/**
* Counts the number of characters (text nodes) in this element excluding footnotes etc.
*
* @param node the node to count in
* @return the number of characters
*/
public static int getCharacterCount(Node node) {
Node child = node.getFirstChild();
int nCount = 0;
while (child != null) {
short nodeType = child.getNodeType();
switch (nodeType) {
case Node.TEXT_NODE:
nCount += child.getNodeValue().length();
break;
case Node.ELEMENT_NODE:
String sName = child.getNodeName();
if (sName.equals(XMLString.TEXT_S)) {
nCount += Misc.getPosInteger(Misc.getAttribute(child, XMLString.TEXT_C), 1);
} else if (sName.equals(XMLString.TEXT_TAB_STOP)) {
nCount++; // treat as single space
} else if (sName.equals(XMLString.TEXT_TAB)) { // oasis
nCount++; // treat as single space
} else if (isNoteElement(child)) {
// ignore
} else if (isTextElement(child)) {
nCount += getCharacterCount(child);
}
}
child = child.getNextSibling();
}
return nCount;
}
private void loadStylesFromDOM(
Document stylesDOM, Document contentDOM, boolean bAllParagraphsAreSoft) {
// Flat xml: stylesDOM will be null and contentDOM contain everything
// This is only the case for old versions of xmerge; newer versions
// creates DOM for styles, content, meta and settings.
NodeList list;
// font declarations: Try old format first
if (stylesDOM == null) {
list = contentDOM.getElementsByTagName(XMLString.OFFICE_FONT_DECLS);
} else {
list = stylesDOM.getElementsByTagName(XMLString.OFFICE_FONT_DECLS);
}
// If that fails, try oasis format
if (list.getLength() == 0) {
if (stylesDOM == null) {
list = contentDOM.getElementsByTagName(XMLString.OFFICE_FONT_FACE_DECLS);
} else {
list = stylesDOM.getElementsByTagName(XMLString.OFFICE_FONT_FACE_DECLS);
}
}
if (list.getLength() != 0) {
Node node = list.item(0);
if (node.hasChildNodes()) {
NodeList nl = node.getChildNodes();
int nLen = nl.getLength();
for (int i = 0; i < nLen; i++) {
Node child = nl.item(i);
if (child.getNodeType() == Node.ELEMENT_NODE) {
if (child.getNodeName().equals(XMLString.STYLE_FONT_DECL)) {
font.loadStyleFromDOM(child, false);
} else if (child.getNodeName().equals(XMLString.STYLE_FONT_FACE)) {
font.loadStyleFromDOM(child, false);
}
}
}
}
}
// soft formatting:
if (stylesDOM == null) {
list = contentDOM.getElementsByTagName(XMLString.OFFICE_STYLES);
} else {
list = stylesDOM.getElementsByTagName(XMLString.OFFICE_STYLES);
}
if (list.getLength() != 0) {
loadStylesFromDOM(list.item(0), bAllParagraphsAreSoft);
}
// master styles:
if (stylesDOM == null) {
list = contentDOM.getElementsByTagName(XMLString.OFFICE_MASTER_STYLES);
} else {
list = stylesDOM.getElementsByTagName(XMLString.OFFICE_MASTER_STYLES);
}
if (list.getLength() != 0) {
loadStylesFromDOM(list.item(0), bAllParagraphsAreSoft);
}
// hard formatting:
// Load from styles.xml first. Problem: There may be name clashes
// with automatic styles from content.xml
if (stylesDOM != null) {
list = stylesDOM.getElementsByTagName(XMLString.OFFICE_AUTOMATIC_STYLES);
if (list.getLength() != 0) {
loadStylesFromDOM(list.item(0), bAllParagraphsAreSoft);
}
}
list = contentDOM.getElementsByTagName(XMLString.OFFICE_AUTOMATIC_STYLES);
if (list.getLength() != 0) {
loadStylesFromDOM(list.item(0), bAllParagraphsAreSoft);
}
// footnotes configuration:
if (stylesDOM == null) {
list = contentDOM.getElementsByTagName(XMLString.TEXT_FOOTNOTES_CONFIGURATION);
} else {
list = stylesDOM.getElementsByTagName(XMLString.TEXT_FOOTNOTES_CONFIGURATION);
}
if (list.getLength() != 0) {
footnotes = new PropertySet();
footnotes.loadFromDOM(list.item(0));
}
// endnotes configuration:
if (stylesDOM == null) {
list = contentDOM.getElementsByTagName(XMLString.TEXT_ENDNOTES_CONFIGURATION);
} else {
list = stylesDOM.getElementsByTagName(XMLString.TEXT_ENDNOTES_CONFIGURATION);
}
if (list.getLength() != 0) {
endnotes = new PropertySet();
endnotes.loadFromDOM(list.item(0));
}
// if it failed, try oasis format
if (footnotes == null || endnotes == null) {
if (stylesDOM == null) {
list = contentDOM.getElementsByTagName(XMLString.TEXT_NOTES_CONFIGURATION);
} else {
list = stylesDOM.getElementsByTagName(XMLString.TEXT_NOTES_CONFIGURATION);
}
int nLen = list.getLength();
for (int i = 0; i < nLen; i++) {
String sClass = Misc.getAttribute(list.item(i), XMLString.TEXT_NOTE_CLASS);
if ("endnote".equals(sClass)) {
endnotes = new PropertySet();
endnotes.loadFromDOM(list.item(i));
} else {
footnotes = new PropertySet();
footnotes.loadFromDOM(list.item(i));
}
}
}
}
private void loadStylesFromDOM(Node node, boolean bAllParagraphsAreSoft) {
// node should be office:master-styles, office:styles or office:automatic-styles
boolean bAutomatic = XMLString.OFFICE_AUTOMATIC_STYLES.equals(node.getNodeName());
if (node.hasChildNodes()) {
NodeList nl = node.getChildNodes();
int nLen = nl.getLength();
for (int i = 0; i < nLen; i++) {
Node child = nl.item(i);
if (child.getNodeType() == Node.ELEMENT_NODE) {
if (child.getNodeName().equals(XMLString.STYLE_STYLE)) {
String sFamily = Misc.getAttribute(child, XMLString.STYLE_FAMILY);
if ("text".equals(sFamily)) {
text.loadStyleFromDOM(child, bAutomatic);
} else if ("paragraph".equals(sFamily)) {
par.loadStyleFromDOM(child, bAutomatic && !bAllParagraphsAreSoft);
} else if ("section".equals(sFamily)) {
section.loadStyleFromDOM(child, bAutomatic);
} else if ("table".equals(sFamily)) {
table.loadStyleFromDOM(child, bAutomatic);
} else if ("table-column".equals(sFamily)) {
column.loadStyleFromDOM(child, bAutomatic);
} else if ("table-row".equals(sFamily)) {
row.loadStyleFromDOM(child, bAutomatic);
} else if ("table-cell".equals(sFamily)) {
cell.loadStyleFromDOM(child, bAutomatic);
} else if ("graphics".equals(sFamily)) {
frame.loadStyleFromDOM(child, bAutomatic);
} else if ("graphic".equals(sFamily)) { // oasis
frame.loadStyleFromDOM(child, bAutomatic);
} else if ("presentation".equals(sFamily)) {
presentation.loadStyleFromDOM(child, bAutomatic);
} else if ("drawing-page".equals(sFamily)) {
// Bug in OOo 1.x: The same name may be used for a real and an automatic style...
if (drawingPage.getStyle(Misc.getAttribute(child, XMLString.STYLE_NAME)) == null) {
drawingPage.loadStyleFromDOM(child, bAutomatic);
}
}
} else if (child.getNodeName().equals(XMLString.STYLE_PAGE_MASTER)) { // old
pageLayout.loadStyleFromDOM(child, bAutomatic);
} else if (child.getNodeName().equals(XMLString.STYLE_PAGE_LAYOUT)) { // oasis
pageLayout.loadStyleFromDOM(child, bAutomatic);
} else if (child.getNodeName().equals(XMLString.STYLE_MASTER_PAGE)) {
masterPage.loadStyleFromDOM(child, bAutomatic);
if (firstMasterPage == null) {
firstMasterPage =
(MasterPage) masterPage.getStyle(Misc.getAttribute(child, XMLString.STYLE_NAME));
}
} else if (child.getNodeName().equals(XMLString.TEXT_LIST_STYLE)) {
list.loadStyleFromDOM(child, bAutomatic);
} else if (child.getNodeName().equals(XMLString.TEXT_OUTLINE_STYLE)) {
outline.loadStyleFromDOM(child);
} else if (child.getNodeName().equals(XMLString.STYLE_DEFAULT_STYLE)) {
String sFamily = Misc.getAttribute(child, XMLString.STYLE_FAMILY);
if ("paragraph".equals(sFamily)) {
StyleWithProperties defaultPar = new StyleWithProperties();
defaultPar.loadStyleFromDOM(child);
par.setDefaultStyle(defaultPar);
} else if ("graphics".equals(sFamily) || "graphic".equals(sFamily)) { // oasis: no s
StyleWithProperties defaultFrame = new StyleWithProperties();
defaultFrame.loadStyleFromDOM(child);
frame.setDefaultStyle(defaultFrame);
} else if ("table-cell".equals(sFamily)) {
StyleWithProperties defaultCell = new StyleWithProperties();
defaultCell.loadStyleFromDOM(child);
cell.setDefaultStyle(defaultCell);
}
}
}
}
}
}
private void traverseContent(Element node) {
// Handle this node first
String sName = node.getTagName();
if (sName.equals(XMLString.TEXT_P)) {
// collectMasterPage(getParStyle(node.getAttribute(XMLString.TEXT_STYLE_NAME)));
} else if (sName.equals(XMLString.TEXT_H)) {
int nLevel;
if (node.hasAttribute(XMLString.TEXT_OUTLINE_LEVEL)) {
nLevel = Misc.getPosInteger(node.getAttribute(XMLString.TEXT_OUTLINE_LEVEL), 1);
} else {
nLevel = Misc.getPosInteger(node.getAttribute(XMLString.TEXT_LEVEL), 1);
}
StyleWithProperties style = getParStyle(node.getAttribute(XMLString.TEXT_STYLE_NAME));
// collectMasterPage(style);
if (1 <= nLevel && nLevel <= 10 && heading[nLevel] == null) {
if (style != null && style.isAutomatic()) {
heading[nLevel] = getParStyle(style.getParentName());
} else {
heading[nLevel] = style;
}
}
} else if (sName.equals(XMLString.TEXT_SEQUENCE)) {
String sSeqName = Misc.getAttribute(node, XMLString.TEXT_NAME);
String sRefName = Misc.getAttribute(node, XMLString.TEXT_REF_NAME);
if (sSeqName != null) {
Element par = getParagraph(node);
if (!sequenceNames.containsKey(par)) {
// Only the first text:seqence should be registered as possible caption sequence
sequenceNames.put(par, sSeqName);
}
if (sRefName != null) {
seqrefNames.put(sRefName, sSeqName);
}
}
} else if (sName.equals(XMLString.TEXT_FOOTNOTE_REF)) {
collectRefName(footnoteRef, node);
} else if (sName.equals(XMLString.TEXT_ENDNOTE_REF)) {
collectRefName(endnoteRef, node);
} else if (sName.equals(XMLString.TEXT_NOTE_REF)) { // oasis
String sClass = Misc.getAttribute(node, XMLString.TEXT_NOTE_CLASS);
if ("footnote".equals(sClass)) {
collectRefName(footnoteRef, node);
} else if ("endnote".equals(sClass)) {
collectRefName(endnoteRef, node);
}
} else if (sName.equals(XMLString.TEXT_REFERENCE_MARK)) {
collectMarkInHeading(referenceHeading, node);
} else if (sName.equals(XMLString.TEXT_REFERENCE_MARK_START)) {
collectMarkInHeading(referenceHeading, node);
} else if (sName.equals(XMLString.TEXT_REFERENCE_REF)) {
collectRefName(referenceRef, node);
} else if (sName.equals(XMLString.TEXT_BOOKMARK)) {
collectMarkInHeading(bookmarkHeading, node);
} else if (sName.equals(XMLString.TEXT_BOOKMARK_START)) {
collectMarkInHeading(bookmarkHeading, node);
} else if (sName.equals(XMLString.TEXT_BOOKMARK_REF)) {
collectRefName(bookmarkRef, node);
} else if (sName.equals(XMLString.TEXT_SEQUENCE_REF)) {
collectRefName(sequenceRef, node);
} else if (sName.equals(XMLString.TEXT_A)) {
String sHref = node.getAttribute(XMLString.XLINK_HREF);
if (sHref != null && sHref.startsWith("#")) {
links.add(sHref.substring(1));
}
} else if (sName.equals(XMLString.OFFICE_FORMS)) {
forms.read(node);
} else if (sName.equals(XMLString.TEXT_TABLE_OF_CONTENT)) {
TocReader tocReader = new TocReader(node);
indexes.put(node, tocReader);
indexSourceStyles.addAll(tocReader.getIndexSourceStyles());
} else if (sName.equals(XMLString.TEXT_TABLE_INDEX)
|| sName.equals(XMLString.TEXT_ILLUSTRATION_INDEX)) {
LoftReader loftReader = new LoftReader(node);
indexes.put(node, loftReader);
if (loftReader.useCaption()) {
if (loftReader.isTableIndex()) {
tableSequenceNames.add(loftReader.getCaptionSequenceName());
} else {
figureSequenceNames.add(loftReader.getCaptionSequenceName());
}
}
}
// todo: other indexes
// Traverse the children
Node child = node.getFirstChild();
while (child != null) {
if (child.getNodeType() == Node.ELEMENT_NODE) {
traverseContent((Element) child);
}
child = child.getNextSibling();
}
// Collect automatic captions sequences
// Use OOo defaults: Captions have style names Illustration and Table resp.
if ((sAutoFigureSequenceName == null || sAutoTableSequenceName == null)
&& sName.equals(XMLString.TEXT_P)) {
String sStyleName =
getParStyles().getDisplayName(node.getAttribute(XMLString.TEXT_STYLE_NAME));
if (sAutoFigureSequenceName == null) {
if ("Illustration".equals(sStyleName)) {
sAutoFigureSequenceName = getSequenceName(node);
}
}
if (sAutoTableSequenceName == null) {
if ("Table".equals(sStyleName)) {
sAutoTableSequenceName = getSequenceName(node);
}
}
}
}