Did stuff

Cargo.lock

@@ -118,6 +118,7 @@ dependencies = [
  "clap",
  "hence",
  "itertools",
+ "num-parse",
 ]
 
 [[package]]

examples/test.lisp

@@ -1,7 +1,5 @@
-;; test module
-(module test
-  "Main module"
-  (defun main ()
-    "Program entrypoint"
-    (let ((x (+ 40 2))))
-    (debug x)))
+;; main function
+(defun main ()
+  "Program entrypoint"
+  (let ((x (+ 40 2))))
+  (debug x))

hence/src/lib/arg.rs

@@ -32,7 +32,7 @@ impl assembler::ToCode for Arg {
     }
 }
 
-impl assembler::ByteResolvable for Arg {
+impl assembler::ByteResolvable<assembler::Data> for Arg {
     fn resolve_number(&self, data: &mut assembler::Data) -> Result<u16, String> {
         match self {
             Arg::String(x) => {

hence/src/lib/assembler.rs

@@ -9,10 +9,10 @@ pub trait ToCode {
     fn to_code(&self) -> String;
 }
 
-pub trait ByteResolvable {
-    fn resolve_number(&self, data: &mut Data) -> Result<u16, String>;
+pub trait ByteResolvable<T> {
+    fn resolve_number(&self, data: &mut T) -> Result<u16, String>;
 
-    fn resolve_bytes(&self, data: &mut Data) -> Result<Vec<u8>, String>;
+    fn resolve_bytes(&self, data: &mut T) -> Result<Vec<u8>, String>;
 }
 
 #[derive(Debug)]

hence/src/lib/parser.rs

@@ -9,7 +9,7 @@ pub fn parse(tokens: Vec<lexer::Token>) -> Result<ast::AST, &'static str> {
     let mut iter = tokens.iter().peekable();
     let mut body: Vec<ast::Node> = Vec::new();
 
-    while let Some(token) = iter.peek() {
+    while let Some(&token) = iter.peek() {
         match token {
             lexer::Token::Comment(x) => {
                 body.push(ast::Node::Comment(x.trim().to_string()));
@@ -60,7 +60,7 @@ pub fn parse(tokens: Vec<lexer::Token>) -> Result<ast::AST, &'static str> {
             lexer::Token::Whitespace(_) | lexer::Token::Newline(_) => {
                 iter.next();
             }
-            _ => return Err("Unexspected token"),
+            _ => return Err("Unexpected token"),
         }
     }
 

hencelisp/Cargo.toml

@@ -16,3 +16,4 @@ path = "src/bin/main.rs"
 hence = { path = "../hence" }
 clap = { version = "3.2.12", features = ["derive"] }
 itertools = "0.10.3"
+num-parse = "0.1.2"

hencelisp/src/bin/main.rs

@@ -16,6 +16,20 @@ enum Commands {
         #[clap(value_parser)]
         src: String,
     },
+    #[clap(about = "Parses source code and outputs AST")]
+    Parse {
+        #[clap(value_parser)]
+        src: String,
+    },
+    #[clap(about = "Compiles source code to hence assembly")]
+    Compile {
+        #[clap(value_parser)]
+        src: String,
+        #[clap(value_parser)]
+        out: Option<String>,
+        #[clap(long, action)]
+        dump: bool,
+    },
 }
 
 fn main() {
@@ -26,5 +40,20 @@ fn main() {
             let tokens = lexer::lex(source).unwrap();
             dbg!(tokens);
         }
+        Commands::Parse { src } => {
+            let source = fs::read_to_string(src).unwrap();
+            let tokens = lexer::lex(source).unwrap();
+            let ast = parser::parse(tokens).unwrap();
+            dbg!(ast);
+        }
+        Commands::Compile { src, out, dump } => {
+            let source = fs::read_to_string(src).unwrap();
+            let tokens = lexer::lex(source).unwrap();
+            let ast = parser::parse(tokens).unwrap();
+
+            let mut data = compiler::Data::new();
+            compiler::compile(ast, &mut data).unwrap();
+            dbg!(data);
+        }
     }
 }

hencelisp/src/lib/compiler.rs

@@ -0,0 +1,61 @@
+use hence;
+
+use crate::parser;
+
+#[derive(Debug)]
+
+pub struct Data {
+    pub body: Vec<hence::parser::ast::Node>,
+}
+
+impl Data {
+    pub fn new() -> Self {
+        Self {
+            body: Vec::new(),
+        }
+    }
+}
+
+// pub fn create_call(list: Vec<parser::ast::Node>) -> Result<Call, &'static str> {
+//     let mut raw_content = list.iter();
+//     let content = raw_content.by_ref();
+
+//     let name = match content
+//         .skip_while(|n| matches!(n, parser::ast::Node::Comment(_)))
+//         .next()
+//     {
+//         Some(node) => match node {
+//             parser::ast::Node::String(x)
+//             | parser::ast::Node::Symbol(x)
+//             | parser::ast::Node::Literal(x) => x.clone(),
+//             _ => return Err("Invalid call name node literal"),
+//         },
+//         _ => return Err("Cannot create call from empty list"),
+//     };
+//     let args: Vec<_> = content.collect();
+//     dbg!(args);
+
+//     Ok(Call {
+//         name,
+//         args: Vec::new(),
+//     })
+// }
+
+pub fn compile(ast: parser::ast::AST, data: &mut Data) -> Result<(), &str> {
+    let mut iter = ast.body.into_iter().peekable();
+    while let Some(node) = iter.next() {
+        match node {
+            parser::ast::Node::Comment(_) => {}
+            parser::ast::Node::String(_) => return Err("Unexpected string on top level"),
+            parser::ast::Node::Number(_) => return Err("Unexpected number on top level"),
+            parser::ast::Node::Symbol(_) => return Err("Unexpected symbol on top level"),
+            parser::ast::Node::Literal(_) => return Err("Unexpected literal on top level"),
+
+            parser::ast::Node::List(x) => {
+                dbg!(&x);
+            }
+        }
+    }
+
+    Ok(())
+}

hencelisp/src/lib/lexer.rs

@@ -4,7 +4,6 @@ use itertools::Itertools;
 #[derive(Debug)]
 pub enum Token {
     Comment(String),
-    MultiLineComment(String),
 
     Newline(String),
     Whitespace(String),
@@ -12,8 +11,9 @@ pub enum Token {
     LParen,
     RParen,
 
-    StringLiteral(String),
+    String(String),
     Number(String),
+    Symbol(String),
     Literal(String),
 }
 
@@ -21,12 +21,11 @@ impl hence::assembler::ToCode for Token {
     fn to_code(&self) -> String {
         match self {
             Token::Comment(x) => format!(";;{x}"),
-            Token::MultiLineComment(x) => format!("#|{x}|#"),
             Token::Newline(x) | Token::Whitespace(x) => x.clone(),
             Token::LParen => "(".to_string(),
             Token::RParen => ")".to_string(),
-            Token::StringLiteral(x) => format!("\"{x}\""),
-            Token::Number(x) | Token::Literal(x) => x.clone(),
+            Token::String(x) => format!("\"{x}\""),
+            Token::Number(x) | Token::Symbol(x) | Token::Literal(x) => x.clone(),
         }
     }
 }
@@ -67,12 +66,26 @@ pub fn lex(source: String) -> Result<Vec<Token>, String> {
             }
             '"' => {
                 chars.next();
-                tokens.push(Token::StringLiteral(
+                tokens.push(Token::String(
                     chars.by_ref().take_while(|c| *c != '"').collect(),
                 ));
             }
+            _ if ch.is_numeric() => {
+                tokens.push(Token::Number(
+                    chars
+                        .peeking_take_while(|c| c.is_alphanumeric())
+                        .collect::<String>(),
+                ));
+            }
+            ':' => {
+                chars.next();
+                tokens.push(Token::Symbol(
+                    chars
+                        .peeking_take_while(|c| !c.is_whitespace() && *c != '(' && *c != ')')
+                        .collect(),
+                ));
+            }
             _ => {
-                dbg!(ch);
                 tokens.push(Token::Literal(
                     chars
                         .peeking_take_while(|c| !c.is_whitespace() && *c != '(' && *c != ')')

hencelisp/src/lib/lib.rs

@@ -1 +1,3 @@
+pub mod compiler;
 pub mod lexer;
+pub mod parser;

hencelisp/src/lib/parser.rs

@@ -0,0 +1,65 @@
+use crate::lexer;
+
+pub mod ast;
+
+pub fn parse(tokens: Vec<lexer::Token>) -> Result<ast::AST, &'static str> {
+    let mut iter = tokens.into_iter().peekable();
+    let mut body: Vec<ast::Node> = Vec::new();
+
+    while let Some(token) = iter.peek() {
+        match token {
+            lexer::Token::Comment(x) => {
+                body.push(ast::Node::Comment(x.trim().to_string()));
+                iter.next();
+            }
+            lexer::Token::Newline(_) | lexer::Token::Whitespace(_) => {
+                iter.next();
+            }
+            lexer::Token::LParen => {
+                iter.next();
+                let mut depth: usize = 1;
+                body.push(ast::Node::List(
+                    parse(
+                        iter.by_ref()
+                            .take_while(|t| match t {
+                                lexer::Token::LParen => {
+                                    depth += 1;
+                                    true
+                                }
+                                lexer::Token::RParen => {
+                                    depth -= 1;
+                                    depth != 0
+                                }
+                                _ => true,
+                            })
+                            .collect(),
+                    )
+                    .unwrap()
+                    .body,
+                ));
+            }
+            lexer::Token::RParen => return Err("Unexpected right parenthesis"),
+            lexer::Token::String(x) => {
+                body.push(ast::Node::String(x.clone()));
+                iter.next();
+            }
+            lexer::Token::Number(x) => {
+                body.push(ast::Node::Number(match num_parse::parse_int(x) {
+                    Some(y) => y,
+                    _ => return Err("Error parsing number"),
+                }));
+                iter.next();
+            }
+            lexer::Token::Symbol(x) => {
+                body.push(ast::Node::Symbol(x.clone()));
+                iter.next();
+            }
+            lexer::Token::Literal(x) => {
+                body.push(ast::Node::Literal(x.clone()));
+                iter.next();
+            }
+        }
+    }
+
+    Ok(ast::AST { body })
+}

hencelisp/src/lib/parser/ast.rs

@@ -0,0 +1,30 @@
+use hence;
+use itertools::Itertools;
+
+#[derive(Debug)]
+pub enum Node {
+    Comment(String),
+    List(Vec<Node>),
+    String(String),
+    Number(i32),
+    Symbol(String),
+    Literal(String),
+}
+
+impl hence::assembler::ToCode for Node {
+    fn to_code(&self) -> String {
+        match self {
+            Node::Comment(x) => format!(";; {x}"),
+            Node::List(x) => format!("({})", x.iter().map(|n| n.to_code()).join(" ")),
+            Node::String(x) => format!("\"{x}\""),
+            Node::Number(x) => x.to_string(),
+            Node::Symbol(x) => format!(":{x}"),
+            Node::Literal(x) => x.clone(),
+        }
+    }
+}
+
+#[derive(Debug)]
+pub struct AST {
+    pub body: Vec<Node>,
+}

hencelisp/src/main.rs

@@ -1,3 +0,0 @@
-fn main() {
-    println!("Hello, world!");
-}