Prepend date in page name by hook creates improper results on some existing pages

[sebibu]

Hey Procis, wrote this - and my very first - hook „Prepend date in page name“, yesterday.
Once you get it, this system and the API is magnificent. Thank you @ryan and @all contributers!

On new (news) pages it works great so far. E.g.:
Title: Testnews mit Ümläüten
Altered page name: 2024-09-24-testnews-mit-ümläüten

But on existing pages special characters are removed. E.g.
Title: Partnerschaftliche Unterstützung - Ihr werdet gesucht!
Altered page name: 2023-07-27-fans1991-partnerschaftliche-untersttzung-ihr-werdet-gesucht

Another example is attached.
There you can see, that the ProcessPageEdit-message shows the right path!?

I have this in my site/config.php to allow only german special characters:

$config->pageNameCharset = 'UTF8';
$config->pageNameWhitelist = '-_abcdefghijklmnopqrstuvwxyz0123456789äöüß';

Temporary uninstalled Module „PagePathHistory“ with no look.

The code of my hook (suggestions for improvement are very welcome!):

/* NEWS pages only: prepend date in page-name */
$wire->addHookAfter('Pages::saveReady', function($event) {
  // get current Page object — in this case this is the first argument for the $event object
  $page = $event->arguments(0);
  
  // Only for pages with news template
  if($page->template == 'news') {
    // Test if field day exists, otherwise exit
    if (!$page->template->hasField('day')) {
      $event->message("Field 'day' is missing!");
      return;
    }
    $title = $page->get('title');
	// Sanitize title and substitute special characters
    $optimizedTitle = wire('sanitizer')->pageNameUTF8($title);
    //$optimizedTitle = wire('sanitizer')->pageName($title, Sanitizer::okUTF8); // or translate option

	$date = wireDate('Y-m-d', $page->day);
	// Set output formatting state off, for page manipulation
	$page->of(false);
	$page->name = $date.'-'.$optimizedTitle;
	//$page->save('name');
	$event->message("New saved name is $page->name");
    //$event->message("Path of new saved page is $page->path");
  }
});

Any idea why this doesn't works on all existing news, too?

Thx and 👋
Sebastian

[bernhard]

2 hours ago, sebibu said:

Hey Procis, wrote this - and my very first - hook „Prepend date in page name“, yesterday.
Once you get it, this system and the API is magnificent. Thank you @ryan and @all contributers!

Congrats and welcome to a new world of superpowers 😉 

I don't have details but I'm always using pageNameTranslate:

Not sure why that would work differently on new or existing pages...

2 hours ago, sebibu said:

The code of my hook (suggestions for improvement are very welcome!):

Instead of if(...) { ... } you can use early exit strategy which is usually cleaner in hooks. This is also called "guard clause":

So you could write it like this:

/* NEWS pages only: prepend date in page-name */
$wire->addHookAfter('Pages::saveReady', function($event) {
  // get current Page object — in this case this is the first argument for the $event object
  $page = $event->arguments(0);
  
  // Only for pages with news template
  if($page->template != 'news') return;

  // Test if field day exists, otherwise exit
  if (!$page->template->hasField('day')) {
    $event->message("Field 'day' is missing!");
    return;
  }

  $title = $page->get('title');
  // Sanitize title and substitute special characters
  $optimizedTitle = wire('sanitizer')->pageNameUTF8($title);
  //$optimizedTitle = wire('sanitizer')->pageName($title, Sanitizer::okUTF8); // or translate option

  $date = wireDate('Y-m-d', $page->day);
  // Set output formatting state off, for page manipulation
  $page->of(false);
  $page->name = $date.'-'.$optimizedTitle;
  //$page->save('name');
  $event->message("New saved name is $page->name");
  //$event->message("Path of new saved page is $page->path");
});

The difference is minimal here but using if(...) { ... } is usually harder to read, because you could have an else { } somewhere further down the hook, so when trying to understand your hook you have more workload for your brain. When using if($template != 'whatever') return; you instantly know the hook only applies to 'whatever' templates.

[sebibu]

Thank you @bernhard for helping optimizing the code!👍
For checking the template field I already used a guard clause, but for the template check it makes sense and is more readable to use it, too.

I would like to keep the german special characters like äöüß instead of exchanging (->pageNameTranslate) them with ae, oe, ue, ss.
So wire('sanitizer')->pageNameUTF8($title); should be the way to go. Right?

[bernhard]

Ah, sorry, didn't read carefully enough! Hope it was helpful nonetheless 🙂 

[poljpocket]

On 9/25/2024 at 2:13 PM, bernhard said:

When using if($template != 'whatever') return; you instantly know the hook only applies to 'whatever' templates.

For such cases, even more readability:

$wire->addHookAfter('Pages::saveReady(template=news)', function($event) {
   // [...]
}

This way, the hook doesn't even get called if template isn't news.

Also, I think you can even get rid of the second guard clause like this:

$wire->addHookAfter('Pages::saveReady(template=news,day!=)', function($event) {
   // [...]
}

If the page doesn't have a day field, the selector doesn't match and the hook isn't called. This is untested though 😊

Edited September 27 by poljpocket

[sebibu]

Thank you @poljpocket for this good optimizations!👍

While searching for hooks I stumpled across this syntax but didn't tested and then forgot it.🤓

[sebibu]

Unfortunately I have to come back to my still existing pagename problem.

This does not have anything to do with the hook.
In general the pagename is shown correct in the preview before saving:

But after saving special characters are removed:

What am I doing wrong?

I'm using this Extended Page Names feature: https://processwire.com/blog/posts/page-name-charset-utf8/

site/config.php:

$config->pageNameCharset = 'UTF8';
$config->pageNameWhitelist = "$-_.+!*'(),äöüßÄÖÜabcdefghijklmnopqrstuvwxyz0123456789";

.htaccess:

RewriteCond %{REQUEST_URI} "^/~?[-_/a-zA-Z0-9äöüß.+!*'(),]*$"

How to debug this? (Haven't configured PHP debug in VSC with DDEV.😆)

🙏👋

[poljpocket]

This uses the PageNameTranslate rules which you can configure in Modules > Configure > InputfieldPageName. By default, it will replace ä with a, ö with o and ü with u.

One of the first things I change on new sites.

BTW: @bernhard's RockMigrations has an automation for this.

Edited 17 hours ago by poljpocket

[sebibu]

Hi @poljpocket,

thanks for taking the time!

By default PageNameTranslate with this rules is used, that‘s right.

But I‘m using the Extended Page Name Feature for UTF-8 support. Then this rules are hidden and not used. Please see here:
https://processwire.com/blog/posts/page-name-charset-utf8/

[poljpocket]

You are obviously right! I am sorry...

I tried your settings in a quick-and-dirty local Docker install using the blank site profile.

I can use your whitelist no problem:

 

This is what I added to the files:

Edited 14 hours ago by poljpocket

[poljpocket]

You can maybe start by looking into where the problem occurs. You can for example use Tracy's bd() function to output the page name before and after every saveReady and saved hook. This will at least let you see in which part of the code you have to go dig for more.

Do you have any hooks already in use for Page::saveReady and Page::saved?

On which PW version are you?

Edited 13 hours ago by poljpocket

[sebibu]

Thanks for testing, @poljpocket!
Great to see that it is basically working.

Testing with a fresh install would have been my next step to narrow it down.👍

Interesting. I see you are using in your .htaccess

RewriteCond %{REQUEST_URI} "^/~?[-_/a-zA-Z0-9äöüß.+!*'(),]

instead of

RewriteCond %{REQUEST_URI} "^/~?[-_/a-zA-Z0-9äöüß.+!*'(),]*$"

So *$" is missing, which is necessary only for resolving the page request. Not build the pagename.
👋

 

[sebibu]

1 hour ago, poljpocket said:

Do you have any hooks already in use for Page::saveReady and Page::saved?

Only one which is only for pages with one determined template.
But the problem occurs for pages with all templates and even if the hook is disabled.

1 hour ago, poljpocket said:

On which PW version are you?

The newest stable one 3.0.229
All modules up to date.😇

BTW.. PHP mbstring extension is enabled.

[poljpocket]

1 hour ago, sebibu said:

Interesting. I see you are using in your .htaccess

RewriteCond %{REQUEST_URI} "^/~?[-_/a-zA-Z0-9äöüß.+!*'(),]

You are right, copy-pasta mistake. I just copied your stuff over. But I corrected it right now and it doesn't make a difference (and yes, it's only about the rewrite rules and doesn't affect the PW admin in any way). Maybe you will find something with the hooks I have talked about above.