word 转pdf时不保留修订痕迹

许鹏程

2023-08-04 efdccbfcb1f20afba47e0abbd0f1319160d44df2

 src/main/java/com/product/file/util/AsposeUtil.java

@@ -5,17 +5,28 @@
import java.io.FileOutputStream;
import java.io.IOException;
import java.io.InputStream;
import java.math.RoundingMode;
import java.text.DecimalFormat;
import java.util.ArrayList;
import java.util.Iterator;
import java.util.List;

import com.aspose.cells.Cells;
import com.aspose.cells.PdfSaveOptions;
import com.aspose.cells.Workbook;
import com.aspose.cells.Worksheet;
import com.aspose.cells.WorksheetCollection;
import com.aspose.slides.Presentation;
import com.aspose.words.Document;
import com.aspose.words.*;
import com.product.core.exception.BaseException;
import com.product.core.spring.context.SpringMVCContextHolder;
import com.product.file.config.FileCode;

public class AsposeUtil {
    private static int OFFICE_XLS=1;
    private static int OFFICE_WORD=2;
    private static int OFFICE_PPT=3;
    private static boolean OFFICE_XLS_=false;
    private static boolean OFFICE_WORD_=false;
    private static boolean OFFICE_PPT_=false;
   //是否初始了license
   private static boolean isInitLicense=false;
    /**
@@ -30,22 +41,26 @@
        boolean result = true;
        InputStream is = null;
        try {
           is =  AsposeUtil.class.getClassLoader().getResourceAsStream("license.xml");
            com.aspose.cells.License excel = new com.aspose.cells.License();
            excel.setLicense(is);
            
            is =  AsposeUtil.class.getClassLoader().getResourceAsStream("license.xml");
            com.aspose.words.License word = new com.aspose.words.License();
            word.setLicense(is);
            
            is =  AsposeUtil.class.getClassLoader().getResourceAsStream("license.xml");
            com.aspose.slides.License ppt = new com.aspose.slides.License();
            ppt.setLicense(is);

            isInitLicense=true;
           if(type==OFFICE_XLS && !OFFICE_XLS_ ) {
                is = AsposeUtil.class.getClassLoader().getResourceAsStream("license.xml");
                com.aspose.cells.License excel = new com.aspose.cells.License();
                excel.setLicense(is);
                OFFICE_XLS_=true;
            }else if(type==OFFICE_WORD && !OFFICE_WORD_ ){
                is =  AsposeUtil.class.getClassLoader().getResourceAsStream("license.xml");
                com.aspose.words.License word = new com.aspose.words.License();
                word.setLicense(is);
                OFFICE_WORD_=true;
            }else if(type==OFFICE_PPT && ! OFFICE_PPT_) {
                is =  AsposeUtil.class.getClassLoader().getResourceAsStream("license.xml");
                com.aspose.slides.License ppt = new com.aspose.slides.License();
                ppt.setLicense(is);
                OFFICE_PPT_=true;
            }
        } catch (Exception e) {
            e.printStackTrace();
           result=false;
           throw new BaseException(e);
           throw new BaseException(FileCode.DOC_CONVERT_FALL,e);
        }finally{
         if(is!=null) {
            try {
@@ -58,7 +73,7 @@
        return result;
    }
    /**
     * 
     *
     * @param officePath
     * @param OutPutPath
     * @throws BaseException
@@ -71,18 +86,17 @@
        FileOutputStream fileOS=null;
        try {
            File file = new File(OutPutPath);
            if (!file.exists()) {
                file.mkdirs();
            if (!file.getParentFile().exists()) {
                file.getParentFile().mkdirs();
            }
            Workbook wb = new Workbook(officePath);// 原始excel路径
            File pdfFile = new File(OutPutPath);// 输出路径
            fileOS = new FileOutputStream(pdfFile);
            fileOS = new FileOutputStream(OutPutPath);
            //wb.save(fileOS, com.aspose.cells.SaveFormat.PDF);
            PdfSaveOptions pdfSaveOptions = new PdfSaveOptions();
            pdfSaveOptions.setAllColumnsInOnePagePerSheet(true);
            wb.save(fileOS, pdfSaveOptions);
        } catch (Exception e) {
           throw new BaseException(e);
           throw new BaseException(FileCode.DOC_CONVERT_FALL,e);
        }finally{
         if(fileOS!=null) {
            try {
@@ -95,7 +109,98 @@
      }
    }
    /**
     * 
     *
     * @param officePath
     * @param OutPutPath
     * @throws BaseException
     */
    public static void xls2xlsx(String officePath,String OutPutPath,String officeType)throws BaseException {
        // 验证License
        if (!getLicense(1)) {
            return;
        }
        if("xlsx".equals(officeType)) {
           FileUtil.copyFile(new File(officePath), OutPutPath);
           return ;
        }
        FileOutputStream fileOS=null;
        try {
            File file = new File(OutPutPath);
            if (!file.getParentFile().exists()) {
                file.getParentFile().mkdirs();
            }
            Workbook wb = new Workbook(officePath);// 原始excel路径
            fileOS = new FileOutputStream(OutPutPath);
            wb.save(fileOS, com.aspose.cells.SaveFormat.XLSX);
        } catch (Exception e) {
           throw new BaseException(FileCode.DOC_CONVERT_FALL,e);
        }finally{
         if(fileOS!=null) {
            try {
               fileOS.flush();
               fileOS.close();
            }catch(IOException e) {
               throw new BaseException(e);
            }
         }
      }
    }
    public static void Excel2Html(String officePath,String OutPutPath)throws BaseException {
        // 验证License
        if (!getLicense(1)) {
            return;
        }
        FileOutputStream fileOS=null;
        try {
            File file = new File(OutPutPath);
            if (!file.getParentFile().exists()) {
                file.getParentFile().mkdirs();
            }
            Workbook wb = new Workbook(officePath);// 原始excel路径
            WorksheetCollection sheets=wb.getWorksheets();
            System.out.println("sheet个数："+sheets.getCount());
            for(int i=0;i<sheets.getCount();i++) {
               //setAutoWithHeight(sheets.get(i));
            }
//            HtmlSaveOptions options = new HtmlSaveOptions();
//            options.setPresentationPreference(true);
            fileOS = new FileOutputStream(OutPutPath);
            wb.save(fileOS, com.aspose.cells.SaveFormat.HTML);
        } catch (Exception e) {
           throw new BaseException(FileCode.DOC_CONVERT_FALL,e);
        }finally{
         if(fileOS!=null) {
            try {
               fileOS.flush();
               fileOS.close();
            }catch(IOException e) {
               throw new BaseException(e);
            }
         }
      }
    }
    /**
     * 设置单元格的自动适宽、高
     * @param sheet
     * @throws Exception
     */
    public static void setAutoWithHeight(Worksheet sheet)throws Exception
    {
        Cells cells =sheet.getCells();
        int columnCount = cells.getMaxColumn();  //获取表页的最大列数
        int rowCount = cells.getMaxRow();        //获取表页的最大行数
        if(columnCount<0 ||rowCount<0 )return ;
        for (int col = 0; col < columnCount; col++)
        {
            sheet.autoFitColumn(col, 0, rowCount);
        }
        for (int row = 0; row < rowCount; row++)
        {
            sheet.autoFitRow(row, 0, columnCount);
        }
    }
    /**
     *
     * @param officePath
     * @param OutPutPath
     * @throws BaseException
@@ -112,11 +217,12 @@
                file.getParentFile().mkdirs();
            }
            Document doc = new Document(officePath);// 原始word路径
            File pdfFile = new File(OutPutPath);// 输出路径
            fileOS = new FileOutputStream(pdfFile);
            fileOS = new FileOutputStream(OutPutPath);
         //不保留修订痕迹
         doc.acceptAllRevisions();
            doc.save(fileOS, com.aspose.words.SaveFormat.PDF);
        } catch (Exception e) {
           throw new BaseException(e);
           throw new BaseException(FileCode.DOC_CONVERT_FALL,e);
        }finally{
         if(fileOS!=null) {
            try {
@@ -129,7 +235,7 @@
      }
    }
    /**
     * 
     *
     * @param officePath
     * @param OutPutPath
     * @throws BaseException
@@ -142,18 +248,15 @@
        FileOutputStream fileOS=null;
        try {
            File PathFile = new File(OutPutPath);
            if (!PathFile.exists()) {
                PathFile.mkdirs();
            if (!PathFile.getParentFile().exists()) {
                PathFile.getParentFile().mkdirs();
            }
            InputStream slides = new FileInputStream(new File(officePath));// 原始ppt路径
            InputStream slides = new FileInputStream(officePath);// 原始ppt路径
            Presentation pres = new Presentation(slides);
            File file = new File(OutPutPath);// 输出pdf路径
            fileOS = new FileOutputStream(file);
            fileOS = new FileOutputStream(OutPutPath);
            pres.save(fileOS, com.aspose.slides.SaveFormat.Pdf);
            fileOS.flush();
            fileOS.close();
        } catch (Exception e) {
           throw new BaseException(e);
           throw new BaseException(FileCode.DOC_CONVERT_FALL,e);
        }finally{
         if(fileOS!=null) {
            try {
@@ -173,35 +276,27 @@
     * @throws BaseException
     */
    public static String OfficeToPdf(String officePath,String outFile,String officeType )throws BaseException {

       officeType=officeType.toLowerCase();
        SpringMVCContextHolder.getSystemLogger().info("The file of office type:"+officePath);
        SpringMVCContextHolder.getSystemLogger().info("The file of PDF type:"+outFile);
        if("pdf".equals(officeType)) {
           return outFile;
        }
        File file = new File(officePath);
        //判断当前office文件是否已经转为PDF,如果已转为PDF就不需要再次转换。
        if (file.exists()&& !"pdf".equals(officeType)) {
            double bytes = file.length();
            double kilobytes = (bytes / 1024);
            double megabytes = (kilobytes / 1024);

            DecimalFormat df = new DecimalFormat("0.00");
            df.setRoundingMode(RoundingMode.HALF_UP);
            String MB = df.format(megabytes);
            Double Size = Double.parseDouble(MB);
//            if(Size>10){
            SpringMVCContextHolder.getSystemLogger().info("The file size:"+Size+"MB");
//            }
            //"doc", "docx", "xls","xlsx", "ppt", "pptx"
            if(officeType.equals("doc")||officeType.equals("docx")){
        if (file.exists()) {
           if(officeType.equals("doc")||officeType.equals("docx")){
                Word2Pdf(officePath,outFile);
            }else if(officeType.equals("xls")||officeType.equals("xlsx")){
                Excel2Pdf(officePath,outFile);
//                Excel2Pdf(officePath,outFile);
                Excel2Html(officePath,outFile);
//                xls2xlsx(officePath,outFile,officeType);
            }else if(officeType.equals("ppt")||officeType.equals("pptx")){
                PPT2Pdf(officePath,outFile);
            }else{
               SpringMVCContextHolder.getSystemLogger().info("File types that cannot be converted:"+officeType);
               throw new BaseException(FileCode.DOC_CONVERT_NOT_SUPPORT_FILE_FALL);
            }
        } else {
           SpringMVCContextHolder.getSystemLogger().info("File Not Exist");
           throw new BaseException(FileCode.DOC_CONVERT_FILE_NOT_EXIST_FALL);
        }
        return outFile;
    }
@@ -212,43 +307,26 @@
     */
    public static String OfficeToPdf(String officePath) {

        //G:/product/WebApp/fwis_develop/com/is/flywings/oa/attchfile/1000000000/i0002/101951.docx⌒101951.docx⌒feiyu.docx
        String[] split = officePath.split("⌒");
        int lastIndex = split[0].lastIndexOf(".");
        int lastNameIndex = split[0].lastIndexOf("/");

        String officeType = split[0].substring(lastIndex+1);
        String officeName = split[0].substring(lastNameIndex+1,lastIndex)+".pdf";
        String officeName = split[0].substring(lastNameIndex+1,lastIndex);
        String OutPutPath = split[0].substring(0,lastNameIndex+1)+"office/";
        System.out.println("输出目录："+OutPutPath);
        File file = new File(split[0]);
        File pdfFile = new File(OutPutPath+officeName);
        //判断当前office文件是否已经转为PDF,如果已转为PDF就不需要再次转换。
        if(pdfFile.exists()){
            return OutPutPath+officeName;
        }

        if (file.exists()) {

            double bytes = file.length();
            double kilobytes = (bytes / 1024);
            double megabytes = (kilobytes / 1024);

            DecimalFormat df = new DecimalFormat("0.00");
            df.setRoundingMode(RoundingMode.HALF_UP);
            String MB = df.format(megabytes);
            Double Size = Double.parseDouble(MB);
            if(Size>10){
                return Size+"MB";
            }
            //"doc", "docx", "xls","xlsx", "ppt", "pptx"
            try {
                if(officeType.equals("doc")||officeType.equals("docx")){
                    Word2Pdf(split[0],OutPutPath+"/"+officeName);
                    Word2Pdf(split[0],OutPutPath+"/"+officeName+".pdf");
                }else if(officeType.equals("xls")||officeType.equals("xlsx")){
                    Excel2Pdf(split[0],OutPutPath+"/"+officeName);
//                   Excel2Html(split[0],OutPutPath+"/"+officeName+".html");
                   xls2xlsx(split[0],OutPutPath+"/"+officeName+".xlsx",officeType);
                }else if(officeType.equals("ppt")||officeType.equals("pptx")){
                    PPT2Pdf(split[0],OutPutPath+"/"+officeName);
                    PPT2Pdf(split[0],OutPutPath+"/"+officeName+".pdf");
                }else{
                    System.out.println("无法识别该文件！");
                    return "Error";
@@ -261,9 +339,253 @@
        }
        return OutPutPath+officeName;
    }
    
    /**
     * word文档的合并
     * @param targetWordFile
     * @param sourceWordFile
     * @return
     * @throws Exception
     */
    public static File mergeWord(String targetWordFile,String sourceWordFile ) throws BaseException{
       File target=new File(targetWordFile);
       File source=new File(sourceWordFile);
       return mergeWord(target,source);
    }
    /**
     * word文档的合并
     * @param
     * @param
     * @return
     * @throws Exception
     */
    public static File mergeWord(File target,File source ) throws BaseException{

       if(!target.exists() && !source.exists()) {
          return null;
       }else if(!source.exists() && target.exists()) {
          return target;
       }else if(source.exists() && !target.exists()) {
          return source;
       }
       // 验证License
        if (!getLicense(OFFICE_WORD)) {
            return null;
        }
       FileInputStream tin=null;
       FileInputStream sin=null;
       FileOutputStream out=null;
       String temp=target.getParent()+File.separator+"mergeTemp";
       File temp_target=new File(temp+File.separator+target.getName());
       try {
          tin=new FileInputStream(target);
          sin=new FileInputStream(source);
          Document targetDoc = new Document(tin);
          Document sourceDoc = new Document(sin);
//          docBlankProcess(targetDoc);
          targetDoc.appendDocument(sourceDoc, ImportFormatMode.KEEP_SOURCE_FORMATTING);
//          NodeCollection t =sourceDoc.getChildNodes();
//          for(int c=0;c<t.getCount();c++) {
//             targetDoc.getChildNodes().add(t.get(c));
//          }
          File dir=new File(temp);
          if(!dir.exists()) {
             dir.mkdirs();
          }
          if(temp_target.exists()) {
             temp_target.delete();
          }
          if(temp_target.createNewFile()) {
            out=new FileOutputStream(temp_target);
            targetDoc.save(out, SaveFormat.DOCX);
         }
       }catch(Exception e) {
          e.printStackTrace();
          throw new BaseException(FileCode.DOC_MERGE_FALL,e);
       }
       finally{
          try {
             if(tin!=null) {
               tin.close();
            }
             if(sin!=null) {
                sin.close();
            }
            if(out!=null) {
               out.flush();
                out.close();
            }
          }catch(IOException e) {
            throw new BaseException(FileCode.DOC_MERGE_FALL,e);
         }
      }
       return temp_target;
    }

    public static void docBlankProcess(Document targetDoc) {
       NodeCollection nodes=targetDoc.getChildNodes(NodeType.PARAGRAPH, true);
       System.out.println(nodes.getCount());
       List <Node> ss=new ArrayList<>();
       for(Paragraph para :(Iterable<Paragraph>) nodes) {
          System.out.println("===========================");
          System.out.println(para.isEndOfSection());
          System.out.println(para.hasChildNodes());
          String text=para.getText();
          if("\r".equals(text))System.out.println("\\\\r");
          else if("\f".equals(text))System.out.println("\\\\f");
          else if("".equals(text.trim())) System.out.println("空格");
          else System.out.println(text);

          NodeCollection pnodes =para.getChildNodes();
          if(pnodes.getCount()==0)para.appendChild(new Run(targetDoc,"a"));
          if(para.isEndOfSection())para.appendChild(new Run(targetDoc,"B"));
          System.out.println("sub nodes:"+pnodes.getCount());
          for(int i=0;i<pnodes.getCount();i++) {
             Node d=pnodes.get(i);
             System.out.println("type:"+NodeType.getName(d.getNodeType()));

             if("\r".equals(text))System.out.println("\\\\r");
              else if("\f".equals(text))System.out.println("\\\\f");
              else if("".equals(text.trim())) System.out.println("空格");
              else System.out.println(text);
             if("BOOKMARK_END".equals(NodeType.getName(d.getNodeType()))) {
                BookmarkEnd b=(BookmarkEnd)d;
             }
             if(d.getNodeType()==NodeType.SHAPE) {
                ss.add(d);
             }
          }


       }
       for(Node d:ss) {
          d.remove();
       }

    }

    private static String appendDoc(String sohuyepath, String xuyepath, boolean includeSection) throws Exception {
        // 验证License
        if (!getLicense(OFFICE_WORD)) {
            return null ;
        }
        Document dstDoc = new Document(sohuyepath);
        Document srcDoc = new Document(xuyepath);
        System.out.println("目标文档页数："+dstDoc.getPageCount()+" | 源文档页数："+srcDoc.getPageCount());
        if (includeSection) {
         Iterator<Section> var3 = srcDoc.getSections().iterator();
         while (var3.hasNext()) {
            Section srcSection = (Section) var3.next();
            Node dstNode = dstDoc.importNode(srcSection, true, 0);
            dstDoc.appendChild(dstNode);
         }
      } else {
         Node node = dstDoc.getLastSection().getBody().getLastParagraph();
         if (node == null) {
            node = new Paragraph(srcDoc);
            dstDoc.getLastSection().getBody().appendChild(node);
         }

         if (node.getNodeType() != 8 & node.getNodeType() != 5) {
            throw new Exception("Use appendDoc(dstDoc, srcDoc, true) instead of appendDoc(dstDoc, srcDoc, false)");
         }

         insertDocumentAfterNode(node, dstDoc, srcDoc);
      }
        File f=new File(sohuyepath);
        dstDoc.save(f.getParent()+"/merge/"+f.getName());
        return f.getParent()+"/merge/"+f.getName();
   }

   private static void insertDocumentAfterNode(Node insertAfterNode, Document mainDoc, Document srcDoc)
         throws Exception {
      if (insertAfterNode.getNodeType() != 8 & insertAfterNode.getNodeType() != 5) {
         throw new Exception("The destination node should be either a paragraph or table.");
      } else {
         CompositeNode dstStory = insertAfterNode.getParentNode();

         while (null != srcDoc.getLastSection().getBody().getLastParagraph()
               && !srcDoc.getLastSection().getBody().getLastParagraph().hasChildNodes()) {
            srcDoc.getLastSection().getBody().getLastParagraph().remove();
         }

         NodeImporter importer = new NodeImporter(srcDoc, mainDoc, 1);
         int sectCount = srcDoc.getSections().getCount();

         for (int sectIndex = 0; sectIndex < sectCount; ++sectIndex) {
            Section srcSection = srcDoc.getSections().get(sectIndex);
            int nodeCount = srcSection.getBody().getChildNodes().getCount();

            for (int nodeIndex = 0; nodeIndex < nodeCount; ++nodeIndex) {
               Node srcNode = srcSection.getBody().getChildNodes().get(nodeIndex);
               Node newNode = importer.importNode(srcNode, true);
               dstStory.insertAfter(newNode, insertAfterNode);
               insertAfterNode = newNode;
            }
         }

      }
   }

    /**
     * 把源文档内容放到最后一页的页末
     * @param sohuyepath
     * @return
     * @throws Exception
     */
    private static void  appendDocToPageTail(String sohuyepath) throws Exception {
        Document dstDoc = new Document(sohuyepath);
        int destPages=dstDoc.getPageCount();
        System.out.println("合并合的最后总页数："+destPages);
        Node lastNode=dstDoc.getLastSection().getBody().getLastChild();
        Paragraph lp=null;
        if(lastNode.getNodeType()==NodeType.PARAGRAPH){
            lp=(Paragraph)lastNode;
            Node ld=lp.getFirstChild();
            while(ld!=null && ld.getNodeType()!=NodeType.SHAPE && ld.getNodeType()!=NodeType.RUN &&  ld.getNodeType()!=NodeType.TABLE){
                System.out.println("type name "+NodeType.getName(ld.getNodeType()));
                ld=ld.getPreviousSibling();
            }
        }
        System.out.println("最后一节点："+NodeType.getName(lastNode.getNodeType()));
        for(int i=0;i<40;i++) {
            Run t = new Run(dstDoc, "======\r");
            if(lp!=null){
                lp.insertBefore(t,lp.getLastChild());
                System.out.println("子元素数："+lp.getChildNodes().getCount());
            }
            System.out.println("合并合的最后总页数："+dstDoc.getPageCount());
            if(destPages < dstDoc.getPageCount()){
                System.out.println("================分页=================");
            }
        }
        File ff=new File(sohuyepath);
        dstDoc.save(ff.getParent()+"/merge/"+ff.getName(),SaveFormat.DOCX);
    }

    public static void main(String args[]) {
       String file="E:/work/沄崃/FE6.6产品报价工具-新 - 副本.xlsx";
       AsposeUtil.OfficeToPdf(file);
//       String file="E:/工作内容.xlsx";
//       String file="E:/爆破母线.xls";
//       String file="E:/采集提取日志表结构.xlsx";
       String file="E:/FE信息系统国产化解决方案V1.0.docx";
//       AsposeUtil.OfficeToPdf(file);

//       AsposeUtil.mergeWord("E:\\QQ\\住房担保置换公司.docx","E:\\QQ\\住房担保置换公司2.docx");
//       try {
//          processDocPage();
//       }catch(Exception e) {
//          e.printStackTrace();
//       }
       try {
//         String sohuyepath="E:/QQ/住房担保置换公司.docx";
//         String xuyepath="E:/QQ/FE信息系统国产化解决方案V1.0.docx";
//         String npath=appendDoc(sohuyepath, xuyepath, false);
         String tail="E:/QQ/住房担保置换公司2.docx";
//            appendDoc(npath, tail, false);
            appendDocToPageTail(tail);


      }catch(Exception e) {
         e.printStackTrace();
      }
    }
}