Make Fortran, Verilog, and F# parsers opt-in

Cargo.toml

@@ -90,8 +90,8 @@ tree-sitter-elixir = "0.3.4"
 tree-sitter-elm = "5.8.0"
 tree-sitter-erlang = "0.20.0"
 tree-sitter-fish = "3.6.0"
-tree-sitter-fortran = "0.6.0"
-tree-sitter-fsharp = "0.3.0"
+tree-sitter-fortran = { version = "0.6.0", optional = true }
+tree-sitter-fsharp = { version = "0.3.0", optional = true }
 tree-sitter-gleam = "1.0.0"
 tree-sitter-go = "0.25.0"
 tree-sitter-haskell = "0.23.1"
@@ -124,7 +124,7 @@ tree-sitter-solidity = "1.2.13"
 tree-sitter-swift = "0.7.1"
 tree-sitter-toml-ng = "0.7.0"
 tree-sitter-typescript = "0.23.2"
-tree-sitter-verilog = "1.0.3"
+tree-sitter-verilog = { version = "1.0.3", optional = true }
 tree-sitter-vhdl = "1.4.0"
 tree-sitter-xml = "0.7.0"
 tree-sitter-yaml = "0.7.0"
@@ -235,4 +235,8 @@ name = "wasm"
 required-features = ["wasm-plugin-tests"]
 
 [features]
+all-languages = ["lang-fortran", "lang-fsharp", "lang-verilog"]
+lang-fortran = ["dep:tree-sitter-fortran"]
+lang-fsharp = ["dep:tree-sitter-fsharp"]
+lang-verilog = ["dep:tree-sitter-verilog"]
 wasm-plugin-tests = []

src/parse/guess_language.rs

@@ -38,7 +38,9 @@ pub(crate) enum Language {
     EmacsLisp,
     Erlang,
     Fish,
+    #[cfg(feature = "lang-fsharp")]
     FSharp,
+    #[cfg(feature = "lang-fortran")]
     Fortran,
     Gleam,
     Go,
@@ -79,6 +81,7 @@ pub(crate) enum Language {
     Toml,
     TypeScript,
     TypeScriptTsx,
+    #[cfg(feature = "lang-verilog")]
     Verilog,
     Vhdl,
     Xml,
@@ -139,7 +142,9 @@ pub(crate) fn language_name(language: Language) -> &'static str {
         EmacsLisp => "Emacs Lisp",
         Erlang => "Erlang",
         Fish => "Fish",
+        #[cfg(feature = "lang-fsharp")]
         FSharp => "F#",
+        #[cfg(feature = "lang-fortran")]
         Fortran => "Fortran",
         Gleam => "Gleam",
         Go => "Go",
@@ -180,6 +185,7 @@ pub(crate) fn language_name(language: Language) -> &'static str {
         Toml => "TOML",
         TypeScript => "TypeScript",
         TypeScriptTsx => "TypeScript TSX",
+        #[cfg(feature = "lang-verilog")]
         Verilog => "Verilog",
         Vhdl => "VHDL",
         Xml => "XML",
@@ -289,7 +295,9 @@ pub(crate) fn language_globs(language: Language) -> Vec<glob::Pattern> {
             "rebar.lock",
         ],
         Fish => &["*.fish"],
+        #[cfg(feature = "lang-fsharp")]
         FSharp => &["*.fs", "*.fsx", "*.fsi"],
+        #[cfg(feature = "lang-fortran")]
         Fortran => &["*.f", "*.for", "*.f90", "*.F", "*.FOR", "*.F90"],
         Gleam => &["*.gleam"],
         Go => &["*.go"],
@@ -396,6 +404,7 @@ pub(crate) fn language_globs(language: Language) -> Vec<glob::Pattern> {
         ],
         TypeScript => &["*.ts", "*.cts", "*.mts"],
         TypeScriptTsx => &["*.tsx"],
+        #[cfg(feature = "lang-verilog")]
         Verilog => &["*.v", "*.sv", "*.vh"],
         Vhdl => &["*.vhdl", "*.vhd"],
         Xml => &[
@@ -545,7 +554,9 @@ fn from_emacs_mode_header(src: &str) -> Option<Language> {
             "elm" => Elm,
             "emacs-lisp" => EmacsLisp,
             "fish" => Fish,
+            #[cfg(feature = "lang-fsharp")]
             "fsharp" => FSharp,
+            #[cfg(feature = "lang-fortran")]
             "fortran" => Fortran,
             "gleam" => Gleam,
             "go" => Go,
@@ -572,6 +583,7 @@ fn from_emacs_mode_header(src: &str) -> Option<Language> {
             "toml" => Toml,
             "tuareg" => OCaml,
             "typescript" => TypeScript,
+            #[cfg(feature = "lang-verilog")]
             "verilog" => Verilog,
             "vhdl" => Vhdl,
             "yaml" => Yaml,

src/parse/tree_sitter_parser.rs

@@ -456,6 +456,7 @@ fn build_config(language: guess::Language) -> TreeSitterConfig {
                 sub_languages: vec![],
             }
         }
+        #[cfg(feature = "lang-fsharp")]
         FSharp => {
             let language_fn = tree_sitter_fsharp::LANGUAGE_FSHARP;
             let language = tree_sitter::Language::new(language_fn);
@@ -471,6 +472,7 @@ fn build_config(language: guess::Language) -> TreeSitterConfig {
                 sub_languages: vec![],
             }
         }
+        #[cfg(feature = "lang-fortran")]
         Fortran => {
             let language_fn = tree_sitter_fortran::LANGUAGE;
             let language = tree_sitter::Language::new(language_fn);
@@ -1246,6 +1248,7 @@ fn build_config(language: guess::Language) -> TreeSitterConfig {
                 sub_languages: vec![],
             }
         }
+        #[cfg(feature = "lang-verilog")]
         Verilog => {
             let language_fn = tree_sitter_verilog::LANGUAGE;
             let language = tree_sitter::Language::new(language_fn);

tests/cli.rs

@@ -25,6 +25,60 @@ fn list_languages() {
     cmd.assert().stdout(predicate_fn);
 }
 
+#[test]
+fn optional_languages_follow_build_features() {
+    let listed = debug_command()
+        .arg("--list-languages")
+        .assert()
+        .success()
+        .get_output()
+        .stdout
+        .clone();
+    let listed = String::from_utf8(listed).unwrap();
+    let dir = tempfile::tempdir().unwrap();
+    let config = dir.path().join("config.toml");
+    std::fs::write(&config, "[plugins]\norder = []\n").unwrap();
+
+    for (name, fixture, extension, enabled) in [
+        ("Fortran", "fortran", "f90", cfg!(feature = "lang-fortran")),
+        ("Verilog", "verilog", "sv", cfg!(feature = "lang-verilog")),
+        ("F#", "f_sharp", "fs", cfg!(feature = "lang-fsharp")),
+    ] {
+        assert_eq!(listed.contains(name), enabled, "{name}");
+        let output = get_base_command()
+            .arg("--config")
+            .arg(&config)
+            .args([
+                "--format",
+                "ndjson",
+                "--no-index",
+                &format!("sample_files/{fixture}_1.{extension}"),
+                &format!("sample_files/{fixture}_2.{extension}"),
+            ])
+            .assert()
+            .success()
+            .get_output()
+            .stdout
+            .clone();
+        let events: Vec<serde_json::Value> = std::str::from_utf8(&output)
+            .unwrap()
+            .lines()
+            .map(|line| serde_json::from_str(line).unwrap())
+            .collect();
+        let file = events.iter().find(|event| event["type"] == "file").unwrap();
+        assert_eq!(file["diff"]["type"], "text", "{name}: {file}");
+        assert_eq!(
+            file["diff"]["stats"]["fallback"]["code"].as_str(),
+            if enabled {
+                None
+            } else {
+                Some("unsupported_language")
+            },
+            "{name}",
+        );
+    }
+}
+
 #[test]
 fn dump_tree_sitter() {
     let mut cmd = debug_command();