Grouping a column with dates and times (by date, month, and year) in an Excel Pivot Table using Java Apache POI

Question

I'm currently working on a Java application using Java Apache POI to create a Pivot table. I have a column that contain datetime values like this "6/21/2019 15:17:17":

+---------------------------------+
| Date                | Values    |
+---------------------|-----------+
| 6/21/2019 15:17:17  | 4.31      |
| 6/21/2019 15:17:42  | 3.00      |
| 6/21/2019 15:17:42  | 1.45      |
| 6/21/2019 16:51:28  | 3.00      |
| 6/24/2019 20:08:33  | 3.00      |
| 6/24/2019 20:08:33  | 4.31      |
| 6/24/2019 20:08:33  | 10.15     |
| 6/25/2019 17:57:05  | 21.55     |
| 6/25/2019 18:12:35  | 4.35      |
+---------------------------------+

and I need to group this column in the Pivot table into days (months, years) like that using Java Apache POI

+---------------------------------+
| Result         | Sum of Values  |
+----------------|----------------+
| 21-Jun         | 11.46          |
| 24-Jun         | 17.46          |
| 25-Jun         | 25.9           |
+----------------|----------------+
| Grand Total    | 55.12          |
+---------------------------------+

Any ideas on how to do this?

PS a similar question already exists, but remained unanswered

Answer 1

This is not as simple as you might think. The pivot table creation of apache poi is very rudimentary until now. This is fine as long only default pivot tables are needed. Excel will calculate all necessary while opening the file. But if the need is customizing the pivot table, then more efford is needed. And apache poi does not provide classes and methods for this directly. Using the underlaying ooxml-schemas classes is necessary.

In case of your request, we need set a proper pivot cache since the field group settings are set there. Apache poi only sets a rudimentary pivot cache in hope that Excel will correct this while opening the file.

But to set aa proper pivot cache we need at first determine unique data in data column which ist used as the row label. That is because only unique data will be stored in pivot cache.

If done that we can build the pivot cache and then set field group settings there.

Complete example:

import java.io.FileOutputStream;

import org.apache.poi.ss.*;
import org.apache.poi.ss.usermodel.*;
import org.apache.poi.ss.util.*;
import org.apache.poi.xssf.usermodel.*;

import java.util.TimeZone;
import java.util.GregorianCalendar;

class CreatePivotTableDateValue {

 public static void main(String[] args) throws Exception {

  try (Workbook workbook = new XSSFWorkbook();
       FileOutputStream fileout = new FileOutputStream("Excel.xlsx") ) {

   DataFormat format = workbook.createDataFormat();
   CellStyle dateStyle = workbook.createCellStyle();
   dateStyle.setDataFormat(format.getFormat("M\\/d\\/yyyy hh:mm:ss"));

   Sheet sheet = workbook.createSheet();

   String[] headers = new String[]{"Date", "Value"};
   Row row = sheet.createRow(0);
   Cell cell;
   for (int c = 0; c < headers.length; c++) {
    cell = row.createCell(c); cell.setCellValue(headers[c]);
   }

   Object[][] data = new Object[][]{
    new Object[]{new GregorianCalendar(2019, 5, 21, 15, 17, 17), 4.31},
    new Object[]{new GregorianCalendar(2019, 5, 21, 15, 17, 42), 3.00},
    new Object[]{new GregorianCalendar(2019, 5, 21, 15, 17, 42), 1.45},
    new Object[]{new GregorianCalendar(2019, 5, 21, 16, 51, 28 ), 3.00},
    new Object[]{new GregorianCalendar(2019, 5, 24, 20, 8, 33), 3.00},
    new Object[]{new GregorianCalendar(2019, 5, 24, 20, 8, 33), 4.31},
    new Object[]{new GregorianCalendar(2019, 5, 24, 20, 8, 33), 10.15},
    new Object[]{new GregorianCalendar(2019, 5, 25, 17, 57, 05), 21.55},
    new Object[]{new GregorianCalendar(2019, 5, 25, 18, 12, 35), 4.35 }
   };

   for (int r = 0; r < data.length; r++) {
    row = sheet.createRow(r+1);
    Object[] rowData = data[r];
    for (int c = 0; c < rowData.length; c++) {
     cell = row.createCell(c);
     if (rowData[c] instanceof GregorianCalendar) {
      cell.setCellValue((GregorianCalendar)rowData[c]);
      cell.setCellStyle(dateStyle);
     } else if (rowData[c] instanceof Double) {
      cell.setCellValue((Double)rowData[c]);
     }
    }
   }

   sheet.setColumnWidth(0, 19 * 256);

   // create default pivot table
   AreaReference areaReference = new AreaReference("A1:B10",  SpreadsheetVersion.EXCEL2007);
   XSSFPivotTable pivotTable = ((XSSFSheet)sheet).createPivotTable(areaReference, new CellReference("D4"));
   pivotTable.addRowLabel(0);
   pivotTable.addColumnLabel(DataConsolidateFunction.SUM, 1);

   // here ends direct apache poi support

   // customize pivot table
   // determine unique data in data column 0, the row label
   java.util.TreeSet<GregorianCalendar> uniqueItems = new java.util.TreeSet<GregorianCalendar>();
   for (int r = 0; r < data.length; r++) {
    GregorianCalendar calendar = (GregorianCalendar)data[r][0];
    calendar.setTimeZone(TimeZone.getTimeZone("GMT"));
    uniqueItems.add(calendar);
   }

   // we need min date and max date for grouping
   GregorianCalendar minDate = uniqueItems.first();
   GregorianCalendar maxDate = uniqueItems.last();

   // set proper pivot cache
   pivotTable.getPivotCacheDefinition().getCTPivotCacheDefinition().getCacheFields().getCacheFieldArray(0)
    .getSharedItems().setCount(uniqueItems.size());
   pivotTable.getPivotCacheDefinition().getCTPivotCacheDefinition().getCacheFields().getCacheFieldArray(0)
    .getSharedItems().setMinDate(minDate);
   pivotTable.getPivotCacheDefinition().getCTPivotCacheDefinition().getCacheFields().getCacheFieldArray(0)
    .getSharedItems().setMaxDate(maxDate);
   pivotTable.getPivotCacheDefinition().getCTPivotCacheDefinition().getCacheFields().getCacheFieldArray(0)
    .getSharedItems().setContainsDate(true);
   pivotTable.getPivotCacheDefinition().getCTPivotCacheDefinition().getCacheFields().getCacheFieldArray(0)
    .getSharedItems().setContainsString(false);
   pivotTable.getPivotCacheDefinition().getCTPivotCacheDefinition().getCacheFields().getCacheFieldArray(0)
    .getSharedItems().setContainsNonDate(false);
   pivotTable.getPivotCacheDefinition().getCTPivotCacheDefinition().getCacheFields().getCacheFieldArray(0)
    .getSharedItems().setContainsSemiMixedTypes(false);
   for (GregorianCalendar item : uniqueItems) {
    pivotTable.getPivotCacheDefinition().getCTPivotCacheDefinition().getCacheFields().getCacheFieldArray(0)
     .getSharedItems().addNewD().setV(item);
   }

   // set field group settings
   pivotTable.getPivotCacheDefinition().getCTPivotCacheDefinition().getCacheFields().getCacheFieldArray(0)
    .addNewFieldGroup().setBase(0);
   pivotTable.getPivotCacheDefinition().getCTPivotCacheDefinition().getCacheFields().getCacheFieldArray(0)
    .getFieldGroup().addNewRangePr().setGroupBy(org.openxmlformats.schemas.spreadsheetml.x2006.main.STGroupBy.DAYS);
   pivotTable.getPivotCacheDefinition().getCTPivotCacheDefinition().getCacheFields().getCacheFieldArray(0)
    .getFieldGroup().getRangePr().setStartDate(minDate);
   pivotTable.getPivotCacheDefinition().getCTPivotCacheDefinition().getCacheFields().getCacheFieldArray(0)
    .getFieldGroup().getRangePr().setEndDate(maxDate);

   // at least one grup item in group settings
   pivotTable.getPivotCacheDefinition().getCTPivotCacheDefinition().getCacheFields().getCacheFieldArray(0)
    .getFieldGroup().addNewGroupItems().setCount(1);
   pivotTable.getPivotCacheDefinition().getCTPivotCacheDefinition().getCacheFields().getCacheFieldArray(0)
    .getFieldGroup().getGroupItems().addNewS().setV("0");

   workbook.write(fileout);

  }

 }
}