Apache POI：从Word模板文件中完全删除图表

Question

PackagePart packagePartChart = xWPFChart.getPackagePart();

PackagePart packagePartWordDoc = xWPFDocument.getPackagePart();
OPCPackage packageWordDoc = packagePartWordDoc.getPackage();

// iterate over all relations the chart has and remove them
for (PackageRelationship chartrelship : packagePartChart.getRelationships()) {
    String partname = chartrelship.getTargetURI().toString();
    PackagePart part = packageWordDoc.getPartsByName(Pattern.compile(partname)).get(0);
    packageWordDoc.removePart(part);
    packagePartChart.removeRelationship(chartrelship.getId());
}

// now remove the chart itself from the word doc                           
Method removeRelation = POIXMLDocumentPart.class.getDeclaredMethod("removeRelation", POIXMLDocumentPart.class); 
removeRelation.setAccessible(true); 
removeRelation.invoke(xWPFDocument, xWPFChart);

null

有人知道这里会出什么问题吗？

Answer 1

事实上，找到正确的段落保存图表是一项挑战。最后，为了简单起见，我添加了一个一行/一列占位符表，其中有一个带有文本的单元格，比方说“目标字串”，就在图表的前面。通过下面的函数，我定位这个表的BodyElementID：

private Integer iBodyElementIterator (XWPFDocument wordDoc,String targetWordString) {
    Iterator<IBodyElement> iter = wordDoc.getBodyElementsIterator();
    Integer bodyElementID = null;
    while (iter.hasNext()) {
       IBodyElement elem = iter.next();
       bodyElementID = wordDoc.getBodyElements().indexOf(elem);
       if (elem instanceof XWPFParagraph) {

           XWPFParagraph paragraph = (XWPFParagraph) elem;
           for (XWPFRun runText : paragraph.getRuns()) {
                String text = runText.getText(0);                               
               Core.getLogger("WordExporter").trace("Body Element ID:  " + bodyElementID + " Text: " + text);
                if (text != null && text.equals(targetWordString)) {                        
                    break;                          
                }
            }

       } else if (elem instanceof XWPFTable) {
           if (((XWPFTable) elem).getRow(0) != null && ((XWPFTable) elem).getRow(0).getCell(0) != null) {
                // the first cell holds the name via the template
                String tableTitle = ((XWPFTable) elem).getRow(0).getCell(0).getText();
                if (tableTitle.equals(targetWordString)) {
                    break;
                }
                Core.getLogger("WordExporter").trace("Body Element ID:  " + bodyElementID + " Text: " + tableTitle);
           } else {
               Core.getLogger("WordExporter").trace("Body Element ID:  " + bodyElementID + " Table removed!");
           }

       }
       else {
           Core.getLogger("WordExporter").trace("Body Element ID:  " + bodyElementID + " Text: ?");
       }
    }
    return bodyElementID;
}

在代码的主要部分，我调用这个函数来定位表，然后首先删除图表(ID+1)，然后删除表(ID)

int elementIDToBeRemoved = iBodyElementIterator(xWPFWordDoc,targetWordString);

xWPFWordDoc.removeBodyElement(elementIDToBeRemoved + 1);
xWPFWordDoc.removeBodyElement(elementIDToBeRemoved);

它需要按这个顺序排列，因为一旦删除中间的一个数字，ID就会重新排序，因此首先删除表意味着图表原则上会得到该ID。