[Merged by Bors] - Extract the parser into a crate

Cargo.toml

@@ -3,6 +3,7 @@ members = [
     "boa_cli",
     "boa_engine",
     "boa_ast",
+    "boa_parser",
     "boa_gc",
     "boa_interner",
     "boa_profiler",
@@ -30,6 +31,7 @@ boa_profiler = { version = "0.16.0", path = "boa_profiler" }
 boa_unicode = { version = "0.16.0", path = "boa_unicode" }
 boa_macros = { version = "0.16.0", path = "boa_macros" }
 boa_ast = { version = "0.16.0", path = "boa_ast" }
+boa_parser = { version = "0.16.0", path = "boa_parser" }
 
 [workspace.metadata.workspaces]
 allow_branch = "main"

boa_cli/Cargo.toml

@@ -15,6 +15,7 @@ rust-version.workspace = true
 boa_engine = { workspace = true, features = ["deser", "console"] }
 boa_ast = { workspace = true, features = ["serde"]}
 boa_interner.workspace = true
+boa_parser.workspace = true
 rustyline = "10.0.0"
 rustyline-derive = "0.7.0"
 clap = { version = "4.0.18", features = ["derive"] }

boa_cli/src/main.rs

@@ -144,11 +144,9 @@ fn parse_tokens<S>(src: S, context: &mut Context) -> Result<StatementList, Strin
 where
     S: AsRef<[u8]>,
 {
-    use boa_engine::syntax::parser::Parser;
-
     let src_bytes = src.as_ref();
-    Parser::new(src_bytes)
-        .parse_all(context)
+    boa_parser::Parser::new(src_bytes)
+        .parse_all(context.interner_mut())
         .map_err(|e| format!("ParsingError: {e}"))
 }
 

boa_engine/Cargo.toml

@@ -34,6 +34,7 @@ boa_gc.workspace = true
 boa_profiler.workspace = true
 boa_macros.workspace = true
 boa_ast.workspace = true
+boa_parser.workspace = true
 gc = "0.4.1"
 serde = { version = "1.0.147", features = ["derive", "rc"] }
 serde_json = "1.0.87"

boa_engine/src/builtins/eval/mod.rs

@@ -15,10 +15,13 @@ use crate::{
     error::JsNativeError,
     object::FunctionBuilder,
     property::Attribute,
-    Context, JsResult, JsValue,
+    Context, JsResult, JsString, JsValue,
+};
+use boa_ast::operations::{
+    contains, contains_arguments, top_level_var_declared_names, ContainsSymbol,
 };
-use boa_ast::operations::top_level_var_declared_names;
 use boa_gc::Gc;
+use boa_parser::Parser;
 use boa_profiler::Profiler;
 
 #[derive(Debug, Clone, Copy)]
@@ -68,6 +71,17 @@ impl Eval {
         mut strict: bool,
         context: &mut Context,
     ) -> JsResult<JsValue> {
+        bitflags::bitflags! {
+            /// Flags used to throw early errors on invalid `eval` calls.
+            #[derive(Default)]
+            struct Flags: u8 {
+                const IN_FUNCTION = 0b0001;
+                const IN_METHOD = 0b0010;
+                const IN_DERIVED_CONSTRUCTOR = 0b0100;
+                const IN_CLASS_FIELD_INITIALIZER = 0b1000;
+            }
+        }
+
         /// Possible actions that can be executed after exiting this function to restore the environment to its
         /// original state.
         #[derive(Debug)]
@@ -94,19 +108,92 @@ impl Eval {
         debug_assert!(direct || !strict);
 
         // 2. If Type(x) is not String, return x.
-        let Some(x) = x.as_string() else {
+        // TODO: rework parser to take an iterator of `u32` unicode codepoints
+        let Some(x) = x.as_string().map(JsString::to_std_string_escaped) else {
             return Ok(x.clone());
         };
 
         // Because of implementation details the following code differs from the spec.
-        // TODO: rework parser to take an iterator of `u32` unicode codepoints
 
-        // Parse the script body and handle early errors (6 - 11)
-        let body = match context.parse_eval(x.to_std_string_escaped().as_bytes(), direct, strict) {
-            Ok(body) => body,
-            Err(e) => return Err(JsNativeError::syntax().with_message(e.to_string()).into()),
+        // 5. Perform ? HostEnsureCanCompileStrings(evalRealm).
+        let mut parser = Parser::new(x.as_bytes());
+        if strict {
+            parser.set_strict();
+        }
+        // 11. Perform the following substeps in an implementation-defined order, possibly interleaving parsing and error detection:
+        //     a. Let script be ParseText(StringToCodePoints(x), Script).
+        //     b. If script is a List of errors, throw a SyntaxError exception.
+        //     c. If script Contains ScriptBody is false, return undefined.
+        //     d. Let body be the ScriptBody of script.
+        let body = parser.parse_eval(direct, context.interner_mut())?;
+
+        // 6. Let inFunction be false.
+        // 7. Let inMethod be false.
+        // 8. Let inDerivedConstructor be false.
+        // 9. Let inClassFieldInitializer be false.
+        // a. Let thisEnvRec be GetThisEnvironment().
+        let flags = match context
+            .realm
+            .environments
+            .get_this_environment()
+            .as_function_slots()
+        {
+            // 10. If direct is true, then
+            //     b. If thisEnvRec is a Function Environment Record, then
+            Some(function_env) if direct => {
+                let function_env = function_env.borrow();
+                // i. Let F be thisEnvRec.[[FunctionObject]].
+                let function_object = function_env.function_object().borrow();
+
+                // ii. Set inFunction to true.
+                let mut flags = Flags::IN_FUNCTION;
+
+                // iii. Set inMethod to thisEnvRec.HasSuperBinding().
+                if function_env.has_super_binding() {
+                    flags |= Flags::IN_METHOD;
+                }
+
+                // iv. If F.[[ConstructorKind]] is derived, set inDerivedConstructor to true.
+                if function_object
+                    .as_function()
+                    .expect("must be function object")
+                    .is_derived_constructor()
+                {
+                    flags |= Flags::IN_DERIVED_CONSTRUCTOR;
+                }
+
+                // TODO:
+                // v. Let classFieldInitializerName be F.[[ClassFieldInitializerName]].
+                // vi. If classFieldInitializerName is not empty, set inClassFieldInitializer to true.
+
+                flags
+            }
+            _ => Flags::default(),
         };
 
+        if !flags.contains(Flags::IN_FUNCTION) && contains(&body, ContainsSymbol::NewTarget) {
+            return Err(JsNativeError::syntax()
+                .with_message("invalid `new.target` expression inside eval")
+                .into());
+        }
+        if !flags.contains(Flags::IN_METHOD) && contains(&body, ContainsSymbol::SuperProperty) {
+            return Err(JsNativeError::syntax()
+                .with_message("invalid `super` reference inside eval")
+                .into());
+        }
+        if !flags.contains(Flags::IN_DERIVED_CONSTRUCTOR)
+            && contains(&body, ContainsSymbol::SuperCall)
+        {
+            return Err(JsNativeError::syntax()
+                .with_message("invalid `super` call inside eval")
+                .into());
+        }
+        if flags.contains(Flags::IN_CLASS_FIELD_INITIALIZER) && contains_arguments(&body) {
+            return Err(JsNativeError::syntax()
+                .with_message("invalid `arguments` reference inside eval")
+                .into());
+        }
+
         strict |= body.strict();
 
         // Because our environment model does not map directly to the spec, this section looks very different.

boa_engine/src/builtins/function/mod.rs

@@ -26,7 +26,6 @@ use crate::{
     property::{Attribute, PropertyDescriptor, PropertyKey},
     string::utf16,
     symbol::WellKnownSymbols,
-    syntax::Parser,
     value::IntegerOrInfinity,
     Context, JsResult, JsString, JsValue,
 };
@@ -37,6 +36,7 @@ use boa_ast::{
 };
 use boa_gc::{self, custom_trace, Finalize, Gc, Trace};
 use boa_interner::Sym;
+use boa_parser::Parser;
 use boa_profiler::Profiler;
 use dyn_clone::DynClone;
 use std::{

boa_engine/src/builtins/regexp/mod.rs

@@ -25,10 +25,10 @@ use crate::{
     property::{Attribute, PropertyDescriptorBuilder},
     string::{utf16, CodePoint},
     symbol::WellKnownSymbols,
-    syntax::lexer::regex::RegExpFlags,
     value::JsValue,
     Context, JsResult, JsString,
 };
+use boa_parser::lexer::regex::RegExpFlags;
 use boa_profiler::Profiler;
 use regress::Regex;
 use std::str::FromStr;

boa_engine/src/context/mod.rs

@@ -20,14 +20,14 @@ use crate::{
     object::{FunctionBuilder, GlobalPropertyMap, JsObject, ObjectData},
     property::{Attribute, PropertyDescriptor, PropertyKey},
     realm::Realm,
-    syntax::{parser::ParseError, Parser},
     vm::{CallFrame, CodeBlock, FinallyReturn, GeneratorResumeKind, Vm},
     JsResult, JsString, JsValue,
 };
 
 use boa_ast::StatementList;
 use boa_gc::Gc;
 use boa_interner::{Interner, Sym};
+use boa_parser::{Error as ParseError, Parser};
 use boa_profiler::Profiler;
 
 #[cfg(feature = "intl")]
@@ -162,24 +162,7 @@ impl Context {
         S: AsRef<[u8]>,
     {
         let mut parser = Parser::new(src.as_ref());
-        parser.parse_all(self)
-    }
-
-    /// Parse the given source text with eval specific handling.
-    pub(crate) fn parse_eval<S>(
-        &mut self,
-        src: S,
-        direct: bool,
-        strict: bool,
-    ) -> Result<StatementList, ParseError>
-    where
-        S: AsRef<[u8]>,
-    {
-        let mut parser = Parser::new(src.as_ref());
-        if strict {
-            parser.set_strict();
-        }
-        parser.parse_eval(direct, self)
+        parser.parse_all(&mut self.interner)
     }
 
     /// `Call ( F, V [ , argumentsList ] )`
@@ -471,7 +454,7 @@ impl Context {
     {
         let main_timer = Profiler::global().start_event("Evaluation", "Main");
 
-        let statement_list = Parser::new(src.as_ref()).parse_all(self)?;
+        let statement_list = Parser::new(src.as_ref()).parse_all(&mut self.interner)?;
 
         let code_block = self.compile(&statement_list)?;
         let result = self.execute(code_block);

boa_engine/src/error.rs

@@ -5,7 +5,6 @@ use crate::{
     object::JsObject,
     object::ObjectData,
     property::PropertyDescriptor,
-    syntax::parser,
     Context, JsString, JsValue,
 };
 use boa_gc::{Finalize, Trace};
@@ -320,8 +319,8 @@ impl JsError {
     }
 }
 
-impl From<parser::ParseError> for JsError {
-    fn from(err: parser::ParseError) -> Self {
+impl From<boa_parser::Error> for JsError {
+    fn from(err: boa_parser::Error) -> Self {
         Self::from(JsNativeError::from(err))
     }
 }
@@ -655,8 +654,8 @@ impl JsNativeError {
     }
 }
 
-impl From<parser::ParseError> for JsNativeError {
-    fn from(err: parser::ParseError) -> Self {
+impl From<boa_parser::Error> for JsNativeError {
+    fn from(err: boa_parser::Error) -> Self {
         Self::syntax().with_message(err.to_string())
     }
 }

boa_engine/src/lib.rs

@@ -89,7 +89,6 @@ pub mod property;
 pub mod realm;
 pub mod string;
 pub mod symbol;
-pub mod syntax;
 pub mod value;
 pub mod vm;
 

boa_examples/Cargo.toml

@@ -16,4 +16,5 @@ boa_engine = { workspace = true, features = ["console"] }
 boa_ast.workspace = true
 boa_interner.workspace = true
 boa_gc.workspace = true
+boa_parser.workspace = true
 gc = "0.4.1"

boa_examples/src/bin/commuter_visitor.rs

@@ -10,8 +10,9 @@ use boa_ast::{
     visitor::{VisitWith, VisitorMut},
     Expression,
 };
-use boa_engine::{syntax::Parser, Context};
+use boa_engine::Context;
 use boa_interner::ToInternedString;
+use boa_parser::Parser;
 use core::ops::ControlFlow;
 use std::{convert::Infallible, fs::File, io::BufReader};
 
@@ -69,7 +70,7 @@ fn main() {
     ));
     let mut ctx = Context::default();
 
-    let mut statements = parser.parse_all(&mut ctx).unwrap();
+    let mut statements = parser.parse_all(ctx.interner_mut()).unwrap();
 
     let mut visitor = CommutorVisitor::default();