformationGroup();
     this._styleStack.push(this._getStyleState());
     this._transformMatrixStack.push(this.getTransform());
   }
   restore() {
-    this._currentGroup = this._groupStack.pop() ?? this._root;
+    this._currentGroup = this._groupStack.pop() ?? null;
     if (this._styleStack.length) {
       this._applyStyleState(this._styleStack.pop());
     }
@@ -27048,9 +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 {
@@ -27061,9 +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];
@@ -27076,7 +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: {
@@ -27218,13 +27230,20 @@ var SvgCanvasContext = class {
       this._currentMarked.group,
       this._defs.children
     );
+    const x = Math.min(Math.max(bbox?.x ?? 0, 0), this._width);
+    const y = Math.min(Math.max(bbox?.y ?? 0, 0), this._height);
+    const width = Math.min(Math.max(bbox?.width ?? 0, 0), this._width - x);
+    const height = Math.min(Math.max(bbox?.height ?? 0, 0), this._height - y);
     this._currentMarked.node.attrs = {
       ...this._currentMarked.node.attrs,
-      x: bbox?.x ?? 0,
-      y: bbox?.y ?? 0,
-      width: bbox?.width ?? 0,
-      height: bbox?.height ?? 0
+      x,
+      y,
+      width,
+      height
     };
+    if (this._currentGroup && this._currentMarked.elements.includes(this._currentGroup)) {
+      this._currentGroup = this._groupStack.pop() ?? null;
+    }
     this._currentMarked = this._markedStack.pop() ?? null;
   }
   _ensureTransformationGroup() {
@@ -27261,7 +27280,7 @@ var SvgCanvasContext = class {
         markedStackToAdd.unshift(this._markedStack[i]);
       }
       for (const marked of markedStackToAdd) {
-        const parentMarked = this._markedStack[++i - 1];
+        const parentMarked = this._markedStack[i++];
         if (!parentMarked || parentMarked.elements.includes(group)) {
           this._addNode(marked.node, group);
         } else {
@@ -27496,7 +27515,7 @@ var SvgCanvasContext = class {
     this._defs.children = [];
     this._root.children = [];
     this._addNode(this._defs, this._root);
-    this._currentGroup = this._root;
+    this._currentGroup = null;
     this._groupStack = [];
     this._parents.clear();
   }
@@ -28281,11 +28300,10 @@ async function createTextLayer(page, {
     const scaleX = Math.sqrt(finalMatrix[0] ** 2 + finalMatrix[1] ** 2);
     const scaleY = Math.sqrt(finalMatrix[2] ** 2 + finalMatrix[3] ** 2);
     const finalFontSize = fontSize * scaleY;
-    const ascentOffset = finalFontSize * (font.ascent + font.descent);
+    const ascentOffset = font.ascent / (font.ascent - font.descent) * finalFontSize;
     const angle = -Math.atan2(finalMatrix[1], finalMatrix[0]);
     const top = y - ascentOffset * Math.cos(angle);
     const left = x + ascentOffset * Math.sin(angle);
-    appendNewLineAsNeeded(top);
     return {
       glyphs: [],
       font,
@@ -28391,14 +28409,66 @@ async function createTextLayer(page, {
     markedContent.children.push(lastTextLayer);
     resetTextItem();
   };
+  const checkNewLine = (oldTextItem, newTextItem, sameStyle = false) => {
+    if (sameStyle) {
+      if (oldTextItem.fontSize !== newTextItem.fontSize || oldTextItem.font.name !== newTextItem.font.name) {
+        return false;
+      }
+    }
+    if (newTextItem.top > oldTextItem.top) {
+      return oldTextItem.top + oldTextItem.fontSize < newTextItem.top;
+    }
+    return newTextItem.top + newTextItem.fontSize < oldTextItem.top;
+  };
   const createParagraphIfNeeded = (oldTextItem, newTextItem) => {
     if (structTree) {
+      if (oldTextItem && newTextItem) {
+        if (checkNewLine(oldTextItem, newTextItem)) {
+          appendNewLineAsNeeded();
+        }
+      }
       return;
     }
-    if (oldTextItem && newTextItem && oldTextItem.fontSize === newTextItem.fontSize) {
-      return;
-    }
-    if (oldTextItem) {
+    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;
+          }
+        }
+      }
       closeMarkedContent();
     }
     const currentMarkedContent = markedContent || rootContainer;
@@ -28424,15 +28494,11 @@ async function createTextLayer(page, {
       }
     }
   };
-  const appendNewLineAsNeeded = (top) => {
+  const appendNewLineAsNeeded = () => {
     if (!lastTextLayer) {
       return;
     }
     const computedText = getNodeContents(lastTextLayer);
-    const diff = Math.abs(lastTextLayer.top - top);
-    if (diff < lastTextLayer.fontSize) {
-      return;
-    }
     if (!computedText.match(/[\s\n]$/)) {
       lastTextLayer.text.push({
         text: "\n",

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.42",
+	"version": "1.0.0-alpha.44",
 	"type": "module",
 	"author": "Chialab <dev@chialab.it>",
 	"license": "MIT",