Home

Awesome

<img src="https://raw.githubusercontent.com/bminer/node-blade/master/logo.png" alt="node-blade - HTML Template Compiler" height="200"/>

Blade is a HTML Template Compiler for Node.js, inspired by Jade/Pug and Haml.

It works like this...

  1. Write up your template in Blade (which is a Pug-like language)
  2. Use the Blade compiler to generate a Blade template (which is a JavaScript function)
  3. Pass variables into your generated template to produce HTML or XML

View a simple example

Never write HTML again. Please.

This is NOT the Blade templating engine developed by Laravel. Laravel Blade was added in Sept, 2011; whereas, I did not begin development on Blade until May, 2012. Nevertheless, I still blame Laravel for choosing the same name and for creating any confusion. :)

Table of Contents

Why use Blade instead of Jade?

Features

Project Status

I'd say that Blade itself is stable. There are very few (if any) known issues, and I think that Blade is ready for production environments. I use Blade for a few of my projects.

Meteor support for Blade is deprecated. The Meteor team has been changing the Templating API for Meteor significantly, and it is quite likely that any code that has been written for Blade + Meteor support is broken.

If you find a bug, please [report it here] (https://github.com/bminer/node-blade/issues). If you include the Blade code that failed along with the expected HTML output, that is always splendid. Full stack traces for Errors are quite nice, too.

By all means, please feel free to submit pull requests for new features, new tests, or whatever! For big changes, say ~100 lines of code, you might want to contact me first or submit an issue before getting started.

Installation

for Node (via npm): sudo npm install -g blade

Runtime for Browsers: wget https://raw.github.com/bminer/node-blade/master/lib/runtime.js Minified runtime is about 7-8 KB, uncompressed.

Syntax

Tags

Like Jade, a tag is simply a word. For example, the string html will render to <html></html>.

You can have 'id's:

div#awesome

which renders as <div id="awesome"></div>.

Any number of classes work, separated by a dot (.)

div.task-details.container

which renders as <div class="task-details container"></div>.

Tag attributes? Yep, they work pretty much like Jade, too. Even string [interpolation] (#interpolation) works. Put attributes in parenthesis, separate attributes with a comma, space, newline, or whatever.

a(href="/homepage", onclick="return false;") renders as:

<a href="/homepage" onclick="return false;"></a>

You can also have line feeds or weird whitespace between attributes, just like in Jade. Whatever. This works, for example:

input(
		type="text"
		name="email"
		value="Your email here"
	)

You can also put substitute an attribute value with vanilla JS code like this: input(type="text" name="contact-"+name value=value). For example, if you passed the object {name: "fred", value: "testing"} to your view, the above would render to: <input type="text" name="contact-fred" value="testing"/>

You cannot put whitespace, commas, newlines, or parentheses in the vanilla JavaScript code, though. Blade uses these characters to separate each attribute or to end the tag definition.

And, yes... the class attribute is handled with extra special care. Pass an array or string. Classes (delimited by ".") from before will be merged with the value of the class attribute.

For example:

div#foo.bar.dummy(class="another dude") renders as: <div id="foo" class="bar dummy another dude"></div>

Boolean attributes are allowed, as well. If the attribute value is boolean true, then the attribute is set; if the attribute value is boolean false, then the attribute is ignored entirely. For example:

input(type="text" checked=true) renders as: <input type="text" checked="checked"/>.

Or... you can write it HTML 5 style like this:

input(type="text" checked) which renders as: <input type="text" checked="checked"/>.

div, div, div can get annoying... so, we can omit the tag specifier if we specify an id or some classes:

#foo
.bar
#this.is.cool

renders as:

<div id="foo"></div><div class="bar"></div><div id="this" class="is cool"></div>

Blade just assumes anything without a tag name specifier is a <div> tag.

Also, tags without matching ending tags like <img/> render properly.

Escaping Blade keywords

Finally, you can start a tag name with a backslash to escape Blade keywords. Normally, include test would include a file, but \include test renders as:

<include>test</include>

This allows you to be flexible with tag names, so you are not restricted to rendering HTML, for example. You can render any XML document with Blade.

Indenting

Simply intent to put content inside of a tag.

You can indent with any number of spaces or with a single tab character. The only rule is to be consistent within a given file. Jade gives you a lot of weird indent flexibility. Blade, by design, does not.

html
	head
		title Welcome
	body
		#content

renders as:

<html>
	<head>
		<title>Welcome</title>
	</head>
	<body>
		<div id="content"></div>
	</body>
</html>

Text

Simply place content after the tag like this:

p This text is "escaped" by default. Kinda neat.

renders as:

<p>This text is &quot;escaped&quot; by default. Kinda neat.</p>

Want unescaped text? Large blocks of text? Done. Start a line of text with a |.

p! This will be <strong>unescaped</strong> text.
	|
		How about a block? (this is "escaped", btw)
		Yep. It just works!
		Neato.

renders as:

<p>This will be <strong>unescaped</strong> text.
How about a block? (this is &quot;escaped&quot;, btw)
Yep. It just works!
Neato.</p>

Rules are:

String Interpolation

String interpolation is supported for text blocks (and text attributes and text filters). Use #{var_name} notation, and anything between the curly braces is treated as vanilla JavaScript code.

For example, you can write:

(caution: indents are required on line 4 even though it is blank)

p
	|
		I am just testing #{whatever + ", alright?"}

		Relax...

instead of writing the equivalent, but arguably less awesome...

p
	|=
		"I am just testing " + whatever + ", alright?" +
		"\n\n" +
		"Relax..."

Assuming a local variable whatever is passed to the template with value "Blade", both of the examples above will render to this:

<p>I am just testing Blade, alright?

Relax...</p>

Interpolation comes in two forms: escaped and unescaped. If you want escaped (i.e. the resulting string has >, <, ", and other HTML characters escaped), use #{foo}; if you want unescaped, use !{foo}. If you literally want to insert "#{foo}" in your text, just prepend with a backslash like this: \#{foo}.

Whitespace between tags

In Blade, whitespace is only added when it's explicitly needed. For example:

input(type="text")
input(type="text")

renders as: <input type="text"><input type="text">

If you need something like... <input type="text"> <input type="text"> (notice the space between the elements), then you have some options...

One way is to use a text block:

input(type="text")
|
input(type="text")

Notice on line 2 that the | is followed by two spaces.

Another way is to prepend a tag with a <:

input(type="text")
<input(type="text")

Or append the tagname with a >:

input>(type="text")
input(type="text")

Whatever you like!

Text filters

Need <br/> tags inserted? Use a built-in filter, perhaps?

p
	:nl2br
		How about some text with some breaks?

		Yep! It works!

renders as:

<p>How about some text with some breaks?<br/><br/>Yep! It works!</p>

Built-in text filters include:

Filters are essentially functions that accept a text string and return HTML. They cannot modify the AST directly. Also, you cannot inject JavaScript code into filters.

You can add custom filters at compile-time using the API.

Variable interpolation is supported for certain text filters, as well. If a text filter returns text in #{var_name} or !{var_name} notation, then anything between the curly braces is replaced with vanilla JavaScript code. To avoid this behavior, text filters can either escape the #{stuff} with a backslash, or it can set its interpolation property to false. See lib/filters.js for some examples if you want to write your own filter.

Code

Use dash (-) to indicate that JavaScript code follows, which will not output into the template. As before, use equals (=) to specify code output. A few examples, please?

Using dash (-):

#taskStatus
	- if(task.completed)
		p You are done. Do more! >:O
	- else
		p Get to work, slave!

When inserting lines of code with -, curly braces or semicolons are inserted, as appropriate. In the example above, we have an if statement followed by an indented paragraph tag. In this case, Blade wraps the indented content with curly braces. If there is no indented content beneath the line of code, then a semicolon is appended instead.

Disclosure for Jade users: Unlike Jade, you can't write conditional or iterative statements in Blade without the use of the dash -. This is by design -- IMHO it helps one differentiate between logic and markup.

If you have longer lines of JavaScript code, and you want to extend the line over multiple lines, you can do that. Just use the underscore _ to tell Blade that you are extending the previous code statement.

Pay close attention to when the dash - and underscore _ are used.

Here's an example:

#taskStatus
	- if(task.completed &&
		_ tasks.length == 0)
		p Well... I suppose there is nothing left to do.

Or... another example...

- var longString = "This particular line of JavaScript " +
	_ "is rather long, so we decided to break it up into " +
	_ "multiple lines."
p=longString

You should never use the underscore _ if you don't need multi-line code statements.

If you need to insert a tag that begins with an underscore _, escape it with a backslash \ (just like you would if you were [escaping Blade keywords] (#escaping-blade-keywords) ).

Code that Outputs

Code that outputs (i.e. a code block or at the end of a tag). As mentioned before, it's just like a text block, except with an =.

#taskStatus= task.completed ? "Yay!" : "Awww... it's ok."
p
	| The task was due on
	|= task.dueDate

When using code that outputs, the default is to escape all text. To turn off escaping, just prepend a "!", as before:

p
	|!= some_var_containing_html

Missing "|" characters are okay, too. Just don't forget that stuff after the "=" needs to be valid JavaScript code!

p
	= "escape me" + " away & away"

renders <p>escape me away &amp; away</p>

Variable names to avoid

Blade, like other template engines, defines local variables within every single view. You should avoid using these names in your view templates whenever possible:

Doctypes

Don't forget a doctype! Actually, you can, whatever...

Add a doctype using the doctype keyword or !!! like this:

!!! 5 means use HTML 5 doctype.

Use the list of built-in doctypes or pass your own like this:

doctype html PUBLIC "-//W3C//DTD XHTML Basic 1.1//EN"
html

which renders as <!DOCTYPE html PUBLIC "-//W3C//DTD XHTML Basic 1.1//EN"><html></html>

Put the doctype at the top of your Blade files, please. Please refer to [doctypes.js] (https://github.com/bminer/node-blade/blob/master/lib/doctypes.js) for the list of built-in doctypes.

You can modify the list of built-in doctypes through the API, if you insist.

Comments

Use // for a line comment. Use //- if you don't want the comment to be rendered. Block comments work, too.

//Comment example 1
//-Comment example 2
//
	#wow
	p Block comments work, too

renders as:

<!--Comment example 1--><!--<div id="wow"></div><p>Block comments work, too</p>-->

Conditional comments work like this:

head
	//if lt IE 8
		script(src="/dear-microsoft-plz-stop-making-browsers-kthxbye.js")

renders as:

<head><!--[if lt IE 8]><script src="/dear-microsoft-plz-stop-making-browsers-kthxbye.js"></script><![endif]--></head>

To comment out entire sections of Blade code, you can use non-rendering block comments with a text block.

//-
	|
		anything can go here... Blade code, JavaScript code, whatever...
		just make sure that the indenting is right.

or... even better... just use C-style block comments. Begin with /* to generate a non-rendering block comment, or begin with /** to generate a regular comment. End the comment with */. These comments are not parsed like // comments.

/* h1 Testing */
/**
#header
	h3 Notice that this chunk of Blade code is not parsed
*/

renders as:

<!--
#header
	h3 Notice that this chunk of Blade code is not parsed
-->

Foreach

This feature is now deprecated. Consider the use of the native JavaScript Array.prototype.foreach.

It is likely that this syntax will be removed in future versions of Blade, but rather than removing it entirely, there may be some alternative syntax added.

The exact syntax of a foreach region is the word "foreach", followed by the variable name of the JavaScript Array or Cursor Object, optionally followed by "as" and an item alias. Finally, it is possible to follow the foreach region by an "else" block, which is only rendered if there were no items in the collection.

As a side note, a Cursor Object, as described above, is an Object with an observe() method, as described by cursor.observe(callbacks)

For example:

ul
	foreach users as user
		li #{user.firstName} #{user.lastName} (#{user.age})
	else
		li No users were found

Assuming that users is an Array, the above would produce the same as:

ul
	- for(var i = 0; i < users.length; i++)
		- var user = users[i];
		li #{user.firstName} #{user.lastName} (#{user.age})
	- if(users.length == 0)
		li No users were found

The foreach region is preferred over the example above not only because of readability and brevity, but because it also provides Blade with the ability to better integrate with live page updating engines (specifically Meteor and Spark). That is, if the live page update engine supports tracking reactive collections, the most efficient DOM operations may occur to update the view's results in-place, without re-rendering the entire Blade template.

Blocks don't work well inside of foreach regions. Specifically, while inside of a foreach region: (1) you cannot access blocks declared outside of the foreach region; and (2) blocks declared inside of the foreach region are not accessible once you leave the foreach region. If this causes a problem, just use regular JavaScript for loops.

Event Handlers

You can write inline event handlers right into your Blade templates. Here's an example:

form(method="post" action="/login")
	input(type="text" name="username")
		{change}
			//javascript code goes here
			//e refers to the browser's event Object
			//e.currentTarget refers to this DOM element
			validate(e.currentTarget.value);
	input(type="password" name="password")
		{change}
			checkPasswordStrength(this.value);

The above code will automatically register the 'onchange' event handler with the corresponding input tags.

As shown in the example, your event handler may reference e (the browser's event Object). Be aware that every browser's event Object might be slightly different, especially in legacy browsers.

It is also worthwhile to note: If you are rendering the template in the browser (i.e. using client-side templates), your event handler will have access to the view's locals due to JavaScript closures. This can be rather convenient. :)

Functions

Functions are reusable mini-templates. They are similar to 'mixins' in Jade.

Defining a function:

function textbox(name, value)
	input(type="text", name=name, value=value)

Calling a function and inserting into template structure:

form
	call textbox("firstName", "Blake")

Or... maybe just putting the generated HTML into a variable?

call textbox("firstName", "Blake") > text
//alternative syntax: call text = textbox("firstName", "Blake")
form
	!=text

Both examples would render:

<form><input type="text" name="firstName" value="Blake"/></form>

You can also append content rendered by a function to a variable: call textbox("firstName", "Blake") >> text or... alternatively... call text += textbox("firstName", "Blake")

Note: when you define a block (see below) within a function, and you output the rendered content to a variable, the block will be destroyed immediately after the function call.

Yes, you can use arguments within your function, just like a "real" JavaScript function. In fact, functions are "real" JavaScript functions, so even closures work! Although, remember that functions have access to the variables in scope at the time the function was defined, not the variables in scope when the function is called.

Example:

- var x = 12;
function test(foo)
	h1=foo
	- if(x)
		p=x
#example
	call test("Header")

would render: <div id="example"><h1>Header</h1><p>12</p></div>

Adding classes or an id to rendered function content

Yes, you can add a class name or id to the first element rendered by a function:

function dialog(msg)
	.dialog
		= msg
call dialog("Blade is awesome")#foobar.foo.bar

which would render as <div id="foobar" class="dialog foo bar">Blade is awesome</div>.

Although, if you try it with something like this, you get an error because the first child rendered by the function is not a tag.

function dialog(msg)
	= msg
call dialog("Blade is awesome")#foobar.foo.bar
//compiler might generate an error, or it might just ignore the id and classes

Dynamic file includes

include "file.blade"

This will insert "file.blade" right into the current view at runtime, as if the contents of the included file were copied right into the current view.

If you don't know the name of the file to be included until runtime, that's no problem. The include statement can also be followed by the name of a JavaScript variable containing the filename to be included. These are called dynamic filename includes.

- var filename = "file.blade"
include filename

CAUTION: When using dynamic filename includes in the browser, be sure that you have properly loaded all views that might be included into the browser's cache before executing the view containing the dynamic filename include. See the implementation details for a more detailed explanation.

If you do not specifiy a file extension, .blade will be appended to your string internally.

You may also place an include inside of a function or block.

Finally, you can specify which local variables should be passed to the included view template by using the exposing keyword. By default, Blade will pass the parent's view locals (not local variables declared with - var foo = ...) to the included template; however, when using the exposing keyword, you can specify exactly which variables (either view locals or variables declared with - var foo = ...) are to be exposed to the included template.

For example:

- header = "Header: 1, 2, 3"
- text = "This is some text: 1, 2, 3"
- for(var i = 0; i < 10; i++)
	include "foobar" exposing i, text

In the example above, variables i and text are exposed to "foobar.blade"; the header variable will not be accessible from "foobar.blade".

Note: when using Meteor or another live page update engine, preserve and constant regions only work properly in an included template if and only if the template does not define any blocks. In other words, don't include a template that declares blocks and has some preserve/constant regions.

Blocks

Blocks allow you to mark places in your template with code that may or may not be rendered later.

You can do a lot with blocks, including template inheritance, etc. They behave quite differently from Jade.

STOP! If you are planning on using blocks with Meteor, beware! First of all, blocks might not make much sense when building Meteor applications, and furthermore, blocks don't work well with reactive HTML. For more information about why blocks are not recommended for use with Meteor, checkout [this section] (https://github.com/bminer/node-blade/wiki/Using-Blade-with-Meteor#wiki-block_limitations) of the [Using Blade with Meteor wiki page] (https://github.com/bminer/node-blade/wiki/Using-Blade-with-Meteor).

OK. I digress...

There are two types of blocks: regular blocks and parameterized blocks.

Regular blocks

Regular blocks are defined using the "block" keyword followed by a block name. Then, you optionally put indented block content below. Like this:

block regular_block
	h1 Hello
	p This is a test

Assuming nothing else happens to the block, it will be rendered as <h1>Hello</h1><p>This is a test</p> as expected. Empty blocks are also permitted. A simple, empty block looks like this: block block_name

Of course, the purpose of declaring/defining a block is to possibly modify it later. You can modify a block using three different commands:

Example:

append regular_block
	p This is also a test

Replacing a block

Replacing a block is somewhat confusing, so I will explain further. If you replace a block, you are not changing the location of the defined block; you are only replacing the content of the block at its pre-defined location. If you want to change the location of a block, simply re-define a new block ([see below] (#what-happens-if-i-define-the-same-block-more-than-once)).

In addition, when you replace a block, all previously appended and prepended content is lost. The behavior is usually desired, but it can sometimes be a source of confusion.

If you replace a parameterized block (described below) with a regular block, you cannot call "render" on that block.

You can replace a regular block with a parameterized block (described below). This will also clear the contents of the block, as expected.

Parameterized blocks

The other type of block is called a parameterized block, and it looks like this:

block param_block_yo(headerText, text)
	h1= headerText
	p= text

Parameterized blocks do not render automatically because they require parameters. Therefore, assuming nothing else happens to the block, the block will not be rendered at all.

To render a block, use the "render" keyword like this:

render param_block_yo("Some header text", 'Some "paragraph" text')

Now, assuming nothing else happens to the block, the block will be rendered as:

<h1>Some header text</h1><p>Some &quot;paragraph&quot; text</p>

You can render as many times as you wish, and by default, the rendered content will be appended to the block. You can also prepend the rendered content to the block or replace the contents of the block with rendered content. Here are the variations:

Parameterized blocks are really cool because regular "append", "prepend", and "replace" all work, too. Just remember that order matters.

Another example:

!!! 5
html
	head
		block header(pageTitle)
			title= pageTitle
	body
		h1 Hello
		render header("Page Title")
		append header
			script(type="text/javascript")
		render header("Page Title")
		prepend header
			meta

Will output:

<!DOCTYPE html>
<html>
	<head>
		<meta/>
		<title>Page Title</title>
		<script type="text/javascript"></script>
		<title>Page Title</title>
	</head>
	<body>
		<h1>Hello</h1>
	</body>
</html>

Obviously, the example above is rather contrived, but it shows how blocks work. A more realistic example of using blocks is below.

What happens if I define the same block more than once?

You can re-define a block that has already been defined with another "block" statement. This completely destroys the previously defined block. Previously executed "append", "prepend", "replace", and "render" blocks do not affect the re-defined block.

In summary...

You may not render, append to, prepend to, or replace undefined blocks. If you do so, an error message will occur.

When you define a block within a function, and you output the function's rendered content to a variable, the defined block will be destroyed immediately after the function call.

Template Inheritance

There is no extends keyword. Just use blocks and includes:

layout.blade:

!!! 5
html
	head
		block title(pageTitle)
			title=pageTitle
		block scripts
			script(type="text/javascript" src="/js/jquery.min.js")
			script(type="text/javascript" src="/js/jquery-ui.min.js")
		block stylesheets
	body
		block body

homepage.blade:

include "layout.blade"
render title("Homepage")
replace block body
	h1 Hello, World

If you render layout.blade, you get:

<!DOCTYPE html>
<html>
	<head>
		<script type="text/javascript" src="/js/jquery.min.js"></script>
		<script type="text/javascript" src="/js/jquery-ui.min.js"></script>
	</head>
	<body></body>
</html>

but if you render homepage.blade, you get:

<!DOCTYPE html>
<html>
	<head>
		<title>Homepage</title>
		<script type="text/javascript" src="/js/jquery.min.js"></script>
		<script type="text/javascript" src="/js/jquery-ui.min.js"></script>
	</head>
	<body>
		<h1>Hello, World</h1>
	</body>
</html>

The idea here is that you can derive many pages in a website from a basic template. In this case, layout.blade provides the generic template for each page in your website. homepage.blade simply uses the layout and modifies some blocks to generate the actual page.

Preserve and constant regions

Preserve and constant regions are deprecated.

Preserve and constant regions are only useful when using a live page update engine. Anything in a "constant" region is marked by the live page update engine as a region that is not subject to re-rendering.

Example of constant region (using Meteor):

- console.log("Rendering header")
h1 This is a header
constant
	- console.log("Rendering constant...")
	p The current user is: #{Session.get("user")}

In the above example, "Rendering constant..." will only be printed once to the console, even if the Session variable changes.

"Preserve" regions can be used to preserve certain DOM elements during re-rendering, leaving the existing element in place in the document while replacing the surrounding DOM nodes. This means that re-rendering a template need not disturb text fields, iframes, and other sensitive elements it contains.

Example of preserve region (using Meteor):

form
	preserve {"input[id]": function (node) {return node.id;}}
		label First Name
		input#firstName(type="text")
		label Last Name
		input#lastName(type="text")

The example above will preserve the <input> DOM elements when the template is re-rendered. Notice that the code following the "preserve" keyword {"input[id]": function (node) {return node.id;}} is passed directly to the Landmark's "preserve" option (see [the Meteor documentation] (http://docs.meteor.com/#template_preserve) for more info).

You must specify how elements in a preserve region are to be preserved. To do this, you need to tell Blade how to uniquely identify each element that you want to preserve in the block. In the example above, <input> fields with id attributes are preserved, and their id is used to uniquely identify them.

The following is an example of preserving a single element. Since there is only one element in this preserve block, identifying it with the * selector is acceptable.

preserve ["*"]
	input(type="text")

If the "preserve" keyword is not followed by any code, then nothing is preserved.

Isolates

Isolates are deprecated.

Isolates are only useful when using a live page update engine. Creating an isolate ensures that if data dependencies relating only to that isolate are updated, then only the part of the template within isolate will be re-rendered. All other parts of the Blade template will not be re-rendered.

Example (using Meteor):

- console.log("Rendering header...")
h1 This is a header
isolate
	p The current user is: #{Session.get("user")}

In the example above, "Rendering header..." will be printed to the console when the whole template is rendered, but if the reactive variable is updated using Session.set("user", ...), only the isolate will be re-rendered. In this case, nothing will print to the console.

Note: As with Blade functions, any blocks defined within an isolate will be deleted and unaccessible outside the isolate block.

See Reactivity isolation on the Meteor documentation for more details.

Chunks

Chunks are no longer supported as of Blade 3.0.0beta. You should use isolates instead.

API

var blade = require('blade');

blade.compile(string, [options,] cb)

Compiles a Blade template from a string.

Note: if there is a problem with the Blade compiler, or more likely, if there is a syntax error with the JavaScript code in your template, Node.js will not provide any line number or other information about the error. See issue #40 for more details.

You can render a compiled template by calling the function: tmpl(locals, cb)

In addition, a compiled template has these properties and methods:

blade.compileFile(filename, [options,] cb)

Asynchronously compile a Blade template from a filename on the filesystem.

blade.renderFile(filename, options, cb)

Convenience function to compile a template and render it.

blade.middleware(sourcePath, options)

Express middleware for serving compiled client-side templates to the browser. For example, if you visit the URL "/views/homepage.blade" on your server, you can compile the view stored at sourcePath + "/homepage.blade"

{
	'cache': process.env.NODE_ENV == "production",
	'minify': process.env.NODE_ENV == "production",
	'includeSource': process.env.NODE_ENV == "development"
};

blade.Compiler

The compiler itself. It has some useful methods and properties.

blade.Compiler.parse(string)

Just generates the parse tree for the string. For debugging purposes only.

Example using the API:

var blade = require('blade');
blade.compile("string of blade", options, function(err, tmpl) {
	tmpl(locals, function(err, html) {
		console.log(html);
	});
});

Here is a sample Express application that uses Blade for server-side and client-side templates:

var express = require('express'),
	blade = require('blade');
var app = express.createServer();
app.use(blade.middleware(__dirname + '/views') ); //for client-side templates
app.use(express.static(__dirname + "/public") ); //maybe we have some static files
app.set('views', __dirname + '/views'); //tells Express where our views are stored
app.set('view engine', 'blade'); //Yes! Blade works with Express out of the box!
app.get('/', function(req, res, next) {
	res.render('homepage');
});
app.listen(8000);

Browser Usage

The Blade runtime should work on every browser, and since Blade provides an Express middleware for serving compiled templates to the browser ([see above] (#blademiddlewaresourcepath-options)), rendering Blade templates in the browser is a breeze.

Once you have the middleware setup, you can now serve your compiled Blade views to the client. Simply include the /blade/blade.js file in your <script> tags, and then call blade.Runtime.loadTemplate.

blade.Runtime.loadTemplate(filename, cb)

Your template will be stored in blade.cachedViews and will be cached until the user reloads the page or navigates to another page.

Yes, included files work, too. Like magic.

Example client-side JavaScript:

blade.Runtime.loadTemplate("homepage.blade", function(err, tmpl) {
	tmpl({'users': ['John', 'Joe']}, function(err, html) {
		console.log(html); //YAY! We have rendered HTML
	});
});

Additionally, you can set blade.Runtime.options to control how the templates are loaded:

As a side note, you can override the blade.Runtime.loadTemplate function with your own implementation.

Simple Example

The following Blade document ...

!!! 5
html
	head
		title Blade
	body
		#nav
			ul
				- for(var i in nav)
					li
						a(href=nav[i])= i
		#content.center
			h1 Blade is cool

... compiles to this JavaScript function ...

function tmpl(locals,cb,__){__=__||[],__.r=__.r||blade.Runtime,__.func||(__.func={},__.blocks={}),__.locals=locals||{};with(__.locals){__.push("<!DOCTYPE html>","<html",">","<head",">","<title",">","Blade","</title>","</head>","<body",">","<div",' id="nav"',">","<ul",">");for(var i in nav)__.push("<li",">","<a"),__.r.attrs({href:{v:nav[i],e:1}},__),__.push(">",__.r.escape(i),"</a>","</li>");__.push("</ul>","</div>","<div",' id="content"',' class="center"',">","<h1",">","Blade is cool","</h1>","</div>","</body>","</html>")}__.inc||__.r.done(__),cb(null,__.join(""),__)}

... now you call the function like this...

tmpl({
	'nav': {
		'Home': '/',
		'About Us': '/about',
		'Contact': '/contact'
	}
}, function(err, html) {
	if(err) throw err;
	console.log(html);
});

... and you get this (indented for readability):

<!DOCTYPE html>
<html>
	<head>
		<title>Blade</title>
	</head>
	<body>
		<div id="nav">
			<ul>
				<li><a href="/">Home</a></li>
				<li><a href="/about">About Us</a></li>
				<li><a href="/contact">Contact</a></li>
			</ul>
		</div>
		<div id="content" class="center">
			<h1>Blade is cool</h1>
		</div>
	</body>
</html>

Syntax Highlighting

There are a few resources available to get syntax highlighting for your favorite text editor.

If you find (or create yourself) syntax highlighting plugins for other text editors, please write me, and I will post the links here. Or, simply submit a pull request.

Plugins

Live UI - deprecated

Blade provides a Live UI plugin that allows Blade to use the Spark live page update engine independently from Meteor.

Live UI provides automatic two-way synchronization between your models and views on a given web page. That is, when data in your Model is updated, the rendered Blade views on the client's browser are automatically updated with the new content, and similarly, when a Blade view is rendered in the browser, the Blade event handlers can update data in the model.

Complete documentation for the Live UI plugin (including several examples) can be found on the [Live UI Plugin wiki page] (https://github.com/bminer/node-blade/wiki/Live-UI-Blade-Plugin).

definePropertyIE8

This plugin is a prerequisite for the Live UI plugin if you plan on using Live UI in Internet Explorer 8.

Meteor Support

Meteor support has been deprecated and will be removed in a future release.

Implementation Details

PEG.js

The Blade parser is built using PEG.js. Thanks to the PEG.js team for making this project much easier than I had anticipated! To modify the parser, simply change ./lib/parser/blade-grammer.pegjs, and the new parser will be automatically built the next time you run tests.

Running tests

To install all devDependencies, just do: npm link or install manually. To run tests, ensure devDependencies are installed, then run: npm test

Compiler-runtime relationship

Also, I'd like to mention here that the Blade compiler and Blade runtime are rather closely coupled. Unfortunately, that means that templates compiled with an older Blade compiler might not be compatible with a newer runtime and vice versa. To avoid issues, be sure that your Blade templates were compiled with the compiler of the same version as the runtime on which they will run. If you think this is too inconvenient, please feel free to complain, but I probably will ignore you. :)

<a name="fileIncludeDetails"></a> File Includes

Included Blade templates MUST be loaded synchronously, and if this is not possible, an error will be thrown. Obviously, when rendering views on the server, this is not a problem since Node provides synchronous file system calls; however, on the client, it is only possible to include a file synchronously when the file is already in the browser's cache. When the name of the file to be included is known at compile-time (i.e. you are not using a dynamic filename include), the compiler will notify the Blade middleware of a particular view's dependencies. This allows the client-side template loader to also load and cache any dependent views in advance, preventing any issues from occurring. Nevertheless, when dynamic filename includes are used, the compiler has no way of determining which views will be included at runtime, and if a dynamically included view is not loaded into the browser's cache when the include statement is reached, the included view must be be loaded asynchronously and, as such, an error will be thrown.

Loading and compiling files synchronously may temporarily reduce your application's responsiveness, but because compiled views are often cached, this is not really much of an issue.

Event Handlers

Event handlers in Blade work by injecting the event handler function as an HTML comment directly before the bound element. Then, the appropriate event attribute (i.e. onclick, onchange, etc.) on the element is set to call blade.Runtime.trigger. The trigger function basically grabs the HTML comment, passes the contents through eval(), and binds the event handler directly to the element. This means that the event handlers work on templates rendered on the browser or on the server. Everything gets wired up the first time that the event occurs on the browser.

The Blade runtime also keeps track of any event handlers bound to a specific element by assigning each element an 'id' attribute, if necessary. When the view has finished rendering, the Blade runtime will pass a bunch of information (blocks, functions, or event handlers that were defined, etc.) to the 3rd (undocumented) argument of the render callback function. If you are rendering Blade templates on the browser, you can access the list of event handlers and bind the defined event handler directly to the element by looking up the element by its 'id' instead of letting the trigger function do its magic. The advantage of binding direclty to the defined event handler is that (thanks to closures) you can still reference the locals that were passed to your view and modify them, as needed... directly from your event handler. This allows your view code to automatically synchronize with your model, providing one-way view-to-model synchronization capabilties. Very cool! For examples of this and for more information, check out the [Live UI plugin] (https://github.com/bminer/node-blade/wiki/Live-UI-Blade-Plugin).

Benchmarks

See the Benchmark wiki page for more information.

License

See the LICENSE.txt file.