The context
I'm working on an AEM 6 project that uses Sightly as the templating language. I'm facing a use case, in which I want to show or hide certain parts of the markup depending on the presence of a Sling selector.
For example, a request to /content/my-project/my-page.html should yield a basic page view and when a request is made to /content/my-project/my-page.ubermode.html, Sling should return the same content represented by a slightly different HTML document.
According to the Sling Cheat Sheet, it should be possible to use a different script.
I managed to implement this in a component by placing two Sightly scripts, mycomponent.html and ubermode.html (named after the selector)
/apps/(...)/mycomponent
|- .content.xml
|- _cq_editConfig.xml
|- dialog.xml
|- mycomponent.html
|- ubermode.html
This does require some code duplication when it comes to the HTML structure but it works fine.
However, in this particular case, I need to do the same think at the renderer level (let's call this myapp/core/renderers/fancyPageRenderer). What's more, the renderer has a different renderer as its sling:resourceSuperType (let's call this parent renderer myapp/core/renderers/genericPageRenderer) and relies on a moderately complex series of includes (data-sly-include).
In fancyPageRenderer, I override one of the scripts initially defined and included in genericPageRenderer. This is the part that I'd like to be different when the ubermode selector is used. Let's call this script mainColumn.html
I've tried different naming conventions to match the selector but none of them worked in a satisfactory way.
This was the initial structure
/apps/(...)/renderers/fancyPageRenderer
|- .content.xml
|- mainColumn.html //this overrides a script included by a parent renderer
Here's what I tried:
/apps/(...)/renderers/fancyPageRenderer
|- .content.xml
|- mainColumn.uber.html
|- mainColumn.html
This simply didn't work and mainColumn.html would be included every time.
/apps/(...)/renderers/fancyPageRenderer
|- .content.xml
|- uber.html
|- mainColumn.html
This caused the uber.html script to be used but the resulting page did not contain any markup defined in the other scripts included in the genericPageRenderer
I imagine I could just copy all the relevant scripts and includes to fancyPageRenderer but that would result in massive and completely unacceptable code duplication.
I also know that it's possible to manually add, remove or replace selectors using data-sly-resource or just have the original selectors used but in my case, it's data-sly-include and not data-sly-resource that is widely used.
Is there an elegant way to work around this problem?
Eventually, I gave up on using script naming conventions to solve this problem and exposed a very simple Sling Model in the Sightly script of my renderer.
Here's the current structure of fancyPageRenderer (which didn't change from the original):
/apps/(...)/renderers/fancyPageRenderer
|- .content.xml
|- mainColumn.html //this overrides a script included by the parent renderer
Here's what I used in the mainColumn.html Sightly script:
<div class="fancy main-column" data-sly-use.uberMode="com.foo.bar.myapp.fancy.UberMode">
<div data-sly-test="uberMode.enabled" >Uber-mode-only-content</div>
<!-- Lots of markup here -->
<div data-sly-test="!uberMode.enabled" >Explicitly non-uber-mode content</div>
<div>Common content (but some uber-mode-dependend, nested divs as well, rendered the same way as above)</div>
</div>
and the underlying Sling Model, UberMode
#Model(adaptables = SlingHttpServletRequest.class)
public class UberMode {
#Inject
SlingHttpServletRequest request;
private boolean enabled = false;
#PostConstruct
public void postConstruct() {
if (request != null) {
List<String> selectors = Arrays.asList(request.getRequestPathInfo().getSelectors());
enabled = selectors.contains("ubermode");
}
}
public boolean isEnabled() {
return enabled;
}
}
This allows me to avoid code duplication in Sightly and makes the selector-based logic unit-testable. Also, relying on naming conventions would get really tricky in a situation when I'd require more than one selector to decide what to render. Adding support for another relevant selector to this class would be quite straightforward in comparison.
It also leaves me a lot of refactoring options. I could switch from using a selector to a query parameter or a header and only write a couple lines of code without even touching the Sigthly script that's effectively the client code of my class.
Add a file ubermode.html which will be called if there's a Sling Selector ubermode present. To avoid code duplocation, extract the common parts (header.html, footer.html, etc.) and include them where needed.
For the case with mainColumn.html, you could try putting it into a sub-directory named ubermode (/ubermode/mainColumn.html). That's another way to catch a selector.
Related
Need: A dynamic template loaded from a DB
I need to present a JSON object via "template plugins" that will come from a database (defined outside the application itself). The plugin will be applied only if a plugin exists.
For this example, let's assume I have this object of type "Reservation" that contains a sub-object of type "Flight":
{
"id": "ABC-XYZ",
"reservationDate": "2020-09-23",
"state": "paid",
"flight":
{
"origin": "BCN",
"destination": "MAD",
"airline": "VY"
}
}
Rules
The controller will call a page template passing multiple objects.
The page template will "use" or "include" (see later more info) an object.html.twig and will display it.
The object.html.twig will do this:
If there's not any known plugin able to handle this type of object, the template will display a <pre> with the object converted to YAML. This will be mainly act as a "default" plugin or "fallback" plugin.
If there's a known plugin able to handle the object, the object will be sent to the plugin (which is not anything else than another twig template).
In this case, the plugin should be able to separate "parts" of "interpretable" results to make them "nice" and leave the rest into an object that will be, in turn, displayed again with the original "default" plugin.
If there are parts of the object that are in turn interpretable, they will be in turn passed to other plugins.
Example "desired" outputs
Sample 1. No plugin available
Sample 2. Plugin 'reservation' available. Interprets the state in green. Also removes the redundant data of the ID
Sample 3. Same than 2 but also plugin 'flight' available, able to process the flight block. Formats the texts and makes a lookup of the airline full-name
Sample 4. Plugin 'flight' available, able to process the flight, that in turn knows that the origin and the destination are "airports" and passes them into the corresponding nested 'airport' plugin, because they are "reusable" objects not only in the "flight" plugin but also in many other places in the application, so they are defined appart
What I have already explored
I know that doing it via 'include' it could work. But let's take a look at the differences between 'use' and 'include':
The major difference between include and use in twig is that:
When you include a template, it's direct HTML 'inserted there' where you can use the {{ }} operator for printing and {% %} for control flow (set variables, ifs, fors, etc.). The renderer will process it straight forward. But defining a new block via {% block myNiceBlock %} is forbidden.
When you 'use' a template, it's pre-loaded, and blocks are permitted. There's no rendering of the included block. Then, from the caller, you use a {{ block( 'whatever' ) }} to tell the renderer to go and render that specific block.
The 'include' is more rudimentary. The use allows horizontal reusing and allows itself to auto-organize itself with other sub-blocks called by the parent block, all in one single file.
For example, in the airports example, if there are N images, in a include you should put the wrapper HTML directly in the file, do a loop and inside the loop write the inner HTML.
Instead in the use approach you'd do an airport block which in turn loops over the images and just calls the block airportImage which is defined in another block in the same file, thus facilitating clean-coding.
Requirement
The application should not be re-deployed when new plugins are created. They must be loaded from a DB or any other dynamic system, as the plugins will be written by "users of the application" as they need it. No deploy allowed.
The plugins should be written in terms of a "block-able" twig template, so need to be 'use'-able.
Question
Discovering "which" plugin to call is not a problem. Assume that whoever (the controller, the twig itself, whoever really) can discover for this example that a "reservation" plugin exists somewhere. How can I use it from the page? When the reservation is rendering it "asks if a flight plugin" is available. If not, all to the YAML. If it is, how can then dynamically tell the reservation to use the flight?
In short: How do I force a template to dynamically use (not include) templates that, in turn, comes from the database (not from fixed files)?
NOTE: If this info is useful: I'm using Symfony 5 with webpack.
Thanks!
I'm trying to build an experience fragment (XF) template in AEM 6.5. We have some custom clientlibs that I want to include when designers are authoring the experience fragment but I don't want to include when the experience fragment is injected via Adobe Target as the clientlibs will already be included on the base page template.
I want the clientlibs on the XF template so components render properly while designing. I've tried building a new page component based on /libs/cq/experience-fragments/components/xfpage and then checking the runmode for author or publish and using the result of that in a data-sly-test to conditionally include them. But I think because the Export to Target option happens on Author, it's including the scripts in the html output when it's exported to Target.
How do I conditionally include clientlibs during authoring of an XF, but not include them when the experience fragment is exported to target and added to a page from there?
There are couple of things you need to do in order to achieve this.
AEM offers the Link Rewriter Provider Interface. This is a ConsumerType interface that you can implement in your bundles, as a service.
It bypasses the modifications AEM performs on internal links of an HTML offer as rendered from an Experience Fragment. This interface allows you to customize the process of rewriting internal HTML links to align with your business needs.
To use the interface you first need to create a bundle containing a new service component that implements the Link Rewriter Provider interface.
This service will be used to plug into the Experience Fragment Export to Target rewriting in order to have access to the various links.
import com.adobe.cq.xf.ExperienceFragmentLinkRewriterProvider;
import com.adobe.cq.xf.ExperienceFragmentVariation;
import org.osgi.service.component.annotations.Service;
import org.osgi.service.component.annotations.Component;
#Component
#Service
public class GeneralLinkRewriter implements ExperienceFragmentLinkRewriterProvider {
#Override
public String rewriteLink(String link, String tag, String attribute) {
return null;
}
#Override
public boolean shouldRewrite(ExperienceFragmentVariation experienceFragment) {
return false;
}
#Override
public int getPriority() {
return 0;
}
}
shouldRewrite
You need to indicate to the system whether it needs to rewrite the links when a call is made for Export to Target on a certain Experience Fragment variation. You do this by implementing the method:
shouldRewrite(ExperienceFragmentVariation experienceFragment);
For example:
#Override
public boolean shouldRewrite(ExperienceFragmentVariation experienceFragment) {
return experienceFragment.getPath().equals("/content/experience-fragment/master");
}
This method receives, as a parameter, the Experience Fragment Variation that the Export to Target system is currently rewriting.
In the example above, we would like to rewrite:
links present in src
href attributes only
for a specific Experience Fragment:
/content/experience-fragment/
Any other Experience Fragments that pass through the Export to Target system are ignored and not affected by changes implemented in this Service.
rewriteLink
For the Experience Fragment variation impacted by the rewriting process, it will then proceed to let the service handle the link rewriting. Everytime a link is encountered in the internal HTML, the following method is invoked:
rewriteLink(String link, String tag, String attribute)
As input, the method receives the parameters:
link
The String representation of the link that is currently being processed. This is usually a relative URL pointing to the resource on the author instance.
tag
The name of the HTML element that is currently being processed.
attribute
The exact attribute name.
If, for example, the Export to Target system is currently processing this element, you can define CSSInclude as:
<link rel="stylesheet" href="/etc.clientlibs/foundation/clientlibs/main.css" type="text/css">
The call to the rewriteLink() method is done using these parameters:
rewriteLink(link="/etc.clientlibs/foundation/clientlibs/main.css", tag="link", attribute="href" )
When you create the service you can make decisions based on the given input, and then rewrite the link accordingly.
When generating forms with the form module the corresponding yaml files get stored in fileadmin/user_upload.
Now I want to integrate those yaml files into my sitepackage and thus into my CVS. Where is the correct place for them? In the example extension they are stored in Resources/... while I would think they have to go into Configuration/Yaml
And how do I configure the form extension to search them in that place?
While it's basically a matter of taste where exactly one saves his form definitions, I try to separate form configuration and form definitions.
From the official documentation:
[...] the form configuration allows you to define:
which form elements, finishers, and validators are available,
how those objects are pre-configured,
how those objects will be displayed within the frontend and backend.
In contrast, the form definition describes the specific form,
including
all form elements and their corresponding validators,
the order of the form elements within the form, and
the finishers which are fired as soon as the form has been submitted.
Furthermore, it defines the concrete values of each property of the mentioned aspects.
So, for more clarity I save all form configuration in a sitepackage under Configuration/Yaml/ and the form definitions under Resources/Private/Forms, neighbouring the templates.
I wrote a full tutorial how to use custom templates with EXT:form, which also includes the answers to your question.
In short:
Register YAML configuration with TypoScript in your extension root folder as ext_typoscript_setup.txt (as recommended1)
plugin.tx_form.settings.yamlConfigurations {
100 = EXT:my_extension/Configuration/Yaml/CustomFormSetup.yaml
}
module.tx_form.settings.yamlConfigurations {
100 = EXT:my_extension/Configuration/Yaml/CustomFormSetup.yaml
}
CustomFormSetup.yaml – setting up a new storage path
TYPO3:
CMS:
Form:
persistenceManager:
allowedExtensionPaths:
10: EXT:my_extension/Resources/Private/Forms/
allowSaveToExtensionPaths: true
allowDeleteFromExtensionPaths: true
1TypoScript inside an ext_typoscript_setup.txt is automatically loaded in both frontend and backend of your TYPO3 installation directly after installing your extension. This differs from other TypoScript files, which have to be included manually, e.g. as static templates. See official Form Framework documentation.
I'd suggest Resources/Private/Forms for your form definitions. The form extension clarifies how to register additional form definition paths.
Let's say, I have a template A and sling:resourceType is /apps/myproject/components/basePage. In this component I've body.html and header.html and footer.html script included through slightly in body.html.
Now I am creating another template B, and sling:resourceType is /apps/myproject/components/compB and sling:resourceSuperType of compB is /apps/myproject/components/basePage.
In /apps/myproject/components/compB I have added content.html and selector.html
If I create a page (mytest.html) of type template B, then header and footer script is included correctly but when I hit this mytest.selector.html then header and footer script is not included. I want template B will have two different view based on selector.
Please let me know where I am missing.
I believe you are trying to include multiple scripts within same template to achieve different views. This is correct approach todo in AEM. But missing part is the moment you create the second script (selector.html in this case), it becomes another template and you need to code to include your entire page scripts into this script as well.
When you override scripts from /libs/wcm/foundation/components/page component, they ll work fine when your custom script names matches to parent component. For example your body.html will override /libs/wcm/foundation/components/page/body.html and page will render how it is coded. When you create selector.html it becomes independent script as there is no /libs/wcm/foundation/components/page/selector.html.
You need to define all behavior (to include header, footer script etc) explicitly against your custom script. In this case you need include header/footer scripts explicitly into your selector.html
Using a selector means that you're using some special implementation of your component. For instance, your component may have several charts and you want to encapsulate those in your selectors and use it through AJAX from browser, and reuse those selectors in your main component as well.
Currently, you're trying to achieve is to use your header and footer to another component which breaks the encapsulation rule. Rather do this, take out your header.html and footer.html and make those individual components and you it in your basePage as well as your child pages.
See the snippet below:
<div data-sly-resource="${'header' # resourceType='/apps/myproject/components/header'}">
<p>Your body and anything you want to put here</p>
<div data-sly-resource="${'footer' # resourceType='/apps/myproject/components/footer'}">
This way, you can reuse your headers whereever you want even in your selectors.
I am new to Play Framework and still trying to wrap my head around some things with the new Scala template engine.
Let's say I have the following package structure:
app/
app/controllers/Items.scala
app/models/Item.scala
app/views/layouts/page.scala.html
app/views/item/show.scala.html
app/views/item/details.scala.html //partial
And this is my item/show template:
#(item: Item, form: Form[Item])(implicit flash: Flash)
#layout.page() {
#*want to include details partial, wont work due to item param*#
#item.details(item)
}
Since including another template (e.g. including item/details above) is the exact same syntax as accessing a template parameter (e.g. item above), obviously this existing naming convention won't work without something changing.
I know I can rename my "app.views.item" package to "app.views.items", and rely on singular/plural forms to differentiate the view from the param name, but this does not seem like a very straightforward solution. Also what if I really want the parameter name to be the same as the view package?
One idea I have is to prepend all my views with an extra top level package:
app/views/views/item/details.scala.html
So the include syntax would be #views.item.details(), but again this is obviously a hack.
What is a good way to avoid this issue? How can I better organize my code to avoid such naming collisions?
Most other template engines use operations like "include" or "render" to specify a partial include. I don't mean to offend anyone here, but is the Play Scala template engine syntax so terse that it actually dictates the organization of code?
3 solutions:
First
Typpicaly for partial templates you should use tags as described in the docs, where app/views/tags folder is a base:
file: app/views/tags/product.scala.html
in the templates (no initial import required in the parent view full syntax will allow you to avoid name-clash: #tags.packageName.tagName()):
<div id="container">
#tags.product(item)
</div>
Of course in your case you can also use packages in the base folder
file: app/views/tags/item/product.scala.html
<div id="container">
#tags.item.product(item)
</div>
I'm pretty sure that'll solve your problem.
Second
To avoid clash without changing package's name you can just rename the item in your view, also I recommend do not use a form name for the Form[T] as it can conflict with helpers:
#(existingItem: Item, existingItemForm: Form[Item])(implicit flash: Flash)
#layout.page() {
#item.details(existingItem)
}
Third
If you'll fill your Form[Item] before passing to the view with given Item object, you don't need to pass both, as most probably you can get data from the form:
#(itemForm: Form[Item])(implicit flash: Flash)
#layout.page() {
<div>Name of item is: #itemForm("name").value (this is a replacemnet for ##existingItem.name </div>
#item.details(itemForm)
}
Of course in you product.scala.html you'll need to change the #(item: Item) param to #(itemForm: Form[Item])