feat(extraction): index Terraform/HCL (.tf/.tfvars/.tofu) 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 35 languages (TypeScript,
+**Does:** deterministic code-structure extraction across 36 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 **35 languages** grouped by extraction depth. Quick overview:
+CodeGraph supports **36 languages** grouped by extraction depth. Quick overview:
 
-- **Tier 1 — Full symbol extraction (26):** TypeScript, TSX, JavaScript, JSX, ArkTS, Python, Go, Rust, Java, C, C++, C#, PHP, Ruby, Swift, Kotlin, Dart, Scala, Lua, Luau, Objective-C, R, Solidity, Nix, GDScript, Pascal.
+- **Tier 1 — Full symbol extraction (27):** TypeScript, TSX, JavaScript, JSX, ArkTS, Python, Go, Rust, Java, C, C++, C#, PHP, Ruby, Swift, Kotlin, Dart, Scala, Lua, Luau, Objective-C, R, Solidity, Nix, Terraform, 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/terraform/main.tf

@@ -0,0 +1,30 @@
+terraform {}
+
+provider "aws" {}
+
+variable "region" {}
+
+locals {
+  name   = "demo"
+  tagged = var.region
+}
+
+data "aws_ami" "ubuntu" {}
+
+resource "aws_s3_bucket" "b" {
+  bucket     = var.region
+  kms_key_id = aws_kms_key.logs.arn
+}
+
+module "vpc" {
+  source = "./vpc"
+  region = var.region
+}
+
+output "bucket_id" {
+  value = aws_s3_bucket.b.id
+}
+
+output "vpc_id" {
+  value = module.vpc.id
+}

crates/codegraph-bench/tests/equivalence.rs

@@ -179,6 +179,31 @@ fn nix_db_is_self_equivalent_to_nix_golden() {
     assert_equivalent(&nix_db(), &nix_golden_dir()).unwrap();
 }
 
+#[test]
+fn generated_golden_matches_committed_terraform_fixture() {
+    // Guards Terraform/HCL extraction (upstream #1173, extraction slice only):
+    // the `.tf`/`.tfvars`/`.tofu`->Terraform mapping, block-type dispatch
+    // (resource/data->Class, module->Module, variable/output->Variable,
+    // provider->Namespace, locals->Constant per attr) with qualified names, and
+    // plain attribute-expression traversal refs
+    // (var.X/local.X/module.M/data.T.N/<type>.<name>)->References with built-ins
+    // skipped. The module-boundary TerraformResolver, emitModuleWiring's
+    // :-scoped refs, and the .tfvars var ref are DEFERRED, so no :-scoped ref is
+    // emitted; the undeclared aws_kms_key.logs stays unresolved.
+    let tempdir = TestDir::new("generated-golden-terraform");
+    write_golden(&terraform_db(), tempdir.path()).unwrap();
+
+    let expected = load_golden(&terraform_golden_dir()).unwrap();
+    let actual = load_golden(tempdir.path()).unwrap();
+
+    diff_canonical(&expected, &actual, None).unwrap();
+}
+
+#[test]
+fn terraform_db_is_self_equivalent_to_terraform_golden() {
+    assert_equivalent(&terraform_db(), &terraform_golden_dir()).unwrap();
+}
+
 #[test]
 fn tier1_node_drift_is_reported() {
     let expected = load_golden(&mini_golden_dir()).unwrap();
@@ -297,6 +322,14 @@ fn nix_golden_dir() -> PathBuf {
     workspace_root().join("reference/golden/nix")
 }
 
+fn terraform_db() -> PathBuf {
+    workspace_root().join("reference/golden/terraform/colby.db")
+}
+
+fn terraform_golden_dir() -> PathBuf {
+    workspace_root().join("reference/golden/terraform")
+}
+
 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; 39] = [
+pub const LANGUAGE_STRINGS: [&str; 40] = [
     "typescript",
     "javascript",
     "tsx",
@@ -80,6 +80,7 @@ pub const LANGUAGE_STRINGS: [&str; 39] = [
     "r",
     "solidity",
     "nix",
+    "terraform",
     "yaml",
     "twig",
     "xml",
@@ -329,6 +330,8 @@ pub enum Language {
     Solidity,
     #[serde(rename = "nix")]
     Nix,
+    #[serde(rename = "terraform")]
+    Terraform,
     #[serde(rename = "yaml")]
     Yaml,
     #[serde(rename = "twig")]
@@ -350,7 +353,7 @@ pub enum Language {
 }
 
 impl Language {
-    pub const ALL: [Self; 39] = [
+    pub const ALL: [Self; 40] = [
         Self::TypeScript,
         Self::JavaScript,
         Self::Tsx,
@@ -381,6 +384,7 @@ impl Language {
         Self::R,
         Self::Solidity,
         Self::Nix,
+        Self::Terraform,
         Self::Yaml,
         Self::Twig,
         Self::Xml,
@@ -424,6 +428,7 @@ impl Language {
             Self::R => "r",
             Self::Solidity => "solidity",
             Self::Nix => "nix",
+            Self::Terraform => "terraform",
             Self::Yaml => "yaml",
             Self::Twig => "twig",
             Self::Xml => "xml",

crates/codegraph-extract/Cargo.toml

@@ -31,6 +31,7 @@ tree-sitter-css = "0.25.0"
 tree-sitter-dart = "0.2.0"
 tree-sitter-go = "0.25.0"
 tree-sitter-gdscript = "6.1.0"
+tree-sitter-hcl = "1.1.0"
 tree-sitter-html = "0.23.2"
 tree-sitter-java = "0.23.5"
 tree-sitter-javascript = "0.25.0"

crates/codegraph-extract/src/engine.rs

@@ -84,6 +84,7 @@ pub fn builtin_language_for_ext(ext: &str) -> Option<Language> {
         "r" => Language::R,
         "sol" => Language::Solidity,
         "nix" => Language::Nix,
+        "tf" | "tfvars" | "tofu" => Language::Terraform,
         "yml" | "yaml" => Language::Yaml,
         "twig" => Language::Twig,
         "xml" => Language::Xml,
@@ -872,7 +873,7 @@ 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(), 39);
+        assert_eq!(Language::ALL.len(), 40);
     }
 
     #[test]
@@ -890,6 +891,13 @@ mod tests {
         assert_eq!(detect_language("flake.nix"), Language::Nix);
     }
 
+    #[test]
+    fn terraform_extensions_map_to_terraform() {
+        assert_eq!(detect_language("main.tf"), Language::Terraform);
+        assert_eq!(detect_language("prod.tfvars"), Language::Terraform);
+        assert_eq!(detect_language("main.tofu"), Language::Terraform);
+    }
+
     #[test]
     fn plain_ts_stays_typescript() {
         assert_eq!(detect_language("m.ts"), Language::TypeScript);

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

@@ -27,6 +27,7 @@ mod rust;
 mod scala;
 mod solidity;
 mod swift;
+mod terraform;
 mod tsx;
 mod typescript;
 
@@ -64,6 +65,11 @@ pub use rust::RUST_SPEC;
 pub use scala::SCALA_SPEC;
 pub use solidity::SOLIDITY_SPEC;
 pub use swift::SWIFT_SPEC;
+pub use terraform::TERRAFORM_SPEC;
+pub(crate) use terraform::{
+    TerraformBlockDecl, collect_terraform_references, describe_terraform_block,
+    read_terraform_block_header, terraform_block_body,
+};
 pub use tsx::TSX_SPEC;
 pub use typescript::TYPESCRIPT_SPEC;
 
@@ -94,6 +100,7 @@ pub fn spec_for_language(language: Language) -> Option<&'static dyn LanguageSpec
         Language::R => Some(&R_SPEC),
         Language::Solidity => Some(&SOLIDITY_SPEC),
         Language::Nix => Some(&NIX_SPEC),
+        Language::Terraform => Some(&TERRAFORM_SPEC),
         Language::Gdscript => Some(&GDSCRIPT_SPEC),
         _ => None,
     }