feat(tree-sitter): add tree sitter bindings

flake.nix

@@ -128,6 +128,7 @@
               shellHook = ''
                 # This is a hack to work around the hack used in the dune files
                 export OPAM_SWITCH_PREFIX="${ligo.OPAM_SWITCH_PREFIX}";
+                export LIGO_TREE_SITTER="${ligo.LIGO_TREE_SITTER}";
               '';
             };
 

lib/tree_sitter/bindings.ml

+open Ctypes
+open Foreign
+
+(* ////////// Types ////////// *)
+
+let ts_symbol = uint16_t 
+let ts_state_id = uint16_t 
+let ts_field_id = uint16_t 
+
+(* ////////// struct TSPoint ////////// *)
+
+type ts_point 
+let ts_point: ts_point structure typ = structure "TSPoint" 
+let row = field ts_point "row" uint 
+let column = field ts_point "column" uint
+let () = seal ts_point
+
+(* ////////// struct TSRange ////////// *)
+
+type ts_range 
+let ts_range: ts_range structure typ = structure "TSRange" 
+let start_point = field ts_range "start_point" ts_point 
+let endpoint = field ts_range "endpoint" ts_point
+let start_byte = field ts_range "start_byte" uint32_t
+let end_byte = field ts_range "end_byte" uint32_t
+let () = seal ts_range
+
+(* ////////// struct TSLanguage ////////// *)
+
+type ts_language 
+let ts_language: ts_language structure typ = structure "TSLanguage" 
+
+(* ////////// struct TSTree ////////// *)
+
+type ts_tree 
+let ts_tree: ts_tree structure typ = structure "TSTree" 
+
+(* ////////// struct TSParser ////////// *)
+
+type ts_parser
+let ts_parser: ts_parser structure typ = structure "TSParser" 
+
+(* ////////// struct TSNode ////////// *)
+
+type ts_node 
+let ts_node: ts_tree structure typ = structure "TSNode" 
+(* CTypes doesn't suport an array so instead we declare 4 uint32 *)
+let c1 = field ts_node "" (uint32_t)
+let c2 = field ts_node "" (uint32_t)
+let c3 = field ts_node "" (uint32_t)
+let c4 = field ts_node "" (uint32_t)
+let id = field ts_node "id" (ptr void)
+let tree = field ts_node "tree" (ptr ts_tree)
+let () = seal ts_node
+
+
+(* FFI *)
+
+let ts_parser_new = foreign "ts_parser_new" (void @-> returning (ptr ts_parser))
+
+let ts_parser_delete = foreign "ts_parser_delete" (ptr ts_parser @-> returning void)
+
+let ts_parser_set_language = foreign "ts_parser_set_language" (ptr ts_parser @-> ptr ts_language @-> returning bool)
+
+let ts_parser_parse_string = foreign "ts_parser_parse_string" (ptr ts_parser @-> ptr ts_tree @-> string @-> uint32_t @-> returning (ptr ts_tree))
+
+let ts_tree_delete = foreign "ts_tree_delete" (ptr ts_tree @-> returning void)
+
+let ts_tree_root_node = foreign "ts_tree_root_node" (ptr ts_tree @-> returning (ts_node))
+
+let ts_node_named_child = foreign "ts_node_named_child" (ts_node @-> uint32_t @-> returning ts_node)
+
+let ts_node_type = foreign "ts_node_type" (ts_node @-> returning string)
+
+let ts_node_child_count = foreign "ts_node_child_count" (ts_node @-> returning uint32_t)
+
+let ts_node_string = foreign "ts_node_string" (ts_node @-> returning string)
\ No newline at end of file

lib/tree_sitter/dune

+(library
+ (name tree_sitter)
+ (libraries ctypes ctypes-foreign)
+ (modules bindings))
+
+(executable
+ (name example)
+ (libraries ctypes ctypes-foreign tree_sitter tree_sitter_typescript)
+ (modules example)
+ (link_flags
+  :standard
+  -cclib
+  -ltree-sitter
+  -cclib
+  %{env:LIGO_TREE_SITTER=}/parser))

lib/tree_sitter/example.ml

+open Tree_sitter.Bindings
+open Tree_sitter_typescript.Bindings
+open Ctypes 
+open Unsigned
+
+(* export DYLD_LIBRARY_PATH=$(pwd)/tree_sitter:$(pwd)/tree_sitter_typescript && dune build && dune exec tree_sitter/example.exe *)
+
+let parse_typescript_string source_code = 
+  let parser = ts_parser_new () in
+  let language = tree_sitter_typescript () in
+  let _ = ts_parser_set_language parser language in
+  let null_tree = from_voidp ts_tree null in
+  let tree = ts_parser_parse_string parser null_tree source_code (UInt32.of_int @@ String.length source_code) in ts_parser_delete parser; tree
+
+let print_node node = 
+  let str = ts_node_string node in (Printf.printf "\n\n TypeScript CST : %s\n\n" str)
+
+let () = 
+let code = "[1, null]" in   
+let tree = parse_typescript_string code in
+(* extract the node *)
+let root_node = ts_tree_root_node tree in
+let expr_stmt_node = ts_node_named_child root_node (UInt32.of_int 0) in
+let array_node = ts_node_named_child expr_stmt_node (UInt32.of_int 0) in
+let number_node = ts_node_named_child array_node (UInt32.of_int 0) in
+let null_node = ts_node_named_child array_node (UInt32.of_int 1) in
+(* get the node type *)
+assert ((ts_node_type root_node) = "program");
+assert ((ts_node_type expr_stmt_node) = "expression_statement");
+assert ((ts_node_type array_node) = "array");
+assert ((ts_node_type number_node) = "number");
+assert ((ts_node_type null_node) = "null");
+
+(* get the child count *)
+assert ((ts_node_child_count root_node) = (UInt32.of_int 1));
+assert ((ts_node_child_count array_node) = (UInt32.of_int 5));
+assert ((ts_node_child_count number_node) = (UInt32.of_int 0));
+
+print_node root_node; ts_tree_delete tree ;
\ No newline at end of file

lib/tree_sitter_typescript/api.h

+#include "../tree_sitter/api.h"
+// Declare the `tree_sitter_typescript` function, which is
+// implemented by the `tree_sitter_typescript` library.
+const TSLanguage *tree_sitter_typescript(void);
\ No newline at end of file

lib/tree_sitter_typescript/bindings.ml

+open Foreign
+open Ctypes
+open Tree_sitter.Bindings
+
+let tree_sitter_typescript = foreign "tree_sitter_typescript" (void @-> returning (ptr ts_language))
\ No newline at end of file

lib/tree_sitter_typescript/dune

+(library
+ (name tree_sitter_typescript)
+ (libraries ctypes ctypes-foreign tree_sitter)
+ (modules bindings))  ; Ensure that the module name matches the file name

nix/ligo.nix

@@ -3,6 +3,8 @@
   lib,
   pkgs,
   tezos-ligo,
+  tree-sitter,
+  tree-sitter-grammars
 }: let
   inherit (pkgs) darwin ocamlPackages python3Packages coq_8_13 tezos-rust-libs;
 in
@@ -13,6 +15,7 @@ in
       src = ./..;
 
       OPAM_SWITCH_PREFIX = "${tezos-rust-libs}";
+      LIGO_TREE_SITTER = "${tree-sitter-grammars.tree-sitter-typescript}";
 
       postPatch = ''
         mkdir -p vendors/tezos-ligo
@@ -26,6 +29,8 @@ in
         crunch
         odoc
         python3Packages.jsonschema
+        tree-sitter
+        tree-sitter-grammars.tree-sitter-typescript
       ];
 
       propagatedBuildInputs =
@@ -97,6 +102,8 @@ in
           alcotest # with-test
           ppx_expect # with-test
           ppx_inline_test # with-test
+          ctypes
+          ctypes-foreign
         ]
         ++ lib.optionals stdenv.isDarwin [
           darwin.apple_sdk.frameworks.Security