tree-sitter: initial tree-sitter support

CMakeLists.txt

@@ -405,6 +405,11 @@ if(MSGPACK_HAS_FLOAT32)
   add_definitions(-DNVIM_MSGPACK_HAS_FLOAT32)
 endif()
 
+set(TREESITTER_PATH "${DEPS_BUILD_DIR}/build/src/treesitter")
+set(TREESITTER_C_PATH "${DEPS_BUILD_DIR}/build/src/treesitter-c")
+set(TREESITTER_JAVASCRIPT_PATH "${DEPS_BUILD_DIR}/build/src/treesitter-javascript")
+set(LIBUTF8PROC_INCLUDE_DIRS ${DEPS_BUILD_DIR}/build/src/utf8proc)
+
 option(FEAT_TUI "Enable the Terminal UI" ON)
 
 if(FEAT_TUI)

runtime/lua/treesitter_rt.lua

@@ -0,0 +1,119 @@
+local a = vim.api
+
+if __treesitter_rt_ns == nil then
+    __treesitter_rt_ns = a.nvim_buf_add_highlight(0, 0, "", 0, 0, 0)
+    __treesitter_rt_syn_ns = a.nvim_buf_add_highlight(0, 0, "", 0, 0, 0)
+end
+local my_ns = __treesitter_rt_ns
+local my_syn_ns = __treesitter_rt_syn_ns
+
+local path = '.deps/build/src/treesitter-javascript/src/highlights.json'
+a.nvim_set_var("_ts_path", path)
+obj = a.nvim_eval("json_decode(readfile(g:_ts_path,'b'))")
+
+
+--luadev = require'luadev'
+--i = require'inspect'
+
+
+function parse_tree(tsstate, force)
+  if tsstate.valid and not force then
+    return tsstate.tree
+  end
+  tsstate.tree = tsstate.parser:parse_buf(tsstate.bufnr)
+  tsstate.valid = true
+  return tsstate.tree
+end
+
+function the_cb(tsstate, ev, bufnr, tick, start_row, oldstopline, stop_row)
+  local start_byte = a.nvim_buf_get_offset(bufnr,start_row)
+  -- a bit messy, should we expose edited but not reparsed tree?
+  -- are multiple edits safe in general?
+  local root = tsstate.parser:tree():root()
+  -- TODO: add proper lookup function!
+  local inode = root:descendant_for_point_range(oldstopline+9000,0, oldstopline,0)
+  local edit
+  if inode == nil then
+    local stop_byte = a.nvim_buf_get_offset(bufnr,stop_row)
+    tsstate.parser:edit(start_byte,stop_byte,stop_byte,start_row,0,stop_row,0,stop_row,0)
+  else
+    local fakeoldstoprow, fakeoldstopcol, fakebyteoldstop = inode:start()
+    local fake_rows = fakeoldstoprow-oldstopline
+    local fakestop = stop_row+fake_rows
+    local fakebytestop = a.nvim_buf_get_offset(bufnr,fakestop)+fakeoldstopcol
+    tsstate.parser:edit(start_byte,fakebyteoldstop,fakebytestop,start_row,0,fakeoldstoprow,fakeoldstopcol,fakestop,fakeoldstopcol)
+  end
+  tsstate.valid = false
+  --luadev.append_buf({i{edit.start_byte,edit.old_end_byte,edit.new_end_byte},
+  --                   i{edit.start_point, edit.old_end_point, edit.new_end_point}})
+end
+
+function attach_buf(tsstate)
+  local function cb(ev, ...)
+    return the_cb(tsstate, ev, ...)
+  end
+  a.nvim_buf_attach(tsstate.bufnr, false, {on_lines=cb})
+end
+
+function create_parser(bufnr)
+  if bufnr == 0 then
+    bufnr = a.nvim_get_current_buf()
+  end
+  local ft = a.nvim_buf_get_option(bufnr, "filetype")
+  local tsstate = {}
+  tsstate.bufnr = bufnr
+  tsstate.parser = vim.ts_parser(ft)
+  parse_tree(tsstate)
+  attach_buf(tsstate)
+  return tsstate
+end
+
+function ts_inspect_pos(row,col)
+  local tree = parse_tree(theparser)
+  local root = tree:root()
+  local node = root:descendant_for_point_range(row,col,row,col)
+  show_node(node)
+end
+
+function show_node(node)
+  if node == nil then
+    return
+  end
+  a.nvim_buf_clear_highlight(0, my_ns, 0, -1)
+  shown_node = node
+  print(node:type())
+  local start_row, start_col, end_row, end_col = node:range()
+
+  a.nvim_buf_add_highlight(0, my_ns, "ErrorMsg", start_row, start_col, start_col+1)
+
+  if end_col >= 1 then
+    end_col = end_col - 1
+  end
+  a.nvim_buf_add_highlight(0, my_ns, "ErrorMsg", end_row, end_col, end_col+1)
+end
+
+function ts_expand_node()
+  if shown_node == nil then
+    return
+  end
+  parent = shown_node:parent()
+  show_node(parent)
+end
+
+function ts_cursor()
+  local row, col = unpack(a.nvim_win_get_cursor(0))
+  ts_inspect_pos(row-1, col)
+end
+
+if false then
+  ctree = theparser.tree
+  root = ctree:root()
+  cursor = root:to_cursor()
+  node = cursor:forward(5000) if true then return node end
+  print(#root)
+  c = root:child(50)
+  print(require'inspect'{c:extent()})
+  type(ctree.__tostring)
+  root:__tostring()
+  print(_tslua_debug())
+end

runtime/plugin/ts_test.vim

@@ -0,0 +1,32 @@
+let g:ts_test_path = expand("<sfile>:p:h:h")
+let g:has_ts = v:false
+
+func! TSTest()
+  if g:has_ts
+    return
+  end
+  " TODO: module!
+  lua require'treesitter_rt'
+  lua theparser = create_parser(vim.api.nvim_get_current_buf())
+  let g:has_ts = v:true
+endfunc
+
+func! TSCursor()
+  " disable matchparen
+  NoMatchParen
+  call TSTest()
+  au CursorMoved <buffer> lua ts_cursor()
+  au CursorMovedI <buffer> lua ts_cursor()
+  map <buffer> <Plug>(ts-expand) <cmd>lua ts_expand_node()<cr>
+endfunc
+
+func! TSSyntax()
+  " disable matchparen
+  set syntax=
+  call TSTest()
+  lua ts_syntax()
+endfunc
+
+command! TSTest call TSTest()
+command! TSCursor call TSCursor()
+command! TSSyntax call TSSyntax()

src/nvim/CMakeLists.txt

@@ -74,6 +74,8 @@ include_directories(${GENERATED_DIR})
 include_directories(${CACHED_GENERATED_DIR})
 include_directories(${GENERATED_INCLUDES_DIR})
 
+include_directories(${LIBUTF8PROC_INCLUDE_DIRS})
+
 file(MAKE_DIRECTORY ${TOUCHES_DIR})
 file(MAKE_DIRECTORY ${GENERATED_DIR})
 file(MAKE_DIRECTORY ${GENERATED_INCLUDES_DIR})
@@ -85,6 +87,10 @@ file(GLOB NVIM_HEADERS *.h)
 file(GLOB XDIFF_SOURCES xdiff/*.c)
 file(GLOB XDIFF_HEADERS xdiff/*.h)
 
+# when LIBUTF8PROC build is fixed, don't use lib.c with amalgamated utf8proc.c
+#file(GLOB TS_SOURCES tree_sitter/*.c)
+file(GLOB TS_SOURCES ../tree_sitter/lib.c)
+
 foreach(subdir
         os
         api
@@ -141,6 +147,7 @@ set(CONV_SOURCES
   ex_cmds.c
   ex_docmd.c
   fileio.c
+  lua/tree_sitter.c
   mbyte.c
   memline.c
   message.c
@@ -158,7 +165,7 @@ foreach(sfile ${CONV_SOURCES})
   endif()
 endforeach()
 # xdiff: inlined external project, we don't maintain it. #9306
-list(APPEND CONV_SOURCES ${XDIFF_SOURCES})
+list(APPEND CONV_SOURCES ${XDIFF_SOURCES} ${TS_SOURCES})
 
 if(NOT MSVC)
   set_source_files_properties(
@@ -414,7 +421,7 @@ endif()
 
 add_executable(nvim ${NVIM_GENERATED_FOR_SOURCES} ${NVIM_GENERATED_FOR_HEADERS}
   ${NVIM_GENERATED_SOURCES} ${NVIM_SOURCES} ${NVIM_HEADERS}
-  ${XDIFF_SOURCES} ${XDIFF_HEADERS})
+  ${XDIFF_SOURCES} ${XDIFF_HEADERS} ${TS_SOURCES})
 target_link_libraries(nvim ${NVIM_EXEC_LINK_LIBRARIES})
 install_helper(TARGETS nvim)
 
@@ -500,7 +507,7 @@ add_library(
   EXCLUDE_FROM_ALL
   ${NVIM_SOURCES} ${NVIM_GENERATED_SOURCES}
   ${NVIM_HEADERS} ${NVIM_GENERATED_FOR_SOURCES} ${NVIM_GENERATED_FOR_HEADERS}
-  ${XDIFF_SOURCES} ${XDIFF_HEADERS}
+  ${XDIFF_SOURCES} ${XDIFF_HEADERS} ${TS_SOURCES}
 )
 set_property(TARGET libnvim APPEND PROPERTY
              INCLUDE_DIRECTORIES ${LUA_PREFERRED_INCLUDE_DIRS})
@@ -525,7 +532,7 @@ else()
     EXCLUDE_FROM_ALL
     ${NVIM_SOURCES} ${NVIM_GENERATED_SOURCES}
     ${NVIM_HEADERS} ${NVIM_GENERATED_FOR_SOURCES} ${NVIM_GENERATED_FOR_HEADERS}
-    ${XDIFF_SOURCES} ${XDIFF_HEADERS}
+    ${XDIFF_SOURCES} ${XDIFF_HEADERS} ${TS_SOURCES}
     ${UNIT_TEST_FIXTURES}
   )
   target_link_libraries(nvim-test ${NVIM_TEST_LINK_LIBRARIES})

src/nvim/lua/executor.c

@@ -31,6 +31,7 @@
 
 #include "nvim/lua/executor.h"
 #include "nvim/lua/converter.h"
+#include "nvim/lua/tree_sitter.h"
 
 #include "luv/luv.h"
 
@@ -310,7 +311,11 @@ static int nlua_state_init(lua_State *const lstate) FUNC_ATTR_NONNULL_ALL
   lua_setfield(lstate, -2, "luv");
   lua_pop(lstate, 3);
 
+  // internal vim._treesitter... API
+  nlua_add_treesitter(lstate);
+
   lua_setglobal(lstate, "vim");
+
   return 0;
 }
 
@@ -816,3 +821,40 @@ void ex_luafile(exarg_T *const eap)
     return;
   }
 }
+
+static int unsafe_ptr_to_ts_tree(lua_State *L)
+{
+  if (!lua_gettop(L)) {
+    return 0;
+  }
+  TSTree *const *ptr = lua_topointer(L,1);
+  tslua_push_tree(L, *ptr);
+  return 1;
+}
+
+static int create_tslua_parser(lua_State *L)
+{
+  TSLanguage *tree_sitter_c(void), *tree_sitter_javascript(void);
+
+  if (!lua_gettop(L)) {
+    return 0;
+  }
+  char *str = lua_tostring(L,1);
+
+  TSLanguage *lang = tree_sitter_c();
+  if (str && striequal(str, "javascript")) {
+    lang = tree_sitter_javascript();
+  }
+  tslua_push_parser(L, lang);
+  return 1;
+}
+
+static void nlua_add_treesitter(lua_State *const lstate) FUNC_ATTR_NONNULL_ALL
+{
+  tslua_init(lstate);
+  lua_pushcfunction(lstate, unsafe_ptr_to_ts_tree);
+  lua_setfield(lstate, -2, "unsafe_ts_tree");
+
+  lua_pushcfunction(lstate, create_tslua_parser);
+  lua_setfield(lstate, -2, "ts_parser");
+}

src/nvim/lua/tree_sitter.c

@@ -0,0 +1,472 @@
+// This is an open source non-commercial project. Dear PVS-Studio, please check
+// it. PVS-Studio Static Code Analyzer for C, C++ and C#: http://www.viva64.com
+
+// lua bindings for tree-siter.
+// NB: this file should contain a generic lua interface for
+// tree-sitter trees and nodes, and could be broken out as a reusable library
+
+#include <stdbool.h>
+#include <stdlib.h>
+#include <string.h>
+#include <inttypes.h>
+#include <assert.h>
+
+#include <lua.h>
+#include <lualib.h>
+#include <lauxlib.h>
+
+#include "tree_sitter/api.h"
+
+// NOT state-safe, delete when GC is confimed working:
+static int debug_n_trees = 0, debug_n_cursors = 0;
+
+#define REG_KEY "tree_sitter-private"
+
+#include "nvim/lua/tree_sitter.h"
+#include "nvim/api/private/handle.h"
+#include "nvim/memline.h"
+
+typedef struct {
+    TSParser *parser;
+    TSTree *tree;
+} Tslua_parser;
+
+#ifdef INCLUDE_GENERATED_DECLARATIONS
+# include "lua/tree_sitter.c.generated.h"
+#endif
+
+static struct luaL_Reg parser_meta[] = {
+  {"__gc", parser_gc},
+  {"__tostring", parser_tostring},
+  {"parse_buf", parser_parse_buf},
+  {"edit", parser_edit},
+  {"tree", parser_tree},
+  {NULL, NULL}
+};
+
+static struct luaL_Reg tree_meta[] = {
+  {"__gc", tree_gc},
+  {"__tostring", tree_tostring},
+  {"root", tree_root},
+  {NULL, NULL}
+};
+
+static struct luaL_Reg node_meta[] = {
+  {"__tostring", node_tostring},
+  {"__len", node_child_count},
+  {"range", node_range},
+  {"start", node_start},
+  {"type", node_type},
+  {"symbol", node_symbol},
+  {"child_count", node_child_count},
+  {"child", node_child},
+  {"descendant_for_point_range", node_descendant_for_point_range},
+  {"parent", node_parent},
+  {NULL, NULL}
+};
+
+void build_meta(lua_State *L, const luaL_Reg *meta)
+{
+  // [env, target]
+  for (size_t i = 0; meta[i].name != NULL; i++) {
+    lua_pushcfunction(L, meta[i].func);  // [env, target, func]
+    lua_pushvalue(L, -3);  // [env, target, func, env]
+    lua_setfenv(L, -2);  // [env, target, func]
+    lua_setfield(L, -2, meta[i].name);  // [env, target]
+  }
+
+  lua_pushvalue(L, -1);  // [env, target, target]
+  lua_setfield(L, -2, "__index");  // [env, target]
+}
+
+
+
+/// init the tslua library
+///
+/// all global state is stored in the regirstry of the lua_State
+void tslua_init(lua_State *L)
+{
+  lua_createtable(L, 0, 0);
+
+  // type metatables
+  lua_createtable(L, 0, 0);
+  build_meta(L, parser_meta);
+  lua_setfield(L, -2, "parser-meta");
+
+  lua_createtable(L, 0, 0);
+  build_meta(L, tree_meta);
+  lua_setfield(L, -2, "tree-meta");
+
+  lua_createtable(L, 0, 0);
+  build_meta(L, node_meta);
+  lua_setfield(L, -2, "node-meta");
+
+  lua_setfield(L, LUA_REGISTRYINDEX, REG_KEY);
+
+  lua_pushcfunction(L, tslua_debug);
+  lua_setglobal(L, "_tslua_debug");
+}
+
+static int tslua_debug(lua_State *L)
+{
+  lua_pushinteger(L, debug_n_trees);
+  lua_pushinteger(L, debug_n_cursors);
+  return 2;
+}
+
+void tslua_push_parser(lua_State *L, TSLanguage *lang)
+{
+  TSParser *parser = ts_parser_new();
+  ts_parser_set_language(parser, lang);
+  Tslua_parser *p = lua_newuserdata(L, sizeof(Tslua_parser));  // [udata]
+  p->parser = parser;
+  p->tree = NULL;
+
+  lua_getfield(L, LUA_REGISTRYINDEX, REG_KEY);  // [udata, env]
+  lua_getfield(L, -1, "parser-meta");  // [udata, env, meta]
+  lua_setmetatable(L, -3);  // [udata, env]
+  lua_pop(L, 1);  // [udata]
+}
+
+static Tslua_parser *parser_check(lua_State *L)
+{
+  if (!lua_gettop(L)) {
+    return 0;
+  }
+  if (!lua_isuserdata(L, 1)) {
+    return 0;
+  }
+  // TODO: typecheck!
+  return lua_touserdata(L, 1);
+}
+
+static int parser_gc(lua_State *L)
+{
+  Tslua_parser *p = parser_check(L);
+  if (!p) {
+    return 0;
+  }
+
+  ts_parser_delete(p->parser);
+  if (p->tree) {
+    ts_tree_delete(p->tree);
+  }
+
+  return 0;
+}
+
+static int parser_tostring(lua_State *L)
+{
+  lua_pushstring(L, "<parser>");
+  return 1;
+}
+
+static const char *input_cb(void *payload, uint32_t byte_index, TSPoint position, uint32_t *bytes_read)
+{
+   buf_T *bp  = payload;
+   static char buf[200];
+   if ((linenr_T)position.row >= bp->b_ml.ml_line_count) {
+     *bytes_read = 0;
+     return "";
+   }
+   char_u *line = ml_get_buf(bp, position.row+1, false);
+   size_t len = STRLEN(line);
+   size_t tocopy = MIN(len-position.column,200);
+
+   // TODO: translate embedded \n to \000
+   memcpy(buf, line+position.column, tocopy);
+   *bytes_read = (uint32_t)tocopy;
+   if (tocopy < 200) {
+     buf[tocopy] = '\n';
+     (*bytes_read)++;
+   }
+   return buf;
+}
+
+static int parser_parse_buf(lua_State *L)
+{
+  Tslua_parser *p = parser_check(L);
+  if (!p) {
+    return 0;
+  }
+
+  long bufnr = lua_tointeger(L, 2);
+  void *payload = handle_get_buffer(bufnr);
+  TSInput input = {payload, input_cb, TSInputEncodingUTF8};
+  TSTree *new_tree = ts_parser_parse(p->parser, p->tree, input);
+  if (p->tree) {
+    ts_tree_delete(p->tree);
+  }
+  p->tree = new_tree;
+
+  tslua_push_tree(L, ts_tree_copy(p->tree));
+  return 1;
+}
+
+static int parser_tree(lua_State *L)
+{
+  Tslua_parser *p = parser_check(L);
+  if (!p) {
+    return 0;
+  }
+
+  if (p->tree) {
+    tslua_push_tree(L, ts_tree_copy(p->tree));
+  } else {
+    lua_pushnil(L);
+  }
+  return 1;
+}
+
+static int parser_edit(lua_State *L)
+{
+  if(lua_gettop(L) < 10) {
+    lua_pushstring(L, "not enough args to parser:edit()");
+    lua_error(L);
+    return 0; // unreachable
+  }
+
+  Tslua_parser *p = parser_check(L);
+  if (!p) {
+    return 0;
+  }
+
+  if (!p->tree) {
+    return 0;
+  }
+
+  long start_byte = lua_tointeger(L, 2);
+  long old_end_byte = lua_tointeger(L, 3);
+  long new_end_byte = lua_tointeger(L, 4);
+  TSPoint start_point = { lua_tointeger(L, 5), lua_tointeger(L, 6) };
+  TSPoint old_end_point = { lua_tointeger(L, 7), lua_tointeger(L, 8) };
+  TSPoint new_end_point = { lua_tointeger(L, 9), lua_tointeger(L, 10) };
+
+  TSInputEdit edit = { start_byte, old_end_byte, new_end_byte,
+                       start_point, old_end_point, new_end_point };
+
+  ts_tree_edit(p->tree, &edit);
+
+  return 0;
+}
+
+
+// Tree methods
+
+/// push tree interface on lua stack.
+///
+/// This takes "ownership" of the tree and will free it
+/// when the wrapper object is garbage collected
+void tslua_push_tree(lua_State *L, TSTree *tree)
+{
+  TSTree **ud = lua_newuserdata(L, sizeof(TSTree *));  // [udata]
+  *ud = tree;
+  lua_getfield(L, LUA_REGISTRYINDEX, REG_KEY);  // [udata, env]
+  lua_getfield(L, -1, "tree-meta");  // [udata, env, meta]
+  lua_setmetatable(L, -3);  // [udata, env]
+  lua_pop(L, 1);  // [udata]
+
+  // table used for node wrappers to keep a reference to tree wrapper
+  // NB: in lua 5.3 the uservalue for the node could just be the tree, but
+  // in lua 5.1 the uservalue (fenv) must be a table.
+  lua_createtable(L, 1, 0); // [udata, reftable]
+  lua_pushvalue(L, -2); // [udata, reftable, udata]
+  lua_rawseti(L, -2, 1); // [udata, reftable]
+  lua_setfenv(L, -2); // [udata]
+  debug_n_trees++;
+}
+
+static TSTree *tree_check(lua_State *L)
+{
+  if (!lua_gettop(L)) {
+    return 0;
+  }
+  if (!lua_isuserdata(L, 1)) {
+    return 0;
+  }
+  // TODO: typecheck!
+  TSTree **ud = lua_touserdata(L, 1);
+  return *ud;
+}
+
+static int tree_gc(lua_State *L)
+{
+  TSTree *tree = tree_check(L);
+  if (!tree) {
+    return 0;
+  }
+
+  ts_tree_delete(tree);
+  debug_n_trees--;
+  return 0;
+}
+
+static int tree_tostring(lua_State *L)
+{
+  lua_pushstring(L, "<tree>");
+  return 1;
+}
+
+static int tree_root(lua_State *L)
+{
+  TSTree *tree = tree_check(L);
+  if (!tree) {
+    return 0;
+  }
+  TSNode root = ts_tree_root_node(tree);
+  push_node(L, root);
+  return 1;
+}
+
+// Node methods
+
+/// push node interface on lua stack
+///
+/// top of stack must either be the tree this node belongs to or another node
+/// of the same tree! This value is not popped. Can only be called inside a
+/// cfunction with the tslua environment.
+static void push_node(lua_State *L, TSNode node)
+{
+  if (ts_node_is_null(node)) {
+    lua_pushnil(L); // [src, nil]
+    return;
+  }
+  TSNode *ud = lua_newuserdata(L, sizeof(TSNode));  // [src, udata]
+  *ud = node;
+  lua_getfield(L, LUA_ENVIRONINDEX, "node-meta");  // [src, udata, meta]
+  lua_setmetatable(L, -2);  // [src, udata]
+  lua_getfenv(L, -2);  // [src, udata, reftable]
+  lua_setfenv(L, -2);  // [src, udata]
+}
+
+static bool node_check(lua_State *L, TSNode *res)
+{
+  if (!lua_gettop(L)) {
+    return 0;
+  }
+  if (!lua_isuserdata(L, 1)) {
+    return 0;
+  }
+  // TODO: typecheck!
+  TSNode *ud = lua_touserdata(L, 1);
+  *res = *ud;
+  return true;
+}
+
+
+static int node_tostring(lua_State *L)
+{
+  TSNode node;
+  if (!node_check(L, &node)) {
+    return 0;
+  }
+  lua_pushstring(L, "<node ");
+  lua_pushstring(L, ts_node_type(node));
+  lua_pushstring(L, ">");
+  lua_concat(L, 3);
+  return 1;
+}
+
+static int node_range(lua_State *L)
+{
+  TSNode node;
+  if (!node_check(L, &node)) {
+    return 0;
+  }
+  TSPoint start = ts_node_start_point(node);
+  TSPoint end = ts_node_end_point(node);
+  lua_pushnumber(L, start.row);
+  lua_pushnumber(L, start.column);
+  lua_pushnumber(L, end.row);
+  lua_pushnumber(L, end.column);
+  return 4;
+}
+
+static int node_start(lua_State *L)
+{
+  TSNode node;
+  if (!node_check(L, &node)) {
+    return 0;
+  }
+  TSPoint start = ts_node_start_point(node);
+  uint32_t start_byte = ts_node_start_byte(node);
+  lua_pushnumber(L, start.row);
+  lua_pushnumber(L, start.column);
+  lua_pushnumber(L, start_byte);
+  return 3;
+}
+
+static int node_child_count(lua_State *L)
+{
+  TSNode node;
+  if (!node_check(L, &node)) {
+    return 0;
+  }
+  uint32_t count = ts_node_child_count(node);
+  lua_pushnumber(L, count);
+  return 1;
+}
+
+static int node_type(lua_State *L)
+{
+  TSNode node;
+  if (!node_check(L, &node)) {
+    return 0;
+  }
+  lua_pushstring(L, ts_node_type(node));
+  return 1;
+}
+
+static int node_symbol(lua_State *L)
+{
+  TSNode node;
+  if (!node_check(L, &node)) {
+    return 0;
+  }
+  TSSymbol symbol = ts_node_symbol(node);
+  lua_pushnumber(L, symbol);
+  return 1;
+}
+
+static int node_child(lua_State *L)
+{
+  TSNode node;
+  if (!node_check(L, &node)) {
+    return 0;
+  }
+  long num = lua_tointeger(L, 2);
+  TSNode child = ts_node_child(node, (uint32_t)num);
+
+  lua_pushvalue(L, 1);
+  push_node(L, child);
+  return 1;
+}
+
+static int node_descendant_for_point_range(lua_State *L)
+{
+  TSNode node;
+  if (!node_check(L, &node)) {
+    return 0;
+  }
+  TSPoint start = {(uint32_t)lua_tointeger(L, 2),
+                   (uint32_t)lua_tointeger(L, 3)};
+  TSPoint end = {(uint32_t)lua_tointeger(L, 4),
+                 (uint32_t)lua_tointeger(L, 5)};
+  TSNode child = ts_node_descendant_for_point_range(node, start, end);
+
+  lua_pushvalue(L, 1);
+  push_node(L, child);
+  return 1;
+}
+
+static int node_parent(lua_State *L)
+{
+  TSNode node;
+  if (!node_check(L, &node)) {
+    return 0;
+  }
+  TSNode parent = ts_node_parent(node);
+  push_node(L, parent);
+  return 1;
+}
+

src/nvim/lua/tree_sitter.h

@@ -0,0 +1,10 @@
+#ifndef NVIM_LUA_TREE_SITTER_H
+#define NVIM_LUA_TREE_SITTER_H
+
+#include "tree_sitter/api.h"
+
+#ifdef INCLUDE_GENERATED_DECLARATIONS
+# include "lua/tree_sitter.h.generated.h"
+#endif
+
+#endif  // NVIM_LUA_TREE_SITTER_H

third-party/CMakeLists.txt

@@ -195,6 +195,9 @@ set(GETTEXT_SHA256 ff942af0e438ced4a8b0ea4b0b6e0d6d657157c5e2364de57baa279c1c125
 set(LIBICONV_URL https://ftp.gnu.org/pub/gnu/libiconv/libiconv-1.15.tar.gz)
 set(LIBICONV_SHA256 ccf536620a45458d26ba83887a983b96827001e92a13847b45e4925cc8913178)
 
+set(UTF8PROC_URL https://github.com/JuliaStrings/utf8proc/archive/v2.2.0.tar.gz)
+set(UTF8PROC_SHA256 3f8fd1dbdb057ee5ba584a539d5cd1b3952141c0338557cb0bdf8cb9cfed5dbf)
+
 if(USE_BUNDLED_UNIBILIUM)
   include(BuildUnibilium)
 endif()
@@ -246,6 +249,9 @@ if(USE_BUNDLED_LIBICONV)
   include(BuildLibiconv)
 endif()
 
+# TODO: build it as a normal lib, so it can be used as a distro lib when available
+include(BuildUtf8proc)
+
 if(WIN32)
   include(GetBinaryDeps)
 

third-party/cmake/BuildUtf8proc.cmake

@@ -0,0 +1,16 @@
+ExternalProject_Add(utf8proc
+PREFIX ${DEPS_BUILD_DIR}
+URL ${UTF8PROC_URL}
+DOWNLOAD_DIR ${DEPS_DOWNLOAD_DIR}/utf8proc
+DOWNLOAD_COMMAND ${CMAKE_COMMAND}
+  -DPREFIX=${DEPS_BUILD_DIR}
+  -DDOWNLOAD_DIR=${DEPS_DOWNLOAD_DIR}/utf8proc
+  -DURL=${UTF8PROC_URL}
+  -DEXPECTED_SHA256=${UTF8PROC_SHA256}
+  -DTARGET=utf8proc
+  -DUSE_EXISTING_SRC_DIR=${USE_EXISTING_SRC_DIR}
+  -P ${CMAKE_CURRENT_SOURCE_DIR}/cmake/DownloadAndExtractFile.cmake
+CONFIGURE_COMMAND true
+BUILD_COMMAND true
+INSTALL_COMMAND true
+)