Dynamically load grammar libraries at runtime

1 files changed, 30 insertions, 85 deletions

diff --git a/helix-syntax/src/lib.rs b/helix-syntax/src/lib.rs
index 5e3bb3ea..b6c0ecf3 100644
--- a/helix-syntax/src/lib.rs
+++ b/helix-syntax/src/lib.rs
@@ -1,94 +1,39 @@
-use serde::{Deserialize, Serialize};
+use anyhow::{Context, Result};
+use libloading::{Library, Symbol};
 use tree_sitter::Language;
 
-#[macro_export]
-macro_rules! mk_extern {
-    ( $( $name:ident ),* ) => {
-        $(
-            extern "C" { pub fn $name() -> Language; }
-        )*
-    };
-}
-
-#[macro_export]
-macro_rules! mk_enum {
-    ( $( $camel:ident ),* ) => {
-        #[derive(Clone, Copy, Debug, PartialEq, Serialize, Deserialize)]
-        #[serde(rename_all = "lowercase")]
-        pub enum Lang {
-            $(
-                $camel,
-            )*
+fn replace_dashes_with_underscores(name: &str) -> String {
+    let mut result = String::with_capacity(name.len());
+    for c in name.chars() {
+        if c == '-' {
+            result.push('_');
+        } else {
+            result.push(c);
         }
-    };
+    }
+    result
 }
+#[cfg(unix)]
+const DYLIB_EXTENSION: &str = "so";
 
-#[macro_export]
-macro_rules! mk_get_language {
-    ( $( ($camel:ident, $name:ident) ),* ) => {
-        #[must_use]
-        pub fn get_language(lang: Lang) -> Language {
-            unsafe {
-                match lang {
-                    $(
-                        Lang::$camel => $name(),
-                    )*
-                }
-            }
-        }
-    };
-}
+#[cfg(windows)]
+const DYLIB_EXTENSION: &str = "dll";
 
-#[macro_export]
-macro_rules! mk_get_language_name {
-    ( $( $camel:ident ),* ) => {
-        #[must_use]
-        pub const fn get_language_name(lang: Lang) -> &'static str {
-            match lang {
-                $(
-                    Lang::$camel => stringify!($camel),
-                )*
-            }
-        }
-    };
-}
+pub fn get_language(runtime_path: &std::path::Path, name: &str) -> Result<Language> {
+    let name = name.to_ascii_lowercase();
+    let mut library_path = runtime_path.join("grammars").join(&name);
+    // TODO: duplicated under build
+    library_path.set_extension(DYLIB_EXTENSION);
 
-#[macro_export]
-macro_rules! mk_langs {
-    ( $( ($camel:ident, $name:ident) ),* ) => {
-        mk_extern!($( $name ),*);
-        mk_enum!($( $camel ),*);
-        mk_get_language!($( ($camel, $name) ),*);
-        mk_get_language_name!($( $camel ),*);
+    let library = unsafe { Library::new(&library_path) }
+        .with_context(|| format!("Error opening dynamic library {:?}", &library_path))?;
+    let language_fn_name = format!("tree_sitter_{}", replace_dashes_with_underscores(&name));
+    let language = unsafe {
+        let language_fn: Symbol<unsafe extern "C" fn() -> Language> = library
+            .get(language_fn_name.as_bytes())
+            .with_context(|| format!("Failed to load symbol {}", language_fn_name))?;
+        language_fn()
     };
+    std::mem::forget(library);
+    Ok(language)
 }
-
-mk_langs!(
-    // 1) Name for enum
-    // 2) tree-sitter function to call to get a Language
-    (Agda, tree_sitter_agda),
-    (Bash, tree_sitter_bash),
-    (Cpp, tree_sitter_cpp),
-    (CSharp, tree_sitter_c_sharp),
-    (Css, tree_sitter_css),
-    (C, tree_sitter_c),
-    (Elixir, tree_sitter_elixir),
-    (Go, tree_sitter_go),
-    // (Haskell, tree_sitter_haskell),
-    (Html, tree_sitter_html),
-    (Javascript, tree_sitter_javascript),
-    (Java, tree_sitter_java),
-    (Json, tree_sitter_json),
-    (Julia, tree_sitter_julia),
-    (Latex, tree_sitter_latex),
-    (Nix, tree_sitter_nix),
-    (Php, tree_sitter_php),
-    (Python, tree_sitter_python),
-    (Ruby, tree_sitter_ruby),
-    (Rust, tree_sitter_rust),
-    (Scala, tree_sitter_scala),
-    (Swift, tree_sitter_swift),
-    (Toml, tree_sitter_toml),
-    (Tsx, tree_sitter_tsx),
-    (Typescript, tree_sitter_typescript)
-);