Documentation versions

Developing your own Widgets

A programmer should be able to build web applications entirely from Widgets that are available, without ever having to think about the underlying technologies. That is what Reahl is about. However, it is possible for programmers to directly use all the lower-level web technologies in order to create Widgets so that others can reap the benefit of re-use of these Widgets.

This is obviously an advanced topic, and a big one which falls outside the scope of this tutorial. Knowing a little about what is possible on that level is useful though: it gives you a better sense of how Reahl works, and how it is different from other Widget-sporting frameworks.

This section explains the gist of what is possible, without the gory details.

What is a Widget, really

If you search web designer and web development forums, you will find a lot of tricks that are presented in order to achieve a certain effect on a web page. These tricks invariably use a combination of web technologies: you may have to write a bit of HTML and a bit of CSS and a bit of JavaScript to make it work. You may also have to provide some server-side code, exposed via one or more URLs.

Imagine if you can take all the bits of code and URLs in all the various languages that make up one trick and package it in a way that not only makes it reusable, but that hides all that stuff and make the programmer’s world consist of Python only.

That is the dream fuelling Reahl Widgets.

To fulfill the dream, the developer of a low-level Widget needs to be able to use all those web development tricks, and wrap them in something with the face of a Widget class.

For a start, a Widget is built around its Python class. Server-side logic goes into the Python class but the Python class is also responsible for generating some HTML. Widgets exist for most basic HTML elements. Thus, you can easily compose a Widget from other Widgets to achieve the required basic HTML representation wrapped in a Python class.

Adding JavaScript

To add JavaScript to the mix, you use something akin to a class, but written in JavaScript in which you write all code that needs to be executed in the browser for your Widget. This ‘JavaScript class’ is instantiated on the browser for the HTML of each instance of your Python Widget present on a page.

If a Widget needs a ‘JavaScript class’ to be instantiated on the browser end, it achieves this by implementing a special method. This method (.get_js()) returns a small snippet of JavaScript code which will instantiate the JavaScript for that Widget on a page when executed. When rendering an HTML5Page, the framework collects all these snippets of JavaScript from all Widgets on that page and include them on the page in a <script> element. This takes care of instantiating the JavaScript side of Widgets.

For this to work, the code implementing each ‘JavaScript class’ also needs to be included somewhere on the page – this is the code that is called in the ‘instantiating’ snippets. Code for such ‘JavaScript classes’ is written in a separate .js files. Each such .js file is listed in the .reahlproject file of the component containing it. When a web application is started, Reahl builds one .js file that contains all .js files of all the components used by your application and makes the big file available via an URL that is included in any HTML5Page via a <script type="text/javascript"> element. (This file is thus the bulk of JavaScript code, and it will fetched once by browsers and then cached for other pages.)

That’s really all there is to it... Some more details to make it more concrete:

Reahl uses the so-called Widget Factory from the JQuery UI project to simplify the development of the equivalent of a class, but written in JavaScript. If you don’t know the Widget Factory, here’s how it works: the Widget Factory is a function called widget(). You call widget() with the name of the JavaScript class to be created and a JavaScript object that contains a function for each method you want on the ‘JavaScript class’. The special function _create() is the constructor of the ‘JavaScript class’.

What the Widget Factory does is to add a function to the JQuery object which will create an instance of your ‘JavaScript class’ for each HTML element selected by a normal JQuery selection.

For an example, have a look at the implementation of LabelOverInput. A LabelOverInput looks like an Input with its label printed where you’d expect a user to type some input. When a user clicks there to supply some input, the label disappears. The HTML for LabelOverInput is basically a <span> that contains a <label> as well as another <span> that wraps the actual <input> at stake.

To make this plan work, the LabelOverInput needs some CSS for positioning the <label> on top of the <input>. It also needs some JavaScript to make the label appear and disappear at the right moments.

Here is the ‘JavaScript class’ for LabelOverInput:

/* Copyright 2013, 2014 Reahl Software Services (Pty) Ltd. All rights reserved. */
/*
    This file is part of Reahl.

    Reahl is free software: you can redistribute it and/or modify
    it under the terms of the GNU Affero General Public License as
    published by the Free Software Foundation; version 3 of the License.

    This program is distributed in the hope that it will be useful,
    but WITHOUT ANY WARRANTY; without even the implied warranty of
    MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
    GNU Affero General Public License for more details.

    You should have received a copy of the GNU Affero General Public License
    along with this program.  If not, see <http://www.gnu.org/licenses/>.
*/

/* implements plan: http://www.alistapart.com/articles/makingcompactformsmoreaccessible */
(function($) {
"use strict";

$.widget("reahl.labeloverinput", {
	_create: function() {
	    var o = this.options;
	    this.element.addClass("reahl-labeloverinput");
	    this.element.children().filter("label").each( function() {
		    var label = $(this);
		    //		    var input = label.siblings('*[name=' + label.attr("for") +']');
		    var span = label.siblings('span:first');
		    var input = span.find('*[name=' + label.attr("for") +']');
		    label.bind("click", function() {
			    input.focus();
			});
		    input
			.bind("focus", function(){ 
				if ( $(this).val() == "" ) {
    				    label.attr('hidden', 'true');
				}
			    })
			.bind("blur", function(){ 
				if ( $(this).val() == "" ) {
    				    label.removeAttr('hidden');
				}
			    });
		    if ( input.val() == "" ) {
			label.removeAttr('hidden');
		    }
		    else {
			label.attr('hidden', 'true');
		    }
		});
	}
});

$.extend($.reahl.labeloverinput, {
	version: "1.8",
});

})(jQuery);

In the code sample only one method is needed: _create(). This method is the constructor of our ‘JavaScript class’. The variable this is bound to an instance of our JavaScript class inside the functions. Similarly, this.element refers to the HTML element to which this instance of the JavaScript class is bound (the outer <span>, which represents a complete LabelOverInput in HTML).

On the Python side, the Widget needs the .get_js() method that should return a small line of JavaScript needed to instantiate the ‘JavaScript class’, browser side. Using Jquery and the JQuery UI widget function, this is something small, like:

def get_js(self, context=None):
    return ['$(".myclass").labeloverinput();']

The Python implementation of LabelledInlineInput does not illustrate how the HTML is composed, since it inherits all of that from LabelledInlineInput. It does illustrate attaching of JavaScript well though, since adding an HTML class attribute, JavaScript and CSS is all it really does:

class LabelOverInput(LabelledInlineInput):
    """A :class:`LabelledInlineWidget` that shows the Label superimposed over the Input itself. 
       The label is only visible while the Input is empty.
    
       .. admonition:: Styling
       
          Rendered like a :class:`LabelledInlineWidget` with reahl-labeloverinput appended to the class 
          of the containing <div> element.

       :param html_input: (See :class:`InputLabel`)
       :param css_id: (See :class:`HTMLElement`)
    """
    label_class = AutoHideLabel
    @property
    def attributes(self):
        attributes = super(LabelOverInput, self).attributes
        attributes.add_to('class', ['reahl-labeloverinput'])
        return attributes
        
    def get_js(self, context=None):
        js = ['$(%s).labeloverinput();' % self.contextualise_selector('".reahl-labeloverinput"', context)]
        return super(LabelOverInput, self).get_js(context=context) + js

Adding CSS

Sometimes it is necessary to make some CSS also part of the implementation of a Widget. While the end-user may want to add site-specific CSS to deal with the fonts, colors and sizes of an item, sometimes the client-side functionality depends on a little bit of CSS.

The LabelOverInput above is again an example of that. It needs some CSS to help hide its label. CSS is attached in a similar way to how JavaScript is attached. It is written in a file for that Widget, and registered in a .reahlproject from where the framework can pick up all the bits of CSS for a web application and serve it up as one file.

To attach the CSS to the HTML, the CSS would also have to reference the class attribute of a piece of HTML. The CSS for LabelOverInput is as follows:

/* Copyright 2013, 2014 Reahl Software Services (Pty) Ltd. All rights reserved. */
/*
    This file is part of Reahl.

    Reahl is free software: you can redistribute it and/or modify
    it under the terms of the GNU Affero General Public License as
    published by the Free Software Foundation; version 3 of the License.

    This program is distributed in the hope that it will be useful,
    but WITHOUT ANY WARRANTY; without even the implied warranty of
    MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
    GNU Affero General Public License for more details.

    You should have received a copy of the GNU Affero General Public License
    along with this program.  If not, see <http://www.gnu.org/licenses/>.
*/

/*------------------------------------------------------------------- LabelOverInput */

.reahl-labeloverinput {
  position: relative;
}

.reahl-labeloverinput > label[hidden] {
  position: absolute;
  left: -99999999px;
}

.reahl-labeloverinput > label {
  cursor: text;
  position: absolute;
  bottom: 0;
  left: 0;
  z-index:1;
}

.reahl-labeloverinput label.error { 
  position: absolute;
  top: 1.5em;
  left: 0;
  z-index: 2;
}

Adding server-side URLs

Let’s assume you are working on a web application that allows users to store and browse Photos online. Perhaps you’d like to show smaller “thumbnail” versions of photos on some overview page of sorts. You would need a Thumbnail Widget for this.

In order to show a small picture, the HTML for a Thumbnail Widget would need to include an <img> element, and the src attribute of that element needs an URL to be available on the server from where it will fetch a shrinked version of the original image. Users of your Thumbnail Widget do not want to know about this though: it is low-level stuff – it must just happen.

When a Widget is instantiated, it can register so-called SubResources with the framework. A SubResource can be a static file which is reachable via an URL, or a method that can be called server-side by visiting (or posting to) an URL, for example. The URLs of the SubResources of a particular instance of a Widget will all be made available underneath the URL of all the Views on which the Widget is used: If a Widget appears on a View with URL ‘/a/b’, the URL for a SubResource of a Widget will be something like ‘/a/b/__my_small_pic’. (Each SubResource has to have a unique name from which its URL is derived.)

SubResource URLs can also be parameterised, just like a View can be parameterised. This has the effect that the SubResource is only created once its URL is actually accessed.

Table Of Contents

Previous topic

Using a different persistence mechanism

Next topic

Styling your application

Documentation versions