jamesT jamesT - 5 months ago 29
Java Question

Error While Reading Large Excel Files (xlsx) Via Apache POI

I am trying to read large excel files xlsx via Apache POI, say 40-50 MB. I am getting out of memory exception. The current heap memory is 3GB.

I can read smaller excel files without any issues. I need a way to read large excel files and then them back as response via Spring excel view.

public class FetchExcel extends AbstractView {

protected void renderMergedOutputModel(
Map model, HttpServletRequest request, HttpServletResponse response)
throws Exception {

String fileName = "SomeExcel.xlsx";


OPCPackage pkg = OPCPackage.open("/someDir/SomeExcel.xlsx");

XSSFWorkbook workbook = new XSSFWorkbook(pkg);

ServletOutputStream respOut = response.getOutputStream();


workbook = null;

response.setHeader("Content-disposition", "attachment;filename=\"" +fileName+ "\"");



I first started off using
XSSFWorkbook workbook = new XSSFWorkbook(FileInputStream in);

but that was costly per Apache POI API, so I switched to OPC package way but still the same effect. I don't need to parse or process the file, just read it and return it.


You don't mention whether you need to modify the spreadsheet or not.

This may be obvious, but if you don't need to modify the spreadsheet, then you don't need to parse it and write it back out, you can simply read bytes from the file, and write out bytes, as you would with, say an image, or any other binary format.

If you do need to modify the spreadsheet before sending it to the user, then to my knowledge, you may have to take a different approach.

Every library that I'm aware of for reading Excel files in Java reads the whole spreadsheet into memory, so you'd have to have 50MB of memory available for every spreadsheet that could possibly be concurrently processed. This involves, as others have pointed out, adjusting the heap available to the VM.

If you need to process a large number of spreadsheets concurrently, and can't allocate enough memory, consider using a format that can be streamed, instead of read all at once into memory. CSV format can be opened by Excel, and I've had good results in the past by setting the content-type to application/vnd.ms-excel, setting the attachment filename to something ending in ".xls", but actually returning CSV content. I haven't tried this in a couple of years, so YMMV.