Auto-language detection

Cargo.toml

@@ -16,6 +16,7 @@ rayon = "1.7.0"
 regex = "1.8.2"
 termcolor = "1.2.0"
 tree-sitter = "0.20.10"
+tree-sitter-javascript = "0.20.0"
 tree-sitter-rust = "0.20.3"
 tree-sitter-typescript = "0.20.2"
 

src/language.rs

@@ -1,24 +1,30 @@
+use std::{collections::HashMap, ffi::OsStr, path::Path};
+
 use clap::ValueEnum;
 use tree_sitter::Language;
 
-#[derive(Copy, Clone, Debug, Eq, PartialEq, ValueEnum)]
+#[derive(Copy, Clone, Debug, Eq, PartialEq, ValueEnum, Hash)]
 pub enum SupportedLanguageName {
     Rust,
     Typescript,
+    Javascript,
 }
 
 impl SupportedLanguageName {
     pub fn get_language(self) -> Box<dyn SupportedLanguage> {
         match self {
             Self::Rust => Box::new(get_rust_language()),
             Self::Typescript => Box::new(get_typescript_language()),
+            Self::Javascript => Box::new(get_javascript_language()),
         }
     }
 }
 
 pub trait SupportedLanguage {
     fn language(&self) -> Language;
+    fn name(&self) -> SupportedLanguageName;
     fn name_for_ignore_select(&self) -> &'static str;
+    fn extensions(&self) -> Vec<&'static str>;
 }
 
 pub struct SupportedLanguageRust;
@@ -28,9 +34,17 @@ impl SupportedLanguage for SupportedLanguageRust {
         tree_sitter_rust::language()
     }
 
+    fn name(&self) -> SupportedLanguageName {
+        SupportedLanguageName::Rust
+    }
+
     fn name_for_ignore_select(&self) -> &'static str {
         "rust"
     }
+
+    fn extensions(&self) -> Vec<&'static str> {
+        vec!["rs"]
+    }
 }
 
 pub fn get_rust_language() -> SupportedLanguageRust {
@@ -44,11 +58,67 @@ impl SupportedLanguage for SupportedLanguageTypescript {
         tree_sitter_typescript::language_tsx()
     }
 
+    fn name(&self) -> SupportedLanguageName {
+        SupportedLanguageName::Typescript
+    }
+
     fn name_for_ignore_select(&self) -> &'static str {
         "ts"
     }
+
+    fn extensions(&self) -> Vec<&'static str> {
+        vec!["ts", "tsx"]
+    }
 }
 
 pub fn get_typescript_language() -> SupportedLanguageTypescript {
     SupportedLanguageTypescript
 }
+
+pub struct SupportedLanguageJavascript;
+
+impl SupportedLanguage for SupportedLanguageJavascript {
+    fn language(&self) -> Language {
+        tree_sitter_javascript::language()
+    }
+
+    fn name(&self) -> SupportedLanguageName {
+        SupportedLanguageName::Javascript
+    }
+
+    fn name_for_ignore_select(&self) -> &'static str {
+        "js"
+    }
+
+    fn extensions(&self) -> Vec<&'static str> {
+        vec!["js", "jsx", "vue", "cjs", "mjs"]
+    }
+}
+
+pub fn get_javascript_language() -> SupportedLanguageJavascript {
+    SupportedLanguageJavascript
+}
+
+pub fn get_all_supported_languages() -> HashMap<SupportedLanguageName, Box<dyn SupportedLanguage>> {
+    HashMap::from_iter([
+        (
+            SupportedLanguageName::Rust,
+            Box::new(get_rust_language()) as Box<dyn SupportedLanguage>,
+        ),
+        (
+            SupportedLanguageName::Typescript,
+            Box::new(get_typescript_language()) as Box<dyn SupportedLanguage>,
+        ),
+        (
+            SupportedLanguageName::Javascript,
+            Box::new(get_javascript_language()) as Box<dyn SupportedLanguage>,
+        ),
+    ])
+}
+
+pub fn maybe_supported_language_from_path(path: &Path) -> Option<Box<dyn SupportedLanguage>> {
+    let extension = path.extension().and_then(OsStr::to_str)?;
+    get_all_supported_languages()
+        .into_values()
+        .find(|language| language.extensions().contains(&extension))
+}

src/lib.rs

@@ -14,13 +14,18 @@ use once_cell::unsync::OnceCell;
 use rayon::iter::IterBridge;
 use rayon::prelude::*;
 use std::cell::RefCell;
+use std::collections::HashMap;
 use std::fs;
 use std::path::Path;
 use std::path::PathBuf;
+use std::process;
 use std::rc::Rc;
+use std::sync::atomic::AtomicU32;
+use std::sync::atomic::Ordering;
 use std::sync::mpsc;
 use std::sync::mpsc::Receiver;
 use std::sync::Arc;
+use std::sync::Mutex;
 use std::thread;
 use std::thread::JoinHandle;
 use termcolor::Buffer;
@@ -33,9 +38,12 @@ mod macros;
 mod plugin;
 mod treesitter;
 
-use language::{SupportedLanguage, SupportedLanguageName};
+use language::{
+    get_all_supported_languages, maybe_supported_language_from_path, SupportedLanguage,
+    SupportedLanguageName,
+};
 use plugin::get_loaded_filter;
-use treesitter::{get_matches, get_query};
+use treesitter::{get_matches, maybe_get_query};
 
 #[derive(Parser)]
 pub struct Args {
@@ -45,7 +53,7 @@ pub struct Args {
     #[arg(short, long = "capture")]
     pub capture_name: Option<String>,
     #[arg(short, long, value_enum)]
-    pub language: SupportedLanguageName,
+    pub language: Option<SupportedLanguageName>,
     #[arg(short, long)]
     pub filter: Option<String>,
     #[arg(short = 'a', long)]
@@ -91,25 +99,77 @@ fn get_output_mode(args: &Args) -> OutputMode {
     }
 }
 
+struct MaybeInitializedCaptureIndex(AtomicU32);
+
+impl MaybeInitializedCaptureIndex {
+    fn mark_failed(&self) {
+        self.0.store(u32::MAX - 1, Ordering::Relaxed);
+    }
+
+    pub fn get(&self) -> Result<Option<u32>, ()> {
+        let loaded = self.0.load(Ordering::Relaxed);
+        match loaded {
+            loaded if loaded == u32::MAX => Ok(None),
+            loaded if loaded == u32::MAX - 1 => Err(()),
+            loaded => Ok(Some(loaded)),
+        }
+    }
+
+    pub fn get_or_initialize(&self, query: &Query, capture_name: Option<&str>) -> Result<u32, ()> {
+        if let Some(already_initialized) = self.get()? {
+            return Ok(already_initialized);
+        }
+        let capture_index = match capture_name {
+            None => 0,
+            Some(capture_name) => {
+                let capture_index = query.capture_index_for_name(capture_name);
+                if capture_index.is_none() {
+                    self.mark_failed();
+                    fail(&format!("invalid capture name '{}'", capture_name));
+                }
+                capture_index.unwrap()
+            }
+        };
+        self.set(capture_index);
+        Ok(capture_index)
+    }
+
+    fn set(&self, capture_index: u32) {
+        self.0.store(capture_index, Ordering::Relaxed);
+    }
+}
+
+impl Default for MaybeInitializedCaptureIndex {
+    fn default() -> Self {
+        Self(AtomicU32::new(u32::MAX))
+    }
+}
+
 pub fn run(args: Args) {
     let query_source = match args.query_args.path_to_query_file.as_ref() {
         Some(path_to_query_file) => fs::read_to_string(path_to_query_file).unwrap(),
         None => args.query_args.query_source.clone().unwrap(),
     };
-    let supported_language = args.language.get_language();
-    let language = supported_language.language();
-    let query = Arc::new(get_query(&query_source, language));
-    let capture_index = args.capture_name.as_ref().map_or(0, |capture_name| {
-        query
-            .capture_index_for_name(capture_name)
-            .expect(&format!("Unknown capture name: `{}`", capture_name))
-    });
+    let specified_supported_language = args.language.map(|language| language.get_language());
+    let query_or_failure_by_language: Mutex<HashMap<SupportedLanguageName, Option<Arc<Query>>>> =
+        Default::default();
+    let capture_index = MaybeInitializedCaptureIndex::default();
     let output_mode = get_output_mode(&args);
     let buffer_writer = BufferWriter::stdout(ColorChoice::Never);
 
-    get_project_file_walker(&*supported_language, &args.use_paths())
+    get_project_file_walker(specified_supported_language.as_deref(), &args.use_paths())
         .into_parallel_iterator()
         .for_each(|project_file_dir_entry| {
+            let language = maybe_supported_language_from_path(project_file_dir_entry.path())
+                .expect("Walker should've been pre-filtered to just supported file types");
+            let query = return_if_none!(get_and_cache_query_for_language(
+                &query_source,
+                &query_or_failure_by_language,
+                &*language,
+            ));
+            let capture_index = return_if_none!(capture_index
+                .get_or_initialize(&query, args.capture_name.as_deref())
+                .ok());
             let printer = get_printer(&buffer_writer, output_mode);
             let mut printer = printer.borrow_mut();
             let path =
@@ -118,7 +178,7 @@ pub fn run(args: Args) {
             let matcher = TreeSitterMatcher::new(
                 &query,
                 capture_index,
-                language,
+                language.language(),
                 args.filter.clone(),
                 args.filter_arg.clone(),
             );
@@ -129,6 +189,38 @@ pub fn run(args: Args) {
                 .unwrap();
             buffer_writer.print(printer.get_mut()).unwrap();
         });
+
+    error_if_no_successful_query_parsing(&query_or_failure_by_language);
+}
+
+fn error_if_no_successful_query_parsing(
+    query_or_failure_by_language: &Mutex<HashMap<SupportedLanguageName, Option<Arc<Query>>>>,
+) {
+    let query_or_failure_by_language = query_or_failure_by_language.lock().unwrap();
+    if !query_or_failure_by_language
+        .values()
+        .any(|query| query.is_some())
+    {
+        fail("invalid query");
+    }
+}
+
+fn fail(message: &str) -> ! {
+    eprintln!("error: {message}");
+    process::exit(1);
+}
+
+fn get_and_cache_query_for_language(
+    query_source: &str,
+    query_or_failure_by_language: &Mutex<HashMap<SupportedLanguageName, Option<Arc<Query>>>>,
+    language: &dyn SupportedLanguage,
+) -> Option<Arc<Query>> {
+    query_or_failure_by_language
+        .lock()
+        .unwrap()
+        .entry(language.name())
+        .or_insert_with(|| maybe_get_query(query_source, language.language()).map(Arc::new))
+        .clone()
 }
 
 type Printer = Standard<Buffer>;
@@ -253,16 +345,22 @@ impl Iterator for WalkParallelIterator {
     }
 }
 
-fn get_project_file_walker(language: &dyn SupportedLanguage, paths: &[PathBuf]) -> WalkParallel {
+fn get_project_file_walker(
+    language: Option<&dyn SupportedLanguage>,
+    paths: &[PathBuf],
+) -> WalkParallel {
     assert!(!paths.is_empty());
     let mut builder = WalkBuilder::new(&paths[0]);
-    builder.types(
-        TypesBuilder::new()
-            .add_defaults()
-            .select(language.name_for_ignore_select())
-            .build()
-            .unwrap(),
-    );
+    let mut types_builder = TypesBuilder::new();
+    types_builder.add_defaults();
+    if let Some(language) = language {
+        types_builder.select(language.name_for_ignore_select());
+    } else {
+        for language in get_all_supported_languages().values() {
+            types_builder.select(language.name_for_ignore_select());
+        }
+    }
+    builder.types(types_builder.build().unwrap());
     for path in &paths[1..] {
         builder.add(path);
     }

src/macros.rs

@@ -3,5 +3,17 @@ macro_rules! regex {
     ($re:literal $(,)?) => {{
         static RE: once_cell::sync::OnceCell<regex::Regex> = once_cell::sync::OnceCell::new();
         RE.get_or_init(|| regex::Regex::new($re).unwrap())
-    }}
+    }};
+}
+
+#[macro_export]
+macro_rules! return_if_none {
+    ($expr:expr $(,)?) => {
+        match $expr {
+            None => {
+                return;
+            }
+            Some(expr) => expr,
+        }
+    };
 }

src/treesitter/mod.rs

@@ -11,8 +11,8 @@ pub fn get_parser(language: Language) -> Parser {
     parser
 }
 
-pub fn get_query(source: &str, language: Language) -> Query {
-    Query::new(language, source).unwrap()
+pub fn maybe_get_query(source: &str, language: Language) -> Option<Query> {
+    Query::new(language, source).ok()
 }
 
 pub fn get_matches(

tests/fixtures/mixed_project/javascript_src/index.js

@@ -0,0 +1 @@
+const js_foo = () => {}

tests/fixtures/mixed_project/rust_src/lib.rs

@@ -0,0 +1 @@
+fn foo() {}

tests/fixtures/mixed_project/typescript_src/index.tsx

@@ -0,0 +1 @@
+const foo = () => {}

tests/fixtures/rust_project/function-itemz.scm

@@ -0,0 +1 @@
+(function_itemz) @function_item