Blade is a HTML Template Compiler, inspired by Jade & Haml, implemented in JavaScript, so it will run on your microwave oven.
Never write HTML again. Please.
"Blade's blood is the key" :P Sorry... I had to...
- Why use Blade instead of Jade?
- Features
- Project Status
- Installation
- Language Syntax
- API
- Browser Usage
- Implementation Details
- License
Here are the reasons Blade might be considered "better" than Jade:
- Jade is an ornamental stone. Blade is a badass vampire hunter.
- Client-side templates can be served to the browser, no problem. See Browser Usage and [Blade Middleware] (#blademiddlewaresourcepath-options) for more info.
- In Blade, file includes happen dynamically at run-time, instead of at compile-time. This means that files compiled in Blade are generally smaller than Jade files when you are using file includes. In addition, if you re-use the same included file among multiple parent views, the included file does not need to be re-compiled. This can significantly decrease the size of client-side templates, and reduce the overall bandwidth required to transfer the templates over the Internet.
- Blocks in Blade are awesome. We removed features from Jade like explicit template inheritance and static file includes and then added features like blocks and parameterized blocks. You might find our idea of a block to be similar to Jade's, but just wait until you realize how much more flexible they are!
- Just Functions, not mixins or partials. In Blade, there are no "mixins" or partial
templates. There are only functions, and they work just like regular JavaScript
functions that you've come to know and love. You can put your functions into separate
files and include them into other templates, you can take advantage of the
arguments
Array-like Object, closures (not necessarily recommended), or whatever you want! Can you define functions within other functions? Yep. - Load function output into a variable. Blade has a built-in syntax for taking content rendered by a function and loading into a variable within your view template. Then, you can pass the rendered HTML content to another function, for example.
OK... it's admittedly not as funny as I thought it would be. But, I tried.
- Write extremely readable short-hand HTML
- Insert escaped and unescaped text and vanilla JavaScript code
- Code and text are escaped by default for security/convenience
- Functions (like Jade mixins)
- Dynamic file includes
- Regular blocks and Parameterized blocks (aids in supporting template inheritance)
- True client-side template support with caching, etc.
- Supports Express.JS
- HTML Comments and block comments
- Text filters
Blade does more than Jade, and it does less than Jade. Your Jade templates will probably need some modifications before they will work with Blade.
I'd say that Blade is in beta. There are some [known issues] (https://github.com/bminer/node-blade/issues), and I would not recommend Blade for production environments. That being said, I am using Blade for a few projects, and it will be receiving more and more test coverage.
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.
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.
for Node (via npm): npm install blade
for Browsers:
Runtime only: wget https://raw.github.com/bminer/node-blade/master/dist/blade-runtime.min.js
Runtime is about 3-4 KB, uncompressed.
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. 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 in the vanilla JavaScript code, though. Blade uses whitespace to separate each attribute.
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>
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.
It works. 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
body
#content
renders as:
<html><head></head><body><div id="content"></div></body></html>
It works, too. Simply place content after the tag like this:
p This text is "escaped" by default. Kinda neat.
renders as:
<p>This text is "escaped" by default. Kinda neat.</p>
Want unescaped text? Large blocks of text? Done.
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 "escaped", btw)
Yep. It just works!
Neato.</p>
Rules are:
- Text is escaped by default
- Want unescaped text? Precede with a
!
- Large text block? Use
|
and indent properly. - Unescaped text block? Use
|!
or even just!
works.
Variable interpolation is supported for text blocks. Use #{var_name}
notation, and
anything between the curly braces is treated as vanilla JavaScript code.
For example, you can write:
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 value with value "Blade",
both of the examples above will render to this:
<p>I am just testing Blade, alright?
Relax...</p>
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:
- :nl2br - Escapes the content and converts newline characters to
<br/>
- :cdata - Surrounds text like this:
<![CDATA[
...text goes here...]]>
Text should not contain]]>
. - :markdown (must have markdown-js installed)
- :md (alias for :markdown)
- :coffeescript - Generates a
<script>
tag for the generated JavaScript. (must have coffee-script installed) - :cs (alias for :coffeescript)
- :stylus - Generates a
<style>
tag for the generated CSS. (must have stylus installed) - :less - Generates a
<style>
tag for the generated CSS. (must have less installed) - :sass - Generates a
<style>
tag for the generated CSS. (must have sass installed)
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.
Use dash (-
) to specify a code block. Use equals (=
) to specify code output. A few examples, please?
Code blocks:
#taskStatus
- if(task.completed)
p You are done. Do more! >:O
- else
p Get to work, slave!
Code that outputs (i.e. in a text block or at the end of a tag).
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
Extra "|" characters are okay, too. Just don't forget that stuff after the "=" means JavaScript code!
p
|= "escape me" + " away & away"
renders <p>escape me away & away</p>
Blade, like other template engines, defines local variables within every single view. You should avoid using these names in your view templates whenever possible:
locals
cb
__
(that's two underscores)- Any of the compiler options (i.e.
debug
,minify
, etc.)
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
which renders as <!DOCTYPE html PUBLIC "-//W3C//DTD XHTML Basic 1.1//EN>
Put the doctype at the top of your Blade files, please. Here is the list of built-in doctypes:
exports.doctypes = {
'5': '<!DOCTYPE html>',
'xml': '<?xml version="1.0" encoding="utf-8" ?>',
'default': '<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd">',
'transitional': '<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd">',
'strict': '<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Strict//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-strict.dtd">',
'frameset': '<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Frameset//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-frameset.dtd">',
'1.1': '<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.1//EN" "http://www.w3.org/TR/xhtml11/DTD/xhtml11.dtd">',
'basic': '<!DOCTYPE html PUBLIC "-//W3C//DTD XHTML Basic 1.1//EN" "http://www.w3.org/TR/xhtml-basic/xhtml-basic11.dtd">',
'mobile': '<!DOCTYPE html PUBLIC "-//WAPFORUM//DTD XHTML Mobile 1.2//EN" "http://www.openmobilealliance.org/tech/DTD/xhtml-mobile12.dtd">'
};
Yes, you can modify the list of built-in doctypes through the API. Why would you, though?
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="/ie-really-sux.js")
renders as:
<head><!--[if lt IE 8]><script src="/ie-really-sux.js"></script><![endif]--></head>
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>
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 will generate an error
include "file.blade"
This will dynamically (at runtime) insert "file.blade" right into the current view, as if it was a single file.
The include statement can also be followed by the name of a JavaScript variable containing the filename to be included.
- var filename = "file.blade"
include filename
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.
There are two types of blocks: regular blocks and parameterized 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:
- Use the
append
keyword to append to the matching block. - Use the
prepend
keyword to prepend to the matching block. - Use the
replace
keyword to replace the matching block.
Example:
append regular_block
p This is also a test
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).
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), you cannot call "render" on that block anymore.
At this time, you cannot replace any defined block with a parameterized block. The "replace" command will not accept parameters.
The other type of block is called a parameterized block, and it looks like this:
block regular_block(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 regular_block("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 "paragraph" text</p>
Parameterized blocks are really cool because "append", "prepend", and "replace" all work, too. Just remember that order matters. A render call always "appends" to the block, and you can render as many times as you wish.
Another example:
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:
<head>
<meta/>
<title>Page Title</title>
<script type="text/javascript"></script>
<title>Page Title</title>
</head>
<body>
<h1>Hello</h1>
</body>
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...
- Use the
block
keyword to mark where the block will go (block definition). - Use the
render
keyword to render the matching "parameterized" block. Do not use this on a regular block. - Use the
append
keyword to append to the matching block. - Use the
prepend
keyword to prepend to the matching block. - Use the
replace
keyword to replace the matching block.
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.
There is no extends
keyword. Just use blocks and includes:
layout.blade:
html
head
block title(pageTitle)
title=pageTitle
body
block body
homepage.blade:
include "layout.blade"
render title("Homepage")
replace block body
h1 Hello, World
If you render layout.blade, you get: <html><head></head><body></body></html>
, but if you
render homepage.blade, you get:
<html>
<head>
<title>Homepage</title>
</head>
<body>
<h1>Hello, World</h1>
</body>
</html>
var blade = require('blade');
Asynchronously compiles a Blade template from a string.
string
is a string of Bladeoptions
include:filename
- the filename being compiled (required when using includes or thecache
option)cache
- if true, the compiled template will be cacheddebug
- outputs debug info to the console (defaults to false)minify
- if true, Blade generates a minified template without debugging information (defaults to false)includeSource
- if true, Blade inserts the Blade source file directly into the compiled template, which can further improve error reporting, although the size of the template is increased significantly. (defaults to true if and only ifprocess.env.NODE_ENV
is "development" and minify is false; defaults to false, otherwise)doctypes
- use this Object instead ofblade.Compiler.doctypes
inlineTags
- use this array instead ofblade.Compiler.inlineTags
selfClosingTags
- use this array instead ofblade.Compiler.selfClosingTags
filters
- use this Object instead ofblade.Compiler.filters
templateNamespace
- the name of the reserved variable in the view (defaults to two underscores: __). Other reserved names are listed here
cb
is a function of the form:cb(err, tmpl)
whereerr
contains any parse or compile errors andtmpl
is the compiled template. If an error occurs,err
may contain the following properties:message
- The error messageexpected
- If the error is a 'SyntaxError', this is an array of expected tokensfound
- If the error is a 'SyntaxError', this is the token that was foundfilename
- The filename where the error occurredoffset
- The offset in the string where the error occurredline
- The line # where the error occurredcolumn
- The column # where the error occurred
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. At the time of this writing, this is a limitation of the Google V8 engine.
You can render a compiled template by calling the function: tmpl(locals, cb)
locals
are the local variables to be passed to the view templatecb
is a function of the formfunction(err, html)
whereerr
contains any runtime errors andhtml
contains the rendered HTML.
In addition, a compiled template has these properties and methods:
template
- a function that also renders the template but accepts 3 parameters:tmpl.template(locals, runtime, cb)
. This simply allows you to use a custom runtime environment, if you choose to do so.filename
- the filename of the compiled template (if provided)toString()
- a function that converts the view template function into a string of JavaScript code. If you need a client-side template for example, you can use this function. Uglify-JS is now used if you have it installed.
Asynchronously compile a Blade template from a filename on the filesystem.
filename
is the filenameoptions
- same asblade.compile
above, exceptfilename
option is always overwritten with thefilename
specified.cb
- same asblade.compile
above
Convenience function to asynchronously compile a template and render it.
filename
is the filenameoptions
- same asblade.compileFile
above. This object is also passed directly to the view, so it should also contain your view's local variables.cb
- a function of the formfunction(err, html)
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"
sourcePath
- the path on the server where your views are storedoptions
include:mount
- the URL path where you can request compiled views (defaults to "/views/")clientNamespace
- the variable where downloaded client-side templates are stored (defaults to "blade.templates").blade.templates
is not defined by the Blade runtime, so be sure to define it yourself.clientCache
- turns on client-side caching of views (defaults toprocess.env.NODE_ENV == "production"
). Caching works until the user navigates to another page. Then, you have to rely on the middleware to do things like [weak and strong caching] (https://developers.google.com/speed/docs/best-practices/caching).compileOptions
- options passed toblade.compile()
. Defaults to:
{
'cache': process.env.NODE_ENV == "production",
'minify': process.env.NODE_ENV == "production",
'includeSource': process.env.NODE_ENV == "development"
};
The compiler itself. It has some useful methods and properties.
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') );
app.use(express.static(__dirname + "/public") );
app.set('views', __dirname + '/views');
app.set('view engine', 'blade');
app.get('/', function(req, res, next) {
res.render('homepage');
});
app.listen(8000);
The Blade compiler doesn't work on browsers yet, but the runtime should work on every browser. That means that you can compile your templates on the server and serve them up to any browser. Blade provides a built-in Express middleware to do just that (see above).
Once you have the middleware setup, you can now serve your compiled Blade views
to the client. Simply include the blade-runtime.js file in your <script>
tags, and then call blade.runtime.loadTemplate
.
baseDir
- just put an empty string here. This argument is ignored.filename
- the filename of the view you wish to retrieve, relative to thesourcePath
you setup in the Blade middleware.compileOptions
- just put an empty Object here. This argument is ignored for now.cb
- your callback of the formcb(err, tmpl)
Your template will be stored in blade.templates
or whatever you put as the
clientNamespace
when you setup the Blade middleware.
Yes, included files work, too. Like magic.
Example client-side JavaScript:
blade.templates = {};
blade.runtime.loadTemplate("", "homepage.blade", {}, function(err, tmpl) {
tmpl({}, function(err, html) {
console.log(html); //YAY! We have rendered HTML
});
});
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.
To install all devDependencies, just do: npm link
or install manually
To run tests, ensure devDependencies are installed, then run: npm test
See the LICENSE.txt file.