parser: reject reserved keywords in qualified identifiers

Validate namespace and literal parts of ns::name against gawk's reserved
keyword list when building identifiers, including @namespace directives.

Closes: #37

Author: Franklin-Qi

parser/src/diagnostics.rs

@@ -75,6 +75,12 @@ pub enum ParsingError {
     SpecialVariableIndirectCall(Span, String),
     #[error("Can't chain non-associative operators.")]
     NonAssociativeOperator(Span),
+    #[error("Using reserved identifier `{1}` as a namespace is not allowed.")]
+    ReservedNamespace(Span, String),
+    #[error(
+        "Using reserved identifier `{1}` as second component of a qualified name is not allowed."
+    )]
+    ReservedQualifiedLiteral(Span, String),
 }
 
 impl ParsingError {
@@ -121,6 +127,8 @@ impl ParsingError {
             Self::SpecialVariableCall(span, _) => Some(span.clone()),
             Self::SpecialVariableIndirectCall(span, _) => Some(span.clone()),
             Self::NonAssociativeOperator(span) => Some(span.clone()),
+            Self::ReservedNamespace(span, _) => Some(span.clone()),
+            Self::ReservedQualifiedLiteral(span, _) => Some(span.clone()),
         }
     }
     fn hint(&self) -> Option<&'static str> {

parser/src/keywords.rs

@@ -0,0 +1,35 @@
+// This file is part of the uutils awk package.
+//
+// For the full copyright and license information, please view the LICENSE
+// files that was distributed with this source code.
+
+/// Returns whether `name` is a reserved keyword that cannot appear in a
+/// qualified identifier (`ns::name`), matching gawk.
+pub fn is_reserved_keyword(name: &str) -> bool {
+    matches!(
+        name,
+        "BEGIN"
+            | "END"
+            | "if"
+            | "else"
+            | "switch"
+            | "case"
+            | "default"
+            | "do"
+            | "while"
+            | "for"
+            | "in"
+            | "print"
+            | "printf"
+            | "getline"
+            | "next"
+            | "nextfile"
+            | "exit"
+            | "break"
+            | "continue"
+            | "return"
+            | "delete"
+            | "function"
+            | "func"
+    )
+}

parser/src/lib.rs

@@ -8,6 +8,7 @@
 mod ast;
 mod diagnostics;
 mod idempotency;
+mod keywords;
 mod lex;
 mod pratt;
 mod sexpr;
@@ -134,7 +135,14 @@ impl<'a> Parser<'a> {
                     }
                     Token::NamespaceDirective => {
                         let namespace = lex.expect_string()?;
-                        self.namespace = lex.lex_ident(namespace.as_ref(), self.arena)?;
+                        let namespace = lex.lex_ident(namespace.as_ref(), self.arena)?;
+                        if keywords::is_reserved_keyword(namespace) {
+                            return Err(ParsingError::ReservedNamespace(
+                                lex.span(),
+                                namespace.to_string(),
+                            ));
+                        }
+                        self.namespace = namespace;
                         lex.expect_with(Token::is_stmnt_end, "expected statement end.".into())?;
                     }
                     Token::ConcurrentDirective => {
@@ -544,8 +552,13 @@ impl<'a> Parser<'a> {
 
     fn parse_delete(&mut self, lex: &mut Lexer<'a>) -> Result<SimpleStatement<'a>> {
         let next = lex.expect_next()?;
-        let Ok(var) = self.get_place(lex, next) else {
-            return Err(ParsingError::OperatorExpectsVariable(lex.span()));
+        let var = match self.get_place(lex, next) {
+            Ok(var) => var,
+            Err(
+                err @ (ParsingError::ReservedNamespace(..)
+                | ParsingError::ReservedQualifiedLiteral(..)),
+            ) => return Err(err),
+            Err(_) => return Err(ParsingError::OperatorExpectsVariable(lex.span())),
         };
         let index = if lex.consume(&Token::OpenBracket) {
             let mut pratt = Pratt::new(self, false);
@@ -561,7 +574,9 @@ impl<'a> Parser<'a> {
 
     #[tracing::instrument]
     fn parse_function(&mut self, lex: &mut Lexer<'a>) -> Result<()> {
-        let name = lex.expect_identifier()?.qualify(self.namespace);
+        let name = lex
+            .expect_identifier()?
+            .try_qualify(self.namespace, &lex.span())?;
         let args = self.parse_signature(lex, &name)?;
         lex.consume(&Token::Newline);
         let body = self.parse_body(lex)?;
@@ -586,7 +601,9 @@ impl<'a> Parser<'a> {
         }
 
         loop {
-            let name = lex.expect_identifier()?.qualify(self.namespace);
+            let name = lex
+                .expect_identifier()?
+                .try_qualify(self.namespace, &lex.span())?;
             // Linear search is fine for the numbers we are working with.
             if let Some(arg) = args.iter().find(|&a| a == &name) {
                 return Err(ParsingError::DuplicatedArgument(
@@ -658,6 +675,10 @@ impl<'a> Parser<'a> {
             Token::TypedRegex(_) => Err(ParsingError::UnexpectedTypedRegex(lex.span())),
             token => match self.get_place(lex, token) {
                 Ok(var) => Ok(Atom::Variable(var)),
+                Err(
+                    err @ (ParsingError::ReservedNamespace(..)
+                    | ParsingError::ReservedQualifiedLiteral(..)),
+                ) => Err(err),
                 Err(_) => Err(ParsingError::UnexpectedToken(
                     lex.span(),
                     "is not valid data.".into(),
@@ -667,10 +688,10 @@ impl<'a> Parser<'a> {
     }
 
     #[tracing::instrument]
-    fn get_place(&self, lex: &mut Lexer<'a>, token: Token<'a>) -> Result<Variable<'a>, Token<'a>> {
+    fn get_place(&self, lex: &mut Lexer<'a>, token: Token<'a>) -> Result<Variable<'a>> {
         match token {
             Token::Identifier(a) if !(lex.peek_is(&Token::OpenParent) && lex.is_yuxtaposed()) => {
-                Ok(a.qualify(self.namespace).into())
+                Ok(a.try_qualify(self.namespace, &lex.span())?.into())
             }
             Token::NrVariable => Ok(Variable::Nr),
             Token::NfVariable => Ok(Variable::Nf),
@@ -687,7 +708,10 @@ impl<'a> Parser<'a> {
             Token::RstartVariable => Ok(Variable::Rstart),
             Token::RlengthVariable => Ok(Variable::Rlength),
             Token::EnvironVariable => Ok(Variable::Environ),
-            tok => Err(tok),
+            tok => Err(ParsingError::UnexpectedToken(
+                lex.span(),
+                format!("{tok:?}"),
+            )),
         }
     }
 }
@@ -717,22 +741,35 @@ impl Preprocessor {
 }
 
 trait IdentifierExt<'a> {
-    fn qualify(self, namespace: &'a str) -> Identifier<'a>
+    fn try_qualify(self, namespace: &'a str, span: &Span) -> Result<Identifier<'a>>
     where
         Self: 'a;
 }
 
 impl<'a> IdentifierExt<'a> for lexer::Identifier<'_> {
-    fn qualify(self, namespace: &'a str) -> Identifier<'a>
+    fn try_qualify(self, namespace: &'a str, span: &Span) -> Result<Identifier<'a>>
     where
         Self: 'a,
     {
         let literal = self.literal;
-        if let Some(namespace) = self.namespace {
-            Identifier { namespace, literal }
+        let namespace = if let Some(ns) = self.namespace {
+            if keywords::is_reserved_keyword(ns) {
+                return Err(ParsingError::ReservedNamespace(
+                    span.clone(),
+                    ns.to_string(),
+                ));
+            }
+            if keywords::is_reserved_keyword(literal) {
+                return Err(ParsingError::ReservedQualifiedLiteral(
+                    span.clone(),
+                    literal.to_string(),
+                ));
+            }
+            ns
         } else {
-            Identifier { namespace, literal }
-        }
+            namespace
+        };
+        Ok(Identifier { namespace, literal })
     }
 }
 

parser/src/pratt.rs

@@ -269,10 +269,12 @@ impl<'a, 'b> Pratt<'a, 'b> {
                     name.literal.to_string(),
                 ));
             }
+            let span = lex.span();
+            let qualified = name.try_qualify(self.parser.namespace, &span)?;
             self.parser.parse_function_call(
                 lex,
-                |args| ExprNode::FunctionCall(name.qualify(self.parser.namespace), args),
-                lex.span(),
+                move |args| ExprNode::FunctionCall(qualified, args),
+                span,
             )
         } else if let Some(builtin) = next.maps_to_builtin() {
             self.parser.parse_function_call(
@@ -289,16 +291,17 @@ impl<'a, 'b> Pratt<'a, 'b> {
                     name.literal.to_string(),
                 ));
             }
-            let name = Variable::User(name.qualify(self.parser.namespace));
+            let span = lex.span();
+            let name = Variable::User(name.try_qualify(self.parser.namespace, &span)?);
             self.parser.parse_function_call(
                 lex,
-                |args| ExprNode::IndirectCall(name, args),
-                lex.span(),
+                move |args| ExprNode::IndirectCall(name, args),
+                span,
             )
         } else if next.is_place() && lex.peek_is(&Token::OpenParent) && lex.is_yuxtaposed() {
             let name = match self.parser.get_place(lex, next) {
                 Ok(var) => var.to_string(),
-                Err(tok) => format!("{tok:?}"),
+                Err(err) => err.to_string(),
             };
             Err(ParsingError::SpecialVariableCall(lex.span(), name))
         } else {

parser/src/tests.rs

@@ -182,6 +182,18 @@ fn test_parser_invalid_patterns() {
     test_parser!(is_err!("BEGIN", "END", "BEGINFILE", "ENDFILE", "print 1;"));
 }
 
+#[test]
+fn test_parser_reserved_qualified_identifiers() {
+    test_parser!(is_err!(
+        "{ if::while }",
+        "{ foo::while }",
+        "{ while::foo }",
+        "@namespace \"if\"; BEGIN {}",
+        "function foo::while() {}",
+        "function while::foo() {}"
+    ));
+}
+
 #[test]
 fn test_parser_non_assoc() {
     test_parser!(is_err!(