Add an option to disable markdown elements

[Guichaguri]

A simple change that let you disable any markdown element.
It's better creating an option than using unsafe tricks and hacks

Examples

marked.setOptions({
    disabledElements: ['hr', 'heading', 'list']
});

marked('---'); // Returns "---"
marked('# I am not a heading'); // Returns "# I am not a heading"
marked('* Item'); // Returns "* Item"

[Guichaguri]

I just found #579.

Differences

• This one works with GFM.
• It's an option instead of a new function.
• Can be applied to a single marked call (marked('---', { disableElements: ['hr'] }))

[Guichaguri]

Fixed merge conflicts

[joshbruce]

Thanks for the PR @Guichaguri! Much appreciated. Can you write the tests demonstrating the examples? (Note: I am wondering about the flexibility of our testing harness to actually test all the things we need it to.)

[Guichaguri]

How should I write it? After a quick look through the test code, it seems like it can only test with predefined options.

[joshbruce]

@Guichaguri: Let me tag in @UziTech, @matt-, and @Feder1co5oave to help on possibly getting a test written on this.

[UziTech]

Ya, the tests aren't very option friendly at the moment. I am working on redoing the tests to allow options to be set with front-matter

stay tuned...

[Feder1co5oave]

I think this feature is destructive: once you've disabled an element, you cannot re-enable it. Correct me if I'm wrong

[UziTech]

I submitted pr #1002 to fix the tests and allow options to be specified with front-matter

@Feder1co5oave it looks like you can re-enable all elements by setting the disabledElements option to an empty array (or removing the elements that you want to re-enable)

[Guichaguri]

Once the test PR gets merged, I'll write tests for all cases, including re-enabling disabled elements.

I should also write documentation for it.

[joshbruce]

@Guichaguri: Can you also handle @Feder1co5oave's question. I'm not sure I would choose the word "destructive" just because of connotation...maybe "global" - the entire markdown string being parsed??

Which I think is true, yeah?

It would be interesting if it could be open to extension a little bit...like "disable the frist 5 instances of H1" or "disable the first, second, and fifth H1". Having said that, given the developer requests to date, being able to stop Marked from rendering an element in a markdown is the request...not granular control.

Am I missing something in @Feder1co5oave's question.

[Feder1co5oave]

That's what I'm talking about:

> marked('# heading?')
'<h1 id="heading-">heading?</h1>\n'
> marked('# heading?', {disabledElements: ['heading']})
'<p># heading?</p>\n'
> marked('# heading?')
'<p># heading?</p>\n'
> marked('# heading?', {disabledElements: []})
'<p># heading?</p>\n'

[UziTech]

I was wrong, I didn't realize the rules were overwritten in the lexer. @Guichaguri maybe a better approach would be to copy the rules to this.rules only if they are not in this.options.disabledElements otherwise set the rule as noop. That way the static rule sets are left alone.

Also should it be called disabledRules instead of disabledElements. I feel like elements refer to tags like h1, h2, etc.

[joshbruce]

My two cents would be that, yes, "elements" are individual elements (skip h1) while "rules" are all the types (skip headings). Can we have the ability to specify both?

[joshbruce]

Closing due to merge conflicts and lack of tests. Should be easy enough to resurrect or continue the spirit of in 1.x or 2.x.

[sungwoncho]

What can do to help this feature get merged to master?

[joshbruce]

@sungwoncho: Hello! Right now our focus is on complying with our chosen specifications and that's where we need the most help. This feature, if approved, would not be put in until after 1.0 or later. Therefore, from an over-arching project perspective, if you are still interested in moving toward at least the possibility of having this feature include sooner rather than later, then I recommend helping us with the specification compliance mission of 0.x: #1216

Up to you! Thanks again.