我处理的平均XLSX表格约为18-22个含有13-20列的750000行。它与其他许多功能一起在Spring Web应用程序中运行。我没有给整个应用程序太多的内存:
-Xms1024m -Xmx4096m
,效果很好!
首先,解决代码问题:将每一行数据全部加载到内存中再开始转储是错误的。对于我的情况(从PostgreSQL数据库报告),我重新设计了数据转储过程,使用
RowCallbackHandler
写入我的XLSX。当达到“限制”的750000行时,我会创建新的工作表。并且工作簿的可见窗口为50行。通过这种方式,我能够转储大量数据:XLSX文件的大小约为1230Mb。
以下是编写工作表的一些代码:
jdbcTemplate.query(
new PreparedStatementCreator() {
@Override
public PreparedStatement createPreparedStatement(Connection connection) throws SQLException {
PreparedStatement statement = connection.prepareStatement(finalQuery, ResultSet.TYPE_FORWARD_ONLY, ResultSet.CONCUR_READ_ONLY);
statement.setFetchSize(100);
statement.setFetchDirection(ResultSet.FETCH_FORWARD);
return statement;
}
}, new RowCallbackHandler() {
Sheet sheet = null;
int i = 750000;
int tableId = 0;
@Override
public void processRow(ResultSet resultSet) throws SQLException {
if (i == 750000) {
tableId++;
i = 0;
sheet = wb.createSheet(sheetName.concat(String.format("%02d%n", tableId)));
Row r = sheet.createRow(0);
Cell c = r.createCell(0);
c.setCellValue("id");
c = r.createCell(1);
c.setCellValue("Дата");
c = r.createCell(2);
c.setCellValue("Комментарий");
c = r.createCell(3);
c.setCellValue("Сумма операции");
c = r.createCell(4);
c.setCellValue("Дебет");
c = r.createCell(5);
c.setCellValue("Страхователь");
c = r.createCell(6);
c.setCellValue("Серия договора");
c = r.createCell(7);
c.setCellValue("Номер договора");
c = r.createCell(8);
c.setCellValue("Основной агент");
c = r.createCell(9);
c.setCellValue("Кредит");
c = r.createCell(10);
c.setCellValue("Программа");
c = r.createCell(11);
c.setCellValue("Дата начала покрытия");
c = r.createCell(12);
c.setCellValue("Дата планового окончания покрытия");
c = r.createCell(13);
c.setCellValue("Периодичность уплаты взносов");
}
i++;
PremiumEntity e = PremiumEntity.builder()
.Id(resultSet.getString("id"))
.OperationDate(resultSet.getDate("operation_date"))
.Comments(resultSet.getString("comments"))
.SumOperation(resultSet.getBigDecimal("sum_operation").doubleValue())
.DebetAccount(resultSet.getString("debet_account"))
.Strahovatelname(resultSet.getString("strahovatelname"))
.Seria(resultSet.getString("seria"))
.NomPolica(resultSet.getLong("nom_polica"))
.Agentname(resultSet.getString("agentname"))
.CreditAccount(resultSet.getString("credit_account"))
.Program(resultSet.getString("program"))
.PoliciStartDate(resultSet.getDate("polici_start_date"))
.PoliciPlanEndDate(resultSet.getDate("polici_plan_end_date"))
.Periodichn(resultSet.getString("id_periodichn"))
.build();
Row r = sheet.createRow(i);
Cell c = r.createCell(0);
c.setCellValue(e.getId());
if (e.getOperationDate() != null) {
c = r.createCell(1);
c.setCellStyle(dateStyle);
c.setCellValue(e.getOperationDate());
}
c = r.createCell(2);
c.setCellValue(e.getComments());
c = r.createCell(3);
c.setCellValue(e.getSumOperation());
c = r.createCell(4);
c.setCellValue(e.getDebetAccount());
c = r.createCell(5);
c.setCellValue(e.getStrahovatelname());
c = r.createCell(6);
c.setCellValue(e.getSeria());
c = r.createCell(7);
c.setCellValue(e.getNomPolica());
c = r.createCell(8);
c.setCellValue(e.getAgentname());
c = r.createCell(9);
c.setCellValue(e.getCreditAccount());
c = r.createCell(10);
c.setCellValue(e.getProgram());
if (e.getPoliciStartDate() != null) {
c = r.createCell(11);
c.setCellStyle(dateStyle);
c.setCellValue(e.getPoliciStartDate());
}
;
if (e.getPoliciPlanEndDate() != null) {
c = r.createCell(12);
c.setCellStyle(dateStyle);
c.setCellValue(e.getPoliciPlanEndDate());
}
c = r.createCell(13);
c.setCellValue(e.getPeriodichn());
}
});
在重新编写将数据转储为XLSX的代码后,我遇到了一个问题,它需要64位Office才能打开它们。因此,我需要将具有许多工作表的工作簿拆分为单个工作表的单独XLSX文件,以使它们可在普通计算机上阅读。再次使用小的可见窗口和流处理,并保持整个应用程序正常工作,没有任何OutOfMemory的迹象。
一些读取和拆分工作表的代码:
OPCPackage opcPackage = OPCPackage.open(originalFile, PackageAccess.READ);
ReadOnlySharedStringsTable strings = new ReadOnlySharedStringsTable(opcPackage);
XSSFReader xssfReader = new XSSFReader(opcPackage);
StylesTable styles = xssfReader.getStylesTable();
XSSFReader.SheetIterator iter = (XSSFReader.SheetIterator) xssfReader.getSheetsData();
int index = 0;
while (iter.hasNext()) {
InputStream stream = iter.next();
String sheetName = iter.getSheetName();
DataFormatter formatter = new DataFormatter();
InputSource sheetSource = new InputSource(stream);
SheetToWorkbookSaver saver = new SheetToWorkbookSaver(sheetName);
try {
XMLReader sheetParser = SAXHelper.newXMLReader();
ContentHandler handler = new XSSFSheetXMLHandler(
styles, null, strings, saver, formatter, false);
sheetParser.setContentHandler(handler);
sheetParser.parse(sheetSource);
} catch(ParserConfigurationException e) {
throw new RuntimeException("SAX parser appears to be broken - " + e.getMessage());
}
stream.close();
FileDto partFile = new FileDto("report_".concat(StringUtils.trimToEmpty(sheetName)).concat(".xlsx"));
File cloneFile = fileStorage.read(partFile);
FileOutputStream cloneFos = new FileOutputStream(cloneFile);
saver.getWb().write(cloneFos);
cloneFos.close();
}
并且。
public class SheetToWorkbookSaver implements XSSFSheetXMLHandler.SheetContentsHandler {
private SXSSFWorkbook wb;
private Sheet sheet;
private CellStyle dateStyle ;
private Row currentRow;
public SheetToWorkbookSaver(String workbookName) {
this.wb = new SXSSFWorkbook(50);
this.dateStyle = this.wb.createCellStyle();
this.dateStyle.setDataFormat(this.wb.getCreationHelper().createDataFormat().getFormat("dd.mm.yyyy"));
this.sheet = this.wb.createSheet(workbookName);
}
@Override
public void startRow(int rowNum) {
this.currentRow = this.sheet.createRow(rowNum);
}
@Override
public void endRow(int rowNum) {
}
@Override
public void cell(String cellReference, String formattedValue, XSSFComment comment) {
int thisCol = (new CellReference(cellReference)).getCol();
Cell c = this.currentRow.createCell(thisCol);
c.setCellValue(formattedValue);
c.setCellComment(comment);
}
@Override
public void headerFooter(String text, boolean isHeader, String tagName) {
}
public SXSSFWorkbook getWb() {
return wb;
}
}
因此它可以读写数据。我猜在你的情况下,你应该重新设计你的代码,使其保持小内存占用的模式。所以我建议为读取创建自定义SheetContentsReader
,它将把数据推送到某个数据库,在那里可以轻松地处理、聚合等。