private boolean shouldSkip(StartElement se) {
try {
if (skips == null) {
skips = new HashSet<StartElement>();
XMLEventFactory xef = StAXEventFactoryPool.getInstance().acquire();
Set<Attribute> attrs;
skips.add(
xef.createStartElement(
new QName(Namespaces.Z2005_DTBOOK_NS_URI, "pagenum"), null, null));
skips.add(
xef.createStartElement(
new QName(Namespaces.Z2005_DTBOOK_NS_URI, "noteref"), null, null));
skips.add(
xef.createStartElement(
new QName(Namespaces.Z2005_DTBOOK_NS_URI, "annoref"), null, null));
attrs = new HashSet<Attribute>();
attrs.add(xef.createAttribute("class", "page-normal"));
skips.add(
xef.createStartElement(
new QName(Namespaces.XHTML_10_NS_URI, "span"), attrs.iterator(), null));
attrs = new HashSet<Attribute>();
attrs.add(xef.createAttribute("class", "page-front"));
skips.add(
xef.createStartElement(
new QName(Namespaces.XHTML_10_NS_URI, "span"), attrs.iterator(), null));
attrs = new HashSet<Attribute>();
attrs.add(xef.createAttribute("class", "page-special"));
skips.add(
xef.createStartElement(
new QName(Namespaces.XHTML_10_NS_URI, "span"), attrs.iterator(), null));
attrs = new HashSet<Attribute>();
attrs.add(xef.createAttribute("class", "noteref"));
skips.add(
xef.createStartElement(
new QName(Namespaces.XHTML_10_NS_URI, "span"), attrs.iterator(), null));
StAXEventFactoryPool.getInstance().release(xef);
}
for (StartElement test : skips) {
boolean matchesName = true;
boolean matchesAttrs = true;
if (se.getName().equals(test.getName())) {
Iterator<?> iter = test.getAttributes();
// real elem must have all attrs of test elem
while (iter.hasNext()) {
Attribute testAttr = (Attribute) iter.next();
Attribute realAttr = AttributeByName.get(testAttr.getName(), se);
if (realAttr == null || !realAttr.getValue().equals(testAttr.getValue())) {
matchesAttrs = false;
}
}
} else {
matchesName = false;
}
if (matchesName && matchesAttrs) return true;
}
} catch (Exception e) {
e.printStackTrace();
}
return false;
}
private void scramble(EFile input) throws XMLStreamException, IOException {
File output =
new File(
input.getParentFile(),
input.getNameMinusExtension() + ".scrambled." + input.getExtension());
Map<String, Object> xifProperties = null;
Map<String, Object> xofProperties = null;
XMLInputFactory xif = null;
XMLOutputFactory xof = null;
XMLEventWriter writer = null;
XMLEventFactory xef = null;
FileInputStream fis = null;
FileOutputStream fos = null;
try {
xifProperties = StAXInputFactoryPool.getInstance().getDefaultPropertyMap(false);
xofProperties = StAXOutputFactoryPool.getInstance().getDefaultPropertyMap();
xif = StAXInputFactoryPool.getInstance().acquire(xifProperties);
xif.setXMLResolver(new StaxEntityResolver(CatalogEntityResolver.getInstance()));
xof = StAXOutputFactoryPool.getInstance().acquire(xofProperties);
xef = StAXEventFactoryPool.getInstance().acquire();
fis = new FileInputStream(input);
fos = new FileOutputStream(output);
XMLEventReader baseReader = xif.createXMLEventReader(fis);
writer = xof.createXMLEventWriter(fos);
BookmarkedXMLEventReader reader = new BookmarkedXMLEventReader(baseReader);
ContextStack context = new ContextStack(true);
boolean skipElemTextScrambling = false;
int c = 0;
while (reader.hasNext()) {
XMLEvent xe = reader.nextEvent();
context.addEvent(xe);
if (xe.isStartElement()) {
skipElemTextScrambling = shouldSkip(xe.asStartElement());
if (isMetaElement(xe.asStartElement())) {
xe = handleMetaElement(xe.asStartElement(), xef);
} else if (isImageElement(xe.asStartElement())) {
xe = handleImageElement(xe.asStartElement(), xef);
} else if (isAcronymElement(xe.asStartElement())) {
xe = handleAcronymElement(xe.asStartElement(), xef);
} else if (isLinkElement(xe.asStartElement())) {
xe = handleLinkElement(xe.asStartElement(), xef);
} else if (isAnchorElement(xe.asStartElement())) {
xe = handleAnchorElement(xe.asStartElement(), xef);
}
} else if (xe.isCharacters()
&& !skipElemTextScrambling
&& !CharUtils.isXMLWhiteSpace(xe.asCharacters().getData())) {
xe =
xef.createCharacters(
"["
+ Integer.toString(++c)
+ "] "
+ context.getContextXPath(
ContextStack.XPATH_SELECT_ELEMENTS_ONLY,
ContextStack.XPATH_PREDICATES_NONE));
} else if (xe.getEventType() == XMLEvent.PROCESSING_INSTRUCTION) {
xe = handleProcessingInstruction((ProcessingInstruction) xe, xef);
} else if (xe.isEndElement()) {
skipElemTextScrambling = false;
}
if (xe != null) writer.add(xe);
}
} catch (CatalogExceptionNotRecoverable e) {
e.printStackTrace();
} finally {
fis.close();
fos.close();
writer.flush();
writer.close();
StAXInputFactoryPool.getInstance().release(xif, xifProperties);
StAXOutputFactoryPool.getInstance().release(xof, xofProperties);
StAXEventFactoryPool.getInstance().release(xef);
}
}