I have tested this item ✅ successfully on 035c38b

I have tested this item ? unsuccessfully on 035c38b

facepalm

facepalm

hmm is something not working than expected?

This fails for multiple reasons.

1. If show last is enabled and this pr is applied then you have changed the last part from text to a link. It is deliberately not a link because it is then a link to itself - easy to see this with a single article menu item

2. This will change the visual display of any website with show last enabled - we really shouldnt do that.

3. The code actually says
   // Make a link if not the last item in the breadcrumbs
   Which you haven't removed even though its not true any more

4. Even if you check the template override in beez you will see that the last element is not a link

So what does google say about breadcrumbs?

Breadcrumb trails on a page indicate the page's position in the site hierarchy. A user can navigate all the way up in the site hierarchy, one level at a time, by starting from the last breadcrumb in the breadcrumb trail.

Which is why most people dont have show last enabled and if you do then you dont expect it to be a link to itself

Never rely on automated tools to test something :)

This is definitely a case where the user should create an override.

And even if it is decided to accept this it needs to be applied to beex as well

Which is why most people dont have show last enabled [...]

Well it is enabled by default. So i think most sites have this option enabled.

[...]and if you do then you dont expect it to be a link to itself

Ok got the point so the last is not part of the breadcrumb right? So this would be the correct fix == don't just add them partly to the breadcrumb microdata (5b10fe0)

@zero-24 I am confused now what the intention is here now. Can you please update the title and description please as now this isnt adding anything ;)

Done thanks.

Add missing itemprop item to the last breadcrumb item

can you correct that please

Before applying this PR the tool shows two errors
This PR fixes one of those errors

The tool still shows me the following error on a single article menu item

  <li itemprop="itemListElement" itemscope itemtype="https://schema.org/ListItem">
    <link
       itemprop="item" href="https://example.com/books/sciencefiction/ancillaryjustice">
      <span itemprop="name">Ancillary Justice</span></link>
    <meta itemprop="position" content="3" />
  </li>

That's valid for Google.

...but looks wrong somehow from my point of view. Couldn't find any reliable source if LINK in BODY is allowed that way.

The main problem is that G changes its criterias and things that were valid yesterday aren't valid today...

Don't forget to push #25117 ;-) and remove MicroData from Joomla.

The tool still shows me the following error on a single article menu item

hmm confirmed but it seams the tool is broken there? When I just enter the breadcrumb part of the page it works good but when i paste the complete page it also shows this error here.

I stick by my original comment that this is something a determined user can fix for themselves in a template override.

EDIT: the following is wrong, read below.
-- remove start --
The solution is to not say that the last item is a part of that breadcrumb list which it isn't, because it is not part of the way back.

Removing all the itemprop, itemscope and itemtype attributes from the $show_last part would do such a thing and it would be semantically appropriate.

At the moment the last item is an itemListElement of https://schema.org/BreadcrumbList, which is wrong, since it is not part of the way back.

So the following code would suffice:

<? php elseif ($show_last) :
	// Render last item if reqd. ?>
	<li class="active">
		<span>
			<?php echo $item->name; ?>
		</span>
	</li>
<? php endif;

-- remove end --

I just tested @StefanSTS suggestion and at least it makes google happy

according to https://developers.google.com/search/docs/data-types/breadcrumb

A breadcrumb trail on a page indicates the page's position in the site hierarchy. A user can navigate all the way up in the site hierarchy, one level at a time, by starting from the last breadcrumb in the breadcrumb trail.

Interesting is the last line stating: by starting from the last breadcrumb

If I understand the PR correctly the last part of the breadcrumb trail is excluded from the list.
Why is this? Why don't we show the last item as a link too? Even if it just is a link to the current page.

add the top code after 53 & 68 of modules/mod_breadcrumbs/tmpl/default.php

just stumbled upon an edge case we have to solve as well.

https://www.oeteldonk.org/veur-carnaval/proklemaosie-van-de-peer (Dutch website)
"veur-carnaval" is a menu item with type = heading.
Menu items with this type have no URL, therefor the Google Testing Tool will fail.
See screenshot below.

You can solve it with adding <meta itemprop="item" content="CURRENT URL"> to last item
Here a sample code

		<?php elseif ($show_last) :
			$uri = JFactory::getURI();
			$absolute_url = $uri->toString();
			// Render last item if reqd. ?>
			<li itemprop="itemListElement" itemscope itemtype="https://schema.org/ListItem" class="active">
				<span itemprop="name">
					<?php echo $item->name; ?>
				</span>
				<meta itemprop="position" content="<?php echo $key + 1; ?>">
				<meta itemprop="item" content="<?php echo $absolute_url; ?>">
			</li>
		<?php endif;

But you have another error when $item->link is empt y(in case you use separator for eg.)
I mean that in this case, the itemprop="itemListElement" should not be set, but this break the way breadcrumb is displayed by Google

None of both codes about the last item worked 100% for me. Is this issue still open?

I also encountered this issue. While it does seem silly to show the last (the current page) item linked, that is also how the Microdata examples have done it for a very long time. It was also mentioned elsewhere that in menus we do link to the current page (just styling it differently), so in that sense not too different.

In an earlier PR for Joomla 4.0 @zwiastunsw made a reasonable argument why having the last item in the breadcrumb as a link, also make sense for accessibility.
#24113

For my own quick-fix of the Missing field "item" error, I only changed the "Render last item if reqd." section of the breadcrumb template in an override (but we should not expect most users to know how to do this.)

So that part I changed to:

// Render last item if reqd. ?> <li itemprop="itemListElement" itemscope itemtype="https://schema.org/ListItem" class="breadcrumb-item active"> <a itemtype="https://schema.org/Thing" itemprop="item" href="<?php echo JURI::getInstance()->toString(); ?>"> <span itemprop="name"><?php echo $item->name; ?></span></a> <meta itemprop="position" content="<?php echo $key + 1; ?>"> </li>

I don't know if echo JURI::getInstance()->toString(); is the best way to get the correct link, but it worked for a few different scenarios that I tested it with.

One could add an option to the breadcrumb module, to display the last item as a link or not. That would avoid changing old layouts unless they want to change.

@hans2103 I don't know how we would deal with your situation. Someone with good accessibility knowledge should perhaps comment if the practice (that I also use) of having some menu items that are not actual links (at all or just # ) will create other issues.

The information page about breadcrumbs written by Google tells us that the implementation is depending upon the search context. There is a example given where a specific book title is search for and the breadcrumbs include this page. But, they also specify that the last item should be omitted if "the page could be found in a larger set of search result". To me it is unclear what the best solution is, but it also indicates that both could be right and valid.

The current implementation is semantically wrong. I am (and probably nobody is) not quite sure about how Google processes semantically incorrect microdata, but it should be fixed for sure. Since the last item never has been a true HTML link <a> but was a sort off link with the microdata is seems logic to keep it that way. A possible semantically correct way to do this without changing the page visually could be this:

<li itemprop="itemListElement" itemscope itemtype="http://schema.org/ListItem">
    <span itemscope itemtype="http://schema.org/Thing" itemprop="item" itemid="/ul/to/page">
      <span itemprop="name">This page</span>
    </span>
    <meta itemprop="position" content="3" />
</li>

Note the itemscope on the first <span> element. This is mandatory, without the validation fails. This will also add a additional <span> element to the last page. And as last note, this PR should be ported to J4 whatever the outcome may be. The current implementation on that branch also won't pass the validation.

I have just implemented the suggested changes by @n9iels thanks.

Haven't tried the latest changes and, therefore, my comments may be inappropriate. See my comments posted at the J! forum a couple of days ago: https://forum.joomla.org/viewtopic.php?f=712&t=974526#p3581156

Tested the last-mentioned update to the changed file. While it render's valid microdata (and passes GSC's requirements) it still does not meet my needs as I described at https://forum.joomla.org/viewtopic.php?f=712&t=974526#p3581156

Looks like GSC is being unnecessarily finnicky (?)

Tested the last-mentioned update to the changed file. While it render's valid microdata (and passes GSC's requirements) it still does not meet my needs as I described at https://forum.joomla.org/viewtopic.php?f=712&t=974526#p3581156

Looks like GSC is being unnecessarily finnicky (?)

I've tested it and it looks like that the $item->link is empty for the last item (the current active item) when opening a article that is linked from a category page has no direct menu link. On normale menu pages it all seems to work fine.

This behaviour is caused by this line, that is writen 9 years ago as far as I can tell ? No idea what the intention was.

This could be replaced by the following line:

 $path     = array(array('title' => $this->item->title, 'link' => ContentHelperRoute::getArticleRoute($this->item->slug, $this->item->catid, $this->item->language)));

Not sure if this could have any side effects, in what situation there is no direct link to the article and what will happen on that point.

$item->link did not work for me for the last item on pages that don't have a menu link. (like pages in a news category)

I don't know if echo JURI::getInstance()->toString(); is the best way to get the correct link, but it worked for a few different scenarios that I tested it with.

@n9iels Where does itemid come from? Can it be used with BreadcrumbList? Is Joomla following the Microdata breadcrumb specification from schema.org or somewhere else?

@sozzled the "issue" that you describe in the forum does not seem to be an issue? Searching for a popular term, all results are more or less the same irrespective of Joomla or another CMS or even if microdata was used or just a div called breadcrumbs. (Google shows the base URL at the start, which makes sense, since the user may want to see where they are going before clicking on the link)

@JacquesR See this page: https://schema.org/docs/datamodel.html#identifierBg

Hello,

I have made an attempt for showing the last item without a link and still getting the Google Tool happy. I described it here: https://forum.joomla.org/viewtopic.php?f=712&t=974526&p=3581290#p3581290

My approach to just say, the last item is not part of the breadcrumb list was nonsense, the first sentence in the description on schema.org says definitely something different:
"A BreadcrumbList is an ItemList consisting of a chain of linked Web pages, typically described using at least their URL and their name, and typically ending with the current page."
That was trying the easy way out. Sorry for that. So I put a bit time in, the result now works and is valid structured data. But @sozzled had test cases for the other approaches that did not work, so it would be great if you could test, sozzled.

There's some confusion about whether Google is making changes behind-the-scenes as we speak (see https://forum.joomla.org/viewtopic.php?f=712&t=974526&p=3581379#p3581370) and so I'm waiting for things to settle down until we know what we know about GSC and its enforcement of b/crumbs microdata parsing. I'm a bit lost with this.

I saw that post, but that one error about the @id will not go away. So if anyone would test the default.php in the forum post, it could help.
You can compare the result of a plain j installation with this result:
https://search.google.com/structured-data/testing-tool#url=https%3A%2F%2Fphp72.sts-hosting.de%2Fvm3%2Fde%2Funited-states-progress-map-2

Did you also see https://forum.joomla.org/viewtopic.php?f=712&t=974526#p3581373 ?

I'm happy enough to test things but, if Google is making changes as we speak, I'm wondering if the exercise is useful? I'm not concerned about "valid microdata". I am concerned that the "last" segment of the b/crumbs doesn't appear on G's SERPs. The 3rd-party extension Krizalys Breadcrumbs (https://extensions.joomla.org/extension/krizalys-breadcrumbs/) took care of that however, GSC now (or, at least a few days ago) doesn't like it.

The last segment got an @id and a name now. So the SERP's should show the last item.
In both cases, that i added as option, showing the last item as link or with a simple span element.

I saw that post, but that one error about the @id will not go away. So if anyone would test the default.php in the forum post, it could help.
You can compare the result of a plain j installation with this result:
https://search.google.com/structured-data/testing-tool#url=https%3A%2F%2Fphp72.sts-hosting.de%2Fvm3%2Fde%2Funited-states-progress-map-2

The default.php in that formpost is basically the same implementation as this PR, with the exception that the link for the last item does work. In the formpost it is fixed by add <?php echo JUri::getInstance(); ?> but I'am not a big fan of that since it is returning the absolute url of the page. All other items have a relative url, also in my opinion it should be fixed in the view.html.php as I proposed a few comments back.

The current situation for me is that for the current core implementation the test fails, and when testing this PR both versions with a empty itemid and the version with a itemid that contains a valid url both pass. In case of the empty itemid the tool is using the given URL as @id value for the last breadcrumb. I agree with @sozzled to give it a few days.

The absolute URI is needed for itemid, because it is not automatically converted to the absolute URI. For the href JUri:: getInstance() could be changed to the relative URI, since there it is resolved against the base URI.
Since this is not a Google problem I disagree that waiting a few days will give more clarity. If the Microdata is properly implemented, Google will not show errors. At the moment the Microdata is not valid since the identifier for the last item is missing.

Great let me know when you get feedback from google on this ?

I just tested the Joomla! Core implementation against the testing tool and it passed. The @id has no value but also gives no error or warning. Someone else that can confirm this behavior? And if Google fixed this issue, should we still implement this PR?

@n9iels Well sounds like google fixed the issue and we can close this issue here? Feel free to re-open when needed.