schala/src/tokenizer.rs

186 lines
4.8 KiB
Rust
Raw Normal View History

extern crate itertools;
2017-01-01 20:15:17 -08:00
use std::iter::Peekable;
use std::str::Chars;
use self::itertools::Itertools;
2017-01-01 20:15:17 -08:00
2015-12-25 02:03:11 -08:00
#[derive(Debug, Clone, PartialEq)]
2015-07-22 03:02:55 -07:00
pub enum Token {
Newline,
Semicolon,
2015-07-22 03:02:55 -07:00
LParen,
RParen,
Comma,
2015-07-26 01:51:15 -07:00
Period,
Colon,
2015-07-22 03:02:55 -07:00
NumLiteral(f64),
StrLiteral(String),
2015-07-22 04:01:56 -07:00
Identifier(String),
2016-01-15 03:27:24 -08:00
Operator(Op),
2016-12-29 02:04:03 -08:00
Keyword(Kw),
2015-07-22 04:01:56 -07:00
}
2016-01-15 03:27:24 -08:00
#[derive(Debug, Clone, PartialEq)]
pub struct Op(pub String);
2016-01-15 03:27:24 -08:00
2015-07-22 04:01:56 -07:00
#[derive(Debug, Clone, PartialEq)]
pub enum Kw {
If,
Then,
Else,
While,
End,
Let,
Fn,
2015-08-08 00:27:40 -07:00
Null,
2015-07-22 03:02:55 -07:00
}
2015-07-22 03:12:01 -07:00
2016-12-28 22:52:23 -08:00
pub type TokenizeResult = Result<Vec<Token>, TokenizeError>;
#[derive(Debug)]
pub struct TokenizeError {
pub msg: String,
}
impl TokenizeError {
fn new(msg: &str) -> TokenizeError {
2016-12-29 02:04:03 -08:00
TokenizeError { msg: msg.to_string() }
2016-12-28 22:52:23 -08:00
}
}
fn is_digit(c: &char) -> bool {
c.is_digit(10)
}
2016-12-28 22:52:23 -08:00
pub fn tokenize(input: &str) -> TokenizeResult {
use self::Token::*;
2015-07-22 03:12:01 -07:00
let mut tokens = Vec::new();
2017-01-01 20:15:17 -08:00
let mut iter: Peekable<Chars> = input.chars().peekable();
while let Some(c) = iter.next() {
if c == '#' {
while let Some(c) = iter.next() {
2016-12-29 02:04:03 -08:00
if c == '\n' {
break;
}
}
}
let cur_tok = match c {
c if char::is_whitespace(c) && c != '\n' => continue,
'\n' => Newline,
';' => Semicolon,
'(' => LParen,
')' => RParen,
':' => Colon,
',' => Comma,
2017-01-01 20:15:17 -08:00
'"' => try!(tokenize_str(&mut iter)),
c if !char::is_alphanumeric(c) => try!(tokenize_operator(c, &mut iter)),
c if is_digit(&c) || c == '.' => try!(tokenize_number_or_period(c, &mut iter)),
c => try!(tokenize_identifier(c, &mut iter)),
};
tokens.push(cur_tok);
}
2016-12-28 22:52:23 -08:00
Ok(tokens)
2015-07-22 03:12:01 -07:00
}
2015-07-22 04:01:56 -07:00
2017-01-01 20:15:17 -08:00
fn tokenize_str(iter: &mut Peekable<Chars>) -> Result<Token, TokenizeError> {
let mut buffer = String::new();
loop {
// TODO handle string escapes, interpolation
match iter.next() {
Some(x) if x == '"' => break,
Some(x) => buffer.push(x),
None => return Err(TokenizeError::new("Unclosed quote")),
}
}
Ok(Token::StrLiteral(buffer))
}
fn tokenize_operator(c: char, iter: &mut Peekable<Chars>) -> Result<Token, TokenizeError> {
let mut buffer = String::new();
buffer.push(c);
buffer.extend(iter.peeking_take_while(|x| !char::is_alphanumeric(*x) && !char::is_whitespace(*x)));
2017-01-01 20:15:17 -08:00
Ok(Token::Operator(Op(buffer)))
}
fn tokenize_number_or_period(c: char, iter: &mut Peekable<Chars>) -> Result<Token, TokenizeError> {
if c == '.' && !iter.peek().map_or(false, is_digit) {
return Ok(Token::Period);
}
let mut buffer = String::new();
buffer.push(c);
buffer.extend(iter.peeking_take_while(|x| is_digit(x) || *x == '.'));
2017-01-01 20:15:17 -08:00
match buffer.parse::<f64>() {
Ok(f) => Ok(Token::NumLiteral(f)),
Err(_) => Err(TokenizeError::new("Failed to parse digit")),
}
}
fn tokenize_identifier(c: char, iter: &mut Peekable<Chars>) -> Result<Token, TokenizeError> {
fn ends_identifier(c: &char) -> bool {
let c = *c;
char::is_whitespace(c) || is_digit(&c) || c == ';' || c == '(' || c == ')' ||
c == ',' || c == '.' || c == ',' || c == ':'
}
use self::Token::*;
let mut buffer = String::new();
buffer.push(c);
buffer.extend(iter.peeking_take_while(ends_identifier));
2017-01-01 20:15:17 -08:00
Ok(match &buffer[..] {
"if" => Keyword(Kw::If),
"then" => Keyword(Kw::Then),
"else" => Keyword(Kw::Else),
"while" => Keyword(Kw::While),
"end" => Keyword(Kw::End),
"let" => Keyword(Kw::Let),
"fn" => Keyword(Kw::Fn),
"null" => Keyword(Kw::Null),
b => Identifier(b.to_string()),
})
}
2015-12-20 17:03:03 -08:00
#[cfg(test)]
mod tests {
2016-01-09 00:07:48 -08:00
macro_rules! tokentest {
($input:expr, $output:expr) => {
{
let tokens = tokenize($input).unwrap();
assert_eq!(format!("{:?}", tokens), $output);
}
}
}
2015-12-20 17:03:03 -08:00
use super::*;
#[test]
fn tokeniziation_tests() {
2016-01-09 00:07:48 -08:00
tokentest!("let a = 3\n",
2017-01-01 20:15:17 -08:00
"[Keyword(Let), Identifier(\"a\"), Operator(Op(\"=\")), \
2016-12-29 02:04:03 -08:00
NumLiteral(3), Newline]");
2016-01-09 00:07:48 -08:00
tokentest!("2+1",
2017-01-01 20:15:17 -08:00
"[NumLiteral(2), Operator(Op(\"+\")), NumLiteral(1)]");
2016-01-09 00:07:48 -08:00
tokentest!("2 + 1",
2017-01-01 20:15:17 -08:00
"[NumLiteral(2), Operator(Op(\"+\")), NumLiteral(1)]");
2016-01-09 00:07:48 -08:00
2016-01-09 00:09:48 -08:00
tokentest!("2.3*49.2",
2017-01-01 20:15:17 -08:00
"[NumLiteral(2.3), Operator(Op(\"*\")), NumLiteral(49.2)]");
2016-01-09 00:09:48 -08:00
2016-12-28 22:52:23 -08:00
assert!(tokenize("2.4.5").is_err());
}
#[test]
#[ignore]
fn more_tokenization() {
2016-12-29 02:04:03 -08:00
// it would be nice to support complicated operators in a nice, haskell-ish way
tokentest!("a *> b",
2016-12-29 02:04:03 -08:00
"[Identifier(\"a\"), Identifier(\"*>\"), Identifier(\"b\"), EOF]");
2016-01-09 00:09:48 -08:00
2015-12-20 17:03:03 -08:00
}
}