文档处理

shichongfu

2023-06-19 d67cc1b3c08a2a7c5212e51c9776d095b4503b0a

文档处理

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

@@ -5,29 +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.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.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;
    /**
@@ -42,20 +41,24 @@
        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(FileCode.DOC_CONVERT_FALL,e);
        }finally{
@@ -283,8 +286,8 @@
                Word2Pdf(officePath,outFile);
            }else if(officeType.equals("xls")||officeType.equals("xlsx")){
//                Excel2Pdf(officePath,outFile);
//                Excel2Html(officePath,outFile);
                xls2xlsx(officePath,outFile,officeType);
                Excel2Html(officePath,outFile);
//                xls2xlsx(officePath,outFile,officeType);
            }else if(officeType.equals("ppt")||officeType.equals("pptx")){
                PPT2Pdf(officePath,outFile);
            }else{
@@ -348,8 +351,8 @@
    }
    /**
     * word文档的合并
     * @param targetWordFile
     * @param sourceWordFile
     * @param
     * @param
     * @return
     * @throws Exception
     */
@@ -363,7 +366,7 @@
            return source;
        }
        // 验证License
        if (!getLicense(2)) {
        if (!getLicense(OFFICE_WORD)) {
            return null;
        }
        FileInputStream tin=null;
@@ -376,15 +379,11 @@
            sin=new FileInputStream(source);
            Document targetDoc = new Document(tin);
            Document sourceDoc = new Document(sin); 
            docBlankProcess(targetDoc);
//            docBlankProcess(targetDoc);
            targetDoc.appendDocument(sourceDoc, ImportFormatMode.KEEP_SOURCE_FORMATTING);
	    	
  
//            for(Section sect : targetDoc.getSections())
//            {
//                if (sect.getBody().getParagraphs().getCount() == 0) {
//                    sect.remove();
//                }
//            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()) {
@@ -398,6 +397,7 @@
                targetDoc.save(out, SaveFormat.DOCX);
            }
        }catch(Exception e) {
            e.printStackTrace();
            throw new BaseException(FileCode.DOC_MERGE_FALL,e);
        }
        finally{
@@ -422,46 +422,168 @@
    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("rrrrrrrrrrrr");
            }else if("\f".equals(text)) {
                System.out.println("ffffffffffff");
            }else {
                System.out.println(text);
            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);
                }
            }
            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();
            }
        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:/工作内容.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);
        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();
        }
    }
}