Whitespace prevents random component to be interpreted about markdown-to-jsx HOT 4 CLOSED

Released as 6.5.1, thanks for your contribution!

from markdown-to-jsx.

After further investigations, it seems a more global problem:

These markdown input render differently:

<div class="green">
   Hello
</div>

<div class ="green">
   Hello
</div>

<div class= "green">
   Hello
</div>

<div class = "green">
   Hello
</div>

if this is parsed using regex, would it be possible to add something like:

/ *= */

if yes, I'd be glad to help

from markdown-to-jsx.

Got into the code and found these bits:

const ATTR_EXTRACTOR_R = /([-A-Z0-9_:]+)(?:\s*=\s*(?:(?:"((?:\\.|[^"])*)")|(?:'((?:\\.|[^'])*)')|(?:\{((?:\\.|{[^}]*?}|[^}])*)\})))?/gi;

and

function attrStringToMap (str) {
        const attributes = str.match(ATTR_EXTRACTOR_R);

        return attributes ? attributes.reduce(function (map, raw, index) {
            const delimiterIdx = raw.indexOf('=');

            if (delimiterIdx !== -1) {
                const key = normalizeAttributeKey(raw.slice(0, delimiterIdx));
                const value = unquote(raw.slice(delimiterIdx + 1));
...

Should be where the planets align together.
Now, I could suggest another approach to prevent parsing attributes twice. In fact, the regex oddly catches relevant groups into one single part, then it is parsed again using indexOf('='), right ?
Couldn't we leverage regex groups to naturally handle attributes key/value split for us ? something like:

/([-A-Z0-9_:]+)\s*=\s*(?:"([^"]+)"|'([^']+)'|\{([^{]+)\})/gi

where 4 groups can be extracted: attribute key, double quoted value, single quoted value and curly brackets enclosed value. Then unquote would not even be needed.

What do you think ?

from markdown-to-jsx.

Submitted PR in a very lightweight fashion (by just trimming down spaces before parsing) + corresponding tests :)

from markdown-to-jsx.