Update grammars (#1141)

* Bump tree-sitter to 0.25.3

* Bump tree-sitter-java to 0.23.5

* Replace tree-sitter-kotlin with tree-sitter-kotlin-ng 1.1.0

* Bump tree-sitter-typescript to 0.23.2

* Bump tree-sitter-javascript to 0.23.1

* Bump tree-sitter-python to 0.23.6

* Bump tree-sitter-rust to 0.23.2

* Bump tree-sitter-preproc to 0.20.3

* Bump tree-sitter-ccomment to 0.20.3

* Bump tree-sitter-mozcpp to 0.20.4

* Bump tree-sitter-mozjs to 0.20.3

* Sync rca-output

Author: alexle0nte

Cargo.toml

@@ -24,17 +24,17 @@ serde = { version = "^1.0", features = ["derive"] }
 termcolor = "^1.2"
 walkdir = "^2.3"
 
-tree-sitter = "=0.22.6"
-tree-sitter-java = "=0.21.0"
-tree-sitter-kotlin = "0.3.6"
-tree-sitter-typescript = "=0.21.2"
-tree-sitter-javascript = "=0.21.4"
-tree-sitter-python = "=0.21.0"
-tree-sitter-rust = "=0.21.2"
-tree-sitter-preproc = { path = "./tree-sitter-preproc", version = "=0.20.2" }
-tree-sitter-ccomment = { path = "./tree-sitter-ccomment", version = "=0.20.2" }
-tree-sitter-mozcpp = { path = "./tree-sitter-mozcpp", version = "=0.20.3" }
-tree-sitter-mozjs = { path = "./tree-sitter-mozjs", version = "=0.20.2" }
+tree-sitter = "=0.25.3"
+tree-sitter-java = "=0.23.5"
+tree-sitter-kotlin-ng = "1.1.0"
+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-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"] }

enums/Cargo.toml

@@ -8,17 +8,17 @@ edition = "2021"
 clap = { version = "^4.0", features = ["derive"] }
 askama = "^0.12"
 
-tree-sitter = "=0.22.6"
-tree-sitter-java = "=0.21.0"
-tree-sitter-kotlin = "0.3.6"
-tree-sitter-typescript = "=0.21.2"
-tree-sitter-javascript = "=0.21.4"
-tree-sitter-python = "=0.21.0"
-tree-sitter-rust = "=0.21.2"
-tree-sitter-preproc = { path = "../tree-sitter-preproc", version = "=0.20.2" }
-tree-sitter-ccomment = { path = "../tree-sitter-ccomment", version = "=0.20.2" }
-tree-sitter-mozcpp = { path = "../tree-sitter-mozcpp", version = "=0.20.3" }
-tree-sitter-mozjs = { path = "../tree-sitter-mozjs", version = "=0.20.2" }
+tree-sitter = "=0.25.3"
+tree-sitter-java = "=0.23.5"
+tree-sitter-kotlin-ng = "1.1.0"
+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-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" }
 
 [profile.release]
 strip = "debuginfo"

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),
+    (Kotlin, tree_sitter_kotlin_ng),
     (Java, tree_sitter_java),
     (Rust, tree_sitter_rust),
     (Cpp, tree_sitter_cpp),

enums/src/macros.rs

@@ -19,17 +19,17 @@ macro_rules! mk_get_language {
     ( $( ($camel:ident, $name:ident) ),* ) => {
         pub fn get_language(lang: &Lang) -> Language {
             match lang {
-                Lang::Kotlin => tree_sitter_kotlin::language(),
-                Lang::Java => tree_sitter_java::language(),
-                Lang::Typescript => tree_sitter_typescript::language_typescript(),
-                Lang::Tsx => tree_sitter_typescript::language_tsx(),
-                Lang::Javascript => tree_sitter_javascript::language(),
-                Lang::Python => tree_sitter_python::language(),
-                Lang::Rust => tree_sitter_rust::language(),
-                Lang::Preproc => tree_sitter_preproc::language(),
-                Lang::Ccomment => tree_sitter_ccomment::language(),
-                Lang::Cpp => tree_sitter_mozcpp::language(),
-                Lang::Mozjs => tree_sitter_mozjs::language(),
+                Lang::Kotlin => tree_sitter_kotlin_ng::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(),
+                Lang::Javascript => tree_sitter_javascript::LANGUAGE.into(),
+                Lang::Python => tree_sitter_python::LANGUAGE.into(),
+                Lang::Rust => tree_sitter_rust::LANGUAGE.into(),
+                Lang::Preproc => tree_sitter_preproc::LANGUAGE.into(),
+                Lang::Ccomment => tree_sitter_ccomment::LANGUAGE.into(),
+                Lang::Cpp => tree_sitter_mozcpp::LANGUAGE.into(),
+                Lang::Mozjs => tree_sitter_mozjs::LANGUAGE.into(),
             }
         }
     };

generate-grammars/generate-mozcpp.sh

@@ -53,8 +53,14 @@ npm install -y
 # Exit tree-sitter-cpp directory
 popd
 
-# Copy tree-sitter-cpp `scanner.cc` functions into the `src` directory
-cp --verbose $TS_CPP_CRATE/src/scanner.cc ./src/scanner.cc
+# Copy tree-sitter-cpp `scanner.c` functions into the `src` directory
+cp --verbose $TS_CPP_CRATE/src/scanner.c ./src/scanner.c
+
+# Since the tree-sitter-mozcpp `scanner.c` file contains the very same functions
+# present in the tree-sitter-cpp `scanner.c` file, to avoid having a
+# multiple symbol definition error during the linking phase,
+# those functions will be assigned a new prefix.
+sed -i 's/tree_sitter_cpp/tree_sitter_mozcpp/g' ./src/scanner.c
 
 # Exit tree-sitter-mozcpp directory
 popd

generate-grammars/generate-mozjs.sh

@@ -12,7 +12,7 @@ TS_JS_CRATE="tree-sitter-javascript"
 JSON_CRATE_FILENAME=".cargo_vcs_info.json"
 
 # Get the current tree-sitter-javascript crate version
-TS_JS_VERSION=`grep $TS_JS_CRATE Cargo.toml | cut -f2 -d "=" | tr -d ' ' | tr -d \"`
+TS_JS_VERSION=`grep -m 1 $TS_JS_CRATE tree-sitter-mozjs/Cargo.toml | cut -f2 -d "," | cut -f2 -d "=" | tr -d ' ' | tr -d } | tr -d \"`
 
 # Name assigned to the compressed binary crate downloaded from crates.io
 CRATE_OUTPUT="$TS_JS_CRATE-download.gz"
@@ -50,13 +50,14 @@ git checkout FETCH_HEAD
 # Exit tree-sitter-javascript directory
 popd
 
-# Since the tree-sitter-mozjs `scanner.cc` file contains the very same functions
-# present in the tree-sitter-javascript `scanner.cc` file, to avoid having a
-# multiple symbols definitions error during the linking phase,
-# to those functions will be assigned a new prefix and the relative
-# output file will be saved into the `src` directory.
-SED_PATTERN="s/tree_sitter_javascript_external_scanner_/tree_sitter_javascript_external_scanner_mozjs_/g"
-sed $SED_PATTERN $TS_JS_CRATE/src/scanner.c > ./src/tree_sitter_javascript_scanner.c
+# Copy tree-sitter-cpp `scanner.c` functions into the `src` directory
+cp --verbose $TS_JS_CRATE/src/scanner.c ./src/scanner.c
+
+# Since the tree-sitter-mozjs `scanner.c` file contains the very same functions
+# present in the tree-sitter-javascript `scanner.c` file, to avoid having a
+# multiple symbol definition error during the linking phase,
+# those functions will be assigned a new prefix.
+sed -i 's/tree_sitter_javascript/tree_sitter_mozjs/g' ./src/scanner.c
 
 # Exit tree-sitter-mozjs directory
 popd

src/checker.rs

@@ -253,7 +253,7 @@ impl Checker for CppCode {
             return false;
         }
         if let Some(parent) = node.parent() {
-            return node.kind_id() == Cpp::IfStatement && parent.kind_id() == Cpp::IfStatement;
+            return parent.kind_id() == Cpp::ElseClause;
         }
         false
     }

src/getter.rs

@@ -396,10 +396,24 @@ impl Getter for RustCode {
         use Rust::*;
 
         match node.kind_id().into() {
+            // `||` is treated as an operator only if it's part of a binary expression.
+            // This prevents misclassification inside macros where closures without arguments (e.g., `let closure = || { /* ... */ };`)
+            // are not recognized as `ClosureExpression` and their `||` node is identified as `PIPEPIPE` instead of `ClosureParameters`.
+            //
+            // Similarly, exclude `/` when it corresponds to the third slash in `///` (`OuterDocCommentMarker`)
+            PIPEPIPE | SLASH => match node.parent() {
+                Some(parent) if matches!(parent.kind_id().into(), BinaryExpression) => HalsteadType::Operator,
+                _ => HalsteadType::Unknown
+            }
+            // Ensure `!` is counted as an operator unless it belongs to an `InnerDocCommentMarker` `//!`
+            BANG => match node.parent() {
+                Some(parent) if !matches!(parent.kind_id().into(), InnerDocCommentMarker) => HalsteadType::Operator,
+                _ => HalsteadType::Unknown
+            }
             LPAREN | LBRACE | LBRACK | EQGT | PLUS | STAR | Async | Await | Continue | For | If
-            | Let | Loop | Match | Return | Unsafe | While | BANG | EQ | COMMA | DASHGT | QMARK
-            | LT | GT | AMP | MutableSpecifier | DOTDOT | DOTDOTEQ | DASH | AMPAMP | PIPEPIPE
-            | PIPE | CARET | EQEQ | BANGEQ | LTEQ | GTEQ | LTLT | GTGT | SLASH | PERCENT
+            | Let | Loop | Match | Return | Unsafe | While | EQ | COMMA | DASHGT | QMARK
+            | LT | GT | AMP | MutableSpecifier | DOTDOT | DOTDOTEQ | DASH | AMPAMP
+            | PIPE | CARET | EQEQ | BANGEQ | LTEQ | GTEQ | LTLT | GTGT | PERCENT
             | PLUSEQ | DASHEQ | STAREQ | SLASHEQ | PERCENTEQ | AMPEQ | PIPEEQ | CARETEQ
             | LTLTEQ | GTGTEQ | Move | DOT | PrimitiveType | Fn | SEMI => HalsteadType::Operator,
             Identifier | StringLiteral | RawStringLiteral | IntegerLiteral | FloatLiteral
@@ -478,13 +492,17 @@ impl Getter for CppCode {
         match node.kind_id().into() {
             DOT | LPAREN | LPAREN2 | COMMA | STAR | GTGT | COLON | SEMI | Return | Break
             | Continue | If | Else | Switch | Case | Default | For | While | Goto | Do | Delete
-            | New | Try | Catch | Throw | EQ | AMPAMP | PIPEPIPE | DASH | DASHDASH | DASHGT
+            | New | Try | Try2 | Catch | Throw | EQ | AMPAMP | PIPEPIPE | DASH | DASHDASH | DASHGT
             | PLUS | PLUSPLUS | SLASH | PERCENT | PIPE | AMP | LTLT | TILDE | LT | LTEQ | EQEQ
             | BANGEQ | GTEQ | GT | GT2 | PLUSEQ | BANG | STAREQ | SLASHEQ | PERCENTEQ | GTGTEQ
             | LTLTEQ | AMPEQ | CARET | CARETEQ | PIPEEQ | LBRACK | LBRACE | QMARK | COLONCOLON
             | PrimitiveType | TypeSpecifier | Sizeof => HalsteadType::Operator,
             Identifier | TypeIdentifier | FieldIdentifier | RawStringLiteral | StringLiteral
-            | NumberLiteral | True | False | Null | Nullptr | DOTDOTDOT => HalsteadType::Operand,
+            | NumberLiteral | True | False | Null | DOTDOTDOT => HalsteadType::Operand,
+            NamespaceIdentifier => match node.parent() {
+                Some(parent) if matches!(parent.kind_id().into(), NamespaceDefinition) => HalsteadType::Operand,
+                _ => HalsteadType::Unknown,
+            },
             _ => HalsteadType::Unknown,
         }
     }

src/langs.rs

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