6 Commits
Author SHA1 Message Date
Giulio Agostini b93a1394cd Update parser.rs 2026-07-06 19:10:01 +02:00
Giulio Agostini 842216729b wip:
Add type system and refator for having only epression
2026-07-06 10:43:17 +02:00
Giulio Agostini 9f15a00b98 Add static variable resolution with scope tracking
- Add distance-based get_at on EnvironmentStack
- Add distance-based assign_at on EnvironmentStack
- Introduce ErrorSink to accumulate diagnostics across passes
- Implement Resolver with a ScopeStack and per-node distance map
- Extend interpreter to store locals distances for runtime lookup
- Add tests for resolution behavior and error accumulation
  Add static variable resolution with scope tracking
2026-06-30 15:05:34 +02:00
Giulio Agostini d40fe2a550 Introduce NodeId for AST and synthetic slices 2026-06-30 14:05:46 +02:00
Giulio Agostini ef8abda048 Move variable resolution to middleend
- Introduce AST-based visitor and resolver in middleend
- Update guard field and parser to use AstNode<Expr>
- Remove backend variable_resolution and adjust exports
- Expose middleend in the library
2026-06-30 09:47:22 +02:00
Giulio Agostini 29c86d278d Make environment generic and adapt closures
Added unit test
2026-06-29 20:50:29 +02:00
29 changed files with 1978 additions and 1340 deletions
Executable → Regular
View File
Executable
+36
View File
@@ -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
+6 -6
View File
@@ -1,12 +1,12 @@
closure :: fn() do
closure :: fn(): Any do
value := "Closure";
internal :: fn() do
internal :: fn(): String do
print value;
return value;
end
end;
return internal;
end
end;
internal := closure();
internal();
@@ -17,9 +17,9 @@ c := b;
c = b = a;
d := c = b = a;
do
showA :: fn() do
showA :: fn(): Nil do
print a;
end
end;
showA();
a := "Local";
+3 -3
View File
@@ -4,7 +4,7 @@
// 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;
print param2;
while param < param2 do
@@ -26,6 +26,6 @@ func_name :: fn(param: Int, param2: String) do
end
return False or True;
end
end;
func_name(1,20)
func_name(1,20);
+14 -14
View File
@@ -1,30 +1,30 @@
function :: fn() do
function :: fn(): Nil do
print "Function";
end
end;
function_factory :: fn() do
function_factory :: fn(): Any do
print "Function Factory";
function
end
function;
end;
function_fatcoty_factory :: fn() do
function_fatcoty_factory :: fn(): Any do
print "Function Factory Factory";
return function_factory;
end
end;
//function();
//function_factory()();
//function_fatcoty_factory()()();
clock()
clock();
closure :: fn() do
value = "Closure";
internal :: fn() do
closure :: fn(): Any do
value := "Closure";
internal :: fn(): Nil do
print value;
end
end;
return internal;
end
end;
closure()
closure();
+57
View File
@@ -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 )* ")"
+34 -2
View File
@@ -48,7 +48,7 @@ impl<T: Clone + Debug + PartialEq> EnvironmentStack<T> {
}
}
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),
)
}
@@ -58,6 +58,38 @@ impl<T: Clone + Debug + PartialEq> EnvironmentStack<T> {
Ok(value)
}
/// 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<T> {
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<T> {
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<T> {
let size = self.stack.len();
@@ -75,7 +107,7 @@ impl<T: Clone + Debug + PartialEq> EnvironmentStack<T> {
}
}
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),
)
}
+171 -175
View File
@@ -1,79 +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<BaseValue>,
/// 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<NodeId, usize>,
}
pub trait EvaluateInterpreter<T> {
fn evaluate(&mut self, stmt: T) -> LoxResult<BaseValue>;
}
impl<R: AstNodeKind + Clone + Debug + Display> EvaluateInterpreter<AstNode<R>> for Interpreter
where
Interpreter: EvaluateInterpreter<R>,
{
fn evaluate(&mut self, stmt: AstNode<R>) -> LoxResult<BaseValue> {
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<Expr> for Interpreter {
fn evaluate(&mut self, expr: Expr) -> LoxResult<BaseValue> {
match expr {
Expr::Literal { value } => match value {
BaseValue::Function(mut func) => {
func.closure = Some(self.enviorment.clone());
Ok(BaseValue::Function(func))
}
_ => 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<AstNode<Stmt>> for Interpreter {
fn evaluate(&mut self, node: AstNode<Stmt>) -> LoxResult<BaseValue> {
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<AstNode> for Interpreter {
fn evaluate(&mut self, node: AstNode) -> LoxResult<BaseValue> {
self.eval_expr(&node)
}
}
@@ -91,27 +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<NodeId, usize>) {
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<BaseValue> {
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<BaseValue> {
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<AstNode<Expr>>,
arguments: Vec<AstNode<Expr>>,
name: &str,
id: NodeId,
value: BaseValue,
) -> LoxResult<BaseValue> {
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<BaseValue> {
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() {
@@ -120,7 +193,7 @@ impl Interpreter {
let evaluated_arguments = arguments
.iter()
.map(|arg| self.evaluate(arg.clone()))
.map(|arg| self.eval_expr(arg))
.collect::<Result<Vec<BaseValue>, LoxError>>()?;
match function {
@@ -142,7 +215,7 @@ impl Interpreter {
}
// Execute the function body
let result = match self.evaluate(func.body) {
let result = match self.eval_expr(&func.body) {
Ok(value) => Ok(value),
Err(LoxError::Return { value, .. }) => Ok(value),
Err(err) => Err(err),
@@ -180,7 +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),
}),
}
@@ -191,114 +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<BaseValue> {
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::Return { expression, .. } => self.evaluate(*expression),
Stmt::VarDeclaration {
name, initializer, ..
} => {
let value = match initializer {
Some(expr_node) => 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, ..
} => {
let mut ret = BaseValue::Nil;
while self.evaluate(*condition.clone())?.is_truthy() {
match self.evaluate(*body.clone()) {
Ok(val) => ret = val,
Err(LoxError::Return { value, .. }) => {
ret = value;
break;
}
Err(err) => return Err(err),
};
}
Ok(ret)
}
Stmt::For {
variable,
condition,
increment,
body,
..
} => {
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() {
match self.evaluate(*body.clone()) {
Ok(val) => ret = val,
Err(LoxError::Return { value, .. }) => {
ret = value;
break;
}
Err(err) => return Err(err),
};
self.evaluate(*increment.clone())?;
}
Ok(ret)
}
}
}
fn evaluate_if(
&mut self,
condition: Box<AstNode<Expr>>,
then_branch: Box<AstNode<Stmt>>,
elif_branch: Vec<(Box<AstNode<Expr>>, Box<AstNode<Stmt>>)>,
else_branch: Option<Box<AstNode<Stmt>>>,
condition: &AstNode,
then_branch: &AstNode,
elif_branch: &[(Box<AstNode>, Box<AstNode>)],
else_branch: &Option<Box<AstNode>>,
) -> LoxResult<BaseValue> {
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(),
});
@@ -306,38 +304,36 @@ 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<AstNode<Stmt>>) -> LoxResult<BaseValue> {
fn evaluate_block(&mut self, statements: &[AstNode]) -> LoxResult<BaseValue> {
self.enviorment.push_new_scope();
// Ora elements è sempre disponibile
let mut result = Ok(BaseValue::Nil);
for statement in statements.iter() {
let node = statement.node.clone();
match node {
Stmt::Return { expression, .. } => {
let value = self.evaluate(*expression)?;
match &statement.node {
Expr::Return { expression, .. } => {
let value = self.eval_expr(expression)?;
result = Err(LoxError::Return {
source_slice: statement.source_slice.clone(),
value: value,
value,
return_label: "Hi".to_string(),
});
break;
}
_ => result = self.evaluate((*statement).clone()),
_ => result = self.eval_expr(statement),
};
}
self.enviorment.pop_scope();
@@ -460,7 +456,7 @@ mod tests {
#[test]
fn defines_and_calls_a_function() {
let src = "add :: fn (a, b) do return a + b; end add(2, 3);";
let src = "add :: fn (a, b): Number do return a + b; end; add(2, 3);";
assert_eq!(eval(src).unwrap().to_string(), "5");
}
+195 -327
View File
@@ -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<BaseValue>,
},
Binary {
left: Box<AstNode<Expr>>,
left: Box<AstNode>,
operator: TokenType,
right: Box<AstNode<Expr>>,
right: Box<AstNode>,
},
Unary {
operator: TokenType,
operand: Box<AstNode<Expr>>,
operand: Box<AstNode>,
},
Grouping {
expression: Box<AstNode<Expr>>,
expression: Box<AstNode>,
},
Identifier {
name: String,
},
Call {
callee: Box<AstNode<Expr>>,
arguments: Vec<AstNode<Expr>>,
callee: Box<AstNode>,
arguments: Vec<AstNode>,
},
Assign {
name: String,
value: Box<AstNode>,
},
Print {
expression: Box<AstNode>,
},
VarDeclaration {
name: String,
var_type: Type,
initializer: Option<Box<AstNode>>,
},
Return {
expression: Box<AstNode>,
label: String,
},
Block {
statements: Box<Vec<AstNode>>,
label: String,
},
If {
condition: Box<AstNode>,
then_branch: Box<AstNode>,
elif_branches: Vec<(Box<AstNode>, Box<AstNode>)>,
else_branch: Option<Box<AstNode>>,
},
While {
condition: Box<AstNode>,
body: Box<AstNode>,
},
For {
variable: Box<AstNode>,
condition: Box<AstNode>,
increment: Box<AstNode>,
body: Box<AstNode>,
},
}
@@ -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 ({}({}))",
@@ -86,6 +110,60 @@ impl std::fmt::Display for Expr {
.collect::<Vec<String>>()
.join(", ")
),
Expr::Assign { name, value } => write!(f, "Assign ({} = {})", name, value.node),
Expr::If {
condition,
then_branch,
elif_branches,
else_branch,
} => {
let mut result = 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,
));
}
if let Some(else_branch) = else_branch {
result.push_str(&format!(" ELSE {{\n{}\n}}", else_branch.node));
}
write!(f, "IfStmt {}", result)
}
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),
},
Expr::Return { expression, label } => {
write!(f, "Return({}, {})", expression.node, label)
}
Expr::Block { statements, label } => write!(
f,
"Block [{}] ([\n{}\n])",
label,
statements
.iter()
.map(|stmt| format!("\t \t{}", stmt.node))
.collect::<Vec<_>>()
.join("\n"),
),
Expr::While { condition, body } => {
write!(f, "While({}) {{\n{}\n}}", condition.node, body.node)
}
Expr::For {
variable,
condition,
increment,
body,
} => write!(
f,
"For({} = {} in {}) {{\n{}\n}}",
variable.node, condition.node, increment.node, body.node
),
}
}
}
@@ -93,15 +171,21 @@ impl std::fmt::Display for Expr {
impl Debug for Expr {
fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
match self {
Expr::Literal { value } => write!(f, "{:?}", value),
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),
} => 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 ({}({}))",
@@ -110,320 +194,90 @@ impl Debug for Expr {
.iter()
.map(|arg| arg.node.to_string())
.collect::<Vec<String>>()
.join(", ")
.join(", "),
),
}
}
}
#[derive(Clone, PartialEq)]
pub enum Stmt {
Expression {
expression: Box<AstNode<Expr>>,
return_value: Box<BaseValue>,
},
Print {
expression: Box<AstNode<Expr>>,
return_value: Box<BaseValue>,
},
VarDeclaration {
name: String,
initializer: Option<Box<AstNode<Expr>>>,
return_value: Box<BaseValue>,
},
VarAssigment {
name: String,
value: Box<AstNode<Expr>>,
return_value: Box<BaseValue>,
},
Return {
expression: Box<AstNode<Expr>>,
label: String,
return_value: Box<BaseValue>,
},
Block {
statements: Box<Vec<AstNode<Stmt>>>,
label: String,
return_value: Box<BaseValue>,
},
If {
condition: Box<AstNode<Expr>>,
then_branch: Box<AstNode<Stmt>>,
elif_branch: Vec<(Box<AstNode<Expr>>, Box<AstNode<Stmt>>)>,
else_branch: Option<Box<AstNode<Stmt>>>,
return_value: Box<BaseValue>,
},
While {
condition: Box<AstNode<Expr>>,
body: Box<AstNode<Stmt>>,
return_value: Box<BaseValue>,
},
For {
variable: Box<AstNode<Stmt>>,
condition: Box<AstNode<Expr>>,
increment: Box<AstNode<Stmt>>,
body: Box<AstNode<Stmt>>,
return_value: Box<BaseValue>,
},
}
impl Display for Stmt {
fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
match self {
Stmt::Expression {
expression,
return_value,
} => write!(f, "Expression ({}) -> {:?}", expression.node, return_value),
Stmt::If {
Expr::Assign { name, value } => {
write!(f, "(assign {:?} {:?})", name, value.node,)
}
Expr::If {
condition,
then_branch,
elif_branch,
elif_branches,
else_branch,
return_value,
} => {
let mut result = format!("IF ({}) {{\n{}\n}}", condition.node, then_branch.node);
for (condition, branch) in elif_branch {
let mut result =
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, return_value
" ELIF ({:?}) {{\n{:?}\n}}",
condition.node, branch.node,
));
}
if let Some(else_branch) = else_branch {
result.push_str(&format!(
" ELSE {{\n{}\n}} -> {:?}",
else_branch.node, return_value
));
}
write!(f, "IfStmt {}", result)
}
Stmt::Print {
expression,
return_value,
} => write!(f, "Print({}) -> {:?};", expression.node, return_value),
Stmt::VarDeclaration {
name,
initializer,
return_value,
} => match initializer {
Some(init) => write!(f, "Var({} = {}) -> {:?};", name, init.node, return_value),
None => write!(f, "Var({}) -> {:?};", name, return_value),
},
Stmt::VarAssigment {
name,
value,
return_value,
} => write!(
f,
"Assign({} = {}) -> {:?};",
name, value.node, return_value
),
Stmt::Return {
expression,
return_value,
label,
} => write!(
f,
"Return({}, {}) -> {:?};",
expression.node, label, return_value
),
Stmt::Block {
statements,
label,
return_value,
} => write!(
f,
"Block [{}] ([\n{}\n]) -> {:?}",
label,
statements
.iter()
.map(|stmt| format!("\t \t{}", stmt.node))
.collect::<Vec<_>>()
.join("\n"),
return_value
),
Stmt::While {
condition,
body,
return_value,
} => {
write!(
f,
"While({}) {{\n{}\n}} -> {:?}",
condition.node, body.node, return_value
)
}
Stmt::For {
variable,
condition,
increment,
body,
return_value,
} => write!(
f,
"For({} = {} in {}) {{\n{}\n}} -> {:?}",
variable.node, condition.node, increment.node, body.node, return_value
),
}
}
}
impl Debug for Stmt {
fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
match self {
Stmt::Expression {
expression,
return_value,
} => write!(
f,
" Expression ({:?}) -> {:?}",
expression.node, return_value
),
Stmt::If {
condition,
then_branch,
elif_branch,
else_branch,
return_value,
} => {
let mut result = format!(
"IF ({:?}) {{\n{:?}\n}} -> {:?} ",
condition.node, then_branch.node, return_value
);
for (condition, branch) in elif_branch {
result.push_str(&format!(
" ELIF ({:?}) {{\n{:?}\n}} -> {:?} ",
condition.node, branch.node, return_value
));
}
if let Some(else_branch) = else_branch {
result.push_str(&format!(
" ELSE {{\n{:?}\n}} -> {:?}",
else_branch.node, return_value
));
result.push_str(&format!(" ELSE {{\n{:?}\n}}", else_branch.node,));
}
write!(f, "IfStmt {:?}", result)
}
Stmt::Print {
expression,
return_value,
} => write!(f, "Print({:?}) -> {:?};", expression.node, return_value),
Stmt::VarDeclaration {
Expr::Print { expression } => {
write!(f, "Print({:?});", expression.node,)
}
Expr::VarDeclaration {
name,
initializer,
return_value,
var_type,
} => match initializer {
Some(init) => write!(
f,
"Var({:?} = {:?}) -> {:?};",
name, init.node, return_value
),
None => write!(f, "Var({:?}) -> {:?};", name, return_value),
Some(init) => write!(f, "Var({:?}: {:?} = {:?});", name, var_type, init.node,),
None => write!(f, "Var({:?}: {:?});", name, var_type,),
},
Stmt::VarAssigment {
name,
value,
return_value,
} => {
write!(
f,
"Assign({:?} = {:?}) -> {:?};",
name, value.node, return_value
)
Expr::Return { expression, label } => {
write!(f, "Return({:?}, {}) ;", expression.node, label,)
}
Stmt::Return {
expression,
return_value,
label,
} => write!(
Expr::Block { label, statements } => write!(
f,
"Return({:?}, {}) -> {:?};",
expression.node, label, return_value
),
Stmt::Block {
label,
statements,
return_value,
} => write!(
f,
"Block [{}] ([\n{:?}\n]) -> {:?}",
"Block [{}] ([\n{:?}\n])",
label,
statements
.iter()
.map(|stmt| format!("{:?}", stmt.node))
.collect::<Vec<_>>()
.join("\t\t\n"),
return_value
),
Stmt::While {
condition,
body,
return_value,
} => {
write!(
f,
"While({:?}) {{\n{:?}\n}} -> {:?}",
condition.node, body.node, return_value
)
Expr::While { condition, body } => {
write!(f, "While({:?}) {{\n{:?}\n}}", condition.node, body.node,)
}
Stmt::For {
Expr::For {
variable,
condition,
increment,
body,
return_value,
} => write!(
f,
"For({:?} = {:?} in {:?}) {{\n{:?}\n}} -> {:?}",
variable.node, condition.node, increment.node, body.node, return_value
"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::While { .. } => "while statement",
Stmt::For { .. } => "for statement",
}
}
}
#[derive(Clone, PartialEq, Default)]
pub struct AstNode<T: AstNodeKind + Debug + Display> {
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<T: AstNodeKind + Debug + Display> Display for AstNode<T> {
// 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,
@@ -433,7 +287,7 @@ impl<T: AstNodeKind + Debug + Display> Display for AstNode<T> {
}
}
impl<T: AstNodeKind + Debug + Display> Debug for AstNode<T> {
impl Debug for AstNode {
fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
write!(
f,
@@ -443,8 +297,22 @@ impl<T: AstNodeKind + Debug + Display> Debug for AstNode<T> {
}
}
impl<T: AstNodeKind + Debug + Display> AstNode<T> {
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)
}
}
+35 -24
View File
@@ -1,12 +1,12 @@
use core::fmt;
use std::collections::HashMap;
use std::fmt::Display;
use std::ops::{Add, Div, Mul, Not, Rem, Sub};
use crate::backend::environment::Environment;
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)]
@@ -272,15 +272,20 @@ pub enum BaseValue {
Nil,
Function(LoxFunction),
NativeFunction(NativeFunction),
Struct(Struct),
}
struct Dict {
map: HashMap<String, BaseValue>,
}
struct ReturnValue {
label: Vec<String>,
value: BaseValue,
value: Type,
}
impl ReturnValue {
pub fn new(label: Vec<String>, value: BaseValue) -> Self {
pub fn new(label: Vec<String>, value: Type) -> Self {
Self { label, value }
}
}
@@ -304,39 +309,40 @@ impl Display for BaseValue {
BaseValue::Nil => write!(f, "nil"),
BaseValue::Function(..) => write!(f, "<fn lox function>"),
BaseValue::NativeFunction(..) => write!(f, "<native native function>"),
BaseValue::Struct(_) => write!(f, "<struct>"),
}
}
}
#[derive(Debug, Clone, PartialEq)]
pub struct LoxFunction {
pub parameters: Vec<(String, String)>,
pub return_type: Option<String>,
pub body: AstNode<Stmt>,
pub parameters: Vec<(String, Type)>,
pub return_type: Type,
pub body: AstNode,
pub closure: Option<EnvironmentStack<BaseValue>>,
pub guard: Option<Box<AstNode<Expr>>>,
pub guard: Option<Box<AstNode>>,
}
impl LoxFunction {
pub fn new(
parameters: Vec<(String, String)>,
body: AstNode<Stmt>,
parameters: Vec<(String, Type)>,
body: AstNode,
closure: Option<EnvironmentStack<BaseValue>>,
guard: Option<Box<AstNode<Expr>>>,
guard: Option<Box<AstNode>>,
) -> Self {
LoxFunction {
parameters,
return_type: None,
return_type: Type::Any,
body,
closure,
guard,
}
}
pub fn anonymous_function(parameters: Vec<(String, String)>, body: AstNode<Stmt>) -> Self {
pub fn anonymous_function(parameters: Vec<(String, Type)>, body: AstNode) -> Self {
LoxFunction {
parameters,
return_type: None,
return_type: Type::Any,
body,
closure: None,
guard: None,
@@ -346,12 +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,
@@ -359,6 +365,11 @@ impl NativeFunction {
}
}
#[derive(Debug, Clone, PartialEq)]
pub struct Struct {
pub fields: Vec<(String, BaseValue)>,
}
// Trait implementations for BaseValue
pub trait Truthy {
@@ -398,7 +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(),
),
}
@@ -428,7 +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(),
),
}
@@ -442,10 +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(),
),
}
@@ -459,7 +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(),
),
}
@@ -473,10 +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(),
),
}
+87
View File
@@ -479,3 +479,90 @@ pub fn io_error<T>(message: impl Into<String>) -> LoxResult<T> {
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<LoxError>,
}
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<LoxError> {
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);
}
}
+1
View File
@@ -1,3 +1,4 @@
pub mod ast;
pub mod base_value;
pub mod lox_result;
pub mod types;
+70
View File
@@ -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<StructType>),
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<Type>)>,
pub return_type: Box<Type>,
}
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::<Vec<_>>()
.join(", ");
write!(f, "fn({}) -> {}", rendered_params, self.return_type)
}
}
#[derive(Debug, Clone, PartialEq)]
pub struct StructType {
pub fields: Vec<(String, Box<Type>)>,
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::<Vec<_>>()
.join(", ");
let rendered_method = self
.methods
.iter()
.map(|(name, ty)| format!("{}: {}", name, ty))
.collect::<Vec<_>>()
.join(", ");
write!(f, "struct {{ {}, {} }}", rendered_field, rendered_method)
}
}
+74 -41
View File
@@ -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<TokenType> {
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),
@@ -48,28 +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<Vec<Token>> {
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),
@@ -82,32 +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,
@@ -119,7 +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,
@@ -179,9 +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() {
@@ -201,10 +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(),
@@ -236,7 +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()),
)))
}
@@ -266,7 +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 {
@@ -303,7 +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)),
@@ -509,4 +493,53 @@ mod tests {
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<TokenType> = 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");
}
}
-1
View File
@@ -2,4 +2,3 @@ pub mod lexer;
pub mod parser;
pub mod source_registry;
pub mod tokens;
pub mod variable_resolution;
+293 -293
View File
File diff suppressed because it is too large Load Diff
+29 -2
View File
@@ -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,
+4 -2
View File
@@ -35,11 +35,12 @@ pub enum TokenType {
Identifier,
String,
Number,
Atom,
// Keywords
Fn,
And,
Class,
Struct,
StartBlock,
EndBlock,
False,
@@ -80,8 +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"),
-204
View File
@@ -1,204 +0,0 @@
use crate::{
backend::environment::EnvironmentStack,
common::{
ast::{AstNode, AstNodeKind, Expr, Stmt},
lox_result::{runtime_error, LoxError, LoxResult},
},
frontend::source_registry::SourceSlice,
};
use std::fmt::{Debug, Display};
struct Resolver {
scopes: EnvironmentStack<bool>,
}
impl Resolver {
pub fn new() -> Self {
Resolver {
scopes: EnvironmentStack::new(),
}
}
fn declare(&mut self, name: &String) {
if self.scopes.is_empty() {
return;
}
let _ = self.scopes.set(name.clone(), false);
}
fn define(&mut self, name: &String) {
if self.scopes.is_empty() {
return;
}
let _ = self.scopes.set(name.clone(), true);
}
}
pub trait StaticAnalyzer<T> {
fn resolve(&mut self, node: &T) -> LoxResult<bool>;
}
impl<R: AstNodeKind + Clone + Debug + Display> StaticAnalyzer<AstNode<R>> for Resolver
where
Resolver: StaticAnalyzer<R>,
{
fn resolve(&mut self, node: &AstNode<R>) -> LoxResult<bool> {
self.resolve(&node.node)
}
}
impl StaticAnalyzer<Stmt> for Resolver {
fn resolve(&mut self, node: &Stmt) -> LoxResult<bool> {
match node {
Stmt::Expression { expression, .. } => {
// Visit the expression
self.resolve(expression.as_ref())
}
Stmt::Print { expression, .. } => {
// Visit the expression
self.resolve(expression.as_ref())
}
Stmt::VarDeclaration {
initializer, name, ..
} => {
self.declare(name);
// Visit the initializer if present
if let Some(init) = initializer {
self.resolve(init.as_ref())?;
}
self.define(name);
Ok(true)
}
Stmt::VarAssigment { value, name, .. } => {
match self.scopes.get(name) {
Ok(true) => (),
Ok(false) => {
return Err(LoxError::RuntimeError {
source_slice: SourceSlice::default(),
message: "Cant read loac variable in it own lintilizer".to_string(),
})
}
Err(err) => return Err(err),
}
self.resolve(value.as_ref())
}
Stmt::Return { expression, .. } => {
// Visit the return expression
self.resolve(expression.as_ref())
}
Stmt::Block { statements, .. } => {
self.scopes.push_new_scope();
// Visit all statements in the block
for stmt in statements.iter() {
self.resolve(stmt)?;
}
self.scopes.pop_scope();
Ok(true)
}
Stmt::If {
condition,
then_branch,
elif_branch,
else_branch,
..
} => {
// Visit the condition
self.resolve(condition.as_ref())?;
// Visit the then branch
self.resolve(then_branch.as_ref())?;
// Visit all elif branches
for (elif_condition, elif_stmt) in elif_branch.iter() {
self.resolve(elif_condition.as_ref())?;
self.resolve(elif_stmt.as_ref())?;
}
// Visit the else branch if present
if let Some(else_stmt) = else_branch {
self.resolve(else_stmt.as_ref())?;
}
Ok(true)
}
Stmt::While {
condition, body, ..
} => {
// Visit the condition
self.resolve(condition.as_ref())?;
// Visit the body
self.resolve(body.as_ref())?;
Ok(true)
}
Stmt::For {
variable,
condition,
increment,
body,
..
} => {
// Visit the variable initialization
self.resolve(variable.as_ref())?;
// Visit the condition
self.resolve(condition.as_ref())?;
// Visit the increment
self.resolve(increment.as_ref())?;
// Visit the body
self.resolve(body.as_ref())?;
Ok(true)
}
}
}
}
impl StaticAnalyzer<Expr> for Resolver {
fn resolve(&mut self, node: &Expr) -> LoxResult<bool> {
match node {
Expr::Literal { .. } => {
// Leaf node - no children to visit
Ok(true)
}
Expr::Binary { left, right, .. } => {
// Visit left operand
self.resolve(left.as_ref())?;
// Visit right operand
self.resolve(right.as_ref())
}
Expr::Unary { operand, .. } => {
// Visit the operand
self.resolve(operand.as_ref())
}
Expr::Grouping { expression } => {
// Visit the grouped expression
self.resolve(expression.as_ref())
}
Expr::Identifier { name, .. } => {
if !self.scopes.is_empty() && self.scopes.get(name).is_ok() {
return Err(LoxError::ParseError {
source_slice: SourceSlice::default(),
message: "Cant read local varialbe in it own initializer".to_string(),
});
}
Ok(true)
}
Expr::Call {
callee, arguments, ..
} => {
// Visit the callee
self.resolve(callee.as_ref())?;
// Visit all arguments
for arg in arguments.iter() {
self.resolve(arg)?;
}
Ok(true)
}
}
}
}
+1
View File
@@ -1,4 +1,5 @@
pub mod backend;
pub mod common;
pub mod frontend;
pub mod logging;
pub mod middleend;
+30 -80
View File
@@ -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 +277,7 @@ impl PrettyPrint for Stmt {
write!(f, "{}}}", indent)
}
}
Stmt::VarDeclaration {
Expr::VarDeclaration {
name, initializer, ..
} => {
if ctx.config.compact {
@@ -317,21 +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());
@@ -344,9 +318,7 @@ impl PrettyPrint for Stmt {
write!(f, "{}}}", indent)
}
}
Stmt::Block {
statements, label, ..
} => {
Expr::Block { statements, label } => {
if ctx.config.compact {
write!(f, "{{ ... [{}] ({} statements) }}", label, statements.len())
} else {
@@ -365,12 +337,11 @@ impl PrettyPrint for Stmt {
write!(f, "{}}}", indent)
}
}
Stmt::If {
Expr::If {
condition,
then_branch,
elif_branch,
elif_branches,
else_branch,
..
} => {
if ctx.config.compact {
let cond_str =
@@ -385,10 +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())?;
@@ -420,21 +391,18 @@ impl PrettyPrint for Stmt {
write!(f, "{}}}", indent)
}
}
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,
body,
..
} => {
write!(f, "{}for (", ctx.child_context(true).indent())?;
variable.pretty_print(&ctx.child_context(true), f)?;
@@ -450,7 +418,7 @@ impl PrettyPrint for Stmt {
}
}
impl<T: AstNodeKind + fmt::Debug + fmt::Display + PrettyPrint> PrettyPrint for AstNode<T> {
impl PrettyPrint for AstNode {
fn pretty_print(&self, ctx: &PrettyContext, f: &mut fmt::Formatter) -> fmt::Result {
let indent = ctx.indent();
@@ -462,9 +430,9 @@ impl<T: AstNodeKind + fmt::Debug + fmt::Display + PrettyPrint> PrettyPrint for A
if ctx.config.show_types {
writeln!(
f,
"{}kind: {:?},",
"{}type: {},",
ctx.child_context(false).indent(),
self.node.kind()
self.return_type
)?;
}
@@ -518,25 +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<T: AstNodeKind + fmt::Debug + fmt::Display + PrettyPrint> PrettyPrintExt for AstNode<T> {
impl PrettyPrintExt for AstNode {
fn pretty(&self) -> String {
pretty_print(self)
}
+4 -2
View File
@@ -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",
@@ -219,7 +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
@@ -239,6 +240,7 @@ impl Token {
| 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())
+30 -11
View File
@@ -1,13 +1,12 @@
mod backend;
mod common;
mod frontend;
mod logging;
mod middleend;
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::{
@@ -15,6 +14,7 @@ use crate::{
parser::Parser,
source_registry::{SourceId, SourceRegistry},
},
middleend::variable_resolution::Resolver,
};
use std::env;
use std::fs;
@@ -174,7 +174,7 @@ impl LoxInterpreter {
}
// Funzione per parsare fino all'AST
fn parse_to_ast(&self, source_id: SourceId) -> LoxResult<Vec<AstNode<Stmt>>> {
fn parse_to_ast(&self, source_id: SourceId) -> LoxResult<Vec<AstNode>> {
let tokens = self.tokenize(source_id)?;
Parser::new(tokens).parse().or_else(|err| {
@@ -184,8 +184,12 @@ impl LoxInterpreter {
}
// Funzione per eseguire l'AST
fn execute_ast(&self, ast: Vec<AstNode<Stmt>>, debug: bool) -> LoxResult<String> {
fn execute_ast(&self, ast: Vec<AstNode>, debug: bool) -> LoxResult<String> {
// 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() {
@@ -204,6 +208,19 @@ impl LoxInterpreter {
}
}
// Static variable resolution; prints and returns the first error, if any.
fn resolve(
&self,
ast: &[AstNode],
) -> LoxResult<std::collections::HashMap<crate::common::ast::NodeId, usize>> {
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,
@@ -245,8 +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() {
@@ -275,7 +294,7 @@ fn format_tokens(tokens: &[crate::frontend::tokens::Token]) -> String {
.join("\n")
}
fn format_ast(ast: &[AstNode<Stmt>]) -> String {
fn format_ast(ast: &[AstNode]) -> String {
use crate::logging::display_ast::{pretty_print_with_config, PrettyConfig};
let config = PrettyConfig {
+1
View File
@@ -1,2 +1,3 @@
pub mod scope_stack;
pub mod variable_resolution;
pub mod visit_ast;
+146
View File
@@ -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<T> {
scopes: Vec<HashMap<String, T>>,
}
impl<T> Default for ScopeStack<T> {
fn default() -> Self {
Self::new()
}
}
impl<T> ScopeStack<T> {
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<String>, 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<usize> {
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<bool> = 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<bool> = 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<bool> = 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<bool> = 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<bool> = 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);
}
}
+221 -62
View File
@@ -1,99 +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::{
backend::environment::EnvironmentStack,
common::{
ast::{AstNode, Expr, Stmt},
lox_result::{LoxError, LoxResult},
ast::{AstNode, Expr, NodeId},
base_value::{BaseValue, LoxFunction},
lox_result::{ErrorSink, LoxError, LoxResult},
},
middleend::{
scope_stack::ScopeStack,
visit_ast::{walk_expr, walk_function, Visitor},
},
frontend::source_registry::SourceSlice,
middleend::visit_ast::{walk_expr, walk_stmt, Visitor},
};
struct Resolver {
scopes: EnvironmentStack<bool>,
locals: HashMap<SourceSlice, usize>,
/// 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<bool>,
locals: HashMap<NodeId, usize>,
errors: ErrorSink,
current_function: FunctionType,
}
impl Resolver {
pub fn new() -> Self {
Resolver {
scopes: EnvironmentStack::new(),
scopes: ScopeStack::new(),
locals: HashMap::new(),
errors: ErrorSink::new(),
current_function: FunctionType::None,
}
}
fn declare(&mut self, name: &String) {
/// Resolve a whole program, returning the per-reference scope distances or
/// the accumulated resolution errors.
pub fn resolve_program(
statements: &[AstNode],
) -> Result<HashMap<NodeId, usize>, Vec<LoxError>> {
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;
}
let _ = self.scopes.set(name.clone(), false);
self.scopes.set_local(name, true);
}
fn define(&mut self, name: &String) {
if self.scopes.is_empty() {
return;
fn resolve_local(&mut self, id: NodeId, name: &str) {
if let Some(distance) = self.scopes.resolve(name) {
self.locals.insert(id, distance);
}
let _ = self.scopes.set(name.clone(), true);
// Not found locally: assume global, record nothing.
}
}
fn resolve_local(&mut self, name: &String) {
let depth = self.scopes.depth();
for i in (0..depth).rev() {
if self.scopes.scope_contains(i, name) {
self.locals.insert(, i);
}
}
impl Default for Resolver {
fn default() -> Self {
Self::new()
}
}
impl Visitor for Resolver {
fn visit_stmt(&mut self, stmt: &AstNode<Stmt>) -> LoxResult<()> {
match &stmt.node {
Stmt::VarDeclaration { name, .. } => {
self.declare(name);
// `walk_stmt` resolves the initializer (if present).
walk_stmt(self, stmt)?;
self.define(name);
Ok(())
}
Stmt::VarAssigment { name, .. } => {
match self.scopes.get(name) {
Ok(true) => (),
Ok(false) => {
return Err(LoxError::RuntimeError {
source_slice: SourceSlice::default(),
message: "Cant read local variable in it own lintilizer".to_string(),
})
}
Err(err) => return Err(err),
}
// `walk_stmt` resolves the assigned value.
walk_stmt(self, stmt)
}
Stmt::Block { .. } => {
self.scopes.push_new_scope();
walk_stmt(self, stmt)?;
self.scopes.pop_scope();
Ok(())
}
// Expression, Print, Return, If, While, For: default traversal.
_ => walk_stmt(self, stmt),
}
}
fn visit_expr(&mut self, expr: &AstNode<Expr>) -> LoxResult<()> {
fn visit_expr(&mut self, expr: &AstNode) -> LoxResult<()> {
match &expr.node {
Expr::Identifier { name, .. } => {
if !self.scopes.is_empty() && self.scopes.get(name).is_ok() {
return Err(LoxError::ParseError {
source_slice: SourceSlice::default(),
message: "Cant read local varialbe in it own initializer".to_string(),
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, &param_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<AstNode> {
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<HashMap<NodeId, usize>, Vec<LoxError>> {
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());
}
}
+71 -91
View File
@@ -23,7 +23,7 @@
//! struct IdentifierCounter { count: usize }
//!
//! impl Visitor for IdentifierCounter {
//! fn visit_expr(&mut self, expr: &AstNode<Expr>) -> LoxResult<()> {
//! fn visit_expr(&mut self, expr: &AstNode) -> LoxResult<()> {
//! if let Expr::Identifier { .. } = &expr.node {
//! self.count += 1;
//! }
@@ -33,7 +33,7 @@
//! ```
use crate::common::{
ast::{AstNode, Expr, Stmt},
ast::{AstNode, Expr},
base_value::{BaseValue, LoxFunction},
lox_result::LoxResult,
};
@@ -43,13 +43,8 @@ use crate::common::{
/// 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 a statement node. Defaults to [`walk_stmt`].
fn visit_stmt(&mut self, stmt: &AstNode<Stmt>) -> LoxResult<()> {
walk_stmt(self, stmt)
}
/// Visit an expression node. Defaults to [`walk_expr`].
fn visit_expr(&mut self, expr: &AstNode<Expr>) -> LoxResult<()> {
fn visit_expr(&mut self, expr: &AstNode) -> LoxResult<()> {
walk_expr(self, expr)
}
@@ -62,71 +57,13 @@ pub trait Visitor: Sized {
}
}
/// Recurse into the children of `stmt`, calling back into `visitor`.
pub fn walk_stmt<V: Visitor>(visitor: &mut V, stmt: &AstNode<Stmt>) -> LoxResult<()> {
match &stmt.node {
Stmt::Expression { expression, .. } => visitor.visit_expr(expression),
Stmt::Print { expression, .. } => visitor.visit_expr(expression),
Stmt::VarDeclaration { initializer, .. } => {
if let Some(initializer) = initializer {
visitor.visit_expr(initializer)?;
}
Ok(())
}
Stmt::VarAssigment { value, .. } => visitor.visit_expr(value),
Stmt::Return { expression, .. } => visitor.visit_expr(expression),
Stmt::Block { statements, .. } => {
for statement in statements.iter() {
visitor.visit_stmt(statement)?;
}
Ok(())
}
Stmt::If {
condition,
then_branch,
elif_branch,
else_branch,
..
} => {
visitor.visit_expr(condition)?;
visitor.visit_stmt(then_branch)?;
for (elif_condition, elif_body) in elif_branch.iter() {
visitor.visit_expr(elif_condition)?;
visitor.visit_stmt(elif_body)?;
}
if let Some(else_body) = else_branch {
visitor.visit_stmt(else_body)?;
}
Ok(())
}
Stmt::While {
condition, body, ..
} => {
visitor.visit_expr(condition)?;
visitor.visit_stmt(body)
}
Stmt::For {
variable,
condition,
increment,
body,
..
} => {
visitor.visit_stmt(variable)?;
visitor.visit_expr(condition)?;
visitor.visit_stmt(increment)?;
visitor.visit_stmt(body)
}
}
}
/// Recurse into the children of `expr`, calling back into `visitor`.
pub fn walk_expr<V: Visitor>(visitor: &mut V, expr: &AstNode<Expr>) -> LoxResult<()> {
pub fn walk_expr<V: Visitor>(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 {
if let BaseValue::Function(function) = &**value {
visitor.visit_function(function)?;
}
Ok(())
@@ -137,6 +74,7 @@ pub fn walk_expr<V: Visitor>(visitor: &mut V, expr: &AstNode<Expr>) -> LoxResult
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, ..
@@ -147,6 +85,57 @@ pub fn walk_expr<V: Visitor>(visitor: &mut V, expr: &AstNode<Expr>) -> LoxResult
}
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)
}
}
}
@@ -155,7 +144,7 @@ pub fn walk_function<V: Visitor>(visitor: &mut V, function: &LoxFunction) -> Lox
if let Some(guard) = &function.guard {
visitor.visit_expr(guard)?;
}
visitor.visit_stmt(&function.body)
visitor.visit_expr(&function.body)
}
#[cfg(test)]
@@ -165,7 +154,7 @@ mod tests {
use crate::frontend::lexer::Lexer;
use crate::frontend::parser::Parser;
fn parse(src: &str) -> Vec<AstNode<Stmt>> {
fn parse(src: &str) -> Vec<AstNode> {
let tokens = Lexer::new(src.to_string(), 0)
.scans_tokens()
.expect("source should lex");
@@ -173,29 +162,23 @@ mod tests {
}
/// A visitor that relies entirely on the default traversal and just counts
/// how many statement and expression nodes it sees.
/// how many nodes it sees.
#[derive(Default)]
struct Counter {
stmts: usize,
exprs: usize,
nodes: usize,
}
impl Visitor for Counter {
fn visit_stmt(&mut self, stmt: &AstNode<Stmt>) -> LoxResult<()> {
self.stmts += 1;
walk_stmt(self, stmt)
}
fn visit_expr(&mut self, expr: &AstNode<Expr>) -> LoxResult<()> {
self.exprs += 1;
walk_expr(self, expr)
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_stmt(stmt).unwrap();
counter.visit_expr(stmt).unwrap();
}
counter
}
@@ -204,19 +187,16 @@ mod tests {
fn counts_every_node_via_default_traversal() {
// 1 + 2 * 3 => Binary(+){ Literal, Binary(*){ Literal, Literal } }
let counter = count("1 + 2 * 3;");
assert_eq!(counter.stmts, 1);
assert_eq!(counter.exprs, 5);
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) do return a; end");
// VarDeclaration + Block + Return
assert_eq!(counter.stmts, 3);
// Function literal + identifier `a`
assert_eq!(counter.exprs, 2);
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
@@ -224,7 +204,7 @@ mod tests {
struct FailOnIdentifier;
impl Visitor for FailOnIdentifier {
fn visit_expr(&mut self, expr: &AstNode<Expr>) -> LoxResult<()> {
fn visit_expr(&mut self, expr: &AstNode) -> LoxResult<()> {
if let Expr::Identifier { .. } = &expr.node {
return runtime_error(expr.source_slice.clone(), "found an identifier");
}
@@ -238,7 +218,7 @@ mod tests {
let mut visitor = FailOnIdentifier;
let mut result = Ok(());
for stmt in stmts.iter() {
result = visitor.visit_stmt(stmt);
result = visitor.visit_expr(stmt);
if result.is_err() {
break;
}
+296
View File
@@ -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<Vec<AstNode>, 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<AstNode> {
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());
}
+69
View File
@@ -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<BaseValue, String> {
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"));
}