private XMLEvent handleAnchorElement(StartElement se, XMLEventFactory xef) {
Set<Attribute> attrs = new HashSet<Attribute>();
Attribute href = AttributeByName.get(new QName("href"), se);
if (href != null) {
String value = "http";
if (href.getValue().toLowerCase().contains(("mailto"))) {
value = "mailto:[email protected]";
} else if (href.getValue().trim().startsWith("#")) {
value = href.getValue();
} else if (href.getValue().toLowerCase().contains("smil")) {
value = href.getValue();
} else {
value = "http://dummy.org";
}
attrs.add(xef.createAttribute(href.getName(), value));
}
Iterator<?> i = se.getAttributes();
while (i.hasNext()) {
Attribute a = (Attribute) i.next();
if (!a.getName().getLocalPart().equals("href")) {
attrs.add(a);
}
}
return xef.createStartElement(se.getName(), attrs.iterator(), se.getNamespaces());
}
private XMLEvent handleLinkElement(StartElement se, XMLEventFactory xef) {
// remove all links since they may refer to copyrighted stuff
// unless its a css link in which case we create a new one
Attribute type = AttributeByName.get(new QName("type"), se);
Attribute rel = AttributeByName.get(new QName("rel"), se);
if ((type != null && type.getValue().toLowerCase().equals("text/css"))
|| (rel != null && rel.getValue().toLowerCase().equals("stylesheet"))) {
Set<Attribute> attrs = new HashSet<Attribute>();
attrs.add(xef.createAttribute(new QName("href"), "dummy.css"));
Iterator<?> i = se.getAttributes();
while (i.hasNext()) {
Attribute a = (Attribute) i.next();
if (a.getName().getLocalPart() != "href") {
attrs.add(a);
}
}
return xef.createStartElement(se.getName(), attrs.iterator(), se.getNamespaces());
}
return null;
}
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;
}