feat(extraction): index CFML (.cfc/.cfm/.cfs) 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 37 languages (TypeScript,
+**Does:** deterministic code-structure extraction across 38 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 **37 languages** grouped by extraction depth. Quick overview:
+CodeGraph supports **38 languages** grouped by extraction depth. Quick overview:
 
-- **Tier 1 — Full symbol extraction (28):** 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, Erlang, GDScript, Pascal.
+- **Tier 1 — Full symbol extraction (29):** 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, Erlang, CFML, 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/cfml/Base.cfc

@@ -0,0 +1,5 @@
+component {
+  public function ping() {
+    return true;
+  }
+}

crates/codegraph-bench/fixtures/cfml/Gadget.cfs

@@ -0,0 +1,6 @@
+component extends="Base" {
+  property name="x";
+  public function doThing() {
+    helper();
+  }
+}

crates/codegraph-bench/fixtures/cfml/Widget.cfm

@@ -0,0 +1,5 @@
+<cfcomponent name="Widget" extends="Base">
+  <cffunction name="doThing" access="public" returntype="void">
+    <cfset x = 1>
+  </cffunction>
+</cfcomponent>

crates/codegraph-bench/tests/equivalence.rs

@@ -230,6 +230,31 @@ fn erlang_db_is_self_equivalent_to_erlang_golden() {
     assert_equivalent(&erlang_db(), &erlang_golden_dir()).unwrap();
 }
 
+#[test]
+fn generated_golden_matches_committed_cfml_fixture() {
+    // Guards CFML extraction (upstream #1153, scope-B extraction slice): the
+    // `.cfc`/`.cfm`/`.cfs`->Cfml mapping, the dual-grammar dialect switch
+    // (`is_bare_script_cfml`: cfscript for bare-script, cfml tag grammar for
+    // tag files), a bare-script `component`->Class named from the FILE (unnamed
+    // in the grammar) + script-style `extends`->Extends, a tag
+    // `<cfcomponent>`->Class from `name` attr + `<cffunction>`->Method with
+    // access/returntype + tag `extends`->Extends. Both extends resolve to
+    // `Base.cfc`; `helper()` stays unresolved. The `<cfscript>`-in-tag delegation
+    // + cfquery SQL-body extraction + framework resolvers are DEFERRED.
+    let tempdir = TestDir::new("generated-golden-cfml");
+    write_golden(&cfml_db(), tempdir.path()).unwrap();
+
+    let expected = load_golden(&cfml_golden_dir()).unwrap();
+    let actual = load_golden(tempdir.path()).unwrap();
+
+    diff_canonical(&expected, &actual, None).unwrap();
+}
+
+#[test]
+fn cfml_db_is_self_equivalent_to_cfml_golden() {
+    assert_equivalent(&cfml_db(), &cfml_golden_dir()).unwrap();
+}
+
 #[test]
 fn tier1_node_drift_is_reported() {
     let expected = load_golden(&mini_golden_dir()).unwrap();
@@ -364,6 +389,14 @@ fn erlang_golden_dir() -> PathBuf {
     workspace_root().join("reference/golden/erlang")
 }
 
+fn cfml_db() -> PathBuf {
+    workspace_root().join("reference/golden/cfml/colby.db")
+}
+
+fn cfml_golden_dir() -> PathBuf {
+    workspace_root().join("reference/golden/cfml")
+}
+
 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; 41] = [
+pub const LANGUAGE_STRINGS: [&str; 42] = [
     "typescript",
     "javascript",
     "tsx",
@@ -82,6 +82,7 @@ pub const LANGUAGE_STRINGS: [&str; 41] = [
     "nix",
     "terraform",
     "erlang",
+    "cfml",
     "yaml",
     "twig",
     "xml",
@@ -335,6 +336,8 @@ pub enum Language {
     Terraform,
     #[serde(rename = "erlang")]
     Erlang,
+    #[serde(rename = "cfml")]
+    Cfml,
     #[serde(rename = "yaml")]
     Yaml,
     #[serde(rename = "twig")]
@@ -356,7 +359,7 @@ pub enum Language {
 }
 
 impl Language {
-    pub const ALL: [Self; 41] = [
+    pub const ALL: [Self; 42] = [
         Self::TypeScript,
         Self::JavaScript,
         Self::Tsx,
@@ -389,6 +392,7 @@ impl Language {
         Self::Nix,
         Self::Terraform,
         Self::Erlang,
+        Self::Cfml,
         Self::Yaml,
         Self::Twig,
         Self::Xml,
@@ -434,6 +438,7 @@ impl Language {
             Self::Nix => "nix",
             Self::Terraform => "terraform",
             Self::Erlang => "erlang",
+            Self::Cfml => "cfml",
             Self::Yaml => "yaml",
             Self::Twig => "twig",
             Self::Xml => "xml",

crates/codegraph-extract/Cargo.toml

@@ -26,6 +26,7 @@ tracing = { workspace = true }
 tree-sitter-arkts = "0.2.0"
 tree-sitter-c = "0.24.2"
 tree-sitter-c-sharp = "0.23.5"
+tree-sitter-cfml = "0.26.30"
 tree-sitter-cpp = "0.23.4"
 tree-sitter-css = "0.25.0"
 tree-sitter-dart = "0.2.0"

crates/codegraph-extract/src/engine.rs

@@ -86,6 +86,7 @@ pub fn builtin_language_for_ext(ext: &str) -> Option<Language> {
         "nix" => Language::Nix,
         "tf" | "tfvars" | "tofu" => Language::Terraform,
         "erl" | "hrl" => Language::Erlang,
+        "cfc" | "cfm" | "cfs" => Language::Cfml,
         "yml" | "yaml" => Language::Yaml,
         "twig" => Language::Twig,
         "xml" => Language::Xml,
@@ -203,7 +204,15 @@ pub fn extract_source(
     };
 
     let mut parser = Parser::new();
-    let ts_language = spec.tree_sitter_language();
+    // Run `pre_parse` first so the grammar selection can see the (possibly
+    // rewritten) source. Only CFML overrides `tree_sitter_language_for_source`
+    // to pick between its cfscript / cfml tag grammars per file dialect; every
+    // other spec inherits the default (`tree_sitter_language`), so this is a
+    // behavior-neutral reorder (their `pre_parse` is the identity default, and
+    // the two that override it — C++/embedded — don't override the grammar
+    // hook, so their parse path is byte-identical).
+    let parsed_source = spec.pre_parse(source, file_path);
+    let ts_language = spec.tree_sitter_language_for_source(&parsed_source);
     if let Err(error) = parser.set_language(&ts_language) {
         return ExtractionResult {
             nodes: Vec::new(),
@@ -213,7 +222,6 @@ pub fn extract_source(
             duration_ms: start.elapsed().as_millis() as i64,
         };
     }
-    let parsed_source = spec.pre_parse(source, file_path);
     let Some(tree) = parser.parse(&parsed_source, None) else {
         return ExtractionResult {
             nodes: Vec::new(),
@@ -874,7 +882,7 @@ 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(), 41);
+        assert_eq!(Language::ALL.len(), 42);
     }
 
     #[test]
@@ -905,6 +913,13 @@ mod tests {
         assert_eq!(detect_language("defs.hrl"), Language::Erlang);
     }
 
+    #[test]
+    fn cfml_extensions_map_to_cfml() {
+        assert_eq!(detect_language("Widget.cfc"), Language::Cfml);
+        assert_eq!(detect_language("page.cfm"), Language::Cfml);
+        assert_eq!(detect_language("Gadget.cfs"), Language::Cfml);
+    }
+
     #[test]
     fn plain_ts_stays_typescript() {
         assert_eq!(detect_language("m.ts"), Language::TypeScript);