[lexer] Correctly set line/offset for emitted Lexes

This commit is contained in:
Eryn Wells 2018-09-05 22:18:27 -07:00
parent 8b96eb190c
commit 9365e51893
3 changed files with 23 additions and 15 deletions

View file

@ -49,16 +49,14 @@ impl<T> Lexer<T> where T: Iterator<Item=char> {
fn prepare_offsets(&mut self) { } fn prepare_offsets(&mut self) { }
fn update_offsets(&mut self, c: char) { fn update_offsets(&mut self, c: char) {
self.offset += 1;
match c { match c {
'\n' => { '\n' => {
self.line += 1; self.line += 1;
self.offset = 0; self.offset = 0;
}, },
_ => { _ => self.offset += 1
self.offset += 1;
},
} }
println!("incremented offsets {}:{}", self.line, self.offset);
} }
} }
@ -68,6 +66,10 @@ impl<T> Iterator for Lexer<T> where T: Iterator<Item=char> {
fn next(&mut self) -> Option<Self::Item> { fn next(&mut self) -> Option<Self::Item> {
self.prepare_offsets(); self.prepare_offsets();
let mut token_line = self.line;
let mut token_offset = self.offset;
println!("beginning token at {}:{}", token_line, token_offset);
let mut buffer = String::new(); let mut buffer = String::new();
let mut state: Box<states::State> = Box::new(Begin::new()); let mut state: Box<states::State> = Box::new(Begin::new());
let mut out: Option<Self::Item> = None; let mut out: Option<Self::Item> = None;
@ -78,7 +80,7 @@ impl<T> Iterator for Lexer<T> where T: Iterator<Item=char> {
None => match state.none() { None => match state.none() {
Ok(None) => break, Ok(None) => break,
Ok(Some(token)) => { Ok(Some(token)) => {
out = Some(Ok(Lex::new(token, &buffer, self.line, self.offset))); out = Some(Ok(Lex::new(token, &buffer, token_line, token_offset)));
break; break;
}, },
Err(err) => self.handle_error(err) Err(err) => self.handle_error(err)
@ -101,13 +103,15 @@ impl<T> Iterator for Lexer<T> where T: Iterator<Item=char> {
if resume == Resume::AtNext { if resume == Resume::AtNext {
self.next(); self.next();
} }
token_line = self.line;
token_offset = self.offset;
}, },
StateResult::Emit(token, resume) => { StateResult::Emit(token, resume) => {
if resume == Resume::AtNext { if resume == Resume::AtNext {
buffer.push(c); buffer.push(c);
self.next(); self.next();
} }
out = Some(Ok(Lex::new(token, &buffer, self.line, self.offset))); out = Some(Ok(Lex::new(token, &buffer, token_line, token_offset)));
break; break;
}, },
StateResult::Fail(err) => self.handle_error(err), StateResult::Fail(err) => self.handle_error(err),
@ -115,6 +119,7 @@ impl<T> Iterator for Lexer<T> where T: Iterator<Item=char> {
}, },
} }
} }
println!("emitting {:?}", out);
out out
} }
} }

View file

@ -2,7 +2,7 @@
* Eryn Wells <eryn@erynwells.me> * Eryn Wells <eryn@erynwells.me>
*/ */
#[derive(Debug, Eq)] #[derive(Debug, Eq, PartialEq)]
pub struct Lex { pub struct Lex {
token: Token, token: Token,
value: String, value: String,
@ -33,8 +33,3 @@ impl Lex {
pub fn value(&self) -> &str { self.value.as_str() } pub fn value(&self) -> &str { self.value.as_str() }
} }
impl PartialEq for Lex {
fn eq(&self, rhs: &Lex) -> bool {
self.token == rhs.token && self.value == rhs.value
}
}

View file

@ -67,16 +67,24 @@ fn bool_long_false() {
#[test] #[test]
fn bool_with_spaces() { fn bool_with_spaces() {
// See issue #12 // See issue #12
let expected_lex = Lex::new(Token::Bool(false), "#f", 0, 0); let expected_lex = Lex::new(Token::Bool(false), "#f", 0, 2);
let mut lex = Lexer::new(" #f ".chars()); let mut lex = Lexer::new(" #f ".chars());
assert_eq!(lex.next(), Some(Ok(expected_lex))); assert_eq!(lex.next(), Some(Ok(expected_lex)));
assert_eq!(lex.next(), None); assert_eq!(lex.next(), None);
} }
#[test] #[test]
fn integer() { fn simple_integers() {
let mut lex = Lexer::new("23 42".chars()); let mut lex = Lexer::new("23 42".chars());
assert_eq!(lex.next(), Some(Ok(Lex::new(Token::Num(23), "23", 0, 0)))); assert_eq!(lex.next(), Some(Ok(Lex::new(Token::Num(23), "23", 0, 0))));
assert_eq!(lex.next(), Some(Ok(Lex::new(Token::Num(42), "42", 0, 0)))); assert_eq!(lex.next(), Some(Ok(Lex::new(Token::Num(42), "42", 0, 3))));
assert_eq!(lex.next(), None);
}
#[test]
fn integers_in_alternative_bases() {
let mut lex = Lexer::new("#x2A #b11001".chars());
assert_eq!(lex.next(), Some(Ok(Lex::new(Token::Num(0x2A), "#x2A", 0, 0))));
assert_eq!(lex.next(), Some(Ok(Lex::new(Token::Num(0b11001), "#b11001", 0, 5))));
assert_eq!(lex.next(), None); assert_eq!(lex.next(), None);
} }