Leaderboard

This week I've been busy developing a site. It's the same one as the last few weeks, which is an established site moving out of WordPress and into ProcessWire. Next week the whole thing is getting uploaded to its final server for collaboration and for client preview. I've been pretty focused on getting that ready and don't have any core updates to report this week, though should next week. One thing about the prior version of this site (and perhaps many WordPress sites) is that there wasn't much structure to the pages used in the site, and hundreds of unrelated pages in the site confusingly live off the root, such as /some-product/ and /some-press-release/ and /some-other-thing/. There's no apparent structure or order to it. And those pages that do have some loose structure in the wp-admin have URLs that don't represent that structure on the front-end. There's very little relation between the structure one sees in the wp-admin and the structure that one sees in the URLs, or in the front-end navigation. They all seem to be completely unrelated. That's one thing that I've tried to fix, so that there is some logic and structure rather than having a bunch of unrelated pages all in the same bucket (is this common in WordPress?) But there's one big caveat. We didn't want to change anything about the actual URLs that are used on the site. This is a site with a long history, a lot of incoming links, and a lot of search traffic. The current URLs have been in place a long time and we didn't want to introduce more redirects into the site (there are already a ton of 301 redirects accumulated over time). So we wanted to make sure the existing URLs in the new ProcessWire-powered site are identical to what they were in the WordPress site. That might seem difficult to do, going from an unstructured WordPress site into a highly structured ProcessWire site... but actually it's very simple. Here's how: Making ProcessWire render pages from WordPress URLs We created a new URL field named "href" and added it to most of the site's templates. For established pages that came in from the old WP site, this "href" field contains the WordPress URL for the page. Depending on the page, it contains values like /some-product/ or /some-press-release/, /some-country/some-town/, etc. In most cases this is different from the actual ProcessWire URL, since the page structure is now much more orderly in the back-end. And for newly added pages, we'll be using the more logical ProcessWire URL. But for all the old WordPress pages, we'll make them render from their original URL. This is certainly preferable from an SEO standpoint but also helps to limit the redirects in the site. In order to make $page->url return the old/original WordPress URL (value in $page->href), a hook was added to the /site/init.php file: /** * Update page paths and URLs to use the 'href' field value on pages that have it * */ $wire->addHookBefore('Page::path', function(HookEvent $event) { $page = $event->object; /** @var Page $page */ $href = $page->get('href'); if(!$href) return; // skip if page has no 'href' value $event->return = $href; $event->replace = true; }); Now we've got $page->url calls returning the URLs we want, but how do we get ProcessWire to accept those URLs for rendering pages? The first thing we'll need to do is enable URL segments for the homepage template. We do this by going to: Setup > Templates > home > URLs > and check the box to enable URL segments. Save. Then we need to edit our /site/templates/home.php to identify when URL segments are present and render the appropriate page, rather than the homepage: $href = $input->urlSegmentStr; if($href) { $href = '/' . trim($href, '/') . '/'; $page = $pages->get("href=$href"); if(!$page->id || !$page->viewable()) wire404(); $wire->wire('page', $page); // set new $page API var include($page->template->filename); // include page's template file } else { // render homepage output } As you can see from the above, when URL segments are present, we find a page that has an "href" field value matching those URL segments ($input->urlSegmentStr). If we don't find one, we stop with a 404. But if we do find one, then we set the $page API variable to it and then include its template file, making that page render rather than the homepage. If there is no $input->urlSegmentStr present then of course we just render the homepage. That's it! By using these little bits of code snippets to replace ProcessWire's URL routing, now all the URLs of the old WordPress site are handled by ProcessWire. Like most things in ProcessWire, there's more than one way to accomplish this… We could have used URL/path hooks, or we could have hooked before Page::render to take over homepage requests with URL segments, before the homepage even got involved. Or perhaps we could have hooked in even earlier, to something in the ProcessPageView module or PagesRequest class. Or we could have used an existing module. Any of these might be equally good (or even better) solutions, but I just went with what seemed like the simplest route, one that I could easily see and adjust. Plus, it'll work in any version of ProcessWire. The actual solution I used is a little more than what's presented here, as it has a few fallbacks for finding pages and scanning redirect lists, plus passes along remaining pagination/URL segments to rendered pages. I'm guessing most don't need that stuff, and it adds a decent chunk of code, so I left that out. But there are a couple of optional additions that I would recommend adding in a lot of cases: Forcing pages to only render from their "href" URL and not their ProcessWire URL (optional) Our existing hooks ensure that any URLs output for pages having "href" values are always based on that "href" value. But what if someone accesses a given page at its ProcessWire path/url? The page would render normally. We might want to prevent that from happening, ensuring that it only ever renders at its defined "href" URL instead. If the page is requested at its ProcessWire URL, then we redirect to its "href" URL. We can do that by adding the following code to our /site/templates/_init.php file: if($page->id > 1) { // ensure that we are rendering from the 'href' URL // rather than native PW url, when href is populated $href = $page->get('href'); $requestUrl = isset($_SERVER['REQUEST_URI']) ? $_SERVER['REQUEST_URI'] : ''; if($href && $requestUrl && strpos($requestUrl, $href) === false) { // href value is not present in request URL, so redirect to it $session->redirect($config->urls->root . ltrim($href, '/')); } } Enforcing uniqueness and slashes in "href" values (optional) It's worthwhile to enforce our "href" values being consistent in having both a leading and trailing slash, as well as making sure they are always unique, so no two pages can have the same "href" value. To do that, I added this hook to my /site/ready.php file (/site/init.php would also work): /** * Ensure that any saved 'href' values are unique and have leading/trailing slashes * */ $wire->addHookAfter('Pages::savePageOrFieldReady', function(HookEvent $event) { $page = $event->arguments(0); /** @var Page $page */ $fieldName = $event->arguments(1); /** @var string $fieldName */ if($fieldName === 'href' || in_array('href', $page->getChanges())) { // the href field is being saved $href = $page->get('href'); if(!strlen($href)) return; // make sure value has leading and trailing slashes if(strpos($href, '/') === 0 && substr($href, -1) === '/') { // already has slashes } else { $href = '/' . trim($href, '/') . '/'; $page->set('href', $href); } // make sure that the 'href' value is unique $pages = $event->object; /** @var Pages $pages */ $p = $pages->get("id!=$page->id, href=$href"); if($p->id && !$p->isTrash()) { $page->set('href', ''); $page->error( "Path of '$href' is already in use by page $p->id “{$p->title}” - " . "Please enter a different “href” path and save again" ); } } }); That's all for this week. Thanks for reading, have a great weekend!

I've also been using this great module on another WP conversion project to do the same thing: https://processwire.com/modules/custom-paths/ I'm not 100% sure if it checks for uniqueness (probably) but since everything's being imported by a script I wrote I know they are unique for now.

New search engine on the block .... will their business model work ? yep.com ******************

From PW Weekly: This week we're happy to introduce a brand new module from Bernhard Baumrock, called RockLanguage. As the name suggests, RockLanguage is a tool for dealing with translations, and more specifically with translations related to ProcessWire modules. While ProcessWire natively ships with extensive language support, including the ability to ship module translations with the module itself, this does still require some manual work. That is exactly what RockLanguage aims to solve. Here's what the module translation process looks like with RockLanguage: Install RockLanguage on your site and (optionally) configure custom language code mapping via the module configuration screen. Add a directory for the language you'd like to include translations for within your own module's directory, e.g. /site/modules/MyModule/RockLanguage/FI/ for the Finnish language. Translate your module for said language via ProcessWire's translations manager. RockLanguage will automatically notice the update and duplicate the translation file from its original source to the directory you've just created. Now if you install this module to another site with the language folder included, and the site has RockLanguage installed and Finnish as one of its languages, the translation files for your module will be automatically synced with ProcessWire. What's nice about this workflow is that it takes some manual steps out of the equation, thus streamlining translation management. It's too early to say how widely this module will be adopted among public third party modules, but if you like the concept, you can easily start using it for your own modules right away. Download + Docs: baumrock.com/RockLanguage

https://www.autohaus-bendel.at/ Highlights/Features: New Logo and matching Backend using module AdminStyleRock Very good google- and performance-ratings thx to UIkit and ProCache Flexible content pages (using private module RockMatrix) Cars are fetched automatically from an austrian cardealer API and converted into ProcessWire pages with a custom filter page: https://www.autohaus-bendel.at/fahrzeuge/ Forms with honeypots and live validation (using private module RockForms based on nette forms) Web-Coach The client gets automated reminders if he does not add new content to the website. Thx @wbmnfktr Bendel Web Coach The last news entry was created 21 days ago. There must be something new that will interest your clients ;) > To the website > Create news entry Animated page transitions using barba.js Migrations and deployment using RockMigrations. Debugging using TracyDebugger. ?

I missed the XML sitemap generator that I used in a previous CMS so I built my own module to achieve the same functionality. This module outputs an XML sitemap of your site that is readable by Google Webmaster Tools etc. I've generally found that it reduces the time it takes for new sites and pages to be listed in search engines using one in combination with Webmaster Tools etc (since you're specifically telling the service that a new site/new pages exist) so thought I may as well create a module for it. The module ignores any hidden pages and their children, assuming that since you don't want these to be visible on the site then you don't want them to be found via search engines either. It also adds a field called sitemap_ignore that you can add to your templates and exclude specific pages on a per-page basis. Again, this assumes that you wish to ignore that page's children as well. The sitemap is accessible at yoursite.com/sitemap.xml - the module checks to see whether this URL has been called and outputs the sitemap, then does a hard exit before PW gets a chance to output a 404 page. If there's a more elegant way of doing this I'll happily change the code to suit. Feedback and suggestions welcome On a slightly different note, I wanted to call the file XMLSitemap originally so as to be clearer about what it does in the filename, but if you have a module that begins with more than one uppercase letter then a warning containing only the module name is displayed on the Modules page, so I changed it to Sitemap instead which is fine as the description still says what it does. File can be downloaded via GitHub here: https://github.com/N.../zipball/master

I've got a scenario where I use ProcessWire as a CRM for a client and they have several companies using this now, so it's important to be able to update things easily across all installations. A tricky one recently was working out how to manage translations via API to avoid going into several installations, enabling language support and translating the file manually (in this case they wanted to tweak a message in LRP's confirmation routine). On your master installation, do your translation as normal Export to CSV Upload the CSV file somewhere where your updater code can reference it - I created a folder in site/assets called "uploads" for some CRM-specific stuff, so I added another folder under there for ' Use code similar to the below in your updater script on other installations to first create the JSON file to be translated (it may not already exist) and then import your CSV file and tidy up afterwards - something like the below: <?php $this->modules->refresh(); // I've found this helps first, mostly for modules that have just been pushed to the Git repo but doesn't hurt to do it every time $this->modules->install('LanguageSupport'); // This wasn't yet installed $languages = $this->wire()->languages; $default = $languages->get('default'); // We only use the default language as it's only for the UK market $default->of(false); // It will complain without outputformatting switched off under some scenarios $translator = new LanguageTranslator($default); $translator->addFileToTranslate('site/modules/LoginRegisterPro/LoginRegisterProConfirm.php'); // This generates the JSON file that you would see in PW on the translations page $languages->importTranslationsFile($default, $this->config->paths->assets . 'uploads/imports/default-site.csv'); // This imports the CSV we exported from our master installation unlink($this->config->paths->assets . 'uploads/imports/default-site.csv'); // And we don't need the file afterwards I appreciate it's not a particularly in-depth tutorial, but may help someone out there with a similar need. Our CRM installations are all on an Amazon EC2 instance under separate subdomains, so separate databases too and there's a shell script I use to iterate all the installation folders, get the latest updates from the Github repo and then run an updater script that will do more complex updates like the above code or perhaps alter tables etc. This may get trickier in future if we shift to using something like Docker, or that may make life easier, but at the moment this is all nicely manageable the way it is.

Hi @MoritzLost that's interesting! I've recently fallen in love with latte and there I have the same issue. I didn't think of that until know as I've not been using it on a multilang site yet. Your approach with the table field is interesting, especially the fact that it creates rows automatically from code and populates a default value. Though I'm not sure if I really like that approach or not. I think if the default value is changed by the client and a developer looks into the code and tries to change the wording there this might lead to unexpected results as that change would not have any effect since the new value will not be written to the translations? On the other hand having the msgid has the benefit that if you change the wording in the initial language you don't lose the relation to the translation which would be the case in PW's internal system... There's always a pro and con ? Anyway.. it's not really a big issue for me since I'm always using custom page classes and I can easily add a method there that returns a translatable label, eg $page->buttonLabel()

I am still working on this one. It works great and is quite flexible now. I made a couple of improvements. Now the data to style the items is stored directly on the item with $page->meta(). This way, cloning pages and page reference fields work out of the box. Also there is no need for hidden fields anymore. The style panel now supports adding custom classes and assigning styles to them. These classes can be used globally on all pages (a css class is also a page). The style panel now supports selecting html tags to style tags globally across the whole site. I did a lot of refactoring to make the code cleaner and more efficient. There are still things where I am looking for feedback before releasing an open beta for this module: Sharing blocks and migrating pages I played with the core ProcessPagesExportImport and the migrator module and it seems that it's not exporting the meta() data, which is a bummer. So for now I might not be able to have a migration feature out of the box. Maybe someone else can take a look at this, but it seems to be too much hassle for me now. Importing/exporting the whole database with ProcessDatabaseBackups works as expected though. Also I did develop some block modules that you can install optionally, these will create the needed templates and fields for that block and give you a quick way to start working with PageGrid (without the need to create your own templates or install a site profile), more on that later. (Item pages are based on a block template). Structure of PageGrid items/pages I decided to always store the items under the admin page in a seperate parent, instead of making the parent customisable (like PageTable is doing). So there is a clear separation between pages for your site and item pages. PageGrid then creates a structure looking like this (makes it easy to access the pages): Admin – PageGrid Items – Home items (name: for-1213) – Item Editor – Item XY – Another Page items (name: for-1315) – Item Group – Item XY – Item XY – Item Editor The structure of these pages is an exact representation of the PageGrid items/pages (sorting, child/parent relation, hidden/published). While working with PageGrid via the UI this is not so important, but I think it will make interacting with the item pages via API more flexible and clean. Adding a page to any of the item pages via API or the backend also adds the page to PageGrid. If you delete a page containing a PageGrid field, the items also get deleted. You can also sort via API (or manually under the Admin page) and the changes are reflected for the items as well. The connection is via a name like for-1213, where 1213 is the ID of the page holding the PageGrid field (inspired by repeaters). I like this setup because it gives you a lot of options to access items via API and keeps the pagetree clean. Also it will make it easier to handle migrations for pages containing a PageGrid field in the future (just import/export the page and the item page container). One downside, though, is that the methods of PageTable like $page->pt->add($newpage) or $page->pt->remove($newpage) are no longer working (PageTable is using a PageArray that is saved as the field value). Also this will not work with multiple PagGrid fields on one page (A PageGrid field can however have a page reference field to another page containing a PageGrid field, as well as having nested pages, so I see no benefit of supporting multiple fields). You can add a page to the item parent like this: // Add an PageGrid item via API $itemParent = $pages->get("for-$page->id"); $itemParent->add($newpage) I am not decided if my approach is better or if I should just stick with the PageTable methods (keeping the items stored as PageArray inside the field). Since PageTable can only have one parent that you select in the field settings, all items will than be stored under the same parent, which is a bit messy. Not sure if any of that makes sense to you ? Any feedback is welcome. More info and website is coming soon!

Data is fetched via cron every night and then PW pages are created so that the client can enrich those pages with data that is not part of the API but is helpful on the website (eg a PDF sheet for the car or the point of contact): API-Fields are read-only. If a car is created by the cron and the point of contact is empty, the client gets a link to directly edit this page and select the POC ? If cars are sold, the cron automatically trashes those pages on the website. Ah... I forgot a nice detail! They have a custom branded PDF viewer:

More than a year ago I said I would try TailwindCSS... now it finally happened and... and the result is disastrous! Well... disastrous for my opinion back then and my thoughts about it until a few weeks back. I am quite surprised and happy with the result, to be honest. I did a rebuild of an ongoing side-project, with a full rebuild of the whole site structure all the way through to any breakpoints (CSS only, HTML was kept except additional classes). Yet some parts are still outside of TailwindCSS as those are really custom and are way better placed in an additional CSS file. Even though you can customize TailwindCSS to almost all of your needs but I didn't want to bother with it for now. The change in file size is quite dramatic... or even super awesome. However you want to see it. I went from... 59kb (the whole project) down to 17kb (structure - based on TailwindCSS CLI generated .css file) PLUS 5kb custom CSS for colors, custom properties PLUS 10kb custom CSS for icons, backgrounds, and fonts PLUS unknown size of additional kb within HTML due to those TailwindCSS classes Ignore those other files as those are helpers or part of other instances within that setup. Which is in total: 32kb for the full project, all pages, all icons, all fonts. Yet... the whole setup isn't ready for production right now as my setup only really works with 11ty as a middle-man for now and not within my PW workflows... but that's another story. There are probably even more things I could optimise in terms of fonts, icons and custom classes for colors. But for now... WOW... I always was against those utility frameworks and didn't like them at all. I am and always was a VanillaCSS Purist... didn't ever really like Bootstrap, UIKIT or anything like that, but TailwindCSS could and might have changed my mind. From now on... I might have to think and work different for some or another reason. Right now my test was built on a project I really knew in and out with all its details. Let's see how it works out in a real and new project. Results based on: TailwindCSS 3.1.2 (latest - as of now, 2022-06-14. all file sizes unzipped, without plugins like Typography)

So i tried the other way around and formatted a date using the IntlDateFormatter. This thread helped me find my way: https://stackoverflow.com/questions/71874030/local-language-without-strftime-deprecated-but-with-date So imagine we want to output "today" like this (formatted for german) Mittwoch 8 Juni 2022 This was the old way to to it: echo strftime('%A %e %B %Y', strtotime('now')); Since this will be deprecated in PHP 8.1 and removed in PHP 9 now we have to make use of this instead: $fmt = new IntlDateFormatter('de_DE', IntlDateFormatter::FULL, IntlDateFormatter::FULL ); $fmt->setPattern('EEEE d LLLL yyyy'); echo $fmt->format(strtotime("now")); The only thing that freaked my out about this was the weird formatting pattern. At first sight it makes no sense (to me), but the ICU documentation offers a big list of available options for formatting the date in any way you could imagine: https://unicode-org.github.io/icu/userguide/format_parse/datetime/ If you are looking for a "one liner solution" then this seems to be the right way (https://stackoverflow.com/questions/12038558/php-timestamp-into-datetime) echo IntlDateFormatter::formatObject( new DateTime('@' . strtotime("now")), "EEEE d LLLL yyyy", 'de_DE' ); EDIT: Beware that timezone settings get ignored when you work with timestamps. This might get you into some trouble. My solution was to format the timestamp into a date string: echo IntlDateFormatter::formatObject( new DateTime(date("d.m.Y", $timestamp)), "EEEE d LLLL yyyy", 'de_DE' );

Thanks Radek, this has been updated.