83 lines
1.7 KiB
Go
83 lines
1.7 KiB
Go
package shsh;
|
|
|
|
import (
|
|
"strings"
|
|
"unicode"
|
|
)
|
|
|
|
type token_t int
|
|
const (
|
|
LIST token_t = iota
|
|
STRING token_t = iota
|
|
NUMBER token_t = iota
|
|
SYMBOL token_t = iota
|
|
)
|
|
|
|
type Token struct {
|
|
next *Token
|
|
tag token_t
|
|
position int
|
|
_inner interface{}
|
|
}
|
|
|
|
func Lex(input string) *Token {
|
|
ret := new(Token)
|
|
iter := &ret
|
|
delim := ' '
|
|
var tok strings.Builder
|
|
iter_alloced := false
|
|
is_list := false
|
|
is_str := false
|
|
is_num := true
|
|
|
|
for pos, char := range input {
|
|
switch char {
|
|
// TODO: User configurable delimiters
|
|
case '\'', '"', '`':
|
|
is_str = true
|
|
delim = char
|
|
case '(':
|
|
is_list = true
|
|
delim = ')'
|
|
|
|
case delim:
|
|
*iter = new(Token)
|
|
(*iter).position = pos
|
|
|
|
if is_list {
|
|
(*iter)._inner = Lex(tok.String())
|
|
is_list = false
|
|
(*iter)._inner = LIST
|
|
|
|
} else {
|
|
(*iter)._inner = tok.String()
|
|
if is_str {
|
|
(*iter).tag = STRING
|
|
is_str = false
|
|
|
|
} else if is_num {
|
|
(*iter).tag = NUMBER
|
|
|
|
} else {
|
|
(*iter).tag = SYMBOL
|
|
}
|
|
}
|
|
|
|
iter_alloced = true
|
|
delim = ' '
|
|
|
|
default:
|
|
is_num = is_num && unicode.IsDigit(char)
|
|
tok.WriteRune(char)
|
|
}
|
|
|
|
if iter_alloced {
|
|
iter = &((*iter).next)
|
|
iter_alloced = false
|
|
tok.Reset()
|
|
}
|
|
}
|
|
|
|
// TODO: Throw parsing error here if there is leftover in tok
|
|
return ret
|
|
}
|