Code

Calling parent function in Symfony/Twig form theme

17th November, 2020 - Posted by david

I’ve recently started building a new web app and have decided to build it in Symfony (and hopefully ultimately Vue for the front-end), as a way of learning what looks like a really cool framework. I’d had an introduction to Symfony through work, then did some online courses, including a load of the excellent SymfonyCasts, so have a reasonable amount of knowledge now on the subject.

One cool feature of Symfony is its form handling abilities. Using simple annotations and generated code, you can get alot of functionality around submitting, validating and saving your form data for very little effort. When rendering a form to the user, Symfony’s template handler, Twig, has a load of built in functions to render each field, such as form_widget to display the field’s HTML tag, form_label to display a field’s label, form_row to combine the above 2 and 2 more, etc. These can be overwritten, so you can style/theme your form using your own CSS, while keeping Symfony and Twig’s powerful form functionality. More details can be found on Symfony’s form customization page.

One thing I found was that sometimes you don’t really need to modify the original form_widget, you just need to say wrap its contents in a div with a specific class. So, ideally there’d be a way to overwrite form_widget but still call the parent/original one. Luckily there is, but it was a little tricky to figure out.

Let’s say I need to add a <div class="select"> around a particular field. For this, we can overwrite the choice_widget function, which generates <select> tags. In your form theme you can do something like:

1
2
3
4
5

{% block choice_widget %}
<div class="select">
{{ parent() }}
</div>
{% endblock choice_widget %}

And you’d think that would work no issues! However, I kept getting errors along the lines of “Can’t call parent function” or “parent function does not exist”, or something like that. I found it surprising, because you can simply call choice_widget before you add your code to overwrite it, so the function must be in the global scope, right? Well, somehow it isn’t! Luckily the fix is simple: import the base Twig theme at the top of your own theme:

1	{% use 'form_div_layout.html.twig' %}

And voilá! You should now have your select wrapped in its stlystyled div.

KnockoutJS: multiple jquery actions in one binding handler

5th August, 2016 - Posted by david

A general rule I follow when using KnockoutJS is that there should be no DOM manipulation in the viewModel. The viewModel should be completely independent of any markup, while any changes to the DOM (via jQuery or otherwise) should be handled in the binding handler. This makes your viewModels much more portable and testable.

As I’m sure you’re aware if you’re reading this article(!), KnockoutJS’s binding handlers are applied to an element and have init and update functions that get called when a certain value in your viewModel changes. Within your init function, you can set up various DOM-element-specific jQuery handlers, while within your update function you can perform various DOM manipulations, trigger events etc., as well as reading from/updating your viewModel and much more.

A common situation I’ve come across a number of times is: say you have a big div with plenty of buttons and links that are tied into external jQuery plugins and DOM elements and you want to perform certain actions when they’re clicked or when other changes happen in your viewModel. You don’t really want to have loads of binding handlers for each separate change that might happen in your viewModel, your codebase could get quite big quite quickly. What I’m about to propose is a structure of how to apply 1 binding handler to the entire div, then call various functions to manipulate the DOM outside of your update binding handler function, via the viewModel.

So, I’ll start with the viewModel. I’m going to have an observable action attribute and 2 functions linkClicked and buttonClicked. (Please bear in mind, this is a very simple example for illustration purposes, you wouldn’t really call viewModel functions linkClicked etc.!) There’ll also be a resetAction function, which will be explained shortly.

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18

exampleViewModel = (function($) {
var viewModel = function() {
this.action = ko.observable('');
};
viewModel.prototype.resetAction = function() {
this.action('');
};
viewModel.prototype.linkClicked = function() {
this.action('jQLinkClicked'); // prepended "jQ" to the function name to help the reader later
};
viewModel.prototype.buttonClicked = function() {
this.action('jQButtonClicked');
};
//JS Module Pattern
return viewModel;
}(jQuery));

ko.applyBindings(new exampleViewModel(), $('#example')[0]);

And now our HTML markup:

1
2
3
4

So now we can see that whenever we click either the link or the button, our action attribute will be updated and thus trigger the update function in the exampleBindingHandler binding handler that’s applied to the div. Let’s look at that binding handler now:

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23

ko.bindingHandlers.exampleBindingHandler = {
init: function(element, valueAccessor, allBindingsAccessor, viewModel) {
// do whatever initial set up you need to do here, e.g.
$('body').addClass('whatever');
},
update: function(element, valueAccessor, allBindingsAccessor, viewModel) {
// so this will be called whenever our observable 'action' changes

// get the value
var action = valueAccessor();
// reset to empty
viewModel.resetAction();

switch (action) {
case 'jQLinkClicked':
alert('link');
break;
case 'jQButtonClicked':
alert('button');
break;
}
}
};

So you can see from the above how we can move from various different viewModel changes out to the binding handler and maniuplate the DOM in there. We read and save action from the valueAccessor, then reset it via the viewModel’s resetAction function, just to keep things clean.

At this point we have very simple alerts for each of our actions but of course in real life you’ll want to call your jQuery plugins, change the DOM etc. To keep things clean what we can do is have a simple JSON object with functions for each of the actions and within those functions do our heavy jQuery lifting, something along the lines of:

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34

So, in summary:

Have an observable action attribute in your viewModel
Apply the binding handler to your main div, with the observable action variable as the valueAccessor parameter
Set action accordingly in your viewModel
In your binding handler, figure out what jQuery/DOM manipulation etc. you want to do based on the value of action

Passing options to Reveal modal in Foundation via HTML

18th March, 2016 - Posted by david

In my current job, we use Foundation for stuff like modal popups, fancy drop downs etc. I haven’t used it too much but I know for the modal dialogs you can either instantiate them via Javascript ($('#elem').foundation('reveal', 'open', {option: 'value'});) or via HTML attributes (<a href="#" data-reveal-id="elem">Open</a> and <div id="elem" data-reveal>).

Passing options to Foundation via Javascript is pretty trivial, as can be viewed in the example above. However, doing this via HTML attributes isn’t so straight-forward and I found the documentation online pretty hard to find. Luckily I was able to figure it out and it’s simple enough: you add a data-reveal-init attribute and a data-options attribute on your modal div. Each of the options are separated by semi-colons and are of the format option: value, e.g.

1
2
3

So, as I said, easy enough in the end but finding it documented proved tircky! Hope this post helps.

Autocomplete with KnockoutJS

3rd February, 2016 - Posted by david

Recently I started a new job at a company that is looking to transition away from a customised, unstructured, jQuery module set up to use KnockoutJS and RequireJS for it’s modules. This approach was chosen because the core platform is based on Magento and the forthcoming Megento 2 uses KnockoutJS heavily throughout it’s frontend templates. As a good starting point and proof of concept, we decided to look at converting our existing custom-autocomplete module from a combination of EJS and jQuery to pure KnockoutJS. Luckily for me, I was the one who got to implement it, and thus learn a new skill!

I’m not going to go into the ins and outs of how KnockoutJS works but in short it’s a MVVM system, where you have Models, Views and ViewModels, the latter being the interface between the other 2, the client and the server. This autocomplete was a standard input field, whereby on typing 3 characters, an AJAX call is made to the server looking for strings that matched the search string and displayed a clickable list of results underneath the input field. Additionally, you could use the arrow keys to select items in the menu, as well as the mouse. We also have different instances of the autocomplete, to search for different types of entities (e.g. searching for a product vs. search for a place), so we need the code to work with each.

From this point on I’m going to assume at least a basic knowledge of KnockoutJS, how it uses data-bind etc.

The View Model

So, first up we’ll want an Autocomplete viewModel, to handle the DOM events in the view (e.g. keyup etc.), fetch data from the server and call the correct model to format the received data. It’ll have 2 observable attibutes: suggestions, an array of suggestion objects, and q, the incoming query from the user. As a parameter we’ll pass it the model type to format the suggestions (e.g. LocationSuggestion below) and we’ll have functions to fetch suggestions as JSON from the server (loadSuggestions), add them to our suggestions array (addSuggestion, formatting the data via the model along the way) and clear our array (clearSuggestions), as well a helper function to look for valid character key presses (validKey). None of this is overly complex and it’s well commented, so I’ll just leave the whole class here:

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83

/**
* AutoComplete viewModel. Handles the observable events from the view, requests data from the server and calls
* the corresponding Model above to format fetched data
*
* @param options JSON object of options, to contain:
* - url: URL to request the search results from
* - suggestionEntry: required model (i.e. one of the above) to format the data
*/
function AutoComplete(options) {

// KnockoutJS standard is to refer to 'self' instead of 'this' throughout the class.
// It's because 'this' in a sub-function refers to the function, not the viewModel
var self = this;

$.extend(self, options);

// Array to store suggestions received from the server
self.suggestions = ko.observableArray([]);

// Value of input field that user queries
self.q = ko.observable('');

// Attribute to store the current AJAX request. Means we can cancel the current request if the observable 'q' changes
self.ajaxRequest = null;

/**
* Append a JSON search result to our suggestions array. Instantiates the correct model to format the data
* (view is rendered automatically by KnockoutJS)
*
* @param suggestion JSON object, returned from search server
*/
self.addSuggestion = function (suggestion) {
self.suggestions.push(new self.suggestionEntry(suggestion, self.q()));
}

/**
* If the user has entered a valid search string (more than 3 latin-ish or punctuation characters), cancel the current AJAX request (if any),
* fetch the data from the server, format it and store in 'suggestions' array
*
* @param obj HTML <input> element (not used)
* @param event The event object for the triggered event (keydown)
*/
self.loadSuggestions = function(obj, event) {
// if a valid, non-control, character has been typed
if (self.validKey(event)) {
self.clearAjaxRequest(); // cancel current request
var q = self.q();
// if they've entered less than 3 characters, just clear the array, which clears the suggestions drop down
if (q.length < 3) {
self.clearSuggestions();
return;
}

// request data from the server
self.ajaxRequest = $.getJSON(self.url, {term: q}, function(response) {
self.clearSuggestions(); // clear out current values
for (var i = 0; i < response['suggestions'].length; i++) {
self.addSuggestion(response['suggestions'][i]); // add search result
}
});
}
}

self.clearSuggestions = function() {
self.suggestions([]);;
}

self.clearAjaxRequest = function() {
if (self.ajaxRequest) {
self.ajaxRequest.abort();
self.ajaxRequest = null;
}
}

/**
* Check what key was pressed is valid: if it was alphanumeric, space, punctuation or backspace/delete
*/
self.validKey = function(event) {
var keyCode = event.keyCode ? event.keyCode : event.which;
// 8 is backspace, 46 is delete
return keyCode == 8 || keyCode == 46 || /^[a-zA-Z0-9\s\-_\+=!"£$%^&*\[\]\{\}:;@'#~<>,\.\/\?ÀÁÂÃÄÅàáâãäåÒÓÔÕÕÖØòóôõöøÈÉÊËèéêëðÇçÐÌÍÎÏìíîïÙÚÛÜùúûüÑñŠšŸÿýŽž]$/.test(event.key);
}
}

We also store the current AJAX request with the object in the ajaxRequest attribute. By doing this, we can cancel any existing requests as the user types more keys. So, when they type the first 3 characters, a request is fired off; when they type the 4th character, we’ll cancel the existing request if it hasn’t finished and do a new search for the longer string.

The Model(s)

For this example, I mentioned above that the user could be searching for locations or products; let’s go with a location search for this example. Below, we have a class LocationSearch, which takes a JSON object that was returned from our server, formats the matched string by wrapping <strong> tags around the bit of the string that was matched (via the global accentInsensitiveRegex function, which I unfortunately don’t have the code for), generates the URL for the result and translates the type of location found (.e.g city, county etc.).

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22

/**
* Model for a location search result. Formats data to be displayed in the HTML view
*
* @param data JSON object
* @param q User's original query
*
*/
function LocationSuggestionEntry(data, q) {
this.type = (data.type === 'region' && Number(data.id) > 999) ? 'country' : data.type;

var separator = data.url.indexOf('?') !== -1 ? '&' : '?';
this.url = data.url + separator + 'autocomplete=1&ac-text=' + q;

this.id = data.id;

this.label = data.label;
// wrap what the user typed in a <strong> tag
var regexp = new RegExp('(' + accentInsensitiveRegex(q) + ')', 'gi');
this.labelFormatted = data.label.replace(regexp, '<strong>$1</strong>');

this.typeTranslated = Translator.translate('autocomplete::type::' + this.type);
}

The View

So, for the HTML side, we need an <input> field for the user’s query and a <ul> for the search results. The <ul> will obviously be hidden if we’ve no results to show. We wrap the whole thing in a <div> with class autocomplete, which we’ll use when binding the whole thing together later.

For the <input> field, we bind our AutoComplete‘s q attribute to Knockout’s textInput data binding (textInput: q), so that every time the value of the <input> changes, q will too. Additionally, we want to fire our loadSuggestions function, which will check the length of q and fetch suggestions from the server if it’s greater than 3 characters; this is achieved by calling loadSuggestions when a Javascript keyup event is fired on the <input> (event: {keyup: loadSuggestions}).

The HTML for the <ul> is also fairly straight-forward. If we have any suggestions to show, we want to add the has-results class to the <ul> (css: {'has-results': suggestions().length > 0}) and of course hide the <ul> when there’s less than 3 characters typed in the <input> (visible: q().length > 2). Assuming we have suggestions to show, we loop through the suggestions array, displaying an <li> for each, containing the suggestion’s labelFormatted and translatedType, as well as adding some attributes to the surrounding <a> (data-bind="attr: {href: url ...).

1
2
3
4
5
6
7
8
9
10
11
12

<div class="autocomplete">
<input class="search-input" data-bind="textInput: q, event: {keyup: loadSuggestions}" type="text" name="text" placeholder="Enter a location..." />
<ul class="autocomplete-results" data-bind="css: {'has-results': suggestions().length > 0}, visible: q().length > 2">

<li>
<a data-bind="attr: {href: url, 'data-type': type, 'data-id': id, 'data-label': label}">
<span data-bind="html: labelFormatted"></span> - <span data-bind="text: translatedType"></span>
</a>
</li>

</ul>
</div>

Binding It All Together

At this point we need to bind our AutoComplete object to our div.autocomplete, passing it the search URL and suggestion type as parameters:

1
2
3
4
5
6

$('.autocomplete').each(function() {
ko.applyBindings(new AutoComplete({
url: '/suggestions/locations/',
suggestion: LocationSuggestion
}), this);
});

One last bit…

To get all this working nicely, you’ll need CSS for the <ul> and it’s <li> children. Additionally, you might want code to look out for when the up and down arrows are pressed on the keyboard and highlight the next row correctly. The code I have for this isn’t mine, so I’m not going to put it here. However, I will point out that to add any fancy jQuery on your view, i.e. to handle these up/down arrow keypress events, you can use KnockoutJS’s custom binding handlers. This is to keep business and presentation logic separate from each other. So, in JS you’d have something like:

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17

/**
* This custom binding is how knockout lets you set up your HTML elements. It's separate from the viewModel, which
* should purely deal with business logic, not display stuff.
*
* Sample usage: <input type="text" data-bind="autoComplete">
*/
ko.bindingHandlers.autoComplete = {
/**
* Called when the HTML element is instantiated
*/
init: function(element, valueAccessor, allBindings, viewModel) {
var $el = $(element);

// specific jQuery code goes here
},
update: function() {} // not needed here
};

The HTML for your <ul> would change to <ul class="autocomplete-results" data-bind="autoComplete, css: {'has-results': suggestions ... – note the addition of autoComplete on the data-bind.

Deeplink to a sub-page within a Facebook iframe app

22nd May, 2015 - Posted by david

Facebook have long had the ability for developers to write custom apps and embed them as tabs in people’s or company’s Facebook profile pages. What I’m talking about here is when you write your own HTML app and host it at e.g. https://fb.mysite.com, which is then embedded into profile pages via an iframe. These apps then have URLs like https://www.facebook.com/profile.site/app_123456789, where 123456789 is your app ID within Facebook.

I’ve written one such app, in PHP, which has sub-pages, so while the root is https://fb.mysite.com, using the app will call pages such as https://fb.mysite.com/product/1. Seeing as this is within an iframe, the URL within the browser remains at https://www.facebook.com/profile.site/app_123456789 while you browse around the app. I recently had a request from a client on how they could link to a sub-page from within a post on their profile page, so they wanted to post something like ‘Check out my product at <link>’, where clicking on the link will load up the iframe app and bring them to the specific product. This is achievable, but it’s not exactly straight-forward and requires some work on the developers behalf. In an ideal world the link would simply be https://www.facebook.com/profile.site/app_123456789/product/1!

The way I managed to achieve this was using Facebook’s app_data parameter. Here, you can pass any data you want to the variable and it’ll end up as part of the signed_request variable in $_REQUEST at https://fb.mysite.com/index.php . The way we’re going to structure these deeplinks is to pass a JSON object to app_data containing a page key, with the sub-page we want, in this instance products/1, so our deeplink is going to look like https://www.facebook.com/profile.site/app_123456789?app_data={page:products/1} . Not exactly elegant but it’ll have to do! You could simply set app_data to products/1, but there may come a time when you want to pass other data in addition to the page, so I opted to go down the JSON route.

Now that we know what to expect, we need to decode $_REQUEST['signed_request'] (which should be available to your https://fb.mysite.com/index.php), json_decode app_data from the result, validate page, then redirect the browser accordingly.

To decode $_REQUEST['signed_request'] I used Facebook’s PHP SDK. Once we have the signed request as an array, we decode the JSON from app_data. Then, we check for the presence of page, validate it (I’ll leave the validation code up to yourself!) and send them on their way. This is pretty straight-forward, so is probably best illustrated with some code:

1
2
3
4
5
6
7
8
9
10
11
12
13
14

// index.php
require_once 'base_facebook.php';
$fb = new Facebook(array(
'appId' => 'YOUR_APP_ID',
'secret' => 'YOUR_APP_SECRET',
));
$req = $fb->getSignedRequest(); // parses $_REQUEST['signed_request']
$app_data = json_decode($req['app_data'], true); // true to ensure an associative array
if (!empty($app_data['page']) && valid_page($app_data['page'])) {
header('Location: /'.$app_data['page']);
exit;
}

// rest of regular index.php

Pity it’s not more straight-forward, but this is the way I got it to work. Hope this helps!

How to call some Pixastic functions

21st February, 2015 - Posted by david

I was doing a bit of work with Canvas recently, manipulating images in the browser and writing the results out to files. I was looking for a package that could do various effects on the images, such as sharpen, blur etc. and came across the Pixastic package used in similar applications. However, unfortunately, the website for the package is currently down and I couldn’t find much documentation on it anywhere. So, I had to look at the source code to figure out how to call the various functions. Not the end of the world, but I just thought I’d stick some simple examples here, to maybe help get others started with the package and to direct them to the source code for more information!

I got the source code from JSdeilvr.net, which is minified, but there is unminified source on GitHub.

The 3 functions I was looking to use were sharpen, brighten/darken and blur. I’ll go through each individually. Firstly though, I’ll mention that the first parameter to each function is a Javascript Image object, which obviously has a src attribute with your image’s contents. The whole point of this was that I was building a tool with buttons which, when clicked, would perform each of the functions above on the current Image. When applying a filter, it’s best track the current level of e.g. brightness, increase/decrease this value, reset the image to how it originally looked and then apply the filter with the new value. This is better than say brightening it by 10%, then brightening the result by 10% again.

Seeing as I used brightness for the example above, I’ll start with that. Also, to darken an image, you simply reduce the brightness value, obviously.

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22

// initial code setup
var img = new Image();
img.src = 'whatever';
img.original_src = img.src; // for doing the resets
// add the current value of each filter to the Image
$.extend(img, {brightness: 0, sharpen: 0, blur: 0});

// ...

// 'Brighten' button click handler
$('#brighten').click(function() {
img.brightness += 2; // brighten by 2 (for darken, reduce by an amount, will work with negative values)
// max brightness of 255
if (img.brightness > 255) img.brightness = 255;

// now we reset the image by creating a copy
img.src = img.original_src;
// now, apply the filter
img = Pixastic.process(img, "brightness", {
brightness : img.brightness
});
});

For sharpen, I found it best not to sharpen by the same amount each time, it just didn’t lead to nice results.

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21

$('#sharpen').click(function() {
if (img.sharpen < 0.5){
img.sharpen += 0.1;
}
else if (img.sharpen > 0.75){
img.sharpen += 0.01;
}
else {
img.sharpen += 0.02;
}
if (img.sharpen < 0.88) {
img.sharpen = 0.88;
}

// now we reset the image by creating a copy
img.src = img.original_src;
// now, apply the filter
img = Pixastic.process(img, "sharpen", {
amount : img.sharpen
});
});

Lastly, blur was a bit more straightforward, increasing by 1 every time:

1
2
3
4
5
6
7
8
9
10

$('#blur').click(function() {
img.blur++;

// now we reset the image by creating a copy
img.src = img.original_src;
// now, apply the filter
img = Pixastic.process(imageObjs[selected_image].image, "blur", {
amount : img.blur
});
});

As I’m sure you’ve spotted, there’s a certain amount of repetition in the code above, starting with the line “// now we reset…”. What I did was to write a function called applyFilters, which you call once you’ve calculated your new value for brightness/sharpen/blur. This will then reset the image and apply all 3 filters. With the code above, if you were to say brighten, then blur, only the blur would be applied, as the image is reset each time. Doing it this way removes that problem.

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25

So, that should be enough to get you working with Pixastic. As for working with Canvas, that’s another blog post!

Homepage takeover with existing skyscraper ads

14th February, 2015 - Posted by david

Sometimes in work we can be asked to do things we don’t like and recently I was asked to look into implementing one of those homepage takovers. Personally, I think these are awful and would like to think I wouldn’t degrade my site by implementing one, but they do make money and have a high click rate, so I can see why sites like to use them.

Normally they’re done using a fixed background wallpaper that’s clickable all the way to the edge of the page. However, I was asked to simulate this look using 2 existing skyscraper ads, 170px in width, to be positioned either side of the main content and fixed to the top of the page. Since it wasn’t entirely straightforward, I thought I’d block about it here, to help anyone else in a similar situation. I’m not going to go into the specifics of displaying the ads, simply the CSS and Javascript involved in positioning them where you want them.

I should point out, this might be possible with just CSS, but changing a site’s fundamental structure to accomodate the new columns isn’t always possible. Also, you might only want the takeover on the homepage and not other pages. This solution should have minimal impact, as it simply adds 2 divs, that can go anywhere in the HTML.

So, to describe the set-up, let’s say our main content is 1000px in width, centred in the page and we want 2 170px x 1068px divs to contain our ads and line up on the right and left of that content, as well as for the 2 ads to remain fixed at the top of the page, no matter how far we scroll down. We’ll give each of these divs a class of side-banner, with left-banner and right-banner IDs. Since these are going to be positioned explicitly, it doesn’t really matter where in the HTML you put them, maybe just inside your main content div. Initially, we’re simply going to position them in the extreme top corners of each side. I’m also going to give them different background colours, so we can know they’re positioned correctly without having to worry about loading the ads (which can come later).

1
2
3
4
5
6
7
8
9
10
11
12
13
14

.side-banner {
width: 170px;
height: 1068px;
position: fixed;
top: 0;
}
#right-banner {
right: 0;
background: red;
}
#left-banner {
left: 0;
background: green;
}

To align these alongside the content, I needed to write a Javascript function (position_banners()) to position them correctly. This function is called when the page finishes loading, as well as when the window is resized. It simply gets the body’s width, subtracts the width of our main content (remember 1000px), divides the result by 2 (as we’ve 2 sides), then further subtracts the width of our banners. This fairly basic formula works out the amount to move each div in from their corresponding edge, to line up with our main content. Then, we just use CSS left and right to position them nicely.

1
2
3
4
5
6
7
8
9
10
11
12

$(document).ready(function() {
position_banners();
$(window).resize(position_banners);
});

function position_banners() {
var margin = ($('body').width() - 1000) / 2 - $('#left-banner').width(),
left = Math.floor(margin),
right = Math.ceil(margin);
$('#left-banner').css({left: left + 'px'});
$('#right-banner').css({right: left + 'px'});
}

I know this code isn’t the tidiest, but should be enough to get the idea of what you need to do.

To further enhance the ‘takeover’ effect, you could display a 970px x 250px ‘billboard’ right at the top of your main content.

Opening hours done right, in PHP and MySQL

7th February, 2015 - Posted by david

I often see different ways of storing and displaying opening hours for various businesses when browsing the web. Some seem to simply store them as an unstructured blob of plaintext and spit that back to the user. Others will store the exact times for each day and display all 7 days on 7 separate rows.

I think neither of these options are great and I’ve come up with what I think is the ideal solution. Basically, we want to store the exact time for each day, but group similar days together. So, you could have lines like ‘Mon – Thurs: 9 – 5’, but if it’s say Tuesday at 10am, you’d also know that the business is currently open. I also think you could have alternative text to display when the business is closed (indicated by open and closed being NULL), so you could have something like ‘Sat – Sun: Open by appointment’.

So, let’s start with our table layout, see below. The primary key is id, which is just your standard auto_increment value. business_id is if you have multiple businesses, each with their own opening hours, as was the case for me. You might want to build an index on this field too. If you’re just storing your own opening hours, dow could be the primary key and you could drop those last 2 fields. open and closed are just simple strings, to store the time in 24-hour ‘HH:MM’ format. Doing it this way, you can still to comparisons like WHERE open > '09:00' and get the result you were expecting.

1
2
3
4
5
6
7
8
9
10

+---------------+-----------------------+------+-----+---------+----------------+
| Field | Type | Null | Key | Default | Extra |
+---------------+-----------------------+------+-----+---------+----------------+
| id | mediumint(9) unsigned | NO | PRI | NULL | auto_increment |
| business_id | mediumint(8) unsigned | NO | MUL | NULL | |
| dow | tinyint(1) unsigned | NO | | NULL | |
| open | char(5) | YES | | 09:00 | |
| closed | char(5) | YES | | 17:30 | |
| optional_text | char(100) | YES | | NULL | |
+---------------+-----------------------+------+-----+---------+----------------+

Next up, I want to show you a quick function to format the time. Being a programmer, the time ’13:30′ is easily translated to ‘1.30pm’, but for the general public it might not be so simple, so this function will display your time in a more human readable format, such as the example given above. Basically, we want to drop any leading 0s, any 0-value minutes, convert the time to a 12-hour version with ‘am’ and ‘pm’ and change the ‘:’ to a ‘.’ (this last bit is probably a bit region-specific). For midnight, we’ll store that as ’24:00′ (which I know is technically the start of the next day!) and display that to the user as ‘midnight’, instead of the slightly confusing ‘0am’.

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26

/**
* Function to change an opening hour to a more human readable version,
* e.g., 09:00 to 9am or 13:30 to 1.30pm
*
* @param String $time Time to format, in HH:MM format
*
* @return String Formatted time
*
*/
function format_opening_hour($time) {
if ($time == '24:00') {
$new_time = 'midnight';
}
else {
list($hours, $minutes) = explode(':', $time);
$hours = ltrim($hours, '0');
$am_pm = ($hours >= 12) ? 'pm' : 'am';
if ($hours > 12) $hours -= 12;
$new_time = $hours;
if ($minutes != '00') {
$new_time .= '.'.$minutes;
}
$new_time .= $am_pm;
}
return $new_time;
}

OK, so displaying the data in a nice table, with similar days grouped together, is the next bit. We’ll have 2 columns, one for the day/days, the other for the time. If a day has a value for optional_text, then that value will be displayed and the times are ignored. I’m also going to add another block of optional text ($extra_text below) that will be displayed at the end of the table and is applied for all days, to be used for something like ‘phone anytime’. Finally, there’s a $short_day_names option, so you can choose between say ‘Mon’ and ‘Monday’.

I should also mention at this point: I’m returning a block of HTML here from a function, as well as mixing business logic with display logic; I realise this is generally a bad idea and some of this could be split into a function and a template, but seeing as it’s a simple 2-column table, I just kept it all together.

/**
* Function to generate a simple html table for a business' opening hours
*
* @param Array $opening_hours Array of rows from opening_hours table, sorted by dow (0-indexed, starting with Monday)
* @param String $extra_text Extra block of generic text that applies to all days, goes at end of table
* @param String $short_day_names Whether to use e.g. 'Mon' or 'Monday'
*
* @return String HTML <table> output
*
*/
function opening_hours_table($opening_hours, $extra_text='', $short_day_names=false) {
$dow = array(
array('long' => 'Monday', 'short' => 'Mon'),
array('long' => 'Tuesday', 'short' => 'Tue'),
array('long' => 'Wednesday', 'short' => 'Wed'),
array('long' => 'Thursday', 'short' => 'Thu'),
array('long' => 'Friday', 'short' => 'Fri'),
array('long' => 'Saturday', 'short' => 'Sat'),
array('long' => 'Sunday', 'short' => 'Sun')
);
$key = ($short_day_names) ? 'short' : 'long';

// first, find similar days and group them together
if (!empty($opening_hours)) {
$opening_short = array();
// start with current day
for ($i=0; $i<7; $i++) {
$temp = array($i);
// try to find matching adjacent days
for ($j=$i+1;$j<7;$j++) {
if (empty($opening_hours[$i]['optional_text']) &&
empty($opening_hours[$j]['optional_text']) &&
$opening_hours[$i]['open'] == $opening_hours[$j]['open'] &&
$opening_hours[$i]['closed'] == $opening_hours[$j]['closed'] ||
!empty($opening_hours[$i]['optional_text']) &&
!empty($opening_hours[$j]['optional_text']) &&
strtolower($opening_hours[$i]['optional_text']) == strtolower($opening_hours[$j]['optional_text']) ) {
// we have a match, store the day
$temp[] = $j;
if ($j == 6) $i = 6; // edge case
}
else {
// otherwise, move on to the next day
$i = $j-1;
$j = 7; // break
}
}
$opening_short[] = $temp; // $temp will be an array of matching days (possibly only 1 day)
}
}

$html = '<table>';
$colspan = '';

if (!empty($opening_short)) {
$colspan = ' colspan="2"';
foreach ($opening_short as $os) {
$day_text = $dow[$os[0]][$key];
if (count($os) > 1) { // if there's another, adjacent day with the same time
$end = array_pop($os); // get the last one
$end = $dow[$end][$key];
$day_text = $day_text . ' - ' . $end; // append the day to the string
}
// at this point, $day_text will be something like 'Monday' or 'Monday - Thursday'
if (!empty($opening_hours[$os[0]]['optional_text'])) {
// optional string takes precedent over any opening hours that may be set
$hours_text = htmlentities($opening_hours[$os[0]]['optional_text']);
}
elseif (!empty($opening_hours[$os[0]]['open'])) {
// otherwise generate something like '9am - 5.30pm'
$hours_text = format_opening_hour($opening_hours[$os[0]]['open']) . ' - ' .format_opening_hour($opening_hours[$os[0]]['closed']);
}
else {
// if nothing, it must be closed on that day/days
$hours_text = 'Closed';
}
// new row for our table
$html .= '<tr>
<td>'.$day_text.':</td>
<td>'.$hours_text.'</td>
</tr>';
}
}

// append the extra block of text at the end of the table
if (!empty($extra_text)) {
$html .= '<tr>
<td'.$colspan.'>'.htmlentities($extra_text).'</td>
</tr>';
}

$html .= '</table>';
return $html;
}

So, with the following data…

1
2
3
4
5
6
7
8
9
10
11

+-----+-------+--------+----------------+
| dow | open | closed | extra |
+-----+-------+--------+----------------+
| 0 | 09:00 | 17:30 | NULL |
| 1 | 09:00 | 17:30 | NULL |
| 2 | 09:00 | 17:30 | NULL |
| 3 | 09:30 | 19:00 | NULL |
| 4 | 09:30 | 19:00 | NULL |
| 5 | 09:30 | 16:00 | NULL |
| 6 | NULL | NULL | By appointment |
+-----+-------+--------+----------------+

you should end up with the following table:

Monday – Wednesday	9am – 5.30pm
Thursday – Friday	9.30am – 7pm
Saturday	9.30am – 4pm
Sunday	By appointment
Phone anytime before 11pm

Backbone JS Models and Collections in same file with RequireJS

31st January, 2015 - Posted by david

In BackboneJS, classes/objects are called Models and groups of these models go into Collections. Using RequireJS, you define something and return that something for use in other files. In many BackboneJS examples, I’ve seen say a Person Model and a Person Collection stored in separate files, with each one being imported separately to the script they’re used in. Seeing as you generally use Models and Collections together at the same time, I was looking for a way to have them both in the one file. The solution is pretty simple and involves some simple Javascript object manipulation. I should give credit for this solution to my friend Ronan, although I’m not sure if he came up with it originally!

So, just to illustrate the normal Backbone way of doing things, sticking wih our Person class/object, you might have 2 files in your JS folder, models/person.js and collections/persons.js, as follows:

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28

// models/person.js
define(['underscore', 'backbone'],
function(_, Backbone) {
Person = Backbone.Model.extend({
url: '/person',
// ... etc
});

return Person;
}
);

//collections/persons.js
define(['underscore', 'backbone', 'models/person']
function(_, Backbone, Person) {
Persons = Backbone.Collection.extend({
model: Person,
// ... etc
});
return Persons;
}
);

// some other JS file
require('models/person');
require('collections/persons');
var people = new Persons();
people.add(new Person());

So, I’m looking for a way to just have one require‘d file, that will import both my Model and Collection. A way this can be achieved is by defining a Person as a standard Javascript object, then having Model and Collection as attributes of that object. We need to add each attribute one step at a time, we can’t just do it all in one big Person = { // ... } block. Then, once the file is imported, we can do stuff like var people = new Person.Collection();.

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22

// person.js
define(['underscore', 'backbone'],
function(_, Backbone) {
var Person = {}; // create initial object
Person.Model = Backbone.Model.extend({
url: '/person',
// ... etc
});

// in order to access the Person.Model, we have to specify each attribute separately
Person.Collection = Backbone.Collection.extend({
model: Person.Model,
// ... etc
});
return Person;
}
);

// some other JS file
require('person');
var people = new Person.Collection();
people.add(new Person.Model());

Extending Google Maps class with RequireJS and Underscore

24th January, 2015 - Posted by david

I’ve been working alot with RequireJS lately and love the whole modular JS movement, where blocks of code are safely contained within separate functions, that can be loaded in when necessary, without worrying about variable/function name clashes. Generally each module is stored in it’s own file but you can combine several modules into one giant file, in theory containing your entire JS code for your site!

When working with Google Maps, I was thinking it’d be nice to simply extend the google.maps.Map class and add in my own functions, e.g. showLoading to display a nice, semi-transparent ‘Loading…’ div in the centre of the map while the code does an AJAX request to get some data. So, you’d end up with something like:

1 2	var map = new Map(); map.showLoading();

So, first thing you need to do is add the Asynch plugin for requireJS to your requireJS config. For external plugins, I tend to source them from https://cdnjs.com/ if possible, cutting down on some of my own server bandwidth. I have my RequireJS config in my require.min.js file, after the main block of minified code. Here’s an edited version, with the asynch plugin on line 6:

1
2
3
4
5
6
7
8
9
10
11
12
13

requirejs.config({
'baseUrl': '//cdn.mydomain.com/js',
'paths': {
'jquery': 'libs/jquery.min',
'underscore': 'libs/underscore.min',
'async': '//cdnjs.cloudflare.com/ajax/libs/requirejs-async/0.1.1/async'
},
'shim': {
'underscore': {
exports: '_'
},
}
});

Next up, we have a map.js file, which, via the define function, we’ll pass UnderscoreJS and the Google Maps JS code using the Asynch plugin. This file is going to return a Map object, which we can then use to instantiate new instances of the class. The file starts off with a Map function, which takes as a parameter the id of the div you want to put the map into and the latitude & longitude of the map’s centre. Obviously you could pass more parameters if you wish, e.g. initial zoom level etc. This Map function will then create a variable of type google.maps.Map, set the map up and return the variable as output from the function. This way when you do var myMap = new Map();, your myMap will be able to call all of google.maps.Map‘s functions.

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16

We also need to give our Map‘s protoype the google.maps.Map prototype, which is simply:

1	Map.prototype = google.maps.Map.prototype;

Finally, if we want to add our own functions to our Map, we use UnderscoreJS to extend the Map.prototype with our functions. So, for example, I have one called isBigDrag to detect if the user has dragged the map a reasonable amount to fetch more data, as well as showLoading and hideLoading to show/hide my nice ‘Loading…’ div.

1
2
3
4
5
6
7
8
9
10
11
12
13
14

_.extend(Map.prototype, {
isBigDrag: function (prevCentre) {
var currentCentre = this.getCenter();
// .. etc
},

showLoading: function() {
// etc.
},

hideLoading: function() {
// etc.
}
});

See below for the entire file, which is stored in ‘classes/map.js’ in baseUrl above.

To pass this new Map class into a script and to be able to do var myMap = new Map();, we do so via a call to requirejs in the script, as follows:

1
2
3
4
5
6
7

requirejs([
'underscore', 'classes/map',
], function(_, Map) {
var myMap = new Map('id', 50.0, 10.0);
myMap.showLoading();
// etc.
});

Took me a while to figure it all out but works pretty well now. The classes/maps.js file in it’s entirety is:

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32

define(
['underscore', 'async!https://maps.googleapis.com/maps/api/js?sensor=false'],
function(_) {

var Map = function (div, lat, lng) {
zoom: 15,
center: new google.maps.LatLng(lat, lng)
});

return map;
}

Map.prototype = google.maps.Map.prototype;

_.extend(Map.prototype, {
isBigDrag: function (prevCentre) {
var currentCentre = this.getCenter();
// .. etc
},

showLoading: function() {
// etc.
},

hideLoading: function() {
// etc.
}
});

return Map;

});