Add lexer, parser, ast, and basic command line interaction

* Add lexer and parser using lrpar crate
* AST is added and constructed via lexer and parser
* Command line program `not` will read a file and parse it

Signed-off-by: Alek Ratzloff <alekratz@gmail.com>

src/bin/not.rs

@@ -1,5 +1,38 @@
-use not_python::syn;
+use not_python::syn::ast;
 
-fn main() {
+use std::{fs, path::PathBuf};
-    println!("hell world");
+use structopt::StructOpt;
+
+type Result<T> = std::result::Result<T, Box<dyn std::error::Error>>;
+
+#[derive(StructOpt, Debug)]
+struct Options {
+    /// The input file to work with.
+    ///
+    /// By default, the file will be executed. If -c is passed, it will not run and only compile.
+    /// Supplying - for the path will read from STDIN.
+    #[structopt(name = "FILE", parse(from_os_str))]
+    input: PathBuf,
+    /*
+    /// Disassemble object that would be passed to VM and exit before running it.
+    #[structopt(short = "d", long)]
+    disassemble: bool,
+    */
+}
+
+fn main() -> Result<()> {
+    let opt = Options::from_args();
+    let text = fs::read_to_string(&opt.input)?;
+    let ast = match ast::parse(text.as_str()) {
+        Ok(ast) => ast,
+        Err(errs) => {
+            for err in errs {
+                eprintln!("{}", err);
+            }
+            return Err("errors reported, exiting".into());
+        }
+    };
+    println!("{:#?}", ast);
+
+    Ok(())
 }

src/syn/ast.rs

@@ -115,14 +115,16 @@ pub enum Atom {
     String(String),
 }
 
-pub fn parse(text: &str) -> Result<Option<Vec<Stmt>>, Vec<lrpar::LexParseError<u32>>> {
+pub fn parse(text: &str) -> Result<Option<Vec<Stmt>>, Vec<String>> {
     use crate::syn::{lexer, parser};
     let lexerdef = lexer::lexerdef();
     let lexer = lexerdef.lexer(text);
     let (res, errors) = parser::parse(&lexer);
 
     if !errors.is_empty() {
-        return Err(errors)
+        return Err(errors.into_iter()
+            .map(|e| e.pp(&lexer, &parser::token_epp))
+            .collect())
     }
 
     Ok(res.transpose().unwrap())

src/syn/lexer.l

@@ -1,5 +1,5 @@
 %%
-[\n;]+                      "EOL"
+[\r\n;]+                    "EOL"
 [a-zA-Z_][a-zA-Z0-9_]*      "IDENT"
 :[a-zA-Z_][a-zA-Z0-9_]*     "SYM"
 [0-9]+                      "NUM"

src/syn/parser.y

@@ -12,6 +12,7 @@ Body -> Result<Vec<Stmt>>:
         Body 'EOL' Stmt {
             flatten($1, $3)
         }
+    |   Body 'EOL' { $1 }
     |   Stmt { Ok(vec![$1?]) }
     |   { Ok(Vec::new()) }
     ;
@@ -48,6 +49,9 @@ AccessExpr -> Result<Expr>:
         AtomExpr AccessExprTail {
             Ok(AccessExpr::new_expr($1?, $2?))
         }
+    |   AtomExpr {
+            Ok(AccessExpr::new_expr($1?, Default::default()))
+        }
     ;
 
 AccessExprTail -> Result<Vec<Access>>: