/**
* VIPS Rule Nine
*
* <p>If the child of the node with maximum size are small than a threshold (relative size), do
* not divide this node. <br>
* Set the DoC based on the html tag and size of this node.
*
* @param node Input node
* @return True, if rule is applied, otherwise false.
*/
private boolean ruleNine(ElementBox node) {
// System.err.println("Applying rule Nine on " + node.getNode().getNodeName() + " node");
if (node.getSubBoxList().isEmpty()) return false;
int maxSize = 0;
for (Box childNode : node.getSubBoxList()) {
int childSize = childNode.getWidth() * childNode.getHeight();
if (maxSize < childSize) {
maxSize = childSize;
}
}
if (maxSize > _sizeTresholdWidth * _sizeTresholdHeight) return true;
// TODO set DOC
_currentVipsBlock.setIsVisualBlock(true);
_currentVipsBlock.setIsDividable(false);
if (node.getNode().getNodeName().equals("Xdiv")) _currentVipsBlock.setDoC(7);
if (node.getNode().getNodeName().equals("a")) _currentVipsBlock.setDoC(11);
else _currentVipsBlock.setDoC(8);
return true;
}
/**
* Counts number of visual blocks in visual structure
*
* @param vipsBlock Visual structure
*/
private void getVisualBlocksCount(VipsBlock vipsBlock) {
if (vipsBlock.isVisualBlock()) _visualBlocksCount++;
for (VipsBlock vipsBlockChild : vipsBlock.getChildren()) {
if (!(vipsBlockChild.getBox() instanceof TextBox)) getVisualBlocksCount(vipsBlockChild);
}
}
/**
* Finds previous sibling node's VIPS block.
*
* @param node Node
* @param vipsBlock Actual VIPS block
* @param foundBlock VIPS block for given node
*/
private void findPreviousSiblingNodeVipsBlock(Node node, VipsBlock vipsBlock) {
if (vipsBlock.getBox().getNode().equals(node)) {
_tempVipsBlock = vipsBlock;
return;
} else
for (VipsBlock vipsBlockChild : vipsBlock.getChildren())
findPreviousSiblingNodeVipsBlock(node, vipsBlockChild);
}
/**
* Construct VIPS block tree from viewport.
*
* <p>Starts from <body> element.
*
* @param element Box that represents element
* @param node Visual structure tree node
*/
private void constructVipsBlockTree(Box element, VipsBlock node) {
node.setBox(element);
if (!(element instanceof TextBox)) {
for (Box box : ((ElementBox) element).getSubBoxList()) {
node.addChild(new VipsBlock());
constructVipsBlockTree(box, node.getChildren().get(node.getChildren().size() - 1));
}
}
}
/**
* VIPS Rule Three
*
* <p>If the DOM node is the root node of the sub-DOM tree (corresponding to the block), and there
* is only one sub DOM tree corresponding to this block, divide this node.
*
* @param node Input node
* @return True, if rule is applied, otherwise false.
*/
private boolean ruleThree(ElementBox node) {
// System.err.println("Applying rule Three on " + node.getNode().getNodeName() + " node");
if (!node.isRootElement()) return false;
boolean result = true;
int cnt = 0;
for (VipsBlock vipsBlock : _vipsBlocks.getChildren()) {
if (vipsBlock.getBox().getNode().getNodeName().equals(node.getNode().getNodeName())) {
result = true;
isOnlyOneDomSubTree(node.getNode(), vipsBlock.getBox().getNode(), result);
if (result) cnt++;
}
}
return (cnt == 1) ? true : false;
}
/*
* Checks if node has valid children nodes
*/
private boolean hasValidChildrenNodes(ElementBox node) {
if (node.getNode().getNodeName().equals("img")
|| node.getNode().getNodeName().equals("input")) {
if (node.getContentWidth() > 0 && node.getContentHeight() > 0) {
_currentVipsBlock.setIsVisualBlock(true);
_currentVipsBlock.setDoC(8);
return true;
} else return false;
}
if (node.getSubBoxList().isEmpty()) return false;
_cnt = 0;
for (Box child : node.getSubBoxList()) {
checkValidChildrenNodes(child);
}
return (_cnt > 0) ? true : false;
}
/**
* VIPS Rule One
*
* <p>If the DOM node is not a text node and it has no valid children, then this node cannot be
* divided and will be cut.
*
* @param node Input node
* @return True, if rule is applied, otherwise false.
*/
private boolean ruleOne(ElementBox node) {
// System.err.println("Applying rule One on " + node.getNode().getNodeName() + " node");
if (!isTextNode(node)) {
if (!hasValidChildrenNodes(node)) {
_currentVipsBlock.setIsDividable(false);
return true;
}
}
return false;
}
/**
* VIPS Rule Ten
*
* <p>If previous sibling node has not been divided, do not divide this node
*
* @param node Input node
* @return True, if rule is applied, otherwise false.
*/
private boolean ruleTen(ElementBox node) {
// System.err.println("Applying rule Ten on " + node.getNode().getNodeName() + " node");
// VipsBlock previousSiblingVipsBlock = null;
// findPreviousSiblingNodeVipsBlock(node.getNode().getPreviousSibling(), _vipsBlocks,
// previousSiblingVipsBlock);
_tempVipsBlock = null;
findPreviousSiblingNodeVipsBlock(node.getNode().getPreviousSibling(), _vipsBlocks);
if (_tempVipsBlock == null) return false;
if (_tempVipsBlock.isAlreadyDivided()) return true;
return false;
}
/**
* VIPS Rule Twelve
*
* <p>Do not divide this node <br>
* Set the DoC value based on the html tag and size of this node.
*
* @param node Input node
* @return True, if rule is applied, otherwise false.
*/
private boolean ruleTwelve(ElementBox node) {
// System.err.println("Applying rule Twelve on " + node.getNode().getNodeName() + " node");
_currentVipsBlock.setIsDividable(false);
_currentVipsBlock.setIsVisualBlock(true);
if (node.getNode().getNodeName().equals("Xdiv")) _currentVipsBlock.setDoC(7);
else if (node.getNode().getNodeName().equals("li")) _currentVipsBlock.setDoC(8);
else if (node.getNode().getNodeName().equals("span")) _currentVipsBlock.setDoC(8);
else if (node.getNode().getNodeName().equals("sup")) _currentVipsBlock.setDoC(8);
else if (node.getNode().getNodeName().equals("img")) _currentVipsBlock.setDoC(8);
else _currentVipsBlock.setDoC(333);
// TODO DoC Part
return true;
}
/**
* VIPS Rule Seven
*
* <p>If the background color of this node is different from one of its children’s, divide this
* node and at the same time, the child node with different background color will not be divided
* in this round. Set the DoC value (6-8) for the child node based on the <html> tag of the
* child node and the size of the child node.
*
* @param node Input node
* @return True, if rule is applied, otherwise false.
*/
private boolean ruleSeven(ElementBox node) {
// System.err.println("Applying rule Seven on " + node.getNode().getNodeName() + " node");
if (node.getSubBoxList().isEmpty()) return false;
if (isTextNode(node)) return false;
// String nodeBgColor = node.getStylePropertyValue("background-color");
String nodeBgColor = _currentVipsBlock.getBgColor();
for (VipsBlock vipsStructureChild : _currentVipsBlock.getChildren()) {
if (!(vipsStructureChild.getBgColor().equals(nodeBgColor))) {
vipsStructureChild.setIsDividable(false);
vipsStructureChild.setIsVisualBlock(true);
// TODO DoC values
vipsStructureChild.setDoC(7);
return true;
}
}
return false;
}
/**
* VIPS Rule Eight
*
* <p>If the node has at least one text node child or at least one virtual text node child, and
* the node's relative size is smaller than a threshold, then the node cannot be divided. Set the
* DoC value (from 5-8) based on the html tag of the node.
*
* @param node Input node
* @return True, if rule is applied, otherwise false.
*/
private boolean ruleEight(ElementBox node) {
// System.err.println("Applying rule Eight on " + node.getNode().getNodeName() + " node");
if (node.getSubBoxList().isEmpty()) return false;
List<Box> children = new ArrayList<Box>();
findTextChildrenNodes(node, children);
int cnt = children.size();
if (cnt == 0) return false;
if (node.getWidth() == 0 || node.getHeight() == 0) {
children.clear();
getAllChildren(node, children);
for (Box child : children) {
if (child.getWidth() != 0 && child.getHeight() != 0) return true;
}
}
if (node.getWidth() * node.getHeight() > _sizeTresholdHeight * _sizeTresholdWidth) return false;
if (node.getNode().getNodeName().equals("ul")) {
return true;
}
_currentVipsBlock.setIsVisualBlock(true);
_currentVipsBlock.setIsDividable(false);
if (node.getNode().getNodeName().equals("Xdiv")) _currentVipsBlock.setDoC(7);
else if (node.getNode().getNodeName().equals("code")) _currentVipsBlock.setDoC(7);
else if (node.getNode().getNodeName().equals("div")) _currentVipsBlock.setDoC(5);
else _currentVipsBlock.setDoC(8);
return true;
}
/**
* Append node from given visual structure to parent node
*
* @param parentNode Given visual structure
* @param visualStructure Parent node
*/
private void writeVisualBlocks(Element parentNode, VisualStructure visualStructure) {
Element layoutNode = doc.createElement("LayoutNode");
layoutNode.setAttribute(
"FrameSourceIndex", String.valueOf(visualStructure.getFrameSourceIndex()));
layoutNode.setAttribute("SourceIndex", visualStructure.getSourceIndex());
layoutNode.setAttribute("DoC", String.valueOf(visualStructure.getDoC()));
layoutNode.setAttribute("ContainImg", String.valueOf(visualStructure.containImg()));
layoutNode.setAttribute("IsImg", String.valueOf(visualStructure.isImg()));
layoutNode.setAttribute("ContainTable", String.valueOf(visualStructure.containTable()));
layoutNode.setAttribute("ContainP", String.valueOf(visualStructure.containP()));
layoutNode.setAttribute("TextLen", String.valueOf(visualStructure.getTextLength()));
layoutNode.setAttribute("LinkTextLen", String.valueOf(visualStructure.getLinkTextLength()));
Box parentBox = visualStructure.getNestedBlocks().get(0).getBox().getParent();
layoutNode.setAttribute(
"DOMCldNum", String.valueOf(parentBox.getNode().getChildNodes().getLength()));
layoutNode.setAttribute("FontSize", String.valueOf(visualStructure.getFontSize()));
layoutNode.setAttribute("FontWeight", String.valueOf(visualStructure.getFontWeight()));
layoutNode.setAttribute("BgColor", visualStructure.getBgColor());
layoutNode.setAttribute("ObjectRectLeft", String.valueOf(visualStructure.getX()));
layoutNode.setAttribute("ObjectRectTop", String.valueOf(visualStructure.getY()));
layoutNode.setAttribute("ObjectRectWidth", String.valueOf(visualStructure.getWidth()));
layoutNode.setAttribute("ObjectRectHeight", String.valueOf(visualStructure.getHeight()));
layoutNode.setAttribute("ID", visualStructure.getId());
layoutNode.setAttribute("order", String.valueOf(_order));
_order++;
if (_pDoC >= visualStructure.getDoC()) {
// continue segmenting
if (visualStructure.getChildrenVisualStructures().size() == 0) {
if (visualStructure.getNestedBlocks().size() > 0) {
String src = "";
String content = "";
for (VipsBlock block : visualStructure.getNestedBlocks()) {
ElementBox elementBox = block.getElementBox();
if (elementBox == null) continue;
if (!elementBox.getNode().getNodeName().equals("Xdiv")
&& !elementBox.getNode().getNodeName().equals("Xspan"))
src += getSource(elementBox.getElement());
else src += elementBox.getText();
content += elementBox.getText() + " ";
}
layoutNode.setAttribute("SRC", src);
layoutNode.setAttribute("Content", content);
}
}
parentNode.appendChild(layoutNode);
for (VisualStructure child : visualStructure.getChildrenVisualStructures())
writeVisualBlocks(layoutNode, child);
} else {
// "stop" segmentation
if (visualStructure.getNestedBlocks().size() > 0) {
String src = "";
String content = "";
for (VipsBlock block : visualStructure.getNestedBlocks()) {
ElementBox elementBox = block.getElementBox();
if (elementBox == null) continue;
if (!elementBox.getNode().getNodeName().equals("Xdiv")
&& !elementBox.getNode().getNodeName().equals("Xspan"))
src += getSource(elementBox.getElement());
else src += elementBox.getText();
content += elementBox.getText() + " ";
}
layoutNode.setAttribute("SRC", src);
layoutNode.setAttribute("Content", content);
}
parentNode.appendChild(layoutNode);
}
}
private void findVisualBlocks(VipsBlock vipsBlock, List<VipsBlock> list) {
if (vipsBlock.isVisualBlock()) list.add(vipsBlock);
for (VipsBlock vipsStructureChild : vipsBlock.getChildren())
findVisualBlocks(vipsStructureChild, list);
}
/**
* VIPS Rule Four
*
* <p>If all of the child nodes of the DOM node are text nodes or virtual text nodes, do not
* divide the node. <br>
* If the font size and font weight of all these child nodes are same, set the DoC of the
* extracted block to 10. Otherwise, set the DoC of this extracted block to 9.
*
* @param node Input node
* @return True, if rule is applied, otherwise false.
*/
private boolean ruleFour(ElementBox node) {
// System.err.println("Applying rule Four on " + node.getNode().getNodeName() + " node");
if (node.getSubBoxList().isEmpty()) return false;
for (Box box : node.getSubBoxList()) {
if (box instanceof TextBox) continue;
if (!isTextNode((ElementBox) box) || !isVirtualTextNode((ElementBox) box)) return false;
}
_currentVipsBlock.setIsVisualBlock(true);
_currentVipsBlock.setIsDividable(false);
if (node.getSubBoxList().size() == 1) {
/*
if (node.getSubBox(0) instanceof TextBox)
{
_currentVipsBlock.setIsVisualBlock(false);
_currentVipsBlock.setIsDividable(true);
_currentVipsBlock.getChildren().get(0).setIsVisualBlock(true);
_currentVipsBlock.getChildren().get(0).setIsDividable(false);
_currentVipsBlock.getChildren().get(0).setDoC(11);
}
*/
if (node.getSubBox(0).getNode().getNodeName().equals("em")) _currentVipsBlock.setDoC(11);
else _currentVipsBlock.setDoC(10);
return true;
}
String fontWeight = "";
int fontSize = 0;
for (Box childNode : node.getSubBoxList()) {
int childFontSize = childNode.getVisualContext().getFont().getSize();
if (childNode instanceof TextBox) {
if (fontSize > 0) {
if (fontSize != childFontSize) {
_currentVipsBlock.setDoC(9);
break;
} else _currentVipsBlock.setDoC(10);
} else fontSize = childFontSize;
continue;
}
ElementBox child = (ElementBox) childNode;
if (child.getStylePropertyValue("font-weight") == null) return false;
if (fontSize > 0) {
if (child.getStylePropertyValue("font-weight").toString().equals(fontWeight)
&& childFontSize == fontSize) {
_currentVipsBlock.setDoC(10);
} else {
_currentVipsBlock.setDoC(9);
break;
}
} else {
fontWeight = child.getStylePropertyValue("font-weight").toString();
fontSize = childFontSize;
}
}
return true;
}
/**
* Tries to divide DOM elements and finds visual blocks.
*
* @param vipsBlock Visual structure
*/
private void divideVipsBlockTree(VipsBlock vipsBlock) {
_currentVipsBlock = vipsBlock;
ElementBox elementBox = (ElementBox) vipsBlock.getBox();
// System.err.println(elementBox.getNode().getNodeName());
// System.out.println(elementBox.getText());
if (elementBox.getElement().getAttribute("id").equals("logosLine")) {
System.out.println();
}
// With VIPS rules it tries to determine if element is dividable
if (applyVipsRules(elementBox) && vipsBlock.isDividable() && !vipsBlock.isVisualBlock()) {
// if element is dividable, let's divide it
_currentVipsBlock.setAlreadyDivided(true);
for (VipsBlock vipsBlockChild : vipsBlock.getChildren()) {
if (!(vipsBlockChild.getBox() instanceof TextBox)) divideVipsBlockTree(vipsBlockChild);
}
} else {
if (vipsBlock.isDividable()) {
// System.err.println("Element " + elementBox.getNode().getNodeName() + " is visual block");
vipsBlock.setIsVisualBlock(true);
vipsBlock.setDoC(11);
}
if (!verifyValidity(elementBox)) {
_currentVipsBlock.setIsVisualBlock(false);
}
/*
if (vipsBlock.isVisualBlock())
//System.err.println("Element " + elementBox.getNode().getNodeName() + " is visual block");
else
System.err.println("Element " + elementBox.getNode().getNodeName() + " is not visual block");*/
}
}
