Merge #239: fix(fuzz): avoid keywords in identifiers

6384adc fix(fuzz): avoid keywords in identifiers (Volodymyr Herashchenko)

Pull request description:

  Fixes #236. 
  
  Sometimes the fuzzer would generate keywords in identifiers and parser would reject this. This PR simply pads any keyword with `_`, like it was done with other built-ins.


ACKs for top commit:
  apoelstra:
    ACK 6384adc; successfully ran local tests


Tree-SHA512: 09a1b0c5ba75541f54eacb6f6f9fb5cf4b3a971586d2646a360c9c8819d653a2841d8f54fa69babb748c679807959d5504c8ef8554f983e24642e5a3b7f1e18e

Author: apoelstra

src/lexer.rs

@@ -237,6 +237,15 @@ pub fn lex<'src>(input: &'src str) -> (Option<Tokens<'src>>, Vec<crate::error::R
     )
 }
 
+/// Checks whether a given string is a keyword.
+#[cfg(feature = "arbitrary")]
+pub fn is_keyword(s: &str) -> bool {
+    matches!(
+        s,
+        "fn" | "let" | "type" | "mod" | "const" | "match" | "true" | "false"
+    )
+}
+
 #[cfg(test)]
 mod tests {
     use chumsky::error::Rich;

src/str.rs

@@ -47,9 +47,9 @@ macro_rules! wrapped_string {
 /// Implementation of [`arbitrary::Arbitrary`] for wrapped string types,
 /// such that strings of 1 to 10 letters `a` to `z` are generated.
 ///
-/// The space of lowercase letter strings includes values that are invalid
-/// according to the grammar of the particular string type. For instance,
-/// keywords are reserved. However, this should not affect fuzzing.
+/// The space of lowercase letter strings includes reserved keywords,
+/// which cannot be used as identifiers. To ensure valid grammar
+/// for fuzzing, any generated keywords are padded with the `_`.
 macro_rules! impl_arbitrary_lowercase_alpha {
     ($wrapper:ident) => {
         #[cfg(feature = "arbitrary")]
@@ -61,6 +61,9 @@ macro_rules! impl_arbitrary_lowercase_alpha {
                     let offset = u.int_in_range(0..=25)?;
                     string.push((b'a' + offset) as char)
                 }
+                if crate::lexer::is_keyword(string.as_str()) {
+                    string.push('_');
+                }
                 Ok(Self::from_str_unchecked(string.as_str()))
             }
         }
@@ -103,7 +106,7 @@ impl<'a> arbitrary::Arbitrary<'a> for FunctionName {
             let offset = u.int_in_range(0..=25)?;
             string.push((b'a' + offset) as char)
         }
-        if RESERVED_NAMES.contains(&string.as_str()) {
+        if RESERVED_NAMES.contains(&string.as_str()) || crate::lexer::is_keyword(string.as_str()) {
             string.push('_');
         }
 
@@ -202,7 +205,7 @@ impl<'a> arbitrary::Arbitrary<'a> for AliasName {
             let offset = u.int_in_range(0..=25)?;
             string.push((b'a' + offset) as char)
         }
-        if RESERVED_NAMES.contains(&string.as_str()) {
+        if RESERVED_NAMES.contains(&string.as_str()) || crate::lexer::is_keyword(string.as_str()) {
             string.push('_');
         }