From c734bfd63583657ae14210cbfab5a9985980dc75 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Simon=20W=C3=BClker?= <simon.wuelker@arcor.de>
Date: Sat, 20 Jan 2024 17:05:46 +0100
Subject: [PATCH] js: Track some static semantics during parsing

---
 crates/js/src/parser/identifiers.rs | 32 +++++++++++++++++++++++++--
 crates/js/src/parser/script.rs      |  1 +
 crates/js/src/parser/tokenizer.rs   | 34 ++++++++++++++++++++++++-----
 3 files changed, 60 insertions(+), 7 deletions(-)

diff --git a/crates/js/src/parser/identifiers.rs b/crates/js/src/parser/identifiers.rs
index 5c33bbdb..f8b6d325 100644
--- a/crates/js/src/parser/identifiers.rs
+++ b/crates/js/src/parser/identifiers.rs
@@ -1,6 +1,6 @@
 //! <https://262.ecma-international.org/14.0/#sec-identifiers>
 
-use super::{SyntaxError, Tokenizer};
+use super::{tokenizer::GoalSymbol, SyntaxError, Tokenizer};
 
 const RESERVED_WORDS: [&str; 37] = [
     "await",
@@ -47,11 +47,17 @@ pub(crate) fn parse_binding_identifier<const YIELD: bool, const AWAIT: bool>(
     tokenizer: &mut Tokenizer<'_>,
 ) -> Result<String, SyntaxError> {
     let binding_identifier = if let Ok(identifier) = tokenizer.attempt(Identifier::parse) {
+        if tokenizer.is_strict() && matches!(identifier.0.as_str(), "arguments" | "eval") {
+            return Err(tokenizer.syntax_error());
+        }
+
         identifier.0
     } else {
         let identifier_name = tokenizer.consume_identifier()?;
 
-        if matches!(identifier_name.as_str(), "yield" | "await") {
+        if !YIELD && identifier_name.as_str() == "yield" {
+            identifier_name
+        } else if !AWAIT && identifier_name.as_str() == "await" {
             identifier_name
         } else {
             return Err(tokenizer.syntax_error());
@@ -65,6 +71,18 @@ pub(crate) fn parse_binding_identifier<const YIELD: bool, const AWAIT: bool>(
 #[derive(Clone, Debug)]
 pub struct Identifier(String);
 
+const DISALLOWED_IDENTIFIERS_IN_STRICT_MODE: [&str; 9] = [
+    "implements",
+    "interface",
+    "let",
+    "package",
+    "private",
+    "protected",
+    "public",
+    "static",
+    "yield",
+];
+
 impl Identifier {
     /// <https://262.ecma-international.org/14.0/#prod-Identifier>
     pub(crate) fn parse(tokenizer: &mut Tokenizer<'_>) -> Result<Self, SyntaxError> {
@@ -73,6 +91,16 @@ impl Identifier {
             return Err(tokenizer.syntax_error());
         }
 
+        if tokenizer.is_strict()
+            && DISALLOWED_IDENTIFIERS_IN_STRICT_MODE.contains(&identifier_name.as_str())
+        {
+            return Err(tokenizer.syntax_error());
+        }
+
+        if tokenizer.goal_symbol() == GoalSymbol::Module && identifier_name.as_str() == "await" {
+            return Err(tokenizer.syntax_error());
+        }
+
         Ok(Self(identifier_name))
     }
 }
diff --git a/crates/js/src/parser/script.rs b/crates/js/src/parser/script.rs
index 6b32d86c..20efc69d 100644
--- a/crates/js/src/parser/script.rs
+++ b/crates/js/src/parser/script.rs
@@ -89,6 +89,7 @@ impl CompileToBytecode for StatementListItem {
 
 impl CompileToBytecode for Statement {
     fn compile(&self, builder: &mut bytecode::Builder) -> Self::Result {
+        _ = builder;
         todo!()
     }
 }
diff --git a/crates/js/src/parser/tokenizer.rs b/crates/js/src/parser/tokenizer.rs
index 8d313f4c..851da235 100644
--- a/crates/js/src/parser/tokenizer.rs
+++ b/crates/js/src/parser/tokenizer.rs
@@ -96,19 +96,43 @@ pub enum Punctuator {
     CurlyBraceClose,
 }
 
+#[derive(Clone, Copy, Debug, PartialEq, Eq)]
+pub enum GoalSymbol {
+    Script,
+    Module,
+}
+
 #[derive(Clone, Copy)]
 pub struct Tokenizer<'a> {
     source: ReversibleCharIterator<&'a str>,
+    strict: bool,
+    goal_symbol: GoalSymbol,
 }
 
 impl<'a> Tokenizer<'a> {
     #[must_use]
-    pub fn new(source_text: &'a str) -> Self {
+    pub fn new(source_text: &'a str, goal_symbol: GoalSymbol) -> Self {
         Self {
             source: ReversibleCharIterator::new(source_text),
+            strict: false,
+            goal_symbol,
         }
     }
 
+    #[must_use]
+    pub const fn is_strict(&self) -> bool {
+        self.strict
+    }
+
+    #[must_use]
+    pub const fn goal_symbol(&self) -> GoalSymbol {
+        self.goal_symbol
+    }
+
+    pub fn set_strict(&mut self, strict: bool) {
+        self.strict = strict;
+    }
+
     #[must_use]
     pub fn syntax_error(&self) -> SyntaxError {
         SyntaxError::from_position(self.source.position())
@@ -493,12 +517,12 @@ mod tests {
     #[test]
     fn tokenize_punctuator() {
         assert_eq!(
-            Tokenizer::new("?.").consume_punctuator(),
+            Tokenizer::new("?.", GoalSymbol::Script).consume_punctuator(),
             Ok(Punctuator::OptionalChaining)
         );
 
         assert_ne!(
-            Tokenizer::new("?.5").consume_punctuator(),
+            Tokenizer::new("?.5", GoalSymbol::Script).consume_punctuator(),
             Ok(Punctuator::OptionalChaining)
         );
     }
@@ -506,14 +530,14 @@ mod tests {
     #[test]
     fn tokenize_string_literal() {
         assert_eq!(
-            Tokenizer::new("\"foobar\"")
+            Tokenizer::new("\"foobar\"", GoalSymbol::Script)
                 .consume_string_literal()
                 .as_deref(),
             Ok("foobar")
         );
 
         assert_eq!(
-            Tokenizer::new("'foobar'")
+            Tokenizer::new("'foobar'", GoalSymbol::Script)
                 .consume_string_literal()
                 .as_deref(),
             Ok("foobar")