@chialab/pdfjs-lib 1.0.0-alpha.30 → 1.0.0-alpha.32

package/dist/browser/index.js

@@ -27858,26 +27858,6 @@ async function toDataUrl(data, type = "image/png") {
   }
   return `data:${type};base64,${Buffer.from(data).toString("base64")}`;
 }
-function makeSerializable(object) {
-  if (typeof object !== "object" || object === null) {
-    return object;
-  }
-  if (object instanceof Int8Array || object instanceof Uint8Array || object instanceof Uint8ClampedArray || object instanceof Int16Array || object instanceof Uint16Array || object instanceof Int32Array || object instanceof Uint32Array || object instanceof Float32Array || object instanceof Float64Array) {
-    return makeSerializable(Array.from(object));
-  }
-  if (object instanceof BigInt64Array || object instanceof BigUint64Array) {
-    return makeSerializable(Array.from(object));
-  }
-  if (Array.isArray(object)) {
-    return object.map(makeSerializable);
-  }
-  return Object.fromEntries(
-    Object.entries(object).map(([key, value]) => [
-      key,
-      makeSerializable(value)
-    ])
-  );
-}
 function colorToRgb(color) {
   if (color.startsWith("#")) {
     const hex = color.slice(1);
@@ -29121,7 +29101,7 @@ PDFPageProxy.prototype.getAnnotations = async function(params) {
       }
     }
   }
-  return makeSerializable(annotations);
+  return annotations;
 };
 
 // src/lib/AnnotationData.ts
@@ -29357,6 +29337,30 @@ var renderTextLayer = (root, options = {}) => {
     }
     if (isTextNode(node)) {
       const isHeading = parents.find((p) => p.role.match(/^h[1-6]$/));
+      let contents;
+      if (typeof node.text === "string") {
+        contents = node.text;
+      } else {
+        contents = node.text.map((chunk) => {
+          if (!classes && !styles) {
+            return chunk.text;
+          }
+          const tag3 = !isHeading && node.fontWeight >= 700 && node.fontStyle === "italic" ? "em" : "span";
+          const serializedAttrs3 = serializeAttributes(
+            classes ? {
+              class: "tl-span"
+            } : {}
+          );
+          const serializedStyle2 = styles ? serializeStyles({
+            "--tl-margin": `${chunk.margin}px`,
+            "--tl-transform": typeof chunk.scale === "number" && chunk.scale !== 1 ? `scaleX(${chunk.scale ?? 1})` : null
+          }) : "";
+          return `<${tag3}${serializedAttrs3 ? ` ${serializedAttrs3}` : ""}${serializedStyle2 ? ` style="${serializedStyle2}` : ""}">${chunk.text}</${tag3}>`;
+        }).join("");
+      }
+      if (!classes && !styles) {
+        return contents;
+      }
       const tag2 = isHeading ? "span" : node.fontWeight >= 700 ? "strong" : node.fontStyle === "italic" ? "em" : "span";
       const serializedAttrs2 = serializeAttributes(
         classes ? {
@@ -29374,25 +29378,6 @@ var renderTextLayer = (root, options = {}) => {
           typeof node.angle === "number" && node.angle !== 0 ? `rotate(${node.angle}deg)` : null
         ].filter((v) => v !== null).join(" ")
       }) : "";
-      let contents;
-      if (typeof node.text === "string") {
-        contents = node.text;
-      } else {
-        contents = node.text.map((chunk) => {
-          const tag3 = !isHeading && node.fontWeight >= 700 && node.fontStyle === "italic";
-          const serializedAttrs3 = serializeAttributes(
-            classes ? {
-              class: "tl-span"
-            } : {}
-          );
-          const serializedStyle2 = styles ? serializeStyles({
-            "--tl-left": `${chunk.left}px`,
-            "--tl-top": `${chunk.top}px`,
-            "--tl-transform": typeof chunk.scale === "number" && chunk.scale !== 1 ? `scaleX(${chunk.scale ?? 1})` : null
-          }) : "";
-          return `<${tag3}${serializedAttrs3 ? ` ${serializedAttrs3}` : ""}${serializedStyle2 ? ` style="${serializedStyle2}` : ""}">${chunk.text}</${tag3}>`;
-        }).join("");
-      }
       return `<${tag2}${serializedAttrs2 ? ` ${serializedAttrs2}` : ""}${serializedStyle ? ` style="${serializedStyle}"` : ""}>${contents}</${tag2}>`;
     }
     if (isAnchorNode(node)) {
@@ -29668,6 +29653,10 @@ var decorateStructTree = (node, rootContainer, graphics, annotations, parents =
       if (child.id !== contentId) {
         return;
       }
+      if (child.attrs) {
+        parent.attrs ?? (parent.attrs = {});
+        Object.assign(parent.attrs, child.attrs);
+      }
       previousParent.children = previousParent.children.filter(
         (c) => c !== child
       );
@@ -42378,7 +42367,7 @@ async function createTextLayerV2(page, {
         if (!glyph[0]) {
           return acc;
         }
-        if (!glyph[0].isSpace) {
+        if (glyph[0].unicode.trim()) {
           acc[acc.length - 1].push(glyph);
         } else {
           acc.push([glyph]);
@@ -42389,21 +42378,50 @@ async function createTextLayerV2(page, {
       [[]]
     );
     const chunks = [];
-    for (const glyphs of glyphsBlocks) {
+    let currentLeft = 0;
+    for (let i = 0; i < glyphsBlocks.length; i++) {
+      const glyphs = glyphsBlocks[i];
       if (glyphs.length === 0) {
         continue;
       }
       const text = glyphs.map((g) => g[0].unicode).join("");
+      const textWidth = textFont.getAdvanceWidth(text, fontSize);
+      if (!textWidth) {
+        continue;
+      }
+      const isSpaceOnly = text.trim() === "";
+      const previousBlock = glyphsBlocks.at(i - 1);
+      const nextBlock = glyphsBlocks.at(i + 1);
+      const previousGlyphs = previousBlock?.at(-1);
+      const nextGlyphs = nextBlock?.[0];
       const firstGlyph = glyphs[0];
       const lastGlyph = glyphs[glyphs.length - 1];
-      const graphicWidth = lastGlyph[1] - firstGlyph[1] + lastGlyph[2];
-      const textWidth = textFont.getAdvanceWidth(text, fontSize);
+      let margin = firstGlyph[1] / fontSize - currentLeft;
+      let graphicWidth = lastGlyph[1] - firstGlyph[1] + lastGlyph[2];
+      if (isSpaceOnly) {
+        if (previousGlyphs) {
+          margin = (previousGlyphs[1] + previousGlyphs[2]) / fontSize - currentLeft;
+        }
+        if (nextGlyphs) {
+          if (previousGlyphs) {
+            graphicWidth = nextGlyphs[1] - (previousGlyphs[1] + previousGlyphs[2]);
+          } else {
+            graphicWidth = nextGlyphs[1] - firstGlyph[1];
+          }
+        }
+      } else if (previousGlyphs) {
+        const space = (previousGlyphs[1] + previousGlyphs[2]) / fontSize - currentLeft;
+        if (margin > 0 && margin - space <= 0.1) {
+          graphicWidth += (margin - space) * fontSize;
+          margin = space;
+        }
+      }
       chunks.push({
         text,
-        top: 0,
-        left: firstGlyph[1] * finalFontSize,
-        scale: textWidth > 0 ? graphicWidth / textWidth : 1
+        margin: margin * finalFontSize,
+        scale: graphicWidth / textWidth
       });
+      currentLeft += margin + textWidth / fontSize;
     }
     markedContent.children.push({
       role: "text",
@@ -42567,11 +42585,12 @@ async function createTextLayerV2(page, {
         break;
       case OPS.beginMarkedContent:
       case OPS.beginMarkedContentProps: {
-        const [role, idRef] = args;
+        const [role, idRef, props] = args;
         const id2 = normalizeMarkedContentId(idRef);
         markedContent.children.push({
           id: id2,
           role: role?.toString().toLowerCase() || "span",
+          attrs: props ?? void 0,
           children: []
         });
         markedContent = markedContent.children.at(-1) || rootContainer;
@@ -42736,7 +42755,6 @@ export {
   isWidgetAnnotation,
   loadDefaultFonts,
   loadTextLayerFonts,
-  makeSerializable,
   noContextMenu,
   normalizeUnicode,
   parseRgbaColor,

package/dist/browser/worker.js

@@ -78280,7 +78280,67 @@ AnnotationFactory.saveNewAnnotations = async (evaluator, task, annotations, imag
   return data;
 };
 
-// src/lib/Font.ts
+// src/lib/Evaluator.ts
+var collectMarkedContentLanguage = (stream, xref) => {
+  const map = /* @__PURE__ */ new Map();
+  const pos = stream.pos;
+  const preprocessor = new EvaluatorPreprocessor(stream, xref);
+  const operation = {
+    fn: null,
+    args: null
+  };
+  while (preprocessor.read(operation)) {
+    if (operation.fn === null) {
+      operation.args = null;
+      continue;
+    }
+    if (operation.fn !== OPS.beginMarkedContentProps) {
+      operation.args = null;
+      continue;
+    }
+    if (!operation.args?.length) {
+      operation.args = null;
+      continue;
+    }
+    const [, dict] = operation.args.slice();
+    operation.args = null;
+    if (!(dict instanceof Dict)) {
+      continue;
+    }
+    const mcid = dict.get("MCID");
+    if (typeof mcid !== "number") {
+      continue;
+    }
+    const lang = dict.get("Lang");
+    if (!lang || typeof lang !== "string") {
+      continue;
+    }
+    map.set(mcid, lang);
+  }
+  stream.pos = pos;
+  return map;
+};
+var getOperatorList = PartialEvaluator.prototype.getOperatorList;
+PartialEvaluator.prototype.getOperatorList = async function(options, ...args) {
+  const { stream, operatorList } = options;
+  const languages = collectMarkedContentLanguage(stream, this.xref);
+  const addOp = operatorList.addOp;
+  operatorList.addOp = function(fn, args2) {
+    if (fn === OPS.beginMarkedContentProps) {
+      const mcid = args2[1];
+      if (typeof mcid === "number" && languages.has(mcid)) {
+        const lang = languages.get(mcid);
+        if (lang) {
+          args2[2] = {
+            lang: stringToPDFString(lang)
+          };
+        }
+      }
+    }
+    return addOp.call(this, fn, args2);
+  };
+  await getOperatorList.call(this, options, ...args);
+};
 var translateFont = PartialEvaluator.prototype.translateFont;
 PartialEvaluator.prototype.translateFont = async function(options, ...args) {
   const { descriptor } = options;

package/dist/lib/TextLayer.d.ts

@@ -20,14 +20,14 @@ export interface TextLayerAnchor extends TextLayerElement {
     role: 'a';
     href: string;
 }
+export interface TextLayerChunk {
+    text: string;
+    margin: number;
+    scale?: number;
+}
 export interface TextLayerText extends TextLayerNode {
     role: 'text';
-    text: string | {
-        text: string;
-        top: number;
-        left: number;
-        scale?: number;
-    }[];
+    text: /* v1 */ string | /* v2 */ TextLayerChunk[];
     top: number;
     left: number;
     fontSize: number;

package/dist/lib/utils.d.ts

@@ -16,13 +16,6 @@ export declare function canvasToData(canvas: HTMLCanvasElement | Canvas): Promis
  * @returns A promise that resolves to the data url.
  */
 export declare function toDataUrl(data: Uint8Array<ArrayBuffer>, type?: string): Promise<string>;
-/**
- * Ensure the object can be serialized and unserialized as JSON.
- * Internally it converts typed arrays to plain arrays.
- * @param object The object to serialize.
- * @returns The serialized object.
- */
-export declare function makeSerializable<T>(object: T): T;
 /**
  * Convert hash color to RGB array.
  * @param color The color in hex format (e.g., '#ff0000' or '#f00').

package/dist/node/{NodeUtils-PVXQMWFJ.js → NodeUtils-SRP3N4DX.js}

@@ -6,7 +6,7 @@ import {
   NodeWasmFactory,
   fetchData2 as fetchData,
   filtersRegistry
-} from "./chunk-O74KGUUC.js";
+} from "./chunk-R66TN6BM.js";
 import "./chunk-T2JWSGAF.js";
 import "./chunk-ZFIGV5OT.js";
 export {

package/dist/node/{chunk-O74KGUUC.js → chunk-R66TN6BM.js}

@@ -1635,26 +1635,6 @@ async function toDataUrl(data, type = "image/png") {
   }
   return `data:${type};base64,${Buffer.from(data).toString("base64")}`;
 }
-function makeSerializable(object) {
-  if (typeof object !== "object" || object === null) {
-    return object;
-  }
-  if (object instanceof Int8Array || object instanceof Uint8Array || object instanceof Uint8ClampedArray || object instanceof Int16Array || object instanceof Uint16Array || object instanceof Int32Array || object instanceof Uint32Array || object instanceof Float32Array || object instanceof Float64Array) {
-    return makeSerializable(Array.from(object));
-  }
-  if (object instanceof BigInt64Array || object instanceof BigUint64Array) {
-    return makeSerializable(Array.from(object));
-  }
-  if (Array.isArray(object)) {
-    return object.map(makeSerializable);
-  }
-  return Object.fromEntries(
-    Object.entries(object).map(([key, value]) => [
-      key,
-      makeSerializable(value)
-    ])
-  );
-}
 function colorToRgb(color) {
   if (color.startsWith("#")) {
     const hex = color.slice(1);
@@ -2079,7 +2059,6 @@ export {
   DOMFilterFactory,
   canvasToData,
   toDataUrl,
-  makeSerializable,
   colorToRgb,
   rgbToHex,
   parseRgbaColor,

package/dist/node/index.js

@@ -46,7 +46,6 @@ import {
   isDataScheme,
   isPdfFile,
   isValidFetchUrl,
-  makeSerializable,
   noContextMenu,
   parseRgbaColor,
   renderRichText,
@@ -54,7 +53,7 @@ import {
   setLayerDimensions,
   stopEvent,
   toDataUrl
-} from "./chunk-O74KGUUC.js";
+} from "./chunk-R66TN6BM.js";
 import {
   AbortException,
   AnnotationBorderStyleType,
@@ -39834,7 +39833,7 @@ function destroySvgContext(ctx) {
 
 // src/lib/PDFPageProxy.ts
 async function loadNodeCanvasFactory() {
-  const { NodeCanvasFactory: NodeCanvasFactory2 } = await import("./NodeUtils-PVXQMWFJ.js");
+  const { NodeCanvasFactory: NodeCanvasFactory2 } = await import("./NodeUtils-SRP3N4DX.js");
   return new NodeCanvasFactory2({});
 }
 var getAnnotations = PDFPageProxy.prototype.getAnnotations;
@@ -40001,7 +40000,7 @@ PDFPageProxy.prototype.getAnnotations = async function(params) {
       }
     }
   }
-  return makeSerializable(annotations);
+  return annotations;
 };
 
 // src/lib/AnnotationData.ts
@@ -40237,6 +40236,30 @@ var renderTextLayer = (root, options = {}) => {
     }
     if (isTextNode(node)) {
       const isHeading = parents.find((p) => p.role.match(/^h[1-6]$/));
+      let contents;
+      if (typeof node.text === "string") {
+        contents = node.text;
+      } else {
+        contents = node.text.map((chunk) => {
+          if (!classes && !styles) {
+            return chunk.text;
+          }
+          const tag3 = !isHeading && node.fontWeight >= 700 && node.fontStyle === "italic" ? "em" : "span";
+          const serializedAttrs3 = serializeAttributes(
+            classes ? {
+              class: "tl-span"
+            } : {}
+          );
+          const serializedStyle2 = styles ? serializeStyles({
+            "--tl-margin": `${chunk.margin}px`,
+            "--tl-transform": typeof chunk.scale === "number" && chunk.scale !== 1 ? `scaleX(${chunk.scale ?? 1})` : null
+          }) : "";
+          return `<${tag3}${serializedAttrs3 ? ` ${serializedAttrs3}` : ""}${serializedStyle2 ? ` style="${serializedStyle2}` : ""}">${chunk.text}</${tag3}>`;
+        }).join("");
+      }
+      if (!classes && !styles) {
+        return contents;
+      }
       const tag2 = isHeading ? "span" : node.fontWeight >= 700 ? "strong" : node.fontStyle === "italic" ? "em" : "span";
       const serializedAttrs2 = serializeAttributes(
         classes ? {
@@ -40254,25 +40277,6 @@ var renderTextLayer = (root, options = {}) => {
           typeof node.angle === "number" && node.angle !== 0 ? `rotate(${node.angle}deg)` : null
         ].filter((v) => v !== null).join(" ")
       }) : "";
-      let contents;
-      if (typeof node.text === "string") {
-        contents = node.text;
-      } else {
-        contents = node.text.map((chunk) => {
-          const tag3 = !isHeading && node.fontWeight >= 700 && node.fontStyle === "italic";
-          const serializedAttrs3 = serializeAttributes(
-            classes ? {
-              class: "tl-span"
-            } : {}
-          );
-          const serializedStyle2 = styles ? serializeStyles({
-            "--tl-left": `${chunk.left}px`,
-            "--tl-top": `${chunk.top}px`,
-            "--tl-transform": typeof chunk.scale === "number" && chunk.scale !== 1 ? `scaleX(${chunk.scale ?? 1})` : null
-          }) : "";
-          return `<${tag3}${serializedAttrs3 ? ` ${serializedAttrs3}` : ""}${serializedStyle2 ? ` style="${serializedStyle2}` : ""}">${chunk.text}</${tag3}>`;
-        }).join("");
-      }
       return `<${tag2}${serializedAttrs2 ? ` ${serializedAttrs2}` : ""}${serializedStyle ? ` style="${serializedStyle}"` : ""}>${contents}</${tag2}>`;
     }
     if (isAnchorNode(node)) {
@@ -40548,6 +40552,10 @@ var decorateStructTree = (node, rootContainer, graphics, annotations, parents =
       if (child.id !== contentId) {
         return;
       }
+      if (child.attrs) {
+        parent.attrs ?? (parent.attrs = {});
+        Object.assign(parent.attrs, child.attrs);
+      }
       previousParent.children = previousParent.children.filter(
         (c) => c !== child
       );
@@ -41012,7 +41020,7 @@ async function createTextLayerV2(page, {
         if (!glyph[0]) {
           return acc;
         }
-        if (!glyph[0].isSpace) {
+        if (glyph[0].unicode.trim()) {
           acc[acc.length - 1].push(glyph);
         } else {
           acc.push([glyph]);
@@ -41023,21 +41031,50 @@ async function createTextLayerV2(page, {
       [[]]
     );
     const chunks = [];
-    for (const glyphs of glyphsBlocks) {
+    let currentLeft = 0;
+    for (let i = 0; i < glyphsBlocks.length; i++) {
+      const glyphs = glyphsBlocks[i];
       if (glyphs.length === 0) {
         continue;
       }
       const text = glyphs.map((g) => g[0].unicode).join("");
+      const textWidth = textFont.getAdvanceWidth(text, fontSize);
+      if (!textWidth) {
+        continue;
+      }
+      const isSpaceOnly = text.trim() === "";
+      const previousBlock = glyphsBlocks.at(i - 1);
+      const nextBlock = glyphsBlocks.at(i + 1);
+      const previousGlyphs = previousBlock?.at(-1);
+      const nextGlyphs = nextBlock?.[0];
       const firstGlyph = glyphs[0];
       const lastGlyph = glyphs[glyphs.length - 1];
-      const graphicWidth = lastGlyph[1] - firstGlyph[1] + lastGlyph[2];
-      const textWidth = textFont.getAdvanceWidth(text, fontSize);
+      let margin = firstGlyph[1] / fontSize - currentLeft;
+      let graphicWidth = lastGlyph[1] - firstGlyph[1] + lastGlyph[2];
+      if (isSpaceOnly) {
+        if (previousGlyphs) {
+          margin = (previousGlyphs[1] + previousGlyphs[2]) / fontSize - currentLeft;
+        }
+        if (nextGlyphs) {
+          if (previousGlyphs) {
+            graphicWidth = nextGlyphs[1] - (previousGlyphs[1] + previousGlyphs[2]);
+          } else {
+            graphicWidth = nextGlyphs[1] - firstGlyph[1];
+          }
+        }
+      } else if (previousGlyphs) {
+        const space = (previousGlyphs[1] + previousGlyphs[2]) / fontSize - currentLeft;
+        if (margin > 0 && margin - space <= 0.1) {
+          graphicWidth += (margin - space) * fontSize;
+          margin = space;
+        }
+      }
       chunks.push({
         text,
-        top: 0,
-        left: firstGlyph[1] * finalFontSize,
-        scale: textWidth > 0 ? graphicWidth / textWidth : 1
+        margin: margin * finalFontSize,
+        scale: graphicWidth / textWidth
       });
+      currentLeft += margin + textWidth / fontSize;
     }
     markedContent.children.push({
       role: "text",
@@ -41201,11 +41238,12 @@ async function createTextLayerV2(page, {
         break;
       case OPS.beginMarkedContent:
       case OPS.beginMarkedContentProps: {
-        const [role, idRef] = args;
+        const [role, idRef, props] = args;
         const id2 = normalizeMarkedContentId(idRef);
         markedContent.children.push({
           id: id2,
           role: role?.toString().toLowerCase() || "span",
+          attrs: props ?? void 0,
           children: []
         });
         markedContent = markedContent.children.at(-1) || rootContainer;
@@ -41370,7 +41408,6 @@ export {
   isWidgetAnnotation,
   loadDefaultFonts2 as loadDefaultFonts,
   loadTextLayerFonts,
-  makeSerializable,
   noContextMenu,
   normalizeUnicode,
   parseRgbaColor,

package/dist/node/worker.js

@@ -78282,7 +78282,67 @@ AnnotationFactory.saveNewAnnotations = async (evaluator, task, annotations, imag
   return data;
 };
 
-// src/lib/Font.ts
+// src/lib/Evaluator.ts
+var collectMarkedContentLanguage = (stream, xref) => {
+  const map = /* @__PURE__ */ new Map();
+  const pos = stream.pos;
+  const preprocessor = new EvaluatorPreprocessor(stream, xref);
+  const operation = {
+    fn: null,
+    args: null
+  };
+  while (preprocessor.read(operation)) {
+    if (operation.fn === null) {
+      operation.args = null;
+      continue;
+    }
+    if (operation.fn !== OPS.beginMarkedContentProps) {
+      operation.args = null;
+      continue;
+    }
+    if (!operation.args?.length) {
+      operation.args = null;
+      continue;
+    }
+    const [, dict] = operation.args.slice();
+    operation.args = null;
+    if (!(dict instanceof Dict)) {
+      continue;
+    }
+    const mcid = dict.get("MCID");
+    if (typeof mcid !== "number") {
+      continue;
+    }
+    const lang = dict.get("Lang");
+    if (!lang || typeof lang !== "string") {
+      continue;
+    }
+    map.set(mcid, lang);
+  }
+  stream.pos = pos;
+  return map;
+};
+var getOperatorList = PartialEvaluator.prototype.getOperatorList;
+PartialEvaluator.prototype.getOperatorList = async function(options, ...args) {
+  const { stream, operatorList } = options;
+  const languages = collectMarkedContentLanguage(stream, this.xref);
+  const addOp = operatorList.addOp;
+  operatorList.addOp = function(fn, args2) {
+    if (fn === OPS.beginMarkedContentProps) {
+      const mcid = args2[1];
+      if (typeof mcid === "number" && languages.has(mcid)) {
+        const lang = languages.get(mcid);
+        if (lang) {
+          args2[2] = {
+            lang: stringToPDFString(lang)
+          };
+        }
+      }
+    }
+    return addOp.call(this, fn, args2);
+  };
+  await getOperatorList.call(this, options, ...args);
+};
 var translateFont = PartialEvaluator.prototype.translateFont;
 PartialEvaluator.prototype.translateFont = async function(options, ...args) {
   const { descriptor } = options;

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.30",
+	"version": "1.0.0-alpha.32",
 	"type": "module",
 	"author": "Chialab <dev@chialab.it>",
 	"license": "MIT",