"canonical" issue

HaraldJ · 2016-02-04 01:03:58

After the update from CE3 to CE4, the website is now running since a few days. One problem has come up:

In CE3, building sitemaps and image sitemaps has been no problem at all. In CE4, the sitemaps are almost empty.

The reason seems to be related to wrong canonical tags:

In galleries.php, line 50 reads

when it should be

The PHP code line in the exported "galleries.php" is

Does anyone have an idea what's going wrong or, more importantly, how to fix it?

Thanks!

Harald

rod barbee · 2016-02-04 02:20:54

Just looking at the code in the exported php file, the comment for the function currentPageLocation shows that the function "Gets current page parent location". Which, for those pages in the root of the site, will yield the site url.

I don't know where the :443 is coming from in yours. Mine (from the galleries.php page or any of the pages generated by CE4 Pages) looks like this:

<link rel="canonical" href="http://rodbarbee.com/" />

if I come to my site by going to http://rodbarbee.com
or this:

<link rel="canonical" href="http://barbeephoto.com/" />

If I go to my site using http://barbeephoto.com
(I redirect the rodbarbee.com url to barbeephoto.com)

So it doesn't look like this is coming from TTG, rather, it's coming from the server.

that 443 is in several places in your source code and my guess is that it's coming from your host.

Matt or Ben will have to comment about any "fix", if there is one.

HaraldJ · 2016-02-04 02:30:44

rod barbee wrote:

Just looking at the code in the exported php file, the comment for the function currentPageLocation shows that the function "Gets current page parent location". Which, for those pages in the root of the site, will yield the site url.
I don't know where the :443 is coming from in yours. Mine (from the galleries.php page or any of the pages generated by CE4 Pages) looks like this:
<link rel="canonical" href="http://rodbarbee.com/" />
if I come to my site by going to http://rodbarbee.com
or this:
<link rel="canonical" href="http://barbeephoto.com/" />
If I go to my site using http://barbeephoto.com
(I redirect the rodbarbee.com url to barbeephoto.com)
So it doesn't look like this is coming from TTG, rather, it's coming from the server.
that 443 is in several places in your source code and my guess is that it's coming from your host.
Matt or Ben will have to comment about any "fix", if there is one.

Hi Rob,

Thanks for having a look. The 443 is the port for HTTPS, so that's correct, and comes indeed from the server.

On most pages all seems to be correct, I picked a random page and find

which is correct.

The problem seems to be specific to galleries.php, where the canonical tag refers to the root, instead of galleries.php.

Harald

rod barbee · 2016-02-04 03:52:50

The problem seems to be specific to galleries.php, where the canonical tag refers to the root, instead of galleries.php

I think that's what's supposed to happen. All pages in the root of the site will give the same result. The code gives the parent location of the page.
If you were to be looking at a gallery's index page the the canonical url would be something like yoursite.com/galleries/some-index/actual-gallery/
So it shows the folder, not the file.

HaraldJ · 2016-02-04 04:04:50

rod barbee wrote:

The problem seems to be specific to galleries.php, where the canonical tag refers to the root, instead of galleries.php
I think that's what's supposed to happen. All pages in the root of the site will give the same result. The code gives the parent location of the page.
If you were to be looking at a gallery's index page the the canonical url would be something like yoursite.com/galleries/some-index/actual-gallery/
So it shows the folder, not the file.

Hi Rob,

Please have look at this Google information. The canonical should not point to the folder.

I have hardcoded the correct information in galleries.php, and the sitemap generation gets a bit further but is then stuck again, so I guess that problem isn't limited to galleries.php.

Cheers

Harald

rod barbee · 2016-02-04 05:01:27

I don't write the code so this is something Matt will have to comment on.

Ben · 2016-02-04 07:29:07

Hi Harald, can you post the contents of the currentPage functions as found in the template?

HaraldJ · 2016-02-04 07:40:23

Ben wrote:

Hi Harald, can you post the contents of the currentPage functions as found in the template?

Hi Ben,

Here is the code from galleries.php. It's quite different to the very short function from CE3.

Thanks!

Harald

if (!function_exists('currentPageLocation')) {
        // Gets current page parent location
        function currentPageLocation() {

            $currentPageURL = currentPageURL();
            $returnURL = '';

            $finalSlash = strrpos(currentPageURL(), '/');
            if (strrpos($currentPageURL, '.') > strrpos($currentPageURL, '/')) // has a file after the final slash, e.g. http://url/directory/index.php
                $returnURL = substr($currentPageURL, 0, $finalSlash);
            else if (strrpos($currentPageURL, '?') > strrpos($currentPageURL, '/')) // has a ? after the final slash, e.g. http://url/directory/index.php
                $returnURL = substr($currentPageURL, 0, $finalSlash);
            else if ($finalSlash == strlen($currentPageURL)-1) // final character is a slash, e.g. http://url/directory/
                $returnURL = substr($currentPageURL, 0, $finalSlash);
            else // final character is not a slash, e.g. http://url/directory
                $returnURL = $currentPageURL;

            return $returnURL;
        }

Ben · 2016-02-04 08:35:34

Hi Harald, can you also post currentPageURL?

rod barbee · 2016-02-04 08:47:29

Ben, here's the currentPageURL() function

if (!function_exists('currentPageURL')) {
        // Gets current page URL
        function currentPageURL() {
            $pageURL = 'http';

            if (!empty($_SERVER['HTTPS']) && $_SERVER['HTTPS'] == 'on') {$pageURL .= "s";}
            $pageURL .= "://";

            if ($_SERVER["SERVER_PORT"] != "80" && strpos($host, ':') === false) {
                $pageURL .= $_SERVER["SERVER_NAME"].":".$_SERVER["SERVER_PORT"].$_SERVER["REQUEST_URI"];
            } else {
                $pageURL .= $_SERVER["SERVER_NAME"].$_SERVER["REQUEST_URI"];
            }

            return $pageURL;
        }
    }

Ben · 2016-02-04 11:58:51

Thanks Rod. It looks like there may be a logic error in the code. Harald, can you try changing:

=== false

to:

!== false

?

HaraldJ · 2016-02-04 16:49:43

Ben wrote:

Thanks Rod. It looks like there may be a logic error in the code. Harald, can you try changing:
=== false
to:
!== false
?

Hi Ben,

Thanks! The change has a nasty side effect on other lines, and it turns this:

      
      <meta property="og:image" content="https://www.haraldjoergens.com:443/photos/????????.DNG.jpg" />
      <meta property="og:site_name" content="<!--TitleToReplace-->" />

into an "unexpected <" PHP error in the second line. I guess you are right about the problem, but just changing the "equals" to "not equals" has implications.

Harald

Ben · 2016-02-04 18:10:11

Hi Harald, that's strange. Does this line:

if ($_SERVER["SERVER_PORT"] != "80" && strpos($host, ':') === false) {

Now look exactly like this line?

if ($_SERVER["SERVER_PORT"] != "80" && strpos($host, ':') !== false) {

HaraldJ · 2016-02-04 18:15:23

Ben wrote:

Hi Harald, that's strange. Does this line:
if ($_SERVER["SERVER_PORT"] != "80" && strpos($host, ':') === false) {
Now look exactly like this line?
if ($_SERVER["SERVER_PORT"] != "80" && strpos($host, ':') !== false) {

Hi Ben,

Yes, it does. I tried it twice, only changing the "===" to "!==" instead of copying and pasting the line, and the result was the same.
Looking at the code, I find it strange too, but I don't want to fiddle too much with a live website.
Of course I don't have a certificate on my test subdomain or localhost, to I can't really test HTTPS in a test environment.

Cheers

Harald

rod barbee · 2016-02-05 00:09:21

Hi Ben,

As another data point, I made the same change to my galleries.php file and I'm seeing no difference at all in the output page source code. I'm not getting the same error that Harald is seeing.

HaraldJ · 2016-02-05 00:12:33

rod barbee wrote:

Hi Ben,
As another data point, I made the same change to my galleries.php file and I'm seeing no difference at all in the output page source code. I'm not getting the same error that Harald is seeing.

Rob, did you test on an HTTPS site?

HaraldJ · 2016-02-05 00:22:55

HaraldJ wrote:

Ben wrote:
Hi Harald, that's strange. Does this line:
if ($_SERVER["SERVER_PORT"] != "80" && strpos($host, ':') === false) {
Now look exactly like this line?
if ($_SERVER["SERVER_PORT"] != "80" && strpos($host, ':') !== false) {
Hi Ben,
Yes, it does. I tried it twice, only changing the "===" to "!==" instead of copying and pasting the line, and the result was the same.
Looking at the code, I find it strange too, but I don't want to fiddle too much with a live website.
Of course I don't have a certificate on my test subdomain or localhost, to I can't really test HTTPS in a test environment.
Cheers
Harald

Ben,

With a but more testing, I think the function currentPageLocation() is correct after all.

In the line that creates the "canonical" meta-tag I think it should be function getcurrentPageURL() instead of getcurrentPageLocation(), but I'm looking further into it to understand why the sitemap doesn't cover anything within /galleries/.

Cheers

Harald

rod barbee · 2016-02-05 00:32:24

HaraldJ wrote:

rod barbee wrote:
Hi Ben,
As another data point, I made the same change to my galleries.php file and I'm seeing no difference at all in the output page source code. I'm not getting the same error that Harald is seeing.
Rob, did you test on an HTTPS site?

No

Rod

HaraldJ · 2016-02-05 05:18:42

Just had a look at the CE3 code:

<link rel="canonical" href="<?php echo currentPageLocation(); ?>/<?php echo currentPageName(); ?>" />

which means something like

CE4 uses

<link rel="canonical" href="<?php echo currentPageLocation(); ?>/" />

which means something like

CE3 doesn't seem to care about HTTP or HTTPS, but sitemap building worked fine,
CE4 does care about HTTP and HTTPS, and sitemap building doesn't work.

That's not an analysis, just an observation. I still don't know where exactly the sitemap building goes wrong.

Matthew · 2016-02-05 13:16:23

HaraldJ wrote:

In the line that creates the "canonical" meta-tag I think it should be function getcurrentPageURL() instead of getcurrentPageLocation(), but I'm looking further into it to understand why the sitemap doesn't cover anything within /galleries/.

Not so. Using getcurrentPageURL() would, for mobile galleries, return mobile.php; that's exactly what the canonical URLs are in place to prevent. For galleries, we want the mobile gallery pages to canonically be read as the main index.php, hence our use of getcurrentPageLocation().

HaraldJ · 2016-02-05 13:33:26

Matthew wrote:

HaraldJ wrote:
In the line that creates the "canonical" meta-tag I think it should be function getcurrentPageURL() instead of getcurrentPageLocation(), but I'm looking further into it to understand why the sitemap doesn't cover anything within /galleries/.
Not so. Using getcurrentPageURL() would, for mobile galleries, return mobile.php; that's exactly what the canonical URLs are in place to prevent. For galleries, we want the mobile gallery pages to canonically be read as the main index.php, hence our use of getcurrentPageLocation().

Thanks, Matt!

I see your point, but do you have an idea why building a sitemap ignores the galleries then?

Cheers

Harald

Matthew · 2016-02-05 14:15:02

Sounds like the whole thing is misbehaving with Pages, as you have observed. It's possible that we might have unintentionally broken the function through the course of evolving the product, making codebase modifications for the sake of the publisher, etc. We probably made changes to the galleries, echoed those changes through all of the plugins for consistency in code, and neglected to account for this outcome. We will probably need to revisit the functions in CE4 Pages.

HaraldJ · 2016-02-06 04:35:10

Matthew wrote:

Sounds like the whole thing is misbehaving with Pages, as you have observed. It's possible that we might have unintentionally broken the function through the course of evolving the product, making codebase modifications for the sake of the publisher, etc. We probably made changes to the galleries, echoed those changes through all of the plugins for consistency in code, and neglected to account for this outcome. We will probably need to revisit the functions in CE4 Pages.

Today I have replaced, for all files on the website, the CE4 canonical tag code with the code from CE3:

CE4:

<link rel="canonical" href="<?php echo currentPageLocation(); ?>/" />

CE3:

<link rel="canonical" href="<?php echo currentPageLocation(); ?>/<?php echo currentPageName(); ?>" />

It was a test, and if anything goes wrong, I can go back to the original version.

But the sitemap building seems to go well now, and I tried the galleries on mobile devices, where they seem to be working without any problems.

I also added some experimental code to deal with the SEO pagination problem, search engines will see duplicate content when they go through galleries with more than one page (see here for the technical background).

Thanks to the excellent PHP code in CE4, it seemed quite simple to add two additional meta-tags, "prev" and "next". This example is from page 5 of a gallery:

<link rel="canonical" href="https://www.haraldjoergens.com:443/galleries/rowing/the-boat-race/2016-01-31-cuwbc-vs-obubc/index.php" />

<link rel="prev" href="https://www.haraldjoergens.com:443/galleries/rowing/the-boat-race/2016-01-31-cuwbc-vs-obubc/index.php?page=4" />
<link rel="next" href="https://www.haraldjoergens.com:443/galleries/rowing/the-boat-race/2016-01-31-cuwbc-vs-obubc/index.php?page=6" />

I'll find out what the web crawlers will make of it!

Daniel Leu · 2016-02-06 10:33:53

That's really interesting. Thank you for sharing and including the background infos! Please keep us posted on the success of crawlers indexing your paged galleries.

Best

HaraldJ · 2016-02-07 21:33:47

Talking about sitemaps:

If you are using a software or a service to create your sitemaps, it might be a good idea to exclude any links that contain

-single.php
thumbnails
sender.a.getAttribute('data-gps')
http://maps.google.com

If you don't exclude *-single.php, every photo will be seen as a page.

And if you don't exclude *thumbnails*, your image sitemap will include three entries for every photo, one from /photos, one from /thumbnails, and one from /thumbnails-for-mobile.

The sender.a.getAttribute('data-gps') makes sense if you are using map information in your galleries, without this exclusion parameter you might get a large number of broken links.

And the Google Maps parameter stops every single photo with map information from being added as an extra page with Google Map information.

Community @ The Turning Gate

#1 2016-02-04 01:03:58

"canonical" issue

#2 2016-02-04 02:20:54

Re: "canonical" issue

#3 2016-02-04 02:30:44

Re: "canonical" issue

#4 2016-02-04 03:52:50

Re: "canonical" issue

#5 2016-02-04 04:04:50

Re: "canonical" issue

#6 2016-02-04 05:01:27

Re: "canonical" issue

#7 2016-02-04 07:29:07

Re: "canonical" issue

#8 2016-02-04 07:40:23

Re: "canonical" issue

#9 2016-02-04 08:35:34

Re: "canonical" issue

#10 2016-02-04 08:47:29

Re: "canonical" issue

#11 2016-02-04 11:58:51

Re: "canonical" issue

#12 2016-02-04 16:49:43

Re: "canonical" issue

#13 2016-02-04 18:10:11

Re: "canonical" issue

#14 2016-02-04 18:15:23

Re: "canonical" issue

#15 2016-02-05 00:09:21

Re: "canonical" issue

#16 2016-02-05 00:12:33

Re: "canonical" issue

#17 2016-02-05 00:22:55

Re: "canonical" issue

#18 2016-02-05 00:32:24

Re: "canonical" issue

#19 2016-02-05 05:18:42

Re: "canonical" issue

#20 2016-02-05 13:16:23

Re: "canonical" issue

#21 2016-02-05 13:33:26

Re: "canonical" issue

#22 2016-02-05 14:15:02

Re: "canonical" issue

#23 2016-02-06 04:35:10

Re: "canonical" issue

#24 2016-02-06 10:33:53

Re: "canonical" issue

#25 2016-02-07 21:33:47

Re: "canonical" issue

Board footer