Oh ! I forgot the AST.
With this one, i must edit parse5 :
root[2] (1:1-9:1, 0-81)
├─ form[1] (2:1-4:10, 1-38) [data={"hName":"form"}]
│ └─ textarea[1] [data={"hName":"textarea"}]
│ └─ text: "This is a textarea"
└─ html: "<div>\nThis is `dangerous HTML` 🐯\n</div>" (6:1-8:7, 40-80)With that one(hand written), no need to edit parse5 :
root[2] (1:1-9:1, 0-81)
├─ html: "<textarea>This is a textarea</textarea>" (2:1-4:10, 1-38)
└─ html: "<div>\nThis is `dangerous HTML` 🐯\n</div>" (6:1-8:7, 40-80)It's why i don't think the bug comes from parse5 but from rehype raw.
Oh sorry ! I tried my best to be understandable but i'm not used to remark, nor gitter, nor english in fact ... :flushed:
All the code is here http://git.ache.one/markdown-viewer/blob/master/app.js
Important part :
unified()
.use(remark)
.use(mermaid)
.use(textInput) // The plugin who deal with the textarea
.use(math)
.use(kbd)
.use(highlight)
.use(html, {allowDangerousHTML: true}) // This is remark-rehype
.use(rehypeKatex)
.use(raw)
.use(rehypeStringify)
.process(data, // [...]The code source of the plugin : http://git.ache.one/remark-text-input/
I don't think there is a most important part but it's only 64 lines long. The plugin is about creating some interactives stuff from a Markdown document (in a same manner as Markdown)
Maybe, i will check it this nigth and edit this text if i find something revelant. As i say i'm not used to remark.
'html' nodes from your plugin though, instead opting for embedded HAST. remark-math and remark-kbd do so as well: https://github.com/Rokt33r/remark-math/blob/0a0a0f826ff26ea51776faa1576d5f7ee9d1239a/packages/remark-math/block.js#L181-L194
children in that case (which is for markdown nodes), but you should be using data.hChildren (for the embedded HAST children)
mdast-util-to-hast docs for more info! https://github.com/syntax-tree/mdast-util-to-hast#note
Hi @wooorm, I have a question regarding Remark. The mdast-util-to-hast plugin will create raw nodes when there is html element in the markdown parse with remark. As a result hast-util-sanitize will ignore those raw nodes and strip the html element, leaving only the text without markup.
I saw that there is a rehype-rawplugin that rehidrate the raw, yet it will mean I have to use remark-to-rehype somehow.
Is there a better way to reprocess an hast and convert raw nodes that contains html to hast-html-element?
[^footnote]: becomes [^footnote]:
$ echo '[^footnote]:' | remark
[^footnote]:
<stdin>: no issues found
--settings footnotes:true
let g:formatdef_remark_markdown = '"remark --quiet --no-color | sed ''s/:/\:/g''"'
footnotes:true isn't the default?
--settings documented somewhere?
const markdown = remark()
.use(toc)
.stringify(mdast) // The markdown AST
So I need to have different tags to apply a different style