basic parser tests

This commit is contained in:
me 2025-12-16 10:27:34 +02:00
parent 90673854a1
commit 5e40004fd4
5 changed files with 129 additions and 32 deletions

View file

@ -1,8 +1,21 @@
use super::types::*; use super::types::*;
use crate::ast; use crate::ast;
fn parse_definition(tokens: &mut Tokens) -> Result<ast::Definition, Error> { pub fn parse<T>(tokens: &mut Tokens, parser: Parser<T>) -> Result<T, Error> {
if let Some(start) = tokens.next_if(Token::Let) { if let Some(result) = parser(tokens)?
&& tokens.peek().is_none()
{
Ok(result)
} else {
Err(Error::UnexpectedTokenForParser(
"parser".into(),
tokens.next(),
))
}
}
fn parse_definition(tokens: &mut Tokens) -> ParseResult<ast::Definition> {
if let Some(_) = tokens.next_if(Token::Let) {
if let Some(LocatedToken { if let Some(LocatedToken {
token: Token::Identifier(identifier), token: Token::Identifier(identifier),
.. ..
@ -10,30 +23,38 @@ fn parse_definition(tokens: &mut Tokens) -> Result<ast::Definition, Error> {
{ {
if let Some(_eq) = tokens.next_if(Token::Equals) { if let Some(_eq) = tokens.next_if(Token::Equals) {
if let Some(expr) = parse_expr(tokens)? { if let Some(expr) = parse_expr(tokens)? {
Ok(ast::Definition { if let Some(_) = tokens.next_if(Token::Semicolon) {
name: ast::Name(identifier), Ok(Some(ast::Definition {
expr, name: ast::Name(identifier),
}) expr,
}))
} else {
Err(Error::UnexpectedToken {
expected: Token::Semicolon,
got: tokens.next(),
})
}
} else { } else {
Err(Error::UnexpectedToken { Err(Error::UnexpectedTokenForParser(
expected: Token::Equals, "expr".into(),
got: start, tokens.next(),
}) ))
} }
} else { } else {
Err(Error::UnexpectedToken { Err(Error::UnexpectedToken {
expected: Token::Equals, expected: Token::Equals,
got: start, got: tokens.next(),
}) })
} }
} else { } else {
Err(Error::UnexpectedToken { Err(Error::UnexpectedToken {
expected: Token::Equals, expected: Token::Identifier("".into()),
got: start, got: tokens.next(),
}) })
} }
} else { } else {
Err(Error::UnexpectedEndOfInput) println!("not let! {tokens:#?}");
Ok(None)
} }
} }
@ -43,11 +64,11 @@ fn parse_expr(tokens: &mut Tokens) -> ParseResult<ast::Expr> {
fn parse_simple_expr(tokens: &mut Tokens) -> ParseResult<ast::Expr> { fn parse_simple_expr(tokens: &mut Tokens) -> ParseResult<ast::Expr> {
tokens.one_of(vec![ tokens.one_of(vec![
|t| t.between(Token::OpenParen, Token::CloseParen, parse_expr),
parse_identifier,
parse_number, parse_number,
parse_identifier,
parse_string, parse_string,
parse_bool, parse_bool,
|t| t.between(Token::OpenParen, Token::CloseParen, parse_expr),
]) ])
} }
@ -115,3 +136,22 @@ fn parse_bool(tokens: &mut Tokens) -> ParseResult<ast::Expr> {
}, },
]) ])
} }
#[cfg(test)]
mod tests {
use super::super::scanner::*;
use super::*;
#[test]
fn number() {
let program = "108".to_string();
let result = parse(&mut scan(program), parse_expr);
insta::assert_debug_snapshot!(result);
}
#[test]
fn let_number() {
let program = "let x = 108;".to_string();
let result = parse(&mut scan(program), parse_definition);
insta::assert_debug_snapshot!(result);
}
}

View file

@ -3,7 +3,7 @@ use chumsky::text::Char;
use log; use log;
use lyn::Scanner; use lyn::Scanner;
fn scan(source: String) -> Vec<LocatedToken> { pub fn scan(source: String) -> Tokens {
let mut scanner = Scanner::new(&source); let mut scanner = Scanner::new(&source);
let mut tokens = Vec::new(); let mut tokens = Vec::new();
@ -153,7 +153,7 @@ fn scan(source: String) -> Vec<LocatedToken> {
} }
} }
tokens tokens.into()
} }
#[cfg(test)] #[cfg(test)]
@ -163,10 +163,12 @@ mod tests {
#[test] #[test]
fn let_number() { fn let_number() {
let program = "let x = 108;".to_string(); let program = "let x = 108;".to_string();
let result = scan(program) let mut result = scan(program)
.0
.into_iter() .into_iter()
.map(|t| t.token) .map(|t| t.token)
.collect::<Vec<_>>(); .collect::<Vec<_>>();
result.reverse();
insta::assert_debug_snapshot!(result); insta::assert_debug_snapshot!(result);
} }
@ -177,10 +179,12 @@ let main = fn (x) {
console.log(x); console.log(x);
};" };"
.to_string(); .to_string();
let result = scan(program) let mut result = scan(program)
.0
.into_iter() .into_iter()
.map(|t| t.token) .map(|t| t.token)
.collect::<Vec<_>>(); .collect::<Vec<_>>();
result.reverse();
insta::assert_debug_snapshot!(result); insta::assert_debug_snapshot!(result);
} }
@ -206,10 +210,12 @@ let migrate = fn(state) {
}; };
" "
.to_string(); .to_string();
let result = scan(program) let mut result = scan(program)
.0
.into_iter() .into_iter()
.map(|t| t.token) .map(|t| t.token)
.collect::<Vec<_>>(); .collect::<Vec<_>>();
result.reverse();
insta::assert_debug_snapshot!(result); insta::assert_debug_snapshot!(result);
} }
/* /*

View file

@ -0,0 +1,16 @@
---
source: src/parser/parser.rs
expression: result
---
Ok(
Definition {
name: Name(
"x",
),
expr: Value(
Int(
108,
),
),
},
)

View file

@ -0,0 +1,11 @@
---
source: src/parser/parser.rs
expression: result
---
Ok(
Value(
Int(
108,
),
),
)

View file

@ -28,7 +28,15 @@ pub enum Token {
Identifier(String), Identifier(String),
} }
pub struct Tokens(Vec<LocatedToken>); #[derive(Debug, Clone, PartialEq, Eq)]
pub struct Tokens(pub Vec<LocatedToken>);
impl From<Vec<LocatedToken>> for Tokens {
fn from(mut tokens: Vec<LocatedToken>) -> Tokens {
tokens.reverse();
Tokens(tokens)
}
}
impl Tokens { impl Tokens {
pub fn peek(&mut self) -> Option<&LocatedToken> { pub fn peek(&mut self) -> Option<&LocatedToken> {
@ -38,21 +46,30 @@ impl Tokens {
self.0.pop() self.0.pop()
} }
pub fn next_if(&mut self, token: Token) -> Option<LocatedToken> { pub fn next_if(&mut self, token: Token) -> Option<LocatedToken> {
self.0.pop_if(|t| match (&t.token, &token) { self.0.pop_if(|t| match (t.token.clone(), token) {
(a, b) if a == b => true, (a, b) if a == b => true,
(Token::Identifier(_), Token::Identifier(_)) => true, (Token::Identifier(_), Token::Identifier(_)) => true,
(Token::Number(_), Token::Number(_)) => true, (Token::Number(_), Token::Number(_)) => true,
(Token::String(_), Token::String(_)) => true, (Token::String(_), Token::String(_)) => true,
_ => false, x => {
println!("next_if fail: {x:#?}");
false
}
}) })
} }
pub fn between<T>(&mut self, start: Token, end: Token, parser: Parser<T>) -> ParseResult<T> { pub fn between<T>(&mut self, start: Token, end: Token, parser: Parser<T>) -> ParseResult<T> {
let _start = self.next_if(start).ok_or(Error::UnexpectedEndOfInput)?; if let Some(_start) = self.next_if(start) {
if let Some(result) = parser(self)? { if let Some(result) = parser(self)? {
let _end = self.next_if(end).ok_or(Error::UnexpectedEndOfInput)?; let _end = self.next_if(end).ok_or(Error::UnexpectedEndOfInput)?;
Ok(Some(result)) Ok(Some(result))
} else {
Err(Error::UnexpectedTokenForParser(
"between".into(),
self.next(),
))
}
} else { } else {
Err(Error::UnexpectedTokenForParser) Ok(None)
} }
} }
pub fn one_of<T>(&mut self, parsers: Vec<Parser<T>>) -> ParseResult<T> { pub fn one_of<T>(&mut self, parsers: Vec<Parser<T>>) -> ParseResult<T> {
@ -74,7 +91,10 @@ impl Tokens {
if let Some(t2) = parser2(self)? { if let Some(t2) = parser2(self)? {
Ok(Some((t1, t2))) Ok(Some((t1, t2)))
} else { } else {
Err(Error::UnexpectedTokenForParser) Err(Error::UnexpectedTokenForParser(
"sep_by".into(),
self.next(),
))
} }
} else { } else {
Ok(None) Ok(None)
@ -92,8 +112,12 @@ impl Tokens {
pub type Parser<T> = fn(&mut Tokens) -> ParseResult<T>; pub type Parser<T> = fn(&mut Tokens) -> ParseResult<T>;
pub type ParseResult<T> = Result<Option<T>, Error>; pub type ParseResult<T> = Result<Option<T>, Error>;
#[derive(Debug, Clone, PartialEq, Eq)]
pub enum Error { pub enum Error {
UnexpectedToken { expected: Token, got: LocatedToken }, UnexpectedToken {
expected: Token,
got: Option<LocatedToken>,
},
UnexpectedEndOfInput, UnexpectedEndOfInput,
UnexpectedTokenForParser, UnexpectedTokenForParser(String, Option<LocatedToken>),
} }