Import tree sitter queries.

3 years ago · 71c06c11cb
parent 52da68e49a
commit 71c06c11cb
28 changed files with 357 additions and 37 deletions
--- a/.gitmodules
+++ b/.gitmodules
@ -82,6 +82,3 @@
 	path = helix-syntax/languages/tree-sitter-toml
 	url = https://github.com/ikatyang/tree-sitter-toml
 	shallow = true
-[submodule "helix-syntax/nvim-treesitter"]
-	path = helix-syntax/nvim-treesitter
-	url = https://github.com/nvim-treesitter/nvim-treesitter
--- a/Cargo.lock
+++ b/Cargo.lock
@ -486,7 +486,6 @@ version = "0.1.0"
 dependencies = [
 "anyhow",
 "helix-syntax",
- "log",
 "once_cell",
 "regex",
 "ropey",
@ -1227,9 +1226,9 @@ dependencies = [

 [[package]]
 name = "tree-sitter"
-version = "0.17.1"
+version = "0.19.3"
 source = "registry+https://github.com/rust-lang/crates.io-index"
-checksum = "d18dcb776d3affaba6db04d11d645946d34a69b3172e588af96ce9fecd20faac"
+checksum = "1f41201fed3db3b520405a9c01c61773a250d4c3f43e9861c14b2bb232c981ab"
 dependencies = [
 "cc",
 "regex",
--- a/helix-core/Cargo.toml
+++ b/helix-core/Cargo.toml
@ -16,7 +16,7 @@ tendril = "0.4.2"
 unicode-segmentation = "1.6"
 unicode-width = "0.1"
 # slab = "0.4.2"
-tree-sitter = "0.17"
+tree-sitter = "0.19"
 once_cell = "1.4"
 regex = "1"

--- a/helix-core/src/syntax.rs
+++ b/helix-core/src/syntax.rs
@ -1,4 +1,4 @@
-use crate::{Change, Rope, RopeSlice, Transaction};
+use crate::{regex::Regex, Change, Rope, RopeSlice, Transaction};
 pub use helix_syntax::{get_language, get_language_name, Lang};

 use std::{
@ -59,21 +59,53 @@ pub struct IndentationConfiguration {
    pub unit: String,
 }

+fn read_query(language: &str, filename: &str) -> String {
+    static INHERITS_REGEX: Lazy<Regex> =
+        Lazy::new(|| Regex::new(r";+\s*inherits\s*:?\s*([a-z_,()]+)\s*").unwrap());
+
+    let root = PathBuf::from(env!("CARGO_MANIFEST_DIR"));
+
+    let path = root
+        .join("../runtime/queries")
+        .join(language)
+        .join(filename);
+
+    let query = std::fs::read_to_string(&path).unwrap_or_default();
+
+    // TODO: the collect() is not ideal
+    let inherits = INHERITS_REGEX
+        .captures_iter(&query)
+        .flat_map(|captures| {
+            captures[1]
+                .split(',')
+                .map(str::to_owned)
+                .collect::<Vec<_>>()
+        })
+        .collect::<Vec<_>>();
+
+    if inherits.is_empty() {
+        return query;
+    }
+
+    let mut queries = inherits
+        .iter()
+        .map(|language| read_query(language, filename))
+        .collect::<Vec<_>>();
+
+    queries.push(query);
+
+    queries.concat()
+}
+
 impl LanguageConfiguration {
    pub fn highlight_config(&self, scopes: &[String]) -> Option<Arc<HighlightConfiguration>> {
        self.highlight_config
            .get_or_init(|| {
-                // let name = get_language_name(&self.language_id);
-
-                let root = PathBuf::from(env!("CARGO_MANIFEST_DIR"));
+                let language = get_language_name(self.language_id).to_ascii_lowercase();

-                let highlights_query =
-                    std::fs::read_to_string(root.join(&self.path).join("queries/highlights.scm"))
-                        .unwrap_or_default();
+                let highlights_query = read_query(&language, "highlights.scm");

-                let injections_query =
-                    std::fs::read_to_string(root.join(&self.path).join("queries/injections.scm"))
-                        .unwrap_or_default();
+                let injections_query = read_query(&language, "injections.scm");

                let locals_query = "";

--- a/helix-syntax/Cargo.toml
+++ b/helix-syntax/Cargo.toml
@ -7,7 +7,7 @@ edition = "2018"
 # See more keys and their definitions at https://doc.rust-lang.org/cargo/reference/manifest.html

 [dependencies]
-tree-sitter = "0.17"
+tree-sitter = "0.19"
 serde = { version = "1.0", features = ["derive"] }

 [build-dependencies]
--- a/helix-syntax/languages/tree-sitter-bash
+++ b/helix-syntax/languages/tree-sitter-bash
@ -1 +1 @@
-Subproject commit 8ece09ca4c0b5e59b124cd19fa92c76b1a9e9dd4
+Subproject commit a8eb5cb57c66f74c63ab950de081207cccf52017
--- a/helix-syntax/languages/tree-sitter-c
+++ b/helix-syntax/languages/tree-sitter-c
@ -1 +1 @@
-Subproject commit fa408bc9e77f4b770bd1db984ca00c901ddf95fc
+Subproject commit f05e279aedde06a25801c3f2b2cc8ac17fac52ae
--- a/helix-syntax/languages/tree-sitter-c-sharp
+++ b/helix-syntax/languages/tree-sitter-c-sharp
@ -1 +1 @@
-Subproject commit 21ec3c3deb4365085aa353fadbc6a616d7769f9f
+Subproject commit 53a65a908167d6556e1fcdb67f1ee62aac101dda
--- a/helix-syntax/languages/tree-sitter-cpp
+++ b/helix-syntax/languages/tree-sitter-cpp
@ -1 +1 @@
-Subproject commit 3bfe046f3967fef92ebb33f8cd58c3ff373d5e56
+Subproject commit c61212414a3e95b5f7507f98e83de1d638044adc
--- a/helix-syntax/languages/tree-sitter-css
+++ b/helix-syntax/languages/tree-sitter-css
@ -1 +1 @@
-Subproject commit e882c98b5e62d864f7f9e4d855b19b6050c897a8
+Subproject commit 94e10230939e702b4fa3fa2cb5c3bc7173b95d07
--- a/helix-syntax/languages/tree-sitter-go
+++ b/helix-syntax/languages/tree-sitter-go
@ -1 +1 @@
-Subproject commit dadfd9c9aab2630632e61cfce645c13c35aa092f
+Subproject commit 2a83dfdd759a632651f852aa4dc0af2525fae5cd
--- a/helix-syntax/languages/tree-sitter-html
+++ b/helix-syntax/languages/tree-sitter-html
@ -1 +1 @@
-Subproject commit 7f442e1c6163d450c69c75c7a621badc3a0ea98f
+Subproject commit d93af487cc75120c89257195e6be46c999c6ba18
--- a/helix-syntax/languages/tree-sitter-java
+++ b/helix-syntax/languages/tree-sitter-java
@ -1 +1 @@
-Subproject commit 16c07a726c34c9925b3e28716b2d6d60e3643252
+Subproject commit bd6186c24d5eb13b4623efac9d944dcc095c0dad
--- a/helix-syntax/languages/tree-sitter-javascript
+++ b/helix-syntax/languages/tree-sitter-javascript
@ -1 +1 @@
-Subproject commit 452151f7192cc26b70fe4a7ec192a37d99854a6a
+Subproject commit 4a95461c4761c624f2263725aca79eeaefd36cad
--- a/helix-syntax/languages/tree-sitter-json
+++ b/helix-syntax/languages/tree-sitter-json
@ -1 +1 @@
-Subproject commit d3976b27df8622ed17bef6dd5e358b398e73c676
+Subproject commit 65bceef69c3b0f24c0b19ce67d79f57c96e90fcb
--- a/helix-syntax/languages/tree-sitter-julia
+++ b/helix-syntax/languages/tree-sitter-julia
@ -1 +1 @@
-Subproject commit 6a0863f1ce3fcf6f99dc0addb7886dcbd27c5a48
+Subproject commit 0ba7a24b062b671263ae08e707e9e94383b25bb7
--- a/helix-syntax/languages/tree-sitter-php
+++ b/helix-syntax/languages/tree-sitter-php
@ -1 +1 @@
-Subproject commit b0c0367d4b7058921fdc4ba11e257441a64ab809
+Subproject commit 0d63eaf94e8d6c0694551b016c802787e61b3fb2
--- a/helix-syntax/languages/tree-sitter-python
+++ b/helix-syntax/languages/tree-sitter-python
@ -1 +1 @@
-Subproject commit 3196e288650992bca2399dda15ac703c342a22bb
+Subproject commit d6210ceab11e8d812d4ab59c07c81458ec6e5184
--- a/helix-syntax/languages/tree-sitter-ruby
+++ b/helix-syntax/languages/tree-sitter-ruby
@ -1 +1 @@
-Subproject commit 32cd5a04adb4accb0c121f037ab59df3c3488228
+Subproject commit dfff673b41df7fadcbb609c6338f38da3cdd018e
--- a/helix-syntax/languages/tree-sitter-rust
+++ b/helix-syntax/languages/tree-sitter-rust
@ -1 +1 @@
-Subproject commit 1be73c282bb6d9fd2fbb55945b313f9527cad060
+Subproject commit a360da0a29a19c281d08295a35ecd0544d2da211
--- a/helix-syntax/languages/tree-sitter-scala
+++ b/helix-syntax/languages/tree-sitter-scala
@ -1 +1 @@
-Subproject commit 211bb726bb5857f872247b600c7c1808e641a8d4
+Subproject commit fb23ed9a99da012d86b7a5059b9d8928607cce29
--- a/helix-syntax/languages/tree-sitter-toml
+++ b/helix-syntax/languages/tree-sitter-toml
@ -1 +1 @@
-Subproject commit 084da152d85cb8c4bbbe0ab5f3f1f9e5bfb77d3c
+Subproject commit 7cff70bbcbbc62001b465603ca1ea88edd668704
--- a/helix-syntax/languages/tree-sitter-typescript
+++ b/helix-syntax/languages/tree-sitter-typescript
@ -1 +1 @@
-Subproject commit 543cbe44f16189f7f1b739cf268d70f373d94b87
+Subproject commit 3e897ea5925f037cfae2e551f8e6b12eec2a201a
--- a/helix-syntax/nvim-treesitter
+++ b/helix-syntax/nvim-treesitter
@ -1 +0,0 @@
-Subproject commit 1f00ecdfa36ef5e43a4feaf189e8c2c003118c00
--- a/runtime/queries/c/highlights.scm
+++ b/runtime/queries/c/highlights.scm
@ -0,0 +1,81 @@
+"break" @keyword
+"case" @keyword
+"const" @keyword
+"continue" @keyword
+"default" @keyword
+"do" @keyword
+"else" @keyword
+"enum" @keyword
+"extern" @keyword
+"for" @keyword
+"if" @keyword
+"inline" @keyword
+"return" @keyword
+"sizeof" @keyword
+"static" @keyword
+"struct" @keyword
+"switch" @keyword
+"typedef" @keyword
+"union" @keyword
+"volatile" @keyword
+"while" @keyword
+
+"#define" @keyword
+"#elif" @keyword
+"#else" @keyword
+"#endif" @keyword
+"#if" @keyword
+"#ifdef" @keyword
+"#ifndef" @keyword
+"#include" @keyword
+(preproc_directive) @keyword
+
+"--" @operator
+"-" @operator
+"-=" @operator
+"->" @operator
+"=" @operator
+"!=" @operator
+"*" @operator
+"&" @operator
+"&&" @operator
+"+" @operator
+"++" @operator
+"+=" @operator
+"<" @operator
+"==" @operator
+">" @operator
+"||" @operator
+
+"." @delimiter
+";" @delimiter
+
+(string_literal) @string
+(system_lib_string) @string
+
+(null) @constant
+(number_literal) @number
+(char_literal) @number
+
+(call_expression
+  function: (identifier) @function)
+(call_expression
+  function: (field_expression
+    field: (field_identifier) @function))
+(function_declarator
+  declarator: (identifier) @function)
+(preproc_function_def
+  name: (identifier) @function.special)
+
+(field_identifier) @property
+(statement_identifier) @label
+(type_identifier) @type
+(primitive_type) @type
+(sized_type_specifier) @type
+
+((identifier) @constant
+ (#match? @constant "^[A-Z][A-Z\\d_]*$"))
+
+(identifier) @variable
+
+(comment) @comment
--- a/runtime/queries/cpp/highlights.scm
+++ b/runtime/queries/cpp/highlights.scm
@ -0,0 +1,68 @@
+; inherits: c
+
+; Functions
+
+(call_expression
+  function: (scoped_identifier
+    name: (identifier) @function))
+
+(template_function
+  name: (identifier) @function)
+
+(template_method
+  name: (field_identifier) @function)
+
+(template_function
+  name: (scoped_identifier
+    name: (identifier) @function))
+
+(function_declarator
+  declarator: (scoped_identifier
+    name: (identifier) @function))
+
+(function_declarator
+  declarator: (scoped_identifier
+    name: (identifier) @function))
+
+(function_declarator
+  declarator: (field_identifier) @function)
+
+; Types
+
+((namespace_identifier) @type
+ (#match? @type "^[A-Z]"))
+
+(auto) @type
+
+; Constants
+
+(this) @variable.builtin
+(nullptr) @constant
+
+; Keywords
+
+"catch" @keyword
+"class" @keyword
+"constexpr" @keyword
+"delete" @keyword
+"explicit" @keyword
+"final" @keyword
+"friend" @keyword
+"mutable" @keyword
+"namespace" @keyword
+"noexcept" @keyword
+"new" @keyword
+"override" @keyword
+"private" @keyword
+"protected" @keyword
+"public" @keyword
+"template" @keyword
+"throw" @keyword
+"try" @keyword
+"typename" @keyword
+"using" @keyword
+"virtual" @keyword
+
+; Strings
+
+(raw_string_literal) @string
--- a/runtime/queries/rust/highlights.scm
+++ b/runtime/queries/rust/highlights.scm
@ -0,0 +1,135 @@
+; Identifier conventions
+
+; Assume all-caps names are constants
+((identifier) @constant
+ (#match? @constant "^[A-Z][A-Z\\d_]+$'"))
+
+; Assume that uppercase names in paths are types
+((scoped_identifier
+  path: (identifier) @type)
+ (#match? @type "^[A-Z]"))
+((scoped_identifier
+  path: (scoped_identifier
+    name: (identifier) @type))
+ (#match? @type "^[A-Z]"))
+
+; Assume other uppercase names are enum constructors
+((identifier) @constructor
+ (#match? @constructor "^[A-Z]"))
+
+; Function calls
+
+(call_expression
+  function: (identifier) @function)
+(call_expression
+  function: (field_expression
+    field: (field_identifier) @function.method))
+(call_expression
+  function: (scoped_identifier
+    "::"
+    name: (identifier) @function))
+
+(generic_function
+  function: (identifier) @function)
+(generic_function
+  function: (scoped_identifier
+    name: (identifier) @function))
+(generic_function
+  function: (field_expression
+    field: (field_identifier) @function.method))
+
+(macro_invocation
+  macro: (identifier) @function.macro
+  "!" @function.macro)
+
+; Function definitions
+
+(function_item (identifier) @function)
+(function_signature_item (identifier) @function)
+
+; Other identifiers
+
+(type_identifier) @type
+(primitive_type) @type.builtin
+(field_identifier) @property
+
+(line_comment) @comment
+(block_comment) @comment
+
+"(" @punctuation.bracket
+")" @punctuation.bracket
+"[" @punctuation.bracket
+"]" @punctuation.bracket
+
+(type_arguments
+  "<" @punctuation.bracket
+  ">" @punctuation.bracket)
+(type_parameters
+  "<" @punctuation.bracket
+  ">" @punctuation.bracket)
+
+"::" @punctuation.delimiter
+"." @punctuation.delimiter
+";" @punctuation.delimiter
+
+(parameter (identifier) @variable.parameter)
+
+(lifetime (identifier) @label)
+
+"break" @keyword
+"const" @keyword
+"continue" @keyword
+"default" @keyword
+"dyn" @keyword
+"else" @keyword
+"enum" @keyword
+"extern" @keyword
+"fn" @keyword
+"for" @keyword
+"if" @keyword
+"impl" @keyword
+"in" @keyword
+"let" @keyword
+"let" @keyword
+"loop" @keyword
+"macro_rules!" @keyword
+"match" @keyword
+"mod" @keyword
+"move" @keyword
+"pub" @keyword
+"ref" @keyword
+"return" @keyword
+"static" @keyword
+"struct" @keyword
+"trait" @keyword
+"type" @keyword
+"union" @keyword
+"unsafe" @keyword
+"use" @keyword
+"where" @keyword
+"while" @keyword
+(mutable_specifier) @keyword
+(use_list (self) @keyword)
+(scoped_use_list (self) @keyword)
+(scoped_identifier (self) @keyword)
+(super) @keyword
+
+(self) @variable.builtin
+
+(char_literal) @string
+(string_literal) @string
+(raw_string_literal) @string
+
+(boolean_literal) @constant.builtin
+(integer_literal) @constant.builtin
+(float_literal) @constant.builtin
+
+(escape_sequence) @escape
+
+(attribute_item) @attribute
+(inner_attribute_item) @attribute
+
+"as" @operator
+"*" @operator
+"&" @operator
+"'" @operator
--- a/runtime/queries/rust/injections.scm
+++ b/runtime/queries/rust/injections.scm
@ -0,0 +1,9 @@
+((macro_invocation
+  (token_tree) @injection.content)
+ (#set! injection.language "rust")
+ (#set! injection.include-children))
+
+((macro_rule
+  (token_tree) @injection.content)
+ (#set! injection.language "rust")
+ (#set! injection.include-children))
				`@ -1 +0,0 @@`
				`Subproject commit 1f00ecdfa36ef5e43a4feaf189e8c2c003118c00`