This question already has answers here:
jQuery: find() children until a certain threshold element is encountered
(5 answers)
Closed 16 days ago.
TL;DR: How do I get an action like find(), but block traversal (not full stop, just skip) for a certain selector?
ANSWERS: $(Any).find(Selector).not( $(Any).find(Mask).find(Selector) )
There were many truly great answers, I wish I could some how distribute the bounty points more, maybe I should make some 50 pt bounties in response to some of these ;p I choose Karl-André Gagnon's because this answer managed to make findExclude unrequired in one, slightly long, line. While this uses three find calls and a heavy not filter, in most situations jQuery can use very fast implementation that skips traversal for most find()s.
Especially good answers are listed below:
falsarella: Good improvement on my solution, findExclude(), best in many situatoins
Zbyszek: A filter-based solution similar to falsarella's, also good on efficiency
Justin: A completely different, but manageable and functional solution to the underlaying issues
Each of these have their own unique merits and and are deserving of some mention.
I need to descend into an element fully and compare selectors, returning all matched selectors as an array, but skip descending into the tree when another selector is encountered.
Edit: replacing original code sample with some from my site
This is for a message forum which may have reply message-groups nested inside any message.
Notice, however, we cannot use the message or content classes because the script is also used for other components outside of the forum. Only InterfaceGroup, Interface and controls classes are potentially useful - and preferably just Interface and controls.
Interact with the code and see it in JS Fiddle, thanks Dave A, here Click on the buttons while viewing a JavaScript console to see that the controls class is being bound to one extra time per level of .Interface nesting.
Visual A, Forum Layout Struture:
<li class="InterfaceGroup">
<ul class="Interface Message" data-role="MessagePost" >
<li class="instance"> ... condensed ... </li>
<li class="InterfaceGroup"> ... condensed ...</li>
</ul>
<ul class="Interface Message" data-role="MessagePost" >
<li class="instance"> ... condensed ... </li>
</ul>
<ul class="Interface Message" data-role="MessagePost" >
<li class="instance"> ... condensed ... </li>
<li class="InterfaceGroup"> ... condensed ...</li>
</ul>
</li>
Inside of each <li class="InterfaceGroup"> there could be any number of repetitions of the same structure (each group is a thread of messages) and/or deeper nesting such as..
<li class="InterfaceGroup">
<ul class="Interface Message" data-role="MessagePost" >
<li class="instance"> ... condensed ... </li>
<li class="InterfaceGroup">
<ul class="Interface Message" data-role="MessagePost" >
<li class="instance"> ... condensed ... </li>
<li class="InterfaceGroup"> ... condensed ...</li>
</ul>
</li>
</ul>
</li>
Inside of each <li class="instance"> ... </li> there are arbitrary places decided by another team where class="controls" may appear and an event listener should be bound. Though these contain messages, other components structure their markup arbitrarily but will always have .controls inside of .Interface, which are collected into an .InterfaceGroup.A reduced-complexity version of the inner-content (for forum posts) is below for reference.
Visual B, Message Contents with controls class:
<ul class="Interface Message" data-role="MessagePost" >
<li class="instance">
<ul class="profile"> ...condensed, nothing clickable...</ul>
<ul class="contents">
<li class="heading"><h3>Hi there!</h3></li>
<li class="body"><article>TEST Message here</article></li>
<li class="vote controls">
<button class="up" data-role="VoteUp" ><i class="fa fa-caret-up"> </i><br/>1</button>
<button class="down" data-role="VoteDown" >0<br/><i class="fa fa-caret-down"> </i></button>
</li>
<li class="social controls">
<button class="reply-btn" data-role="ReplyButton" >Reply</button>
</li>
</ul>
</li>
<li class="InterfaceGroup" > <!-- NESTING OCCURRED -->
<ul class="Interface Message" data-role="MessagePost" >
<li class="instance">... condensed ... </li>
<li class="InterfaceGroup" >... condensed ... </li>
</ul>
</li>
</ul>
We can only bind to controls that are within an Interface class, instance may or may not exist but Interface will. Events bubble to .controls elements and have a reference to the .Interface which holds them..
So I am trying to $('.Interface').each( bind to any .controls not inside a deeper .Interface )
That's the tricky part, because
.Interface .controls will select the same .control multiple times in the .each()
.not('.Interface .Interface .controls') cancels out controls in any deeper nesting
How can I do this using jQuery.find() or a similar jQuery method for this?
I have been considering that, perhaps, using children with a not selector could work and could be doing the same thing as find under the hood, but I'm not so sure that it actually is or wont cause horrible performance. Still, an answer recursing .children effectively is acceptable.
UPDATE: Originally I tried to use a psuedo-example for brevity, but hopefully seeing a forum structure will help clarify the issue since they're naturally nested structures. Below I'm also posting partial javascript for reference, line two of the init function is most important.
Reduced JavaScript partial:
var Interface=function()
{
$elf=this;
$elf.call=
{
init:function(Markup)
{
$elf.Interface = Markup;
$elf.Controls = $(Markup).find('.controls').not('.Interface .controls');
$elf.Controls.on('click mouseenter mouseleave', function(event){ $elf.call.events(event); });
return $elf;
},
events:function(e)
{
var classlist = e.target.className.split(/\s+/), c=0, L=0;
var role = $(e.target).data('role');
if(e.type == 'click')
{
CurrentControl=$(e.target).closest('[data-role]')[0];
role = $(CurrentControl).data('role');
switch(role)
{
case 'ReplyButton':console.log('Reply clicked'); break;
case 'VoteUp':console.log('Up vote clicked'); break;
case 'VoteDown':console.log('Down vote clicked'); break;
default: break;
}
}
}
}
};
$(document).ready( function()
{
$('.Interface').each(function(instance, Markup)
{
Markup.Interface=new Interface().call.init(Markup);
});
} );
If you want to exclude element in you find, you can use a not filter. As for example, I've taken you function that exclude element and made it way shorter :
$.fn.findExclude = function( Selector, Mask,){
return this.find(Selector).not(this.find(Mask).find(Selector))
}
Now, ill be honest with you, I did not fully understand what you want. But, when i took a look at your function, I saw what you were trying to do.
Anyway, take a look at this fiddle, the result is the same as your : http://jsfiddle.net/KX65p/8/
Well, I really don't want to be answering my own question on a bounty, so if anyone can provide a better or alternative implementation please do..
However, being pressed to complete the project, I ended up working on this quite a bit and came up with a fairly clean jQuery plugin for doing a jQuery.find() style search while excluding child branches from the results as you go.
Usage to work with sets of elements inside nested views:
// Will not look in nested ul's for inputs
$('ul').findExclude('input','ul');
// Will look in nested ul's for inputs unless it runs into class="potato"
$('ul').findExclude('input','.potato');
More complex example found at http://jsfiddle.net/KX65p/3/ where I use this to .each() a nested class and bind elements which occur in each nested view to a class. This let me make components server-side and client-side reflect each other's properties and have cheaper nested event handling.
Implementation:
// Find-like method which masks any descendant
// branches matching the Mask argument.
$.fn.findExclude = function( Selector, Mask, result){
// Default result to an empty jQuery object if not provided
result = typeof result !== 'undefined' ?
result :
new jQuery();
// Iterate through all children, except those match Mask
this.children().each(function(){
thisObject = jQuery( this );
if( thisObject.is( Selector ) )
result.push( this );
// Recursively seek children without Mask
if( !thisObject.is( Mask ) )
thisObject.findExclude( Selector, Mask, result );
});
return result;
}
(Condensed Version):
$.fn.findExclude = function( selector, mask, result )
{
result = typeof result !== 'undefined' ? result : new jQuery();
this.children().each( function(){
thisObject = jQuery( this );
if( thisObject.is( selector ) )
result.push( this );
if( !thisObject.is( mask ) )
thisObject.findExclude( selector, mask, result );
});
return result;
}
Maybe something like this would work:
$.fn.findExclude = function (Selector, Mask) {
var result = new jQuery();
$(this).each(function () {
var $selected = $(this);
$selected.find(Selector).filter(function (index) {
var $closest = $(this).closest(Mask);
return $closest.length == 0 || $closest[0] == $selected[0] || $.contains($closest, $selected);
}).each(function () {
result.push(this);
});
});
return result;
}
http://jsfiddle.net/JCA23/
Chooses those elements that are either not in mask parent or their closest mask parent is same as root or their closest mask parent is a parent of root.
I think that this is the closest the findExclude can be optimized:
$.fn.findExclude = function (Selector, Mask) {
var result = $([]);
$(this).each(function (Idx, Elem) {
$(Elem).find(Selector).each(function (Idx2, Elem2) {
if ($(Elem2).closest(Mask)[0] == Elem) {
result = result.add(Elem2);
}
});
});
return result;
}
Also, see its fiddle with added logs with ellapsed time in milliseconds.
I see that you are worried with the performances. So, I've run some tests, and this implementation takes no longer than 2 milliseconds, while your implementation (as the answer you have posted) sometimes takes around 4~7 millisecods.
From my understanding, I would bind to the .controls elements and allow the event to bubble up to them. From that, you can get the closest .Interface to get the parent, if needed. This way you are added multiple handlers to the same elements as you go further down the rabbit hole.
While I saw you mention it, I never saw it implemented.
//Attach the event to the controls to minimize amount of binded events
$('.controls').on('click mouseenter mouseleave', function (event) {
var target = $(event.target),
targetInterface = target.closest('.Interface'),
role = target.data('role');
if (event.type == 'click') {
if (role) {
switch (role) {
case 'ReplyButton':
console.log('Reply clicked');
break;
case 'VoteUp':
console.log('Up vote clicked');
break;
case 'VoteDown':
console.log('Down vote clicked');
break;
default:
break;
}
}
}
});
Here is a fiddle showing what I mean. I did remove your js in favor of a simplified display.
It does seem that my solution may be a over simplification though...
Update 2
So here is a fiddle that defines some common functions that will help achieve what you are looking for...I think. The getInterfaces provides a simplified function to find the interfaces and their controls, assuming all interfaces always have controls.
There are probably fringe cases that will creep up though. I also feel I need to apologize if you have already ventured down this path and I'm just not seeing/understanding!
Update 3
Ok, ok. I think I understand what you want. You want to get the unique interfaces and have a collection of controls that belong to it, that make sense now.
Using this fiddle as the example, we select both the .Interface and the .Interface .controls.
var interfacesAndControls = $('.Interface, .Interface .controls');
This way we have a neat collection of the interfaces and the controls that belong to them in order they appear in the DOM. With this we can loop through the collection and check to see if the current element has the .Interface associated with it. We can also keep a reference to the current interface object we create for it so we can add the controls later.
if (el.hasClass('Interface')){
currentInterface = new app.Interface(el, [], eventCallback);
interfaces.push(currentInterface);
//We don't need to do anything further with the interface
return;
};
Now when we don't have the .Interface class associate with the element, we got controls. So let's first modify our Interface object to support adding controls and binding events to the controls as they are being added to the collection.
//The init function was removed and the call to it
self.addControls = function(el){
//Use the mouseover and mouseout events so event bubbling occurs
el.on('click mouseover mouseout', self.eventCallback)
self.controls.push(el);
}
Now all we have to do is add the control to the current interfaces controls.
currentInterface.addControls(el);
After all that, you should get an array of 3 objects (interfaces), that have an array of 2 controls each.
Hopefully, THAT has everything you are looking for!
If I understand you:
understanding your needs better and applying the specific classes you need, I think this is the syntax will work:
var targetsOfTopGroups = $('.InterfaceGroup .Interface:not(.Interface .Interface):not(.Interface .InterfaceGroup)')
This Fiddle is an attempt to reproduce your scenario. Feel free to play around with it.
I think I found the problem. You were not including the buttons in your not selector
I changed the binding to be
var Controls = $('.InterfaceGroup .Interface :button:not(.Interface .Interface :button):not(.Interface .InterfaceGroup :button)');
Fiddle
Why not taking the problem upside down?
Select all $(.target) elements and then discard them from further treatment if their .$parents(.group) is empty, that would give sonething like:
$('.target').each(function(){
if (! $(this).parents('.group').length){
//the jqueryElem is empy, do or do not
} else {
//not empty do what you wanted to do
}
});
Note that don't answer the title but literally gives you "Selector B, inside of a result from Selector A"
If your .interface classes had some kind of identifier this would seem to be rather easy.
Perhabs you already have such an identifier for other reasons or choose to include one.
http://jsfiddle.net/Dc4dz/
<div class="interface" name="a">
<div class="control">control</div>
<div class="branch">
<div class="control">control</div>
<div class="interface">
<div class="branch">
<div class="control">control</div>
</div>
</div>
</div>
<div class="interface" name="c">
<div class="branch">
<div class="control">control</div>
</div>
</div> </div>
$( ".interface[name=c] .control:not(.interface[name=c] .interface .control)" ).css( "background-color", "red" );
$( ".interface[name=a] .control:not(.interface[name=a] .interface .control)" ).css( "background-color", "green" );
Edit: And now Im wondering if you're tackling this problem from the wrong angle.
So I am trying to $('.Interface').each( bind to any .controls not
inside a deeper .Interface )
http://jsfiddle.net/Dc4dz/1/
$(".interface").on("click", ".control", function (event) {
alert($(this).text());
event.stopPropagation();
});
The event would be triggered on a .control; it would then bubble up to its .closest( ".interface" ) where it would be processed and further propagation be stopped. Isn't that what you described?
I am making a custom drop down menu with arrow key functions as well as being filtered by :contains.
I can't seem to get past the following code. What is happening is that I need to start the selection at the current hovered li, then proceed to the next or previous li that has the class match.
My following code starts correctly but even though it has .next('.match') it wont pass the <li>'s that either don't have the class (match) or are (hidden)
Does the next() function break on hidden elements?
Jquery Code
$('.dropdown_shell.opened li.match.hovered')
.next('.match')
.addClass('hovered')
.siblings()
.removeClass('hovered');
Html
<ul class="scroll">
<li class="selected default match">None</li>
<li class="" style="display: none;">For Sale</li>
<li class="match">For Rent</li>
<li class="" style="display: none;">For Lease</li>
<li class="match hovered">Sale or Lease</li>
<li class="match">New Listing</li>
<li class="match">Open House</li>
</ul>
Your next() function is not doing what you think it is. It looks at the next element, and only matches it if it makes the given selector. Otherwise it returns an empty jQuery object.
Instead, you want nextAll(), which looks at all future siblings, then use the first() method (or :first) selector to match the first one.
You can see this working here; http://jsfiddle.net/DH3hG/
To answer your question about whether next() considers hidden elements; yes it does. Anything that is inserted into the DOM is considered.
Jquery:
//Jquery-JS search for actors and directors
//Hide extra Actors
$(".actor ul li").not($(".actor ul li").slice(0,11)).hide();
//
function find_and_unhide (keyword,container) {
$(container+':contains("'+keyword+'")').show();
}
find_and_unhide("fra",".actor ul li");
Html:
<div class="option-combo actor">
<h4>Actor →</h4>
<ul class="filter option-set" data-filter-group="actor">
<li>Any</li>
<li>Sandro</li>
<li>Barbara</li>
<li>Ku</li>
<li>Cool</li>
<li>Aid</li>
<li>Leo</li>
<li>John</li>
<li>Kvara</li>
<li>Kuku</li>
<li>Bubu</li>
<li style="display: none;">Fra</li>
</ul>
</div>
Find and Unhide function doesn't work.
I'm also wondering if I should use data-filter-value instead of contains to select correct li. But which one will be faster?
Also How do i implement fuzzy matching?
I would do something like this to fix your find and unhide function:
function find_and_unhide (keyword,container) {
$(container).find('[data-filter-value="' + keyword + '"]').show();
}
I definitely think searching on the data attribute would be both faster and be more extensible in the long term, however, especially since you narrow down the DOM that needs to be searched I imagine the performance difference will be negligible either way. I personally think its cleaner to use the data attribute.
Don't know much about fuzzy matching, but maybe this question will give you a starting point : Getting the closest string match
Let's face this situation:
<ul>
<li>data</li>
<li class="selector">data2</li>
<li class="selector2">data3</li>
</ul>
What i'm trying to do is match lis that either have selector class or have class attribute undefined, something like this:
jQuery(function($) {
$('.selector2').prevAll('li.selector OR li[class==""]');
});
So if I'm running prevAll() on the .selector2, it should return 2 list items. If i run it on .selector, it should return the first list item.
So is there a way to replace that OR ... ?
PS: xpath may work for me too as i'm developing for modern browsers
jQuery(function($) {
$('.selector2').prevAll('li.selector, li:not([class])');
});
DEMO
Adding in important comment from #pimvdb
This is correct, but be careful - something like
.addClass("foo").removeClass("foo") leaves the class attribute
behind, although you (might) expect it to be in it's initial state. So
it's not quite the same as [class=''].
What i'm trying to do is match lis that either have "selector" class
or have class attribute undefined
This XPath expression is equivalent to the pseudo-code in the question:
/ul/li[#class='selector2']/preceding-sibling::li[#class='selector' or not(#class)]
However, a literal translation of the quoted requirement is:
/ul/li[#class='selector' or not(#class)]
How would I only select Item A and Item B pragmatically while excluding the sub item?
<div id="nav">
<ul>
<li>
<p>Item A</p>
<ul>
<li>
<p>Sub Item A</p>
</li>
</ul>
</li>
<li>Item B</li>
</ul>
</div>
Well after a quick test run - this is my contribution to this issue
$("#nav p:first, #nav > ul > li:eq(1)");
You specified that you wanted only those two items and no sub items so this is what jQuery will capture :
[<p>Item A</p>, <li>Item B</li>]
You can easily separate selectors by placing a comma between them.
Now that you have seen my solution I would strongly suggest that you take Xenon06's advice...
Giving your markup classes really helps you to keep track of them. Especially with jQuery. The class attribute while IMO mostly used for styling is a perfectly valid selector to use and abuse in your jQuery code. That is of course if you actually have access to that HTML. If you don't kindly ignore my last paragraph :)
This will select any first level li's that have only text and no children and any children of a li that isnt a ul. Given this is not a good way to do it. You should really put classes on your stuff to start with. But if that's not an option this will get you there.
$($('#nav').children()).children().each(function(){
if($(this).text() !== "" && $(this).children().length === 0 ){
$(this).addClass("IwantThisElement");
}
});
$($($('#nav') .children()) .children()) .children(':not(ul)').each(function(){
if($(this).text() !== ""){
$(this).addClass("IwantThisElement");
}
});
$('.IwantThisElement').text('Assuming Control');
Well, if your structure was more consistent, you could use direct children selectors, ie:
$("#nav ul li > p")
However your Item B is not in a paragraph. Without defining more what you want, you'll need to put classes on the items you want and do
$("#nav .yourclass")