migrate helix-syntax crate into helix-core and helix-term

helix-syntax mostly existed for the sake of the build task which
checks and compiles the submodules. Since we won't be relying on
that process anymore, it doesn't end up making much sense to have
a very thin crate just for some functions that we could port to
helix-core.

The remaining build-related code is moved to helix-term which will
be able to provide grammar builds through the --build-grammars CLI
flag.

Cargo.lock

@@ -358,11 +358,12 @@ dependencies = [
 name = "helix-core"
 version = "0.6.0"
 dependencies = [
+ "anyhow",
  "arc-swap",
  "chrono",
  "encoding_rs",
  "etcetera",
- "helix-syntax",
+ "libloading",
  "log",
  "once_cell",
  "quickcheck",
@@ -415,22 +416,12 @@ dependencies = [
  "which",
 ]
 
-[[package]]
-name = "helix-syntax"
-version = "0.6.0"
-dependencies = [
- "anyhow",
- "cc",
- "libloading",
- "threadpool",
- "tree-sitter",
-]
-
 [[package]]
 name = "helix-term"
 version = "0.6.0"
 dependencies = [
  "anyhow",
+ "cc",
  "chrono",
  "content_inspector",
  "crossterm",
@@ -454,6 +445,7 @@ dependencies = [
  "serde_json",
  "signal-hook",
  "signal-hook-tokio",
+ "threadpool",
  "tokio",
  "tokio-stream",
  "toml",

Cargo.toml

@@ -4,7 +4,6 @@ members = [
   "helix-view",
   "helix-term",
   "helix-tui",
-  "helix-syntax",
   "helix-lsp",
   "helix-dap",
   "xtask",
@@ -14,10 +13,6 @@ default-members = [
   "helix-term"
 ]
 
-# Build helix-syntax in release mode to make the code path faster in development.
-# [profile.dev.package."helix-syntax"]
-# opt-level = 3
-
 [profile.dev]
 split-debuginfo = "unpacked"
 

docs/architecture.md

@@ -1,11 +1,10 @@
 
-| Crate        | Description                                            |
-| -----------  | -----------                                            |
-| helix-core   | Core editing primitives, functional.                   |
-| helix-syntax | Tree-sitter grammars                                   |
-| helix-lsp    | Language server client                                 |
-| helix-view   | UI abstractions for use in backends, imperative shell. |
-| helix-term   | Terminal UI                                            |
+| Crate        | Description                                             |
+| -----------  | -----------                                             |
+| helix-core   | Core editing primitives, functional.                    |
+| helix-lsp    | Language server client                                  |
+| helix-view   | UI abstractions for use in backends, imperative shell.  |
+| helix-term   | Terminal UI                                             |
 | helix-tui    | TUI primitives, forked from tui-rs, inspired by Cursive |
 
 

helix-core/Cargo.toml

@@ -13,8 +13,6 @@ include = ["src/**/*", "README.md"]
 [features]
 
 [dependencies]
-helix-syntax = { version = "0.6", path = "../helix-syntax" }
-
 ropey = "1.3"
 smallvec = "1.8"
 smartstring = "1.0.0"
@@ -40,5 +38,8 @@ encoding_rs = "0.8"
 
 chrono = { version = "0.4", default-features = false, features = ["alloc", "std"] }
 
+libloading = "0.7"
+anyhow = "1"
+
 [dev-dependencies]
 quickcheck = { version = "1", default-features = false }

helix-core/src/syntax.rs

@@ -7,7 +7,9 @@ use crate::{
     Rope, RopeSlice, Tendril,
 };
 
-pub use helix_syntax::get_language;
+use anyhow::{Context, Result};
+use libloading::{Library, Symbol};
+use tree_sitter::Language;
 
 use arc_swap::{ArcSwap, Guard};
 use slotmap::{DefaultKey as LayerId, HopSlotMap};
@@ -25,6 +27,34 @@ use std::{
 use once_cell::sync::{Lazy, OnceCell};
 use serde::{Deserialize, Serialize};
 
+#[cfg(unix)]
+pub const DYLIB_EXTENSION: &str = "so";
+
+#[cfg(windows)]
+pub const DYLIB_EXTENSION: &str = "dll";
+
+fn replace_dashes_with_underscores(name: &str) -> String {
+    name.replace('-', "_")
+}
+
+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);
+    library_path.set_extension(DYLIB_EXTENSION);
+
+    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)
+}
+
 fn deserialize_regex<'de, D>(deserializer: D) -> Result<Option<Regex>, D::Error>
 where
     D: serde::Deserializer<'de>,
@@ -426,7 +456,7 @@ impl LanguageConfiguration {
                 &injections_query,
                 &locals_query,
             )
-            .unwrap(); // TODO: avoid panic
+            .unwrap_or_else(|query_error| panic!("Could not parse queries for language {:?}. Are your grammars out of sync? Try running 'hx --fetch-grammars' and 'hx --build-grammars'. This query could not be parsed: {:?}", self.language_id, query_error));
 
             config.configure(scopes);
             Some(Arc::new(config))
@@ -2023,7 +2053,7 @@ mod test {
         );
 
         let loader = Loader::new(Configuration { language: vec![] });
-        let language = get_language(&crate::RUNTIME_DIR, "Rust").unwrap();
+        let language = get_language("Rust").unwrap();
 
         let query = Query::new(language, query_str).unwrap();
         let textobject = TextObjectQuery { query };

helix-syntax/Cargo.toml

@@ -1,21 +0,0 @@
-[package]
-name = "helix-syntax"
-version = "0.6.0"
-authors = ["Blaž Hrastnik <blaz@mxxn.io>"]
-edition = "2021"
-license = "MPL-2.0"
-description = "Tree-sitter grammars support"
-categories = ["editor"]
-repository = "https://github.com/helix-editor/helix"
-homepage = "https://helix-editor.com"
-include = ["src/**/*", "languages/**/*", "build.rs", "!**/docs/**/*", "!**/test/**/*", "!**/examples/**/*", "!**/build/**/*"]
-
-[dependencies]
-tree-sitter = "0.20"
-libloading = "0.7"
-anyhow = "1"
-
-[build-dependencies]
-cc = { version = "1" }
-threadpool = { version = "1.0" }
-anyhow = "1"

helix-syntax/README.md

@@ -1,13 +0,0 @@
-helix-syntax
-============
-
-Syntax highlighting for helix, (shallow) submodules resides here.
-
-Differences from nvim-treesitter
---------------------------------
-
-As the syntax are commonly ported from
-<https://github.com/nvim-treesitter/nvim-treesitter>.
-
-Note that we do not support the custom `#any-of` predicate which is
-supported by neovim so one needs to change it to `#match` with regex.

helix-syntax/src/lib.rs

@@ -1,31 +0,0 @@
-use anyhow::{Context, Result};
-use libloading::{Library, Symbol};
-use tree_sitter::Language;
-
-fn replace_dashes_with_underscores(name: &str) -> String {
-    name.replace('-', "_")
-}
-#[cfg(unix)]
-const DYLIB_EXTENSION: &str = "so";
-
-#[cfg(windows)]
-const DYLIB_EXTENSION: &str = "dll";
-
-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);
-
-    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)
-}

helix-term/Cargo.toml

@@ -66,5 +66,9 @@ grep-searcher = "0.1.8"
 # Remove once retain_mut lands in stable rust
 retain_mut = "0.1.7"
 
+# compiling grammars
+cc = { version = "1" }
+threadpool = { version = "1.0" }
+
 [target.'cfg(not(windows))'.dependencies]  # https://github.com/vorner/signal-hook/issues/100
 signal-hook-tokio = { version = "0.3", features = ["futures-v0_3"] }

helix-term/build.rs

@@ -14,5 +14,10 @@ fn main() {
         None => env!("CARGO_PKG_VERSION").into(),
     };
 
+    println!(
+        "cargo:rustc-env=BUILD_TARGET={}",
+        std::env::var("TARGET").unwrap()
+    );
+
     println!("cargo:rustc-env=VERSION_AND_GIT_HASH={}", version);
 }

helix-term/src/grammars.rs

@@ -6,11 +6,13 @@ use std::{
     process::Command,
 };
 
-use std::sync::mpsc::channel;
+use helix_core::syntax::DYLIB_EXTENSION;
 
-fn collect_tree_sitter_dirs(ignore: &[String]) -> Result<Vec<String>> {
+const BUILD_TARGET: &str = env!("BUILD_TARGET");
+
+pub fn collect_tree_sitter_dirs(ignore: &[String]) -> Result<Vec<String>> {
     let mut dirs = Vec::new();
-    let path = PathBuf::from(env!("CARGO_MANIFEST_DIR")).join("languages");
+    let path = PathBuf::from(env!("CARGO_MANIFEST_DIR")).join("../helix-syntax/languages");
 
     for entry in fs::read_dir(path)? {
         let entry = entry?;
@@ -32,12 +34,6 @@ fn collect_tree_sitter_dirs(ignore: &[String]) -> Result<Vec<String>> {
     Ok(dirs)
 }
 
-#[cfg(unix)]
-const DYLIB_EXTENSION: &str = "so";
-
-#[cfg(windows)]
-const DYLIB_EXTENSION: &str = "dll";
-
 fn build_library(src_path: &Path, language: &str) -> Result<()> {
     let header_path = src_path;
     // let grammar_path = src_path.join("grammar.json");
@@ -65,7 +61,12 @@ fn build_library(src_path: &Path, language: &str) -> Result<()> {
         return Ok(());
     }
     let mut config = cc::Build::new();
-    config.cpp(true).opt_level(2).cargo_metadata(false);
+    config
+        .cpp(true)
+        .opt_level(2)
+        .cargo_metadata(false)
+        .host(BUILD_TARGET)
+        .target(BUILD_TARGET);
     let compiler = config.get_compiler();
     let mut command = Command::new(compiler.path());
     command.current_dir(src_path);
@@ -148,9 +149,10 @@ fn mtime(path: &Path) -> Result<SystemTime> {
     Ok(fs::metadata(path)?.modified()?)
 }
 
-fn build_dir(dir: &str, language: &str) {
+pub fn build_dir(dir: &str, language: &str) {
     println!("Build language {}", language);
-    if PathBuf::from("languages")
+    if PathBuf::from(env!("CARGO_MANIFEST_DIR"))
+        .join("../helix-syntax/languages")
         .join(dir)
         .read_dir()
         .unwrap()
@@ -158,49 +160,16 @@ fn build_dir(dir: &str, language: &str) {
         .is_none()
     {
         eprintln!(
-            "The directory {} is empty, you probably need to use 'git submodule update --init --recursive'?",
+            "The directory {} is empty, you probably need to use './scripts/grammars sync'?",
             dir
         );
         std::process::exit(1);
     }
 
     let path = PathBuf::from(env!("CARGO_MANIFEST_DIR"))
-        .join("languages")
+        .join("../helix-syntax/languages")
         .join(dir)
         .join("src");
 
     build_library(&path, language).unwrap();
 }
-
-fn main() {
-    let ignore = vec![
-        "tree-sitter-typescript".to_string(),
-        "tree-sitter-ocaml".to_string(),
-    ];
-    let dirs = collect_tree_sitter_dirs(&ignore).unwrap();
-
-    let mut n_jobs = 0;
-    let pool = threadpool::Builder::new().build(); // by going through the builder, it'll use num_cpus
-    let (tx, rx) = channel();
-
-    for dir in dirs {
-        let tx = tx.clone();
-        n_jobs += 1;
-
-        pool.execute(move || {
-            let language = &dir.strip_prefix("tree-sitter-").unwrap();
-            build_dir(&dir, language);
-
-            // report progress
-            tx.send(1).unwrap();
-        });
-    }
-    pool.join();
-    // drop(tx);
-    assert_eq!(rx.try_iter().sum::<usize>(), n_jobs);
-
-    build_dir("tree-sitter-typescript/tsx", "tsx");
-    build_dir("tree-sitter-typescript/typescript", "typescript");
-    build_dir("tree-sitter-ocaml/ocaml", "ocaml");
-    build_dir("tree-sitter-ocaml/interface", "ocaml-interface")
-}

helix-term/src/lib.rs

@@ -7,6 +7,7 @@ pub mod commands;
 pub mod compositor;
 pub mod config;
 pub mod health;
+pub mod grammars;
 pub mod job;
 pub mod keymap;
 pub mod ui;