To continue develop I should clean source code:

• functions and variables name
• sync coding order with original PHP version
• comment for each part

When writing HTML tags inside a paragraph, the tag isn't seen as a tag when it contains attributes. For instance, this will not work (the tag gets escaped):

<img src="blah.png" class="abc">

But interestingly, this will work and pass the tag as-is in the HTML output:

<img>

The problem seems to be in parseSpan. You're suing (?= in two places for non-capture blocks. The original in PHP Markdown Extra uses (?> which consumes the input, unlike (?=. As a general rule you should probably use (?: to replace (?> in JS rather than the non-consuming (?=.

Reference to PHP's once-only patterns (?>: http://www.php.net/manual/en/regexp.reference.onlyonce.php

__strong__, __double__snake, is__double__snake, endsnake__double__

renders as

<strong>strong</strong>, __double__snake, is__double__snake, endsnake<strong>double</strong>

although in php markdown you get the following:

<strong>strong</strong>, <strong>double__snake, is__double__snake, endsnake__double</strong>

Reference-style labels (titles are optional):
An [example][id]. Then, anywhere  
else in the doc, define the link:

  [id]: http://example.com/  "Title"

The above works fine, but when you remove the [example] and just leave the [id] it doesn't work for some strange reasons.
I've tested it in your preview website (http://tanakahisateru.github.com/js-markdown-extra/demo.html), and it doesn't work there either. I found the corresponding code, but i can't figure out why it doens't work:

function _DoAnchors_reference_callback( $0, $1, $2, $3 ) {
        var whole_match = $1;
        var link_text = $2;
        var link_id = $3.toLowerCase( );
        var result = "";

        if( link_id == "" ) {
            link_id = link_text.toLowerCase( );
        }
...

Code ran though the parser indented with the actual tab character (rather than of four spaces) is interpreted just as text rather than code.

The Demo is not working in current Chrome and probably other current Browsers too. That is because you serve the JavaScript directly from raw.github.com. You should probably serve it from github pages where your HTML/CSS is located too.

See related issue in Chromium tracker for more information.

the source is now available on http://www.jsdelivr.com/#!js-markdown-extra

btw: thx for this ;)

Hi!

I think the fix for #27 (which worked) introduced a new bug with inline code blocks.

input     :  `<script></script>`
expected  :  <p><code>&lt;script&gt;&lt;/script&gt;</code></p>
actual    :  <p>`</p>\n\n<script></script>\n\n<p>`</p>

Can you confirm this?

Many thanks!
Phil

For ease to use, I will publish the module to npm.

This would be such a shame, as there are no other Markdown Extra JS plugins.

Currently so many functions are defined in global scope. It should be minimized.

How difficult would it be to add to each generated element, the source line number ? With this information, I could automatically sync the scroll of the editor and the preview in a Markdown editor.

The source line number could be added as a class, for example: class="src-line28". Or as a data attribute: data-src-line="28".

Regexp for fencedCodeBlock does not follow PHP-Markdown implementation correctly.
This issue is follow-up to #18 .

I was using this lib on server-side with node js without any issues.
But the new version won't work at all.
The markdown seems completely ignored by the lib.

I believe it should be JavaScript.

js-markdown-extra is a experimental JapaScript port of PHP Markdown Extra.

Single char codespans, like this:

`x`

will not get a codespan.

I think fenced codeblocks are missing :(

Unlike php markdown extra lists dosen't need a blank line before the list items to be rendered in your version. I don't know if this behavior is expected.

This script doesn't support Google Chrome (or WebKit).

compile() of RegExp object is not supported by Chrome currently...

It seems that your code doesn't support Markdown Extra Abbreviations yet.

http://michelf.com/projects/php-markdown/extra/#abbr

Normal code blocks work as expected:

console.log(Markdown("foo\r\n\r\n    <script>console.log('heya');</scr" + "ipt>"));

... leads to ...

<p>foo</p>
<pre><code>&lt;script&gt;console.log('heya');&lt;/script&gt;
</code></pre>

The script tag is properly escaped. But when using fenced code blocks, that tag is not escaped:

console.log(Markdown("~~~\r\n<script>console.log('heya');</scr" + "ipt>\r\n~~~"));

... leads to ...

<pre><code><br /><br /><script>console.log('heya');</script>
</code></pre>

I'll try to come up with a patch, but I wasn't successful yet. Any help is appreciated.

Best
Phil

I passed to port auto link converter for mailto:. Considering use case of this library, such encryption is needed really?

I've made an copy/paste text from Github markdown.

There's an infinite loop with this text :

```php
function test() { echo "test".$2; };
``

You can reproduce with this (incomplete) test :

QUnit.test("Bug infinite loop", function() {
    var input =
        '```php' + "\n" +
        'function test() { echo "test".$2; }' + "\n" +
        '``' + "\n" +
        "";

    var expected =
        "";

    var result = Markdown(input); // infinite loop
    QUnit.assert.equal(result, expected);
});

Edit: this bug is only for Chrome and Node

やっつけで作ったものですが、よかったらどうぞ

• エラーチェックサボってます。
• コマンドラインオプション足すと、もっとそれっぽくなるかも

var md = require("js-markdown-extra")
var fs = require('fs')

if (process.argv.length < 2) {
    console.error("Must not pass a markdown source")
}

var filePath = process.argv[2]
fs.stat(filePath, function(err, stat) {
    if (! stat.isFile()) return;

    fs.readFile(filePath, "utf-8", function(err, text) {
        console.log(md.Markdown(text))
    })
})

The current version doesn't support Markdown Extra's special attributes. For example, the following Markdown should produce the subsequent HTML.

~~~ .javascript
var x = 7;
~~~

<pre><code class="javascript">var x = 7;</code></pre>

If only half of an emphasis/strong tag is entered, the text before it is deleted, and the text after it is duplicated.

One two _three four five

should render as

<p>One two _three four five</p>

but instead renders as

<p>three four five.three four five</p>

undefined shown before code span in Chrome.

The trailing comma on line 97 is unnecessary doesn't need to (and probably shouldn't) be there.

this.block_gamut = [
     ['doHeaders',         10],
     ['doHorizontalRules', 20],
     ['doLists',           40],
     ['doCodeBlocks',      50],
     ['doBlockQuotes',     60],  /*this is line 97*/
];

.. If the first item in the list is preceded by a blank line. For example:

term

 : definition

term

 : definition

Becomes:

<dl>
<dt>term</dt>
<dd>
<p>definition</p>
</dd>
</dl>

<dl>
<dt>term</dt>
<dd>
<p>definition</p>
</dd>
</dl>

It wraps the definitions in <p> tags correctly, but doesn't recognize that they're part of the same list. The fix is simple enough though. There's an extra space in your regex:

'(?:\\S.*\\n )+?' + // defined term

I believe it should be:

'(?:\\S.*\\n)+?' + // defined term

Could someone give me an (or a few different) example of text that triggers the infinite loop ?

I just noticed, that you can't use the same footnote reference twice. This behaviour is actually the same as in PHP Markdown Extra, but it has also been discussed over there:

    - refer to footnote A[^1]
    - refer to footnote B[^2]
    - refer to footnote A again[^1]

[^1]: footnote A
[^2]: footnote B

The "refer to footnote A again[^1]" line doesn't get linked to footnote A.