[cpp] initial C++ support

I want to demo this on lnav's log files, so need
to add C++ support. These changes are enough to
extract variables from one of lnav's log
statements.

Author: ttiimm

Cargo.lock

@@ -11,6 +11,7 @@ regex = "1.11.1"
 serde = { version = "1.0.219", features = ["derive"] }
 serde_json = "1.0.140"
 tree-sitter = "0.25.3"
+tree-sitter-cpp = "0.23.4"
 tree-sitter-rust-orchard = "0.12.0"
 tree-sitter-java = "0.23.5"
 rayon = "1.10.0"

Cargo.toml

@@ -21,6 +21,7 @@ impl CodeSource {
             Some(ext) => match ext.to_str().unwrap() {
                 "rs" => SourceLanguage::Rust,
                 "java" => SourceLanguage::Java,
+                "h"|"hh"|"hpp"|"cc"|"cpp" => SourceLanguage::Cpp,
                 _ => panic!("Unsupported language"),
             },
             None => panic!("No extension"),
@@ -38,6 +39,7 @@ impl CodeSource {
         match self.language {
             SourceLanguage::Rust => tree_sitter_rust_orchard::LANGUAGE.into(),
             SourceLanguage::Java => tree_sitter_java::LANGUAGE.into(),
+            SourceLanguage::Cpp => tree_sitter_cpp::LANGUAGE.into(),
         }
     }
 
@@ -54,7 +56,7 @@ impl CodeSource {
     }
 }
 
-const SUPPORTED_EXTS: &[&str] = &["java", "rs"];
+const SUPPORTED_EXTS: &[&str] = &["java", "rs", "h", "hh", "hpp", "cc", "cpp"];
 
 fn try_add_file(path: PathBuf, srcs: &mut Vec<CodeSource>, filter: &Option<Vec<String>>) {
     if let Some(filter_list) = filter {

src/code_source.rs

@@ -41,10 +41,12 @@ impl Default for Filter {
 enum SourceLanguage {
     Rust,
     Java,
+    Cpp,
 }
 
 const IDENTS_RS: &[&str] = &["debug", "info", "warn"];
 const IDENTS_JAVA: &[&str] = &["logger", "log", "fine", "debug", "info", "warn", "trace"];
+const IDENTS_CPP: &[&str] = &["debug", "info", "warn", "trace"];
 
 impl SourceLanguage {
     fn get_query(&self) -> &str {
@@ -75,13 +77,25 @@ impl SourceLanguage {
                     )
                 "#
             }
+            SourceLanguage::Cpp => {
+                r#"
+                    (compound_statement
+                        (expression_statement
+                            (call_expression
+                                arguments: (argument_list (string_literal) @arguments)
+                            )
+                        )
+                    )
+                "#
+            }
         }
     }
 
     fn get_identifiers(&self) -> &[&str] {
         match self {
             SourceLanguage::Rust => IDENTS_RS,
             SourceLanguage::Java => IDENTS_JAVA,
+            SourceLanguage::Cpp => IDENTS_CPP,
         }
     }
 }
@@ -511,6 +525,25 @@ fn namedarg(name: &str) {
         );
     }
 
+    const CPP_SOURCE: &str = r#"
+    #include <stdio.h>
+
+    int main(int argc, char* argv[]) {
+        printf("Hello, %s!", argv[1]);
+    }
+    "#;
+
+    #[test]
+    fn test_basic_cpp() {
+        let log_ref = LogRef::new("Hello, Steve!");
+        let code = CodeSource::new(PathBuf::from("in-mem.cc"), Box::new(CPP_SOURCE.as_bytes()));
+        let src_refs = extract_logging(&mut [code]);
+        assert_eq!(src_refs.len(), 1);
+        let vars = extract_variables(log_ref, &src_refs[0]);
+        assert_eq!(vars.len(), 1);
+        assert_eq!(vars.get("argv[1]").map(|val| val.as_str()), Some("Steve"));
+    }
+
     #[test]
     fn test_find_possible_paths() {
         let code = CodeSource::new(PathBuf::from("in-mem.rs"), Box::new(TEST_SOURCE.as_bytes()));

src/lib.rs

@@ -84,6 +84,10 @@ impl<'a> SourceQuery<'a> {
                 let range = node.child_by_field_name("name").unwrap().range();
                 range.start_byte..range.end_byte
             }
+            "function_definition" => {
+                let range = node.child_by_field_name("declarator").unwrap().range();
+                range.start_byte..range.end_byte
+            }
             "method_declaration" => {
                 let range = node.child_by_field_name("name").unwrap().range();
                 range.start_byte..range.end_byte
@@ -101,9 +105,17 @@ impl<'a> SourceQuery<'a> {
                 range.start_byte..range.end_byte
             }
             _ => {
-                let r = Self::find_fn_range(node.parent().unwrap());
-                // println!("*****");
-                r
+                if let Some(parent) = node.parent() {
+                    if parent.kind() == "translation_unit" {
+                        let range = parent.range();
+                        return range.start_byte..range.end_byte;
+                    }
+                    Self::find_fn_range(parent)
+                } else {
+                    let range = node.range();
+
+                    range.start_byte..range.end_byte
+                }
             }
         }
     }

src/source_query.rs

@@ -88,10 +88,14 @@ static RUST_PLACEHOLDER_REGEX: LazyLock<Regex> = LazyLock::new(|| {
 static JAVA_PLACEHOLDER_REGEX: LazyLock<Regex> =
     LazyLock::new(|| Regex::new(r#"\\?\{.*}"#).unwrap());
 
+static CPP_PLACEHOLDER_REGEX: LazyLock<Regex> =
+    LazyLock::new(|| Regex::new(r#"%[-+ #0]*\d*(\.\d+)?[hlLzjt]*[diuoxXfFeEgGaAcspn%]"#).unwrap());
+
 fn placeholder_regex_for(language: SourceLanguage) -> &'static Regex {
     match language {
         SourceLanguage::Rust => RUST_PLACEHOLDER_REGEX.deref(),
         SourceLanguage::Java => JAVA_PLACEHOLDER_REGEX.deref(),
+        SourceLanguage::Cpp => CPP_PLACEHOLDER_REGEX.deref(),
     }
 }
 
@@ -160,4 +164,11 @@ mod tests {
         assert_eq!(Regex::new(r#"^(.+)$"#).unwrap().as_str(), matcher.as_str());
         assert_eq!(args[0], FormatArgument::Positional(2));
     }
+
+    #[test]
+    fn test_build_matcher_cpp() {
+        let (matcher, args) = build_matcher("they are %d years old", SourceLanguage::Cpp);
+        assert_eq!(Regex::new(r#"^they are (.+) years old$"#).unwrap().as_str(), matcher.as_str());
+        assert_eq!(args[0], FormatArgument::Placeholder);
+    }
 }