1
0
forked from M-Labs/nac3

expression type check, but list comprehension done in a bad way for now...

This commit is contained in:
CrescentonC 2021-07-13 16:23:03 +08:00
parent 144b84a612
commit 7eb0ab41d4
2 changed files with 133 additions and 95 deletions

View File

@ -6,12 +6,12 @@ use rustpython_parser::ast;
use std::boxed::Box; use std::boxed::Box;
use std::collections::HashMap; use std::collections::HashMap;
struct ContextStack<'a> { pub struct ContextStack {
/// stack level, starts from 0 /// stack level, starts from 0
level: u32, pub level: u32,
/// stack of symbol definitions containing (name, level) where `level` is the smallest level /// stack of symbol definitions containing (name, level) where `level` is the smallest level
/// where the name is assigned a value /// where the name is assigned a value
sym_def: Vec<(&'a str, u32)>, pub sym_def: Vec<(String, u32)>,
} }
pub struct InferenceContext<'a> { pub struct InferenceContext<'a> {
@ -25,9 +25,9 @@ pub struct InferenceContext<'a> {
/// identifier to (type, readable, location) mapping. /// identifier to (type, readable, location) mapping.
/// an identifier might be defined earlier but has no value (for some code path), thus not /// an identifier might be defined earlier but has no value (for some code path), thus not
/// readable. /// readable.
sym_table: HashMap<&'a str, (Type, bool, Location)>, pub sym_table: HashMap<String, (Type, bool, Location)>,
/// stack /// stack
stack: ContextStack<'a>, pub stack: ContextStack,
} }
// non-trivial implementations here // non-trivial implementations here
@ -52,7 +52,7 @@ impl<'a> InferenceContext<'a> {
/// execute the function with new scope. /// execute the function with new scope.
/// variable assignment would be limited within the scope (not readable outside), and type /// variable assignment would be limited within the scope (not readable outside), and type
/// returns the list of variables assigned within the scope, and the result of the function /// returns the list of variables assigned within the scope, and the result of the function
pub fn with_scope<F, R>(&mut self, f: F) -> (Vec<(&'a str, Type, Location)>, R) pub fn with_scope<F, R>(&mut self, f: F) -> (Vec<(String, Type, Location)>, R)
where where
F: FnOnce(&mut Self) -> R, F: FnOnce(&mut Self) -> R,
{ {
@ -64,7 +64,7 @@ impl<'a> InferenceContext<'a> {
let (_, level) = self.stack.sym_def.last().unwrap(); let (_, level) = self.stack.sym_def.last().unwrap();
if *level > self.stack.level { if *level > self.stack.level {
let (name, _) = self.stack.sym_def.pop().unwrap(); let (name, _) = self.stack.sym_def.pop().unwrap();
let (t, b, l) = self.sym_table.get_mut(name).unwrap(); let (t, b, l) = self.sym_table.get_mut(&name).unwrap();
// set it to be unreadable // set it to be unreadable
*b = false; *b = false;
poped_names.push((name, t.clone(), *l)); poped_names.push((name, t.clone(), *l));
@ -77,8 +77,8 @@ impl<'a> InferenceContext<'a> {
/// assign a type to an identifier. /// assign a type to an identifier.
/// may return error if the identifier was defined but with different type /// may return error if the identifier was defined but with different type
pub fn assign(&mut self, name: &'a str, ty: Type, loc: ast::Location) -> Result<Type, String> { pub fn assign(&mut self, name: String, ty: Type, loc: ast::Location) -> Result<Type, String> {
if let Some((t, x, _)) = self.sym_table.get_mut(name) { if let Some((t, x, _)) = self.sym_table.get_mut(&name) {
if t == &ty { if t == &ty {
if !*x { if !*x {
self.stack.sym_def.push((name, self.stack.level)); self.stack.sym_def.push((name, self.stack.level));
@ -89,7 +89,7 @@ impl<'a> InferenceContext<'a> {
Err("different types".into()) Err("different types".into())
} }
} else { } else {
self.stack.sym_def.push((name, self.stack.level)); self.stack.sym_def.push((name.clone(), self.stack.level));
self.sym_table.insert( self.sym_table.insert(
name, name,
(ty.clone(), true, Location::CodeRange(self.file, loc)), (ty.clone(), true, Location::CodeRange(self.file, loc)),
@ -124,6 +124,11 @@ impl<'a> InferenceContext<'a> {
self.resolver.get_symbol_location(name) self.resolver.get_symbol_location(name)
} }
} }
/// check if an identifier is already defined
pub fn defined(&self, name: &String) -> bool {
self.sym_table.get(name).is_some()
}
} }
// trivial getters: // trivial getters:

View File

@ -1,4 +1,5 @@
use std::convert::TryInto; use std::convert::TryInto;
use std::fs::create_dir_all;
use crate::typecheck::context::InferenceContext; use crate::typecheck::context::InferenceContext;
use crate::typecheck::inference_core; use crate::typecheck::inference_core;
@ -6,11 +7,12 @@ use crate::typecheck::magic_methods;
use crate::typecheck::typedef::{Type, TypeEnum}; use crate::typecheck::typedef::{Type, TypeEnum};
use crate::typecheck::primitives; use crate::typecheck::primitives;
use rustpython_parser::ast; use rustpython_parser::ast;
use rustpython_parser::ast::fold::Fold;
use super::inference_core::resolve_call; use super::inference_core::resolve_call;
pub struct ExpressionTypeInferencer<'a> { pub struct ExpressionTypeInferencer<'a> {
pub ctx: InferenceContext<'a> //FIXME: may need to remove this pub pub ctx: InferenceContext<'a>
} }
impl<'a> ExpressionTypeInferencer<'a> { // NOTE: add location here in the function parameter for better error message? impl<'a> ExpressionTypeInferencer<'a> { // NOTE: add location here in the function parameter for better error message?
@ -211,25 +213,6 @@ impl<'a> ExpressionTypeInferencer<'a> { // NOTE: add location here in the functi
} }
} }
fn infer_slice(&self, lower: &Option<Box<ast::Expr<Option<Type>>>>, upper: &Option<Box<ast::Expr<Option<Type>>>>, step: &Option<Box<ast::Expr<Option<Type>>>>) -> Result<Option<Type>, String> {
let int32_type = self.ctx.get_primitive(primitives::INT32_TYPE);
let l = lower.as_ref().map_or(
Ok(&int32_type),
|x| x.custom.as_ref().ok_or("lower bound cannot be typped".to_string()))?;
let u = upper.as_ref().map_or(
Ok(&int32_type),
|x| x.custom.as_ref().ok_or("upper bound cannot be typped".to_string()))?;
let s = step.as_ref().map_or(
Ok(&int32_type),
|x| x.custom.as_ref().ok_or("step cannot be typped".to_string()))?;
if l == &int32_type && u == &int32_type && s == &int32_type {
Ok(Some(self.ctx.get_primitive(primitives::SLICE_TYPE)))
} else {
Err("slice must be int32 type".into())
}
}
fn infer_subscript(&self, value: &Box<ast::Expr<Option<Type>>>, slice: &Box<ast::Expr<Option<Type>>>) -> Result<Option<Type>, String> { fn infer_subscript(&self, value: &Box<ast::Expr<Option<Type>>>, slice: &Box<ast::Expr<Option<Type>>>) -> Result<Option<Type>, String> {
// let tt = value.custom.ok_or_else(|| "no value".to_string())?.as_ref(); // let tt = value.custom.ok_or_else(|| "no value".to_string())?.as_ref();
@ -239,13 +222,28 @@ impl<'a> ExpressionTypeInferencer<'a> { // NOTE: add location here in the functi
return Err("subscript is not supported for types other than list".into()); return Err("subscript is not supported for types other than list".into());
}; };
if slice.custom == Some(self.ctx.get_primitive(primitives::SLICE_TYPE)) { if let ast::ExprKind::Slice {lower, upper, step} = &slice.node {
Ok(value.custom.clone()) let int32_type = self.ctx.get_primitive(primitives::INT32_TYPE);
} else if slice.custom == Some(self.ctx.get_primitive(primitives::INT32_TYPE)) { let l = lower.as_ref().map_or(
Ok(Some(t)) Ok(&int32_type),
} else { |x| x.custom.as_ref().ok_or("lower bound cannot be typped".to_string()))?;
Err("slice or index must be int32 type".into()) let u = upper.as_ref().map_or(
} Ok(&int32_type),
|x| x.custom.as_ref().ok_or("upper bound cannot be typped".to_string()))?;
let s = step.as_ref().map_or(
Ok(&int32_type),
|x| x.custom.as_ref().ok_or("step cannot be typped".to_string()))?;
if l == &int32_type && u == &int32_type && s == &int32_type {
Ok(value.custom.clone())
} else {
Err("slice must be int32 type".into())
}
} else if slice.custom == Some(self.ctx.get_primitive(primitives::INT32_TYPE)) {
Ok(Some(t))
} else {
Err("slice or index must be int32 type".into())
}
} }
fn infer_if_expr(&self, test: &Box<ast::Expr<Option<Type>>>, body: &Box<ast::Expr<Option<Type>>>, orelse: &Box<ast::Expr<Option<Type>>>) -> Result<Option<Type>, String> { fn infer_if_expr(&self, test: &Box<ast::Expr<Option<Type>>>, body: &Box<ast::Expr<Option<Type>>>, orelse: &Box<ast::Expr<Option<Type>>>) -> Result<Option<Type>, String> {
@ -260,21 +258,56 @@ impl<'a> ExpressionTypeInferencer<'a> { // NOTE: add location here in the functi
} }
} }
fn infer_simple_binding(&mut self, name: &'a ast::Expr<Option<Type>>, ty: Type) -> Result<(), String> { fn infer_list_comprehesion(&mut self, elt: &Box<ast::Expr<Option<Type>>>, generators: &Vec<ast::Comprehension<Option<Type>>>) -> Result<Option<Type>, String> {
if generators[0]
.ifs
.iter()
.all(|x| x.custom == Some(self.ctx.get_primitive(primitives::BOOL_TYPE))) {
Ok(Some(TypeEnum::ParametricType(
primitives::LIST_TYPE,
vec![elt.custom.clone().ok_or_else(|| "elements should have value".to_string())?]).into()))
} else {
Err("test must be bool".into())
}
}
fn fold_comprehension_first(&mut self, node: ast::Comprehension<Option<Type>>) -> Result<ast::Comprehension<Option<Type>>, String> {
Ok(ast::Comprehension {
target: node.target,
iter: Box::new(self.fold_expr(*node.iter)?),
ifs: node.ifs,
is_async: node.is_async
})
}
fn fold_comprehension_second(&mut self, node: ast::Comprehension<Option<Type>>) -> Result<ast::Comprehension<Option<Type>>, String> {
Ok(ast::Comprehension {
target: Box::new(self.fold_expr(*node.target)?),
iter: node.iter,
ifs: node
.ifs
.into_iter()
.map(|x| self.fold_expr(x))
.collect::<Result<Vec<_>, _>>()?,
is_async: node.is_async
})
}
fn infer_simple_binding(&mut self, name: &ast::Expr<Option<Type>>, ty: Type) -> Result<(), String> {
match &name.node { match &name.node {
ast::ExprKind::Name {id, ctx: _} => { ast::ExprKind::Name {id, ctx: _} => {
if id == "_" { if id == "_" {
Ok(()) Ok(())
} else if self.ctx.defined(id.as_str()) { } else if self.ctx.defined(id) {
Err("duplicated naming".into()) Err("duplicated naming".into())
} else { } else {
self.ctx.assign(id.as_str(), ty, name.location)?; self.ctx.assign(id.clone(), ty, name.location)?;
Ok(()) Ok(())
} }
} }
ast::ExprKind::Tuple {elts, ctx: _} => { ast::ExprKind::Tuple {elts, ctx: _} => {
if let TypeEnum::ParametricType(TUPLE_TYPE, ls) = ty.as_ref() { if let TypeEnum::ParametricType(primitives::TUPLE_TYPE, ls) = ty.as_ref() {
if elts.len() == ls.len() { if elts.len() == ls.len() {
for (a, b) in elts.iter().zip(ls.iter()) { for (a, b) in elts.iter().zip(ls.iter()) {
self.infer_simple_binding(a, b.clone())?; self.infer_simple_binding(a, b.clone())?;
@ -287,37 +320,12 @@ impl<'a> ExpressionTypeInferencer<'a> { // NOTE: add location here in the functi
Err("not supported".into()) Err("not supported".into())
} }
} }
_ => Err("not supported".into()) _ => Err("not supported".into())
} }
} }
fn infer_list_comprehesion(&mut self, elt: &Box<ast::Expr<Option<Type>>>, generators: &Vec<ast::Comprehension<Option<Type>>>) -> Result<Option<Type>, String> {
if generators.len() != 1 {
Err("only 1 generator statement is supported".into())
} else {
let gen = &generators[0];
if gen.is_async {
Err("async is not supported".into())
} else {
let iter_type = gen.iter.custom.as_ref().ok_or("no value".to_string())?.as_ref();
if let TypeEnum::ParametricType(primitives::LIST_TYPE, ref ls) = iter_type {
self.ctx.with_scope(|x| {
// x.infer_simple_binding(&gen.target, ls[0].clone()); // FIXME:
Ok(None)
}).1
} else {
Err("iteration is supported for list only".into())
}
}
}
}
} }
// REVIEW: field custom: from () to Option<Type> or just Option<Type>?
impl<'a> ast::fold::Fold<Option<Type>> for ExpressionTypeInferencer<'a> { impl<'a> ast::fold::Fold<Option<Type>> for ExpressionTypeInferencer<'a> {
type TargetU = Option<Type>; type TargetU = Option<Type>;
type Error = String; type Error = String;
@ -326,28 +334,54 @@ impl<'a> ast::fold::Fold<Option<Type>> for ExpressionTypeInferencer<'a> {
Ok(user) Ok(user)
} }
// override the default fold_comprehension to avoid errors caused by folding locally bound variable
fn fold_comprehension(&mut self, node: ast::Comprehension<Option<Type>>) -> Result<ast::Comprehension<Self::TargetU>, Self::Error> {
Ok(ast::Comprehension {
target: node.target,
iter: Box::new(self.fold_expr(*node.iter)?),
ifs: node.ifs,
is_async: node.is_async
})
}
fn fold_expr(&mut self, node: ast::Expr<Option<Type>>) -> Result<ast::Expr<Self::TargetU>, Self::Error> { fn fold_expr(&mut self, node: ast::Expr<Option<Type>>) -> Result<ast::Expr<Self::TargetU>, Self::Error> {
assert_eq!(node.custom, None); // NOTE: should pass assert_eq!(node.custom, None); // NOTE: should pass
let mut expr = node; let mut expr = node;
if let ast::Expr {location: _, custom: _, node: ast::ExprKind::ListComp {elt, generators } } = expr {
expr = ast::Expr { if let ast::Expr {location, custom, node: ast::ExprKind::ListComp {elt, generators } } = expr {
location: expr.location, // is list comprehension, only fold generators which does not include unknown identifiers introduced by list comprehension
custom: expr.custom, if generators.len() != 1 {
node: ast::ExprKind::ListComp { return Err("only 1 generator statement is supported".into())
elt, }
generators: generators.into_iter().map(|x| self.fold_comprehension(x)).collect::<Result<Vec<_>, _>>()? let generators_first_folded = generators
.into_iter()
.map(|x| self.fold_comprehension_first(x)).collect::<Result<Vec<_>, _>>()?;
let gen = &generators_first_folded[0];
let iter_type = gen.iter.custom.as_ref().ok_or("no value".to_string())?.as_ref();
if let TypeEnum::ParametricType(primitives::LIST_TYPE, ls) = iter_type {
self.ctx.stack.level += 1; // FIXME: how to use with_scope??
self.infer_simple_binding(&gen.target, ls[0].clone())?;
expr = ast::Expr {
location,
custom,
node: ast::ExprKind::ListComp {
elt: Box::new(self.fold_expr(*elt)?),
generators: generators_first_folded
.into_iter()
.map(|x| self.fold_comprehension_second(x))
.collect::<Result<Vec<_>, _>>()?
}
};
self.ctx.stack.level -= 1;
while !self.ctx.stack.sym_def.is_empty() {
let (_, level) = self.ctx.stack.sym_def.last().unwrap();
if *level > self.ctx.stack.level {
let (name, _) = self.ctx.stack.sym_def.pop().unwrap();
let (t, b, l) = self.ctx.sym_table.get_mut(&name).unwrap();
// set it to be unreadable
*b = false;
} else {
break;
}
} }
};
} else {
return Err("iteration is supported for list only".into());
}
} else { } else {
// if not listcomp which requires special handling, skip current level, make sure child nodes have their type // if not listcomp which requires special handling, skip current level, make sure child nodes have their type
expr = ast::fold::fold_expr(self, expr)?; expr = ast::fold::fold_expr(self, expr)?;
@ -429,15 +463,15 @@ impl<'a> ast::fold::Fold<Option<Type>> for ExpressionTypeInferencer<'a> {
node: expr.node node: expr.node
}), }),
// REVIEW: add a new primitive type for slice and do type check of bounds here? /* // REVIEW: add a new primitive type for slice and do type check of bounds here?
ast::ExprKind::Slice {lower, upper, step } => ast::ExprKind::Slice {lower, upper, step } =>
Ok(ast::Expr { Ok(ast::Expr {
location: expr.location, location: expr.location,
custom: self.infer_slice(lower, upper, step)?, custom: self.infer_slice(lower, upper, step)?,
node: expr.node node: expr.node
}), }), */
ast::ExprKind::Subscript {value, slice, ctx} => ast::ExprKind::Subscript {value, slice, ctx: _} =>
Ok(ast::Expr { Ok(ast::Expr {
location: expr.location, location: expr.location,
custom: self.infer_subscript(value, slice)?, custom: self.infer_subscript(value, slice)?,
@ -451,15 +485,14 @@ impl<'a> ast::fold::Fold<Option<Type>> for ExpressionTypeInferencer<'a> {
node: expr.node node: expr.node
}), }),
ast::ExprKind::ListComp {elt, generators} => ast::ExprKind::ListComp {elt, generators} => {
Ok(ast::Expr { Ok(ast::Expr {
location: expr.location, location: expr.location,
custom: self.infer_list_comprehesion(elt, generators)?, custom: self.infer_list_comprehesion(elt, generators)?,
node: expr.node node: expr.node
}), })
}
_ => { // not supported _ => { // not supported
Err("not supported yet".into()) Err("not supported yet".into())