POI Excel 合并数据相同的行

import java.io.Serializable;

/**
 * POI Excel报表导出,列合并实体<br>
 * 
 * @author WQ
 *
 */
public class PoiModel implements Serializable{
    /**
     * 
     */
    private static final long serialVersionUID = 1L;

    private String content;  
      
    private String oldContent;  
  
    private int rowIndex;  
  
    private int cellIndex;

    public String getContent() {
        return content;
    }

    public void setContent(String content) {
        this.content = content;
    }

    public String getOldContent() {
        return oldContent;
    }

    public void setOldContent(String oldContent) {
        this.oldContent = oldContent;
    }

    public int getRowIndex() {
        return rowIndex;
    }

    public void setRowIndex(int rowIndex) {
        this.rowIndex = rowIndex;
    }

    public int getCellIndex() {
        return cellIndex;
    }

    public void setCellIndex(int cellIndex) {
        this.cellIndex = cellIndex;
    }

    public PoiModel() {
    }

    public PoiModel(String content, String oldContent, int rowIndex,
            int cellIndex) {
        this.content = content;
        this.oldContent = oldContent;
        this.rowIndex = rowIndex;
        this.cellIndex = cellIndex;
    }

    @Override
    public String toString() {
        return "PoiModel [content=" + content + ", oldContent=" + oldContent
                + ", rowIndex=" + rowIndex + ", cellIndex=" + cellIndex + "]";
    }  
  
}

 

 

import java.io.File;
import java.io.FileOutputStream;
import java.io.IOException;
import java.util.ArrayList;
import java.util.Date;
import java.util.HashMap;
import java.util.Iterator;
import java.util.List;
import java.util.Map;

import org.apache.poi.ss.usermodel.Cell;
import org.apache.poi.ss.usermodel.Row;
import org.apache.poi.ss.usermodel.Sheet;
import org.apache.poi.ss.usermodel.Workbook;
import org.apache.poi.ss.util.CellRangeAddress;
import org.apache.poi.xssf.usermodel.XSSFWorkbook;

public class Demo {

    /** 
     * @param title 标题集合 tilte的长度应该与list中的model的属性个数一致 
     * @param maps 内容集合 
     * @param mergeIndex 合并单元格的列 
     */  
    public static String createExcel(String[] title, Map<String, List<Map<String, String>>> maps, int[] mergeIndex){  
        if (title.length==0){  
            return null;  
        }  
        /*初始化excel模板*/  
        Workbook workbook = new XSSFWorkbook();  
        Sheet sheet = null;  
        int n = 0;  
        for(Map.Entry<String, List<Map<String, String>>> entry : maps.entrySet()){  
            try {  
                sheet = workbook.createSheet();  
                workbook.setSheetName(n, entry.getKey());  
                workbook.setSelectedTab(0);  
            }catch (Exception e){  
                e.printStackTrace();  
            }  
            /*初始化head,填值标题行(第一行)*/  
            Row row0 = sheet.createRow(0);  
            for(int i = 0; i<title.length; i++){  
                /*创建单元格,指定类型*/  
                Cell cell_1 = row0.createCell(i, Cell.CELL_TYPE_STRING);  
                cell_1.setCellValue(title[i]);  
            }  
            /*得到当前sheet下的数据集合*/  
            List<Map<String, String>> list = entry.getValue();  
            /*遍历该数据集合*/  
            List<PoiModel> poiModels = new ArrayList();  
            if(null!=workbook){  
                Iterator iterator = list.iterator();  
                int index = 1;
                while (iterator.hasNext()){  
                    Row row = sheet.createRow(index);  
                    /*取得当前这行的map,该map中以key,value的形式存着这一行值*/  
                    Map<String, String> map = (Map<String, String>)iterator.next();  
                    /*循环列数,给当前行塞值*/  
                    for(int i = 0; i<title.length; i++){  
                        String old = "";  
                        /*old存的是上一行统一位置的单元的值,第一行是最上一行了,所以从第二行开始记*/  
                        if(index > 1){  
                            old = poiModels.get(i)==null?"":poiModels.get(i).getContent();  
                        }  
                        /*循环需要合并的列*/  
                        for(int j = 0; j < mergeIndex.length; j++){  
                            if(index == 1){  
                                /*记录第一行的开始行和开始列*/  
                                PoiModel poiModel = new PoiModel();  
                                poiModel.setOldContent(map.get(title[i]));  
                                poiModel.setContent(map.get(title[i]));  
                                poiModel.setRowIndex(1);  
                                poiModel.setCellIndex(i);  
                                poiModels.add(poiModel);  
                                break;  
                            }else if(i > 0 && mergeIndex[j] == i){/*这边i>0也是因为第一列已经是最前一列了,只能从第二列开始*/  
                                /*当前同一列的内容与上一行同一列不同时,把那以上的合并, 或者在当前元素一样的情况下,前一列的元素并不一样,这种情况也合并*/  
                                /*如果不需要考虑当前行与上一行内容相同,但是它们的前一列内容不一样则不合并的情况,把下面条件中
                  ||poiModels.get(i).getContent().equals(map.get(title[i])) && !poiModels.get(i - 1).getOldContent().equals(map.get(title[i-1]))去掉就行
*/ if(!poiModels.get(i).getContent().equals(map.get(title[i])) || poiModels.get(i).getContent().equals(map.get(title[i]))
                    && !poiModels.get(i - 1).getOldContent().equals(map.get(title[i-1]))){ /*当前行的当前列与上一行的当前列的内容不一致时,则把当前行以上的合并*/ CellRangeAddress cra=new CellRangeAddress(poiModels.get(i).getRowIndex(), index - 1, poiModels.get(i).getCellIndex(), poiModels.get(i).getCellIndex()); //在sheet里增加合并单元格 sheet.addMergedRegion(cra); /*重新记录该列的内容为当前内容,行标记改为当前行标记,列标记则为当前列*/ poiModels.get(i).setContent(map.get(title[i])); poiModels.get(i).setRowIndex(index); poiModels.get(i).setCellIndex(i); } } /*处理第一列的情况*/ if(mergeIndex[j] == i && i == 0 && !poiModels.get(i).getContent().equals(map.get(title[i]))){ /*当前行的当前列与上一行的当前列的内容不一致时,则把当前行以上的合并*/ CellRangeAddress cra=new CellRangeAddress(poiModels.get(i).getRowIndex(), index - 1, poiModels.get(i).getCellIndex(), poiModels.get(i).getCellIndex()); //在sheet里增加合并单元格 sheet.addMergedRegion(cra); /*重新记录该列的内容为当前内容,行标记改为当前行标记*/ poiModels.get(i).setContent(map.get(title[i])); poiModels.get(i).setRowIndex(index); poiModels.get(i).setCellIndex(i); } /*最后一行没有后续的行与之比较,所有当到最后一行时则直接合并对应列的相同内容*/ if(mergeIndex[j] == i && index == list.size()){ CellRangeAddress cra=new CellRangeAddress(poiModels.get(i).getRowIndex(), index, poiModels.get(i).getCellIndex(), poiModels.get(i).getCellIndex()); //在sheet里增加合并单元格 sheet.addMergedRegion(cra); } } Cell cell = row.createCell(i, Cell.CELL_TYPE_STRING); cell.setCellValue(map.get(title[i])); /*在每一个单元格处理完成后,把这个单元格内容设置为old内容*/ poiModels.get(i).setOldContent(old); } index++; } } n++; } /*生成临时文件*/ FileOutputStream out = null; String localPath = null; File tempFile = null; String fileName = String.valueOf(new Date().getTime()/1000); try { tempFile = File.createTempFile(fileName, ".xlsx"); localPath = tempFile.getAbsolutePath(); out = new FileOutputStream(localPath); workbook.write(out); }catch (IOException e){ e.printStackTrace(); }finally { try { out.flush(); out.close(); }catch (IOException e){ e.printStackTrace(); } } return localPath; } public static void main(String[] args) throws IOException{ /*此处标题的数组则对应excel的标题*/ String[] title = {"id","标题","描述","负责人","开始时间"}; List<Map<String, String>> list = new ArrayList(); /*这边是制造一些数据,注意每个list中map的key要和标题数组中的元素一致*/ for(int i = 0; i<10; i++){ HashMap<String, String> map = new HashMap(); if(i > 5){ if(i<7){ map.put("id","333"); map.put("标题","mmmm"); }else { map.put("id","333"); map.put("标题","aaaaa"); } }else if (i >3){ map.put("id","222"); map.put("标题","哈哈哈哈"); }else if (i>1 && i<3){ map.put("id","222"); map.put("标题","hhhhhhhh"); }else { map.put("id","222"); map.put("标题","bbbb"); } map.put("描述","sssssss"); map.put("负责人","vvvvv"); map.put("开始时间","2017-02-27 11:20:26"); list.add(map); } Map<String, List<Map<String, String>>> map = new HashMap(); map.put("测试合并数据", list); System.out.println(createExcel(title, map, new int[]{0,1,2})); } }

 

思路:

后来就通过建一个单元格对象,属性有单元格内容,与单元格同一列的上一行内容,起始行,起始列,就这四个属性

 

生成的文件效果(两种情况的):

 

 
 
posted @ 2017-11-30 18:16  Mr_伍先生  阅读(4744)  评论(0编辑  收藏  举报