// Bookmarklet to calculate the most generic xpath for the current selection
// (helper utility for scraping websites)

function selectionxpath() {

	function calculateShortestXpathOfElement( sel ) {
		var node = sel;
		var nextId = null;
		var stop = null;
		var xpath = "";
	
		//find next element with an id
		while (true) {
			if (node.id && node.id != "") {
				nextId = node.id;
				break;
			} 
			node = node.parentNode;
			if (node == stop) break;
		}
	
		if (nextId != null) {
			xpath = "//" + node.tagName.toLowerCase() + "[@id='" + nextId + "']";
			if ( node == sel ) {
				return xpath;
			} else {
				stop = node;
			}
		} 
	
		//find next element with unique tag+class
		node = sel;
		var nextUniqueClass = null;
		while (true) {
			if (node.nodeType === 1) {
				var styleClass = node.getAttribute("class"); 
	
				if (styleClass != null) {
					var tmpXpath = xpath+"//"+node.tagName+"[@class='"+styleClass+"']";
					var tempResult = document.evaluate(tmpXpath, sel.ownerDocument, null, XPathResult.FIRST_ORDERED_NODE_TYPE, null);
					if (tempResult.singleNodeValue == node) {
						nextUniqueClass = styleClass;
						break;
					}
				}
			}
			node = node.parentNode;
			if (node == stop) break;
		}
		if (nextUniqueClass != null) {
			xpath += "//"+node.tagName.toLowerCase()+"[@class='"+nextUniqueClass+"']";
			if ( node == sel ) {
				return xpath;
			} else {
				stop = node;
			}
		}

		//find next element with unique tag
		node = sel;
		var nextUniqueTag = null;
		while (true) {
			if (node.nodeType === 1) {
				var tmpXpath = xpath+"//"+node.tagName;
				var tempResult = document.evaluate(tmpXpath, sel.ownerDocument, null, XPathResult.FIRST_ORDERED_NODE_TYPE, null);
				if (tempResult.singleNodeValue == node) {
					nextUniqueTag = node.tagName;
					break;
				}
			}
			node = node.parentNode;
			if (node == stop) break;
		}
		if (nextUniqueTag != null) {
			xpath += "//"+node.tagName.toLowerCase();
			if ( node == sel ) {
				return xpath;
			} else {
				stop = node;
			}
		}	
	
		//get absolute path for the rest
		var restPath = "";
		for (node = sel; node && node.nodeType == 1; node = node.parentNode) {
			if (node == stop) break;
			var idx = 1;
			for (var sib = node.previousSibling; sib ; sib = sib.previousSibling) {
			    if(sib.nodeType == 1 && sib.tagName == node.tagName) idx++;
			}
			var xname = node.tagName.toLowerCase();
			if (idx > 1) xname += "[" + idx + "]";
			restPath = "/" + xname + restPath;
		}
	
		var result = xpath + restPath;
		return result;
	}

	function depthOf( el ) {
	  i = 0;
	  while (el) {
	    el = el.parentNode;
			i++;
		}
		return i;
	}

	function calculateShortestXpathOfSelection() {
		var sel = window.getSelection().getRangeAt(0);
		if (!sel) return null;
		var start = sel.startContainer;
		var end =  sel.endContainer;

		var i = depthOf( start );
		var j = depthOf( end );
	
		while (start != end && i != 0 && j != 0) {
			if (i > j) {
				start = start.parentNode;
				i--;
			} else {
				end = end.parentNode;
				j--;
			}
		}
	
		return calculateShortestXpathOfElement(start);
	}

	var xpath = calculateShortestXpathOfSelection();
	var node = document.evaluate(xpath, document, null, XPathResult.FIRST_ORDERED_NODE_TYPE, null).singleNodeValue;
	var border = node.style.border;
	if (!border) border = "";
	node.style.border = "2px dashed red";
	if (xpath) {
		prompt("Most generic xpath for selection:", xpath);
		node.style.border = border;		
	} else {
		alert("Could not determine generic xpath for selection");
	}

}

selectionxpath();