Merge pull request #624 from ratmice/module_identifier_chars

Improve error when `CTParserBuilder::mod_name` is not a valid rust ident

Author: ltratt

lrlex/src/lib/ctbuilder.rs

@@ -753,7 +753,15 @@ where
                 format!("{}_l", stem)
             }
         };
-        let mod_name = format_ident!("{}", mod_name);
+        let mod_name =
+            match syn::parse_str::<proc_macro2::Ident>(&mod_name) {
+                Ok(s) => s,
+                Err(e) => return Err(format!(
+                    "CTLexerBuilder::mod_name(\"{}\") is not a valid rust identifier due to '{}'",
+                    mod_name, e
+                )
+                .into()),
+            };
         let mut lexerdef_func_impl = {
             let LexFlags {
                 allow_wholeline_comments,
@@ -1488,4 +1496,34 @@ mod test {
                 .unwrap();
         }
     }
+
+    #[test]
+    /// Tests a yacc .y filename containing a dash character leading to an invalid rust identifier
+    /// when that dash is subsequently used as the default `CTParserBuilder::mod_name`.
+    fn test_invalid_identifier_in_derived_mod_name() {
+        let mut lex_path = std::path::PathBuf::from(env!("OUT_DIR"));
+        lex_path.push("contains-a-dash.l");
+        let mut f = File::create(&lex_path).unwrap();
+        let _ = f.write_all(
+            r#"
+%%
+A  "A"
+"#
+            .as_bytes(),
+        );
+        match CTLexerBuilder::new()
+            .output_path(format!("{}.rs", lex_path.display()))
+            .lexer_path(lex_path.clone())
+            .build()
+        {
+            Ok(_) => panic!("Expected error"),
+            Err(e) => {
+                let err_string = e.to_string();
+                assert_eq!(
+                    err_string,
+                    "CTLexerBuilder::mod_name(\"contains-a-dash_l\") is not a valid rust identifier due to 'unexpected token'"
+                );
+            }
+        }
+    }
 }

lrpar/src/lib/ctbuilder.rs

@@ -973,7 +973,15 @@ where
                 None
             };
 
-        let mod_name = format_ident!("{}", mod_name);
+        let mod_name =
+            match syn::parse_str::<proc_macro2::Ident>(mod_name) {
+                Ok(s) => s,
+                Err(e) => return Err(format!(
+                    "CTParserBuilder::mod_name(\"{}\") is not a valid rust identifier due to '{}'",
+                    mod_name, e
+                )
+                .into()),
+            };
         let out_tokens = quote! {
             #visibility mod #mod_name {
                 // At the top so that `user_actions` may contain #![inner_attribute]
@@ -1801,6 +1809,37 @@ C : 'a';"
         }
     }
 
+    #[test]
+    /// Tests a yacc .y filename containing a dash character leading to an invalid rust identifier
+    /// when that dash is subsequently used as the default `CTParserBuilder::mod_name`.
+    fn test_invalid_identifier_in_derived_mod_name() {
+        let temp = TempDir::new().unwrap();
+        let mut file_path = PathBuf::from(temp.as_ref());
+        file_path.push("contains-a-dash.y");
+        let mut f = File::create(&file_path).unwrap();
+        let _ = f.write_all(
+            "%start A
+%%
+A : 'a';"
+                .as_bytes(),
+        );
+        match CTParserBuilder::<TestLexerTypes>::new()
+            .yacckind(YaccKind::Original(YaccOriginalActionKind::GenericParseTree))
+            .grammar_path(file_path.to_str().unwrap())
+            .output_path(file_path.with_extension("ignored"))
+            .build()
+        {
+            Ok(_) => panic!("Expected error"),
+            Err(e) => {
+                let err_string = e.to_string();
+                assert_eq!(
+                    err_string,
+                    "CTParserBuilder::mod_name(\"contains-a-dash_y\") is not a valid rust identifier due to 'unexpected token'"
+                );
+            }
+        }
+    }
+
     #[cfg(test)]
     #[test]
     fn test_recoverer_header() -> Result<(), Box<dyn std::error::Error>> {