Thomas Rake 77d15cf099 Update to amber 0.14.3, node_modules added to .gitignore, trysmalltalk 0.1.1,fix copyright dates. | 9 years ago | |
---|---|---|
.. | ||
compressed | 9 years ago | |
src | 9 years ago | |
.bower.json | 9 years ago | |
.jshintignore | 9 years ago | |
.travis.yml | 9 years ago | |
Gruntfile.js | 9 years ago | |
README.md | 9 years ago | |
bower.json | 9 years ago | |
license.txt | 9 years ago | |
package.json | 9 years ago |
A JavaScript port of Markdown
Showdown is now maintained by the showdownjs organization on Github.
The organization needs members to maintain Showdown.
Please see this issue to express interest or comment on this note.
Showdown Copyright (c) 2007 John Fraser. http://www.attacklab.net/
Original Markdown Copyright (c) 2004-2005 John Gruber http://daringfireball.net/projects/markdown/
Redistributable under a BSD-style open source license. See license.txt for more information.
var Showdown = require('showdown');
var converter = new Showdown.converter();
converter.makeHtml('#hello markdown!');
// <h1 id="hellomarkdown">hello, markdown</h1>
Developers can use Showdown to:
Add in-browser preview to existing Markdown apps
Showdown's output is (almost always) identical to markdown.pl's, so the server can reproduce exactly the output that the user saw. (See below for exceptions.)
Add Markdown input to programs that don't support it
Any app that accepts HTML input can now be made to speak
Markdown by modifying the input pages's HTML. If your
application lets users edit documents again later,
then they won't have access to the original Markdown
text. But this should be good enough for many
uses -- and you can do it with just a two-line
onsubmit
function!
Add Markdown input to closed-source web apps
You can write bookmarklets or userscripts to extend any standard textarea on the web so that it accepts Markdown instead of HTML. With a little more hacking, the same can probably be done with many rich edit controls.
Build new web apps from scratch
A Showdown front-end can send back text in Markdown, HTML or both, so you can trade bandwidth for server load to reduce your cost of operation. If your app requires JavaScript, you won't need to do any Markdown processing on the server at all. (For most uses, you'll still need to sanitize the HTML before showing it to other users -- but you'd need to do that anyway if you're allowing raw HTML in your Markdown.)
Showdown has been tested successfully with:
In theory, Showdown will work in any browser that supports ECMA 262 3rd Edition (JavaScript 1.5). The converter itself might even work in things that aren't web browsers, like Acrobat. No promises.
Showdown allows additional functionality to be loaded via extensions.
<script src="src/showdown.js" />
<script src="src/extensions/twitter.js" />
var converter = new Showdown.converter({ extensions: 'twitter' });
// Using a bundled extension
var Showdown = require('showdown');
var converter = new Showdown.converter({ extensions: ['twitter'] });
// Using a custom extension
var mine = require('./custom-extensions/mine');
var converter = new Showdown.converter({ extensions: ['twitter', mine] });
In most cases, Showdown's output is identical to that of Perl Markdown v1.0.2b7. What follows is a list of all known deviations. Please file an issue if you find more.
This release uses the HTML parser from Markdown 1.0.2b2,
which means it fails Inline HTML (Advanced).text
from
the Markdown test suite:
<div>
<div>
unindented == broken
</div>
</div>
Showdown doesn't support the markdown="1" attribute:
<div markdown="1">
Markdown does *not* work in here.
</div>
This is half laziness on my part and half stubbornness. Markdown is smart enough to process the contents of span- level tags without screwing things up; shouldn't it be able to do the same inside block elements? Let's find a way to make markdown="1" the default.
You can only nest square brackets in link titles to a depth of two levels:
[[fine]](http://www.attacklab.net/)
[[[broken]]](http://www.attacklab.net/)
If you need more, you can escape them with backslashes.
When sublists have paragraphs, Showdown produces equivalent HTML with a slightly different arrangement of newlines:
+ item
- subitem
The HTML has a superfluous newline before this
paragraph.
- subitem
The HTML here is unchanged.
- subitem
The HTML is missing a newline after this
list subitem.
Markdown.pl creates empty title attributes for inline-style images:
Here's an empty title on an inline-style
![image](http://w3.org/Icons/valid-xhtml10).
I tried to replicate this to clean up my diffs during testing, but I went too far: now Showdown also makes empty titles for reference-style images:
Showdown makes an empty title for
reference-style ![images][] too.
[images]: http://w3.org/Icons/valid-xhtml10
With crazy input, Markdown will mistakenly put
<strong>
or <em>
tags in URLs:
<a href="<*Markdown adds em tags in here*>">
improbable URL
</a>
Showdown won't. But still, don't do that.
A suite of tests is available which require node.js. Once node is installed, run the following command from the project root to install the development dependencies:
npm install --dev
Once installed the tests can be run from the project root using:
npm test
New test cases can easily be added. Create a markdown file (ending in .md
) which contains the markdown to test. Create a .html
file of the exact same name. It will automatically be tested when the tests are executed with mocha
.
A showdown extension is simply a function which returns an array of extensions. Each single extension can be one of two types:
^^youtube http://www.youtube.com/watch?v=oHg5SJYRHA0
to automatically render as an embedded YouTube video, that would be a language extension.<div class="header">
to be <header>
, that would be an output modifier.Each extension can provide two combinations of interfaces for showdown.
Regex/replace style extensions are very similar to javascripts string.replace
function. Two properties are given, regex
and replace
. regex
is a string and replace
can be either a string or a function. If replace
is a string, it can use the $1
syntax for group substitution, exactly as if it were making use of string.replace
(internally it does this actually); The value of regex
is assumed to be a global replacement.
Example:
var demo = function(converter) {
return [
// Replace escaped @ symbols
{ type: 'lang', regex: '\\@', replace: '@' }
];
}
Alternately, if you'd just like to do everything yourself, you can specify a filter which is a callback with a single input parameter, text (the current source text within the showdown engine).
Example:
var demo = function(converter) {
return [
// Replace escaped @ symbols
{ type: 'lang', filter: function(text) {
return text.replace(/\\@/g, '@');
}}
];
}
One bit which should be taken into account is maintaining both client-side and server-side compatibility. This can be achieved with a few lines of boilerplate code. First, to prevent polluting the global scope for client-side code, the extension definition should be wrapped in a self-executing function.
(function(){
// Your extension here
}());
Second, client-side extensions should add a property onto Showdown.extensions
which matches the name of the file. As an example, a file named demo.js
should then add Showdown.extensions.demo
. Server-side extensions can simply export themselves.
(function(){
var demo = function(converter) {
// ... extension code here ...
};
// Client-side export
if (typeof window !== 'undefined' && window.Showdown && window.Showdown.extensions) { window.Showdown.extensions.demo = demo; }
// Server-side export
if (typeof module !== 'undefined') module.exports = demo;
}());
The showdown test runner is setup to automatically test cases for extensions. To add test cases for an extension, create a new folder under ./test/extensions
which matches the name of the .js
file in ./src/extensions
. Place any test cases into the filder using the md/html format and they will automatically be run when tests are run.