lang: basic parser

crates/lang/Cargo.toml

@@ -7,6 +7,10 @@ edition = "2021"
 
 [dependencies]
 logos = "0.14"
+chumsky = "1.0.0-alpha.6"
+petgraph = { workspace = true}
+indexmap = "2.2.6"
+clap = { version = "4", features = ["derive"] }
 
 [lints]
 workspace = true

crates/lang/src/lib.rs

@@ -1 +1,5 @@
+pub mod parser;
 pub mod tokens;
+pub mod err_reporting {
+    pub struct GlobalReporter {}
+}

crates/lang/src/main.rs

@@ -0,0 +1,17 @@
+use std::{fs, path::PathBuf};
+
+use clap::Parser;
+use lang::parser::parse;
+
+#[derive(Parser)]
+struct Args {
+    file: PathBuf,
+}
+
+fn main() {
+    let args = Args::parse();
+    let f = fs::read_to_string(args.file).expect("failed to read file");
+
+    println!("file: {f}\n");
+    println!("parsed: {:?}", parse(&f))
+}

crates/lang/src/parser.rs

@@ -0,0 +1,134 @@
+use std::ops::Range;
+
+use chumsky::{
+    error::Rich,
+    extra,
+    input::{Stream, ValueInput},
+    prelude::*,
+    primitive::just,
+    recursive::recursive,
+    select,
+    span::SimpleSpan,
+    IterParser, Parser,
+};
+use indexmap::IndexMap;
+use logos::{Logos, Source};
+
+use crate::tokens::Token;
+
+pub mod ast;
+use self::ast::{Expr, File};
+
+type Span = SimpleSpan;
+type Spanned<T> = (T, Span);
+
+pub fn parse<'src>(src: &'src str) -> ParseResult<File<'_>, Rich<'_, Token<'_>>> {
+    let toks: Vec<_> = Token::lexer(src)
+        .spanned()
+        .into_iter()
+        .map(|(t, s)| (t.expect("TODO: add lexer error(s)"), Span::from(s)))
+        .collect();
+    let tok_stream = Stream::from_iter(toks).spanned((src.len()..src.len()).into());
+    expr_parser().parse(tok_stream)
+}
+
+fn expr_parser<'tokens, 'src: 'tokens, I: ValueInput<'tokens, Token = Token<'src>, Span = Span>>(
+) -> impl Parser<'tokens, I, File<'src>, extra::Err<Rich<'tokens, Token<'src>, Span>>> {
+    let word = select! { Token::Word(word) => word };
+
+    let expr = recursive(|expr| {
+        let var = select! {
+            Token::VarIdent(name) => (Expr::Var as fn(_) -> _, name),
+            Token::InputIdent(name) => (Expr::InputVar as fn(_) -> _, name)
+        }
+        .map_with(|(item_type, name), extra| item_type((name, extra.span())));
+
+        let attrset = word
+            .map_with(|n, e| (n, e.span()))
+            .then_ignore(just(Token::Colon))
+            .then(expr)
+            .separated_by(just(Token::Comma))
+            .collect::<Vec<_>>()
+            .map(IndexMap::from_iter)
+            .delimited_by(just(Token::BracketOpen), just(Token::BracketClose))
+            .map_with(|v, e| (v, e.span()));
+
+        let node = word
+            .map_with(|v, e| (v, e.span()))
+            .then(attrset.clone().or_not())
+            .map(|(name, params)| Expr::Node(name, params))
+            .or(var)
+            .or(attrset.map(Expr::AttrSet));
+
+        let pipeline = node
+            .clone()
+            .then(choice((
+                just(Token::Pipe).to(Expr::SimplePipe as fn(_, _) -> _),
+                just(Token::MappingPipe).to(Expr::MappingPipe as fn(_, _) -> _),
+                just(Token::NullPipe).to(Expr::NullPipe as fn(_, _) -> _),
+            )))
+            .repeated()
+            .foldr(node, |(curr, pipe), next| {
+                pipe(Box::new(curr), Box::new(next))
+            });
+
+        pipeline
+    });
+
+    let decl = just(Token::Def).ignore_then(
+        word.map_with(|n, e| (n, e.span()))
+            .then_ignore(just(Token::Equals))
+            .then(expr.clone().map_with(|expr, extra| (expr, extra.span())))
+            .then_ignore(just(Token::SemiColon)),
+    );
+
+    expr.map_with(|expr, extra| File {
+        decls: IndexMap::from_iter([(("main", (0..0).into()), (expr, extra.span()))]),
+    })
+    .or(decl.repeated().collect::<Vec<_>>().map(|decls| File {
+        decls: IndexMap::from_iter(decls),
+    }))
+}
+
+#[cfg(test)]
+mod tests {
+    use crate::parser::ast::{Expr, File};
+    use crate::parser::parse;
+    use crate::tokens::Token;
+    use chumsky::input::Stream;
+    use chumsky::prelude::*;
+    use indexmap::IndexMap;
+    use logos::Logos;
+
+    #[test]
+    fn test_parse_node_with_params() {
+        const INPUT: &str = "meow [ hello: $foo, world: @bar]";
+        assert_eq!(
+            parse(INPUT).unwrap(),
+            File {
+                decls: IndexMap::from_iter([(
+                    ("main", (0..0).into()),
+                    (
+                        Expr::Node(
+                            ("meow", (0..4).into()),
+                            Some((
+                                IndexMap::from_iter([
+                                    (
+                                        ("hello", (7..12).into()),
+                                        Expr::Var(("foo", (14..18).into()))
+                                    ),
+                                    (
+                                        ("world", (20..25).into()),
+                                        Expr::InputVar(("bar", (27..31).into()))
+                                    )
+                                ]),
+                                (5..32).into()
+                            ))
+                        ),
+                        (0..0).into()
+                    )
+                )])
+            }
+        );
+    }
+}

crates/lang/src/parser/ast.rs

@@ -0,0 +1,36 @@
+use std::collections::{BTreeMap, HashMap};
+
+use indexmap::IndexMap;
+
+use super::Spanned;
+
+#[derive(Debug, PartialEq, Eq)]
+pub struct File<'src> {
+    pub decls: IndexMap<Spanned<&'src str>, Spanned<Expr<'src>>>,
+}
+
+#[derive(Debug, PartialEq, Eq)]
+pub enum Expr<'src> {
+    Node(
+        Spanned<&'src str>,
+        Option<Spanned<IndexMap<Spanned<&'src str>, Expr<'src>>>>,
+    ),
+    SimplePipe(Box<Expr<'src>>, Box<Expr<'src>>),
+    NamingPipe(
+        Box<Expr<'src>>,
+        (Vec<Spanned<&'src str>>, Vec<Spanned<&'src str>>),
+        Box<Expr<'src>>,
+    ),
+    MappingPipe(Box<Expr<'src>>, Box<Expr<'src>>),
+    NullPipe(Box<Expr<'src>>, Box<Expr<'src>>),
+    MultiPipe(IndexMap<Spanned<&'src str>, Expr<'src>>),
+    LetIn(
+        IndexMap<Spanned<&'src str>, Box<Expr<'src>>>,
+        Box<Expr<'src>>,
+    ),
+    // $
+    Var(Spanned<&'src str>),
+    // @
+    InputVar(Spanned<&'src str>),
+    AttrSet(Spanned<IndexMap<Spanned<&'src str>, Expr<'src>>>),
+}

crates/lang/src/tokens.rs

@@ -1,8 +1,15 @@
 use logos::Logos;
 
-#[derive(Logos, Debug, PartialEq, Eq)]
+#[derive(Logos, Debug, PartialEq, Eq, Clone)]
 #[logos(skip r"[ \t\n\f]+")]
 pub enum Token<'a> {
+    // hack!
+    // this isn't actually supposed to be in the language.
+    // i just can't figure out how to automatically choose between a top level declaration
+    // or a top level expression
+    // so a declaration needs the keyword def until i can figure this out
+    #[token("def")]
+    Def,
     #[regex("[a-zA-Z0-9_\\-]+", |lex| lex.slice())]
     Word(&'a str),
     #[regex("\\$[a-zA-Z0-9_\\-]+", |lex| &lex.slice()[1..])]
@@ -27,6 +34,8 @@ pub enum Token<'a> {
     Equals,
     #[token(":")]
     Colon,
+    #[token(";")]
+    SemiColon,
     #[token("[")]
     BracketOpen,
     #[token("]")]