package llvm

import "strings"

const (
	// decimal specifies the decimal digit characters.
	decimal = "0123456789"
	// upper specifies the uppercase letters.
	upper = "ABCDEFGHIJKLMNOPQRSTUVWXYZ"
	// lower specifies the lowercase letters.
	lower = "abcdefghijklmnopqrstuvwxyz"
	// alpha specifies the alphabetic characters.
	alpha = upper + lower
	// head is the set of valid characters for the first character of an
	// identifier.
	head = alpha + "$-._"
	// tail is the set of valid characters for the remaining characters of an
	// identifier (i.e. all characters in the identifier except the first). All
	// characters of a label may be from the tail set, even the first character.
	tail = head + decimal
	// quotedIdent is the set of valid characters in quoted identifiers, which
	// excludes ASCII control characters, double quote, backslash and extended
	// ASCII characters.
	quotedIdent = " !#$%&'()*+,-./0123456789:;<=>?@ABCDEFGHIJKLMNOPQRSTUVWXYZ[]^_`abcdefghijklmnopqrstuvwxyz{|}~"
)

func EscapeIdent (s string) string {
	replace := false
	extra := 0
	for i := 0; i < len(s); i++ {
		if strings.IndexByte(tail, s[i]) == -1 {
			// Check if a replacement is required.
			//
			// Note, there are characters which are not valid in an identifier
			// (e.g. '#') but are valid in a quoted identifier, and therefore
			// require a replacement (i.e. quoted identifier), but no extra
			// characters for the escape sequence.
			replace = true
		}
		if strings.IndexByte(quotedIdent, s[i]) == -1 {
			// Two extra bytes are required for each byte not valid in a quoted
			// identifier; e.g.
			//
			//    "\t" -> `\09`
			//    "世" -> `\E4\B8\96`
			extra += 2
		}
	}
	if !replace {
		return s
	}
	// Replace invalid characters.
	const hextable = "0123456789ABCDEF"
	buf := make([]byte, len(s)+extra)
	j := 0
	for i := 0; i < len(s); i++ {
		b := s[i]
		if strings.IndexByte(quotedIdent, b) != -1 {
			buf[j] = b
			j++
			continue
		}
		buf[j] = '\\'
		buf[j+1] = hextable[b>>4]
		buf[j+2] = hextable[b&0x0F]
		j += 3
	}
	// Add surrounding quotes.
	return `"` + string(buf) + `"`
}

func Escape (s []byte, valid func(b byte) bool) string {
	// Check if a replacement is required.
	extra := 0
	for i := 0; i < len(s); i++ {
		if !valid(s[i]) {
			// Two extra bytes are required for each invalid byte; e.g.
			//    "#" -> `\23`
			//    "世" -> `\E4\B8\96`
			extra += 2
		}
	}
	if extra == 0 {
		return string(s)
	}
	// Replace invalid characters.
	const hextable = "0123456789ABCDEF"
	buf := make([]byte, len(s)+extra)
	j := 0
	for i := 0; i < len(s); i++ {
		b := s[i]
		if valid(b) {
			buf[j] = b
			j++
			continue
		}
		buf[j] = '\\'
		buf[j+1] = hextable[b>>4]
		buf[j+2] = hextable[b&0x0F]
		j += 3
	}
	return string(buf)
}

func EscapeString (s []byte) string {
	valid := func(b byte) bool {
		return ' ' <= b && b <= '~' && b != '"' && b != '\\'
	}
	return string(Escape(s, valid))
}

func EncodeTypeName (name string) string {
	return "%" + EscapeIdent(name)
}

func EncodeRegisterName (name string) string {
	return "%" + EscapeIdent(name)
}

func EncodeGlobalName (name string) string {
	return "@" + EscapeIdent(name)
}

func EncodeFunctionName (name string) string {
	return EncodeGlobalName(name)
}

func EncodeLabelName (name string) string {
	return EscapeIdent(name) + ":"
}

func EscapeQuoteString (s []byte) string {
	return `"` + string(EscapeString(s)) + `"`
}