[WIP] Feat: Add C++

.gitmodules

@@ -65,3 +65,9 @@
 [submodule "resources/language-submodules/tree-sitter-php"]
 	path = resources/language-submodules/tree-sitter-php
 	url = https://github.com/tree-sitter/tree-sitter-php
+[submodule "resources/language-submodules/tree-sitter-cpp"]
+	path = resources/language-submodules/tree-sitter-cpp
+	url = https://github.com/tree-sitter/tree-sitter-cpp
+[submodule "resources/language-submodules/tree-sitter-c"]
+	path = resources/language-submodules/tree-sitter-c
+	url = https://github.com/tree-sitter/tree-sitter-c

crates/core/src/test.rs

@@ -14987,3 +14987,54 @@ fn or_file() {
     })
     .unwrap();
 }
+
+#[test]
+fn cpp_simple() {
+    run_test_expected({
+        TestArgExpected {
+            pattern: r#"
+                |language cpp
+                |`char* editor = $string_literal;` => `char* editor = "emacs";`
+                |"#
+            .trim_margin()
+            .unwrap(),
+            source: r#"char* editor = "vim";"#.to_owned(),
+            expected: r#"char* editor = "emacs";"#.to_owned(),
+        }
+    })
+    .unwrap();
+}
+
+#[test]
+fn cpp_rename_variable_name() {
+    run_test_expected({
+        TestArgExpected {
+            pattern: r#"
+                |language cpp
+                |`char* $name = $val;` => `char* new_name = $val;`
+                |"#
+            .trim_margin()
+            .unwrap(),
+            source: r#"char* my_string = "hey";"#.to_owned(),
+            expected: r#"char* new_name = "hey";"#.to_owned(),
+        }
+    })
+    .unwrap();
+}
+
+#[test]
+fn cpp_change_float_to_double() {
+    run_test_expected({
+        TestArgExpected {
+            pattern: r#"
+                |language cpp
+                |`int $foo` => `int what`
+                |"#
+            .trim_margin()
+            .unwrap(),
+            source: r#"int foo"#.to_owned(),
+            expected: r#"int foo"#.to_owned(),
+        }
+    })
+    .unwrap();
+}

crates/language/Cargo.toml

@@ -15,6 +15,7 @@ rust.unused_crate_dependencies = "warn"
 [dependencies]
 tree-sitter = { path = "../../vendor/tree-sitter-facade", package = "tree-sitter-facade-sg" }
 tree-sitter-gritql = { path = "../../vendor/tree-sitter-gritql", optional = true }
+tree-sitter-cpp = { path = "../../resources/language-metavariables/tree-sitter-cpp", optional = true }
 tree-sitter-css = { path = "../../resources/language-metavariables/tree-sitter-css", optional = true }
 tree-sitter-json = { path = "../../resources/language-metavariables/tree-sitter-json", optional = true }
 tree-sitter-solidity = { path = "../../resources/language-metavariables/tree-sitter-solidity", optional = true }
@@ -64,6 +65,7 @@ builtin-parser = [
     "tree-sitter-javascript",
     "tree-sitter-html",
     "tree-sitter-java",
+    "tree-sitter-cpp",
     "tree-sitter-c-sharp",
     "tree-sitter-python",
     "tree-sitter-md",

crates/language/src/cpp.rs

@@ -0,0 +1,105 @@
+use crate::language::{fields_for_nodes, Field, MarzanoLanguage, NodeTypes, SortId, TSLanguage};
+use grit_util::Language;
+use marzano_util::node_with_source::NodeWithSource;
+use std::sync::OnceLock;
+
+static NODE_TYPES_STRING: &str = include_str!("../../../resources/node-types/cpp-node-types.json");
+static NODE_TYPES: OnceLock<Vec<Vec<Field>>> = OnceLock::new();
+static LANGUAGE: OnceLock<TSLanguage> = OnceLock::new();
+
+#[cfg(not(feature = "builtin-parser"))]
+fn language() -> TSLanguage {
+    unimplemented!(
+        "tree-sitter parser must be initialized before use when [builtin-parser] is off."
+    )
+}
+#[cfg(feature = "builtin-parser")]
+fn language() -> TSLanguage {
+    tree_sitter_cpp::language().into()
+}
+
+#[derive(Debug, Clone)]
+pub struct Cpp {
+    node_types: &'static [Vec<Field>],
+    metavariable_sort: SortId,
+    comment_sort: SortId,
+    language: &'static TSLanguage,
+}
+
+impl Cpp {
+    pub(crate) fn new(lang: Option<TSLanguage>) -> Self {
+        let language = LANGUAGE.get_or_init(|| lang.unwrap_or_else(language));
+        let node_types = NODE_TYPES.get_or_init(|| fields_for_nodes(language, NODE_TYPES_STRING));
+        let metavariable_sort = language.id_for_node_kind("grit_metavariable", true);
+        let comment_sort = language.id_for_node_kind("comment", true);
+        Self {
+            node_types,
+            metavariable_sort,
+            comment_sort,
+            language,
+        }
+    }
+
+    pub(crate) fn is_initialized() -> bool {
+        LANGUAGE.get().is_some()
+    }
+}
+
+impl NodeTypes for Cpp {
+    fn node_types(&self) -> &[Vec<Field>] {
+        self.node_types
+    }
+}
+
+impl Language for Cpp {
+    use_marzano_delegate!();
+
+    fn language_name(&self) -> &'static str {
+        // TODO: Confirm C++ vs Cpp, etc. here.
+        "C++"
+    }
+
+    fn snippet_context_strings(&self) -> &[(&'static str, &'static str)] {
+        // TODO: Unclear about other good additions here. Presumably, there are many.
+        // TODO: Unclear if the following additions are good or bad.
+        &[
+            ("", ""),
+            ("", ";"),
+            ("{", "}"),
+            ("", "  void GRIT_FN() {}"),
+            ("void GRIT_FN(", "){}"),
+            ("void GRIT_FN(){", "}"),
+        ]
+    }
+}
+
+impl<'a> MarzanoLanguage<'a> for Cpp {
+    fn get_ts_language(&self) -> &TSLanguage {
+        self.language
+    }
+
+    fn is_comment_sort(&self, id: SortId) -> bool {
+        id == self.comment_sort
+    }
+
+    fn metavariable_sort(&self) -> SortId {
+        self.metavariable_sort
+    }
+}
+
+#[cfg(test)]
+mod tests {
+
+    use crate::language::nodes_from_indices;
+
+    use super::*;
+
+    #[test]
+    fn snippet_nodes_not_empty() {
+        let snippet = r#"std::cout << "Hello World!";"#;
+        let lang = Cpp::new(None);
+        let snippets = lang.parse_snippet_contexts(snippet);
+        let nodes = nodes_from_indices(&snippets);
+        assert!(!nodes.is_empty());
+    }
+}

crates/language/src/lib.rs

@@ -56,6 +56,7 @@ macro_rules! use_marzano_delegate {
     };
 }
 
+pub mod cpp;
 pub mod csharp;
 pub mod css;
 pub mod foreign_language;

resources/edit_grammars.mjs

@@ -42,6 +42,8 @@ process.on('uncaughtException', (err) => {
 ///////////////////////////////////////////////////
 const allLanguages = [
   'c-sharp',
+  'c',
+  'cpp',
   'css',
   'go',
   'hcl',

resources/language-metavariables/tree-sitter-c/.editorconfig

@@ -0,0 +1,39 @@
+root = true
+
+[*]
+charset = utf-8
+end_of_line = lf
+insert_final_newline = true
+trim_trailing_whitespace = true
+
+[*.{json,toml,yml,gyp}]
+indent_style = space
+indent_size = 2
+
+[*.js]
+indent_style = space
+indent_size = 2
+
+[*.rs]
+indent_style = space
+indent_size = 4
+
+[*.{c,cc,h}]
+indent_style = space
+indent_size = 4
+
+[*.{py,pyi}]
+indent_style = space
+indent_size = 4
+
+[*.swift]
+indent_style = space
+indent_size = 4
+
+[*.go]
+indent_style = tab
+indent_size = 8
+
+[Makefile]
+indent_style = tab
+indent_size = 8

resources/language-metavariables/tree-sitter-c/.gitattributes

@@ -0,0 +1,11 @@
+* text eol=lf
+
+src/*.json linguist-generated
+src/parser.c linguist-generated
+src/tree_sitter/* linguist-generated
+
+bindings/** linguist-generated
+binding.gyp linguist-generated
+setup.py linguist-generated
+Makefile linguist-generated
+Package.swift linguist-generated

resources/language-metavariables/tree-sitter-c/.gitignore

@@ -0,0 +1,38 @@
+# Rust artifacts
+Cargo.lock
+target/
+
+# Node artifacts
+build/
+prebuilds/
+node_modules/
+*.tgz
+
+# Swift artifacts
+.build/
+
+# Go artifacts
+go.sum
+_obj/
+
+# Python artifacts
+.venv/
+dist/
+*.egg-info
+*.whl
+
+# C artifacts
+*.a
+*.so
+*.so.*
+*.dylib
+*.dll
+*.pc
+
+# Example dirs
+/examples/*/
+
+# Grammar volatiles
+*.wasm
+*.obj
+*.o

resources/language-metavariables/tree-sitter-c/Cargo.toml

@@ -0,0 +1,26 @@
+[package]
+name = "tree-sitter-c"
+description = "C grammar for tree-sitter"
+version = "0.21.4"
+authors = [
+  "Max Brunsfeld <maxbrunsfeld@gmail.com>",
+  "Amaan Qureshi <amaanq12@gmail.com>",
+]
+license = "MIT"
+keywords = ["incremental", "parsing", "tree-sitter", "c"]
+categories = ["parsing", "text-editors"]
+repository = "https://github.com/tree-sitter/tree-sitter-c"
+edition = "2021"
+autoexamples = false
+
+build = "bindings/rust/build.rs"
+include = ["bindings/rust/*", "grammar.js", "queries/*", "src/*"]
+
+[lib]
+path = "bindings/rust/lib.rs"
+
+[dependencies]
+tree-sitter = "~0.20"
+
+[build-dependencies]
+cc = "1.0.90"