rlidwka / render-readme Goto Github PK

Pulled out of an example containing a multipart/form-data upload. As more dashes are added the rendering seems to take exponentially more time.

→ time node -e 'require("render-readme")("\`\`\`\n------------\n\`\`\`")'                                      
node -e   0.48s user 0.07s system 102% cpu 0.541 total
→ time node -e 'require("render-readme")("\`\`\`\n------------------------------\n\`\`\`")'
node -e   60.09s user 1.42s system 105% cpu 58.180 total

The sanitization is a bit overzealous, which can remove real content. Example:

render('```sh\n$ gulp test\n```');

Yields

<pre><code class="lang-sh">$ gulp \n</code></pre>\n

The word "test" is removed because pre-sanitization it looks like...

<pre><code class="lang-sh">$ gulp <span class="hljs-built_in">test</span>\n</code></pre>\n

Note the underscore in the hljs-built_in class, which doesn't match the regexp in the sanitizer.

Hi,

It looks like render-readme does not render a header line if the previous line contains any text. Every other Mark Down render that I have seen (GitHub, MacDown, WebStorm) renders this scenario by adding the <h#> tag on the header line. I believe this is the expected result.

Render Readme Example

//Multiple new line examples
console.log(require('render-readme')('<a name="MyClass"></a>\n\n' +
    '## MyClass(config)'));
/*
Expected
 <p><a name="MyClass"></a></p>
 <h2><a name="myclass-config-"></a>MyClass(config)</h2>
 */

console.log(require('render-readme')('Some Text\n\n' +
    '## MyClass(config)'));
/*
Expected
 <p>Some Text</p>
 <h2><a name="myclass-config-"></a>MyClass(config)</h2>
 */

//only one new line examples
console.log(require('render-readme')('<a name="MyClass"></a>\n' +
    '## MyClass(config)'));
/*
Not Expected
 <p><a name="MyClass"></a>
 ## MyClass(config)</p>
 */

console.log(require('render-readme')('Some Text\n' +
    '## MyClass(config)'));
/*
Not Expected
 <p>Some Text
 ## MyClass(config)</p>
 */

GitHub example

some text followed by
###### header on the next line

some text follow by

header on the next line

Related to #4 long strings of dashes in code blocks seem to become a single dash

→ node -e 'console.log(require("render-readme")("\`\`\`\n------------------------------\n\`\`\`"))'       
<pre><code>-</code></pre>

they come through fine from markdown-it

→ node -e 'console.log(require("markdown-it")().render("\`\`\`\n------------------------------\n\`\`\`"))'
<pre><code>------------------------------
</code></pre>

Hi,

I would like to have base64 images, so I changed the allowedSchemes.

allowedSchemes    : $w('http https mailto github-windows github-mac data'),

I would recommend to enable this for everyone using this module. What do you think?

Hello!

Thanks for putting this package together! I've run across an issue where the heading anchor tags generated do not quite match what Github generates. This causes some issues when where we can't have links between pages that work in both Github as well as on our site.

There are two distinct issues:

1. Punctuation is transformed into dashes when it should be filtered

• Example: ## ((test))
  • markup anchor: test
  • render-readme anchor: --test--

1. Duplicate anchors are not handled
   • If two anchor names are duplicated, Github will make them unique
   • Example: ## test\n ## test\n ## test\n
     • markup: test test-1 test-2
     • render-readme: test test test

Here is the ruby code behind generating anchors:
https://github.com/jch/html-pipeline/blob/master/lib/html/pipeline/toc_filter.rb#L35

I think filtering out punctuation in the same way (/[^\w\- ]/) will solve number 1 above.

Number 2 above will need some extra code...

Would you be willing to accept PRs for these updates?

We use AsciiDoc for all textual documents, including readmes in npm modules. It would be great if Sinopia can render even AsciiDoc. There’s a JS version of Asciidoctor, asciidoctor.js, that can be used for it.

I’d like to implement it, but this module doesn’t seem to be prepared extending for other markups…?