Java Code Examples for org.docx4j.XmlUtils

private SectPr createSectPr() throws Exception {
		
		String openXML = "<w:sectPr xmlns:w=\"http://schemas.openxmlformats.org/wordprocessingml/2006/main\" xmlns:r=\"http://schemas.openxmlformats.org/officeDocument/2006/relationships\">"
				
				// Word adds the background image in each of 3 header parts
	            + "<w:headerReference r:id=\"" + createHeaderPart("even").getId() + "\" w:type=\"even\"/>"
	            + "<w:headerReference r:id=\"" + createHeaderPart("default").getId() + "\" w:type=\"default\"/>"
	            + "<w:headerReference r:id=\"" + createHeaderPart("first").getId() + "\" w:type=\"first\"/>"
	            
	            // Word adds empty footer parts when you create a watermark, but its not necessary
	            
//	            + "<w:footerReference r:id=\"rId9\" w:type=\"even\"/>"
//	            + "<w:footerReference r:id=\"rId10\" w:type=\"default\"/>"
//	            + "<w:footerReference r:id=\"rId12\" w:type=\"first\"/>"
	            
	            + "<w:pgSz w:h=\"15840\" w:w=\"12240\"/>"
	            + "<w:pgMar w:bottom=\"1440\" w:footer=\"708\" w:gutter=\"0\" w:header=\"708\" w:left=\"1440\" w:right=\"1440\" w:top=\"1440\"/>"
	            + "<w:cols w:space=\"708\"/>"
	            + "<w:docGrid w:linePitch=\"360\"/>"
	        +"</w:sectPr>";
	 
		return (SectPr)XmlUtils.unmarshalString(openXML);
	
	}
 

public static <T> List<T> getChildrenElements(Object source, Class<T> targetClass) {  
    List<T> result = new ArrayList<T>();  
    //获取真实的对象
	Object target = XmlUtils.unwrap(source);
	//if (target.getClass().equals(targetClass)) {
    if (targetClass.isAssignableFrom(target.getClass())) { 
        result.add((T)target);
    } else if (target instanceof ContentAccessor) {  
        List<?> children = ((ContentAccessor) target).getContent();  
        //if (children.getClass().equals(targetClass)) {
        if (targetClass.isAssignableFrom(children.getClass())) { 
            result.add((T)children);
        }
    }
    return result;  
}
 

/**
 * @Description:删除指定位置的表格,删除后表格数量减一
 */
public static boolean removeTableByIndex(WordprocessingMLPackage wordMLPackage, int index) throws Exception {
    boolean flag = false;
    if (index < 0) {
        return flag;
    }
    List<Object> objList = wordMLPackage.getMainDocumentPart().getContent();
    if (objList == null) {
        return flag;
    }
    int k = -1;
    for (int i = 0, len = objList.size(); i < len; i++) {
        Object obj = XmlUtils.unwrap(objList.get(i));
        if (obj instanceof Tbl) {
            k++;
            if (k == index) {
                wordMLPackage.getMainDocumentPart().getContent().remove(i);
                flag = true;
                break;
            }
        }
    }
    return flag;
}
 

/**
 * @Description: 删除指定行 删除后行数减一
 */
public static boolean removeTrByIndex(Tbl tbl, int index) {
    boolean flag = false;
    if (index < 0) {
        return flag;
    }
    List<Object> objList = tbl.getContent();
    if (objList == null) {
        return flag;
    }
    int k = -1;
    for (int i = 0, len = objList.size(); i < len; i++) {
        Object obj = XmlUtils.unwrap(objList.get(i));
        if (obj instanceof Tr) {
            k++;
            if (k == index) {
                tbl.getContent().remove(i);
                flag = true;
                break;
            }
        }
    }
    return flag;
}
 

public static void writeToStream(WordprocessingMLPackage wmlPackage,OutputStream output) throws IOException, Docx4JException {
	Assert.notNull(wmlPackage, " wmlPackage is not specified!");
	Assert.notNull(output, " output is not specified!");
	InputStream input = null;
	try {
		//Document对象
		MainDocumentPart document = wmlPackage.getMainDocumentPart();	
		//Document XML
		String documentXML = XmlUtils.marshaltoString(wmlPackage);
		//转成字节输入流
		input = IOUtils.toBufferedInputStream(new ByteArrayInputStream(documentXML.getBytes()));
		//输出模板
		IOUtils.copy(input, output);
	} finally {
		IOUtils.closeQuietly(input);
	}
}
 

public void writeToWriter(WordprocessingMLPackage wmlPackage,Writer output) throws IOException, Docx4JException {
	Assert.notNull(wmlPackage, " wmlPackage is not specified!");
	Assert.notNull(output, " output is not specified!");
	InputStream input = null;
	try {
		//Document对象
		MainDocumentPart document = wmlPackage.getMainDocumentPart();	
		//Document XML
		String documentXML = XmlUtils.marshaltoString(wmlPackage.getPackage());
		//转成字节输入流
		input = IOUtils.toBufferedInputStream(new ByteArrayInputStream(documentXML.getBytes()));
		//获取模板输出编码格式
		String charsetName = Docx4jProperties.getProperty(Docx4jConstants.DOCX4J_CONVERT_OUT_WMLTEMPLATE_CHARSETNAME, Docx4jConstants.DEFAULT_CHARSETNAME );
		//输出模板
		IOUtils.copy(input, output, Charset.forName(charsetName));
	} finally {
		IOUtils.closeQuietly(input);
	}
}
 

/**
 * @Description:删除指定位置的表格,删除后表格数量减一
 */
public boolean removeTableByIndex(WordprocessingMLPackage wordMLPackage,
        int index) throws Exception {
    boolean flag = false;
    if (index < 0) {
        return flag;
    }
    List<Object> objList = wordMLPackage.getMainDocumentPart().getContent();
    if (objList == null) {
        return flag;
    }
    int k = -1;
    for (int i = 0, len = objList.size(); i < len; i++) {
        Object obj = XmlUtils.unwrap(objList.get(i));
        if (obj instanceof Tbl) {
            k++;
            if (k == index) {
                wordMLPackage.getMainDocumentPart().getContent().remove(i);
                flag = true;
                break;
            }
        }
    }
    return flag;
}
 

/**
 * @Description: 删除指定行 删除后行数减一
 */
public boolean removeTrByIndex(Tbl tbl, int index) {
    boolean flag = false;
    if (index < 0) {
        return flag;
    }
    List<Object> objList = tbl.getContent();
    if (objList == null) {
        return flag;
    }
    int k = -1;
    for (int i = 0, len = objList.size(); i < len; i++) {
        Object obj = XmlUtils.unwrap(objList.get(i));
        if (obj instanceof Tr) {
            k++;
            if (k == index) {
                tbl.getContent().remove(i);
                flag = true;
                break;
            }
        }
    }
    return flag;
}
 

private static boolean deleteCommentReference(ContentAccessor parent,
		BigInteger commentId) {
	for (int i = 0; i < parent.getContent().size(); i++) {
		Object contentObject = XmlUtils.unwrap(parent.getContent().get(i));
		if (contentObject instanceof ContentAccessor) {
			if (deleteCommentReference((ContentAccessor) contentObject, commentId)) {
				return true;
			}
		}
		if (contentObject instanceof R) {
			for (Object runContentObject : ((R) contentObject).getContent()) {
				Object unwrapped = XmlUtils.unwrap(runContentObject);
				if (unwrapped instanceof R.CommentReference) {
					BigInteger foundCommentId = ((R.CommentReference) unwrapped)
							.getId();
					if (foundCommentId.equals(commentId)) {
						parent.getContent().remove(i);
						return true;
					}
				}
			}
		}
	}
	return false;
}
 

public void walkJAXBElements(Object parent) {
    List children = getChildren(parent);
    if (children != null) {

        for (Object object : children) {
            object = XmlUtils.unwrap(object);

            if (object instanceof Child && !(parent instanceof SdtBlock)) {
                ((Child) object).setParent(parent);
            }

            this.apply(object);

            if (this.shouldTraverse(object)) {
                walkJAXBElements(object);
            }
        }
    }
}
 

public List<Object> apply(Object o) {  
    if (o instanceof javax.xml.bind.JAXBElement  
            && (((JAXBElement) o).getName().getLocalPart()  
                    .equals("commentReference")  
                    || ((JAXBElement) o).getName().getLocalPart()  
                            .equals("commentRangeStart") || ((JAXBElement) o)  
                    .getName().getLocalPart().equals("commentRangeEnd"))) {  
        System.out.println(((JAXBElement) o).getName().getLocalPart());  
        commentElements.add((Child) XmlUtils.unwrap(o));  
    } else if (o instanceof CommentReference  
            || o instanceof CommentRangeStart  
            || o instanceof CommentRangeEnd) {  
        System.out.println(o.getClass().getName());  
        commentElements.add((Child) o);  
    }  
    return null;  
}
 

public void walkJAXBElements(Object parent) {  
    List children = getChildren(parent);  
    if (children != null) {  
        for (Object o : children) {  
            if (o instanceof javax.xml.bind.JAXBElement  
                    && (((JAXBElement) o).getName().getLocalPart()  
                            .equals("sdt"))) {  
                ((Child) ((JAXBElement) o).getValue()).setParent(XmlUtils  
                        .unwrap(parent));  
            } else {  
                o = XmlUtils.unwrap(o);  
                if (o instanceof Child) {  
                    ((Child) o).setParent(XmlUtils.unwrap(parent));  
                }  
            }  
            this.apply(o);  
            if (this.shouldTraverse(o)) {  
                walkJAXBElements(o);  
            }  
        }  
    }  
}
 

/**
 * Use RunFontSelector to determine the correct font for the list item label.
 * 
 * @param context
 * @param foListItemLabelBody
 * @param pPr
 * @param rPr
 * @param text
 */
protected static void setFont(RunFontSelector runFontSelector, Element foListItemLabelBody, PPr pPr, RPr rPr, String text) {
	
	DocumentFragment result = (DocumentFragment)runFontSelector.fontSelector(pPr, rPr, text);
	log.debug(XmlUtils.w3CDomNodeToString(result));
	// eg <fo:inline xmlns:fo="http://www.w3.org/1999/XSL/Format" font-family="Times New Roman">1)</fo:inline>
	
	// Now get the attribute value
	if (result!=null && result.getFirstChild()!=null) {
		Attr attr = ((Element)result.getFirstChild()).getAttributeNode("font-family");
		if (attr!=null) {
			foListItemLabelBody.setAttribute("font-family", attr.getValue());
		}
	}
			
}
 

public static DocumentFragment getLayoutMasterSetFragment(AbstractWmlConversionContext context) {

		LayoutMasterSet lms = getFoLayoutMasterSet(context);	
		
		// Set suitable extents, for which we need area tree 
		FOSettings foSettings = (FOSettings)context.getConversionSettings();
		if ( !foSettings.lsLayoutMasterSetCalculationInProgress()) // Avoid infinite loop
			// Can't just do it where foSettings.getApacheFopMime() is not MimeConstants.MIME_FOP_AREA_TREE,
			// since TOC functionality uses that.
		{
			fixExtents( lms, context, true);
		}
		
		org.w3c.dom.Document document = XmlUtils.marshaltoW3CDomDocument(lms, Context.getXslFoContext() );
		DocumentFragment docfrag = document.createDocumentFragment();
		docfrag.appendChild(document.getDocumentElement());
		
		
		return docfrag;		
	}
 

/**
    * For XSLFOExporterNonXSLT
    * @since 3.0
    * 
    */	
public static void appendLayoutMasterSetFragment(AbstractWmlConversionContext context, Node foRoot) {

	LayoutMasterSet lms = getFoLayoutMasterSet(context);	
	
	// Set suitable extents, for which we need area tree 
	FOSettings foSettings = (FOSettings)context.getConversionSettings();
	if ( !foSettings.lsLayoutMasterSetCalculationInProgress()) // Avoid infinite loop
		// Can't just do it where foSettings.getApacheFopMime() is not MimeConstants.MIME_FOP_AREA_TREE,
		// since TOC functionality uses that.
	{
		fixExtents( lms, context, false);
	}
	
	org.w3c.dom.Document document = XmlUtils.marshaltoW3CDomDocument(lms, Context.getXslFoContext() );
	XmlUtils.treeCopy(document.getDocumentElement(), foRoot);
}
 

/**
	 * table says 40, Normal says nothing.
	 * don't override table style
	 * @throws Exception
	 */
	@Test 
	public void testTblStyle_AllSilent() throws Exception {
				
		WordprocessingMLPackage wordMLPackage = WordprocessingMLPackage.createPackage();
		wordMLPackage.getMainDocumentPart().setContents(
				(Document)XmlUtils.unmarshalString(mdpXml_tblStyle) );
		wordMLPackage.getMainDocumentPart().getStyleDefinitionsPart().setContents(
				(Styles)XmlUtils.unmarshalString(styles_no_font_sz) );
		
//		// NB PropertyResolver puts 10pt in DocDefaults, if nothing is specified!
		
		setSetting(wordMLPackage, OVERRIDE); // resulting text in table is 40pt

		wordMLPackage.save(new File(System.getProperty("user.dir") + "/OUT_PStyleInTableTest.docx"));
		
		ParagraphStylesInTableFix.process(wordMLPackage);
		
		Style s = getStyle(wordMLPackage, getStyleName());
		this.assertSz(s, 40);
	}
 

@Test 
	@Ignore
	public void testTblStyle_AllSilent() throws Exception {
		
		WordprocessingMLPackage wordMLPackage = WordprocessingMLPackage.createPackage();
		wordMLPackage.getMainDocumentPart().setContents(
				(Document)XmlUtils.unmarshalString(mdpXml_tblStyle) );
		wordMLPackage.getMainDocumentPart().getStyleDefinitionsPart().setContents(
				(Styles)XmlUtils.unmarshalString(styles_no_font_sz) );

		setSetting(wordMLPackage, OVERRIDE); 

		wordMLPackage.save(new File(System.getProperty("user.dir") + "/OUT_PStyleInTableTest.docx"));
		
//		// NB createVirtualStylesForDocDefaults() puts 10pt there, if nothing is specified!
//		// So we need to delete that!
//		wordMLPackage.getMainDocumentPart().getStyleDefinitionsPart().createVirtualStylesForDocDefaults();
//		Style dd = wordMLPackage.getMainDocumentPart().getStyleDefinitionsPart().getStyleById("DocDefaults");
//		dd.getRPr().setSz(null);
//		dd.getRPr().setSzCs(null);
		
		ParagraphStylesInTableFix.process(wordMLPackage);
		
		Style s = getStyle(wordMLPackage, getStyleName());
		Assert.assertTrue(s.getRPr().getSz().getVal().intValue()==EXPECTED_RESULT); 
	}
 

@Test 
	public void testTblStyle_AllSilent() throws Exception {
		
		// Result is to use the implicit 10pt		
	
		WordprocessingMLPackage wordMLPackage = WordprocessingMLPackage.createPackage();
		wordMLPackage.getMainDocumentPart().setContents(
				(Document)XmlUtils.unmarshalString(mdpXml_tblStyle) );
		wordMLPackage.getMainDocumentPart().getStyleDefinitionsPart().setContents(
				(Styles)XmlUtils.unmarshalString(styles_no_font_sz) );

		setSetting(wordMLPackage, OVERRIDE);  // table style should get overridden
		
		wordMLPackage.save(new File(System.getProperty("user.dir") + "/OUT_PStyleInTableTest.docx"));
		
		// NB PropertyResolver puts 10pt in DocDefaults, if nothing is specified!
		
		ParagraphStylesInTableFix.process(wordMLPackage);
		
		Style s = getStyle(wordMLPackage, getStyleName());
		this.assertSz(s, 20);
//		Assert.assertTrue(s.getRPr().getSz().getVal().intValue()==20); 
	}
 

private void repeatRows(final WordprocessingMLPackage document) {
    for (TableRowCoordinates rCoords : tableRowsToRepeat.keySet()) {
        List<Object> expressionContexts = tableRowsToRepeat.get(rCoords);
        int index = rCoords.getIndex();
        for (final Object expressionContext : expressionContexts) {
            Tr rowClone = XmlUtils.deepCopy(rCoords.getRow());
            DocumentWalker walker = new BaseDocumentWalker(rowClone) {
                @Override
                protected void onParagraph(P paragraph) {
                    placeholderReplacer.resolveExpressionsForParagraph(paragraph, expressionContext, document);
                }
            };
            walker.walk();
            rCoords.getParentTableCoordinates().getTable().getContent().add(++index, rowClone);
        }
        rCoords.getParentTableCoordinates().getTable().getContent().remove(rCoords.getRow());
    }
}
 

public Set<Text> mergeMatchedTexts() {
    for (Object paragraphContentObject : paragraph.getContent()) {
        if (paragraphContentObject instanceof R) {
            R currentRun = (R) paragraphContentObject;
            for (Object runContentObject : currentRun.getContent()) {
                Object unwrappedRunContentObject = XmlUtils.unwrap(runContentObject);
                if (unwrappedRunContentObject instanceof Text) {
                    handleText((Text) unwrappedRunContentObject);
                }
            }
        }
    }

    removeUnnecessaryTexts();

    return resultingTexts;
}
 

private void walkTableCell(TableCellCoordinates cellCoordinates) {
    onTableCell(cellCoordinates);
    int elementIndex = 0;
    for (Object cellContentElement : cellCoordinates.getCell().getContent()) {
        if (XmlUtils.unwrap(cellContentElement) instanceof P) {
            P p = (P) cellContentElement;
            ParagraphCoordinates paragraphCoordinates = new ParagraphCoordinates(p, elementIndex, cellCoordinates);
            onParagraph(paragraphCoordinates);
        } else if (XmlUtils.unwrap(cellContentElement) instanceof Tbl) {
            Tbl nestedTable = (Tbl) ((JAXBElement) cellContentElement).getValue();
            TableCoordinates innerTableCoordinates = new TableCoordinates(nestedTable, elementIndex, cellCoordinates);
            walkTable(innerTableCoordinates);
        }
        elementIndex++;
    }
}
 

public void walk() {
    for (Object contentElement : contentAccessor.getContent()) {
        Object unwrappedObject = XmlUtils.unwrap(contentElement);
        if (unwrappedObject instanceof P) {
            P p = (P) unwrappedObject;
            walkParagraph(p);
        } else if (unwrappedObject instanceof Tbl) {
            Tbl table = (Tbl) unwrappedObject;
            walkTable(table);
        } else if (unwrappedObject instanceof Tr) {
            Tr row = (Tr) unwrappedObject;
            walkTableRow(row);
        } else if (unwrappedObject instanceof Tc) {
            Tc cell = (Tc) unwrappedObject;
            walkTableCell(cell);
        } else if (unwrappedObject instanceof CommentRangeStart) {
            CommentRangeStart commentRangeStart = (CommentRangeStart) unwrappedObject;
            onCommentRangeStart(commentRangeStart);
        } else if (unwrappedObject instanceof CommentRangeEnd) {
            CommentRangeEnd commentRangeEnd = (CommentRangeEnd) unwrappedObject;
            onCommentRangeEnd(commentRangeEnd);
        }
    }
}
 

private void walkTable(Tbl table) {
    onTable(table);
    for (Object contentElement : table.getContent()) {
        if (XmlUtils.unwrap(contentElement) instanceof Tr) {
            Tr row = (Tr) contentElement;
            walkTableRow(row);
        }
    }
}
 

private static List<Object> convertToWmlObject(
		WordprocessingMLPackage wordMLPackage, String content)
		throws Docx4JException, JAXBException {
	MainDocumentPart document = wordMLPackage.getMainDocumentPart();
	//获取Jsoup参数
	String charsetName = Docx4jProperties.getProperty(Docx4jConstants.DOCX4J_CONVERT_OUT_WMLTEMPLATE_CHARSETNAME, Docx4jConstants.DEFAULT_CHARSETNAME );
	
	List<Object> wmlObjList = null;
	String templateString = XmlUtils.marshaltoString(document.getContents().getBody());
	System.out.println(templateString);
	Body templateBody = document.getContents().getBody();
	try {
		document.getContents().setBody(XmlUtils.deepCopy(templateBody));
		document.getContent().clear();
		Document doc = Jsoup.parse(content);
		doc.outputSettings().syntax(Document.OutputSettings.Syntax.xml).escapeMode(Entities.EscapeMode.xhtml);
		//XHTMLImporterImpl xhtmlImporter = new XHTMLImporterImpl(wordMLPackage);
		
		AlternativeFormatInputPart  part = document.addAltChunk(AltChunkType.Xhtml,doc.html().getBytes(Charset.forName(charsetName)));
		
		WordprocessingMLPackage tempPackage = document.convertAltChunks();
		File file = new File("d://temp.docx");
		tempPackage.save(file);
		wmlObjList = document.getContent();
		//part.getOwningRelationshipPart().getSourceP().get
		//wmlObjList = xhtmlImporter.convert(doc.html(), doc.baseUri());
	} finally {
		document.getContents().setBody(templateBody);
	}
	return wmlObjList;
}
 

static void addObject(MainDocumentPart wordDocumentPart, String template, String fontName ) throws JAXBException {
	
    HashMap substitution = new HashMap();
    substitution.put("fontname", fontName);
    Object o = XmlUtils.unmarshallFromTemplate(template, substitution);
    wordDocumentPart.addObject(o);    		    
	
}
 

private void walkTableRow(Tr row) {
    onTableRow(row);
    for (Object rowContentElement : row.getContent()) {
        if (XmlUtils.unwrap(rowContentElement) instanceof Tc) {
            Tc cell = rowContentElement instanceof Tc ? (Tc) rowContentElement : (Tc) ((JAXBElement) rowContentElement).getValue();
            walkTableCell(cell);
        }
    }
}
 

public static <T> List<T> getTargetElements(Object source, Class<T> targetClass) {  
    List<T> result = new ArrayList<T>();  
    //获取真实的对象
	Object target = XmlUtils.unwrap(source);
    //if (target.getClass().equals(targetClass)) {
	if (targetClass.isAssignableFrom(target.getClass())) { 
        result.add((T) target);  
    } else if (target instanceof ContentAccessor) {  
        List<?> children = ((ContentAccessor) target).getContent();  
        for (Object child : children) {  
            result.addAll(getTargetElements(child, targetClass));  
        }
    }
    return result;  
}
 

public void walkJAXBElements(Object parent) {
    List children = getChildren(parent);
    if (children != null) {

        for (Object o : children) {
            o = XmlUtils.unwrap(o);

            if (o instanceof Child && !(parent instanceof SdtBlock)) {
                ((Child) o).setParent(parent);
            }

            if (o instanceof Tbl) {
                currentTables.push(new TableManager(docxFormatter, (Tbl) o));
            }

            this.apply(o);

            if (this.shouldTraverse(o)) {
                walkJAXBElements(o);
            }

            if (o instanceof Tbl) {
                TableManager currentTable = currentTables.pop();
                currentTable.setSkipIt(false);
            }
        }
    }
}
 

private void walkTableCell(Tc cell) {
    onTableCell(cell);
    for (Object cellContentElement : cell.getContent()) {
        if (XmlUtils.unwrap(cellContentElement) instanceof P) {
            P p = (P) cellContentElement;
            walkParagraph(p);
        } else if (XmlUtils.unwrap(cellContentElement) instanceof Tbl) {
            Tbl nestedTable = (Tbl) ((JAXBElement) cellContentElement).getValue();
            walkTable(nestedTable);
        }
    }
}
 

/** 
 * 获取单元格字符串 
 */  
private String[] getTcMarshalStr(Tr tr) {  
    List<Object> tcObjList = tr.getContent();  
    String[] marshaArr = new String[7];  
    // 跳过层次  
    for (int i = 1, len = tcObjList.size(); i < len; i++) {  
        marshaArr[i - 1] = XmlUtils.marshaltoString(tcObjList.get(i), true, false);  
    }  
    return marshaArr;  
}