Question

I am using the apache poi to handle the .xlsx file.

I have two .xlsx files,
part.xlsx,and full.xlsx,

they own the same structure.

Each record(The Row object in the poi) have three colmn:name,age,location.

There are almost 5000 rows in the part.xlsx,and 40000 rows in the full.xlsx.

Now I want to extra the rows from the full.xlsx who have the same value with that in the part.xlsx.

for example:

The part.xlsx:

Name age location
kk   23  USA
bb   24  England
......

The full.xlsx

Name age location
kk   23  USA
bb   24  England
xx   25  USA
......

Now I want to extra the ‘kk’ and ‘bb’ rows and save them to a new file.

This is the code code:

List<User> usersInpart=new ArrayList<User>();
List<Row> rows_to_be_saved=new ArrayList<Row>();

//read the part.xlsx and save them.
FileInputStream fis_part=new FileInputStream("part.xlsx");
WorkBook wb_part=WorkbookFactory.create(fis_part);
Sheet st_part=wb_part.getSheetAt(0);
for(Row row : st_part){
    if(row.getRowNum()==0) continue; //skip the first row(the title)
    User u=new User();
    u.setName(row.getCell(0).getRichStringValue().getString().trim());
    u.setAge(row.getCell(1).getNumericCellValue());
    u.setLocation(row.getCell(2).getRichStringValue().getString().trim());
    usersInpart.add(u);
}
fis_part.close();


//read the full.xlsx

FileInputStream fis_full=new FileInputStream("full.xlsx");
WorkBook wb_full=WorkbookFactory.create(fis_full);
Sheet st_full=wb_full.getSheetAt(0);
for(Row row : st_full){
    if(row.getRowNum()==0) continue; //skip the first row(the title)

    String name=row.getCell(0).getRichStringValue().getString().trim();
    double age=row.getCell(1).getNumericCellValue();
    String location=row.getCell(2).getRichStringValue().getString().trim();

    for(User u : usersInpart){
        if(u.getName.equals(name) && u.getAge==age && u.getLocation().equals(location))
            rows_to_be_saved.add(row);
    }
}
fis_full.close();

//write the selected rows to file

WorkBook wb_res=WorkbookFactory.create(fis_full);
Sheet st_res=wb_res.createSheet(0);

    int i=0;
    for (Row row : rows_to_be_saved) {
        Row rw=st_res.createRow(i);

        int k=0;
        for (Cell cell : row) {
            switch (cell.getCellType()) {
                case Cell.CELL_TYPE_STRING:
                    rw.createCell(k).setCellValue(cell.getRichStringCellValue().getString());
                    break;
                case Cell.CELL_TYPE_NUMERIC:
                    if (DateUtil.isCellDateFormatted(cell)) {
                        rw.createCell(k).setCellValue(cell.getDateCellValue());
                    } else {
                        rw.createCell(k).setCellValue(cell.getNumericCellValue());
                    }
                    break;
                case Cell.CELL_TYPE_BOOLEAN:
                    rw.createCell(k).setCellValue(cell.getBooleanCellValue());
                    break;
                case Cell.CELL_TYPE_FORMULA:
                    rw.createCell(k).setCellValue(cell.getCellFormula());
                    break;
                default:
            }
            k++;
        }
        i++;
    }
//save the wb_res
wb_res.write(new FileOutputStrem("xx.xlsx"));

now I wonder any good ideas to save the file?

Since I have saved the selected rows in the “rows_to_be_saved”.

And I create the new sheet “st_res”,any way I can save these rows to the “st_res” directly? Since now,I have create each row according the rows in the “rows_to_be_saved”.

So there will be two List of rows. I think this is a waste of memory.

Any suggestion?

Answer 1

If memory usage is a concern, you could save much more by using the XSSF Event Model to read the full.xlsx file. You’re presently loading a 40,000-row file into memory, whereas the event model would only keep a single row in memory at a time.