feat(kotlin): Implement complete metrics support for Kotlin language

Cargo.toml

@@ -24,20 +24,20 @@ serde = { version = "^1.0", features = ["derive"] }
 termcolor = "^1.2"
 walkdir = "^2.3"
 
-tree-sitter = "=0.25.3"
+tree-sitter = "=0.26.5"
 tree-sitter-java = "=0.23.5"
-tree-sitter-kotlin-ng = "1.1.0"
+tree-sitter-kotlin-codanna = "0.3.9"
 tree-sitter-typescript = "=0.23.2"
-tree-sitter-javascript = "=0.23.1"
-tree-sitter-python = "=0.23.6"
-tree-sitter-rust = "=0.23.2"
+tree-sitter-javascript = "=0.25.0"
+tree-sitter-python = "=0.25.0"
+tree-sitter-rust = "=0.24.0"
 tree-sitter-preproc = { path = "./tree-sitter-preproc", version = "=0.20.3" }
 tree-sitter-ccomment = { path = "./tree-sitter-ccomment", version = "=0.20.3" }
 tree-sitter-mozcpp = { path = "./tree-sitter-mozcpp", version = "=0.20.4" }
 tree-sitter-mozjs = { path = "./tree-sitter-mozjs", version = "=0.20.3" }
 
 [dev-dependencies]
-insta = { version = "1.29.0", features = ["yaml", "json", "redactions"] }
+insta = { version = "1.46.1", features = ["yaml", "json", "redactions"] }
 pretty_assertions = "^1.3"
 
 [profile.dev.package.insta]

enums/Cargo.toml

@@ -6,15 +6,15 @@ edition = "2024"
 
 [dependencies]
 clap = { version = "^4.0", features = ["derive"] }
-askama = "^0.14"
+askama = "^0.15"
 
-tree-sitter = "=0.25.3"
+tree-sitter = "=0.26.5"
 tree-sitter-java = "=0.23.5"
-tree-sitter-kotlin-ng = "1.1.0"
+tree-sitter-kotlin-codanna = "0.3.9"
 tree-sitter-typescript = "=0.23.2"
-tree-sitter-javascript = "=0.23.1"
-tree-sitter-python = "=0.23.6"
-tree-sitter-rust = "=0.23.2"
+tree-sitter-javascript = "=0.25.0"
+tree-sitter-python = "=0.25.0"
+tree-sitter-rust = "=0.24.0"
 tree-sitter-preproc = { path = "../tree-sitter-preproc", version = "=0.20.3" }
 tree-sitter-ccomment = { path = "../tree-sitter-ccomment", version = "=0.20.3" }
 tree-sitter-mozcpp = { path = "../tree-sitter-mozcpp", version = "=0.20.4" }

enums/src/languages.rs

@@ -3,7 +3,7 @@ use tree_sitter::Language;
 mk_langs!(
     // 1) Name for enum
     // 2) tree-sitter function to call to get a Language
-    (Kotlin, tree_sitter_kotlin_ng),
+    (Kotlin, tree_sitter_kotlin_codanna),
     (Java, tree_sitter_java),
     (Rust, tree_sitter_rust),
     (Cpp, tree_sitter_cpp),

enums/src/macros.rs

@@ -19,7 +19,7 @@ macro_rules! mk_get_language {
     ( $( ($camel:ident, $name:ident) ),* ) => {
         pub fn get_language(lang: &Lang) -> Language {
             match lang {
-                Lang::Kotlin => tree_sitter_kotlin_ng::LANGUAGE.into(),
+                Lang::Kotlin => tree_sitter_kotlin_codanna::language().into(),
                 Lang::Java => tree_sitter_java::LANGUAGE.into(),
                 Lang::Typescript => tree_sitter_typescript::LANGUAGE_TYPESCRIPT.into(),
                 Lang::Tsx => tree_sitter_typescript::LANGUAGE_TSX.into(),

src/asttools.rs

@@ -0,0 +1,153 @@
+//! AST traversal and analysis utilities.
+//!
+//! This module provides helper functions and macros for traversing
+//! and analyzing the AST tree structure.
+
+use crate::node::Node;
+
+/// Gets an ancestor at a specific level above the current node.
+///
+/// # Arguments
+/// * `node` - The starting node
+/// * `level` - How many levels up to traverse (0 returns the node itself)
+///
+/// # Returns
+/// The ancestor node at the specified level, or None if the tree isn't deep enough.
+///
+/// # Example
+/// ```ignore
+/// // Get the grandparent (2 levels up)
+/// if let Some(grandparent) = get_parent(&node, 2) {
+///     println!("Grandparent kind: {}", grandparent.kind());
+/// }
+/// ```
+pub fn get_parent<'a>(node: &Node<'a>, level: usize) -> Option<Node<'a>> {
+    let mut level = level;
+    let mut current = *node;
+    while level != 0 {
+        current = current.parent()?;
+        level -= 1;
+    }
+    Some(current)
+}
+
+/// Traverses a tree passing from children to children in search of a specific
+/// token or series of tokens.
+///
+/// # Arguments
+/// * `node` - The starting node
+/// * `token_list` - A slice of predicates, each matching a level of descent
+///
+/// # Returns
+/// The final node after following the token path, or None if any token wasn't found.
+///
+/// # Example
+/// ```ignore
+/// // Find: node -> child matching pred1 -> grandchild matching pred2
+/// let result = traverse_children(&node, &[
+///     |id| id == SomeToken::Foo as u16,
+///     |id| id == SomeToken::Bar as u16,
+/// ]);
+/// ```
+pub fn traverse_children<'a, F>(node: &Node<'a>, token_list: &[F]) -> Option<Node<'a>>
+where
+    F: Fn(u16) -> bool,
+{
+    let mut current = *node;
+    'outer: for token in token_list {
+        for child in current.children() {
+            if token(child.kind_id()) {
+                current = child;
+                continue 'outer;
+            }
+        }
+        // Token not found at this level
+        return None;
+    }
+    Some(current)
+}
+
+/// Checks if a node has specific ancestors in sequence.
+///
+/// This macro checks if the node's ancestors match a specific pattern,
+/// where the first pattern(s) are immediate ancestors and the last pattern
+/// is the final ancestor to match.
+///
+/// # Example
+/// ```ignore
+/// // Check if node is inside a function inside a class
+/// let is_method = has_ancestors!(node, Class | Struct, Function);
+/// ```
+#[macro_export]
+macro_rules! has_ancestors {
+    ($node:expr, $( $typs:pat_param )|*, $( $typ:pat_param ),+) => {{
+        let mut res = false;
+        loop {
+            let mut node = *$node;
+            $(
+                if let Some(parent) = node.parent() {
+                    match parent.kind_id().into() {
+                        $typ => {
+                            node = parent;
+                        },
+                        _ => {
+                            break;
+                        }
+                    }
+                } else {
+                    break;
+                }
+            )*
+            if let Some(parent) = node.parent() {
+                match parent.kind_id().into() {
+                    $( $typs )|+ => {
+                        res = true;
+                    },
+                    _ => {}
+                }
+            }
+            break;
+        }
+        res
+    }};
+}
+
+/// Counts specific ancestors matching a pattern until a stop condition.
+///
+/// This macro traverses up the tree counting ancestors that match the given
+/// patterns, stopping when it encounters an ancestor matching the stop pattern.
+///
+/// # Example
+/// ```ignore
+/// // Count nested if statements until we hit a function boundary
+/// let nesting = count_specific_ancestors!(node, If | ElseIf, Function | Method);
+/// ```
+#[macro_export]
+macro_rules! count_specific_ancestors {
+    ($node:expr, $checker:ty, $( $typs:pat_param )|*, $( $stops:pat_param )|*) => {{
+        let mut count = 0;
+        let mut node = *$node;
+        while let Some(parent) = node.parent() {
+            match parent.kind_id().into() {
+                $( $typs )|* => {
+                    if !<$checker>::is_else_if(&parent) {
+                        count += 1;
+                    }
+                },
+                $( $stops )|* => break,
+                _ => {}
+            }
+            node = parent;
+        }
+        count
+    }};
+}
+
+#[cfg(test)]
+mod tests {
+    #[test]
+    fn test_get_parent_level_zero() {
+        // Level 0 should return the same node
+        // (actual test would need a real node)
+    }
+}

src/checker.rs

@@ -663,39 +663,64 @@ impl Checker for RustCode {
 }
 
 impl Checker for KotlinCode {
-    fn is_comment(_: &Node) -> bool {
-        false
+    fn is_comment(node: &Node) -> bool {
+        matches!(
+            node.kind_id().into(),
+            Kotlin::LineComment | Kotlin::MultilineComment
+        )
     }
 
     fn is_useful_comment(_: &Node, _: &[u8]) -> bool {
         false
     }
 
-    fn is_func_space(_: &Node) -> bool {
-        false
+    fn is_func_space(node: &Node) -> bool {
+        matches!(
+            node.kind_id().into(),
+            Kotlin::SourceFile | Kotlin::ClassDeclaration
+        )
     }
 
-    fn is_func(_: &Node) -> bool {
-        false
+    fn is_func(node: &Node) -> bool {
+        node.kind_id() == Kotlin::FunctionDeclaration
     }
 
-    fn is_closure(_: &Node) -> bool {
-        false
+    fn is_closure(node: &Node) -> bool {
+        node.kind_id() == Kotlin::LambdaLiteral
     }
 
-    fn is_call(_: &Node) -> bool {
-        false
+    fn is_call(node: &Node) -> bool {
+        node.kind_id() == Kotlin::CallExpression
     }
 
-    fn is_non_arg(_: &Node) -> bool {
-        false
+    fn is_non_arg(node: &Node) -> bool {
+        matches!(
+            node.kind_id().into(),
+            Kotlin::LPAREN
+                | Kotlin::COMMA
+                | Kotlin::RPAREN
+                | Kotlin::PIPEPIPE
+                | Kotlin::UnaryExpression
+        )
     }
 
-    fn is_string(_: &Node) -> bool {
-        false
+    fn is_string(node: &Node) -> bool {
+        // StringLiteral covers both single-line and multi-line strings in this grammar
+        // StringContent captures the text content within strings
+        matches!(
+            node.kind_id().into(),
+            Kotlin::StringLiteral | Kotlin::StringContent
+        )
     }
 
-    fn is_else_if(_: &Node) -> bool {
+    fn is_else_if(node: &Node) -> bool {
+        if node.kind_id() != Kotlin::IfExpression {
+            return false;
+        }
+        if let Some(parent) = node.parent() {
+            return parent.kind_id() == Kotlin::Else;
+        }
+
         false
     }
 

src/getter.rs

@@ -575,4 +575,52 @@ impl Getter for JavaCode {
     }
 }
 
-impl Getter for KotlinCode {}
+impl Getter for KotlinCode {
+    fn get_space_kind(node: &Node) -> SpaceKind {
+        use Kotlin::*;
+
+        let typ = node.kind_id().into();
+        match typ {
+            ClassDeclaration => SpaceKind::Class,
+            FunctionDeclaration | Constructor | AnnotatedLambda => SpaceKind::Function,
+            SourceFile => SpaceKind::Unit,
+            _ => SpaceKind::Unknown,
+        }
+    }
+
+    fn get_op_type(node: &Node) -> HalsteadType {
+        use Kotlin::*;
+
+        let typ = node.kind_id();
+
+        match typ.into() {
+            // Operator: function calls
+            CallExpression
+            // Operator: control flow
+            | If | Else | When | Try | Catch | Throw | For | While | Continue | Break | Do | Finally
+            // Operator: keywords
+            | Return | Abstract | Final | Super | This
+            // Operator: brackets and comma and terminators (separators)
+            | SEMI | COMMA | COLONCOLON | LBRACE | LBRACK | LPAREN | RBRACE | RBRACK | RPAREN | DOTDOT | DOT
+            // Operator: operators
+            | EQ | LT | GT | BANG | QMARKCOLON | AsQMARK | COLON // no grammar for lambda operator ->
+            | EQEQ | LTEQ | GTEQ | BANGEQ | AMPAMP | PIPEPIPE | PLUSPLUS | DASHDASH
+            | PLUS | DASH | STAR | SLASH | PERCENT
+            | PLUSEQ | DASHEQ | STAREQ | SLASHEQ |  PERCENTEQ => {
+                HalsteadType::Operator
+            }
+            // Operands: variables, constants, literals
+            // StringLiteral covers both line strings and multi-line strings in this grammar
+            RealLiteral | IntegerLiteral | HexLiteral | BinLiteral | CharacterLiteralToken1 | UniCharacterLiteralToken1
+            | LiteralConstant | StringLiteral | StringContent | LambdaLiteral | FunctionLiteral
+            | ObjectLiteral | UnsignedLiteral | LongLiteral | BooleanLiteral | CharacterLiteral => {
+                HalsteadType::Operand
+            },
+            _ => {
+                HalsteadType::Unknown
+            },
+        }
+    }
+
+    get_operator!(Kotlin);
+}

src/langs.rs

@@ -53,7 +53,7 @@ mk_langs!(
         "kotlin",
         KotlinCode,
         KotlinParser,
-        tree_sitter_kotlin_ng,
+        tree_sitter_kotlin_codanna,
         [kt, kts],
         ["kotlin"]
     ),