OCaml support (#666)

* added some stuff

* add interface

* indent

* highlights and locals

* scope

* change some stuff

* add indents

* fix blanket highlight

* macro

* use inherits

8 files changed, 222 insertions, 0 deletions

diff --git a/.gitmodules b/.gitmodules
index ae4984da..f905b8c7 100644
--- a/.gitmodules
+++ b/.gitmodules
@@ -102,6 +102,10 @@
 	path = helix-syntax/languages/tree-sitter-protobuf
 	url = https://github.com/yusdacra/tree-sitter-protobuf.git
 	shallow = true
+[submodule "helix-syntax/languages/tree-sitter-ocaml"]
+	path = helix-syntax/languages/tree-sitter-ocaml
+	url = https://github.com/tree-sitter/tree-sitter-ocaml
+	shallow = true
 [submodule "helix-syntax/languages/tree-sitter-lua"]
 	path = helix-syntax/languages/tree-sitter-lua
 	url = https://github.com/nvim-treesitter/tree-sitter-lua
diff --git a/helix-syntax/build.rs b/helix-syntax/build.rs
index 75f8c970..473646fd 100644
--- a/helix-syntax/build.rs
+++ b/helix-syntax/build.rs
@@ -177,6 +177,7 @@ fn main() {
     let ignore = vec![
         "tree-sitter-typescript".to_string(),
         "tree-sitter-haskell".to_string(), // aarch64 failures: https://github.com/tree-sitter/tree-sitter-haskell/issues/34
+        "tree-sitter-ocaml".to_string(),
     ];
     let dirs = collect_tree_sitter_dirs(&ignore).unwrap();
 
@@ -202,4 +203,6 @@ fn main() {
 
     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")
 }
diff --git a/helix-syntax/languages/tree-sitter-ocaml b/helix-syntax/languages/tree-sitter-ocaml
new file mode 160000
+Subproject 23d419ba45789c5a47d31448061557716b02750
diff --git a/languages.toml b/languages.toml
index 7164c4df..c04435fe 100644
--- a/languages.toml
+++ b/languages.toml
@@ -225,6 +225,23 @@ comment-token = ";"
 indent = { tab-width = 4, unit = "    " }
 
 [[language]]
+name = "ocaml"
+scope = "source.ocaml"
+injection-regex = "ocaml"
+file-types = ["ml"]
+roots = []
+comment-token = "(**)"
+indent = { tab-width = 2, unit = "  " }
+
+[[language]]
+name = "ocaml-interface"
+scope = "source.ocaml.interface"
+file-types = ["mli"]
+roots = []
+comment-token = "(**)"
+indent = { tab-width = 2, unit = "  "}
+
+[[language]]
 name = "lua"
 scope = "source.lua"
 file-types = ["lua"]
diff --git a/runtime/queries/ocaml-interface/highlights.scm b/runtime/queries/ocaml-interface/highlights.scm
new file mode 100644
index 00000000..6d3dfbcf
--- /dev/null
+++ b/runtime/queries/ocaml-interface/highlights.scm
@@ -0,0 +1 @@
+; inherits: ocaml
diff --git a/runtime/queries/ocaml/highlights.scm b/runtime/queries/ocaml/highlights.scm
new file mode 100644
index 00000000..093b3cce
--- /dev/null
+++ b/runtime/queries/ocaml/highlights.scm
@@ -0,0 +1,160 @@
+; Modules
+;--------
+
+[(module_name) (module_type_name)] @namespace
+
+; Types
+;------
+
+[(class_name) (class_type_name) (type_constructor)] @type
+
+[(constructor_name) (tag)] @constructor
+
+; Functions
+;----------
+
+(let_binding
+  pattern: (value_name) @function
+  (parameter))
+
+(let_binding
+  pattern: (value_name) @function
+  body: [(fun_expression) (function_expression)])
+
+(value_specification (value_name) @function)
+
+(external (value_name) @function)
+
+(method_name) @method
+
+; Variables
+;----------
+
+(value_pattern) @parameter
+
+; Application
+;------------
+
+(infix_expression
+  left: (value_path (value_name) @function)
+  (infix_operator) @operator
+  (#eq? @operator "@@"))
+
+(infix_expression
+  (infix_operator) @operator
+  right: (value_path (value_name) @function)
+  (#eq? @operator "|>"))
+
+(application_expression
+  function: (value_path (value_name) @function))
+
+; Properties
+;-----------
+
+[(label_name) (field_name) (instance_variable_name)] @property
+
+; Constants
+;----------
+
+[(boolean) (unit)] @constant
+
+[(number) (signed_number)] @number
+
+(character) @character
+
+(string) @string
+
+(quoted_string "{" @string "}" @string) @string
+
+(escape_sequence) @string.escape
+
+[
+  (conversion_specification)
+  (pretty_printing_indication)
+] @punctuation.special
+
+; Keywords
+;---------
+
+[
+  "and" "as" "assert" "begin" "class" "constraint"
+  "end" "external" "in"
+  "inherit" "initializer" "lazy" "let" "match" "method" "module"
+  "mutable" "new" "nonrec" "object" "of" "private" "rec" "sig" "struct"
+  "type" "val" "virtual" "when" "with"
+] @keyword
+
+["fun" "function" "functor"] @keyword.function
+
+["if" "then" "else"] @keyword.control.conditional
+
+["exception" "try"] @keyword.control.exception
+
+["include" "open"] @include
+
+["for" "to" "downto" "while" "do" "done"] @keyword.control.loop
+
+; Macros
+;-------
+
+(attribute ["[@" "]"] @attribute)
+(item_attribute ["[@@" "]"] @attribute)
+(floating_attribute ["[@@@" "]"] @attribute)
+(extension ["[%" "]"] @function.macro)
+(item_extension ["[%%" "]"] @function.macro)
+(quoted_extension ["{%" "}"] @function.macro)
+(quoted_item_extension ["{%%" "}"] @function.macro)
+"%" @function.macro
+
+["(" ")" "[" "]" "{" "}" "[|" "|]" "[<" "[>"] @punctuation.bracket
+
+(object_type ["<" ">"] @punctuation.bracket)
+
+[
+  "," "." ";" ":" "=" "|" "~" "?" "+" "-" "!" ">" "&"
+  "->" ";;" ":>" "+=" ":=" ".."
+] @punctuation.delimiter
+
+; Operators
+;----------
+
+[
+  (prefix_operator)
+  (sign_operator)
+  (infix_operator)
+  (hash_operator)
+  (indexing_operator)
+  (let_operator)
+  (and_operator)
+  (match_operator)
+] @operator
+
+(match_expression (match_operator) @keyword)
+
+(value_definition [(let_operator) (and_operator)] @keyword)
+
+;; TODO: this is an error now
+;(prefix_operator "!" @operator)
+
+(infix_operator ["&" "+" "-" "=" ">" "|" "%"] @operator)
+
+(signed_number ["+" "-"] @operator)
+
+["*" "#" "::" "<-"] @operator
+
+; Attributes
+;-----------
+
+(attribute_id) @property
+
+; Comments
+;---------
+
+[(comment) (line_number_directive) (directive) (shebang)] @comment
+
+(ERROR) @error
+
+; Blanket highlights
+; ------------------
+
+[(value_name) (type_variable)] @variable
diff --git a/runtime/queries/ocaml/indents.toml b/runtime/queries/ocaml/indents.toml
new file mode 100644
index 00000000..9b6462d8
--- /dev/null
+++ b/runtime/queries/ocaml/indents.toml
@@ -0,0 +1,13 @@
+indent = [
+    "let_binding",
+    "type_binding",
+    "structure",
+    "signature",
+    "record_declaration",
+    "function_expression",
+    "match_case",
+]
+
+oudent = [
+    "}",
+]
diff --git a/runtime/queries/ocaml/locals.scm b/runtime/queries/ocaml/locals.scm
new file mode 100644
index 00000000..8f3f3fdf
--- /dev/null
+++ b/runtime/queries/ocaml/locals.scm
@@ -0,0 +1,24 @@
+; Scopes
+;-------
+
+[
+  (let_binding)
+  (class_binding)
+  (class_function)
+  (method_definition)
+  (fun_expression)
+  (object_expression)
+  (for_expression)
+  (match_case)
+  (attribute_payload)
+] @local.scope
+
+; Definitions
+;------------
+
+(value_pattern) @local.definition
+
+; References
+;-----------
+
+(value_path . (value_name) @local.reference)