Controlling page URL structure

[Roope]

I remember that I read something about this earlier but can't find any info now when it's needed.

I would like to drop some parent page names from URL structure but don't know how to do this.

Tree is like this:

- Products

  - Vendor

    - Product 1

And sysytem default URL for product page:

/products/vendor/product-1/

How can I setup templates so that I could have product page URL:

/products/product-1/

Or simpy:

/product-1/

[Pete]

I would actually separate out each Vendor and each invidivual Product to be under their own tree like:

/vendors/vendorname/

/products/product-1/

You then just use a Page fieldtype to link products to relevant vendors: http://processwire.com/videos/page-fieldtype/ and then use code on the product name something like this to pull the vendor into the product page:

<?php
$productvendors = $pages->find("template=vendors, id=$page->vendors"); // Assuming you could have more than one vendor per product
$productvendor = $pages->find("template=vendors, id=" . $page->vendors->first()); // Or just one vendor

You can also do it the other way to view a vendor page and list all products they sell:

<?php
$vendorproducts = $pages->find("template=products, vendor=$page"); // Assumes your Page field is called vendor

[Roope]

Thanks Pete!

Actually vendor pages are not needed in front-end so I don't need any data from these.

We're building a site for our customer that holds product data from multiple vendors. We're planing to build some integration for product data import in scv-file.

Reason why I would like to have page tree structured like in the example is that it would be easier to locate products under each vendor since there can be thousands of items per each. I think it's too messy to mix all products from all vendors under same folder.

In MODX we're used to have setting for this in configuration. If URL paths is enabled, all pages use site tree like URL path structure. If disabled, it's just domain.com/pagename no matter where page sits in site tree. PW doesn't have this kind of feature?

What if we would use templates without template files as vendor folder? Then page wouldn't be visible in URL path?

We could still build inputs to admin area for csv import functions (using module) to this template am I right?

[SiNNuT]

Can i ask why exactly you want to omit certain url parts? Is it purely aesthetic or are there other reasons. Also, a product can only ever belong to 1 vendor? If not, you should choose Pete's suggestion.

The thread you where looking for is this one i think: http://processwire.com/talk/topic/1799-routes-and-rewriting-urls/

[Soma]

You can't hide path in between unless you use urls segments.

Pete is right, and it will result in url segments anyway.

Sure you can build any feature other CMS have. You can build that with url segments too.

I would build it the way Pete suggested. You argument that it's not nice to manage under one branch, I have to say I would then build a admin page that let's you list per vender or category, like a data table. Everything possible with ease. See for example my data table POC http://modules.processwire.com/modules/data-table/

[Roope]

Thanks guys! You have to give me some credit since love between me and PW is only couple of weeks old.

I think what happened here is that when I was looking first glance at PW admin I misunderstood URL segments. In MODX you can disable folder paths globally for the site and I thought URL segment would do same in PW for template basis. Shorty after I got URL segments right but still thought that I saw URL path option similar to MODX somewhere. Thanks for posting this article SiNNuT - haven't read it before and it clarified much about how to deal with URLs in PW.

I'm myself a big fan of short and simple URLs. I prefer much more URL like domain.com/drills than domain.com/products/powertools/drills for example. It's good to now I can achive this using URL segments but it needs some hard coding and extra work compared to build in feature (like described above).

With this project I'll do like Pete suggested and move vendors to seperate folder. All products can only belong to one vendor. Also product data from vendors come in different kind of CSVs. Info about vendor is something we don't want end-users to see. That's why first plan was to put products under vendor folders (to ease crawling for specific vendor data) and to put CSV import functionalities to vendor pages (that would be only visible for logged users).

[Soma]

Sure you have to do some extra efforts but it's not that hard as it might first seems coming from others systems. No problem.

But this way you are in full control and can change it the way you like and are not restricted to some built in feature.

[Roope]

What's the right way to build navigation links when using shorter URLs by URL segments? Page url still has folders in it so do I have to code links manually? Same with links inside richtext editor.

EDIT:

Or meta canonical tag: this will be pointing to uncorrect address if I make shorter URLs by URL segments:

<link href="<?php echo $page->httpUrl; ?>" rel="canonical" />

[Soma]

Do you really think "domain.com/drills" is a good strategy? Have all pages on the root? You miss a lot of if you reduce it that way. There's no way to surely determine a single page name, because there can be multiple of it, and you can't control it anymore. I'd consider this for shortcut urls you use in print or something, but not as a general approach to the site on all ends.

At least have a category domain.com/powertools/drills/ that gives you powertools as a category.

There's maybe too many ways and scenarios with build a urlSegment system, and I would advise to use it only where really needed.

How you approach it also depends also a lot how you structure your page tree with what templates.

To construct some category/product urls segments are great. Where category is itself a page you use with a page field to categorize product. Then the category pages with url segments enabled and list products that belong to that category, or if url segment present show a detail page. That's how I use url segments to virtualize a structure that is different from the admin. But leaving the real url intact would sure make sense to use canonical urls or redirects to solve it.

You can alter the url path PW generates, so a hook into page::path or page::url would do the trick to strip out segments.

Once you have that url hook in place it would also effect the Wysiwyg urls generated. But this comes maybe with some overhead added to what might get called thousands of times, though not noticeable unless you go wild. 

You can also have the template file for those pages redirect to a shorter url and use that template only to construct it.

[Soma]

As example to "rewrite" the url of certain pages, this is possible within the template code:

function segmentUrl(HookEvent $event){
    $url = $event->return; // requested url
    $pages = wire("pages");
    $page = $pages->get($url);
    $segment = $pages->get(1001); // get segment we want to strip
    // for example only for certain pages under a certain section
    if($page->template == "basic-page" && $page->parents->has($segment)){
        // set event return the stripped new url; /about/childpage/ becomes /childpage/
        $event->return = str_replace(rtrim($segment->url,"/"),'',$url); 
    }
}

$wire->addHookAfter('page::path', null, 'segmentUrl');
 

Or similar as in a autoload module (HelloWorld.module) it would be almost same code, just

$this->addHookAfter("page::path", $this, 'segmentUrl');

[Roope]

domain.com/drills was there just for example - I'm not saying it would be best policy or anything. Now I know how to achieve this in PW way if I choose to do so.

Thanks alot Soma!

[Soma]

The above example already has lots of overhead so the more optimized way would be to avoid as much get or find and template stuff and do something like this:

function segmentUrl(HookEvent $event){
    $url = $event->return; // requested url
    $segment = "/about/";
    if(strpos($url,$segment) == 0){
        $event->return = str_replace(rtrim($segment,'/'),'',$url);
    }
}
$wire->addHookAfter('page::path', null, 'segmentUrl');

Will be a lot faster.

[fenton]

Hi Soma,

function segmentUrl(HookEvent $event){
    $url = $event->return; // requested url
    $segment = "/about/";
    if(strpos($url,$segment) == 0){
        $event->return = str_replace(rtrim($segment,'/'),'',$url);
    }
}
$wire->addHookAfter('page::path', null, 'segmentUrl');
 

is this a working example? I tried placing it in my template (url segments are enabled) and changed the segment part to the one I want to strip from my url

cheers & thanks, j

[ryan]

I think that the "page::path" needs to be "Page::path" (uppercase first character). Also, where you place this will matter. You want hooks to be attached before any other code executes. I like to use the $config->prependTemplateFile = 'init.php'; in my /site/config.php, and then setup hooks or any common functions I want in /site/templates/init.php.

[fenton]

thanks Ryan! will try it out and report back

*edit: yes, it works