通过使用jsoup解析html,绘画表格生成execl文件

Posted 2020-09-16

1.获取文件或者字符设置绘画表格字符编码

//得到Document并且设置编码格式
    public static Document getDoc(String fileName) throws IOException{

        File myFile=new File(fileName);
        Document doc= Jsoup.parse(myFile, "UTF-8","");
        return doc;
    }

2.根据解析出来的table进行绘画

public static void mergeColRow(Elements trs,Elements tdcol,WritableSheet sheet) throws RowsExceededException, WriteException{

        int[][] rowhb=new int[trs.size()][tdcol.size()];
        for(int i=0;i<trs.size();i++){
            Element tr=trs.get(i);
            Elements tds=tr.getElementsByTag("td");

            int realColNum=0;
            for(int j=0;j<tds.size();j++){

                Element td=tds.get(j);
                if(rowhb[i][realColNum]!=0){
                    realColNum=getRealColNum(rowhb,i,realColNum);
                }
                int rowspan=1;
                int colspan=1;
                if(td.attr("rowspan")!=""){
                    rowspan = Integer.parseInt(td.attr("rowspan"));
                }
                if(td.attr("colspan")!=""){
                    colspan = Integer.parseInt(td.attr("colspan"));
                }
                String text=td.text();
                drawMegerCell(rowspan,colspan,sheet,realColNum,i,text,rowhb);
                realColNum=realColNum+colspan;
            }
        }
    }
    ///这个方法用于根据样式画出单元格，并且根据rowpan和colspan合并单元格
    public static void drawMegerCell(int rowspan,int colspan,WritableSheet sheet,int realColNum,int realRowNum,String text,int[][] rowhb) throws RowsExceededException, WriteException{

        for(int i=0;i<rowspan;i++){
            for(int j=0;j<colspan;j++){
                if(i!=0||j!=0){
                    text="";
                }
                Label label = new Label(realColNum+j,realRowNum+i,text);
                WritableFont countents = new WritableFont(WritableFont.createFont("微软雅黑"),10); // 设置单元格内容，字号12
                WritableCellFormat cellf = new WritableCellFormat(countents );
                cellf.setAlignment(jxl.format.Alignment.CENTRE);//把水平对齐方式指定为居中
                cellf.setVerticalAlignment(jxl.format.VerticalAlignment.CENTRE);//把垂直对齐方式指定为居
                label.setCellFormat(cellf);
                sheet.addCell(label);
                rowhb[realRowNum+i][realColNum+j]=1;
            }
        }
        sheet.mergeCells(realColNum,realRowNum, realColNum+colspan-1,realRowNum+rowspan-1);
    }

    public static int getRealColNum(int[][] rowhb,int i,int realColNum){
        while(rowhb[i][realColNum]!=0){
            realColNum++;
        }
        return realColNum;
    }

    ///根据colgroups设置表格的列宽
    public static void setColWidth(Elements colgroups,WritableSheet sheet){

        if(colgroups.size()>0){
            Element colgroup=colgroups.get(0);
            Elements cols=colgroup.getElementsByTag("col");
            for(int i=0;i<cols.size();i++){

                Element col=cols.get(i);
                String strwd=col.attr("width");
                if(col.attr("width")!=""){
                    int wd=Integer.parseInt(strwd);
                    sheet.setColumnView(i,wd/8);
                }
            }
        }
    }

3.生成根据绘画的表格生成Excel

public  void toExcel(String fileName, String excelName, HttpServletRequest request)throws IOException{

        //通过解析字符
        Document doc = Jsoup.parse(fileName);
        //根据地址文件进行解析
      // Document doc=getDoc(fileName);
        String title = doc.title();

        ///得到样式，以后可以根据正则表达式解析css，暂且没有找到cssparse
        Elements style= doc.getElementsByTag("style");
        ///得到Table，demo只演示输入一个table，以后可以用循环遍历tables集合输入所有table
        Elements tables= doc.getElementsByTag("TABLE");

        if(tables.size()==0){
            return;
        }
            try {
                //文件保存到classpath目录下面
                String path = request.getSession().getServletContext()
                        .getRealPath("/resource/download/");
                path += excelName + ".xls";
                WritableWorkbook book = Workbook.createWorkbook(new File(path));

                for(int i = 0; i < tables.size();i++){

                    //获取table
                    Element table = tables.get(i);
                    String name = table.attr("value");
                    WritableSheet sheet = book.createSheet( name, i);
                    //得到所有行
                    Elements trs = table.getElementsByTag("tr");
                    Elements tdcol=trs.get(0).getElementsByTag("td");
                    ///得到列宽集合
                    Elements colgroups = table.getElementsByTag("colgroup");
                    setColWidth(colgroups, sheet);
                    mergeColRow(trs, tdcol, sheet);
                }
                book.write();
                book.close();

            } catch (RowsExceededException e) {
                e.printStackTrace();
            } catch (WriteException e) {
                e.printStackTrace();
            }

    }