文档处理

shichongfu

2023-06-15 d873c581e95fdabe4f1a32d95e283a1e1c20c980

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

@@ -8,12 +8,24 @@
import java.math.RoundingMode;
import java.text.DecimalFormat;

import com.aspose.cells.Cells;
import com.aspose.cells.HtmlSaveOptions;
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.ImportFormatMode;
import com.aspose.words.NodeCollection;
import com.aspose.words.NodeType;
import com.aspose.words.Paragraph;
import com.aspose.words.SaveFormat;
import com.aspose.words.Section;
import com.aspose.words.SectionStart;
import com.product.core.exception.BaseException;
import com.product.core.spring.context.SpringMVCContextHolder;
import com.product.file.config.FileCode;

public class AsposeUtil {
   //是否初始了license
@@ -45,7 +57,7 @@
            isInitLicense=true;
        } catch (Exception e) {
           result=false;
           throw new BaseException(e);
           throw new BaseException(FileCode.DOC_CONVERT_FALL,e);
        }finally{
         if(is!=null) {
            try {
@@ -71,18 +83,17 @@
        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 {
@@ -93,6 +104,97 @@
            }
         }
      }
    }
    /**
     * 
     * @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);
        }
    }
    /**
     * 
@@ -112,11 +214,10 @@
                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);
            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 {
@@ -142,18 +243,15 @@
        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 +271,27 @@
     * @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 +302,26 @@
     */
    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 +334,134 @@
        }
        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 targetWordFile
     * @param sourceWordFile
     * @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(2)) {
            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);
	    	
  
//          for(Section sect : targetDoc.getSections())
//          {
//              if (sect.getBody().getParagraphs().getCount() == 0) {
//                  sect.remove();
//              }
//          }
          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) {
          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());
       for(Paragraph para :(Iterable<Paragraph>) nodes) {
          String text=para.getText();
          if("\r".equals(text)) {
             System.out.println("rrrrrrrrrrrr");
          }else if("\f".equals(text)) {
             System.out.println("ffffffffffff");
          }else {
             System.out.println(text);
          }
          if ("\r".equals(text) 
                //|| "\f".equals(text) 
                ) {
             para.remove();
          }
       }
    }
    public static void docProcess(Document targetDoc) {
       NodeCollection nodes=targetDoc.getChildNodes();
       for(Paragraph para :(Iterable<Paragraph>) nodes) {
          String text=para.getText();
          if("\r".equals(text)) {
             System.out.println("rrrrrrrrrrrr");
          }else if("\f".equals(text)) {
             System.out.println("ffffffffffff");
          }else {
             System.out.println(text);
          }
          if ("\r".equals(text) 
                //|| "\f".equals(text) 
                ) {
             para.remove();
          }
       }
    }
    public static void main(String args[]) {
       String file="E:/work/沄崃/FE6.6产品报价工具-新 - 副本.xlsx";
       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");
    }
}