html转word或pdf

import com.spire.doc.*;
import java.io.*;

public class htmlStringToWord {

    public static void main(String[] args) throws Exception {

        String inputHtml = "InputHtml.txt";
        //新建Document对象
        Document document = new Document();
        //添加section
        Section sec = document.addSection();

        String htmlText = readTextFromFile(inputHtml);
        //添加段落并写入HTML文本
        sec.addParagraph().appendHTML(htmlText);

        //文档另存为PDF
        document.saveToFile("HTMLstringToPDF.pdf", FileFormat.PDF);
    }
    public static String readTextFromFile(String fileName) throws IOException{
        StringBuffer sb = new StringBuffer();
        BufferedReader br = new BufferedReader(new FileReader(fileName));
        String content = null;
        while ((content = br.readLine()) != null) {
            sb.append(content);
        }
        return sb.toString();
    }
}

/**
 * @MethodName: fillReplaceHtml
 * @Description: 填充替换掉html标签后的文本
 * @Date: 2020/12/28 0028 15:02
 * *
 * @author: <achao1441470436@gmail.com>
 * @param: [doc, bookmarkName, data]
 * @returnValue: void
 */
public static void fillReplaceHtml(Document doc, String bookmarkName, String data) {
    // 定位书签
    BookmarksNavigator bookmarkNavigator = new BookmarksNavigator(doc);
    bookmarkNavigator.moveToBookmark(bookmarkName);
    // 创建段落
    Paragraph para = new Paragraph(doc);
    // 创建文本内容
    TextRange textRange = para.appendText(StringUtils.replaceHtml(data));
    // 设置格式
    CharacterFormat format = textRange.getCharacterFormat();
    // 大小
    format.setFontSize(9);
    // 创建文本域
    TextBodyPart bodyPart = new TextBodyPart(doc);
    // 添加段落到文本域
    bodyPart.getBodyItems().add(para);
    // 替换书签内容
    try {
        bookmarkNavigator.replaceBookmarkContent(bodyPart);
    } catch (Exception e) {
        log.error("书签《" + bookmarkName + "》丢失", e);
    }
}