56 lines
2.0 KiB
Plaintext
56 lines
2.0 KiB
Plaintext
|
// Minimal tokenizer for Ny v0 (ints, + - * /, ( ), return)
|
||
|
|
|
||
|
|
static box Tokenizer {
|
||
|
|
tokenize(input) {
|
||
|
|
local tokens = new ArrayBox()
|
||
|
|
local i = 0
|
||
|
|
local n = input.length()
|
||
|
|
// helper: skip whitespace
|
||
|
|
fn skip_ws() {
|
||
|
|
loop(i < n) {
|
||
|
|
local ch = input.substring(i, i+1)
|
||
|
|
if ch == " " || ch == "\t" || ch == "\r" || ch == "\n" { i = i + 1 } else { return }
|
||
|
|
}
|
||
|
|
}
|
||
|
|
// main loop
|
||
|
|
loop(i < n) {
|
||
|
|
skip_ws()
|
||
|
|
if i >= n { break }
|
||
|
|
local ch = input.substring(i, i+1)
|
||
|
|
if ch == "+" || ch == "-" || ch == "*" || ch == "/" || ch == "(" || ch == ")" {
|
||
|
|
local tok = new MapBox(); tok.set("type", ch)
|
||
|
|
tokens.push(tok); i = i + 1; continue
|
||
|
|
}
|
||
|
|
// keyword: return
|
||
|
|
if i + 6 <= n {
|
||
|
|
local kw = input.substring(i, i+6)
|
||
|
|
if kw == "return" {
|
||
|
|
local t = new MapBox(); t.set("type", "RETURN")
|
||
|
|
tokens.push(t); i = i + 6; continue
|
||
|
|
}
|
||
|
|
}
|
||
|
|
// integer literal
|
||
|
|
if ch >= "0" && ch <= "9" {
|
||
|
|
local j = i
|
||
|
|
loop(j < n) {
|
||
|
|
local cj = input.substring(j, j+1)
|
||
|
|
if cj >= "0" && cj <= "9" { j = j + 1 } else { break }
|
||
|
|
}
|
||
|
|
local num_str = input.substring(i, j)
|
||
|
|
local tnum = new MapBox(); tnum.set("type", "INT"); tnum.set("value", num_str)
|
||
|
|
tokens.push(tnum); i = j; continue
|
||
|
|
}
|
||
|
|
// unknown
|
||
|
|
local err = new MapBox(); err.set("version", 0); err.set("kind", "Error")
|
||
|
|
local e = new MapBox(); e.set("message", "Unknown token");
|
||
|
|
local sp = new MapBox(); sp.set("start", i); sp.set("end", i+1)
|
||
|
|
e.set("span", sp); err.set("error", e)
|
||
|
|
return err
|
||
|
|
}
|
||
|
|
// EOF
|
||
|
|
local eof = new MapBox(); eof.set("type", "EOF"); tokens.push(eof)
|
||
|
|
return tokens
|
||
|
|
}
|
||
|
|
}
|
||
|
|
|