Fix: casing of unicode characters

[armano2]

#688

[armano2]

What i should do with this than? Remove support for utf-8 and add rule to ban non a-zA-Z characters instead?

i feel like its base on preference, some ppl may want to use non ascii characters like #688

[mysticatea]

I think that:

• the utilities have to behave as same as Vue.js's.
• the rule which disallows non-ascii upper case characters in names would be nice because it can cause unexpected behavior.

[michalsnik]

LGTM 👍 Good work @armano2. Now we can also add rule to prevent usage of non-ascii characters

[michalsnik]

@mysticatea or @ota-meshi can you also take a look at the code here? I would appreciate second pair of eyes before merging it :)

[mysticatea]

My opinion is not changed since #694 (comment). I meant that our utilities should behave as same as https://github.com/vuejs/vue/blob/dev/src/shared/util.js. That doesn't have special handling for non-ASCII characters.

[armano2]

@mysticatea what should happen with non A-Za-z0-9: characters than?

current implementation is removing them,

const invalidChars = /[^a-zA-Z0-9:]+/g

---

anyway new implementation is working better even without unicode characters

i fixed also issue with assert.equal(converter(' foo Bar '), 'fooBar') in camelcase fooBar instead of FooBar

[mysticatea]

@armano2 Why does it remove those?

[armano2]

no clue, it was copied from vue :)

[armano2]

i will replace it with

.replace(/[\-!#$%^&*()_+~`"'<>,./?\[\]{}\s\r\n\v\t]+/gu, ' ')

[armano2]

and now it's ignoring unknown (non ascii) characters, and we will have to add new rule to warn about usage of non ascii characters.

[mysticatea]

no clue, it was copied from vue :)

In that case, I think we should keep it as is. My concern is that static analysis gets different behavior to Vue.js compiler and overlooks errors.

[armano2]

@mysticatea but in vue its used in different context https://github.com/vuejs/vue/blob/a7658e03a16dc507f0abeba41aee705f773727d0/src/core/util/options.js#L260
and from what i see it got removed anyway

[mysticatea]

My question in #694 (comment) was "why remove?". But https://github.com/vuejs/vue/blob/a7658e03a16dc507f0abeba41aee705f773727d0/src/core/util/options.js#L260 is not removal. So I should ask again, why does it remove those? If the removal is different behavior to Vue.js compiler, we should not remove those characters.

[armano2]

that's why i changed it now, we are removing now only invalid characters

\-!#$%^&*()_+~`"'<>,./?\[\]{}\s\r\n\v\t

those characters can't be used as attribute names and property names.

[mysticatea]

Why does remove invalid characters? I think that we should handle invalid names as invalid names because this is an error checker. It should not hide errors. If Vue.js compiler removes the invalid characters, yes, we should follow that. But if not, we should follow that.

[armano2]

we have to somehow determine when word ends, and those are characters on which we split text, vue compile is never removing invalid characters, it's going to be runtime error..

[armano2]

@mysticatea than should i change this to only check whitespaces? /\s/ug