使用Apache POI從PowerPoint幻燈片中刪除圖表

Question

我們正在嘗試使用Apache POI 3.16從PowerPoint幻燈片中刪除圖表，但遇到了困難。

我們的代碼執行以下步驟：

1. 打開現有的PowerPoint文檔（模板文檔）
2. 添加和刪​​除幻燈片
3. 更新現有幻燈片中的圖表

這很好。

在某些時候，我們需要從給定的幻燈片中刪除圖表。 這是我們的嘗試：

OPCPackage pkg = ppt.getPackage();

String chartRelationId = slide.getRelationId(chart);
pkg.removeRelationship(chartRelationId);

pkg.removePart(chart.getPackagePart());

pkg.removePart()調用似乎可以正常工作，但是將最終的PowerPoint文檔寫入磁盤失敗，但有一個例外，說明無法刪除零件文件（大概是因為我們已經刪除了零件文件）。

在將文檔寫入磁盤時， pkg.removeRelationship()調用還會觸發一個異常，該異常表明core.xml已經存在。

可以使用Apache POI從PowerPoint幻燈片中刪除圖表嗎？ 如果是這樣，怎么辦？

Answer 1

由於XSLFChart處於@Beta狀態，因此到目前為止，圖表沒有明確的Shape 。 因此，使用apache poi我們只能獲取包含圖表的XSLFGraphicFrame 。 但是從幻燈片中刪除XSLFGraphicFrame不會同時刪除所有相關的圖表部分。 因此，自上而下刪除相關圖表零件意味着從POIXMLDocumentPart級別下降到PackagePart級別，直到現在才實現。 而且，由於POIXMLDocumentPart中的所有相關方法都受到保護並且XSLFChart本身是最終的，因此實際上沒有解決的可能性。

以下代碼顯示了該問題。 如此評論。

該代碼將刪除第一張幻燈片的所有圖表，並刪除所有的關系和相關零部件，這將是： /ppt/embeddings/Microsoft_Excel_WorksheetN.xlsx ， /ppt/charts/colorsN.xml和/ppt/charts/styleN.xml 。 正如注釋所示，僅/ppt/charts/chartN.xml無法刪除。

import java.io.FileInputStream;
import java.io.FileOutputStream;

import org.apache.poi.xslf.usermodel.*;
import org.apache.poi.sl.usermodel.*;

import org.apache.poi.POIXMLDocumentPart;

import org.apache.poi.openxml4j.opc.OPCPackage;
import org.apache.poi.openxml4j.opc.PackagePart;
import org.apache.poi.openxml4j.opc.PackageRelationshipCollection;
import org.apache.poi.openxml4j.opc.PackageRelationship;

import org.apache.xmlbeans.XmlObject;

import java.util.Map;
import java.util.HashMap;

import java.util.regex.Pattern;

public class ReadPPTRemoveChart {

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

  XMLSlideShow slideShow = new XMLSlideShow(new FileInputStream("PPTWithCharts.pptx"));

  XSLFSlide slide = slideShow.getSlides().get(0);

  Map<String, XSLFGraphicFrame> chartFramesToRemove = new HashMap<>();

  for (XSLFShape shape : slide.getShapes()) {
   if (shape instanceof XSLFGraphicFrame) {
    XSLFGraphicFrame graphicframe = (XSLFGraphicFrame)shape;
    XmlObject xmlobject = graphicframe.getXmlObject();
    XmlObject[] graphics = xmlobject.selectPath(
                            "declare namespace a='http://schemas.openxmlformats.org/drawingml/2006/main' " +
                            ".//a:graphic");
    if (graphics.length > 0) { //we have a XSLFGraphicFrame containing a:graphic
     XmlObject graphic = graphics[0];
     XmlObject[] charts = graphic.selectPath(
                           "declare namespace c='http://schemas.openxmlformats.org/drawingml/2006/chart' " +
                           ".//c:chart");
     if (charts.length > 0) { //we have a XSLFGraphicFrame containing c:chart
      XmlObject chart = charts[0];
      String rid = chart.selectAttribute(
                          "http://schemas.openxmlformats.org/officeDocument/2006/relationships", "id")
                          .newCursor().getTextValue();
      chartFramesToRemove.put(rid, graphicframe);
     }
    }
   }
  }

  PackagePart slidepart = slide.getPackagePart();
  OPCPackage opcpackage = slideShow.getPackage();

  for (String rid : chartFramesToRemove.keySet()) {
   //at frist remove the XSLFGraphicFrame
   XSLFGraphicFrame chartFrame = chartFramesToRemove.get(rid);
   slide.removeShape(chartFrame);
   //Here is the problem in my opinion. This **should** remove all related parts too.
   //But since XSLFChart is @Beta, it does not.

   //So we try doing removing the related parts manually.
   //we get the PackagePart of the chart
   PackageRelationship relship = slidepart.getRelationships().getRelationshipByID(rid);
   PackagePart chartpart = slidepart.getRelatedPart(relship);

   //now we get and remove all the relations and related PackageParts from this chartpart
   //this are /ppt/embeddings/Microsoft_Excel_WorksheetN.xlsx, /ppt/charts/colorsN.xml 
   //and /ppt/charts/styleN.xml
   for (PackageRelationship chartrelship : chartpart.getRelationships()) {
    String partname = chartrelship.getTargetURI().toString();
    PackagePart part = opcpackage.getPartsByName(Pattern.compile(partname)).get(0);
    opcpackage.removePart(part);
    chartpart.removeRelationship(chartrelship.getId());
   }
   //this works

   //now we **should** be able removing the relationship to the chartpart from the slide too
   //but this seems not to be possible
   //doing this on PackagePart level works:
   slidepart.removeRelationship(rid);
   for (PackageRelationship sliderelship : slidepart.getRelationships()) {
    System.out.println("rel PP level: " + sliderelship.getTargetURI().toString());
   }
   //all relationships to /ppt/charts/chartN.xml are removed

   //but on POIXMLDocumentPart level this has no effect
   for (POIXMLDocumentPart sliderelpart : slide.getRelations()) {
    System.out.println("rel POIXML level: " + sliderelpart.getPackagePart().getPartName());
   }
   //relationships to /ppt/charts/chartN.xml are **not** removed

   //So we cannot remove the chartpart.
   //If we would do this, then while slideShow.write the 
   //org.apache.poi.xslf.usermodel.XSLFChart.commit in XSLFChart.java fails 
   //because after removing the PackagePart is absent but the relation is still there.
   //opcpackage.removePart(chartpart);

  }


  slideShow.write(new FileOutputStream("PPTWithChartsNew.pptx"));
  slideShow.close();
 }
}

使用PowerPoint打開PPTWithChartsNew.pptx並將其保存后，由於不需要更多的/ppt/charts/styleN.xml部分，因此也將它們刪除。

---

編輯2017年9月24日：

找到了使用反射的解決方案。 如前所述，刪除相關的圖表部分需要自上而下，即從POIXMLDocumentPart級別降至PackagePart級別。 並且由於POIXMLDocumentPart.removeRelation受保護，因此我們需要使用反射進行此操作。

import java.io.FileInputStream;
import java.io.FileOutputStream;

import org.apache.poi.xslf.usermodel.*;
import org.apache.poi.sl.usermodel.*;

import org.apache.poi.POIXMLDocumentPart;

import org.apache.poi.openxml4j.opc.OPCPackage;
import org.apache.poi.openxml4j.opc.PackagePart;
import org.apache.poi.openxml4j.opc.PackageRelationshipCollection;
import org.apache.poi.openxml4j.opc.PackageRelationship;

import org.apache.xmlbeans.XmlObject;

import java.util.Map;
import java.util.HashMap;

import java.util.regex.Pattern;

import java.lang.reflect.Method;

public class ReadPPTRemoveChart {

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

  XMLSlideShow slideShow = new XMLSlideShow(new FileInputStream("PPTWithCharts.pptx"));

  XSLFSlide slide = slideShow.getSlides().get(0);

  Map<String, XSLFGraphicFrame> chartFramesToRemove = new HashMap<>();

  for (XSLFShape shape : slide.getShapes()) {
   if (shape instanceof XSLFGraphicFrame) {
    XSLFGraphicFrame graphicframe = (XSLFGraphicFrame)shape;
    XmlObject xmlobject = graphicframe.getXmlObject();
    XmlObject[] graphics = xmlobject.selectPath(
                            "declare namespace a='http://schemas.openxmlformats.org/drawingml/2006/main' " +
                            ".//a:graphic");
    if (graphics.length > 0) { //we have a XSLFGraphicFrame containing a:graphic
     XmlObject graphic = graphics[0];
     XmlObject[] charts = graphic.selectPath(
                           "declare namespace c='http://schemas.openxmlformats.org/drawingml/2006/chart' " +
                           ".//c:chart");
     if (charts.length > 0) { //we have a XSLFGraphicFrame containing c:chart
      XmlObject chart = charts[0];
      String rid = chart.selectAttribute(
                          "http://schemas.openxmlformats.org/officeDocument/2006/relationships", "id")
                          .newCursor().getTextValue();
      chartFramesToRemove.put(rid, graphicframe);
     }
    }
   }
  }

  PackagePart slidepart = slide.getPackagePart();
  OPCPackage opcpackage = slideShow.getPackage();

  for (String rid : chartFramesToRemove.keySet()) {
   //at frist remove the XSLFGraphicFrame
   XSLFGraphicFrame chartFrame = chartFramesToRemove.get(rid);
   slide.removeShape(chartFrame);
   //Here is the problem in my opinion. This **should** remove all related parts too.
   //But since XSLFChart is @Beta, it does not.

   //So we try doing removing the related parts manually.

   //we get the PackagePart of the chart
   PackageRelationship relship = slidepart.getRelationships().getRelationshipByID(rid);
   PackagePart chartpart = slidepart.getRelatedPart(relship);

   //now we get and remove all the relations and related PackageParts from this chartpart
   //this are /ppt/embeddings/Microsoft_Excel_WorksheetN.xlsx, /ppt/charts/colorsN.xml 
   //and /ppt/charts/styleN.xml
   for (PackageRelationship chartrelship : chartpart.getRelationships()) {
    String partname = chartrelship.getTargetURI().toString();
    PackagePart part = opcpackage.getPartsByName(Pattern.compile(partname)).get(0);
    opcpackage.removePart(part);
    chartpart.removeRelationship(chartrelship.getId());
   }

   //now we remove the chart part from the slide part
   //We need doing this on POIXMLDocumentPart level. 
   //Since POIXMLDocumentPart.removeRelation is protected, we need doing this using reflection
   XSLFChart chart = (XSLFChart)slide.getRelationById(rid);
   Method removeRelation = POIXMLDocumentPart.class.getDeclaredMethod("removeRelation", POIXMLDocumentPart.class); 
   removeRelation.setAccessible(true); 
   removeRelation.invoke(slide, chart);

  }

  slideShow.write(new FileOutputStream("PPTWithChartsNew.pptx"));
  slideShow.close();
 }
}