feat(extraction): index Solidity (.sol) at the extraction tier

README.md

@@ -439,7 +439,7 @@ Omit `--install` to print to stdout. Full per-shell install paths and notes:
 
 ## What CodeGraph Does (and Doesn't)
 
-**Does:** deterministic code-structure extraction across 33 languages (TypeScript,
+**Does:** deterministic code-structure extraction across 34 languages (TypeScript,
 Python, Go, Rust, Java, C/C++, C#, Vue, Svelte, GDScript, and more — see
 [`docs/languages.md`](docs/languages.md)), cross-file resolution (including
 Godot project graphs), graph traversal, FTS5 search, whole-graph export with
@@ -454,9 +454,9 @@ beyond the fixed `LANGUAGES` set.
 
 ## Supported Languages
 
-CodeGraph supports **33 languages** grouped by extraction depth. Quick overview:
+CodeGraph supports **34 languages** grouped by extraction depth. Quick overview:
 
-- **Tier 1 — Full symbol extraction (24):** TypeScript, TSX, JavaScript, JSX, ArkTS, Python, Go, Rust, Java, C, C++, C#, PHP, Ruby, Swift, Kotlin, Dart, Scala, Lua, Luau, Objective-C, R, GDScript, Pascal.
+- **Tier 1 — Full symbol extraction (25):** TypeScript, TSX, JavaScript, JSX, ArkTS, Python, Go, Rust, Java, C, C++, C#, PHP, Ruby, Swift, Kotlin, Dart, Scala, Lua, Luau, Objective-C, R, Solidity, GDScript, Pascal.
 - **Tier 2 — Embedded / template extraction (6):** Vue, Svelte, Astro, Razor/`.cshtml`, Liquid, XML/MyBatis mapper.
 - **Tier 3 — File-level only (3):** YAML, Twig, Properties.
 

crates/codegraph-bench/fixtures/solidity/IERC20.sol

@@ -0,0 +1,6 @@
+// SPDX-License-Identifier: MIT
+pragma solidity ^0.8.0;
+
+interface IERC20 {
+    function transfer(address to, uint256 value) external returns (bool);
+}

crates/codegraph-bench/fixtures/solidity/Token.sol

@@ -0,0 +1,49 @@
+// SPDX-License-Identifier: MIT
+pragma solidity ^0.8.0;
+
+import "./IERC20.sol";
+
+error Unauthorized(address caller);
+
+uint256 constant MAX_SUPPLY = 1000000;
+
+contract Token is IERC20 {
+    uint256 public total;
+
+    event Transfer(address indexed from, uint256 value);
+
+    enum Status {
+        Active,
+        Closed
+    }
+
+    struct Holder {
+        address account;
+        uint256 balance;
+    }
+
+    modifier onlyOwner() {
+        _;
+    }
+
+    constructor() {
+        total = MAX_SUPPLY;
+    }
+
+    fallback() external {
+    }
+
+    receive() external payable {
+    }
+
+    function transfer(address to, uint256 value) public onlyOwner returns (bool) {
+        emit Transfer(to, value);
+        return true;
+    }
+}
+
+library Math {
+    function add(uint256 a, uint256 b) internal pure returns (uint256) {
+        return a + b;
+    }
+}

crates/codegraph-bench/tests/equivalence.rs

@@ -135,6 +135,27 @@ fn arkts_db_is_self_equivalent_to_arkts_golden() {
     assert_equivalent(&arkts_db(), &arkts_golden_dir()).unwrap();
 }
 
+#[test]
+fn generated_golden_matches_committed_solidity_fixture() {
+    // Guards Solidity extraction (upstream #1170): the `.sol`->Solidity mapping,
+    // contract/library->Class + interface->Interface + struct->Struct +
+    // enum->Enum, synthetic constructor/fallback/receive method names,
+    // state-var/struct-member/event/error->Field, `is`-inheritance->Extends
+    // (resolver promotes to Implements), and emit/modifier-guard call edges.
+    let tempdir = TestDir::new("generated-golden-solidity");
+    write_golden(&solidity_db(), tempdir.path()).unwrap();
+
+    let expected = load_golden(&solidity_golden_dir()).unwrap();
+    let actual = load_golden(tempdir.path()).unwrap();
+
+    diff_canonical(&expected, &actual, None).unwrap();
+}
+
+#[test]
+fn solidity_db_is_self_equivalent_to_solidity_golden() {
+    assert_equivalent(&solidity_db(), &solidity_golden_dir()).unwrap();
+}
+
 #[test]
 fn tier1_node_drift_is_reported() {
     let expected = load_golden(&mini_golden_dir()).unwrap();
@@ -237,6 +258,14 @@ fn arkts_golden_dir() -> PathBuf {
     workspace_root().join("reference/golden/arkts")
 }
 
+fn solidity_db() -> PathBuf {
+    workspace_root().join("reference/golden/solidity/colby.db")
+}
+
+fn solidity_golden_dir() -> PathBuf {
+    workspace_root().join("reference/golden/solidity")
+}
+
 struct TestDir {
     path: PathBuf,
 }

crates/codegraph-core/src/types.rs

@@ -49,7 +49,7 @@ pub const EDGE_KIND_STRINGS: [&str; 12] = [
     "decorates",
 ];
 
-pub const LANGUAGE_STRINGS: [&str; 37] = [
+pub const LANGUAGE_STRINGS: [&str; 38] = [
     "typescript",
     "javascript",
     "tsx",
@@ -78,6 +78,7 @@ pub const LANGUAGE_STRINGS: [&str; 37] = [
     "luau",
     "objc",
     "r",
+    "solidity",
     "yaml",
     "twig",
     "xml",
@@ -323,6 +324,8 @@ pub enum Language {
     ObjC,
     #[serde(rename = "r")]
     R,
+    #[serde(rename = "solidity")]
+    Solidity,
     #[serde(rename = "yaml")]
     Yaml,
     #[serde(rename = "twig")]
@@ -344,7 +347,7 @@ pub enum Language {
 }
 
 impl Language {
-    pub const ALL: [Self; 37] = [
+    pub const ALL: [Self; 38] = [
         Self::TypeScript,
         Self::JavaScript,
         Self::Tsx,
@@ -373,6 +376,7 @@ impl Language {
         Self::Luau,
         Self::ObjC,
         Self::R,
+        Self::Solidity,
         Self::Yaml,
         Self::Twig,
         Self::Xml,
@@ -414,6 +418,7 @@ impl Language {
             Self::Luau => "luau",
             Self::ObjC => "objc",
             Self::R => "r",
+            Self::Solidity => "solidity",
             Self::Yaml => "yaml",
             Self::Twig => "twig",
             Self::Xml => "xml",

crates/codegraph-extract/Cargo.toml

@@ -47,6 +47,7 @@ tree-sitter-r = "1.2.0"
 tree-sitter-ruby = "0.23.1"
 tree-sitter-rust = "0.24.2"
 tree-sitter-scala = "0.26.0"
+tree-sitter-solidity = "1.2.13"
 tree-sitter-swift = "0.7.3"
 tree-sitter-typescript = "0.23.2"
 tree-sitter-xml = "0.7.0"

crates/codegraph-extract/src/engine.rs

@@ -82,6 +82,7 @@ pub fn builtin_language_for_ext(ext: &str) -> Option<Language> {
         "luau" => Language::Luau,
         "m" | "mm" => Language::ObjC,
         "r" => Language::R,
+        "sol" => Language::Solidity,
         "yml" | "yaml" => Language::Yaml,
         "twig" => Language::Twig,
         "xml" => Language::Xml,
@@ -870,14 +871,19 @@ mod tests {
         assert_eq!(detect_language("s.metal"), Language::Cpp);
         assert_eq!(detect_language("k.cu"), Language::Cpp);
         assert_eq!(detect_language("k.cuh"), Language::Cpp);
-        assert_eq!(Language::ALL.len(), 37);
+        assert_eq!(Language::ALL.len(), 38);
     }
 
     #[test]
     fn arkts_extension_maps_to_arkts() {
         assert_eq!(detect_language("view.ets"), Language::ArkTs);
     }
 
+    #[test]
+    fn sol_maps_to_solidity() {
+        assert_eq!(detect_language("Token.sol"), Language::Solidity);
+    }
+
     #[test]
     fn plain_ts_stays_typescript() {
         assert_eq!(detect_language("m.ts"), Language::TypeScript);

crates/codegraph-extract/src/lang/mod.rs

@@ -24,6 +24,7 @@ mod r;
 mod ruby;
 mod rust;
 mod scala;
+mod solidity;
 mod swift;
 mod tsx;
 mod typescript;
@@ -54,6 +55,7 @@ pub use r::R_SPEC;
 pub use ruby::RUBY_SPEC;
 pub use rust::RUST_SPEC;
 pub use scala::SCALA_SPEC;
+pub use solidity::SOLIDITY_SPEC;
 pub use swift::SWIFT_SPEC;
 pub use tsx::TSX_SPEC;
 pub use typescript::TYPESCRIPT_SPEC;
@@ -83,6 +85,7 @@ pub fn spec_for_language(language: Language) -> Option<&'static dyn LanguageSpec
         Language::Pascal => Some(&PASCAL_SPEC),
         Language::ObjC => Some(&OBJC_SPEC),
         Language::R => Some(&R_SPEC),
+        Language::Solidity => Some(&SOLIDITY_SPEC),
         Language::Gdscript => Some(&GDSCRIPT_SPEC),
         _ => None,
     }