parser, fmt, gen: support js"string literal"

[StunxFS]

This PR aims to add native JS string support to the compiler. It will now be possible to write js"Hello" just like you would write c"Hello".

This test ran successfully locally (using JS Browser):

fn test_js_string() {
	s := js'hello V'
	assert s.charAt(0) == js'h'
	assert s.charAt(6) == js'V'
	assert s.charCodeAt(0) == JS.Number(104)
	assert s.toUpperCase() == js'HELLO V'
	assert s.toLowerCase() == js'hello v'
	assert s.concat(js' from JS') == js'hello V from JS'
	assert s.includes(js' ') == JS.Boolean(true)
	assert s.startsWith(js'hello') == JS.Boolean(true)
	assert s.endsWith(js'V') == JS.Boolean(true)
}

[huly-for-github]

Connected to Huly®: V_0.6-23012

[enghitalo]

What is a JS string?

[JalonSolov]

JavaScript has 2 types of strings - primitive strings and String objects. Primitive strings do not have any methods or properties, while String objects have multiples of both.

Basically, it is the same as the difference between C strings and V strings.

[spytheman]

Why uncomment it later, and not now?
V is bootstrapped in the PR, and the test should be runnable imho as it is?

[spytheman]

Good work.
@medvednikov what do you think about it (adding syntax support for js'strings') ?

[Delta456]

There's no need to have the default value ast.void_type

[StunxFS]

For some strange reason, test_js_string fails with v test. The error it throws is the following:

>> compilation failed:
jsgen error: could not generate string method voidptr_str for type 'voidptr'

If I generate .js from a normal file with the same content, everything compiles and runs perfectly:

$ cat main.v
fn main() {
        s := js'hello V'
        assert s.charAt(0) == js'h'
        assert s.charAt(6) == js'V'
        assert s.charCodeAt(0) == JS.Number(104)
        assert s.toUpperCase() == js'HELLO V'
        assert s.toLowerCase() == js'hello v'
        assert s.concat(js' from JS') == js'hello V from JS'
        assert s.includes(js' ') == JS.Boolean(true)
        assert s.startsWith(js'hello') == JS.Boolean(true)
        assert s.endsWith(js'V') == JS.Boolean(true)
}
$ v -b js main.v
$ node main.js
$ 

[JalonSolov]

First impression... it's related to line 152 in checker.v, where you explicitly set the type to ast.void_type.

[StunxFS]

@JalonSolov

First impression... it's related to line 152 in checker.v, where you explicitly set the type to ast.void_type.

If I remove the default value from the field, V panics:

V panic: table.sym: invalid type (typ=ast.Type(0x0 = 0) idx=0). Compiler bug. This should never happen. Please report the bug using `v bug file.v`.

[medvednikov]

Looks good to me. A simple change.

[westtrade]

Looks cool, but for what purposes is this necessary?