"jQuery for React AST"? - babeljs

Is there such a thing as a "jQuery for React AST"? That like jQuery allows for elegant search, traversal, creation, mutation of an AST that contains things like JSXNode, etc? I saw that acorn has some basic traversal stuff, but it isn't super usable for repeatedly doing reorders, insertions, wrapping a component in {flag && } to conditionally render, etc, etc. I'm not even sure how to google for this except "jquery for AST" which, uh yeah, didn't work.

Subsecond
You can use Subsecond for this purpose, here is how it will looks like:
Input:
<h1>hello world</h1>
Output:
<h2>hello world</h2>
Transformation:
S('JSXIdentifier').each((node) => {
node.text('h2');
})
πŸ“œ
Here is playground.
🐊Putout
Also there is a way to do similar things using declarative approach with help of 🐊Putout I'm working on:
Change tag
Input:
<h1>hello world</h1>
Output:
<h2>hello world</h2>
Transformation:
export const replace = () => ({
'<h1>__a</h1>': '<h2>__a</h2>',
});
πŸŠπŸ“œ
Here is playground.
Change Attribute
You can also change an attribute className to class:
Input:
const div = <div className="abc">{x}</div>
Output:
const div = <div class="abc">{x}</div>
Transformation:
export const replace = () => ({
'<div className="__a">__jsx_children</div>': '<div class="__a">__jsx_children</div>',
});
πŸŠπŸ“œ
Here is playground

Related

Array of JSX Elements not rendering on second loop

I have a Solid.js code that looks like this:
import { render, For } from "solid-js/web";
const Text = () => <div style="color: red">Example</div>;
const App = () => {
const elements = [<Text/>, <Text/>, <Text/>];
return (
<div>
<div>First For Each</div>
<For each={elements}>{(E) => E}</For>
<div>Second For Each</div>
<For each={elements}>{(E) => E}</For>
</div>
);
}
render(() => <App />, document.getElementById("app")!);
But for some reason Solid.js only renders the second <For>:
And when I change the elements to:
const elements = [() => <Text/>, () => <Text/>, () => <Text/>];
it renders twice (also works fine if I change the elements value to primitive value like int or string. Can someone explain to me why Solid.js behaves this way?
Playground Example
Writing <Text/> executes the Text component which returns an actual DOM node. And dom nodes can be inserted only in one place in the DOM.
The attempt with wrapping the component execution with functions works because you'll get a different element instance every time you execute that function. You're basically creating an array of components vs an array of HTML elements.
Here is a similar Github issue: https://github.com/solidjs/solid/issues/899

How can I make StencilJS component to render without component tag itself with TSX?

While I understand this is probably a terrible practice, I need to build StencilJS component such that inside render(), I don't want to render component tag itself due to already existing style guide and it expect DOM to be constructed in certain way. Here is what I'm trying to achieve - component code (from HTML or within another component):
<tab-header-list>
<tab-header label="tab 1"></tab-header>
<tab-header label="tab 2"></tab-header>
</tab-header-list>
when rendered, I want generated DOM to be something like:
<tab-header-list>
<ul>
<li>tab 1</li>
<li>tab 2</li>
</ul>
</tab-header-list>
so inside tab-header-list render() function, I'm doing
return (
<ul>
<slot/>
</ul>
);
and I can do this inside tab-header render() function
#Element() el: HTMLElement;
#Prop() label: string;
render() {
this.el.outerHTML = `<li>${this.label}</li>`;
}
to get what I want but how can I do this with TSX? (for simplicity sake, above code is really simple but what I really need to build is lot more complicated li tag with events etc so I would like to use TSX)
Tried to store DOM to variable but I'm not sure how I can assign it as this.el (outerHTML seem to be only way I can come up with, but I feel there must be better way)
#Element() el: HTMLElement;
#Prop() label: string;
render() {
var tabheaderDOM = (<li>{this.label}</li>);
// how can I assign above DOM to this.el somehow?
//this.el.outerHTML = ?
}
I appreciate any help I can get - thanks in advance for your time!
Unfortunately, you can't use custom elements without tags, but there is a workaround for it:
You can use Host element as reference to the result tag.
render () {
return (
<Host>....</Host>
)
}
Then in your stylesheet you can set the display property for it:
:host {
display: contents;
}
display: contents causes an element's children to appear as if they were direct children of the element's parent, ignoring the element itself
Beware: it doesn't work in IE, opera mini... https://caniuse.com/#feat=css-display-contents
UPD:
If you are not using the shadowDOM then you need to replace :host by the tag name like:
tab-header {
display: contents;
}
Functional components might be able to help you achieve this. They are merely syntactic sugar for a function that returns a TSX element, so they are completely different to normal Stencil components. The main difference is that they don't compile to web components, and therefore only work within TSX. But they also don't result in an extra DOM node because they simply return the template that the function returns.
Let's take your example:
#Element() el: HTMLElement;
#Prop() label: string;
render() {
this.el.outerHTML = `<li>${this.label}</li>`;
}
you could write it as a functional component:
import { FunctionalComponent } from '#stencil/core';
interface ListItemProps {
label: string;
}
export const ListItem: FunctionalComponent<ListItemProps> = ({ label }) => (
<li>{label}</li>
);
and then you can use it like
import { ListItem } from './ListItem';
#Component({ tag: 'my-comp' })
export class MyComp {
render() {
return (
<ul>
<ListItem label="tab 1" />
<ListItem label="tab 2" />
</ul>
);
}
}
Which will render as
<ul>
<li>tab 1</li>
<li>tab 2</li>
</ul>
Instead of a label prop you could also write your functional component to accept the label as a child instead:
export const ListItem: FunctionalComponent = (_, children) => (
<li>{children}</li>
);
and use it like
<ListItem>tab 1</ListItem>
BTW Host is actually a functional component. To find out more about functional components (and there limitations), see https://stenciljs.com/docs/functional-components.

Is there a way/workaround to have the slot principle in hyperHTML without using Shadow DOM?

I like the simplicity of hyperHtml and lit-html that use 'Tagged Template Literals' to only update the 'variable parts' of the template. Simple javascript and no need for virtual DOM code and the recommended immutable state.
I would like to try using custom elements with hyperHtml as simple as possible
with support of the <slot/> principle in the templates, but without Shadow DOM. If I understand it right, slots are only possible with Shadow DOM?
Is there a way or workaround to have the <slot/> principle in hyperHTML without using Shadow DOM?
<my-popup>
<h1>Title</h1>
<my-button>Close<my-button>
</my-popup>
Although there are benefits, some reasons I prefer not to use Shadow DOM:
I want to see if I can convert my existing SPA: all required CSS styling lives now in SASS files and is compiled to 1 CSS file. Using global CSS inside Shadow DOM components is not easily possible and I prefer not to unravel the SASS (now)
Shadow DOM has some performance cost
I don't want the large Shadow DOM polyfill to have slots (webcomponents-lite.js: 84KB - unminified)
Let me start describing what are slots and what problem these solve.
Just Parked Data
Having slots in your layout is the HTML attempt to let you park some data within the layout, and address it later on through JavaScript.
You don't even need Shadow DOM to use slots, you just need a template with named slots that will put values in place.
<user-data>
<img src="..." slot="avatar">
<span slot="nick-name">...</span>
<span slot="full-name">...</span>
</user-data>
Can you spot the difference between that component and the following JavaScript ?
const userData = {
avatar: '...',
nickName: '...',
fullName: '...'
};
In other words, with a function like the following one we can already convert slots into useful data addressed by properties.
function slotsAsData(parent) {
const data = {};
parent.querySelectorAll('[slot]').forEach(el => {
// convert 'nick-name' into 'nickName' for easy JS access
// set the *DOM node* as data property value
data[el.getAttribute('slot').replace(
/-(\w)/g,
($0, $1) => $1.toUpperCase())
] = el; // <- this is a DOM node, not a string ;-)
});
return data;
}
Slots as hyperHTML interpolations
Now that we have a way to address slots, all we need is a way to place these inside our layout.
Theoretically, we don't need Custom Elements to make it possible.
document.querySelectorAll('user-data').forEach(el => {
// retrieve slots as data
const data = slotsAsData(el);
// place data within a more complex template
hyperHTML.bind(el)`
<div class="user">
<div class="avatar">
${data.avatar}
</div>
${data.nickName}
${data.fullName}
</div>`;
});
However, if we'd like to use Shadow DOM to keep styles and node safe from undesired page / 3rd parts pollution, we can do it as shown in this Code Pen example based on Custom Elements.
As you can see, the only needed API is the attachShadow one and there is a super lightweight polyfill for just that that weights 1.6K min-zipped.
Last, but not least, you could use slots inside hyperHTML template literals and let the browser do the transformation, but that would need heavier polyfills and I would not recommend it in production, specially when there are better and lighter alternatives as shown in here.
I hope this answer helped you.
I have a similar approach, i created a base element (from HyperElement) that check the children elements inside a custom element in the constructor, if the element doesn't have a slot attribute im just sending them to default slot
import hyperHTML from 'hyperhtml/esm';
class HbsBase extends HyperElement {
constructor(self) {
self = super(self);
self._checkSlots();
}
_checkSlots() {
const slots = this.children;
this.slots = {
default: []
};
if (slots.length > 0) {
[...slots].map((slot) => {
const to = slot.getAttribute ? slot.getAttribute('slot') : null;
if (!to) {
this.slots.default.push(slot);
} else {
this.slots[to] = slot;
}
})
}
}
}
custom element, im using a custom rollup plugin to load the templates
import template from './customElement.hyper.html';
class CustomElement extends HbsBase {
render() {
template(this.html, this, hyperHTML);
}
}
Then on the template customElement.hyper.html
<div>
${model.slots.body}
</div>
Using the element
<custom-element>
<div slot="body">
<div class="row">
<div class="col-sm-6">
<label for="" class="">Name</label>
<p>
${model.firstName} ${model.middleInitial} ${model.lastName}
</p>
</div>
</div>
...
</div>
</custom-element>
Slots without shadow dom are supported by multiple utilities and frameworks.
Stencil enables using without shadow DOM enabled. slotted-element gives support without framework.

Use mustache for templating smaller parts for use multiple times in a form

Okay. So I hope this is specific enough.
I'm, kind of new to mustache, but see it has great potential, so why not use it.
I'm making a quite big form, and want to have the form built with mustache. So i have started to make the form in mustache and, then i realized i want to template the form-elements. One template for how i want every narrow input, wide input, select etc. to look like. because now i'm repeating myself.
My template and partials are provided through $.ajax get, where the main form template are defined as a mustache file, with html content, and the partials are defined as a mustache file with every template inside -tags.
Variables for mustache to use. This object is somewhat subject for change.
var jsonForm = {
oneInputField: {
value:'put your title here',
rule_set: {
required: {
strName: 'required',
strErrorMsg: 'error message'
}
}
},
oneSelect: {
options: [
{value: '- Pick one -', helper: 'helper', select_options: {disable_search: true}},
{value: 'option1', selected: true},
{value: 'option2'},
{value: 'option3'},
{value: 'option4'}
],
rule_set: {
required: {
strName: 'required',
strErrorMsg: 'error message'
}
}
}
};
How i fetch the data
$.ajax({
url: 'myForm.mustache',
type: 'get',
success: function(template) {
$.ajax({
url: 'myFormElements.mustache',
type: 'get',
success: function(partials) {
var $html = Mustache.render(template, jsonForm.fields, partials);
$('div#formContent').html($html);
$('div#formContent select').chosen();
if (jsonForm.fields.title.length > 1) {
$('div#header').html(jsonForm.fields.title);
}
}
});
}
});
Partials. Would actually like to have it all in separate files, but it doesn't seem to be possible without making a ton of ajax calls, so I keep with my current two mustache templates.
<script type="text/html" id="inputNarrow">
<label>{{label}}:</label><input value="{{value}}">
{{#rule_set.required}}<div class="required">*</div>{{/rule_set.required}}
{{^rule_set.required}}<div class="not-required"> </div>{{/rule_set.required}}
<div class="clearfix"></div>
</script>
and my form-template
<div id="formContainer" class="border-box">
<div class="group-box">
<p>Fields marked with <span class="required">*</span> are required to complete the form</p>
</div>
<div class="group-box">
<h2>Partial element test</h2>
<div class="form-container">
{{>partials.inputNarrow}} <-- I have to be able to specify what data I want to enter here.
</div>
<div class="form-container">
{{>partials.selectNarrow}} <-- I have to be able to specify what data I want to enter here.
</div>
</div>
So my question is is it possible for a bigger one big unique mustache-template to use "element"-templates (also in mustache), for rendering?
I am not very open to adding additional libraries to my project, like ICanHaz or similar, since this is for work.
Really sorry if this question is answered before, but I couldn't find it
I found an alternative way of doing what I tried to do.
I'm not sure if it's any good, but it does the job.
I used the possibility of making functions inside the object sent to the main-tempalte "form-template".
wich looks like this:
var smallRender = function(url, containerId, objData) {
$.ajax({
url: url,
type: 'get',
success: function(template) {
var $html = Mustache.render(template, objData);
var $objContainer = $('div#' + containerId);
$objContainer.append($html);
}
});
};
the actual object
oneInputField: {
value:'put your title here',
rule_set: {
required: {
strName: 'required',
strErrorMsg: 'error message'
}
},
getRendered: function() { // I would like to cache this somehow
smallRender(templateUrl.input.narrow, jsonForm.fields.title.containerId, {
label: jsonForm.labels.title,
field: jsonForm.fields.title,
required: isRequired(jsonForm.fields.title)
});
}
},
So basicly i send an ajax call to get the main template, and for each element in the form I want to present, i call on the getRendered in main-tamplate mustache, and it basicly renders the content with the smaller template.
tempalte url is basicly a collections with the urls to all the templates. so if I specify templateUrl.input.narrow i will get the narrow input element and so on.
Not sure as I mentioned if this is the "correct" way of doing it, or how good it will scale in a big environment, but it works.
I will mark my this as the correct answer (if I'm even allowed), since there are no other answers that solves my problem.

AngularJs Directive: Using TemplateURL. Replace element. Add form input. Getting form.input.$error object

Not sure if this is possible but I'm trying, and keep coming up short.
http://plnkr.co/edit/Gcvm0X?p=info
I want a 'E' (element) directive that is replaced with a more complex nested HTML node using the 'templateUrl' feature of directives.
HTML defining the directive (form tag included for complete mental image):
<form id="frm" name="frm">
<ds-frm-input-container
class="col-md-1"
frm-Name="frm"
frm-obj="frm"
input-name="txtFName"
ds-model="user.firstName"></ds-frm-input-container>
</form>
TemplateUrl contents which 'replaces' the above directive 'ds-frm-input-container' HTML element:
<div>
<input
required
ng-minlength=0
ng-maxlength=50
class="form-control"
ng-model="dsModel"
placeholder="{{dsPlaceHolder}}" />
<span ng-if="showErrs" class="label label-danger">FFFFF: {{dsModel}}</span>
</div>
Controller and Directive:
var app = angular.module('plunker', []);
app.controller('MainCtrl', function($scope) {
$scope.name = "Nacho";
$scope.user = {};
$scope.user.firstName = "";
})
.directive('dsFrmInputContainer', function(){
var ddo = {
priority: 0,
restrict: 'AE',
scope:
{
frmName: '#',
inputName: '#',
dsPlaceHolder: '#',
dsModel: '=',
frmObj: '='
},
templateUrl: 'template1.html',
replace: true,
controller: function($scope)
{
$scope.showErrs = true;
},
compile: function compile(ele, attr) {
return {
pre: function preLink(scope, ele, attr, controller)
{
},
post: function postLink(scope, ele, attr, controller)
{
var txt = ele.find('input');
txt.attr('id', scope.inputName);
txt.attr('name', scope.inputName);
//BLUR
txt.bind('blur', function () {
console.log("BLUR BLUR BLUR");
angular.forEach(scope.frmObj.$error, function(value, key){
var type = scope.frmObj.$error[key];
for(var x=0; x < type.length; x++){
console.log(type[x]);
}
});
event.stopPropagation();
event.preventDefault();
});
}
};
},
};
return ddo;
});
The directive replaces just fine and the input element is named just fine. The form object however doesn't include the input element name in the error information. This makes it impossible for me to single out the input element during a 'blur' event that is setup in the directive.
I am doing this trying to reduce the show/hide logic 'noise' in the html for error messages (spans) and it should be reusable.
UPDATE (2014.01.28):
2014.01.28:
Added promises. There is a service that allows validation on button clicks. NOT USING built in angular validation anymore found some compatibility issues with another library (or viceversa).
ORIGINAL:
Here is my form validation directive vision completed (plnkr link below). Completed in concert with the help of the stack overflow community. It may not be perfect but neither are butterfingers but they taste good.
http://plnkr.co/edit/bek8WR?p=info
So here is a link that has the name variables set as expected on the given input form error object. http://plnkr.co/edit/MruulPncY8Nja1BUfohp?p=preview
The only difference is that the inputName is read from the attrs object and is not part of the scope. This is then read before the link function is returned, in the compile phase, to set the template DOM correctly.
I have just spent quite a while trying to sort this problem out, and while this is not exactly what you were looking for, his is my attempt. It uses bootstrap for all the styling, and allows for required and blur validation, but its definitely not finished yet. Any thoughts or advice much appreciated.
https://github.com/mylescc/angular-super-input