diff --git a/.gitignore b/.gitignore old mode 100644 new mode 100755 index 4495404..91a8d77 --- a/.gitignore +++ b/.gitignore @@ -26,3 +26,5 @@ rust-project.json # End of https://www.toptal.com/developers/gitignore/api/rust,rust-analyzer .zed/ + +.idea/ diff --git a/.rules b/.rules new file mode 100644 index 0000000..e69de29 diff --git a/Cargo.toml b/Cargo.toml old mode 100644 new mode 100755 diff --git a/README.org b/README.org new file mode 100755 index 0000000..4dce23c --- /dev/null +++ b/README.org @@ -0,0 +1,36 @@ +#+title: README + +* Syntax example: +#+begin_src +funzione :: fn(parametro1: Number, parametro2: String): String do + +end +#+end_src + +* Planning: +allora quello che devo fare è: +** TODO Controllare se ho finito la questione del retourn:labe +** TODO implement struct: +#+begin_src rlox +Cosa :: struct{ + field1: Number +} +#+end_src +*** TODO update lexert +*** TODO update ast +*** TODO update parser +** TODO implement partial function ... +#+begin_src rlox +cosa := make(Cosa, feld1: 42) +function :: fn(self:Cosa, b: Number) -> Number +res := function(cosa,44) +#+end_src +*** TODO ... to undericlty create method +#+begin_src rlox +res := cosa.function(44) +#+end_src +*** TODO ... to make pipeline +#+begin_src rlox +res := cosa |> function(44) +#+end_src +** TODO Understand what is a type system diff --git a/examples/test_error.lox b/examples/fail_error.lox old mode 100644 new mode 100755 similarity index 65% rename from examples/test_error.lox rename to examples/fail_error.lox index 698ea42..4bd04c0 --- a/examples/test_error.lox +++ b/examples/fail_error.lox @@ -1,8 +1,9 @@ // Test script to verify error positioning -var a = 5; -var b = "hello"; +"no"(); +var a := 5; +var b := "hello"; print 0; -var result = a + b; // This should cause a type error +var result := a + b; // This should cause a type error print 1; 43 + "hello"; // This should cause a type error print result; diff --git a/examples/function_call.lox b/examples/function_call.lox deleted file mode 100644 index 663d520..0000000 --- a/examples/function_call.lox +++ /dev/null @@ -1,20 +0,0 @@ - -function :: fn() do - print "Function"; -end - - -function_factory :: fn() do - print "Function Factory"; - function -end - -function_fatcoty_factory :: fn() do - print "Function Factory Factory"; - function_factory -end - -//function(); -//function_factory()(); -function_fatcoty_factory()()(); -clock() diff --git a/examples/test_closure.lox b/examples/test_closure.lox new file mode 100755 index 0000000..7b0586c --- /dev/null +++ b/examples/test_closure.lox @@ -0,0 +1,27 @@ + +closure :: fn(): Any do + value := "Closure"; + internal :: fn(): String do + print value; + return value; + end; + return internal; +end; + +internal := closure(); +internal(); + +a := "Global"; +b := a; +c := b; +c = b = a; +d := c = b = a; +do + showA :: fn(): Nil do + print a; + end; + + showA(); + a := "Local"; + showA(); +end diff --git a/examples/example.lox b/examples/test_example.lox old mode 100644 new mode 100755 similarity index 53% rename from examples/example.lox rename to examples/test_example.lox index b9a3c8b..b439140 --- a/examples/example.lox +++ b/examples/test_example.lox @@ -4,25 +4,28 @@ // address: String, // } //func_name :: fn(param: Number, param2: String) {param is Int} do -func_name :: fn(param: Int, param2: String) do +func_name :: fn(param: Int, param2: String): Any do print param; - cavallo := 5; - print cavallo; - while cavallo < 10 do - cavallo = cavallo + 1; - print cavallo; + print param2; + while param < param2 do + param = param + 1; + print param; + print param2; + //break; end - if True then do + if False then do print "this shoul be stop"; - return False; + return "Hahahaha"; end print "this shoud be un other stop"; - return 42; + //return 42; for i := 11; i <= 21; i = i + 1; do print i; + print "hello"; + break; end return False or True; -end +end; -func_name(1,2) +func_name(1,20); diff --git a/examples/test_function_call.lox b/examples/test_function_call.lox new file mode 100755 index 0000000..013d673 --- /dev/null +++ b/examples/test_function_call.lox @@ -0,0 +1,30 @@ + +function :: fn(): Nil do + print "Function"; +end; + + +function_factory :: fn(): Any do + print "Function Factory"; + function; +end; + +function_fatcoty_factory :: fn(): Any do + print "Function Factory Factory"; + return function_factory; +end; + +//function(); +//function_factory()(); +//function_fatcoty_factory()()(); +clock(); + +closure :: fn(): Any do + value := "Closure"; + internal :: fn(): Nil do + print value; + end; + return internal; +end; + +closure(); diff --git a/examples/test_source_tracking.lox b/examples/test_source_tracking.lox old mode 100644 new mode 100755 index 5fad933..5f879de --- a/examples/test_source_tracking.lox +++ b/examples/test_source_tracking.lox @@ -1,10 +1,10 @@ // Test file for source slice tracking -var x = 42; +var x := 42; print x; if x > 0 then do print "positive"; - var y = x + 1; + var y := x + 1; end elif x == 0 then do print "zero"; @@ -19,7 +19,7 @@ while x > 0 do end do - var z = 10; + var z := 10; print z * 2; end diff --git a/grammar.org b/grammar.org new file mode 100644 index 0000000..aecd0a5 --- /dev/null +++ b/grammar.org @@ -0,0 +1,57 @@ + +statement -> print_statement + | return_statement + | break_statement + | block_statement + | var_statement + | if_statement + | while_statement + | for_statement + | expression_statement + +print_statement -> "print" expression ";" +return_statement -> "return" expression? ";" +break_statement -> "break" ";" +block_statement -> "do" statement* "end" +expression_statement -> expression ";"? + +; leading "var" optional; the ":" is required +var_statement -> "var"? IDENTIFIER ":" IDENTIFIER? + ( variable_declaration + | function_declaration + | struct_declaration ) ; struct = WIP +variable_declaration -> ( "=" expression )? ";" +function_declaration -> ":" "fn" "(" parameters? ")" ( "{" expression "}" )? statement +parameters -> IDENTIFIER ( ":" IDENTIFIER )? + ( "," IDENTIFIER ( ":" IDENTIFIER )? )* + +if_statement -> "if" expression "then" statement + ( "elif" expression "then" statement )* + ( "else" statement )? +while_statement -> "while" expression statement +for_statement -> "for" var_statement expression ";" expression_statement statement + +; ---------- expressions (lowest -> highest precedence) ---------- +expression -> assignment +assignment -> IDENTIFIER "=" assignment ; right-associative + | pipe +pipe -> logic ( "|>" logic )* ; left-assoc; x |> f(a) == f(x, a) +logic -> is ( ( "or" | "and" ) is )* +is -> equality ( "is" equality )* +equality -> comparison ( ( "==" | "!=" ) comparison )* +comparison -> term ( ( ">" | ">=" | "<" | "<=" ) term )* +term -> factor ( ( "+" | "-" ) factor )* +factor -> unary ( ( "*" | "/" ) unary )* +unary -> ( "!" | "-" ) unary | call +call -> primary ( "(" arguments? ")" )* +arguments -> expression ( "," expression )* +primary -> NUMBER | STRING | "true" | "false" | "nil" + | "(" expression ")" | IDENTIFIER + | dictionary | list | ATOM ; WIP + +; ---------- work in progress (lexed; parser support being added) ---------- +struct_declaration -> "struct" "{" field* "}" +field -> IDENTIFIER ( ":" IDENTIFIER )? ";" +dictionary -> "dict" "(" IDENTIFIER ":" expression + ( "," IDENTIFIER ":" expression )* ")" +list -> "list" "(" expression ( "," expression )* ")" diff --git a/src/backend/environment.rs b/src/backend/environment.rs old mode 100644 new mode 100755 index 97166ec..2ba5832 --- a/src/backend/environment.rs +++ b/src/backend/environment.rs @@ -1,34 +1,45 @@ -use std::collections::HashMap; - use crate::{ - common::{ - base_value::BaseValue, - lox_result::{runtime_error, LoxResult}, - }, + common::lox_result::{runtime_error, LoxResult}, frontend::source_registry::SourceSlice, }; +use std::collections::HashMap; +use std::fmt::Debug; -#[derive(Debug, Clone, PartialEq)] -pub struct EnvironmentStack { - stack: Vec>, +pub type Environment = HashMap; + +#[derive(Debug, Clone, PartialEq, Default)] +pub struct EnvironmentStack { + stack: Vec>, } -impl EnvironmentStack { +impl EnvironmentStack { pub fn new() -> Self { EnvironmentStack { stack: vec![HashMap::new()], } } + pub fn is_empty(&self) -> bool { + self.stack.is_empty() + } + pub fn push_new_scope(&mut self) { self.stack.push(HashMap::new()); } + pub fn push_existing_scope(&mut self, scope: HashMap) { + self.stack.push(scope); + } + pub fn pop_scope(&mut self) { self.stack.pop(); } - pub fn get(&self, name: &str) -> LoxResult { + pub fn clone_last_scope(&mut self) -> HashMap { + self.stack.last().unwrap().clone() + } + + pub fn get(&self, name: &str) -> LoxResult { let size = self.stack.len(); for i in (0..size).rev() { @@ -37,17 +48,49 @@ impl EnvironmentStack { } } runtime_error( - SourceSlice::default(), // todo change this to the actual source slice + SourceSlice::synthetic(), // todo change this to the actual source slice format!("Undefined variable '{}'", name), ) } - pub fn declare(&mut self, name: String, value: BaseValue) -> LoxResult { + pub fn declare(&mut self, name: String, value: T) -> LoxResult { self.stack.last_mut().unwrap().insert(name, value.clone()); Ok(value) } - pub fn set(&mut self, name: String, value: BaseValue) -> LoxResult { + /// Read a variable at a known scope `distance` from the innermost scope + /// (0 = innermost). Used with resolver-computed distances. + pub fn get_at(&self, distance: usize, name: &str) -> LoxResult { + let index = self.stack.len().checked_sub(distance + 1); + match index + .and_then(|i| self.stack.get(i)) + .and_then(|s| s.get(name)) + { + Some(value) => Ok(value.clone()), + None => runtime_error( + SourceSlice::synthetic(), + format!("Undefined variable '{}'", name), + ), + } + } + + /// Assign to a variable at a known scope `distance` from the innermost + /// scope (0 = innermost). + pub fn assign_at(&mut self, distance: usize, name: String, value: T) -> LoxResult { + let index = self.stack.len().checked_sub(distance + 1); + match index { + Some(i) if i < self.stack.len() => { + self.stack[i].insert(name, value.clone()); + Ok(value) + } + _ => runtime_error( + SourceSlice::synthetic(), + format!("Undefined variable '{}'", name), + ), + } + } + + pub fn set(&mut self, name: String, value: T) -> LoxResult { let size = self.stack.len(); for i in (0..size).rev() { @@ -64,8 +107,102 @@ impl EnvironmentStack { } } runtime_error( - SourceSlice::default(), // todo change this to the actual source slice + SourceSlice::synthetic(), // todo change this to the actual source slice format!("Undefined variable '{}'", name), ) } + + pub fn depth(&self) -> usize { + self.stack.len() + } + pub fn scope_contains(&self, index: usize, name: &str) -> bool { + self.stack[index].contains_key(name) + } +} + +#[cfg(test)] +mod tests { + use super::*; + + #[test] + fn declare_and_get() { + let mut env: EnvironmentStack = EnvironmentStack::new(); + env.declare("a".to_string(), 1).unwrap(); + assert_eq!(env.get("a").unwrap(), 1); + } + + #[test] + fn get_undefined_is_error() { + let env: EnvironmentStack = EnvironmentStack::new(); + assert!(env.get("nope").is_err()); + } + + #[test] + fn declare_overwrites_in_same_scope() { + let mut env: EnvironmentStack = EnvironmentStack::new(); + env.declare("a".to_string(), 1).unwrap(); + env.declare("a".to_string(), 2).unwrap(); + assert_eq!(env.get("a").unwrap(), 2); + } + + #[test] + fn inner_scope_shadows_outer_and_unshadows_on_pop() { + let mut env: EnvironmentStack = EnvironmentStack::new(); + env.declare("a".to_string(), 1).unwrap(); + env.push_new_scope(); + env.declare("a".to_string(), 2).unwrap(); + assert_eq!(env.get("a").unwrap(), 2); + env.pop_scope(); + assert_eq!(env.get("a").unwrap(), 1); + } + + #[test] + fn get_falls_through_to_outer_scope() { + let mut env: EnvironmentStack = EnvironmentStack::new(); + env.declare("a".to_string(), 1).unwrap(); + env.push_new_scope(); + assert_eq!(env.get("a").unwrap(), 1); + } + + #[test] + fn set_updates_existing_value_in_outer_scope() { + let mut env: EnvironmentStack = EnvironmentStack::new(); + env.declare("a".to_string(), 1).unwrap(); + env.push_new_scope(); + env.set("a".to_string(), 9).unwrap(); + env.pop_scope(); + assert_eq!(env.get("a").unwrap(), 9); + } + + #[test] + fn set_undefined_is_error() { + let mut env: EnvironmentStack = EnvironmentStack::new(); + assert!(env.set("a".to_string(), 1).is_err()); + } + + #[test] + fn pop_scope_discards_inner_declarations() { + let mut env: EnvironmentStack = EnvironmentStack::new(); + env.push_new_scope(); + env.declare("temp".to_string(), 5).unwrap(); + env.pop_scope(); + assert!(env.get("temp").is_err()); + } + + #[test] + fn push_existing_scope_makes_values_visible() { + let mut scope = HashMap::new(); + scope.insert("k".to_string(), 7); + let mut env: EnvironmentStack = EnvironmentStack::new(); + env.push_existing_scope(scope); + assert_eq!(env.get("k").unwrap(), 7); + } + + #[test] + fn clone_last_scope_returns_top_scope() { + let mut env: EnvironmentStack = EnvironmentStack::new(); + env.declare("a".to_string(), 1).unwrap(); + let scope = env.clone_last_scope(); + assert_eq!(scope.get("a"), Some(&1)); + } } diff --git a/src/backend/interpreter.rs b/src/backend/interpreter.rs old mode 100644 new mode 100755 index 235467f..00969ef --- a/src/backend/interpreter.rs +++ b/src/backend/interpreter.rs @@ -1,73 +1,29 @@ use crate::{ backend::environment::EnvironmentStack, common::{ - ast::{AstNode, AstNodeKind, Expr, Stmt}, + ast::{AstNode, Expr, NodeId}, base_value::{BaseValue, NativeFunction, Number, Truthy}, lox_result::{runtime_error, LoxError, LoxResult}, }, frontend::{source_registry::SourceSlice, tokens::TokenType}, }; -use std::fmt::{Debug, Display}; +use std::collections::HashMap; pub struct Interpreter { - enviorment: EnvironmentStack, + enviorment: EnvironmentStack, + /// Per-reference scope distances from the resolver. Empty means "no + /// resolution pass was run", in which case lookups fall back to a dynamic + /// search of the scope chain. + locals: HashMap, } pub trait EvaluateInterpreter { fn evaluate(&mut self, stmt: T) -> LoxResult; } -impl<'a, R: AstNodeKind + Clone + Debug + Display> EvaluateInterpreter> for Interpreter -where - Interpreter: EvaluateInterpreter, -{ - fn evaluate(&mut self, stmt: AstNode) -> LoxResult { - match self.evaluate(stmt.node.clone()) { - Ok(value) => Ok(value), - Err(err) => runtime_error(stmt.source_slice, err.get_message()), - } - } -} - -// Direct Expr evaluation to avoid infinite recursion -impl EvaluateInterpreter for Interpreter { - fn evaluate(&mut self, expr: Expr) -> LoxResult { - match expr { - Expr::Literal { value } => Ok(value), - Expr::Identifier { name } => self.enviorment.get(&name), - Expr::Binary { - left, - operator, - right, - } => { - let left_val = self.evaluate(*left)?; - let right_val = self.evaluate(*right)?; - self.evaluate_binary(left_val, operator, right_val) - } - Expr::Unary { operator, operand } => { - let operand_val = self.evaluate(*operand)?; - self.evaluate_unary(operator, operand_val) - } - Expr::Grouping { expression } => self.evaluate(*expression), - Expr::Call { callee, arguments } => self.evaluate_call(callee, arguments), - } - } -} - -impl EvaluateInterpreter> for Interpreter { - fn evaluate(&mut self, node: AstNode) -> LoxResult { - let stmt = node.node; - let result = self.interpret_stmt_inner(stmt); - match result { - Ok(value) => Ok(value), - Err(LoxError::RuntimeError { - message, - source_slice, - }) if source_slice == SourceSlice::default() => { - runtime_error(node.source_slice.clone(), message) - } - Err(err) => Err(err), - } +impl EvaluateInterpreter for Interpreter { + fn evaluate(&mut self, node: AstNode) -> LoxResult { + self.eval_expr(&node) } } @@ -85,27 +41,150 @@ impl Interpreter { )) })), ); - Self { enviorment: env } + Self { + enviorment: env, + locals: HashMap::new(), + } } - fn evaluate_call( + + /// Install the resolver's per-reference scope distances. + pub fn set_locals(&mut self, locals: HashMap) { + self.locals = locals; + } + + /// Evaluate an expression node, threading its `NodeId` so resolved + /// variables and assignments can use their precomputed scope distance. + fn eval_expr(&mut self, node: &AstNode) -> LoxResult { + let result = match &node.node { + Expr::Literal { value } => match &**value { + BaseValue::Function(func) => { + let mut func = func.clone(); + func.closure = Some(self.enviorment.clone()); + Ok(BaseValue::Function(func)) + } + other => Ok(other.clone()), + }, + Expr::Identifier { name } => self.look_up_variable(name, node.id), + Expr::Binary { + left, + operator, + right, + } => { + let left_val = self.eval_expr(left)?; + let right_val = self.eval_expr(right)?; + self.evaluate_binary(left_val, operator.clone(), right_val) + } + Expr::Unary { operator, operand } => { + let operand_val = self.eval_expr(operand)?; + self.evaluate_unary(operator.clone(), operand_val) + } + Expr::Grouping { expression } => self.eval_expr(expression), + Expr::Call { callee, arguments } => self.evaluate_call(callee, arguments), + Expr::Assign { name, value } => { + let value = self.eval_expr(value)?; + self.assign_variable(name, node.id, value) + } + Expr::Print { expression } => { + let value = self.eval_expr(expression)?; + println!("{}", value); + Ok(BaseValue::Nil) + } + Expr::VarDeclaration { + name, initializer, .. + } => { + let value = match initializer { + Some(expr_node) => self.eval_expr(expr_node)?, + None => BaseValue::Nil, + }; + self.enviorment.declare(name.clone(), value) + } + Expr::Return { expression, .. } => self.eval_expr(expression), + Expr::Block { statements, .. } => self.evaluate_block(statements), + Expr::If { + condition, + then_branch, + elif_branches, + else_branch, + } => self.evaluate_if(condition, then_branch, elif_branches, else_branch), + Expr::While { condition, body } => { + let mut ret = BaseValue::Nil; + while self.eval_expr(condition)?.is_truthy() { + match self.eval_expr(body) { + Ok(val) => ret = val, + Err(LoxError::Return { value, .. }) => { + ret = value; + break; + } + Err(err) => return Err(err), + }; + } + Ok(ret) + } + Expr::For { + variable, + condition, + increment, + body, + } => { + let source_slice = variable.source_slice.clone(); + let val = self.eval_expr(variable)?; + if !matches!(val, BaseValue::Number(..)) { + return runtime_error(source_slice, "Expected number literal"); + } + let mut ret = BaseValue::Nil; + while self.eval_expr(condition)?.is_truthy() { + match self.eval_expr(body) { + Ok(val) => ret = val, + Err(LoxError::Return { value, .. }) => { + ret = value; + break; + } + Err(err) => return Err(err), + }; + self.eval_expr(increment)?; + } + Ok(ret) + } + }; + // Give location-less runtime errors this node's span. + match result { + Err(LoxError::RuntimeError { + message, + source_slice, + }) if source_slice.is_synthetic() => runtime_error(node.source_slice.clone(), message), + other => other, + } + } + + /// Look up a variable: by resolved distance if known, else dynamically. + fn look_up_variable(&self, name: &str, id: NodeId) -> LoxResult { + match self.locals.get(&id) { + Some(&distance) => self.enviorment.get_at(distance, name), + None => self.enviorment.get(name), + } + } + + /// Assign a variable: at its resolved distance if known, else dynamically. + fn assign_variable( &mut self, - callee: Box>, - arguments: Vec>, + name: &str, + id: NodeId, + value: BaseValue, ) -> LoxResult { + match self.locals.get(&id) { + Some(&distance) => self.enviorment.assign_at(distance, name.to_string(), value), + None => self.enviorment.set(name.to_string(), value), + } + } + + fn evaluate_call(&mut self, callee: &AstNode, arguments: &[AstNode]) -> LoxResult { let source_slice = callee.source_slice.clone(); - // Estrai il nome della variabile se il callee è un identificatore - let function_name = match &callee.node { - Expr::Identifier { name } => Some(name.clone()), - _ => None, - }; - - let function = if let Some(name) = function_name { - // Se abbiamo un nome di variabile, ottieni la funzione dall'ambiente - self.enviorment.get(&name)? - } else { - // Altrimenti valuta l'espressione callee (per casi più complessi) - self.evaluate(*callee)? + // A bare identifier callee resolves through `locals`; anything else is + // evaluated as a general expression. + let function = match &callee.node { + Expr::Identifier { name } => self.look_up_variable(name, callee.id)?, + _ => self.eval_expr(callee)?, }; if !function.is_callable() { @@ -114,21 +193,38 @@ impl Interpreter { let evaluated_arguments = arguments .iter() - .map(|arg| self.evaluate(arg.clone())) + .map(|arg| self.eval_expr(arg)) .collect::, LoxError>>()?; match function { BaseValue::NativeFunction(func) => Ok((func.function)(&evaluated_arguments)), BaseValue::Function(func) => { - func.parameters - .iter() - .enumerate() - .map(|(index, par)| { - let value = evaluated_arguments.get(index).unwrap(); - self.enviorment.declare(par.0.clone(), value.clone()) - }) - .collect::>>()?; - self.evaluate(func.body) + let saved_env = self.enviorment.clone(); + // If the function captured a closure, use it as the base environment + if let Some(closure_env) = func.closure { + self.enviorment = closure_env; + } + + // Push a new scope for the function's parameters + self.enviorment.push_new_scope(); + + // Declare parameters in the new scope + for (index, par) in func.parameters.iter().enumerate() { + let value = evaluated_arguments.get(index).unwrap(); + self.enviorment.declare(par.0.clone(), value.clone())?; + } + + // Execute the function body + let result = match self.eval_expr(&func.body) { + Ok(value) => Ok(value), + Err(LoxError::Return { value, .. }) => Ok(value), + Err(err) => Err(err), + }; + + // Restore the original environment (cleanup is automatic) + self.enviorment = saved_env; + + result } _ => runtime_error( source_slice.clone(), @@ -157,7 +253,7 @@ impl Interpreter { TokenType::And => Ok(BaseValue::Boolean(left.is_truthy() && right.is_truthy())), TokenType::Or => Ok(BaseValue::Boolean(left.is_truthy() || right.is_truthy())), _ => Err(LoxError::RuntimeError { - source_slice: SourceSlice::default(), + source_slice: SourceSlice::synthetic(), message: format!("Unsupported binary operator: {:?}", operator), }), } @@ -168,105 +264,39 @@ impl Interpreter { TokenType::Minus => match operand { BaseValue::Number(n) => Ok(BaseValue::Number(n.neg())), _ => Err(LoxError::RuntimeError { - source_slice: SourceSlice::default(), + source_slice: SourceSlice::synthetic(), message: "Cannot negate non-numeric value".to_string(), }), }, TokenType::Bang => Ok(!operand), _ => Err(LoxError::RuntimeError { - source_slice: SourceSlice::default(), + source_slice: SourceSlice::synthetic(), message: format!("Unsupported unary operator: {:?}", operator), }), } } - fn interpret_stmt_inner(&mut self, stmt: Stmt) -> LoxResult { - match stmt { - Stmt::Expression { expression } => self.evaluate(*expression), - Stmt::Print { expression } => { - let value = self.evaluate(*expression)?; - println!("{}", value); - Ok(BaseValue::Nil) - } - Stmt::Block { statements } => self.evaluate_block(*statements), - Stmt::Stmt { expression } => self.evaluate(*expression.clone()), - Stmt::Return { expression } => self.evaluate(*expression), - Stmt::VarDeclaration { name, initializer } => { - let value = match initializer { - Some(expr_node) => match &expr_node.node { - Expr::Literal { value } => { - if value.is_callable() { - value.clone() - } else { - value.clone() - } - } - _ => self.evaluate(*expr_node)?, - }, - None => BaseValue::Nil, - }; - self.enviorment.declare(name.clone(), value) - } - - Stmt::VarAssigment { name, value } => { - let result = self.evaluate(*value)?; - self.enviorment.set(name.clone(), result) - } - Stmt::If { - condition, - then_branch, - elif_branch, - else_branch, - } => self.evaluate_if(condition, then_branch, elif_branch, else_branch), - Stmt::While { condition, body } => { - while self.evaluate(*condition.clone())?.is_truthy() { - self.evaluate(*body.clone())?; - } - Ok(BaseValue::Nil) - } - Stmt::For { - variable, - condition, - increment, - body, - } => { - self.enviorment.push_new_scope(); - let source_slice = variable.source_slice.clone(); - let val = self.evaluate(*variable)?; - if !matches!(val, BaseValue::Number(..)) { - return runtime_error(source_slice, "Expected number literal"); - } - let mut ret = BaseValue::Nil; - while self.evaluate(*condition.clone())?.is_truthy() { - ret = self.evaluate(*body.clone())?; - self.evaluate(*increment.clone())?; - } - - Ok(ret) - } - } - } fn evaluate_if( &mut self, - condition: Box>, - then_branch: Box>, - elif_branch: Vec<(Box>, Box>)>, - else_branch: Option>>, + condition: &AstNode, + then_branch: &AstNode, + elif_branch: &[(Box, Box)], + else_branch: &Option>, ) -> LoxResult { - let condition = self.evaluate(*condition)?; + let condition = self.eval_expr(condition)?; match condition { - BaseValue::Boolean(true) => self.evaluate(*then_branch), + BaseValue::Boolean(true) => self.eval_expr(then_branch), BaseValue::Boolean(false) => { for (elif_condition, elif_then_branch) in elif_branch { - let condition = self.evaluate(*elif_condition)?; + let condition = self.eval_expr(elif_condition)?; match condition { BaseValue::Boolean(true) => { - return self.evaluate(*elif_then_branch); + return self.eval_expr(elif_then_branch); } BaseValue::Boolean(false) => continue, _ => { return Err(LoxError::TypeMismatch { - source_slice: SourceSlice::default(), // todo change this to the actual source slice + source_slice: SourceSlice::synthetic(), // todo change this to the actual source slice expected: "boolean".to_string(), found: condition.to_string(), }); @@ -274,49 +304,179 @@ impl Interpreter { }; } if let Some(else_block) = else_branch { - self.evaluate(*else_block) + self.eval_expr(else_block) } else { Ok(BaseValue::Nil) } } _ => Err(LoxError::TypeMismatch { - source_slice: SourceSlice::default(), // todo change this to the actual source slice + source_slice: SourceSlice::synthetic(), // todo change this to the actual source slice expected: "boolean".to_string(), found: condition.to_string(), }), } } - fn evaluate_block(&mut self, statements: Vec>) -> LoxResult { + fn evaluate_block(&mut self, statements: &[AstNode]) -> LoxResult { self.enviorment.push_new_scope(); - let (elements, final_expr) = match statements.split_last() { - Some((stmt, body)) => match &stmt.node { - Stmt::Expression { expression } => (body, Some(expression)), - Stmt::Return { expression } => (body, Some(expression)), - _ => (statements.as_slice(), None), - }, - None => { - (&[][..], None) // Blocco vuoto - } - }; + let mut result = Ok(BaseValue::Nil); + for statement in statements.iter() { + match &statement.node { + Expr::Return { expression, .. } => { + let value = self.eval_expr(expression)?; - // Ora elements è sempre disponibile - for statement in elements.iter() { - self.evaluate((*statement).clone())?; - } - - // Gestisci l'espressione finale se presente - match final_expr { - Some(expr) => { - let res = self.evaluate(*expr.clone()); - self.enviorment.pop_scope(); - res - } - None => { - self.enviorment.pop_scope(); - Ok(BaseValue::Nil) - } + result = Err(LoxError::Return { + source_slice: statement.source_slice.clone(), + value, + return_label: "Hi".to_string(), + }); + break; + } + _ => result = self.eval_expr(statement), + }; } + self.enviorment.pop_scope(); + result + } +} + +#[cfg(test)] +mod tests { + use super::*; + use crate::frontend::lexer::Lexer; + use crate::frontend::parser::Parser; + + /// Lex, parse and interpret `src`, returning the value of the last statement. + fn eval(src: &str) -> LoxResult { + let tokens = Lexer::new(src.to_string(), 0) + .scans_tokens() + .expect("source should lex without errors"); + let statements = Parser::new(tokens) + .parse() + .expect("source should parse without errors"); + let mut interpreter = Interpreter::new(); + let mut last = BaseValue::Nil; + for stmt in statements { + last = interpreter.evaluate(stmt)?; + } + Ok(last) + } + + #[test] + fn evaluates_arithmetic_with_precedence() { + assert_eq!(eval("1 + 2 * 3;").unwrap().to_string(), "7"); + } + + #[test] + fn grouping_changes_precedence() { + assert_eq!(eval("(1 + 2) * 3;").unwrap().to_string(), "9"); + } + + #[test] + fn division_and_subtraction() { + assert_eq!(eval("10 - 4 / 2;").unwrap().to_string(), "8"); + } + + #[test] + fn comparison_yields_boolean() { + assert_eq!(eval("1 < 2;").unwrap(), BaseValue::Boolean(true)); + assert_eq!(eval("2 < 1;").unwrap(), BaseValue::Boolean(false)); + } + + #[test] + fn equality_operators() { + assert_eq!(eval("1 == 1;").unwrap(), BaseValue::Boolean(true)); + assert_eq!(eval("1 != 2;").unwrap(), BaseValue::Boolean(true)); + } + + #[test] + fn unary_negation() { + assert_eq!(eval("-5;").unwrap().to_string(), "-5"); + } + + #[test] + fn logical_not() { + assert_eq!(eval("!true;").unwrap(), BaseValue::Boolean(false)); + } + + #[test] + fn concatenates_strings() { + match eval("\"a\" + \"b\";").unwrap() { + BaseValue::String(s) => assert!(s.contains('a') && s.contains('b')), + other => panic!("expected string, got {:?}", other), + } + } + + #[test] + fn declares_and_reads_variable() { + assert_eq!(eval("var x: Int = 10; x + 5;").unwrap().to_string(), "15"); + } + + #[test] + fn assignment_updates_variable() { + assert_eq!( + eval("var x: Int = 1; x = 42; x;").unwrap().to_string(), + "42" + ); + } + + #[test] + fn if_takes_then_branch_when_true() { + assert_eq!( + eval("var x: Int = 0; if true then x = 1; x;") + .unwrap() + .to_string(), + "1" + ); + } + + #[test] + fn if_takes_else_branch_when_false() { + assert_eq!( + eval("var x: Int = 0; if false then x = 1; else x = 2; x;") + .unwrap() + .to_string(), + "2" + ); + } + + #[test] + fn while_loop_runs_until_condition_false() { + let src = "var i: Int = 0; while i < 3 do i = i + 1; end i;"; + assert_eq!(eval(src).unwrap().to_string(), "3"); + } + + #[test] + fn block_scope_does_not_leak_outer_assignment() { + // `set` walks outer scopes, so the outer `x` is updated from inside the block. + let src = "var x: Int = 1; do x = 5; end x;"; + assert_eq!(eval(src).unwrap().to_string(), "5"); + } + + #[test] + fn defines_and_calls_a_function() { + let src = "add :: fn (a, b): Number do return a + b; end; add(2, 3);"; + assert_eq!(eval(src).unwrap().to_string(), "5"); + } + + #[test] + fn clock_native_function_returns_a_number() { + assert!(matches!(eval("clock();").unwrap(), BaseValue::Number(..))); + } + + #[test] + fn undefined_variable_is_runtime_error() { + assert!(eval("missing;").is_err()); + } + + #[test] + fn calling_a_non_function_is_runtime_error() { + assert!(eval("var x: Int = 5; x();").is_err()); + } + + #[test] + fn adding_incompatible_types_is_runtime_error() { + assert!(eval("1 + true;").is_err()); } } diff --git a/src/backend/mod.rs b/src/backend/mod.rs old mode 100644 new mode 100755 diff --git a/src/common/ast.rs b/src/common/ast.rs old mode 100644 new mode 100755 index 0127b79..fdfab03 --- a/src/common/ast.rs +++ b/src/common/ast.rs @@ -1,65 +1,87 @@ use crate::{ - common::base_value::BaseValue, + common::{base_value::BaseValue, types::Type}, frontend::{source_registry::SourceSlice, tokens::TokenType}, }; use std::fmt::{Debug, Display}; -/* - * grammar: - * program -> statement* EOF - * statement -> expression_statement - * | print_statement - * | var_statement - * | block_statement - * | assignment_statement - * | if_statement - * - * expression_statement -> expression ";" - * print_statement -> "print" expression ";" - * var_statement -> ("var"|"dyn"|"mut")? IDENTIFIER (":" IDENTIFIER)? (("=" expression)? ";")| function_declaration - * function_declaration -> "::" "(" parameters? ")" ("->" IDENTIFIER)? block_statement - * parameters -> ( IDENTIFIER (":" IDENTIFIER)? ("," IDENTIFIER (":" IDENTIFIER)? )* ) - * assignment_statement -> IDENTIFIER "=" expression ";" - * block_statement -> "do" statement* "end" - * if_statement -> "if" expression "then" statement ("elif" expression "then" statement)* ("else" statement)? "end" - * while_statement -> "while" expression "do" statement "end" - * - * expression -> assignment - * assignment -> IDENTIFIER "=" assignment | logical_or - * logical_or -> logical_and (("or" logical_and)* - * logical_and -> logical_is (("and" logical_is)* - * logical_is -> equality (("is" equality)* - * equality -> comparison (("==" | "!=") comparison)* - * comparison -> term ((">" | ">=" | "<" | "<=") term)* - * term -> factor (("+" | "-") factor)* - * factor -> unary (("*" | "/") unary)* - * unary -> ("!" | "-") unary | call - * call -> primary ("(" arguments ")")* - * arguments -> expression ("," expression)* - * primary -> NUMBER | STRING | "true" | "false" | "nil" | "(" expression ")" | IDENTIFIER - */ +use std::sync::atomic::{AtomicUsize, Ordering}; + +/// A unique identity for an AST node, assigned once at construction. +/// +/// Unlike a [`SourceSlice`] (which describes *where* a node is, for +/// diagnostics), a `NodeId` describes *which* node it is. It is stable across +/// clones, so analyses such as the resolver can key per-reference data on it +/// without relying on source positions being unique. +#[derive(Debug, Clone, Copy, PartialEq, Eq, Hash)] +pub struct NodeId(pub usize); + +static NEXT_NODE_ID: AtomicUsize = AtomicUsize::new(0); + +impl NodeId { + /// Allocate the next globally-unique node id. + pub fn next() -> Self { + NodeId(NEXT_NODE_ID.fetch_add(1, Ordering::Relaxed)) + } +} + #[derive(Clone, PartialEq)] pub enum Expr { Literal { - value: BaseValue, + value: Box, }, Binary { - left: Box>, + left: Box, operator: TokenType, - right: Box>, + right: Box, }, Unary { operator: TokenType, - operand: Box>, + operand: Box, }, Grouping { - expression: Box>, + expression: Box, }, Identifier { name: String, }, Call { - callee: Box>, - arguments: Vec>, + callee: Box, + arguments: Vec, + }, + Assign { + name: String, + value: Box, + }, + Print { + expression: Box, + }, + VarDeclaration { + name: String, + var_type: Type, + initializer: Option>, + }, + Return { + expression: Box, + label: String, + }, + Block { + statements: Box>, + label: String, + }, + If { + condition: Box, + then_branch: Box, + elif_branches: Vec<(Box, Box)>, + else_branch: Option>, + }, + While { + condition: Box, + body: Box, + }, + For { + variable: Box, + condition: Box, + increment: Box, + body: Box, }, } @@ -74,8 +96,10 @@ impl std::fmt::Display for Expr { right, } => write!(f, "Binary ({} {} {})", left.node, operator, right.node), Expr::Unary { operator, operand } => write!(f, "Unary ({} {})", operator, operand.node), - Expr::Grouping { expression } => write!(f, "Grouping (group {})", expression.node), - Expr::Identifier { name } => write!(f, "Identifier (variable {})", name), + Expr::Grouping { expression } => write!(f, "Grouping (group {})", expression.node,), + Expr::Identifier { name } => { + write!(f, "Identifier (variable {})", name) + } Expr::Call { callee, arguments } => write!( f, "Call ({}({}))", @@ -86,94 +110,18 @@ impl std::fmt::Display for Expr { .collect::>() .join(", ") ), - } - } -} - -impl Debug for Expr { - fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result { - match self { - Expr::Literal { value } => write!(f, "{:?}", value), - Expr::Binary { - left, - operator, - right, - } => write!(f, "({:?} {:?} {:?})", operator, left.node, right.node), - Expr::Unary { operator, operand } => write!(f, "({:?} {:?})", operator, operand.node), - Expr::Grouping { expression } => write!(f, "(group {:?})", expression.node), - Expr::Identifier { name } => write!(f, "(variable {:?})", name), - Expr::Call { callee, arguments } => write!( - f, - "Call ({}({}))", - callee.node, - arguments - .iter() - .map(|arg| arg.node.to_string()) - .collect::>() - .join(", ") - ), - } - } -} - -#[derive(Clone, PartialEq)] -pub enum Stmt { - Expression { - expression: Box>, - }, - Print { - expression: Box>, - }, - Stmt { - expression: Box>, - }, - VarDeclaration { - name: String, - initializer: Option>>, - }, - VarAssigment { - name: String, - value: Box>, - }, - Return { - expression: Box>, - }, - Block { - statements: Box>>, - }, - If { - condition: Box>, - then_branch: Box>, - elif_branch: Vec<(Box>, Box>)>, - else_branch: Option>>, - }, - While { - condition: Box>, - body: Box>, - }, - For { - variable: Box>, - condition: Box>, - increment: Box>, - body: Box>, - }, -} - -impl Display for Stmt { - fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result { - match self { - Stmt::Expression { expression } => write!(f, "Expression ({})", expression.node), - Stmt::If { + Expr::Assign { name, value } => write!(f, "Assign ({} = {})", name, value.node), + Expr::If { condition, then_branch, - elif_branch, + elif_branches, else_branch, } => { let mut result = format!("IF ({}) {{\n{}\n}}", condition.node, then_branch.node); - for (condition, branch) in elif_branch { + for (condition, branch) in elif_branches { result.push_str(&format!( " ELIF ({}) {{\n{}\n}}", - condition.node, branch.node + condition.node, branch.node, )); } if let Some(else_branch) = else_branch { @@ -181,27 +129,32 @@ impl Display for Stmt { } write!(f, "IfStmt {}", result) } - Stmt::Print { expression } => write!(f, "Print({});", expression.node), - Stmt::Stmt { expression } => write!(f, "Stmt({});", expression.node), - Stmt::VarDeclaration { name, initializer } => match initializer { - Some(init) => write!(f, "Var({} = {});", name, init.node), - None => write!(f, "Var({});", name), + Expr::Print { expression } => write!(f, "Print({})", expression.node), + Expr::VarDeclaration { + name, + initializer, + var_type, + } => match initializer { + Some(init) => write!(f, "Var({} : {:?} = {})", name, var_type, init.node), + None => write!(f, "Var({} : {:?})", name, var_type), }, - Stmt::VarAssigment { name, value } => write!(f, "Assign({} = {});", name, value.node), - Stmt::Return { expression } => write!(f, "Return({});", expression.node), - Stmt::Block { statements } => write!( + Expr::Return { expression, label } => { + write!(f, "Return({}, {})", expression.node, label) + } + Expr::Block { statements, label } => write!( f, - "Block([\n{}\n])", + "Block [{}] ([\n{}\n])", + label, statements .iter() .map(|stmt| format!("\t \t{}", stmt.node)) .collect::>() - .join("\n") + .join("\n"), ), - Stmt::While { condition, body } => { + Expr::While { condition, body } => { write!(f, "While({}) {{\n{}\n}}", condition.node, body.node) } - Stmt::For { + Expr::For { variable, condition, increment, @@ -215,110 +168,116 @@ impl Display for Stmt { } } -impl Debug for Stmt { +impl Debug for Expr { fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result { match self { - Stmt::Expression { expression } => write!(f, " Expression ({:?})", expression.node), - Stmt::If { + Expr::Literal { value } => write!(f, "{:?} ", value,), + Expr::Binary { + left, + operator, + right, + } => write!(f, "({:?} {:?} {:?}) ", operator, left.node, right.node,), + Expr::Unary { operator, operand } => { + write!(f, "(Unary {:?} {:?}) ", operator, operand.node,) + } + Expr::Grouping { expression } => { + write!(f, "(Grouping {:?})", expression.node,) + } + Expr::Identifier { name } => { + write!(f, "(variable {:?})", name,) + } + Expr::Call { callee, arguments } => write!( + f, + "Call ({}({}))", + callee.node, + arguments + .iter() + .map(|arg| arg.node.to_string()) + .collect::>() + .join(", "), + ), + Expr::Assign { name, value } => { + write!(f, "(assign {:?} {:?})", name, value.node,) + } + Expr::If { condition, then_branch, - elif_branch, + elif_branches, else_branch, } => { let mut result = - format!("IF ({:?}) {{\n{:?}\n}}", condition.node, then_branch.node); - for (condition, branch) in elif_branch { + format!("IF ({:?}) {{\n{:?}\n}}", condition.node, then_branch.node,); + for (condition, branch) in elif_branches { result.push_str(&format!( " ELIF ({:?}) {{\n{:?}\n}}", - condition.node, branch.node + condition.node, branch.node, )); } if let Some(else_branch) = else_branch { - result.push_str(&format!(" ELSE {{\n{:?}\n}}", else_branch.node)); + result.push_str(&format!(" ELSE {{\n{:?}\n}}", else_branch.node,)); } write!(f, "IfStmt {:?}", result) } - Stmt::Print { expression } => write!(f, "Print({:?});", expression.node), - Stmt::Stmt { expression } => write!(f, "Stmt({:?});", expression.node), - Stmt::VarDeclaration { name, initializer } => match initializer { - Some(init) => write!(f, "Var({:?} = {:?});", name, init.node), - None => write!(f, "Var({:?});", name), - }, - Stmt::VarAssigment { name, value } => { - write!(f, "Assign({:?} = {:?});", name, value.node) + Expr::Print { expression } => { + write!(f, "Print({:?});", expression.node,) } - Stmt::Return { expression } => write!(f, "Return({:?});", expression.node), - Stmt::Block { statements } => write!( + Expr::VarDeclaration { + name, + initializer, + var_type, + } => match initializer { + Some(init) => write!(f, "Var({:?}: {:?} = {:?});", name, var_type, init.node,), + None => write!(f, "Var({:?}: {:?});", name, var_type,), + }, + Expr::Return { expression, label } => { + write!(f, "Return({:?}, {}) ;", expression.node, label,) + } + Expr::Block { label, statements } => write!( f, - "Block([\n{:?}\n])", + "Block [{}] ([\n{:?}\n])", + label, statements .iter() .map(|stmt| format!("{:?}", stmt.node)) .collect::>() - .join("\t\t\n") + .join("\t\t\n"), ), - Stmt::While { condition, body } => { - write!(f, "While({:?}) {{\n{:?}\n}}", condition.node, body.node) + Expr::While { condition, body } => { + write!(f, "While({:?}) {{\n{:?}\n}}", condition.node, body.node,) } - Stmt::For { + Expr::For { variable, condition, increment, body, } => write!( f, - "For({:?} = {:?} in {:?}) {{\n{:?}\n}}", - variable.node, condition.node, increment.node, body.node + "For({:?} = {:?} in {:?}) {{\n{:?}\n}} ", + variable.node, condition.node, increment.node, body.node, ), } } } -pub trait AstNodeKind { - fn kind(&self) -> &'static str; -} - -impl AstNodeKind for Expr { - fn kind(&self) -> &'static str { - match self { - Expr::Literal { value: _ } => "literal", - Expr::Binary { - left: _, - operator: _, - right: _, - } => "binary expression", - Expr::Unary { .. } => "unary expression", - Expr::Grouping { .. } => "grouping expression", - Expr::Identifier { .. } => "variable expression", - Expr::Call { .. } => "call expression", - } - } -} - -impl AstNodeKind for Stmt { - fn kind(&self) -> &'static str { - match self { - Stmt::Expression { .. } => "expression", - Stmt::VarDeclaration { .. } => "variable declaration", - Stmt::VarAssigment { .. } => "assignment", - Stmt::Return { .. } => "return statement", - Stmt::Block { .. } => "block statement", - Stmt::If { .. } => "if statement", - Stmt::Print { .. } => "print statement", - Stmt::Stmt { .. } => "statement", - Stmt::While { .. } => "while statement", - Stmt::For { .. } => "for statement", - } - } -} - -#[derive(Clone, PartialEq, Default)] -pub struct AstNode { - pub node: T, +#[derive(Clone)] +pub struct AstNode { + /// Stable identity, assigned at construction and preserved across clones. + pub id: NodeId, + pub node: Expr, + pub is_statement: bool, pub source_slice: SourceSlice, + pub return_type: Type, } -impl Display for AstNode { +// Identity (`id`) deliberately does not participate in equality: two nodes are +// equal when their content and location match, regardless of node id. +impl PartialEq for AstNode { + fn eq(&self, other: &Self) -> bool { + self.node == other.node && self.source_slice == other.source_slice + } +} + +impl Display for AstNode { fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result { write!( f, @@ -328,7 +287,7 @@ impl Display for AstNode { } } -impl Debug for AstNode { +impl Debug for AstNode { fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result { write!( f, @@ -338,8 +297,22 @@ impl Debug for AstNode { } } -impl AstNode { - pub fn new(node: T, source_slice: SourceSlice) -> Self { - AstNode { node, source_slice } +impl AstNode { + fn new(node: Expr, source_slice: SourceSlice, type_name: String, is_statement: bool) -> Self { + AstNode { + id: NodeId::next(), + node, + is_statement, + source_slice, + return_type: Type::Unresolved(type_name), + } + } + + pub fn new_statement(node: Expr, source_slice: SourceSlice) -> Self { + Self::new(node, source_slice, "Nil".to_string(), true) + } + + pub fn new_expression(node: Expr, source_slice: SourceSlice, type_name: String) -> Self { + Self::new(node, source_slice, type_name, false) } } diff --git a/src/common/base_value.rs b/src/common/base_value.rs old mode 100644 new mode 100755 index 9ae8024..9d1862d --- a/src/common/base_value.rs +++ b/src/common/base_value.rs @@ -1,11 +1,12 @@ use core::fmt; +use std::collections::HashMap; use std::fmt::Display; use std::ops::{Add, Div, Mul, Not, Rem, Sub}; use crate::common::lox_result::{runtime_error, LoxResult}; +use crate::common::types::Type; use crate::{ - backend::environment::EnvironmentStack, - common::ast::{AstNode, Expr, Stmt}, + backend::environment::EnvironmentStack, common::ast::AstNode, frontend::source_registry::SourceSlice, }; #[derive(Debug, Clone, PartialEq)] @@ -271,6 +272,22 @@ pub enum BaseValue { Nil, Function(LoxFunction), NativeFunction(NativeFunction), + Struct(Struct), +} + +struct Dict { + map: HashMap, +} + +struct ReturnValue { + label: Vec, + value: Type, +} + +impl ReturnValue { + pub fn new(label: Vec, value: Type) -> Self { + Self { label, value } + } } impl BaseValue { @@ -292,36 +309,40 @@ impl Display for BaseValue { BaseValue::Nil => write!(f, "nil"), BaseValue::Function(..) => write!(f, ""), BaseValue::NativeFunction(..) => write!(f, ""), + BaseValue::Struct(_) => write!(f, ""), } } } #[derive(Debug, Clone, PartialEq)] pub struct LoxFunction { - pub parameters: Vec<(String, String)>, - pub body: AstNode, - pub closure: Option, - pub guard: Option>, + pub parameters: Vec<(String, Type)>, + pub return_type: Type, + pub body: AstNode, + pub closure: Option>, + pub guard: Option>, } impl LoxFunction { pub fn new( - parameters: Vec<(String, String)>, - body: AstNode, - closure: Option, - guard: Option>, + parameters: Vec<(String, Type)>, + body: AstNode, + closure: Option>, + guard: Option>, ) -> Self { LoxFunction { parameters, + return_type: Type::Any, body, closure, guard, } } - pub fn anonymous_function(parameters: Vec<(String, String)>, body: AstNode) -> Self { + pub fn anonymous_function(parameters: Vec<(String, Type)>, body: AstNode) -> Self { LoxFunction { parameters, + return_type: Type::Any, body, closure: None, guard: None, @@ -331,12 +352,12 @@ impl LoxFunction { #[derive(Debug, Clone, PartialEq)] pub struct NativeFunction { - pub parameters: Vec<(String, String)>, + pub parameters: Vec<(String, Type)>, pub function: fn(&[BaseValue]) -> BaseValue, } impl NativeFunction { - pub fn new(parameters: Vec<(String, String)>, function: fn(&[BaseValue]) -> BaseValue) -> Self { + pub fn new(parameters: Vec<(String, Type)>, function: fn(&[BaseValue]) -> BaseValue) -> Self { NativeFunction { parameters, function, @@ -344,6 +365,11 @@ impl NativeFunction { } } +#[derive(Debug, Clone, PartialEq)] +pub struct Struct { + pub fields: Vec<(String, BaseValue)>, +} + // Trait implementations for BaseValue pub trait Truthy { @@ -383,7 +409,7 @@ impl Add for BaseValue { Ok(BaseValue::String(format!("{}{}", a, b))) } _ => runtime_error( - SourceSlice::default(), + SourceSlice::synthetic(), "Cannot add non-numeric values".to_string(), ), } @@ -413,7 +439,7 @@ impl Sub for BaseValue { match (self, other) { (BaseValue::Number(a), BaseValue::Number(b)) => Ok(BaseValue::Number(a.sub(b))), _ => runtime_error( - SourceSlice::default(), + SourceSlice::synthetic(), "Cannot subtract non-numeric values".to_string(), ), } @@ -427,10 +453,10 @@ impl Div for BaseValue { match (self, other) { (BaseValue::Number(a), BaseValue::Number(b)) => match a.div(b) { Some(result) => Ok(BaseValue::Number(result)), - None => runtime_error(SourceSlice::default(), "Division by zero".to_string()), + None => runtime_error(SourceSlice::synthetic(), "Division by zero".to_string()), }, _ => runtime_error( - SourceSlice::default(), + SourceSlice::synthetic(), "Cannot divide non-numeric values".to_string(), ), } @@ -444,7 +470,7 @@ impl Mul for BaseValue { match (self, other) { (BaseValue::Number(a), BaseValue::Number(b)) => Ok(BaseValue::Number(a.mul(b))), _ => runtime_error( - SourceSlice::default(), + SourceSlice::synthetic(), "Cannot multiply non-numeric values".to_string(), ), } @@ -458,10 +484,10 @@ impl Rem for BaseValue { match (self, other) { (BaseValue::Number(a), BaseValue::Number(b)) => match a.rem(b) { Some(result) => Ok(BaseValue::Number(result)), - None => runtime_error(SourceSlice::default(), "Division by zero".to_string()), + None => runtime_error(SourceSlice::synthetic(), "Division by zero".to_string()), }, _ => runtime_error( - SourceSlice::default(), + SourceSlice::synthetic(), "Cannot divide non-numeric values".to_string(), ), } diff --git a/src/common/lox_result.rs b/src/common/lox_result.rs old mode 100644 new mode 100755 index 9f0ec6f..602d7c3 --- a/src/common/lox_result.rs +++ b/src/common/lox_result.rs @@ -1,5 +1,8 @@ -use crate::frontend::source_registry::{SourceRegistry, SourceSlice}; -use std::fmt; +use crate::{ + common::base_value::BaseValue, + frontend::source_registry::{SourceRegistry, SourceSlice}, +}; +use std::{error::Error, fmt}; #[derive(Debug, Clone)] pub enum LoxError { @@ -23,6 +26,11 @@ pub enum LoxError { expected: String, found: String, }, + Return { + source_slice: SourceSlice, + value: BaseValue, + return_label: String, + }, } /// Configuration for error display formatting @@ -81,6 +89,13 @@ impl LoxError { } => { format!("expected {}, found {}", expected, found) } + LoxError::Return { + value, + return_label, + .. + } => { + format!("return value: {} and label: {}", value, return_label) + } } } @@ -91,6 +106,7 @@ impl LoxError { LoxError::ParseError { source_slice, .. } => Some(source_slice.clone()), LoxError::IoError { .. } => None, LoxError::TypeMismatch { source_slice, .. } => Some(source_slice.clone()), + LoxError::Return { .. } => None, } } @@ -101,6 +117,7 @@ impl LoxError { LoxError::ParseError { .. } => "parse error", LoxError::IoError { .. } => "io error", LoxError::TypeMismatch { .. } => "type error", + LoxError::Return { .. } => "return error", } } @@ -393,11 +410,25 @@ impl fmt::Display for LoxError { found ) } + LoxError::Return { + value, + return_label, + source_slice, + } => { + write!( + f, + "Return error at {}:{}: value `{}`, label `{}`", + source_slice.start_position.line + 1, + source_slice.start_position.column + 1, + value, + return_label + ) + } } } } -impl std::error::Error for LoxError {} +impl Error for LoxError {} pub type LoxResult = Result; @@ -448,3 +479,90 @@ pub fn io_error(message: impl Into) -> LoxResult { message: message.into(), }) } + +/// Accumulates diagnostics during a pass instead of bailing on the first one. +/// +/// Static analyses (e.g. the resolver) `report` problems and keep traversing so +/// the user sees every error at once, then surface them at the end. +#[derive(Debug, Default)] +pub struct ErrorSink { + errors: Vec, +} + +impl ErrorSink { + pub fn new() -> Self { + Self::default() + } + + /// Record a diagnostic and keep going. + pub fn report(&mut self, error: LoxError) { + self.errors.push(error); + } + + pub fn has_errors(&self) -> bool { + !self.errors.is_empty() + } + + pub fn errors(&self) -> &[LoxError] { + &self.errors + } + + /// Merge another sink's errors into this one (e.g. from a separate pass). + pub fn extend(&mut self, other: ErrorSink) { + self.errors.extend(other.errors); + } + + pub fn into_errors(self) -> Vec { + self.errors + } + + /// Bridge to the single-error [`LoxResult`] API: `Ok` if empty, otherwise + /// the first reported error. + pub fn into_result(self) -> LoxResult<()> { + match self.errors.into_iter().next() { + Some(error) => Err(error), + None => Ok(()), + } + } +} + +#[cfg(test)] +mod tests { + use super::*; + use crate::frontend::source_registry::SourceSlice; + + fn parse_err(message: &str) -> LoxError { + LoxError::ParseError { + source_slice: SourceSlice::synthetic(), + message: message.to_string(), + } + } + + #[test] + fn empty_sink_is_ok() { + let sink = ErrorSink::new(); + assert!(!sink.has_errors()); + assert!(sink.into_result().is_ok()); + } + + #[test] + fn accumulates_multiple_errors() { + let mut sink = ErrorSink::new(); + sink.report(parse_err("first")); + sink.report(parse_err("second")); + assert!(sink.has_errors()); + assert_eq!(sink.errors().len(), 2); + assert!(sink.into_result().is_err()); + } + + #[test] + fn extend_merges_sinks() { + let mut a = ErrorSink::new(); + a.report(parse_err("a")); + let mut b = ErrorSink::new(); + b.report(parse_err("b1")); + b.report(parse_err("b2")); + a.extend(b); + assert_eq!(a.into_errors().len(), 3); + } +} diff --git a/src/common/mod.rs b/src/common/mod.rs old mode 100644 new mode 100755 index 736f74d..ccbfcd9 --- a/src/common/mod.rs +++ b/src/common/mod.rs @@ -1,3 +1,4 @@ pub mod ast; pub mod base_value; pub mod lox_result; +pub mod types; diff --git a/src/common/types.rs b/src/common/types.rs new file mode 100644 index 0000000..d5a3dd7 --- /dev/null +++ b/src/common/types.rs @@ -0,0 +1,70 @@ +use std::{fmt::Display, rc::Rc}; + +#[derive(Debug, Clone, PartialEq)] +pub enum Type { + Any, + Number, + String, + Boolean, + Nil, + Function(FunctionType), + Struct(Rc), + Unresolved(String), +} + +impl Display for Type { + fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result { + match self { + Type::Any => write!(f, "any"), + Type::Number => write!(f, "number"), + Type::String => write!(f, "string"), + Type::Boolean => write!(f, "boolean"), + Type::Nil => write!(f, "nil"), + Type::Function(fun) => write!(f, "{}", fun), + Type::Struct(struct_) => write!(f, "{}", struct_), + Type::Unresolved(name) => write!(f, "{}", name), + } + } +} + +#[derive(Debug, Clone, PartialEq)] +pub struct FunctionType { + pub params: Vec<(String, Box)>, + pub return_type: Box, +} + +impl Display for FunctionType { + fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result { + let rendered_params = self + .params + .iter() + .map(|(name, ty)| format!("{}: {}", name, ty)) + .collect::>() + .join(", "); + write!(f, "fn({}) -> {}", rendered_params, self.return_type) + } +} + +#[derive(Debug, Clone, PartialEq)] +pub struct StructType { + pub fields: Vec<(String, Box)>, + pub methods: Vec<(String, FunctionType)>, +} + +impl Display for StructType { + fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result { + let rendered_field = self + .fields + .iter() + .map(|(name, ty)| format!("{}: {}", name, ty)) + .collect::>() + .join(", "); + let rendered_method = self + .methods + .iter() + .map(|(name, ty)| format!("{}: {}", name, ty)) + .collect::>() + .join(", "); + write!(f, "struct {{ {}, {} }}", rendered_field, rendered_method) + } +} diff --git a/src/frontend/lexer.rs b/src/frontend/lexer.rs old mode 100644 new mode 100755 index b34d5ec..5a61faf --- a/src/frontend/lexer.rs +++ b/src/frontend/lexer.rs @@ -5,8 +5,9 @@ use crate::frontend::tokens::{Token, TokenType}; pub struct Lexer { input: String, - start_char: usize, - current_char: usize, + // Byte offsets into `input` (not char counts) so slicing is UTF-8 correct. + start: usize, + current: usize, start_pos: SourcePosition, end_pos: SourcePosition, source_id: SourceId, @@ -15,7 +16,7 @@ pub struct Lexer { fn get_keyword_token(word: &str) -> Option { match word { "and" => Some(TokenType::And), - "class" => Some(TokenType::Class), + "struct" => Some(TokenType::Struct), "do" => Some(TokenType::StartBlock), "end" => Some(TokenType::EndBlock), "false" => Some(TokenType::False), @@ -29,6 +30,7 @@ fn get_keyword_token(word: &str) -> Option { "in" => Some(TokenType::In), "print" => Some(TokenType::Print), "return" => Some(TokenType::Return), + "break" => Some(TokenType::Break), "super" => Some(TokenType::Super), "this" => Some(TokenType::This), "true" => Some(TokenType::True), @@ -47,28 +49,18 @@ impl Lexer { pub fn new(input: String, source_id: SourceId) -> Lexer { Lexer { input, - start_char: 0, - current_char: 0, + start: 0, + current: 0, start_pos: SourcePosition::default(), end_pos: SourcePosition::default(), source_id, } } - fn advance_column(&mut self) { - self.current_char += 1; - self.end_pos.column += 1; - } - - fn advance_line(&mut self) { - self.end_pos.line += 1; - self.end_pos.column = 0; - } - pub fn scans_tokens(&mut self) -> LoxResult> { let mut tokens = Vec::new(); while !self.is_at_end() { - self.start_char = self.current_char; + self.start = self.current; self.start_pos = self.end_pos.clone(); match self.scan_token() { Ok(Some(token)) => tokens.push(token), @@ -81,32 +73,31 @@ impl Lexer { } fn is_at_end(&self) -> bool { - self.current_char >= self.input.len() + self.current >= self.input.len() } fn advance(&mut self) -> char { - self.advance_column(); - self.input.chars().nth(self.current_char - 1).unwrap() + let c = self.input[self.current..].chars().next().unwrap(); + self.current += c.len_utf8(); + if c == '\n' { + self.end_pos.line += 1; + self.end_pos.column = 0; + } else { + self.end_pos.column += 1; + } + c } fn peek(&self) -> char { - if self.is_at_end() { - '\0' - } else { - self.input.chars().nth(self.current_char).unwrap() - } + self.input[self.current..].chars().next().unwrap_or('\0') } fn peek_next(&self) -> char { - if self.current_char + 1 >= self.input.len() { - '\0' - } else { - self.input.chars().nth(self.current_char + 1).unwrap() - } + self.input[self.current..].chars().nth(1).unwrap_or('\0') } fn make_token(&self, token_type: TokenType) -> Token { - let text = self.input[self.start_char..self.current_char].to_string(); + let text = self.input[self.start..self.current].to_string(); Token::new( token_type, text, @@ -118,7 +109,7 @@ impl Lexer { ) } fn make_token_with_literal(&self, token_type: TokenType, literal: BaseValue) -> Token { - let text = self.input[self.start_char..self.current_char].to_string(); + let text = self.input[self.start..self.current].to_string(); Token::new_complete( token_type, text, @@ -178,9 +169,6 @@ impl Lexer { ('/', '*') => { // Commento multi-line while (self.peek() != '*' || self.peek_next() != '/') && !self.is_at_end() { - if self.peek() == '\n' { - self.advance_line(); - } self.advance(); } if self.is_at_end() { @@ -200,10 +188,7 @@ impl Lexer { } ('/', _) => Ok(Some(self.make_token(TokenType::Slash))), (' ', _) | ('\r', _) | ('\t', _) => Ok(None), - ('\n', _) => { - self.advance_line(); - Ok(None) - } + ('\n', _) => Ok(None), ('"', _) => self.string(), (c, _) if c.is_digit(10) => self.number(), (c, _) if c.is_alphanumeric() || c == '_' => self.identifier(), @@ -235,7 +220,7 @@ impl Lexer { self.advance(); Ok(Some(self.make_token_with_literal( TokenType::String, - BaseValue::String(self.input[self.start_char..self.current_char].to_string()), + BaseValue::String(self.input[self.start..self.current].to_string()), ))) } @@ -265,7 +250,7 @@ impl Lexer { None }; - let num_str = &self.input[self.start_char..self.current_char]; + let num_str = &self.input[self.start..self.current]; let num_str_without_suffix = if suffix.is_some() { &num_str[..num_str.len() - 1] } else { @@ -302,7 +287,7 @@ impl Lexer { while self.peek().is_alphanumeric() || self.peek() == '_' { self.advance(); } - let text = self.input[self.start_char..self.current_char].to_string(); + let text = self.input[self.start..self.current].to_string(); match get_keyword_token(&text) { Some(TokenType::True) => Ok(Some( self.make_token_with_literal(TokenType::True, BaseValue::Boolean(true)), @@ -321,3 +306,240 @@ impl Lexer { } } } + +#[cfg(test)] +mod tests { + use std::f64; + + use super::*; + + /// Lex `input` and return the resulting tokens, panicking on lexical errors. + fn lex(input: &str) -> Vec { + Lexer::new(input.to_string(), 0) + .scans_tokens() + .expect("expected input to lex without errors") + } + + /// Lex `input` and collect just the token types (including the trailing EOF). + fn token_types(input: &str) -> Vec { + lex(input).into_iter().map(|t| t.token_type).collect() + } + + #[test] + fn scans_single_character_tokens() { + assert_eq!( + token_types("(){}[],.-+;:*%"), + vec![ + TokenType::LeftParen, + TokenType::RightParen, + TokenType::LeftBrace, + TokenType::RightBrace, + TokenType::LeftBracket, + TokenType::RightBracket, + TokenType::Comma, + TokenType::Dot, + TokenType::Minus, + TokenType::Plus, + TokenType::Semicolon, + TokenType::Colon, + TokenType::Star, + TokenType::Percent, + TokenType::Eof, + ] + ); + } + + #[test] + fn scans_one_and_two_char_operators() { + assert_eq!( + token_types("! != = == < <= > >= /"), + vec![ + TokenType::Bang, + TokenType::BangEqual, + TokenType::Equal, + TokenType::EqualEqual, + TokenType::Less, + TokenType::LessEqual, + TokenType::Greater, + TokenType::GreaterEqual, + TokenType::Slash, + TokenType::Eof, + ] + ); + } + + #[test] + fn keyword_lookup_matches_known_words() { + assert_eq!(get_keyword_token("and"), Some(TokenType::And)); + assert_eq!(get_keyword_token("if"), Some(TokenType::If)); + assert_eq!(get_keyword_token("then"), Some(TokenType::Then)); + assert_eq!(get_keyword_token("while"), Some(TokenType::While)); + assert_eq!(get_keyword_token("do"), Some(TokenType::StartBlock)); + assert_eq!(get_keyword_token("end"), Some(TokenType::EndBlock)); + assert_eq!(get_keyword_token("not_a_keyword"), None); + } + + #[test] + fn scans_keywords_in_a_stream() { + assert_eq!( + token_types("if then else while print return"), + vec![ + TokenType::If, + TokenType::Then, + TokenType::Else, + TokenType::While, + TokenType::Print, + TokenType::Return, + TokenType::Eof, + ] + ); + } + + #[test] + fn scans_integer_number() { + let tokens = lex("42"); + assert_eq!(tokens[0].token_type, TokenType::Number); + assert_eq!(tokens[0].literal, Some(BaseValue::Number(Number::I32(42)))); + } + + #[test] + fn scans_float_number() { + let tokens = lex("3.141592653589793"); + assert_eq!( + tokens[0].literal, + Some(BaseValue::Number(Number::F64(f64::consts::PI))) + ); + } + + #[test] + fn scans_number_with_float_suffix() { + let tokens = lex("5f"); + assert_eq!(tokens[0].literal, Some(BaseValue::Number(Number::F64(5.0)))); + } + + #[test] + fn scans_number_with_unsigned_suffix() { + let tokens = lex("7u"); + assert_eq!(tokens[0].literal, Some(BaseValue::Number(Number::U128(7)))); + } + + #[test] + fn scans_string_literal_including_quotes() { + // The lexer slices from the opening quote through the closing quote, + // so the literal currently retains the surrounding quotes. + let tokens = lex("\"hello\""); + assert_eq!(tokens[0].token_type, TokenType::String); + assert_eq!( + tokens[0].literal, + Some(BaseValue::String("\"hello\"".to_string())) + ); + } + + #[test] + fn scans_identifier() { + let tokens = lex("foo_bar"); + assert_eq!(tokens[0].token_type, TokenType::Identifier); + assert_eq!( + tokens[0].literal, + Some(BaseValue::String("foo_bar".to_string())) + ); + } + + #[test] + fn scans_boolean_and_nil_literals() { + let tokens = lex("true false Nil"); + assert_eq!(tokens[0].literal, Some(BaseValue::Boolean(true))); + assert_eq!(tokens[1].literal, Some(BaseValue::Boolean(false))); + assert_eq!(tokens[2].literal, Some(BaseValue::Nil)); + } + + #[test] + fn ignores_line_comments() { + assert_eq!( + token_types("1 // a comment\n2"), + vec![TokenType::Number, TokenType::Number, TokenType::Eof] + ); + } + + #[test] + fn ignores_block_comments() { + assert_eq!( + token_types("1 /* multi\nline */ 2"), + vec![TokenType::Number, TokenType::Number, TokenType::Eof] + ); + } + + #[test] + fn always_appends_eof_even_for_empty_input() { + let tokens = lex(""); + assert_eq!(tokens.len(), 1); + assert_eq!(tokens[0].token_type, TokenType::Eof); + } + + #[test] + fn unterminated_string_is_a_lexical_error() { + let result = Lexer::new("\"oops".to_string(), 0).scans_tokens(); + assert!(matches!(result, Err(LoxError::LexicalError { .. }))); + } + + #[test] + fn unterminated_block_comment_is_a_lexical_error() { + let result = Lexer::new("/* never ends".to_string(), 0).scans_tokens(); + assert!(matches!(result, Err(LoxError::LexicalError { .. }))); + } + + #[test] + fn unexpected_character_is_a_lexical_error() { + let result = Lexer::new("@".to_string(), 0).scans_tokens(); + assert!(result.is_err()); + } + + #[test] + fn handles_multibyte_identifier() { + // `é` is two UTF-8 bytes; the old char-counted slicing would panic or + // slice mid-codepoint here. Byte offsets make this correct. + let tokens = lex("café"); + assert_eq!(tokens[0].token_type, TokenType::Identifier); + assert_eq!( + tokens[0].literal, + Some(BaseValue::String("café".to_string())) + ); + } + + #[test] + fn tracks_line_and_column_across_newlines() { + // "1\n22": the second token sits at the start of line 1. + let tokens = lex("1\n22"); + assert_eq!(tokens[1].lexeme, "22"); + assert_eq!(tokens[1].source_slice.start_position.line, 1); + assert_eq!(tokens[1].source_slice.start_position.column, 0); + } + + #[test] + fn handle_struct() { + let tokens = lex("Cosa :: struct {field1: Number, field2: String}"); + let tokens_tyepe: Vec = tokens.iter().map(|x| x.token_type.clone()).collect(); + let tokent_for_check = vec![ + TokenType::Identifier, + TokenType::Colon, + TokenType::Colon, + TokenType::Struct, + TokenType::LeftBrace, + TokenType::Identifier, + TokenType::Colon, + TokenType::Identifier, + TokenType::Comma, + TokenType::Identifier, + TokenType::Colon, + TokenType::Identifier, + TokenType::RightBrace, + TokenType::Eof, + ]; + assert_eq!(tokens_tyepe, tokent_for_check); + assert_eq!(tokens[0].lexeme, "Cosa"); + assert_eq!(tokens[5].lexeme, "field1"); + assert_eq!(tokens[7].lexeme, "Number"); + assert_eq!(tokens[9].lexeme, "field2"); + assert_eq!(tokens[11].lexeme, "String"); + } +} diff --git a/src/frontend/mod.rs b/src/frontend/mod.rs old mode 100644 new mode 100755 diff --git a/src/frontend/parser.rs b/src/frontend/parser.rs old mode 100644 new mode 100755 index bb705f8..d18dea3 --- a/src/frontend/parser.rs +++ b/src/frontend/parser.rs @@ -1,8 +1,9 @@ use crate::{ common::{ - ast::{AstNode, Expr, Stmt}, + ast::{AstNode, Expr}, base_value::{BaseValue, LoxFunction}, lox_result::{parse_error, runtime_error, LoxResult}, + types::Type, }, frontend::{ source_registry::SourceSlice, @@ -20,7 +21,7 @@ impl Parser { Self { tokens, current: 0 } } - pub fn parse(&mut self) -> LoxResult>> { + pub fn parse(&mut self) -> LoxResult> { let mut statements = Vec::new(); let mut errors = Vec::new(); while !self.is_at_end() { @@ -42,17 +43,28 @@ impl Parser { Ok(statements) } - fn statement(&mut self) -> LoxResult> { + fn statement(&mut self) -> LoxResult { match (&self.peek().token_type, &self.peek_next().token_type) { (TokenType::Print, _) => self.print_statement(), - (TokenType::Return, _) => self.return_statement(), + (TokenType::Return, _) => { + let mut label = None; + if self.peek_next().token_type == TokenType::Colon { + self.advance(); + label = Some( + self.consume(TokenType::Identifier, "Expect label after return colon.")? + .lexeme + .clone(), + ); + } + self.return_statement(label) + } + (TokenType::Break, _) => self.return_statement(Some("loop".to_string())), (TokenType::StartBlock, _) => self.block_statement(), (TokenType::Var, TokenType::Identifier) => { self.advance(); self.var_statement() } (TokenType::Identifier, TokenType::Colon) => self.var_statement(), - (TokenType::Identifier, TokenType::Equal) => self.assignment_statement(), (TokenType::If, _) => self.if_statement(), (TokenType::While, _) => self.while_statement(), (TokenType::For, _) => self.for_statement(), @@ -60,19 +72,15 @@ impl Parser { } } - fn for_statement(&mut self) -> LoxResult> { + fn for_statement(&mut self) -> LoxResult { let start_slice = self.peek().source_slice.clone(); self.advance(); // consume 'for' let variable = self.var_statement()?; let condition = self.expression()?; self.consume(TokenType::Semicolon, "Expected ';' after for condition")?; - let increment = self.assignment_statement()?; - - self.consume(TokenType::StartBlock, "Expected 'do' after for range")?; - + let increment = self.expression_statement()?; let body = self.statement()?; - self.consume(TokenType::EndBlock, "Expected 'end' after for body")?; let end_slice = self.previous().source_slice.clone(); let combined_slice = SourceSlice::from_positions( @@ -81,8 +89,8 @@ impl Parser { end_slice.end_position, ); - Ok(AstNode::new( - Stmt::For { + Ok(AstNode::new_statement( + Expr::For { variable: Box::new(variable), condition: Box::new(condition), increment: Box::new(increment), @@ -92,49 +100,7 @@ impl Parser { )) } - // fn for_statement(&mut self) -> LoxResult> { - // let start_slice = self.peek().source_slice.clone(); - // self.advance(); - // let variable = if let Token { - // token_type: TokenType::Identifier, - // literal: Some(variable_name), - // source_slice, - // .. - // } = self.peek() - // { - // AstNode { - // node: Expr::Literal { - // value: variable_name.clone(), - // }, - // source_slice: source_slice.clone(), - // } - // } else { - // return parse_error( - // self.peek().source_slice.clone(), - // "Expect variable name after 'for' keyword.", - // ); - // }; - // self.advance(); - // self.consume(TokenType::In, "Expect 'in' after for variable.")?; - // let iterable = self.expression()?; - // let body = self.statement()?; - // let end_slice = body.source_slice.clone(); - // let combined_slice = SourceSlice::from_positions( - // start_slice.source_id, - // start_slice.start_position, - // end_slice.end_position, - // ); - // Ok(AstNode::new( - // Stmt::For { - // variable: Box::new(variable), - // iterable: Box::new(iterable), - // body: Box::new(body), - // }, - // combined_slice, - // )) - // } - - fn while_statement(&mut self) -> LoxResult> { + fn while_statement(&mut self) -> LoxResult { let start_slice = self.peek().source_slice.clone(); self.advance(); let condition = self.expression()?; @@ -145,8 +111,8 @@ impl Parser { start_slice.start_position, end_slice.end_position, ); - Ok(AstNode::new( - Stmt::While { + Ok(AstNode::new_statement( + Expr::While { condition: Box::new(condition), body: Box::new(body), }, @@ -154,7 +120,7 @@ impl Parser { )) } - fn if_statement(&mut self) -> LoxResult> { + fn if_statement(&mut self) -> LoxResult { let start_slice = self.peek().source_slice.clone(); self.advance(); // consume 'if' let condition = self.expression()?; @@ -187,54 +153,97 @@ impl Parser { last_slice.end_position, ); - Ok(AstNode::new( - Stmt::If { + Ok(AstNode::new_statement( + Expr::If { condition: Box::new(condition), then_branch: Box::new(then_branch), - elif_branch: elif_branches, - else_branch: else_branch, + elif_branches, + else_branch, }, combined_slice, )) } - fn var_statement(&mut self) -> LoxResult> { + fn var_statement(&mut self) -> LoxResult { let start_slice = self.peek().source_slice.clone(); - let name = self.consume(TokenType::Identifier, "Expect variable name.")?; - let name_lexeme = name.lexeme.clone(); + let name = self + .consume(TokenType::Identifier, "Expect variable name.")? + .lexeme + .clone(); + self.consume(TokenType::Colon, "Expected 'colon' after var identifier")?; + let start_slice_clone = start_slice.clone(); - let mut _type_annotation = BaseValue::Nil; - self.consume(TokenType::Colon, "Expect column")?; - - match self.peek().token_type { - TokenType::Equal | TokenType::Identifier => { - self.variable_declaration(start_slice, name_lexeme) + let mut var_type = Type::Any; + let var_content = match self.peek().token_type { + TokenType::Equal | TokenType::Identifier | TokenType::Nil => { + if self.peek_is_type_name() { + let tyepe_annotation = self.consume_type_name("Expected Identier")?; + var_type = Type::Unresolved(tyepe_annotation); + } + let mut value: Option> = None; + if self.peek().token_type == TokenType::Equal { + self.advance(); + value = Some(Box::new(self.expression()?)); + } + value } - TokenType::Colon => self.function_declaration(start_slice, name_lexeme), + + TokenType::Colon => { + self.consume(TokenType::Colon, "Expected 'colon' after var identifier")?; + var_type = Type::Unresolved("Function".to_string()); + let res = self.struct_or_function_declaration(start_slice)?; + Some(Box::new(res)) + } + _ => runtime_error(start_slice, "this is not supposed to be here")?, + }; + + let semicolon = self.consume( + TokenType::Semicolon, + "Expected 'semicolon' after var declaration", + )?; + + let slice = SourceSlice::from_source_slices( + start_slice_clone.clone(), + semicolon.source_slice.clone(), + ); + let variable = Expr::VarDeclaration { + name, + var_type, + initializer: var_content, + }; + + Ok(AstNode::new_statement(variable, slice)) + } + + fn struct_or_function_declaration(&mut self, start_slice: SourceSlice) -> LoxResult { + match self.peek().token_type { + TokenType::Fn => self.function_declaration(start_slice), + TokenType::Struct => self.struct_declaration(start_slice), _ => runtime_error(start_slice, "this is not supposed to be here"), } } - fn function_declaration( - &mut self, - start_slice: SourceSlice, - name_lexeme: String, - ) -> LoxResult> { - self.advance(); + fn struct_declaration(&mut self, start_slice: SourceSlice) -> LoxResult { + self.consume(TokenType::Struct, "Expect 'struct' after '::'")?; + self.consume(TokenType::LeftBrace, "Expect '{' after 'struct'")?; + + todo!() + } + + fn function_declaration(&mut self, start_slice: SourceSlice) -> LoxResult { self.consume(TokenType::Fn, "Expect 'fn' after '::'")?; self.consume(TokenType::LeftParen, "Expect '(' after 'fn' ")?; - let mut parameters: Vec<(String, String)> = vec![]; + let mut parameters: Vec<(String, Type)> = vec![]; while self.peek().token_type != TokenType::RightParen { - let expr = self + let parameter_name = self .consume(TokenType::Identifier, "Expect identifier after '('")? .clone(); if self.peek().token_type == TokenType::Colon { self.advance(); - let type_ = - self.consume(TokenType::Identifier, "Expected identifier after ':' ")?; - parameters.push((expr.lexeme, type_.lexeme.clone())); + let paramete_type = self.consume_type_name("Expected identifier after ':' ")?; + parameters.push((parameter_name.lexeme, Type::Unresolved(paramete_type))); } else { - parameters.push((expr.lexeme, "Any".to_string())); + parameters.push((parameter_name.lexeme, Type::Any)); } if self.peek().token_type == TokenType::Comma { self.advance(); @@ -248,101 +257,32 @@ impl Parser { end_position: end_position.end_position, }; - let mut guard: Option> = None; + let mut guard: Option> = None; if self.peek().token_type == TokenType::LeftBrace { self.consume(TokenType::LeftBrace, "Expected '{' after guard expression")?; - println!("ho trovato una guradia"); - guard = Some(Box::new(self.expression()?.node.clone())); + guard = Some(Box::new(self.expression()?)); self.consume(TokenType::RightBrace, "Expected '}' after guard expression")?; } + self.consume(TokenType::Colon, "Expected ':' after parameters")?; + let return_type = self.consume_type_name("Expected Type identifier after ':'")?; let body = self.statement()?; - let node = AstNode { - node: Expr::Literal { - value: BaseValue::Function(LoxFunction { + Ok(AstNode::new_expression( + Expr::Literal { + value: Box::new(BaseValue::Function(LoxFunction { parameters, + return_type: Type::Unresolved(return_type.clone()), body, closure: None, guard, - }), - }, - source_slice: combine_position.clone(), - }; - Ok(AstNode::new( - Stmt::VarDeclaration { - name: name_lexeme, - initializer: Some(Box::new(node)), + })), }, combine_position.clone(), + return_type.clone(), )) } - fn variable_declaration( - &mut self, - start_slice: SourceSlice, - name_lexeme: String, - ) -> LoxResult> { - if self.peek().token_type == TokenType::Identifier { - self.advance(); - if self.peek().token_type == TokenType::Identifier { - self.advance(); - } - // todo: make type annotation - } - let mut value = AstNode { - node: Expr::Literal { - value: BaseValue::Nil, - }, - source_slice: start_slice.clone(), - }; - if self.peek().token_type == TokenType::Equal { - self.advance(); - value = self.expression()?; - } - let semicolon = self.consume( - TokenType::Semicolon, - "Expect ';' after variable declaration.", - )?; - let end_slice = semicolon.source_slice.clone(); - let combined_slice = SourceSlice::from_positions( - start_slice.source_id, - start_slice.start_position, - end_slice.end_position, - ); - Ok(AstNode::new( - Stmt::VarDeclaration { - name: name_lexeme, - initializer: Some(Box::new(value)), - }, - combined_slice, - )) - } - fn assignment_statement(&mut self) -> LoxResult> { - let start_slice = self.peek().source_slice.clone(); - let name = self.consume(TokenType::Identifier, "Expect variable name.")?; - let name_lexeme = name.lexeme.clone(); - self.consume(TokenType::Equal, "Expect '=' after variable name.")?; - let value = self.expression()?; - let semicolon = self.consume( - TokenType::Semicolon, - "Expect ';' after variable declaration.", - )?; - let end_slice = semicolon.source_slice.clone(); - let combined_slice = SourceSlice::from_positions( - start_slice.source_id, - start_slice.start_position, - end_slice.end_position, - ); - Ok(AstNode::new( - Stmt::VarAssigment { - name: name_lexeme, - value: Box::new(value), - }, - combined_slice, - )) - } - - fn print_statement(&mut self) -> LoxResult> { + fn print_statement(&mut self) -> LoxResult { // consume the print keyword let start_slice = self.peek().source_slice.clone(); self.advance(); @@ -354,25 +294,32 @@ impl Parser { start_slice.start_position, end_slice.end_position, ); - Ok(AstNode::new( - Stmt::Print { + Ok(AstNode::new_statement( + Expr::Print { expression: Box::new(expr), }, combined_slice, )) } - fn block_statement(&mut self) -> LoxResult> { + fn block_statement(&mut self) -> LoxResult { let start_slice = self.peek().source_slice.clone(); self.advance(); + let mut label = None; + if self.peek().token_type == TokenType::Colon { + self.advance(); + label = Some( + self.consume(TokenType::Identifier, "Expect label after colon.")? + .lexeme + .clone(), + ); + } + let mut statements = Vec::new(); + while self.peek().token_type != TokenType::EndBlock && !self.is_at_end() { let stmt = self.statement()?; - let should_break = matches!(stmt.node, Stmt::Expression { .. }); statements.push(stmt); - if should_break { - break; - } } let end_token = self.consume(TokenType::EndBlock, "Expect 'end' after block.")?; let end_slice = end_token.source_slice.clone(); @@ -381,66 +328,100 @@ impl Parser { start_slice.start_position, end_slice.end_position, ); - Ok(AstNode::new( - Stmt::Block { + let label_str = if let Some(label) = label { + label + } else { + String::default() + }; + Ok(AstNode::new_statement( + Expr::Block { statements: Box::new(statements), + label: label_str, }, combined_slice, )) } - fn expression_statement(&mut self) -> LoxResult> { + fn expression_statement(&mut self) -> LoxResult { let start_slice = self.peek().source_slice.clone(); - let expr = self.expression()?; - if self.peek().token_type == TokenType::Semicolon { - let semicolon = self.advance(); - let end_slice = semicolon.source_slice.clone(); - let combined_slice = SourceSlice::from_positions( - start_slice.source_id, - start_slice.start_position, - end_slice.end_position, - ); - return Ok(AstNode::new( - Stmt::Stmt { - expression: Box::new(expr), - }, - combined_slice, - )); - } - // Use the expression's source slice for expression statements without semicolon - let expr_slice = expr.source_slice.clone(); - Ok(AstNode::new( - Stmt::Expression { - expression: Box::new(expr), - }, - expr_slice, - )) + let mut expr = self.expression()?; + let semicolon = + self.consume(TokenType::Semicolon, "Expected semicolon after expression")?; + let end_slice = semicolon.source_slice.clone(); + let combined_slice = SourceSlice::from_source_slices(start_slice, end_slice); + expr.is_statement = true; + expr.source_slice = combined_slice; + return Ok(expr); } - fn return_statement(&mut self) -> LoxResult> { + fn return_statement(&mut self, label: Option) -> LoxResult { let start_slice = self.peek().source_slice.clone(); self.advance(); + if self.peek().token_type == TokenType::Semicolon { + let semicolon = self.consume(TokenType::Semicolon, "Expect ';' after return value.")?; + let sorce_slice = + SourceSlice::from_source_slices(start_slice, semicolon.source_slice.clone()); + return Ok(AstNode::new_expression( + Expr::Literal { + value: Box::new(BaseValue::Nil), + }, + sorce_slice, + "Nil".to_string(), + )); + } let expr = self.expression()?; + let semicolon = self.consume(TokenType::Semicolon, "Expect ';' after return value.")?; let end_slice = semicolon.source_slice.clone(); - let combined_slice = SourceSlice::from_positions( - start_slice.source_id, - start_slice.start_position, - end_slice.end_position, - ); - Ok(AstNode::new( - Stmt::Return { + let combined_slice = SourceSlice::from_source_slices(start_slice, end_slice); + let label_str = if let Some(label) = label { + label + } else { + String::default() + }; + Ok(AstNode::new_statement( + Expr::Return { expression: Box::new(expr), + label: label_str, }, combined_slice, )) } - fn expression(&mut self) -> LoxResult> { - self.or_and() + fn expression(&mut self) -> LoxResult { + self.assignment() } - fn or_and(&mut self) -> LoxResult> { + fn assignment(&mut self) -> LoxResult { + let expr = self.or_and()?; + if self.peek().token_type == TokenType::Equal { + self.advance(); // consume '=' + // Right-associative: `a = b = c` parses as `a = (b = c)`. + let value = self.assignment()?; + let combined_slice = SourceSlice::from_positions( + expr.source_slice.source_id, + expr.source_slice.start_position.clone(), + value.source_slice.end_position.clone(), + ); + let target_slice = expr.source_slice.clone(); + let var_type = value.return_type.to_string(); + match expr.node { + Expr::Identifier { name } => Ok(AstNode::new_expression( + Expr::Assign { + name, + value: Box::new(value), + }, + combined_slice, + var_type, + )), + _ => parse_error(target_slice, "Invalid assignment target."), + } + } else { + Ok(expr) + } + } + + fn or_and(&mut self) -> LoxResult { let mut expr = self.logical_is()?; while [TokenType::Or, TokenType::And].contains(&self.peek().token_type) { @@ -454,20 +435,21 @@ impl Parser { start_slice.start_position, end_slice.end_position, ); - expr = AstNode::new( + expr = AstNode::new_expression( Expr::Binary { left: Box::new(expr), operator, right: Box::new(right), }, combined_slice, + "bool".to_string(), ); } Ok(expr) } - fn logical_is(&mut self) -> LoxResult> { + fn logical_is(&mut self) -> LoxResult { let mut expr = self.equality()?; while self.peek().token_type == TokenType::Is { @@ -481,20 +463,21 @@ impl Parser { start_slice.start_position, end_slice.end_position, ); - expr = AstNode::new( + expr = AstNode::new_expression( Expr::Binary { left: Box::new(expr), operator, right: Box::new(right), }, combined_slice, + "bool".to_string(), ); } Ok(expr) } - fn equality(&mut self) -> LoxResult> { + fn equality(&mut self) -> LoxResult { let mut expr = self.comparison()?; while [TokenType::BangEqual, TokenType::EqualEqual].contains(&self.peek().token_type) { @@ -508,20 +491,21 @@ impl Parser { start_slice.start_position, end_slice.end_position, ); - expr = AstNode::new( + expr = AstNode::new_expression( Expr::Binary { left: Box::new(expr), operator, right: Box::new(right), }, combined_slice, + "bool".to_string(), ); } Ok(expr) } - fn comparison(&mut self) -> LoxResult> { + fn comparison(&mut self) -> LoxResult { let mut expr = self.term()?; while [ @@ -542,20 +526,21 @@ impl Parser { start_slice.start_position, end_slice.end_position, ); - expr = AstNode::new( + expr = AstNode::new_expression( Expr::Binary { left: Box::new(expr), operator, right: Box::new(right), }, combined_slice, + "bool".to_string(), ); } Ok(expr) } - fn term(&mut self) -> LoxResult> { + fn term(&mut self) -> LoxResult { let mut expr = self.factor()?; while [TokenType::Plus, TokenType::Minus].contains(&self.peek().token_type) { @@ -569,20 +554,22 @@ impl Parser { start_slice.start_position, end_slice.end_position, ); - expr = AstNode::new( + let expr_type = expr.return_type.clone().to_string(); + expr = AstNode::new_expression( Expr::Binary { left: Box::new(expr), operator, right: Box::new(right), }, combined_slice, + expr_type, ); } Ok(expr) } - fn factor(&mut self) -> LoxResult> { + fn factor(&mut self) -> LoxResult { let mut expr = self.unary()?; while [TokenType::Slash, TokenType::Star].contains(&self.peek().token_type) { @@ -596,38 +583,42 @@ impl Parser { start_slice.start_position, end_slice.end_position, ); - expr = AstNode::new( + let expr_type = expr.return_type.clone().to_string(); + expr = AstNode::new_expression( Expr::Binary { left: Box::new(expr), operator, right: Box::new(right), }, combined_slice, + expr_type, ); } Ok(expr) } - fn unary(&mut self) -> LoxResult> { + fn unary(&mut self) -> LoxResult { if [TokenType::Bang, TokenType::Minus].contains(&self.peek().token_type) { let operator = self.peek().token_type.clone(); let source_slice = self.peek().source_slice.clone(); self.advance(); let right = self.unary()?; - return Ok(AstNode::new( + let expr_type = right.return_type.clone().to_string(); + return Ok(AstNode::new_expression( Expr::Unary { operator, operand: Box::new(right), }, source_slice, + expr_type, )); } self.call() } - fn call(&mut self) -> LoxResult> { + fn call(&mut self) -> LoxResult { let mut expr = self.primary()?; loop { @@ -640,7 +631,7 @@ impl Parser { Ok(expr) } - fn finish_call(&mut self, callee: AstNode) -> LoxResult> { + fn finish_call(&mut self, callee: AstNode) -> LoxResult { let start_slice = callee.source_slice.start_position.clone(); self.advance(); // Consume '(' @@ -666,46 +657,50 @@ impl Parser { start_position: start_slice.clone(), end_position: end_slice.clone(), }; - - Ok(AstNode::new( + let expr_type = callee.return_type.clone().to_string(); + Ok(AstNode::new_expression( Expr::Call { callee: Box::new(callee), arguments, }, full_slice, + expr_type, )) } - fn primary(&mut self) -> LoxResult> { + fn primary(&mut self) -> LoxResult { match self.peek().token_type { TokenType::False => { let source_slice = self.peek().source_slice.clone(); self.advance(); - Ok(AstNode::new( + Ok(AstNode::new_expression( Expr::Literal { - value: BaseValue::Boolean(false), + value: Box::new(BaseValue::Boolean(false)), }, source_slice, + "bool".to_string(), )) } TokenType::True => { let source_slice = self.peek().source_slice.clone(); self.advance(); - Ok(AstNode::new( + Ok(AstNode::new_expression( Expr::Literal { - value: BaseValue::Boolean(true), + value: Box::new(BaseValue::Boolean(true)), }, source_slice, + "bool".to_string(), )) } TokenType::Nil => { let source_slice = self.peek().source_slice.clone(); self.advance(); - Ok(AstNode::new( + Ok(AstNode::new_expression( Expr::Literal { - value: BaseValue::Nil, + value: Box::new(BaseValue::Nil), }, source_slice, + "nil".to_string(), )) } TokenType::Number => { @@ -713,7 +708,13 @@ impl Parser { let literal = self.peek().literal.clone(); self.advance(); if let Some(literal) = literal { - Ok(AstNode::new(Expr::Literal { value: literal }, source_slice)) + Ok(AstNode::new_expression( + Expr::Literal { + value: Box::new(literal), + }, + source_slice, + "number".to_string(), + )) } else { parse_error(self.peek().source_slice.clone(), "Expected number literal") } @@ -723,7 +724,13 @@ impl Parser { let literal = self.peek().literal.clone(); self.advance(); if let Some(literal) = literal { - Ok(AstNode::new(Expr::Literal { value: literal }, source_slice)) + Ok(AstNode::new_expression( + Expr::Literal { + value: Box::new(literal), + }, + source_slice, + "string".to_string(), + )) } else { parse_error(self.peek().source_slice.clone(), "Expected string literal") } @@ -733,18 +740,23 @@ impl Parser { self.advance(); let expr = self.expression()?; self.consume(TokenType::RightParen, "Expect ')' after expression.")?; - Ok(AstNode::new( + Ok(AstNode::new_expression( Expr::Grouping { expression: Box::new(expr), }, source_slice, + "grouping".to_string(), )) } TokenType::Identifier => { let name = self.peek().lexeme.clone(); let source_slice = self.peek().source_slice.clone(); self.advance(); - Ok(AstNode::new(Expr::Identifier { name }, source_slice)) + Ok(AstNode::new_expression( + Expr::Identifier { name }, + source_slice, + "identifier".to_string(), + )) } _ => parse_error(self.peek().source_slice.clone(), "Expect expression."), } @@ -798,6 +810,25 @@ impl Parser { } } + /// Consume a type name and return its lexeme. + /// + /// A type name is an identifier (e.g. `Int`, `Number`, a struct name) or a + /// built-in type keyword such as `nil` that overlaps with a value keyword. + fn consume_type_name(&mut self, message: &str) -> LoxResult { + match self.peek().token_type { + TokenType::Identifier | TokenType::Nil => Ok(self.advance().lexeme.clone()), + _ => parse_error(self.peek().source_slice.clone(), message), + } + } + + /// Whether the current token can begin a type name. + fn peek_is_type_name(&self) -> bool { + matches!( + self.peek().token_type, + TokenType::Identifier | TokenType::Nil + ) + } + fn synchronize(&mut self) { self.advance(); @@ -807,7 +838,7 @@ impl Parser { } match self.peek().token_type { - TokenType::Class + TokenType::Struct | TokenType::Fun | TokenType::For | TokenType::If @@ -821,3 +852,165 @@ impl Parser { } } } + +#[cfg(test)] +mod tests { + use super::*; + use crate::common::base_value::Number; + use crate::frontend::lexer::Lexer; + + /// Lex and parse `src`, returning the parser's result. + fn parse_source(src: &str) -> LoxResult> { + let tokens = Lexer::new(src.to_string(), 0) + .scans_tokens() + .expect("source should lex without errors"); + Parser::new(tokens).parse() + } + + /// Parse `src`, panicking if parsing fails. + fn parse_ok(src: &str) -> Vec { + parse_source(src).expect("expected source to parse") + } + + /// Extract the inner expression of an expression statement. + /// + /// Expression statements aren't wrapped in a dedicated `Expr` variant; + /// the parser just marks the node's `is_statement` flag, so the node + /// itself already holds the expression. + fn expression_of(stmt: &AstNode) -> &AstNode { + assert!(stmt.is_statement, "expected an expression statement"); + stmt + } + + #[test] + fn parses_number_literal_expression() { + let stmts = parse_ok("42;"); + assert_eq!(stmts.len(), 1); + match &expression_of(&stmts[0]).node { + Expr::Literal { value } => { + assert_eq!(**value, BaseValue::Number(Number::I32(42))); + } + other => panic!("expected literal, got {:?}", other), + } + } + + #[test] + fn respects_multiplication_precedence_over_addition() { + // 1 + 2 * 3 should parse as 1 + (2 * 3) + let stmts = parse_ok("1 + 2 * 3;"); + match &expression_of(&stmts[0]).node { + Expr::Binary { + operator, right, .. + } => { + assert_eq!(*operator, TokenType::Plus); + match &right.node { + Expr::Binary { operator, .. } => { + assert_eq!(*operator, TokenType::Star) + } + other => panic!("expected nested binary, got {:?}", other), + } + } + other => panic!("expected binary expression, got {:?}", other), + } + } + + #[test] + fn parses_unary_negation() { + let stmts = parse_ok("-5;"); + match &expression_of(&stmts[0]).node { + Expr::Unary { operator, .. } => assert_eq!(*operator, TokenType::Minus), + other => panic!("expected unary expression, got {:?}", other), + } + } + + #[test] + fn parses_grouping_to_change_precedence() { + // (1 + 2) * 3 should have a grouping on the left of the multiply. + let stmts = parse_ok("(1 + 2) * 3;"); + match &expression_of(&stmts[0]).node { + Expr::Binary { operator, left, .. } => { + assert_eq!(*operator, TokenType::Star); + assert!(matches!(left.node, Expr::Grouping { .. })); + } + other => panic!("expected binary expression, got {:?}", other), + } + } + + #[test] + fn parses_identifier_expression() { + let stmts = parse_ok("x;"); + match &expression_of(&stmts[0]).node { + Expr::Identifier { name } => assert_eq!(name, "x"), + other => panic!("expected identifier expression, got {:?}", other), + } + } + + #[test] + fn parses_comparison_expression() { + let stmts = parse_ok("1 < 2;"); + match &expression_of(&stmts[0]).node { + Expr::Binary { operator, .. } => assert_eq!(*operator, TokenType::Less), + other => panic!("expected binary expression, got {:?}", other), + } + } + + #[test] + fn parses_print_statement() { + let stmts = parse_ok("print 1;"); + assert!(matches!(stmts[0].node, Expr::Print { .. })); + } + + #[test] + fn parses_var_declaration_with_initializer() { + let stmts = parse_ok("var x: Int = 5;"); + match &stmts[0].node { + Expr::VarDeclaration { + name, initializer, .. + } => { + assert_eq!(name, "x"); + assert!(initializer.is_some()); + } + other => panic!("expected var declaration, got {:?}", other), + } + } + + #[test] + fn parses_assignment_statement() { + let stmts = parse_ok("x = 5;"); + match &expression_of(&stmts[0]).node { + Expr::Assign { name, .. } => assert_eq!(name, "x"), + other => panic!("expected assign expression, got {:?}", other), + } + } + + #[test] + fn parses_block_statement() { + let stmts = parse_ok("do print 1; print 2; end"); + match &stmts[0].node { + Expr::Block { statements, .. } => assert_eq!(statements.len(), 2), + other => panic!("expected block statement, got {:?}", other), + } + } + + #[test] + fn parses_if_statement() { + let stmts = parse_ok("if true then print 1;"); + assert!(matches!(stmts[0].node, Expr::If { .. })); + } + + #[test] + fn parses_while_statement() { + let stmts = parse_ok("while true do print 1; end"); + assert!(matches!(stmts[0].node, Expr::While { .. })); + } + + #[test] + fn errors_on_missing_semicolon_after_print() { + assert!(parse_source("print 1").is_err()); + } + + #[test] + fn errors_on_unclosed_grouping() { + assert!(parse_source("(1 + 2;").is_err()); + } +} diff --git a/src/frontend/source_registry.rs b/src/frontend/source_registry.rs old mode 100644 new mode 100755 index 3bdaa25..9dc55ef --- a/src/frontend/source_registry.rs +++ b/src/frontend/source_registry.rs @@ -4,7 +4,7 @@ use std::{ io::ErrorKind, }; -use crate::common::lox_result::{io_error, LoxError, LoxResult}; +use crate::common::lox_result::{io_error, LoxResult}; pub struct SourceRegistry { pub sources: Vec, @@ -51,7 +51,7 @@ impl SourceFile { } } -#[derive(Debug, Clone, PartialEq, Default)] +#[derive(Debug, Clone, PartialEq, Eq, Hash, Default)] pub struct SourcePosition { pub line: usize, pub column: usize, @@ -63,7 +63,7 @@ impl Display for SourcePosition { } } -#[derive(Clone, PartialEq, Default)] +#[derive(Clone, PartialEq, Eq, Hash)] pub struct SourceSlice { pub source_id: SourceId, pub start_position: SourcePosition, @@ -91,6 +91,25 @@ impl Display for SourceSlice { } impl SourceSlice { + /// An explicit, non-located placeholder span. + /// + /// Use this only when a real span is genuinely unavailable. Unlike the old + /// `Default` impl, it is greppable and obviously intentional, so it can't be + /// produced by accident. + pub fn synthetic() -> Self { + Self { + source_id: 0, + start_position: SourcePosition::default(), + end_position: SourcePosition::default(), + } + } + + /// Whether this span is the non-located placeholder produced by + /// [`SourceSlice::synthetic`]. + pub fn is_synthetic(&self) -> bool { + *self == Self::synthetic() + } + pub fn from_positions( source_id: SourceId, start_position: SourcePosition, @@ -102,6 +121,14 @@ impl SourceSlice { end_position, } } + + pub fn from_source_slices(start_point: SourceSlice, end_point: SourceSlice) -> Self { + Self { + source_id: start_point.source_id, + start_position: start_point.start_position, + end_position: end_point.end_position, + } + } pub fn tree_point( source_id: SourceId, column_start: usize, diff --git a/src/frontend/tokens.rs b/src/frontend/tokens.rs old mode 100644 new mode 100755 index cdad154..2701b7e --- a/src/frontend/tokens.rs +++ b/src/frontend/tokens.rs @@ -35,11 +35,12 @@ pub enum TokenType { Identifier, String, Number, + Atom, // Keywords Fn, And, - Class, + Struct, StartBlock, EndBlock, False, @@ -57,6 +58,7 @@ pub enum TokenType { Is, Print, Return, + Break, Super, This, Var, @@ -79,8 +81,9 @@ impl fmt::Display for TokenType { TokenType::Identifier => write!(f, "IDENTIFIER"), TokenType::String => write!(f, "STRING"), TokenType::Number => write!(f, "NUMBER"), + TokenType::Atom => write!(f, "ATOM"), TokenType::And => write!(f, "and"), - TokenType::Class => write!(f, "class"), + TokenType::Struct => write!(f, "struct"), TokenType::Else => write!(f, "else"), TokenType::False => write!(f, "false"), TokenType::True => write!(f, "true"), @@ -92,6 +95,7 @@ impl fmt::Display for TokenType { TokenType::Or => write!(f, "or"), TokenType::Print => write!(f, "print"), TokenType::Return => write!(f, "return"), + TokenType::Break => write!(f, "break"), TokenType::Super => write!(f, "super"), TokenType::This => write!(f, "this"), TokenType::Var => write!(f, "var"), diff --git a/src/lib.rs b/src/lib.rs old mode 100644 new mode 100755 index 2d494f2..5f4de5d --- a/src/lib.rs +++ b/src/lib.rs @@ -1,3 +1,5 @@ pub mod backend; pub mod common; pub mod frontend; +pub mod logging; +pub mod middleend; diff --git a/src/logging/display_ast.rs b/src/logging/display_ast.rs old mode 100644 new mode 100755 index da837e1..4055696 --- a/src/logging/display_ast.rs +++ b/src/logging/display_ast.rs @@ -5,7 +5,7 @@ use std::fmt::{self, Write}; -use crate::common::ast::{AstNode, AstNodeKind, Expr, Stmt}; +use crate::common::ast::{AstNode, Expr}; /// Configuration for pretty printing output #[derive(Clone, Debug)] @@ -234,6 +234,20 @@ impl PrettyPrint for Expr { write!(f, "{}Identifier {{ name: {:?} }}", indent, name) } } + Expr::Assign { name, value } => { + if ctx.config.compact { + let value_str = + pretty_print_with_config(value.as_ref(), &PrettyConfig::compact()); + write!(f, "{} = {}", name, value_str) + } else { + writeln!(f, "{}Assign {{", indent)?; + writeln!(f, "{}name: {:?},", ctx.child_context(false).indent(), name)?; + write!(f, "{}value: ", ctx.child_context(true).indent())?; + value.pretty_print(&ctx.child_context(true), f)?; + writeln!(f)?; + write!(f, "{}}}", indent) + } + } Expr::Call { callee, arguments } => { if ctx.config.compact { write!(f, "{}", callee) @@ -250,33 +264,7 @@ impl PrettyPrint for Expr { write!(f, "] }}") } } - } - } -} - -impl PrettyPrint for Stmt { - fn pretty_print(&self, ctx: &PrettyContext, f: &mut fmt::Formatter) -> fmt::Result { - if ctx.should_truncate() { - return write!(f, "{}...", ctx.indent()); - } - - let indent = ctx.indent(); - - match self { - Stmt::Expression { expression } => { - if ctx.config.compact { - let expr_str = - pretty_print_with_config(expression.as_ref(), &PrettyConfig::compact()); - write!(f, "{};", expr_str) - } else { - writeln!(f, "{}Expression {{", indent)?; - write!(f, "{}expression: ", ctx.child_context(true).indent())?; - expression.pretty_print(&ctx.child_context(true), f)?; - writeln!(f)?; - write!(f, "{}}}", indent) - } - } - Stmt::Print { expression } => { + Expr::Print { expression } => { if ctx.config.compact { let expr_str = pretty_print_with_config(expression.as_ref(), &PrettyConfig::compact()); @@ -289,7 +277,9 @@ impl PrettyPrint for Stmt { write!(f, "{}}}", indent) } } - Stmt::VarDeclaration { name, initializer } => { + Expr::VarDeclaration { + name, initializer, .. + } => { if ctx.config.compact { match initializer { Some(init) => { @@ -315,21 +305,7 @@ impl PrettyPrint for Stmt { write!(f, "{}}}", indent) } } - Stmt::VarAssigment { name, value } => { - if ctx.config.compact { - let value_str = - pretty_print_with_config(value.as_ref(), &PrettyConfig::compact()); - write!(f, "{} = {};", name, value_str) - } else { - writeln!(f, "{}Assign {{", indent)?; - writeln!(f, "{}name: {:?},", ctx.child_context(false).indent(), name)?; - write!(f, "{}value: ", ctx.child_context(true).indent())?; - value.pretty_print(&ctx.child_context(true), f)?; - writeln!(f)?; - write!(f, "{}}}", indent) - } - } - Stmt::Return { expression } => { + Expr::Return { expression, .. } => { if ctx.config.compact { let expr_str = pretty_print_with_config(expression.as_ref(), &PrettyConfig::compact()); @@ -342,11 +318,11 @@ impl PrettyPrint for Stmt { write!(f, "{}}}", indent) } } - Stmt::Block { statements } => { + Expr::Block { statements, label } => { if ctx.config.compact { - write!(f, "{{ ... ({} statements) }}", statements.len()) + write!(f, "{{ ... [{}] ({} statements) }}", label, statements.len()) } else { - writeln!(f, "{}Block {{", indent)?; + writeln!(f, "{}Block [{}] {{", label, indent)?; writeln!(f, "{}statements: [", ctx.child_context(false).indent())?; for (i, stmt) in statements.iter().enumerate() { let is_last = i == statements.len() - 1; @@ -361,10 +337,10 @@ impl PrettyPrint for Stmt { write!(f, "{}}}", indent) } } - Stmt::If { + Expr::If { condition, then_branch, - elif_branch, + elif_branches, else_branch, } => { if ctx.config.compact { @@ -380,10 +356,10 @@ impl PrettyPrint for Stmt { then_branch.pretty_print(&ctx.child_context(false), f)?; writeln!(f, ",")?; - if !elif_branch.is_empty() { + if !elif_branches.is_empty() { writeln!(f, "{}elif_branches: [", ctx.child_context(false).indent())?; - for (i, (cond, branch)) in elif_branch.iter().enumerate() { - let is_last = i == elif_branch.len() - 1; + for (i, (cond, branch)) in elif_branches.iter().enumerate() { + let is_last = i == elif_branches.len() - 1; let child_ctx = ctx.child_context(false).child_context(is_last); writeln!(f, "{}(", child_ctx.indent())?; write!(f, "{}condition: ", child_ctx.child_context(false).indent())?; @@ -415,15 +391,14 @@ impl PrettyPrint for Stmt { write!(f, "{}}}", indent) } } - Stmt::Stmt { expression } => expression.pretty_print(ctx, f), - Stmt::While { condition, body } => { + Expr::While { condition, body } => { write!(f, "{}while (", ctx.child_context(true).indent())?; condition.pretty_print(&ctx.child_context(true), f)?; writeln!(f, ") {{")?; body.pretty_print(&ctx.child_context(true), f)?; writeln!(f, "{}}}", ctx.child_context(true).indent()) } - Stmt::For { + Expr::For { variable, condition, increment, @@ -443,7 +418,7 @@ impl PrettyPrint for Stmt { } } -impl PrettyPrint for AstNode { +impl PrettyPrint for AstNode { fn pretty_print(&self, ctx: &PrettyContext, f: &mut fmt::Formatter) -> fmt::Result { let indent = ctx.indent(); @@ -455,9 +430,9 @@ impl PrettyPrint for A if ctx.config.show_types { writeln!( f, - "{}kind: {:?},", + "{}type: {},", ctx.child_context(false).indent(), - self.node.kind() + self.return_type )?; } @@ -511,25 +486,7 @@ impl PrettyPrintExt for Expr { } } -impl PrettyPrintExt for Stmt { - fn pretty(&self) -> String { - pretty_print(self) - } - - fn pretty_compact(&self) -> String { - pretty_print_with_config(self, &PrettyConfig::compact()) - } - - fn pretty_tree(&self) -> String { - pretty_print_with_config(self, &PrettyConfig::tree()) - } - - fn pretty_with_config(&self, config: &PrettyConfig) -> String { - pretty_print_with_config(self, config) - } -} - -impl PrettyPrintExt for AstNode { +impl PrettyPrintExt for AstNode { fn pretty(&self) -> String { pretty_print(self) } diff --git a/src/logging/display_token.rs b/src/logging/display_token.rs old mode 100644 new mode 100755 index f3b645f..4910c95 --- a/src/logging/display_token.rs +++ b/src/logging/display_token.rs @@ -174,7 +174,8 @@ impl Token { TokenType::String => "STR", TokenType::Number => "NUM", TokenType::And => "AND", - TokenType::Class => "CLASS", + TokenType::Struct => "STRUCT", + TokenType::Atom => "ATOM", TokenType::StartBlock => "DO", TokenType::EndBlock => "END", TokenType::False => "FALSE", @@ -190,6 +191,7 @@ impl Token { TokenType::Or => "OR", TokenType::Print => "PRINT", TokenType::Return => "RETURN", + TokenType::Break => "BREAK", TokenType::Super => "SUPER", TokenType::This => "THIS", TokenType::Var => "VAR", @@ -218,7 +220,7 @@ impl Token { | TokenType::Less | TokenType::LessEqual => ("\x1b[31m", self.token_type_symbol()), TokenType::And - | TokenType::Class + | TokenType::Struct | TokenType::False | TokenType::True | TokenType::Fun @@ -232,11 +234,13 @@ impl Token { | TokenType::Or | TokenType::Print | TokenType::Return + | TokenType::Break | TokenType::Super | TokenType::This | TokenType::Var | TokenType::Fn | TokenType::Is + | TokenType::Atom | TokenType::Val => ("\x1b[34m", self.token_type_symbol()), TokenType::Identifier | TokenType::String | TokenType::Number => { ("\x1b[32m", self.token_type_symbol()) diff --git a/src/logging/mod.rs b/src/logging/mod.rs old mode 100644 new mode 100755 diff --git a/src/main.rs b/src/main.rs old mode 100644 new mode 100755 index 98e2653..ff296d4 --- a/src/main.rs +++ b/src/main.rs @@ -1,12 +1,12 @@ -mod backend; -mod common; -mod frontend; -mod logging; - +pub mod backend; +pub mod common; +pub mod frontend; +pub mod logging; +pub mod middleend; use crate::{ backend::interpreter::{EvaluateInterpreter, Interpreter}, common::{ - ast::{AstNode, Stmt}, + ast::AstNode, lox_result::{LoxError, LoxResult}, }, frontend::{ @@ -14,6 +14,7 @@ use crate::{ parser::Parser, source_registry::{SourceId, SourceRegistry}, }, + middleend::variable_resolution::Resolver, }; use std::env; use std::fs; @@ -29,109 +30,35 @@ fn main() -> LoxResult<()> { let args: Vec = env::args().collect(); let (stage, file_path, debug) = parse_args(&args); + println!("running {file_path:?}"); let mut lox = LoxInterpreter::new(debug); - let _ = match file_path { + let res = match file_path { Some(path) => lox.run_file(&path, stage), None => lox.run_prompt(stage), }; - Ok(()) + res } fn parse_args(args: &[String]) -> (ExecutionStage, Option, bool) { - if args.len() == 1 { - // Solo il nome del programma: modalità interattiva completa - (ExecutionStage::Full, None, false) - } else if args.len() == 2 { - // Un argomento: potrebbe essere file o flag - let arg = &args[1]; - if arg == "--tokens" || arg == "--ast" || arg == "--full" || arg == "--debug" { - // Flag senza file: modalità interattiva - let stage = match arg.as_str() { - "--tokens" => ExecutionStage::Tokens, - "--ast" => ExecutionStage::Ast, - "--full" => ExecutionStage::Full, - "--debug" => ExecutionStage::Full, - _ => ExecutionStage::Full, - }; - let debug = arg == "--debug"; - (stage, None, debug) - } else { - // File senza flag: esecuzione completa del file - (ExecutionStage::Full, Some(arg.clone()), false) + println!("{args:?}"); + + let mut stage = ExecutionStage::Full; + let mut file_path = None; + let mut debug = false; + + for arg in args.iter().skip(1) { + match arg.as_str() { + "-t" | "--tokens" => stage = ExecutionStage::Tokens, + "-a" | "--ast" => stage = ExecutionStage::Ast, + "-f" | "--full" => stage = ExecutionStage::Full, + "-d" | "--debug" => debug = true, + _ => file_path = Some(arg.clone()), } - } else if args.len() == 3 { - // Due argomenti: potrebbero essere flag + file o due flag - let mut debug = false; - let mut stage = ExecutionStage::Full; - let mut file_path = None; - - for arg in &args[1..3] { - if arg == "--debug" { - debug = true; - } else if arg == "--tokens" || arg == "--ast" || arg == "--full" { - stage = match arg.as_str() { - "--tokens" => ExecutionStage::Tokens, - "--ast" => ExecutionStage::Ast, - "--full" => ExecutionStage::Full, - _ => ExecutionStage::Full, - }; - } else if !arg.starts_with("--") { - file_path = Some(arg.clone()); - } else { - eprintln!( - "Unknown flag: {}. Use --tokens, --ast, --full, or --debug", - arg - ); - eprintln!( - "Usage: {} [--tokens|--ast|--full] [--debug] [script]", - args[0] - ); - std::process::exit(64); - } - } - - (stage, file_path, debug) - } else if args.len() == 4 { - // Tre argomenti: flag + flag + file - let mut debug = false; - let mut stage = ExecutionStage::Full; - let mut file_path = None; - - for arg in &args[1..4] { - if arg == "--debug" { - debug = true; - } else if arg == "--tokens" || arg == "--ast" || arg == "--full" { - stage = match arg.as_str() { - "--tokens" => ExecutionStage::Tokens, - "--ast" => ExecutionStage::Ast, - "--full" => ExecutionStage::Full, - _ => ExecutionStage::Full, - }; - } else if !arg.starts_with("--") { - file_path = Some(arg.clone()); - } else { - eprintln!( - "Unknown flag: {}. Use --tokens, --ast, --full, or --debug", - arg - ); - eprintln!( - "Usage: {} [--tokens|--ast|--full] [--debug] [script]", - args[0] - ); - std::process::exit(64); - } - } - - (stage, file_path, debug) - } else { - eprintln!( - "Usage: {} [--tokens|--ast|--full] [--debug] [script]", - args[0] - ); - std::process::exit(64); } + + (stage, file_path, debug) } struct LoxInterpreter { @@ -247,7 +174,7 @@ impl LoxInterpreter { } // Funzione per parsare fino all'AST - fn parse_to_ast(&self, source_id: SourceId) -> LoxResult>> { + fn parse_to_ast(&self, source_id: SourceId) -> LoxResult> { let tokens = self.tokenize(source_id)?; Parser::new(tokens).parse().or_else(|err| { @@ -257,8 +184,12 @@ impl LoxInterpreter { } // Funzione per eseguire l'AST - fn execute_ast(&self, ast: Vec>, debug: bool) -> LoxResult { + fn execute_ast(&self, ast: Vec, debug: bool) -> LoxResult { + // Static resolution pass: compute variable scope distances and report + // any resolution errors before executing. + let locals = self.resolve(&ast)?; let mut interpreter = Interpreter::new(); + interpreter.set_locals(locals); let mut result = None; for (index, stmt) in ast.iter().enumerate() { @@ -277,6 +208,19 @@ impl LoxInterpreter { } } + // Static variable resolution; prints and returns the first error, if any. + fn resolve( + &self, + ast: &[AstNode], + ) -> LoxResult> { + Resolver::resolve_program(ast).map_err(|errors| { + for err in &errors { + err.print_with_context(&self.source_registry); + } + errors.into_iter().next().unwrap() + }) + } + fn process_source( &self, source_id: SourceId, @@ -318,8 +262,10 @@ impl LoxInterpreter { return Ok(format_ast(&ast)); } - // Stage 3: Interpretation + // Stage 3: Resolution + Interpretation + let locals = self.resolve(&ast)?; let mut interpreter = Interpreter::new(); + interpreter.set_locals(locals); let mut result = None; for (index, stmt) in ast.iter().enumerate() { @@ -348,7 +294,7 @@ fn format_tokens(tokens: &[crate::frontend::tokens::Token]) -> String { .join("\n") } -fn format_ast(ast: &[AstNode]) -> String { +fn format_ast(ast: &[AstNode]) -> String { use crate::logging::display_ast::{pretty_print_with_config, PrettyConfig}; let config = PrettyConfig { diff --git a/src/middleend/mod.rs b/src/middleend/mod.rs new file mode 100755 index 0000000..b361e3d --- /dev/null +++ b/src/middleend/mod.rs @@ -0,0 +1,3 @@ +pub mod scope_stack; +pub mod variable_resolution; +pub mod visit_ast; diff --git a/src/middleend/scope_stack.rs b/src/middleend/scope_stack.rs new file mode 100644 index 0000000..c19689a --- /dev/null +++ b/src/middleend/scope_stack.rs @@ -0,0 +1,146 @@ +//! A reusable lexical-scope stack for static analyses. +//! +//! Unlike [`EnvironmentStack`](crate::backend::environment::EnvironmentStack), +//! which stores runtime *values*, this stores per-binding *analysis state* +//! (for the resolver, a `bool` meaning "defined yet?"). It is generic over that +//! state so other passes (a type checker, an unused-variable lint, ...) can +//! reuse the same machinery. +//! +//! It starts **empty**: the global scope is intentionally untracked, so +//! [`ScopeStack::resolve`] returning `None` means "not local — assume global". + +use std::collections::HashMap; + +#[derive(Debug)] +pub struct ScopeStack { + scopes: Vec>, +} + +impl Default for ScopeStack { + fn default() -> Self { + Self::new() + } +} + +impl ScopeStack { + pub fn new() -> Self { + ScopeStack { scopes: Vec::new() } + } + + pub fn begin_scope(&mut self) { + self.scopes.push(HashMap::new()); + } + + pub fn end_scope(&mut self) { + self.scopes.pop(); + } + + pub fn is_empty(&self) -> bool { + self.scopes.is_empty() + } + + pub fn depth(&self) -> usize { + self.scopes.len() + } + + /// Insert `name` with `state` into the innermost scope (no-op at global). + pub fn declare(&mut self, name: impl Into, state: T) { + if let Some(scope) = self.scopes.last_mut() { + scope.insert(name.into(), state); + } + } + + /// Update an existing binding in the innermost scope (no-op if absent). + pub fn set_local(&mut self, name: &str, state: T) { + if let Some(scope) = self.scopes.last_mut() { + if let Some(slot) = scope.get_mut(name) { + *slot = state; + } + } + } + + /// Look up `name` in the innermost scope only. + pub fn get_local(&self, name: &str) -> Option<&T> { + self.scopes.last().and_then(|scope| scope.get(name)) + } + + /// Whether the innermost scope already declares `name`. + pub fn declared_in_current(&self, name: &str) -> bool { + self.scopes + .last() + .map_or(false, |scope| scope.contains_key(name)) + } + + /// Distance (in scopes) from the innermost scope to the one declaring + /// `name`, or `None` if it isn't in any scope (i.e. global). + pub fn resolve(&self, name: &str) -> Option { + self.scopes + .iter() + .rev() + .enumerate() + .find_map(|(distance, scope)| scope.contains_key(name).then_some(distance)) + } +} + +#[cfg(test)] +mod tests { + use super::*; + + #[test] + fn declare_and_get_local() { + let mut scopes: ScopeStack = ScopeStack::new(); + scopes.begin_scope(); + scopes.declare("a", false); + assert_eq!(scopes.get_local("a"), Some(&false)); + scopes.set_local("a", true); + assert_eq!(scopes.get_local("a"), Some(&true)); + assert_eq!(scopes.get_local("missing"), None); + } + + #[test] + fn resolve_returns_distance_from_innermost() { + let mut scopes: ScopeStack = ScopeStack::new(); + scopes.begin_scope(); + scopes.declare("a", true); + scopes.begin_scope(); + scopes.declare("b", true); + assert_eq!(scopes.resolve("b"), Some(0)); + assert_eq!(scopes.resolve("a"), Some(1)); + assert_eq!(scopes.resolve("missing"), None); + } + + #[test] + fn shadowing_and_end_scope() { + let mut scopes: ScopeStack = ScopeStack::new(); + scopes.begin_scope(); + scopes.declare("a", true); + scopes.begin_scope(); + scopes.declare("a", false); + assert_eq!(scopes.get_local("a"), Some(&false)); + assert_eq!(scopes.resolve("a"), Some(0)); + scopes.end_scope(); + assert_eq!(scopes.get_local("a"), Some(&true)); + assert_eq!(scopes.resolve("a"), Some(0)); + } + + #[test] + fn declared_in_current_checks_only_innermost() { + let mut scopes: ScopeStack = ScopeStack::new(); + scopes.begin_scope(); + scopes.declare("a", true); + scopes.begin_scope(); + assert!(!scopes.declared_in_current("a")); + scopes.declare("a", true); + assert!(scopes.declared_in_current("a")); + } + + #[test] + fn global_scope_is_untracked() { + let mut scopes: ScopeStack = ScopeStack::new(); + // No scope pushed: declarations are no-ops and nothing resolves locally. + scopes.declare("a", true); + assert!(scopes.is_empty()); + assert_eq!(scopes.resolve("a"), None); + assert_eq!(scopes.get_local("a"), None); + } +} diff --git a/src/middleend/variable_resolution.rs b/src/middleend/variable_resolution.rs new file mode 100755 index 0000000..5827240 --- /dev/null +++ b/src/middleend/variable_resolution.rs @@ -0,0 +1,258 @@ +//! Static variable resolution (Crafting Interpreters, chapter 11). +//! +//! Walks the AST with the shared [`Visitor`] traversal, tracking lexical scopes +//! in a [`ScopeStack`], and records for every variable *reference* how many +//! scopes up its declaration lives (`locals: NodeId -> distance`). It also +//! reports the resolution errors from chapter 11: +//! +//! * reading a local variable in its own initializer (`var a = a;`), +//! * declaring two variables with the same name in one local scope, +//! * `return` outside of any function. +//! +//! Diagnostics accumulate in an [`ErrorSink`] so a single pass surfaces them +//! all. The produced `locals` map is keyed by [`NodeId`] (stable identity), +//! ready for the interpreter to consume via distance-based lookup. + +use std::collections::HashMap; + +use crate::{ + common::{ + ast::{AstNode, Expr, NodeId}, + base_value::{BaseValue, LoxFunction}, + lox_result::{ErrorSink, LoxError, LoxResult}, + }, + middleend::{ + scope_stack::ScopeStack, + visit_ast::{walk_expr, walk_function, Visitor}, + }, +}; + +/// Tracks whether resolution is currently inside a function body, so a +/// top-level `return` can be reported. +#[derive(Clone, Copy, PartialEq)] +enum FunctionType { + None, + Function, +} + +pub struct Resolver { + scopes: ScopeStack, + locals: HashMap, + errors: ErrorSink, + current_function: FunctionType, +} + +impl Resolver { + pub fn new() -> Self { + Resolver { + scopes: ScopeStack::new(), + locals: HashMap::new(), + errors: ErrorSink::new(), + current_function: FunctionType::None, + } + } + + /// Resolve a whole program, returning the per-reference scope distances or + /// the accumulated resolution errors. + pub fn resolve_program( + statements: &[AstNode], + ) -> Result, Vec> { + let mut resolver = Resolver::new(); + for statement in statements { + // Visiting only fails for fatal/internal errors, which this pass + // never produces; user-facing diagnostics go to `errors`. + let _ = resolver.visit_expr(statement); + } + if resolver.errors.has_errors() { + Err(resolver.errors.into_errors()) + } else { + Ok(resolver.locals) + } + } + + fn declare(&mut self, name: &str, slice: &crate::frontend::source_registry::SourceSlice) { + if self.scopes.is_empty() { + return; // global scope is untracked + } + if self.scopes.declared_in_current(name) { + self.errors.report(LoxError::ParseError { + source_slice: slice.clone(), + message: format!("Already a variable named '{}' in this scope.", name), + }); + } + self.scopes.declare(name.to_string(), false); + } + + fn define(&mut self, name: &str) { + if self.scopes.is_empty() { + return; + } + self.scopes.set_local(name, true); + } + + fn resolve_local(&mut self, id: NodeId, name: &str) { + if let Some(distance) = self.scopes.resolve(name) { + self.locals.insert(id, distance); + } + // Not found locally: assume global, record nothing. + } +} + +impl Default for Resolver { + fn default() -> Self { + Self::new() + } +} + +impl Visitor for Resolver { + fn visit_expr(&mut self, expr: &AstNode) -> LoxResult<()> { + match &expr.node { + Expr::Identifier { name } => { + if self.scopes.get_local(name) == Some(&false) { + self.errors.report(LoxError::ParseError { + source_slice: expr.source_slice.clone(), + message: "Can't read local variable in its own initializer.".to_string(), + }); + } + self.resolve_local(expr.id, name); + Ok(()) + } + Expr::Assign { name, .. } => { + walk_expr(self, expr)?; // resolve the assigned value first + self.resolve_local(expr.id, name); + Ok(()) + } + Expr::VarDeclaration { + name, initializer, .. + } => { + // A function declaration is a var bound to a function literal. + // Define its name *before* resolving the body so it can recurse; + // a plain variable is defined *after* its initializer so that + // `var a = a;` is caught. + let is_function = matches!( + initializer.as_deref().map(|node| &node.node), + Some(Expr::Literal { value }) if matches!(**value, BaseValue::Function(_)) + ); + self.declare(name, &expr.source_slice); + if is_function { + self.define(name); + walk_expr(self, expr)?; + } else { + walk_expr(self, expr)?; // resolves the initializer, if any + self.define(name); + } + Ok(()) + } + Expr::Block { .. } => { + self.scopes.begin_scope(); + walk_expr(self, expr)?; + self.scopes.end_scope(); + Ok(()) + } + Expr::Return { .. } => { + if self.current_function == FunctionType::None { + self.errors.report(LoxError::ParseError { + source_slice: expr.source_slice.clone(), + message: "Can't return from top-level code.".to_string(), + }); + } + walk_expr(self, expr) // resolve the returned expression + } + _ => walk_expr(self, expr), + } + } + + fn visit_function(&mut self, function: &LoxFunction) -> LoxResult<()> { + let enclosing = std::mem::replace(&mut self.current_function, FunctionType::Function); + self.scopes.begin_scope(); + // Parameters carry no source slice of their own; use the body's. + let param_slice = function.body.source_slice.clone(); + for (param, _ty) in &function.parameters { + self.declare(param, ¶m_slice); + self.define(param); + } + walk_function(self, function)?; // resolves guard + body + self.scopes.end_scope(); + self.current_function = enclosing; + Ok(()) + } +} + +#[cfg(test)] +mod tests { + use super::*; + use crate::frontend::lexer::Lexer; + use crate::frontend::parser::Parser; + + fn parse(src: &str) -> Vec { + let tokens = Lexer::new(src.to_string(), 0) + .scans_tokens() + .expect("source should lex"); + Parser::new(tokens).parse().expect("source should parse") + } + + fn resolve(src: &str) -> Result, Vec> { + Resolver::resolve_program(&parse(src)) + } + + #[test] + fn global_variables_are_not_resolved() { + // Top-level (global) scope is untracked, so nothing is recorded. + let locals = resolve("x := 1; print x;").expect("should resolve"); + assert!(locals.is_empty()); + } + + #[test] + fn local_read_resolves_to_distance_zero() { + let locals = resolve("do x := 1; print x; end").expect("should resolve"); + assert_eq!(locals.len(), 1); + assert_eq!(*locals.values().next().unwrap(), 0); + } + + #[test] + fn nested_scope_resolves_to_outer_distance() { + let src = "do x := 1; do print x; end end"; + let locals = resolve(src).expect("should resolve"); + assert_eq!(locals.len(), 1); + // `x` is read one scope above where it is read from. + assert_eq!(*locals.values().next().unwrap(), 1); + } + + #[test] + fn reading_a_variable_in_its_own_initializer_is_an_error() { + let errors = resolve("do x := x; end").expect_err("should fail"); + assert!(errors + .iter() + .any(|e| e.get_message().contains("its own initializer"))); + } + + #[test] + fn duplicate_declaration_in_same_scope_is_an_error() { + let errors = resolve("do x := 1; x := 2; end").expect_err("should fail"); + assert!(errors + .iter() + .any(|e| e.get_message().contains("Already a variable"))); + } + + #[test] + fn return_at_top_level_is_an_error() { + let errors = resolve("return 1;").expect_err("should fail"); + assert!(errors.iter().any(|e| e.get_message().contains("top-level"))); + } + + #[test] + fn return_inside_a_function_is_allowed() { + let locals = resolve("f :: fn (n): Number do return n; end;").expect("should resolve"); + // `n` resolves from the body block up to the parameter scope. + assert_eq!(locals.len(), 1); + assert_eq!(*locals.values().next().unwrap(), 1); + } + + #[test] + fn assignment_target_is_resolved() { + let locals = resolve("do x := 1; x = 2; end").expect("should resolve"); + // Both the assignment target and... only the target is a reference here. + assert!(locals.values().all(|&d| d == 0)); + assert!(!locals.is_empty()); + } +} diff --git a/src/middleend/visit_ast.rs b/src/middleend/visit_ast.rs new file mode 100644 index 0000000..75be11e --- /dev/null +++ b/src/middleend/visit_ast.rs @@ -0,0 +1,228 @@ +//! A reusable, read-only AST visitor with default traversal. +//! +//! The traversal is split into two layers so that many different static +//! analyses can share a single definition of "how to walk the tree": +//! +//! * The `walk_*` free functions contain the canonical recursion. For each +//! node they call back into the visitor on every child. This is the only +//! place that needs to know the shape of the AST. +//! * The [`Visitor`] trait's `visit_*` methods are the overridable hooks. Each +//! one defaults to calling the matching `walk_*` function, so a visitor that +//! overrides nothing still performs a complete traversal. +//! +//! To implement an analysis, implement [`Visitor`] and override only the hooks +//! you care about. Inside an override, call the corresponding `walk_*` function +//! whenever you want the default "descend into children" behaviour to happen. +//! Accumulate results in your own struct fields (errors, scope tables, type +//! information, ...) rather than through the return value, which is only used +//! to short-circuit on error. +//! +//! # Example +//! +//! ```ignore +//! struct IdentifierCounter { count: usize } +//! +//! impl Visitor for IdentifierCounter { +//! fn visit_expr(&mut self, expr: &AstNode) -> LoxResult<()> { +//! if let Expr::Identifier { .. } = &expr.node { +//! self.count += 1; +//! } +//! walk_expr(self, expr) // keep descending into children +//! } +//! } +//! ``` + +use crate::common::{ + ast::{AstNode, Expr}, + base_value::{BaseValue, LoxFunction}, + lox_result::LoxResult, +}; + +/// A read-only visitor over the AST. +/// +/// Every hook has a default implementation that performs the standard +/// recursive traversal, so implementors only override the cases they need. +pub trait Visitor: Sized { + /// Visit an expression node. Defaults to [`walk_expr`]. + fn visit_expr(&mut self, expr: &AstNode) -> LoxResult<()> { + walk_expr(self, expr) + } + + /// Visit a function literal (parameters, optional guard and body). + /// + /// Defaults to [`walk_function`], which walks the guard (if any) and the + /// body. Override this to manage a parameter scope before descending. + fn visit_function(&mut self, function: &LoxFunction) -> LoxResult<()> { + walk_function(self, function) + } +} + +/// Recurse into the children of `expr`, calling back into `visitor`. +pub fn walk_expr(visitor: &mut V, expr: &AstNode) -> LoxResult<()> { + match &expr.node { + // A function literal carries an entire sub-tree (its body), so it is + // not a leaf: hand it to the dedicated function hook. + Expr::Literal { value } => { + if let BaseValue::Function(function) = &**value { + visitor.visit_function(function)?; + } + Ok(()) + } + Expr::Identifier { .. } => Ok(()), + Expr::Binary { left, right, .. } => { + visitor.visit_expr(left)?; + visitor.visit_expr(right) + } + Expr::Unary { operand, .. } => visitor.visit_expr(operand), + Expr::Assign { value, .. } => visitor.visit_expr(value), + Expr::Grouping { expression } => visitor.visit_expr(expression), + Expr::Call { + callee, arguments, .. + } => { + visitor.visit_expr(callee)?; + for argument in arguments.iter() { + visitor.visit_expr(argument)?; + } + Ok(()) + } + + Expr::Print { expression, .. } => visitor.visit_expr(expression), + Expr::VarDeclaration { initializer, .. } => { + if let Some(initializer) = initializer { + visitor.visit_expr(initializer)?; + } + Ok(()) + } + Expr::Return { expression, .. } => visitor.visit_expr(expression), + Expr::Block { statements, .. } => { + for statement in statements.iter() { + visitor.visit_expr(statement)?; + } + Ok(()) + } + Expr::If { + condition, + then_branch, + elif_branches, + else_branch, + .. + } => { + visitor.visit_expr(condition)?; + visitor.visit_expr(then_branch)?; + for (elif_condition, elif_body) in elif_branches.iter() { + visitor.visit_expr(elif_condition)?; + visitor.visit_expr(elif_body)?; + } + if let Some(else_body) = else_branch { + visitor.visit_expr(else_body)?; + } + Ok(()) + } + Expr::While { + condition, body, .. + } => { + visitor.visit_expr(condition)?; + visitor.visit_expr(body) + } + Expr::For { + variable, + condition, + increment, + body, + .. + } => { + visitor.visit_expr(variable)?; + visitor.visit_expr(condition)?; + visitor.visit_expr(increment)?; + visitor.visit_expr(body) + } + } +} + +/// Walk the guard (if present) and body of a function literal. +pub fn walk_function(visitor: &mut V, function: &LoxFunction) -> LoxResult<()> { + if let Some(guard) = &function.guard { + visitor.visit_expr(guard)?; + } + visitor.visit_expr(&function.body) +} + +#[cfg(test)] +mod tests { + use super::*; + use crate::common::lox_result::runtime_error; + use crate::frontend::lexer::Lexer; + use crate::frontend::parser::Parser; + + fn parse(src: &str) -> Vec { + let tokens = Lexer::new(src.to_string(), 0) + .scans_tokens() + .expect("source should lex"); + Parser::new(tokens).parse().expect("source should parse") + } + + /// A visitor that relies entirely on the default traversal and just counts + /// how many nodes it sees. + #[derive(Default)] + struct Counter { + nodes: usize, + } + + impl Visitor for Counter { + fn visit_expr(&mut self, node: &AstNode) -> LoxResult<()> { + self.nodes += 1; + walk_expr(self, node) + } + } + + fn count(src: &str) -> Counter { + let mut counter = Counter::default(); + for stmt in parse(src).iter() { + counter.visit_expr(stmt).unwrap(); + } + counter + } + + #[test] + fn counts_every_node_via_default_traversal() { + // 1 + 2 * 3 => Binary(+){ Literal, Binary(*){ Literal, Literal } } + let counter = count("1 + 2 * 3;"); + assert_eq!(counter.nodes, 5); + } + + #[test] + fn descends_into_function_bodies() { + // The function body must be traversed through `visit_function`, so the + // `return a;` inside it should contribute to the counts. + let counter = count("f :: fn (a): Number do return a; end;"); + // VarDeclaration + Function literal + Block + Return + identifier `a` + assert_eq!(counter.nodes, 5); + } + + /// A visitor that aborts as soon as it sees an identifier, used to check + /// that errors short-circuit the traversal. + struct FailOnIdentifier; + + impl Visitor for FailOnIdentifier { + fn visit_expr(&mut self, expr: &AstNode) -> LoxResult<()> { + if let Expr::Identifier { .. } = &expr.node { + return runtime_error(expr.source_slice.clone(), "found an identifier"); + } + walk_expr(self, expr) + } + } + + #[test] + fn errors_propagate_through_traversal() { + let stmts = parse("var x: Int = 1; x;"); + let mut visitor = FailOnIdentifier; + let mut result = Ok(()); + for stmt in stmts.iter() { + result = visitor.visit_expr(stmt); + if result.is_err() { + break; + } + } + assert!(result.is_err()); + } +} diff --git a/tests/frontend.rs b/tests/frontend.rs new file mode 100644 index 0000000..004e082 --- /dev/null +++ b/tests/frontend.rs @@ -0,0 +1,296 @@ +//! End-to-end frontend tests: source text -> lexer -> parser -> AST. +//! +//! These exercise the lexer and parser together through the public API and +//! assert on the resulting AST, complementing the per-module unit tests inside +//! `src/frontend/{lexer,parser}.rs`. + +use rlox::common::ast::{AstNode, Expr}; +use rlox::common::base_value::{BaseValue, Number}; +use rlox::frontend::lexer::Lexer; +use rlox::frontend::parser::Parser; +use rlox::frontend::tokens::TokenType; + +/// Run the full frontend pipeline on `src`. +fn parse(src: &str) -> Result, String> { + let tokens = Lexer::new(src.to_string(), 0) + .scans_tokens() + .map_err(|e| format!("lex error: {e}"))?; + Parser::new(tokens) + .parse() + .map_err(|e| format!("parse error: {e}")) +} + +/// Parse `src`, panicking if the frontend reports any error. +fn parse_ok(src: &str) -> Vec { + parse(src).expect("expected source to lex and parse") +} + +/// Parse `src` and return the single node it should produce. +/// +/// Statements and expressions share the unified `Expr`/`AstNode` type, so this +/// simply returns the node of the sole top-level statement. +fn single_stmt(src: &str) -> Expr { + let mut stmts = parse_ok(src); + assert_eq!(stmts.len(), 1, "expected exactly one statement for {src:?}"); + stmts.remove(0).node +} + +/// Parse `src` and return its single expression-statement's expression. +/// +/// Expression statements aren't wrapped in a dedicated variant anymore; the +/// node itself is the expression. +fn single_expr(src: &str) -> Expr { + single_stmt(src) +} + +// --------------------------------------------------------------------------- +// Expressions +// --------------------------------------------------------------------------- + +#[test] +fn lexes_and_parses_number_literal() { + match single_expr("42;") { + Expr::Literal { value } => assert_eq!(*value, BaseValue::Number(Number::I32(42))), + other => panic!("expected literal, got {other:?}"), + } +} + +#[test] +fn parses_string_and_boolean_literals() { + assert!(matches!(single_expr("true;"), Expr::Literal { .. })); + assert!(matches!(single_expr("\"hi\";"), Expr::Literal { .. })); +} + +#[test] +fn multiplication_binds_tighter_than_addition() { + // 1 + 2 * 3 => (+ 1 (* 2 3)) + match single_expr("1 + 2 * 3;") { + Expr::Binary { + operator, right, .. + } => { + assert_eq!(operator, TokenType::Plus); + assert!(matches!( + right.node, + Expr::Binary { + operator: TokenType::Star, + .. + } + )); + } + other => panic!("expected binary, got {other:?}"), + } +} + +#[test] +fn grouping_overrides_precedence() { + // (1 + 2) * 3 => (* (group (+ 1 2)) 3) + match single_expr("(1 + 2) * 3;") { + Expr::Binary { operator, left, .. } => { + assert_eq!(operator, TokenType::Star); + assert!(matches!(left.node, Expr::Grouping { .. })); + } + other => panic!("expected binary, got {other:?}"), + } +} + +#[test] +fn parses_unary_operators() { + assert!(matches!( + single_expr("-5;"), + Expr::Unary { + operator: TokenType::Minus, + .. + } + )); + assert!(matches!( + single_expr("!true;"), + Expr::Unary { + operator: TokenType::Bang, + .. + } + )); +} + +#[test] +fn parses_comparison_and_equality() { + assert!(matches!( + single_expr("1 < 2;"), + Expr::Binary { + operator: TokenType::Less, + .. + } + )); + assert!(matches!( + single_expr("1 == 2;"), + Expr::Binary { + operator: TokenType::EqualEqual, + .. + } + )); +} + +#[test] +fn parses_logical_operators() { + assert!(matches!( + single_expr("true or false;"), + Expr::Binary { + operator: TokenType::Or, + .. + } + )); + assert!(matches!( + single_expr("true and false;"), + Expr::Binary { + operator: TokenType::And, + .. + } + )); +} + +#[test] +fn parses_call_with_arguments() { + match single_expr("add(1, 2);") { + Expr::Call { callee, arguments } => { + assert!(matches!(callee.node, Expr::Identifier { .. })); + assert_eq!(arguments.len(), 2); + } + other => panic!("expected call, got {other:?}"), + } +} + +// --------------------------------------------------------------------------- +// Assignment (now an expression) +// --------------------------------------------------------------------------- + +#[test] +fn assignment_is_an_expression_statement() { + match single_expr("x = 5;") { + Expr::Assign { name, .. } => assert_eq!(name, "x"), + other => panic!("expected assign, got {other:?}"), + } +} + +#[test] +fn assignment_is_right_associative() { + // a = b = c => (assign a (assign b c)) + match single_expr("a = b = c;") { + Expr::Assign { name, value } => { + assert_eq!(name, "a"); + match value.node { + Expr::Assign { name, .. } => assert_eq!(name, "b"), + other => panic!("expected nested assign, got {other:?}"), + } + } + other => panic!("expected assign, got {other:?}"), + } +} + +#[test] +fn assignment_to_non_identifier_is_an_error() { + assert!(parse("1 = 2;").is_err()); +} + +// --------------------------------------------------------------------------- +// Statements +// --------------------------------------------------------------------------- + +#[test] +fn parses_var_declaration() { + match single_stmt("x := 5;") { + Expr::VarDeclaration { + name, initializer, .. + } => { + assert_eq!(name, "x"); + assert!(initializer.is_some()); + } + other => panic!("expected var declaration, got {other:?}"), + } +} + +#[test] +fn parses_print_statement() { + assert!(matches!(single_stmt("print 1;"), Expr::Print { .. })); +} + +#[test] +fn parses_block_with_multiple_statements() { + match single_stmt("do print 1; print 2; end") { + Expr::Block { statements, .. } => assert_eq!(statements.len(), 2), + other => panic!("expected block, got {other:?}"), + } +} + +#[test] +fn parses_if_else() { + match single_stmt("if true then print 1; else print 2;") { + Expr::If { + else_branch: Some(_), + .. + } => {} + other => panic!("expected if/else, got {other:?}"), + } +} + +#[test] +fn parses_while_loop() { + assert!(matches!( + single_stmt("while true do print 1; end"), + Expr::While { .. } + )); +} + +#[test] +fn parses_for_loop_with_assignment_increment() { + match single_stmt("for i := 0; i <= 2; i = i + 1; do print i; end") { + Expr::For { increment, .. } => { + // The increment is an assignment expression. + assert!(matches!(increment.node, Expr::Assign { .. })) + } + other => panic!("expected for loop, got {other:?}"), + } +} + +#[test] +fn parses_function_declaration() { + match single_stmt("add :: fn (a, b): Number do return a + b; end;") { + Expr::VarDeclaration { + name, initializer, .. + } => { + assert_eq!(name, "add"); + match initializer { + Some(init) => assert!(matches!( + &init.node, + Expr::Literal { value } if matches!(&**value, BaseValue::Function(_)) + )), + None => panic!("expected a function initializer"), + } + } + other => panic!("expected function declaration, got {other:?}"), + } +} + +// --------------------------------------------------------------------------- +// Whole-program / error handling +// --------------------------------------------------------------------------- + +#[test] +fn parses_a_multi_statement_program() { + let stmts = parse_ok("x := 1; y := 2; print x + y;"); + assert_eq!(stmts.len(), 3); +} + +#[test] +fn lexer_errors_surface_through_the_frontend() { + // Unterminated string is a lexical error reported by the lexer stage. + assert!(parse("\"unterminated;").is_err()); +} + +#[test] +fn missing_semicolon_is_a_parse_error() { + assert!(parse("print 1").is_err()); +} + +#[test] +fn unclosed_grouping_is_a_parse_error() { + assert!(parse("(1 + 2;").is_err()); +} diff --git a/tests/number_operations.rs b/tests/number_operations.rs old mode 100644 new mode 100755 diff --git a/tests/resolution.rs b/tests/resolution.rs new file mode 100644 index 0000000..a897f83 --- /dev/null +++ b/tests/resolution.rs @@ -0,0 +1,69 @@ +//! End-to-end pipeline tests: lexer -> parser -> resolver -> interpreter. +//! +//! These verify chapter-11 resolution wired into execution: a variable's +//! resolved scope distance is used for lookup (`get_at`), and resolution errors +//! abort before running. + +use rlox::backend::interpreter::{EvaluateInterpreter, Interpreter}; +use rlox::common::base_value::BaseValue; +use rlox::frontend::lexer::Lexer; +use rlox::frontend::parser::Parser; +use rlox::middleend::variable_resolution::Resolver; + +/// Run the whole pipeline, returning the value of the last statement. +fn run(src: &str) -> Result { + let tokens = Lexer::new(src.to_string(), 0) + .scans_tokens() + .map_err(|e| e.to_string())?; + let ast = Parser::new(tokens).parse().map_err(|e| e.to_string())?; + let locals = Resolver::resolve_program(&ast) + .map_err(|errors| errors.into_iter().next().unwrap().to_string())?; + let mut interpreter = Interpreter::new(); + interpreter.set_locals(locals); + let mut last = BaseValue::Nil; + for stmt in ast { + last = interpreter.evaluate(stmt).map_err(|e| e.to_string())?; + } + Ok(last) +} + +#[test] +fn block_local_variables_evaluate() { + assert_eq!(run("do x := 10; x + 5; end").unwrap().to_string(), "15"); +} + +#[test] +fn resolved_nested_closure_reads_captured_local() { + // `count` is read from inside a nested function, two scopes up; the + // resolver records distance 2 and the interpreter uses `get_at(2, ..)`. + let src = "\ +make :: fn (): Any do + count := 10; + get :: fn (): Number do + return count; + end; + return get; +end; +g := make(); +g(); +"; + assert_eq!(run(src).unwrap().to_string(), "10"); +} + +#[test] +fn use_before_initializer_is_rejected() { + let err = run("do x := x; end").expect_err("should be a resolution error"); + assert!(err.contains("its own initializer")); +} + +#[test] +fn top_level_return_is_rejected() { + let err = run("return 1;").expect_err("should be a resolution error"); + assert!(err.contains("top-level")); +} + +#[test] +fn duplicate_declaration_in_block_is_rejected() { + let err = run("do x := 1; x := 2; end").expect_err("should be a resolution error"); + assert!(err.contains("Already a variable")); +} diff --git a/tests/test_example.rs b/tests/test_example.rs new file mode 100755 index 0000000..8dc261f --- /dev/null +++ b/tests/test_example.rs @@ -0,0 +1,207 @@ +use rlox::backend::interpreter::{EvaluateInterpreter, Interpreter}; +use rlox::frontend::lexer::Lexer; +use rlox::frontend::parser::Parser; +use rlox::frontend::source_registry::SourceRegistry; +use std::fs; +use std::path::Path; + +#[test] +fn test_all_examples() { + let examples_dir = Path::new("examples"); + + // Trova tutti i file .lox nella directory examples + let entries = fs::read_dir(examples_dir).expect("Failed to read examples directory"); + + let mut test_files_passed = Vec::new(); + let mut test_files_failed = Vec::new(); + let mut fail_files_passed = Vec::new(); + let mut fail_files_failed = Vec::new(); + + for entry in entries { + let entry = entry.expect("Failed to read directory entry"); + let path = entry.path(); + + // Controlla se è un file .lox + if path.extension().and_then(|s| s.to_str()) == Some("lox") { + let file_name = path.file_name().unwrap().to_str().unwrap().to_string(); + + // Ignora i file che non iniziano con test_ o fail_ + if !file_name.starts_with("test_") && !file_name.starts_with("fail_") { + continue; + } + + // Leggi il contenuto del file + let source = match fs::read_to_string(&path) { + Ok(content) => content, + Err(e) => { + if file_name.starts_with("test_") { + test_files_failed.push((file_name, format!("Failed to read file: {}", e))); + } else { + fail_files_failed.push((file_name, format!("Failed to read file: {}", e))); + } + continue; + } + }; + + // Esegui il file e categorizza il risultato + match run_lox_file(source, &file_name) { + Ok(_) => { + if file_name.starts_with("test_") { + test_files_passed.push(file_name); + } else { + fail_files_passed.push(file_name); + } + } + Err(error) => { + if file_name.starts_with("test_") { + test_files_failed.push((file_name, error)); + } else { + fail_files_failed.push((file_name, error)); + } + } + } + } + } + + // Stampa i risultati + println!("\n========== TEST RESULTS =========="); + + // File test_ (dovrebbero passare) + if !test_files_passed.is_empty() || !test_files_failed.is_empty() { + println!("\n📝 TEST FILES (should pass):"); + if !test_files_passed.is_empty() { + println!(" ✅ Passed ({}):", test_files_passed.len()); + for file in &test_files_passed { + println!(" - {}", file); + } + } + if !test_files_failed.is_empty() { + println!(" ❌ Failed ({}):", test_files_failed.len()); + for (file, error) in &test_files_failed { + println!("\n - {}", file); + println!("{}", error); + } + } + } + + // File fail_ (dovrebbero fallire) + if !fail_files_failed.is_empty() || !fail_files_passed.is_empty() { + println!("\n🚫 FAIL FILES (should fail):"); + if !fail_files_failed.is_empty() { + println!(" ✅ Failed as expected ({}):", fail_files_failed.len()); + for (file, error) in &fail_files_failed { + println!("\n - {}", file); + println!("{}", error); + } + } + if !fail_files_passed.is_empty() { + println!(" ❌ Passed unexpectedly ({}):", fail_files_passed.len()); + for file in &fail_files_passed { + println!(" - {} (should have failed but passed!)", file); + } + } + } + + println!("\n=================================="); + + // Riepilogo + let total_test_files = test_files_passed.len() + test_files_failed.len(); + let total_fail_files = fail_files_passed.len() + fail_files_failed.len(); + let total_files = total_test_files + total_fail_files; + + println!("Summary:"); + if total_test_files > 0 { + println!( + " test_ files: {}/{} passed", + test_files_passed.len(), + total_test_files + ); + } + if total_fail_files > 0 { + println!( + " fail_ files: {}/{} failed (as expected)", + fail_files_failed.len(), + total_fail_files + ); + } + println!(" Total files tested: {}", total_files); + + // Assicurati che almeno un file sia stato testato + assert!( + total_files > 0, + "No test_ or fail_ .lox files found in examples directory" + ); + + // Asserzioni per far fallire il test se le aspettative non sono rispettate + if !test_files_failed.is_empty() { + println!("\n❌ TEST FAILURE: The following test_ files failed but should have passed:"); + for (file, _) in &test_files_failed { + println!(" - {}", file); + } + panic!( + "{} test_ files failed unexpectedly", + test_files_failed.len() + ); + } + + if !fail_files_passed.is_empty() { + println!("\n❌ TEST FAILURE: The following fail_ files passed but should have failed:"); + for file in &fail_files_passed { + println!(" - {}", file); + } + panic!( + "{} fail_ files passed unexpectedly", + fail_files_passed.len() + ); + } + + println!( + "\n✅ All {} test/fail files behaved as expected!", + total_files + ); +} + +fn run_lox_file(source: String, _filename: &str) -> Result<(), String> { + // Crea un source registry e mantienilo per il pretty printing degli errori + let mut source_registry = SourceRegistry::new(); + + // Aggiungi il source con il nome del file corretto + let source_id = match source_registry.add_source_string(source.clone()) { + Ok(id) => id, + Err(e) => return Err(format!("Failed to add source: {}", e)), + }; + + // Tokenizza + let source_content = source_registry.get_by_id(source_id).content.clone(); + let mut lexer = Lexer::new(source_content, source_id); + let tokens = match lexer.scans_tokens() { + Ok(tokens) => tokens, + Err(err) => { + // Usa il pretty printing con source context + return Err(err.display_with_source(&source_registry)); + } + }; + + // Parsa + let ast = match Parser::new(tokens).parse() { + Ok(ast) => ast, + Err(err) => { + // Usa il pretty printing con source context + return Err(err.display_with_source(&source_registry)); + } + }; + + // Esegui + let mut interpreter = Interpreter::new(); + for stmt in ast.iter() { + match interpreter.evaluate(stmt.clone()) { + Ok(_) => {} + Err(err) => { + // Usa il pretty printing con source context + return Err(err.display_with_source(&source_registry)); + } + } + } + + Ok(()) +}