Javascript get XPath of a node - javascript

Is there anyway to return an XPath string of a DOM element in Javascript?

I refactored this from another example. It will attempt to check or there is for sure a unique id and if so use that case to shorten the expression.
function createXPathFromElement(elm) {
var allNodes = document.getElementsByTagName('*');
for (var segs = []; elm && elm.nodeType == 1; elm = elm.parentNode)
{
if (elm.hasAttribute('id')) {
var uniqueIdCount = 0;
for (var n=0;n < allNodes.length;n++) {
if (allNodes[n].hasAttribute('id') && allNodes[n].id == elm.id) uniqueIdCount++;
if (uniqueIdCount > 1) break;
};
if ( uniqueIdCount == 1) {
segs.unshift('id("' + elm.getAttribute('id') + '")');
return segs.join('/');
} else {
segs.unshift(elm.localName.toLowerCase() + '[#id="' + elm.getAttribute('id') + '"]');
}
} else if (elm.hasAttribute('class')) {
segs.unshift(elm.localName.toLowerCase() + '[#class="' + elm.getAttribute('class') + '"]');
} else {
for (i = 1, sib = elm.previousSibling; sib; sib = sib.previousSibling) {
if (sib.localName == elm.localName) i++; };
segs.unshift(elm.localName.toLowerCase() + '[' + i + ']');
};
};
return segs.length ? '/' + segs.join('/') : null;
};
function lookupElementByXPath(path) {
var evaluator = new XPathEvaluator();
var result = evaluator.evaluate(path, document.documentElement, null,XPathResult.FIRST_ORDERED_NODE_TYPE, null);
return result.singleNodeValue;
}

There's not a unique XPath to a node, so you'll have to decide what's the most appropriate way of constructing a path. Use IDs where available? Numeral position in the document? Position relative to other elements?
See getPathTo() in this answer for one possible approach.

Here is a functional programming style ES6 function for the job:
function getXPathForElement(element) {
const idx = (sib, name) => sib
? idx(sib.previousElementSibling, name||sib.localName) + (sib.localName == name)
: 1;
const segs = elm => !elm || elm.nodeType !== 1
? ['']
: elm.id && document.getElementById(elm.id) === elm
? [`id("${elm.id}")`]
: [...segs(elm.parentNode), `${elm.localName.toLowerCase()}[${idx(elm)}]`];
return segs(element).join('/');
}
function getElementByXPath(path) {
return (new XPathEvaluator())
.evaluate(path, document.documentElement, null,
XPathResult.FIRST_ORDERED_NODE_TYPE, null)
.singleNodeValue;
}
// Demo:
const li = document.querySelector('li:nth-child(2)');
const path = getXPathForElement(li);
console.log(path);
console.log(li === getElementByXPath(path)); // true
<div>
<table id="start"></table>
<div>
<ul><li>option</ul></ul>
<span>title</span>
<ul>
<li>abc</li>
<li>select this</li>
</ul>
</div>
</div>
It will use an id selector, unless the element is not the first one with that id. Class selectors are not used, because in interactive web pages classes may change often.

I've adapted the algorithm Chromium uses to calculate the XPath from devtools below.
To use this as-written you'd call Elements.DOMPath.xPath(<some DOM node>, false). The last parameter controls whether you get the shorter "Copy XPath" (if true) or "Copy full XPath".
// Copyright 2018 The Chromium Authors. All rights reserved.
// Use of this source code is governed by a BSD-style license that can be
// found in the LICENSE file.
Elements = {};
Elements.DOMPath = {};
/**
* #param {!Node} node
* #param {boolean=} optimized
* #return {string}
*/
Elements.DOMPath.xPath = function (node, optimized) {
if (node.nodeType === Node.DOCUMENT_NODE) {
return '/';
}
const steps = [];
let contextNode = node;
while (contextNode) {
const step = Elements.DOMPath._xPathValue(contextNode, optimized);
if (!step) {
break;
} // Error - bail out early.
steps.push(step);
if (step.optimized) {
break;
}
contextNode = contextNode.parentNode;
}
steps.reverse();
return (steps.length && steps[0].optimized ? '' : '/') + steps.join('/');
};
/**
* #param {!Node} node
* #param {boolean=} optimized
* #return {?Elements.DOMPath.Step}
*/
Elements.DOMPath._xPathValue = function (node, optimized) {
let ownValue;
const ownIndex = Elements.DOMPath._xPathIndex(node);
if (ownIndex === -1) {
return null;
} // Error.
switch (node.nodeType) {
case Node.ELEMENT_NODE:
if (optimized && node.getAttribute('id')) {
return new Elements.DOMPath.Step('//*[#id="' + node.getAttribute('id') + '"]', true);
}
ownValue = node.localName;
break;
case Node.ATTRIBUTE_NODE:
ownValue = '#' + node.nodeName;
break;
case Node.TEXT_NODE:
case Node.CDATA_SECTION_NODE:
ownValue = 'text()';
break;
case Node.PROCESSING_INSTRUCTION_NODE:
ownValue = 'processing-instruction()';
break;
case Node.COMMENT_NODE:
ownValue = 'comment()';
break;
case Node.DOCUMENT_NODE:
ownValue = '';
break;
default:
ownValue = '';
break;
}
if (ownIndex > 0) {
ownValue += '[' + ownIndex + ']';
}
return new Elements.DOMPath.Step(ownValue, node.nodeType === Node.DOCUMENT_NODE);
};
/**
* #param {!Node} node
* #return {number}
*/
Elements.DOMPath._xPathIndex = function (node) {
// Returns -1 in case of error, 0 if no siblings matching the same expression,
// <XPath index among the same expression-matching sibling nodes> otherwise.
function areNodesSimilar(left, right) {
if (left === right) {
return true;
}
if (left.nodeType === Node.ELEMENT_NODE && right.nodeType === Node.ELEMENT_NODE) {
return left.localName === right.localName;
}
if (left.nodeType === right.nodeType) {
return true;
}
// XPath treats CDATA as text nodes.
const leftType = left.nodeType === Node.CDATA_SECTION_NODE ? Node.TEXT_NODE : left.nodeType;
const rightType = right.nodeType === Node.CDATA_SECTION_NODE ? Node.TEXT_NODE : right.nodeType;
return leftType === rightType;
}
const siblings = node.parentNode ? node.parentNode.children : null;
if (!siblings) {
return 0;
} // Root node - no siblings.
let hasSameNamedElements;
for (let i = 0; i < siblings.length; ++i) {
if (areNodesSimilar(node, siblings[i]) && siblings[i] !== node) {
hasSameNamedElements = true;
break;
}
}
if (!hasSameNamedElements) {
return 0;
}
let ownIndex = 1; // XPath indices start with 1.
for (let i = 0; i < siblings.length; ++i) {
if (areNodesSimilar(node, siblings[i])) {
if (siblings[i] === node) {
return ownIndex;
}
++ownIndex;
}
}
return -1; // An error occurred: |node| not found in parent's children.
};
/**
* #unrestricted
*/
Elements.DOMPath.Step = class {
/**
* #param {string} value
* #param {boolean} optimized
*/
constructor(value, optimized) {
this.value = value;
this.optimized = optimized || false;
}
/**
* #override
* #return {string}
*/
toString() {
return this.value;
}
};
Update 2022-08-14: Here is a TypeScript version.

A similar solution is given by the function getXPathForElement on the MDN:
function getXPathForElement(el, xml) {
var xpath = '';
var pos, tempitem2;
while(el !== xml.documentElement) {
pos = 0;
tempitem2 = el;
while(tempitem2) {
if (tempitem2.nodeType === 1 && tempitem2.nodeName === el.nodeName) { // If it is ELEMENT_NODE of the same name
pos += 1;
}
tempitem2 = tempitem2.previousSibling;
}
xpath = "*[name()='"+el.nodeName+"' and namespace-uri()='"+(el.namespaceURI===null?'':el.namespaceURI)+"']["+pos+']'+'/'+xpath;
el = el.parentNode;
}
xpath = '/*'+"[name()='"+xml.documentElement.nodeName+"' and namespace-uri()='"+(el.namespaceURI===null?'':el.namespaceURI)+"']"+'/'+xpath;
xpath = xpath.replace(/\/$/, '');
return xpath;
}
Also XMLSerializer might be worth a try.

function getElementXPath (element) {
if (!element) return null
if (element.id) {
return `//*[#id=${element.id}]`
} else if (element.tagName === 'BODY') {
return '/html/body'
} else {
const sameTagSiblings = Array.from(element.parentNode.childNodes)
.filter(e => e.nodeName === element.nodeName)
const idx = sameTagSiblings.indexOf(element)
return getElementXPath(element.parentNode) +
'/' +
element.tagName.toLowerCase() +
(sameTagSiblings.length > 1 ? `[${idx + 1}]` : '')
}
}
console.log(getElementXPath(document.querySelector('#a div')))
<div id="a">
<div>def</div>
</div>

I checked every solution provided here but none of them works with svg elements (code getElementByXPath(getXPathForElement(elm)) === elm returns false for svg or path elements)
So I added the Touko's svg fix to the trincot's solution and got this code:
function getXPathForElement(element) {
const idx = (sib, name) => sib
? idx(sib.previousElementSibling, name||sib.localName) + (sib.localName == name)
: 1;
const segs = elm => !elm || elm.nodeType !== 1
? ['']
: elm.id && document.getElementById(elm.id) === elm
? [`id("${elm.id}")`]
: [...segs(elm.parentNode), elm instanceof HTMLElement
? `${elm.localName}[${idx(elm)}]`
: `*[local-name() = "${elm.localName}"][${idx(elm)}]`];
return segs(element).join('/');
}
The difference is it returns *[local-name() = "tag"][n] instead of tag[n] if element is not an instance of HTMLElement (svgs are SVGElement but I decided not to stick with checking only svg).
Example:
Before:
.../div[2]/div[2]/span[1]/svg[1]/path[1]
After:
.../div[2]/div[2]/span[1]/*[local-name() = "svg"][1]/*[local-name() = "path"][1]

Just pass the element in function getXPathOfElement and you will get the Xpath.
function getXPathOfElement(elt)
{
var path = "";
for (; elt && elt.nodeType == 1; elt = elt.parentNode)
{
idx = getElementIdx(elt);
xname = elt.tagName;
if (idx > 1) xname += "[" + idx + "]";
path = "/" + xname + path;
}
return path;
}
function getElementIdx(elt)
{
var count = 1;
for (var sib = elt.previousSibling; sib ; sib = sib.previousSibling)
{
if(sib.nodeType == 1 && sib.tagName == elt.tagName) count++
}
return count;
}

Get xPath by giving a dom element
This function returns full xPath selector (without any id or class).
This type of selector is helpful when an site generate random id or class
function getXPath(element) {
// Selector
let selector = '';
// Loop handler
let foundRoot;
// Element handler
let currentElement = element;
// Do action until we reach html element
do {
// Get element tag name
const tagName = currentElement.tagName.toLowerCase();
// Get parent element
const parentElement = currentElement.parentElement;
// Count children
if (parentElement.childElementCount > 1) {
// Get children of parent element
const parentsChildren = [...parentElement.children];
// Count current tag
let tag = [];
parentsChildren.forEach(child => {
if (child.tagName.toLowerCase() === tagName) tag.push(child) // Append to tag
})
// Is only of type
if (tag.length === 1) {
// Append tag to selector
selector = `/${tagName}${selector}`;
} else {
// Get position of current element in tag
const position = tag.indexOf(currentElement) + 1;
// Append tag to selector
selector = `/${tagName}[${position}]${selector}`;
}
} else {
//* Current element has no siblings
// Append tag to selector
selector = `/${tagName}${selector}`;
}
// Set parent element to current element
currentElement = parentElement;
// Is root
foundRoot = parentElement.tagName.toLowerCase() === 'html';
// Finish selector if found root element
if(foundRoot) selector = `/html${selector}`;
}
while (foundRoot === false);
// Return selector
return selector;
}

Related

Is there a way to get xpath of an element using javascript [duplicate]

Is there anyway to return an XPath string of a DOM element in Javascript?
I refactored this from another example. It will attempt to check or there is for sure a unique id and if so use that case to shorten the expression.
function createXPathFromElement(elm) {
var allNodes = document.getElementsByTagName('*');
for (var segs = []; elm && elm.nodeType == 1; elm = elm.parentNode)
{
if (elm.hasAttribute('id')) {
var uniqueIdCount = 0;
for (var n=0;n < allNodes.length;n++) {
if (allNodes[n].hasAttribute('id') && allNodes[n].id == elm.id) uniqueIdCount++;
if (uniqueIdCount > 1) break;
};
if ( uniqueIdCount == 1) {
segs.unshift('id("' + elm.getAttribute('id') + '")');
return segs.join('/');
} else {
segs.unshift(elm.localName.toLowerCase() + '[#id="' + elm.getAttribute('id') + '"]');
}
} else if (elm.hasAttribute('class')) {
segs.unshift(elm.localName.toLowerCase() + '[#class="' + elm.getAttribute('class') + '"]');
} else {
for (i = 1, sib = elm.previousSibling; sib; sib = sib.previousSibling) {
if (sib.localName == elm.localName) i++; };
segs.unshift(elm.localName.toLowerCase() + '[' + i + ']');
};
};
return segs.length ? '/' + segs.join('/') : null;
};
function lookupElementByXPath(path) {
var evaluator = new XPathEvaluator();
var result = evaluator.evaluate(path, document.documentElement, null,XPathResult.FIRST_ORDERED_NODE_TYPE, null);
return result.singleNodeValue;
}
There's not a unique XPath to a node, so you'll have to decide what's the most appropriate way of constructing a path. Use IDs where available? Numeral position in the document? Position relative to other elements?
See getPathTo() in this answer for one possible approach.
Here is a functional programming style ES6 function for the job:
function getXPathForElement(element) {
const idx = (sib, name) => sib
? idx(sib.previousElementSibling, name||sib.localName) + (sib.localName == name)
: 1;
const segs = elm => !elm || elm.nodeType !== 1
? ['']
: elm.id && document.getElementById(elm.id) === elm
? [`id("${elm.id}")`]
: [...segs(elm.parentNode), `${elm.localName.toLowerCase()}[${idx(elm)}]`];
return segs(element).join('/');
}
function getElementByXPath(path) {
return (new XPathEvaluator())
.evaluate(path, document.documentElement, null,
XPathResult.FIRST_ORDERED_NODE_TYPE, null)
.singleNodeValue;
}
// Demo:
const li = document.querySelector('li:nth-child(2)');
const path = getXPathForElement(li);
console.log(path);
console.log(li === getElementByXPath(path)); // true
<div>
<table id="start"></table>
<div>
<ul><li>option</ul></ul>
<span>title</span>
<ul>
<li>abc</li>
<li>select this</li>
</ul>
</div>
</div>
It will use an id selector, unless the element is not the first one with that id. Class selectors are not used, because in interactive web pages classes may change often.
I've adapted the algorithm Chromium uses to calculate the XPath from devtools below.
To use this as-written you'd call Elements.DOMPath.xPath(<some DOM node>, false). The last parameter controls whether you get the shorter "Copy XPath" (if true) or "Copy full XPath".
// Copyright 2018 The Chromium Authors. All rights reserved.
// Use of this source code is governed by a BSD-style license that can be
// found in the LICENSE file.
Elements = {};
Elements.DOMPath = {};
/**
* #param {!Node} node
* #param {boolean=} optimized
* #return {string}
*/
Elements.DOMPath.xPath = function (node, optimized) {
if (node.nodeType === Node.DOCUMENT_NODE) {
return '/';
}
const steps = [];
let contextNode = node;
while (contextNode) {
const step = Elements.DOMPath._xPathValue(contextNode, optimized);
if (!step) {
break;
} // Error - bail out early.
steps.push(step);
if (step.optimized) {
break;
}
contextNode = contextNode.parentNode;
}
steps.reverse();
return (steps.length && steps[0].optimized ? '' : '/') + steps.join('/');
};
/**
* #param {!Node} node
* #param {boolean=} optimized
* #return {?Elements.DOMPath.Step}
*/
Elements.DOMPath._xPathValue = function (node, optimized) {
let ownValue;
const ownIndex = Elements.DOMPath._xPathIndex(node);
if (ownIndex === -1) {
return null;
} // Error.
switch (node.nodeType) {
case Node.ELEMENT_NODE:
if (optimized && node.getAttribute('id')) {
return new Elements.DOMPath.Step('//*[#id="' + node.getAttribute('id') + '"]', true);
}
ownValue = node.localName;
break;
case Node.ATTRIBUTE_NODE:
ownValue = '#' + node.nodeName;
break;
case Node.TEXT_NODE:
case Node.CDATA_SECTION_NODE:
ownValue = 'text()';
break;
case Node.PROCESSING_INSTRUCTION_NODE:
ownValue = 'processing-instruction()';
break;
case Node.COMMENT_NODE:
ownValue = 'comment()';
break;
case Node.DOCUMENT_NODE:
ownValue = '';
break;
default:
ownValue = '';
break;
}
if (ownIndex > 0) {
ownValue += '[' + ownIndex + ']';
}
return new Elements.DOMPath.Step(ownValue, node.nodeType === Node.DOCUMENT_NODE);
};
/**
* #param {!Node} node
* #return {number}
*/
Elements.DOMPath._xPathIndex = function (node) {
// Returns -1 in case of error, 0 if no siblings matching the same expression,
// <XPath index among the same expression-matching sibling nodes> otherwise.
function areNodesSimilar(left, right) {
if (left === right) {
return true;
}
if (left.nodeType === Node.ELEMENT_NODE && right.nodeType === Node.ELEMENT_NODE) {
return left.localName === right.localName;
}
if (left.nodeType === right.nodeType) {
return true;
}
// XPath treats CDATA as text nodes.
const leftType = left.nodeType === Node.CDATA_SECTION_NODE ? Node.TEXT_NODE : left.nodeType;
const rightType = right.nodeType === Node.CDATA_SECTION_NODE ? Node.TEXT_NODE : right.nodeType;
return leftType === rightType;
}
const siblings = node.parentNode ? node.parentNode.children : null;
if (!siblings) {
return 0;
} // Root node - no siblings.
let hasSameNamedElements;
for (let i = 0; i < siblings.length; ++i) {
if (areNodesSimilar(node, siblings[i]) && siblings[i] !== node) {
hasSameNamedElements = true;
break;
}
}
if (!hasSameNamedElements) {
return 0;
}
let ownIndex = 1; // XPath indices start with 1.
for (let i = 0; i < siblings.length; ++i) {
if (areNodesSimilar(node, siblings[i])) {
if (siblings[i] === node) {
return ownIndex;
}
++ownIndex;
}
}
return -1; // An error occurred: |node| not found in parent's children.
};
/**
* #unrestricted
*/
Elements.DOMPath.Step = class {
/**
* #param {string} value
* #param {boolean} optimized
*/
constructor(value, optimized) {
this.value = value;
this.optimized = optimized || false;
}
/**
* #override
* #return {string}
*/
toString() {
return this.value;
}
};
Update 2022-08-14: Here is a TypeScript version.
A similar solution is given by the function getXPathForElement on the MDN:
function getXPathForElement(el, xml) {
var xpath = '';
var pos, tempitem2;
while(el !== xml.documentElement) {
pos = 0;
tempitem2 = el;
while(tempitem2) {
if (tempitem2.nodeType === 1 && tempitem2.nodeName === el.nodeName) { // If it is ELEMENT_NODE of the same name
pos += 1;
}
tempitem2 = tempitem2.previousSibling;
}
xpath = "*[name()='"+el.nodeName+"' and namespace-uri()='"+(el.namespaceURI===null?'':el.namespaceURI)+"']["+pos+']'+'/'+xpath;
el = el.parentNode;
}
xpath = '/*'+"[name()='"+xml.documentElement.nodeName+"' and namespace-uri()='"+(el.namespaceURI===null?'':el.namespaceURI)+"']"+'/'+xpath;
xpath = xpath.replace(/\/$/, '');
return xpath;
}
Also XMLSerializer might be worth a try.
function getElementXPath (element) {
if (!element) return null
if (element.id) {
return `//*[#id=${element.id}]`
} else if (element.tagName === 'BODY') {
return '/html/body'
} else {
const sameTagSiblings = Array.from(element.parentNode.childNodes)
.filter(e => e.nodeName === element.nodeName)
const idx = sameTagSiblings.indexOf(element)
return getElementXPath(element.parentNode) +
'/' +
element.tagName.toLowerCase() +
(sameTagSiblings.length > 1 ? `[${idx + 1}]` : '')
}
}
console.log(getElementXPath(document.querySelector('#a div')))
<div id="a">
<div>def</div>
</div>
I checked every solution provided here but none of them works with svg elements (code getElementByXPath(getXPathForElement(elm)) === elm returns false for svg or path elements)
So I added the Touko's svg fix to the trincot's solution and got this code:
function getXPathForElement(element) {
const idx = (sib, name) => sib
? idx(sib.previousElementSibling, name||sib.localName) + (sib.localName == name)
: 1;
const segs = elm => !elm || elm.nodeType !== 1
? ['']
: elm.id && document.getElementById(elm.id) === elm
? [`id("${elm.id}")`]
: [...segs(elm.parentNode), elm instanceof HTMLElement
? `${elm.localName}[${idx(elm)}]`
: `*[local-name() = "${elm.localName}"][${idx(elm)}]`];
return segs(element).join('/');
}
The difference is it returns *[local-name() = "tag"][n] instead of tag[n] if element is not an instance of HTMLElement (svgs are SVGElement but I decided not to stick with checking only svg).
Example:
Before:
.../div[2]/div[2]/span[1]/svg[1]/path[1]
After:
.../div[2]/div[2]/span[1]/*[local-name() = "svg"][1]/*[local-name() = "path"][1]
Just pass the element in function getXPathOfElement and you will get the Xpath.
function getXPathOfElement(elt)
{
var path = "";
for (; elt && elt.nodeType == 1; elt = elt.parentNode)
{
idx = getElementIdx(elt);
xname = elt.tagName;
if (idx > 1) xname += "[" + idx + "]";
path = "/" + xname + path;
}
return path;
}
function getElementIdx(elt)
{
var count = 1;
for (var sib = elt.previousSibling; sib ; sib = sib.previousSibling)
{
if(sib.nodeType == 1 && sib.tagName == elt.tagName) count++
}
return count;
}
Get xPath by giving a dom element
This function returns full xPath selector (without any id or class).
This type of selector is helpful when an site generate random id or class
function getXPath(element) {
// Selector
let selector = '';
// Loop handler
let foundRoot;
// Element handler
let currentElement = element;
// Do action until we reach html element
do {
// Get element tag name
const tagName = currentElement.tagName.toLowerCase();
// Get parent element
const parentElement = currentElement.parentElement;
// Count children
if (parentElement.childElementCount > 1) {
// Get children of parent element
const parentsChildren = [...parentElement.children];
// Count current tag
let tag = [];
parentsChildren.forEach(child => {
if (child.tagName.toLowerCase() === tagName) tag.push(child) // Append to tag
})
// Is only of type
if (tag.length === 1) {
// Append tag to selector
selector = `/${tagName}${selector}`;
} else {
// Get position of current element in tag
const position = tag.indexOf(currentElement) + 1;
// Append tag to selector
selector = `/${tagName}[${position}]${selector}`;
}
} else {
//* Current element has no siblings
// Append tag to selector
selector = `/${tagName}${selector}`;
}
// Set parent element to current element
currentElement = parentElement;
// Is root
foundRoot = parentElement.tagName.toLowerCase() === 'html';
// Finish selector if found root element
if(foundRoot) selector = `/html${selector}`;
}
while (foundRoot === false);
// Return selector
return selector;
}

Can I perform "Copy Selector" with javascript without using Chrome Dev tools [duplicate]

I am moving elements using javascript and I need to create a logic for the combinations happening during the drag/drops
I'm trying to get details from the elements, a CSS like selector could be also good, but dunno if it is possible.. (like copy-selector in chrome dev tools)
document.onmouseup = function(e){
targetDest = e.target;
//console.log('targetDest: ', targetDest);
let
indexA = Array.from(targetCurr.parentNode.children).indexOf(targetCurr),
indexB = Array.from(targetDest.parentNode.children).indexOf(targetDest);
console.log(indexA, indexB);
if(targetDest != targetCurr){
if(targetDest == document.documentElement){
console.log('document');
}
else if(targetDest == undefined){
console.log('undefined');
}
else if(!targetDest){
console.log('!dest');
}
else if(targetDest == null){
console.log('null');
}
else if(targetDest == false){
console.log('false');
}
else{
console.log('else');
//targetCurr.parentNode.insertBefore(targetDest, targetCurr);
//console.log('...');
}
}else{
console.log('itself');
}
}
Keep in mind that this will not necessarily uniquely identify elements. But, you can construct that type of selector by traversing upwards from the node and prepending the element you're at. You could potentially do something like this
var generateQuerySelector = function(el) {
if (el.tagName.toLowerCase() == "html")
return "HTML";
var str = el.tagName;
str += (el.id != "") ? "#" + el.id : "";
if (el.className) {
var classes = el.className.split(/\s/);
for (var i = 0; i < classes.length; i++) {
str += "." + classes[i]
}
}
return generateQuerySelector(el.parentNode) + " > " + str;
}
var qStr = generateQuerySelector(document.querySelector("div.moo"));
alert(qStr);
body
<div class="outer">
div.outer
<div class="inner" id="foo">
div#foo.inner
<div class="moo man">
div.moo.man
</div>
</div>
</div>
I wouldn't suggest using this for much besides presenting the information to a user. Splitting it up and reusing parts are bound to cause problems.
My solution using :nth-child:
function getSelector(elm)
{
if (elm.tagName === "BODY") return "BODY";
const names = [];
while (elm.parentElement && elm.tagName !== "BODY") {
if (elm.id) {
names.unshift("#" + elm.getAttribute("id")); // getAttribute, because `elm.id` could also return a child element with name "id"
break; // Because ID should be unique, no more is needed. Remove the break, if you always want a full path.
} else {
let c = 1, e = elm;
for (; e.previousElementSibling; e = e.previousElementSibling, c++) ;
names.unshift(elm.tagName + ":nth-child(" + c + ")");
}
elm = elm.parentElement;
}
return names.join(">");
}
var qStr = getSelector(document.querySelector("div.moo"));
alert(qStr);
body
<div class="outer">
div.outer
<div class="inner" id="foo">
div#foo.inner
<div class="moo man">
div.moo.man
</div>
</div>
</div>
Please note it won't return the whole path if there's an element with ID in it - every ID should be unique on the page, as valid HTML requires.
I use output of this function in document.querySelector later in the code, because I needed to return focus to the same element after replaceChild of its parent element.
I hope CollinD won't mind I borrowed his markup for the code snippet :-)
I mixed the 2 solutions proposed to have a result readable by humans and which gives the right element if there are several similar siblings:
function elemToSelector(elem) {
const {
tagName,
id,
className,
parentNode
} = elem;
if (tagName === 'HTML') return 'HTML';
let str = tagName;
str += (id !== '') ? `#${id}` : '';
if (className) {
const classes = className.split(/\s/);
for (let i = 0; i < classes.length; i++) {
str += `.${classes[i]}`;
}
}
let childIndex = 1;
for (let e = elem; e.previousElementSibling; e = e.previousElementSibling) {
childIndex += 1;
}
str += `:nth-child(${childIndex})`;
return `${elemToSelector(parentNode)} > ${str}`;
}
Test with:
// Select an element in Elements tab of your navigator Devtools, or replace $0
document.querySelector(elemToSelector($0)) === $0 &&
document.querySelectorAll(elemToSelector($0)).length === 1
Which might give you something like, it's a bit longer but it's readable and it always works:
HTML > BODY:nth-child(2) > DIV.container:nth-child(2) > DIV.row:nth-child(2) > DIV.col-md-4:nth-child(2) > DIV.sidebar:nth-child(1) > DIV.sidebar-wrapper:nth-child(2) > DIV.my-4:nth-child(1) > H4:nth-child(3)
Edit: I just found the package unique-selector
Small improvement of the #CollinD answer :
1/ Return value when the selector is unique
2/ Trim classes value (classes with end blanks make errors)
3/ Split multiple spaces between classes
var getSelector = function(el) {
if (el.tagName.toLowerCase() == "html")
return "html";
var str = el.tagName.toLowerCase();
str += (el.id != "") ? "#" + el.id : "";
if (el.className) {
var classes = el.className.trim().split(/\s+/);
for (var i = 0; i < classes.length; i++) {
str += "." + classes[i]
}
}
if(document.querySelectorAll(str).length==1) return str;
return getSelector(el.parentNode) + " > " + str;
}
Based on previous solutions, I made a typescript solution with a shorter selector and additional checks.
function elemToSelector(elem: HTMLElement): string {
const {
tagName,
id,
className,
parentElement
} = elem;
let str = '';
if (id !== '' && id.match(/^[a-z].*/)) {
str += `#${id}`;
return str;
}
str = tagName;
if (className) {
str += '.' + className.replace(/(^\s)/gm, '').replace(/(\s{2,})/gm, ' ')
.split(/\s/).join('.');
}
const needNthPart = (el: HTMLElement): boolean => {
let sib = el.previousElementSibling;
if (!el.className) {
return true;
}
while (sib) {
if (el.className !== sib.className) {
return false;
}
sib = sib.previousElementSibling;
}
return false;
}
const getNthPart = (el: HTMLElement): string => {
let childIndex = 1;
let sib = el.previousElementSibling;
while (sib) {
childIndex++;
sib = sib.previousElementSibling;
}
return `:nth-child(${childIndex})`;
}
if (needNthPart(elem)) {
str += getNthPart(elem);
}
if (!parentElement) {
return str;
}
return `${elemToSelector(parentElement)} > ${str}`;
}

Google Chrome: Is it possible to copy XPath dynamically (in code)? [duplicate]

Is there anyway to return an XPath string of a DOM element in Javascript?
I refactored this from another example. It will attempt to check or there is for sure a unique id and if so use that case to shorten the expression.
function createXPathFromElement(elm) {
var allNodes = document.getElementsByTagName('*');
for (var segs = []; elm && elm.nodeType == 1; elm = elm.parentNode)
{
if (elm.hasAttribute('id')) {
var uniqueIdCount = 0;
for (var n=0;n < allNodes.length;n++) {
if (allNodes[n].hasAttribute('id') && allNodes[n].id == elm.id) uniqueIdCount++;
if (uniqueIdCount > 1) break;
};
if ( uniqueIdCount == 1) {
segs.unshift('id("' + elm.getAttribute('id') + '")');
return segs.join('/');
} else {
segs.unshift(elm.localName.toLowerCase() + '[#id="' + elm.getAttribute('id') + '"]');
}
} else if (elm.hasAttribute('class')) {
segs.unshift(elm.localName.toLowerCase() + '[#class="' + elm.getAttribute('class') + '"]');
} else {
for (i = 1, sib = elm.previousSibling; sib; sib = sib.previousSibling) {
if (sib.localName == elm.localName) i++; };
segs.unshift(elm.localName.toLowerCase() + '[' + i + ']');
};
};
return segs.length ? '/' + segs.join('/') : null;
};
function lookupElementByXPath(path) {
var evaluator = new XPathEvaluator();
var result = evaluator.evaluate(path, document.documentElement, null,XPathResult.FIRST_ORDERED_NODE_TYPE, null);
return result.singleNodeValue;
}
There's not a unique XPath to a node, so you'll have to decide what's the most appropriate way of constructing a path. Use IDs where available? Numeral position in the document? Position relative to other elements?
See getPathTo() in this answer for one possible approach.
Here is a functional programming style ES6 function for the job:
function getXPathForElement(element) {
const idx = (sib, name) => sib
? idx(sib.previousElementSibling, name||sib.localName) + (sib.localName == name)
: 1;
const segs = elm => !elm || elm.nodeType !== 1
? ['']
: elm.id && document.getElementById(elm.id) === elm
? [`id("${elm.id}")`]
: [...segs(elm.parentNode), `${elm.localName.toLowerCase()}[${idx(elm)}]`];
return segs(element).join('/');
}
function getElementByXPath(path) {
return (new XPathEvaluator())
.evaluate(path, document.documentElement, null,
XPathResult.FIRST_ORDERED_NODE_TYPE, null)
.singleNodeValue;
}
// Demo:
const li = document.querySelector('li:nth-child(2)');
const path = getXPathForElement(li);
console.log(path);
console.log(li === getElementByXPath(path)); // true
<div>
<table id="start"></table>
<div>
<ul><li>option</ul></ul>
<span>title</span>
<ul>
<li>abc</li>
<li>select this</li>
</ul>
</div>
</div>
It will use an id selector, unless the element is not the first one with that id. Class selectors are not used, because in interactive web pages classes may change often.
I've adapted the algorithm Chromium uses to calculate the XPath from devtools below.
To use this as-written you'd call Elements.DOMPath.xPath(<some DOM node>, false). The last parameter controls whether you get the shorter "Copy XPath" (if true) or "Copy full XPath".
// Copyright 2018 The Chromium Authors. All rights reserved.
// Use of this source code is governed by a BSD-style license that can be
// found in the LICENSE file.
Elements = {};
Elements.DOMPath = {};
/**
* #param {!Node} node
* #param {boolean=} optimized
* #return {string}
*/
Elements.DOMPath.xPath = function (node, optimized) {
if (node.nodeType === Node.DOCUMENT_NODE) {
return '/';
}
const steps = [];
let contextNode = node;
while (contextNode) {
const step = Elements.DOMPath._xPathValue(contextNode, optimized);
if (!step) {
break;
} // Error - bail out early.
steps.push(step);
if (step.optimized) {
break;
}
contextNode = contextNode.parentNode;
}
steps.reverse();
return (steps.length && steps[0].optimized ? '' : '/') + steps.join('/');
};
/**
* #param {!Node} node
* #param {boolean=} optimized
* #return {?Elements.DOMPath.Step}
*/
Elements.DOMPath._xPathValue = function (node, optimized) {
let ownValue;
const ownIndex = Elements.DOMPath._xPathIndex(node);
if (ownIndex === -1) {
return null;
} // Error.
switch (node.nodeType) {
case Node.ELEMENT_NODE:
if (optimized && node.getAttribute('id')) {
return new Elements.DOMPath.Step('//*[#id="' + node.getAttribute('id') + '"]', true);
}
ownValue = node.localName;
break;
case Node.ATTRIBUTE_NODE:
ownValue = '#' + node.nodeName;
break;
case Node.TEXT_NODE:
case Node.CDATA_SECTION_NODE:
ownValue = 'text()';
break;
case Node.PROCESSING_INSTRUCTION_NODE:
ownValue = 'processing-instruction()';
break;
case Node.COMMENT_NODE:
ownValue = 'comment()';
break;
case Node.DOCUMENT_NODE:
ownValue = '';
break;
default:
ownValue = '';
break;
}
if (ownIndex > 0) {
ownValue += '[' + ownIndex + ']';
}
return new Elements.DOMPath.Step(ownValue, node.nodeType === Node.DOCUMENT_NODE);
};
/**
* #param {!Node} node
* #return {number}
*/
Elements.DOMPath._xPathIndex = function (node) {
// Returns -1 in case of error, 0 if no siblings matching the same expression,
// <XPath index among the same expression-matching sibling nodes> otherwise.
function areNodesSimilar(left, right) {
if (left === right) {
return true;
}
if (left.nodeType === Node.ELEMENT_NODE && right.nodeType === Node.ELEMENT_NODE) {
return left.localName === right.localName;
}
if (left.nodeType === right.nodeType) {
return true;
}
// XPath treats CDATA as text nodes.
const leftType = left.nodeType === Node.CDATA_SECTION_NODE ? Node.TEXT_NODE : left.nodeType;
const rightType = right.nodeType === Node.CDATA_SECTION_NODE ? Node.TEXT_NODE : right.nodeType;
return leftType === rightType;
}
const siblings = node.parentNode ? node.parentNode.children : null;
if (!siblings) {
return 0;
} // Root node - no siblings.
let hasSameNamedElements;
for (let i = 0; i < siblings.length; ++i) {
if (areNodesSimilar(node, siblings[i]) && siblings[i] !== node) {
hasSameNamedElements = true;
break;
}
}
if (!hasSameNamedElements) {
return 0;
}
let ownIndex = 1; // XPath indices start with 1.
for (let i = 0; i < siblings.length; ++i) {
if (areNodesSimilar(node, siblings[i])) {
if (siblings[i] === node) {
return ownIndex;
}
++ownIndex;
}
}
return -1; // An error occurred: |node| not found in parent's children.
};
/**
* #unrestricted
*/
Elements.DOMPath.Step = class {
/**
* #param {string} value
* #param {boolean} optimized
*/
constructor(value, optimized) {
this.value = value;
this.optimized = optimized || false;
}
/**
* #override
* #return {string}
*/
toString() {
return this.value;
}
};
Update 2022-08-14: Here is a TypeScript version.
A similar solution is given by the function getXPathForElement on the MDN:
function getXPathForElement(el, xml) {
var xpath = '';
var pos, tempitem2;
while(el !== xml.documentElement) {
pos = 0;
tempitem2 = el;
while(tempitem2) {
if (tempitem2.nodeType === 1 && tempitem2.nodeName === el.nodeName) { // If it is ELEMENT_NODE of the same name
pos += 1;
}
tempitem2 = tempitem2.previousSibling;
}
xpath = "*[name()='"+el.nodeName+"' and namespace-uri()='"+(el.namespaceURI===null?'':el.namespaceURI)+"']["+pos+']'+'/'+xpath;
el = el.parentNode;
}
xpath = '/*'+"[name()='"+xml.documentElement.nodeName+"' and namespace-uri()='"+(el.namespaceURI===null?'':el.namespaceURI)+"']"+'/'+xpath;
xpath = xpath.replace(/\/$/, '');
return xpath;
}
Also XMLSerializer might be worth a try.
function getElementXPath (element) {
if (!element) return null
if (element.id) {
return `//*[#id=${element.id}]`
} else if (element.tagName === 'BODY') {
return '/html/body'
} else {
const sameTagSiblings = Array.from(element.parentNode.childNodes)
.filter(e => e.nodeName === element.nodeName)
const idx = sameTagSiblings.indexOf(element)
return getElementXPath(element.parentNode) +
'/' +
element.tagName.toLowerCase() +
(sameTagSiblings.length > 1 ? `[${idx + 1}]` : '')
}
}
console.log(getElementXPath(document.querySelector('#a div')))
<div id="a">
<div>def</div>
</div>
I checked every solution provided here but none of them works with svg elements (code getElementByXPath(getXPathForElement(elm)) === elm returns false for svg or path elements)
So I added the Touko's svg fix to the trincot's solution and got this code:
function getXPathForElement(element) {
const idx = (sib, name) => sib
? idx(sib.previousElementSibling, name||sib.localName) + (sib.localName == name)
: 1;
const segs = elm => !elm || elm.nodeType !== 1
? ['']
: elm.id && document.getElementById(elm.id) === elm
? [`id("${elm.id}")`]
: [...segs(elm.parentNode), elm instanceof HTMLElement
? `${elm.localName}[${idx(elm)}]`
: `*[local-name() = "${elm.localName}"][${idx(elm)}]`];
return segs(element).join('/');
}
The difference is it returns *[local-name() = "tag"][n] instead of tag[n] if element is not an instance of HTMLElement (svgs are SVGElement but I decided not to stick with checking only svg).
Example:
Before:
.../div[2]/div[2]/span[1]/svg[1]/path[1]
After:
.../div[2]/div[2]/span[1]/*[local-name() = "svg"][1]/*[local-name() = "path"][1]
Just pass the element in function getXPathOfElement and you will get the Xpath.
function getXPathOfElement(elt)
{
var path = "";
for (; elt && elt.nodeType == 1; elt = elt.parentNode)
{
idx = getElementIdx(elt);
xname = elt.tagName;
if (idx > 1) xname += "[" + idx + "]";
path = "/" + xname + path;
}
return path;
}
function getElementIdx(elt)
{
var count = 1;
for (var sib = elt.previousSibling; sib ; sib = sib.previousSibling)
{
if(sib.nodeType == 1 && sib.tagName == elt.tagName) count++
}
return count;
}
Get xPath by giving a dom element
This function returns full xPath selector (without any id or class).
This type of selector is helpful when an site generate random id or class
function getXPath(element) {
// Selector
let selector = '';
// Loop handler
let foundRoot;
// Element handler
let currentElement = element;
// Do action until we reach html element
do {
// Get element tag name
const tagName = currentElement.tagName.toLowerCase();
// Get parent element
const parentElement = currentElement.parentElement;
// Count children
if (parentElement.childElementCount > 1) {
// Get children of parent element
const parentsChildren = [...parentElement.children];
// Count current tag
let tag = [];
parentsChildren.forEach(child => {
if (child.tagName.toLowerCase() === tagName) tag.push(child) // Append to tag
})
// Is only of type
if (tag.length === 1) {
// Append tag to selector
selector = `/${tagName}${selector}`;
} else {
// Get position of current element in tag
const position = tag.indexOf(currentElement) + 1;
// Append tag to selector
selector = `/${tagName}[${position}]${selector}`;
}
} else {
//* Current element has no siblings
// Append tag to selector
selector = `/${tagName}${selector}`;
}
// Set parent element to current element
currentElement = parentElement;
// Is root
foundRoot = parentElement.tagName.toLowerCase() === 'html';
// Finish selector if found root element
if(foundRoot) selector = `/html${selector}`;
}
while (foundRoot === false);
// Return selector
return selector;
}

Convert html string to selector syntax [duplicate]

I got this function to get a cssPath :
var cssPath = function (el) {
var path = [];
while (
(el.nodeName.toLowerCase() != 'html') &&
(el = el.parentNode) &&
path.unshift(el.nodeName.toLowerCase() +
(el.id ? '#' + el.id : '') +
(el.className ? '.' + el.className.replace(/\s+/g, ".") : ''))
);
return path.join(" > ");
}
console.log(cssPath(document.getElementsByTagName('a')[123]));
But i got something like this :
html > body > div#div-id > div.site > div.clearfix > ul.choices > li
But to be totally right, it should look like this :html > body > div#div-id > div.site:nth-child(1) > div.clearfix > ul.choices > li:nth-child(5)
Did someone have any idea to implement it simply in javascript ?
The answer above actually has a bug in it — the while loop breaks prematurely when it encounters a non-element node (e.g. a text node) resulting in an incorrect CSS selector.
Here's an improved version that fixes that problem plus:
Stops when it encounters the first ancestor element with an id assigned to it
Uses nth-of-type() to make the selectors more readable
var cssPath = function(el) {
if (!(el instanceof Element))
return;
var path = [];
while (el.nodeType === Node.ELEMENT_NODE) {
var selector = el.nodeName.toLowerCase();
if (el.id) {
selector += '#' + el.id;
path.unshift(selector);
break;
} else {
var sib = el, nth = 1;
while (sib = sib.previousElementSibling) {
if (sib.nodeName.toLowerCase() == selector)
nth++;
}
if (nth != 1)
selector += ":nth-of-type("+nth+")";
}
path.unshift(selector);
el = el.parentNode;
}
return path.join(" > ");
}
To always get the right element, you will need to use :nth-child() or :nth-of-type() for selectors that do not uniquely identify an element. So try this:
var cssPath = function(el) {
if (!(el instanceof Element)) return;
var path = [];
while (el.nodeType === Node.ELEMENT_NODE) {
var selector = el.nodeName.toLowerCase();
if (el.id) {
selector += '#' + el.id;
} else {
var sib = el, nth = 1;
while (sib.nodeType === Node.ELEMENT_NODE && (sib = sib.previousSibling) && nth++);
selector += ":nth-child("+nth+")";
}
path.unshift(selector);
el = el.parentNode;
}
return path.join(" > ");
}
You could add a routine to check for unique elements in their corresponding context (like TITLE, BASE, CAPTION, etc.).
The two other provided answers had a couple of assumptions with browser compatibility that I ran into. Below code will not use nth-child and also has the previousElementSibling check.
function previousElementSibling (element) {
if (element.previousElementSibling !== 'undefined') {
return element.previousElementSibling;
} else {
// Loop through ignoring anything not an element
while (element = element.previousSibling) {
if (element.nodeType === 1) {
return element;
}
}
}
}
function getPath (element) {
// False on non-elements
if (!(element instanceof HTMLElement)) { return false; }
var path = [];
while (element.nodeType === Node.ELEMENT_NODE) {
var selector = element.nodeName;
if (element.id) { selector += ('#' + element.id); }
else {
// Walk backwards until there is no previous sibling
var sibling = element;
// Will hold nodeName to join for adjacent selection
var siblingSelectors = [];
while (sibling !== null && sibling.nodeType === Node.ELEMENT_NODE) {
siblingSelectors.unshift(sibling.nodeName);
sibling = previousElementSibling(sibling);
}
// :first-child does not apply to HTML
if (siblingSelectors[0] !== 'HTML') {
siblingSelectors[0] = siblingSelectors[0] + ':first-child';
}
selector = siblingSelectors.join(' + ');
}
path.unshift(selector);
element = element.parentNode;
}
return path.join(' > ');
}
Doing a reverse CSS selector lookup is an inherently tricky thing. I've generally come across two types of solutions:
Go up the DOM tree to assemble the selector string out of a combination of element names, classes, and the id or name attribute. The problem with this method is that it can result in selectors that return multiple elements, which won't cut it if we require them to select only one unique element.
Assemble the selector string using nth-child() or nth-of-type(), which can result in very long selectors. In most cases the longer a selector is the higher specificity it has, and the higher the specificity the more likely it will break when the DOM structure changes.
The solution below is an attempt at tackling both of these issues. It is a hybrid approach that outputs a unique CSS selector (i.e., document.querySelectorAll(getUniqueSelector(el)) should always return a one-item array). While the returned selector string is not necessarily the shortest, it is derived with an eye towards CSS selector efficiency while balancing specificity by prioritizing nth-of-type() and nth-child() last.
You can specify what attributes to incorporate into the selector by updating the aAttr array. The minimum browser requirement is IE 9.
function getUniqueSelector(elSrc) {
if (!(elSrc instanceof Element)) return;
var sSel,
aAttr = ['name', 'value', 'title', 'placeholder', 'data-*'], // Common attributes
aSel = [],
// Derive selector from element
getSelector = function(el) {
// 1. Check ID first
// NOTE: ID must be unique amongst all IDs in an HTML5 document.
// https://www.w3.org/TR/html5/dom.html#the-id-attribute
if (el.id) {
aSel.unshift('#' + el.id);
return true;
}
aSel.unshift(sSel = el.nodeName.toLowerCase());
// 2. Try to select by classes
if (el.className) {
aSel[0] = sSel += '.' + el.className.trim().replace(/ +/g, '.');
if (uniqueQuery()) return true;
}
// 3. Try to select by classes + attributes
for (var i=0; i<aAttr.length; ++i) {
if (aAttr[i]==='data-*') {
// Build array of data attributes
var aDataAttr = [].filter.call(el.attributes, function(attr) {
return attr.name.indexOf('data-')===0;
});
for (var j=0; j<aDataAttr.length; ++j) {
aSel[0] = sSel += '[' + aDataAttr[j].name + '="' + aDataAttr[j].value + '"]';
if (uniqueQuery()) return true;
}
} else if (el[aAttr[i]]) {
aSel[0] = sSel += '[' + aAttr[i] + '="' + el[aAttr[i]] + '"]';
if (uniqueQuery()) return true;
}
}
// 4. Try to select by nth-of-type() as a fallback for generic elements
var elChild = el,
sChild,
n = 1;
while (elChild = elChild.previousElementSibling) {
if (elChild.nodeName===el.nodeName) ++n;
}
aSel[0] = sSel += ':nth-of-type(' + n + ')';
if (uniqueQuery()) return true;
// 5. Try to select by nth-child() as a last resort
elChild = el;
n = 1;
while (elChild = elChild.previousElementSibling) ++n;
aSel[0] = sSel = sSel.replace(/:nth-of-type\(\d+\)/, n>1 ? ':nth-child(' + n + ')' : ':first-child');
if (uniqueQuery()) return true;
return false;
},
// Test query to see if it returns one element
uniqueQuery = function() {
return document.querySelectorAll(aSel.join('>')||null).length===1;
};
// Walk up the DOM tree to compile a unique selector
while (elSrc.parentNode) {
if (getSelector(elSrc)) return aSel.join(' > ');
elSrc = elSrc.parentNode;
}
}
I somehow find all the implementations unreadable due to unnecessary mutation. Here I provide mine in ClojureScript and JS:
(defn element? [x]
(and (not (nil? x))
(identical? (.-nodeType x) js/Node.ELEMENT_NODE)))
(defn nth-child [el]
(loop [sib el nth 1]
(if sib
(recur (.-previousSibling sib) (inc nth))
(dec nth))))
(defn element-path
([el] (element-path el []))
([el path]
(if (element? el)
(let [tag (.. el -nodeName (toLowerCase))
id (and (not (string/blank? (.-id el))) (.-id el))]
(if id
(element-path nil (conj path (str "#" id)))
(element-path
(.-parentNode el)
(conj path (str tag ":nth-child(" (nth-child el) ")")))))
(string/join " > " (reverse path)))))
Javascript:
const isElement = (x) => x && x.nodeType === Node.ELEMENT_NODE;
const nthChild = (el, nth = 1) => {
if (el) {
return nthChild(el.previousSibling, nth + 1);
} else {
return nth - 1;
}
};
const elementPath = (el, path = []) => {
if (isElement(el)) {
const tag = el.nodeName.toLowerCase(),
id = (el.id.length != 0 && el.id);
if (id) {
return elementPath(
null, path.concat([`#${id}`]));
} else {
return elementPath(
el.parentNode,
path.concat([`${tag}:nth-child(${nthChild(el)})`]));
}
} else {
return path.reverse().join(" > ");
}
};
There are some js libraries that do exactly this:
https://github.com/antonmedv/finder
https://github.com/gmmorris/simmerjs
I am using the first one and with success so far
function cssPath (e, anchor) {
var selector;
var parent = e.parentNode, child = e;
var tagSelector = e.nodeName.toLowerCase();
while (anchor && parent != anchor || !anchor && parent.nodeType === NodeTypes.ELEMENT_NODE) {
var cssAttributes = ['id', 'name', 'class', 'type', 'alt', 'title', 'value'];
var childSelector = tagSelector;
if (!selector || parent.querySelectorAll (selector).length > 1) {
for (var i = 0; i < cssAttributes.length; i++) {
var attr = cssAttributes[i];
var value = child.getAttribute(attr);
if (value) {
if (attr === 'id') {
childSelector = '#' + value;
} else if (attr === 'class') {
childSelector = childSelector + '.' + value.replace(/\s/g, ".").replace(/\.\./g, ".");
} else {
childSelector = childSelector + '[' + attr + '="' + value + '"]';
}
}
}
var putativeSelector = selector? childSelector + ' ' + selector: childSelector;
if (parent.querySelectorAll (putativeSelector).length > 1) {
var siblings = parent.querySelectorAll (':scope > ' + tagSelector);
for (var index = 0; index < siblings.length; index++)
if (siblings [index] === child) {
childSelector = childSelector + ':nth-of-type(' + (index + 1) + ')';
putativeSelector = selector? childSelector + ' ' + selector: childSelector;
break;
}
}
selector = putativeSelector;
}
child = parent;
parent = parent.parentNode;
}
return selector;
};
Better late than never: I came to this question and tried to use the selected answer, but in my case, it didn't worked because it wasn't very specific for my case. So I decided to write my own solution - I hope it may help some.
This solution goes like this: tag.class#id[name][type]:nth-child(?), and targeted with >.
function path(e) {
let a = [];
while (e.parentNode) {
let d = [
e.tagName.toLowerCase(),
e.hasAttribute("class") ? e.getAttribute("class") : "",
e.hasAttribute("id") ? e.getAttribute("id") : "",
e.hasAttribute("name") ? e.getAttribute("name") : "",
e.hasAttribute("type") ? e.getAttribute("type") : "",
0 // nth-child
];
// Trim
for (let i = 0; i < d.length; i++) d[i] = typeof d[i] == "string" ? d[i].trim() : d[i];
if (d[1] != "") d[1] = "."+d[1].split(" ").join(".");
if (d[2] != "") d[2] = "#"+d[2];
if (d[3] != "") d[3] = '[name="'+d[3]+'"]';
if (d[4] != "") d[4] = '[type="'+d[4]+'"]';
// Get child index...
let s = e;
while (s) {
d[5]++;
s = s.previousElementSibling;
}
d[5] = d[5] != "" ? ":nth-child("+d[5]+")" : ":only-child";
// Build the String
s = "";
for (let i = 0; i < d.length; i++) s += d[i];
a.unshift(s);
// Go to Parent
e = e.parentNode;
}
return a.join(">");
}
I know it's not that readable (I use it in my messy code), but it will give you the exact element(s) you're looking for. Just try it.

Get CSS path from Dom element

I got this function to get a cssPath :
var cssPath = function (el) {
var path = [];
while (
(el.nodeName.toLowerCase() != 'html') &&
(el = el.parentNode) &&
path.unshift(el.nodeName.toLowerCase() +
(el.id ? '#' + el.id : '') +
(el.className ? '.' + el.className.replace(/\s+/g, ".") : ''))
);
return path.join(" > ");
}
console.log(cssPath(document.getElementsByTagName('a')[123]));
But i got something like this :
html > body > div#div-id > div.site > div.clearfix > ul.choices > li
But to be totally right, it should look like this :html > body > div#div-id > div.site:nth-child(1) > div.clearfix > ul.choices > li:nth-child(5)
Did someone have any idea to implement it simply in javascript ?
The answer above actually has a bug in it — the while loop breaks prematurely when it encounters a non-element node (e.g. a text node) resulting in an incorrect CSS selector.
Here's an improved version that fixes that problem plus:
Stops when it encounters the first ancestor element with an id assigned to it
Uses nth-of-type() to make the selectors more readable
var cssPath = function(el) {
if (!(el instanceof Element))
return;
var path = [];
while (el.nodeType === Node.ELEMENT_NODE) {
var selector = el.nodeName.toLowerCase();
if (el.id) {
selector += '#' + el.id;
path.unshift(selector);
break;
} else {
var sib = el, nth = 1;
while (sib = sib.previousElementSibling) {
if (sib.nodeName.toLowerCase() == selector)
nth++;
}
if (nth != 1)
selector += ":nth-of-type("+nth+")";
}
path.unshift(selector);
el = el.parentNode;
}
return path.join(" > ");
}
To always get the right element, you will need to use :nth-child() or :nth-of-type() for selectors that do not uniquely identify an element. So try this:
var cssPath = function(el) {
if (!(el instanceof Element)) return;
var path = [];
while (el.nodeType === Node.ELEMENT_NODE) {
var selector = el.nodeName.toLowerCase();
if (el.id) {
selector += '#' + el.id;
} else {
var sib = el, nth = 1;
while (sib.nodeType === Node.ELEMENT_NODE && (sib = sib.previousSibling) && nth++);
selector += ":nth-child("+nth+")";
}
path.unshift(selector);
el = el.parentNode;
}
return path.join(" > ");
}
You could add a routine to check for unique elements in their corresponding context (like TITLE, BASE, CAPTION, etc.).
The two other provided answers had a couple of assumptions with browser compatibility that I ran into. Below code will not use nth-child and also has the previousElementSibling check.
function previousElementSibling (element) {
if (element.previousElementSibling !== 'undefined') {
return element.previousElementSibling;
} else {
// Loop through ignoring anything not an element
while (element = element.previousSibling) {
if (element.nodeType === 1) {
return element;
}
}
}
}
function getPath (element) {
// False on non-elements
if (!(element instanceof HTMLElement)) { return false; }
var path = [];
while (element.nodeType === Node.ELEMENT_NODE) {
var selector = element.nodeName;
if (element.id) { selector += ('#' + element.id); }
else {
// Walk backwards until there is no previous sibling
var sibling = element;
// Will hold nodeName to join for adjacent selection
var siblingSelectors = [];
while (sibling !== null && sibling.nodeType === Node.ELEMENT_NODE) {
siblingSelectors.unshift(sibling.nodeName);
sibling = previousElementSibling(sibling);
}
// :first-child does not apply to HTML
if (siblingSelectors[0] !== 'HTML') {
siblingSelectors[0] = siblingSelectors[0] + ':first-child';
}
selector = siblingSelectors.join(' + ');
}
path.unshift(selector);
element = element.parentNode;
}
return path.join(' > ');
}
Doing a reverse CSS selector lookup is an inherently tricky thing. I've generally come across two types of solutions:
Go up the DOM tree to assemble the selector string out of a combination of element names, classes, and the id or name attribute. The problem with this method is that it can result in selectors that return multiple elements, which won't cut it if we require them to select only one unique element.
Assemble the selector string using nth-child() or nth-of-type(), which can result in very long selectors. In most cases the longer a selector is the higher specificity it has, and the higher the specificity the more likely it will break when the DOM structure changes.
The solution below is an attempt at tackling both of these issues. It is a hybrid approach that outputs a unique CSS selector (i.e., document.querySelectorAll(getUniqueSelector(el)) should always return a one-item array). While the returned selector string is not necessarily the shortest, it is derived with an eye towards CSS selector efficiency while balancing specificity by prioritizing nth-of-type() and nth-child() last.
You can specify what attributes to incorporate into the selector by updating the aAttr array. The minimum browser requirement is IE 9.
function getUniqueSelector(elSrc) {
if (!(elSrc instanceof Element)) return;
var sSel,
aAttr = ['name', 'value', 'title', 'placeholder', 'data-*'], // Common attributes
aSel = [],
// Derive selector from element
getSelector = function(el) {
// 1. Check ID first
// NOTE: ID must be unique amongst all IDs in an HTML5 document.
// https://www.w3.org/TR/html5/dom.html#the-id-attribute
if (el.id) {
aSel.unshift('#' + el.id);
return true;
}
aSel.unshift(sSel = el.nodeName.toLowerCase());
// 2. Try to select by classes
if (el.className) {
aSel[0] = sSel += '.' + el.className.trim().replace(/ +/g, '.');
if (uniqueQuery()) return true;
}
// 3. Try to select by classes + attributes
for (var i=0; i<aAttr.length; ++i) {
if (aAttr[i]==='data-*') {
// Build array of data attributes
var aDataAttr = [].filter.call(el.attributes, function(attr) {
return attr.name.indexOf('data-')===0;
});
for (var j=0; j<aDataAttr.length; ++j) {
aSel[0] = sSel += '[' + aDataAttr[j].name + '="' + aDataAttr[j].value + '"]';
if (uniqueQuery()) return true;
}
} else if (el[aAttr[i]]) {
aSel[0] = sSel += '[' + aAttr[i] + '="' + el[aAttr[i]] + '"]';
if (uniqueQuery()) return true;
}
}
// 4. Try to select by nth-of-type() as a fallback for generic elements
var elChild = el,
sChild,
n = 1;
while (elChild = elChild.previousElementSibling) {
if (elChild.nodeName===el.nodeName) ++n;
}
aSel[0] = sSel += ':nth-of-type(' + n + ')';
if (uniqueQuery()) return true;
// 5. Try to select by nth-child() as a last resort
elChild = el;
n = 1;
while (elChild = elChild.previousElementSibling) ++n;
aSel[0] = sSel = sSel.replace(/:nth-of-type\(\d+\)/, n>1 ? ':nth-child(' + n + ')' : ':first-child');
if (uniqueQuery()) return true;
return false;
},
// Test query to see if it returns one element
uniqueQuery = function() {
return document.querySelectorAll(aSel.join('>')||null).length===1;
};
// Walk up the DOM tree to compile a unique selector
while (elSrc.parentNode) {
if (getSelector(elSrc)) return aSel.join(' > ');
elSrc = elSrc.parentNode;
}
}
I somehow find all the implementations unreadable due to unnecessary mutation. Here I provide mine in ClojureScript and JS:
(defn element? [x]
(and (not (nil? x))
(identical? (.-nodeType x) js/Node.ELEMENT_NODE)))
(defn nth-child [el]
(loop [sib el nth 1]
(if sib
(recur (.-previousSibling sib) (inc nth))
(dec nth))))
(defn element-path
([el] (element-path el []))
([el path]
(if (element? el)
(let [tag (.. el -nodeName (toLowerCase))
id (and (not (string/blank? (.-id el))) (.-id el))]
(if id
(element-path nil (conj path (str "#" id)))
(element-path
(.-parentNode el)
(conj path (str tag ":nth-child(" (nth-child el) ")")))))
(string/join " > " (reverse path)))))
Javascript:
const isElement = (x) => x && x.nodeType === Node.ELEMENT_NODE;
const nthChild = (el, nth = 1) => {
if (el) {
return nthChild(el.previousSibling, nth + 1);
} else {
return nth - 1;
}
};
const elementPath = (el, path = []) => {
if (isElement(el)) {
const tag = el.nodeName.toLowerCase(),
id = (el.id.length != 0 && el.id);
if (id) {
return elementPath(
null, path.concat([`#${id}`]));
} else {
return elementPath(
el.parentNode,
path.concat([`${tag}:nth-child(${nthChild(el)})`]));
}
} else {
return path.reverse().join(" > ");
}
};
There are some js libraries that do exactly this:
https://github.com/antonmedv/finder
https://github.com/gmmorris/simmerjs
I am using the first one and with success so far
function cssPath (e, anchor) {
var selector;
var parent = e.parentNode, child = e;
var tagSelector = e.nodeName.toLowerCase();
while (anchor && parent != anchor || !anchor && parent.nodeType === NodeTypes.ELEMENT_NODE) {
var cssAttributes = ['id', 'name', 'class', 'type', 'alt', 'title', 'value'];
var childSelector = tagSelector;
if (!selector || parent.querySelectorAll (selector).length > 1) {
for (var i = 0; i < cssAttributes.length; i++) {
var attr = cssAttributes[i];
var value = child.getAttribute(attr);
if (value) {
if (attr === 'id') {
childSelector = '#' + value;
} else if (attr === 'class') {
childSelector = childSelector + '.' + value.replace(/\s/g, ".").replace(/\.\./g, ".");
} else {
childSelector = childSelector + '[' + attr + '="' + value + '"]';
}
}
}
var putativeSelector = selector? childSelector + ' ' + selector: childSelector;
if (parent.querySelectorAll (putativeSelector).length > 1) {
var siblings = parent.querySelectorAll (':scope > ' + tagSelector);
for (var index = 0; index < siblings.length; index++)
if (siblings [index] === child) {
childSelector = childSelector + ':nth-of-type(' + (index + 1) + ')';
putativeSelector = selector? childSelector + ' ' + selector: childSelector;
break;
}
}
selector = putativeSelector;
}
child = parent;
parent = parent.parentNode;
}
return selector;
};
Better late than never: I came to this question and tried to use the selected answer, but in my case, it didn't worked because it wasn't very specific for my case. So I decided to write my own solution - I hope it may help some.
This solution goes like this: tag.class#id[name][type]:nth-child(?), and targeted with >.
function path(e) {
let a = [];
while (e.parentNode) {
let d = [
e.tagName.toLowerCase(),
e.hasAttribute("class") ? e.getAttribute("class") : "",
e.hasAttribute("id") ? e.getAttribute("id") : "",
e.hasAttribute("name") ? e.getAttribute("name") : "",
e.hasAttribute("type") ? e.getAttribute("type") : "",
0 // nth-child
];
// Trim
for (let i = 0; i < d.length; i++) d[i] = typeof d[i] == "string" ? d[i].trim() : d[i];
if (d[1] != "") d[1] = "."+d[1].split(" ").join(".");
if (d[2] != "") d[2] = "#"+d[2];
if (d[3] != "") d[3] = '[name="'+d[3]+'"]';
if (d[4] != "") d[4] = '[type="'+d[4]+'"]';
// Get child index...
let s = e;
while (s) {
d[5]++;
s = s.previousElementSibling;
}
d[5] = d[5] != "" ? ":nth-child("+d[5]+")" : ":only-child";
// Build the String
s = "";
for (let i = 0; i < d.length; i++) s += d[i];
a.unshift(s);
// Go to Parent
e = e.parentNode;
}
return a.join(">");
}
I know it's not that readable (I use it in my messy code), but it will give you the exact element(s) you're looking for. Just try it.

Categories