Add HTML.unescape [Closes #3107]

[dukex]

Given the ruby implementation CGI::unescapeHTML and the comments on #3226 I'm made this implementation of HTML.unescape.

I tried to translate the unescape for hexadecimal codes without success
Someone can help me? Thanks @chaniks

[chaniks]

$1.to_i(16) for $1.hex.

[dukex]

❤️ @chaniks thank you

[asterite]

I think this should be n <= charlimit, since 0x10ffff is a valid unicode value

[dukex]

Done, thanks!

[asterite]

this dup doesn't seem to be needed

[dukex]

Ops, done!

[chaniks]

I guess I have too many thoughts on this PR.
I'd better skip them..

But two things,

• Integer overflow (e.g. �, &#x80000000)
•   (can be generated with HTML.escape—reversibility)

And maybe specifying behaviors like � => � and � => � in specs might be a good idea, if it is intended.
And maybe reusing Char::MAX_CODEPOINT.

I guess I should stop here.

[asterite]

@dukex Thank you for this! ❤️

[dukex]

Hi @chaniks

Fell free to send all your comments here, I want write the best code for the crystal and not just copy the ruby code, I appreciate your points here and I want read more.

Thanks for your comments,

I'm using the Char::MAX_CODEPOINT now, I searched for this constant but I maybe I would need to search more.

I update the code to unescape spaces( )

I think the the condition if n <= Char::MAX_CODEPOINT are handle with integer overflow error, you can explain more your comment?

You can help me with behaviors like � => � it is correct change value or the code should returns �?

[chaniks]

@dukex to_i will raise an error if the string representation is out of integer range. :')

  is actually not a space, but I believe people will report if it becomes an issue. (Sorry I wasn't specific. My writing skill is really bad.)

Please never worry about my other comments. I wanted to edit it, but it was a review so I couldn't.. 😢

p.s. If you want dig more, this link might help: https://www.w3.org/TR/html401/sgml/entities.html

[RX14]

here's a list of entities and their unicode code points. nbsp should certainly be fixed to be the correct codepoint, or removed.

[dukex]

@chaniks and @RX14 I'll address this bugs in my next PR, ok?