Value literal inference

[nnnnblaser]

No description provided.

[5nord]

Without conversion table the code almost looks the same:

Suggested change

	kind := kindConvert[n.Tok.Kind()]
	switch kind {
	case Integer, Float, Boolean:
	return Predefined[kind.String()]
	case Verdict:
	return Predefined["verdicttype"]
	switch n.Tok.Kind() {
	case syntax.INT:
	return Predefined["integer"]
	case syntax.FLOAT:
	return Predefined["float"]
	case syntax.TRUE, syntax.FALSE:
	return Predefined["boolean"]

I would go with that.

[5nord]

I find the flow hard to follow. What about:

Suggested change

	if strings.IndexFunc(n.Tok.String(), func(r rune) bool {
	return r <= 32 || r >= 126
	}) > -1 {
	return Predefined["universal charstring"]
	} else {
	return Predefined["charstring"]
	}
	for _, r := range n.Tok.String() {
	if r < 32 || 126 < r {
	return Predefined["universal charstring"]
	}
	}
	return Predefined["charstring"]

[5nord]

Checking the whole string for H or B is too much, also I don't think bitstring is the best default. I would use nil as default until we know better. How about something like:

Suggested change

	if strings.Contains(n.Tok.String(), "H") {
	return Predefined["hexstring"]
	} else if strings.Contains(n.Tok.String(), "O") {
	return Predefined["octetstring"]
	}
	return Predefined["bitstring"]
	s := n.Tok.String()
	if len(s) == 0 {
	return nil
	}
	switch s[len(s)-1] {
	case 'H', 'h':
	return Predefined["hexstring"]
	case 'O','o':
	return Predefined["octetstring"]
	case 'B','b':
	return Predefined["bitstring"]
	}
	return nil