aidl commit almost mvp

programs/aidl/assets/vfs.idl

@@ -1,5 +1,4 @@
-// core provides lots of useful types like String and Byte
+Use core;
-use core;
 
 Constant VERSION Version{
     major: 1,
@@ -21,4 +20,4 @@ Interface File{
     function open accepts(Path) returns(File);
 
     function close accepts(File) returns(None);
-}
+}

programs/aidl/src/ast.rs

@@ -0,0 +1,43 @@
+//! **note** the order of fields is the order of parsing.
+
+/// An IDL module.
+///
+/// Parsing order:
+/// - use declarations,
+/// - items
+#[derive(Debug)]
+pub struct IDLModule {
+    // why: only allow use before other items
+    // parser will error if use is present in any other place
+    pub uses: Vec<UseDecl>,
+    pub items: Vec<Item>
+}
+
+#[derive(Debug)]
+pub enum Item {
+    Interface(ItemInterface),
+    Type(ItemType)
+}
+
+#[derive(Debug)]
+pub struct Function {
+    pub name: String,
+
+}
+
+#[derive(Debug)]
+pub struct ItemInterface {
+    pub name: String,
+    pub functions: Vec<Function>
+}
+
+#[derive(Debug)]
+pub struct ItemType {
+    pub name: String,
+    pub referree: String
+}
+
+#[derive(Debug)]
+pub struct UseDecl {
+    pub module: String
+}

programs/aidl/src/lexer.rs

@@ -0,0 +1,111 @@
+use std::ops::{Range, Add};
+
+use logos::Logos;
+
+#[derive(Logos, Debug, PartialEq)]
+#[logos(skip r"[ \t\n\f]+")]
+pub enum Token {
+    #[token("{")]
+    LeftBrace,
+
+    #[token("}")]
+    RightBrace,
+
+    #[token("(")]
+    LeftParen,
+
+    #[token(")")]
+    RightParen,
+
+    #[token(";")]
+    Semicolon,
+
+    #[token(":")]
+    Colon,
+
+    #[token("<")]
+    LeftArrow,
+
+    #[token(">")]
+    RightArrow,
+
+    #[token(",")]
+    Comma,
+
+    #[token("=")]
+    Equals,
+
+    //#[regex(r#"[A-z]+"#, |lex| lex.slice().parse().ok())]
+    //Literal(String),
+
+    #[regex(r#"[A-z]+"#, |lex| Ident::lexer(lex.slice()).next().and_then(Result::ok))]
+    Ident(Ident),
+
+    #[regex("use [a-zA-Z/]+;", |lex| lex.slice().parse().ok())]
+    Component(String),
+
+    #[regex("U[0-9]+", |lex| lex.slice().parse().ok())]
+    UnsignedType(String),
+
+    #[regex("I[0-9]+", |lex| lex.slice().parse().ok())]
+    SignedType(String),
+
+    #[regex(r"//.*", |lex| lex.slice().parse().ok())]
+    Comment(String),
+}
+
+#[derive(Logos, Debug, PartialEq, Eq)]
+pub enum Ident {
+    #[token("Interface")]
+    Interface,
+    #[token("Function")]
+    Function,
+    #[token("Constant")]
+    Constant,
+    #[token("Structure")]
+    Structure,
+    #[token("Type")]
+    Type,
+    #[token("Use")]
+    Use,
+    #[regex(r"[A-z]+", |lex| lex.slice().parse().ok())]
+    Other(String)
+}
+
+#[derive(Debug, Clone)]
+pub struct Span(pub Range<usize>);
+impl Span {
+    pub const ZERO: Self = Self(0..0);
+
+    pub fn lower(&self) -> usize {
+        self.0.start
+    }
+    pub fn upper(&self) -> usize {
+        self.0.end
+    }
+
+    pub fn concat(self, other: Span) -> Self {
+        use std::cmp::{min, max};
+
+        Self(min(self.lower(), other.lower())..max(self.upper(), other.upper()))
+    }
+}
+impl Add for Span {
+    type Output = Self;
+
+    fn add(self, rhs: Self) -> Self::Output {
+        self.concat(rhs)    
+    }
+}
+
+#[derive(Debug, Clone)]
+pub struct Spanned<T>(pub T, pub Span);
+
+impl<T> Spanned<T> {
+    pub fn new<const N: usize>(thing: T, spans: [Span; N]) -> Self {
+        Self(thing, spans.into_iter().fold(Span::ZERO, Span::concat))
+    }
+    pub fn map<R>(self, f: impl Fn(T) -> R) -> Spanned<R> {
+        Spanned(f(self.0), self.1)
+    }
+}

programs/aidl/src/main.rs

@@ -1,62 +1,26 @@
+#![allow(non_snake_case)]
+
 use logos::Logos;
+use parser::Parser;
 
-#[derive(Logos, Debug, PartialEq)]
+use crate::lexer::Token;
-#[logos(skip r"[ \t\n\f]+")] // Ignore this regex pattern between tokens
+mod ast;
-enum Token {
+mod lexer;
-    #[token("{")]
+mod parser;
-    LeftBrace,
 
-    #[token("}")]
+const VFS: &str = include_str!("../assets/vfs.idl");
-    RightBrace,
-
-    #[token("(")]
-    LeftParen,
-
-    #[token(")")]
-    RightParen,
-
-    #[token(";")]
-    Semicolon,
-
-    #[token(":")]
-    Colon,
-
-    #[token("<")]
-    LeftArrow,
-
-    #[token(">")]
-    RightArrow,
-
-    #[token(",")]
-    Comma,
-
-    #[token("=")]
-    Equals,
-
-    #[regex(r#"[A-z]+"#, |lex| lex.slice().parse().ok())]
-    Literal(String),
-
-    #[regex("use [a-zA-Z/]+;", |lex| lex.slice().parse().ok())]
-    Component(String),
-
-    #[regex("U[0-9]+", |lex| lex.slice().parse().ok())]
-    UnsignedType(String),
-
-    #[regex("I[0-9]+", |lex| lex.slice().parse().ok())]
-    SignedType(String),
-
-    #[regex(r"//[ a-zA-Z!-+]+", |lex| lex.slice().parse().ok())]
-    Comment(String),
-}
 
 fn main() {
-    let mut lex = Token::lexer(include_str!("../../../programs/aidl/assets/vfs.idl"));
+    dbg!(Token::lexer(VFS).for_each(|a| println!("{:#?}", a)));
-
+    dbg!(Parser::new(VFS).parse());
-    for token in lex {
+}
-        // let ok_token = token.ok();
+
-        // if ok_token.is_some() {
+#[macro_export]
-        //     println!("{:?}", ok_token.unwrap());
+macro_rules! unwrap_match {
-        // }
+    ($x:expr, $m:pat => $a:expr) => {
-        println!("{:?}", token);
+        match $x {
-    }
+            $m => $a,
+            _ => unreachable!()
+        }
+    };
 }

programs/aidl/src/parser.rs

@@ -0,0 +1,150 @@
+use logos::{Logos, SpannedIter};
+
+use crate::{
+    ast::{IDLModule, Item, ItemInterface, ItemType, UseDecl},
+    lexer::{Span, Spanned, Token, Ident},
+};
+use std::{
+    iter::{Iterator, Peekable, Filter},
+    ops::Range,
+};
+
+type Wtf<'a> = Peekable<
+        Filter<SpannedIter<'a, Token>, Box<dyn Fn(&(Result<Token, ()>, Range<usize>)) -> bool>>,
+    >;
+
+struct TokenIterator<'a> {
+    spanned: Wtf<'a>,
+}
+
+fn token_is_not_comment((ref a, ..): &(Result<Token, ()>, Range<usize>)) -> bool {
+    !matches!(a, Err(_) | Ok(Token::Comment(..)))
+}
+
+impl<'a> TokenIterator<'a> {
+    pub fn new(src: &'a str) -> Self {
+        let spanned = Token::lexer(src)
+            .spanned()
+            .filter(
+                Box::new(token_is_not_comment) as Box<dyn Fn(&(Result<Token, ()>, Range<usize>)) -> bool>,
+            )
+            .peekable();
+
+        Self { spanned }
+    }
+    pub fn next(&mut self) -> Option<Spanned<Token>> {
+        let nxt = self
+            .spanned
+            .next()
+            .and_then(|(token, span)| Some(Spanned(token.ok()?, Span(span))));
+        println!("[NEXT] {:#?}", nxt);
+        nxt
+    }
+
+    pub fn peek(&mut self) -> Option<Spanned<&Token>> {
+        let peek = self
+            .spanned
+            .peek()
+            .and_then(|(token, span)| Some(Spanned(token.as_ref().ok()?, Span(span.clone()))));
+        println!("[PEEK] {:#?}", peek);
+        peek
+    }
+}
+
+pub struct Parser<'a> {
+    tokens: TokenIterator<'a>,
+}
+
+impl<'a> Parser<'a> {
+    pub fn new(src: &'a str) -> Self {
+        Self {
+            tokens: TokenIterator::new(src),
+        }
+    }
+
+    fn get_real(&mut self, matcher: impl Fn(&Token) -> bool) -> Option<Spanned<Token>> {
+        if matcher(self.tokens.peek()?.0) {
+            self.tokens.next()
+        } else {
+            None
+        }
+    }
+
+    fn semi(&mut self) -> Option<Span> {
+        Some(self.get_real(|token| matches!(token, Token::Semicolon))?.1)
+    }
+
+    fn ask_ident(&mut self) -> Option<Spanned<String>> {
+        Some(crate::unwrap_match!(
+            self.get_real(|token| matches!(token, Token::Ident(Ident::Other(_))))?,
+            Spanned(Token::Ident(Ident::Other(ident)), span) =>
+            Spanned(ident, span)
+        ))
+    }
+
+    fn ask_interface(&mut self) -> Option<Spanned<ItemInterface>> {
+        let Spanned(_, kSp) =
+            self.get_real(|token| matches!(token, Token::Ident(Ident::Interface)))?;
+        let Spanned(ident, iSp) = self.ask_ident()?;
+
+        Some(Spanned::new(
+            ItemInterface {
+                name: ident,
+                functions: vec![],
+            },
+            [kSp, iSp, self.semi()?],
+        ))
+    }
+
+    fn ask_typealias(&mut self) -> Option<Spanned<ItemType>> {
+        let Spanned(_, kSp) =
+            self.get_real(|token| matches!(token, Token::Ident(Ident::Type)))?;
+        let Spanned(name, nSp) = self.ask_ident()?;
+
+        let Spanned(_, eqSp) = self.get_real(|token| matches!(token, Token::Equals))?;
+        let Spanned(referree, rSp) = self.ask_ident()?;
+
+        Some(Spanned::new(
+            ItemType { name, referree },
+            [kSp, nSp, eqSp, rSp, self.semi()?],
+        ))
+    }
+
+    fn ask_item(&mut self) -> Option<Spanned<Item>> {
+        Some(match self.tokens.peek()?.0 {
+            Token::Ident(Ident::Other(_)) => None?,
+            Token::Ident(keyword) => match keyword {
+                Ident::Interface => self.ask_interface()?.map(Item::Interface),
+                Ident::Type => self.ask_typealias()?.map(Item::Type),
+                _ => None?,
+            },
+            _ => None?,
+        })
+    }
+
+    fn ask_use(&mut self) -> Option<Spanned<UseDecl>> {
+        let Spanned(_, kSp) =
+            self.get_real(|token| matches!(token, Token::Ident(Ident::Use)))?;
+        let Spanned(name, nSp) = self.ask_ident()?;
+
+        Some(Spanned::new(
+            UseDecl { module: name },
+            [kSp, nSp, self.semi()?],
+        ))
+    }
+
+    pub fn parse(mut self) -> IDLModule {
+        IDLModule {
+            uses: fill_while(|| self.ask_use()),
+            items: fill_while(|| self.ask_item()),
+        }
+    }
+}
+
+fn fill_while<T>(mut f: impl FnMut() -> Option<Spanned<T>>) -> Vec<T> {
+    let mut real = vec![];
+    while let Some(Spanned(t, _)) = f() {
+        real.push(t);
+    }
+    real
+}

rust-toolchain

@@ -1 +0,0 @@
-nightly

rust-toolchain.toml

@@ -0,0 +1,3 @@
+[toolchain]
+channel = "nightly"
+components = ["cargo", "clippy", "rustfmt", "rust-analyzer"]