public static boolean isEmpty(Node node, boolean doFilter) {
return node == null
|| node.nodeName().equals("#comment")
|| node.nodeName().equals("#data")
|| node.nodeName().equals("style")
|| node.nodeName().equals("script")
|| isHidden(node)
|| (doFilter && isFiltered(node))
|| (node.nodeName().equals("#text") && CommonUtil.isEmpty(node.toString(), true));
}
private Element cleanupElement(Element el) {
Tag newTag = null;
String newText = null;
if (el.nodeName().equals("img")) {
newTag = Tag.valueOf("x");
newText = el.attr("src");
}
if (el.nodeName().equals("em")) {
newTag = Tag.valueOf("b");
}
if (el.nodeName().equals("a")) {
String clazz = el.attr("class");
if (clazz.equals("user")) {
newTag = Tag.valueOf("x");
newText = "@" + el.text().trim();
} else if (clazz.startsWith("postimg video")) {
newTag = Tag.valueOf("x");
newText = "VIDEO: " + el.attr("href") + " THUMBNAIL: " + el.select("img").attr("src");
} else if (clazz.startsWith("postimg")) {
newTag = Tag.valueOf("x");
} else if (clazz.equals("post")) {
newTag = Tag.valueOf("x");
} else {
newTag = Tag.valueOf("x");
newText = el.attr("href");
}
}
if (el.nodeName().equals("div")) {
newTag = Tag.valueOf("x");
}
Element nel;
if (newTag == null) {
// el = el;
nel = new Element(el.tag(), "");
// for(List<Node> children = nel.childNodes(); children.size() > 0; children =
// nel.childNodes()) {
// children.get(0).remove();
// }
} else {
nel = new Element(newTag, "");
}
if (newText != null) {
nel.appendChild(new TextNode(newText, ""));
} else {
List<Node> children = el.childNodes();
for (Node child : children) {
if (child instanceof Element) {
nel.appendChild(cleanupElement((Element) child));
} else {
nel.appendChild(new TextNode(child.toString(), ""));
}
}
}
return nel;
}
private Elements parseNextNode(String query) {
if (!NEXT_NODE_TAG.equals(query)) {
throw new IllegalArgumentException("Argument selector part: " + query + " is illegal");
} else {
Elements eles = new Elements();
if (elements.size() == 1) {
Attributes attributes = new Attributes();
Node nextNode = elements.first().nextSibling();
if (nextNode == null) {
return eles;
}
attributes.put("value", nextNode.toString());
eles.add(new Element(Tag.valueOf("nextnode"), "", attributes));
} else {
eles = elements;
}
return eles;
}
}
@Override
public void readDesign(Element design, DesignContext designContext) {
super.readDesign(design, designContext);
String altText = "";
for (Node child : design.childNodes()) {
if (child instanceof Element
&& ((Element) child).tagName().equals("source")
&& child.hasAttr("href")) {
addSource(DesignAttributeHandler.readAttribute("href", child.attributes(), Resource.class));
} else {
altText += child.toString();
}
}
altText = altText.trim();
if (!altText.isEmpty()) {
setAltText(altText);
}
}
// hit when the node is first seen
public void head(Node node, int depth) {
String name = node.nodeName();
if (name.equals("li")) append('\n');
else if (node.toString().startsWith("<select")) {
append1("[SELECT]");
append(tab);
} else if (node.outerHtml().startsWith("<option")) {
// append1(node.attr("value")+":");append1(" ");
TextNodeVisitor textVisitor = new TextNodeVisitor();
node.traverse(textVisitor);
append1("{" + textVisitor.toString() + "}");
} else if (node.outerHtml().startsWith("<input")) {
if (node.attr("type").equals("input")) append1("[INPUT]" + node.attr("maxLength"));
} else if (node.outerHtml().startsWith("<span")) {
TextNodeVisitor textVisitor = new TextNodeVisitor();
node.traverse(textVisitor);
append1(":" + textVisitor.toString() + " ");
}
}
