如何使用MuPDF.js提取PDF高亮批注对应的文本？

我用MuPDF.js提取PDF批注时，能检测到高亮批注类型，但annot.hasRect()返回false，annot.getContents()也拿不到高亮对应的文本，原代码如下：

import * as fs from "fs";
import * as mupdfjs from "mupdf/mupdfjs";

async function extractAnnotations() {
  try {
    // Read the PDF file as a buffer
    let fileData = fs.readFileSync("Example.pdf");

    // Open the document using mupdfjs.PDFDocument.openDocument
    let document = await mupdfjs.PDFDocument.openDocument(
      fileData,
      "application/pdf"
    );

    // Loop through all pages of the document
    let pageCount = document.countPages(); // Get total number of pages
    let i = 0;

    while (i < pageCount) {
      const page = new mupdfjs.PDFPage(document, i);
      const annots = page.getAnnotations();

      annots.forEach((annot, index) => {
        console.log(`Annotation ${index + 1}:`);
        console.log(`  Has Rect: ${annot.hasRect()}`);
        console.log(`  Contents: ${annot.getContents()}`);
        
      });

      i++;
    }
  } catch (err) {
    console.error("Error extracting annotations:", err);
  }
}

extractAnnotations();

解决方法 #

高亮批注的文本提取需要注意两个关键点：

• 高亮批注的区域不是单一矩形，而是用四边形集合存储，需要用annot.getQuadPoints()获取，而非getRect()
• 需要调用页面的getTextInQuad()方法，传入每个四边形区域来提取对应文本

修改后的完整代码：

import * as fs from "fs";
import * as mupdfjs from "mupdf/mupdfjs";

async function extractAnnotations() {
  try {
    let fileData = fs.readFileSync("Example.pdf");
    let document = await mupdfjs.PDFDocument.openDocument(
      fileData,
      "application/pdf"
    );

    let pageCount = document.countPages();
    let i = 0;

    while (i < pageCount) {
      const page = new mupdfjs.PDFPage(document, i);
      const annots = page.getAnnotations();

      annots.forEach((annot, index) => {
        // 仅处理高亮类型批注
        if (annot.getType() === mupdfjs.PDFAnnotation.Type.Highlight) {
          console.log(`Highlight Annotation ${index + 1}:`);
          
          // 获取高亮的所有四边形区域
          const quadPoints = annot.getQuadPoints();
          if (quadPoints.length > 0) {
            let highlightedText = "";
            // 遍历每个四边形，提取对应文本
            quadPoints.forEach(quad => {
              const text = page.getTextInQuad(quad);
              highlightedText += text + " ";
            });
            console.log(`  Highlighted Text: ${highlightedText.trim()}`);
          } else {
            console.log("  No quad points found for this highlight.");
          }
        } else {
          // 其他类型批注的原有处理逻辑
          console.log(`Annotation ${index + 1} (Type: ${annot.getType()}):`);
          console.log(`  Has Rect: ${annot.hasRect()}`);
          console.log(`  Contents: ${annot.getContents()}`);
        }
      });

      i++;
    }
  } catch (err) {
    console.error("Error extracting annotations:", err);
  }
}

extractAnnotations();

关键说明 #

• annot.getType()用于判断批注类型，mupdfjs.PDFAnnotation.Type.Highlight是MuPDF.js定义的高亮批注常量
• getQuadPoints()返回的是PDFQuad对象数组，每个对象代表高亮区域的一个四边形
• page.getTextInQuad(quad)会提取该四边形覆盖的页面文本，将多个四边形的文本拼接即可得到完整的高亮内容

内容的提问来源于stack exchange，提问作者Joe