🏗️ Use LuaNativeObjects

jieba.cpp

@@ -3,22 +3,21 @@
 
 #define DEFAULT_BUFFER_SIZE 1024
 
-using namespace std;
+using cppjieba::Jieba, cppjieba::Word, std::string, std::vector;
 
-static cppjieba::Jieba *jieba;
-
-extern "C" void init(struct jieba_path jieba_path) {
-  jieba = new cppjieba::Jieba{jieba_path.dict_path, jieba_path.model_path,
-                              jieba_path.user_dict_path, jieba_path.idf_path,
-                              jieba_path.stop_word_path};
+extern "C" jieba jieba_new(char *dict_path, char *model_path,
+                           char *user_dict_path, char *idf_path,
+                           char *stop_word_path) {
+  return reinterpret_cast<jieba>(new Jieba{
+      dict_path, model_path, user_dict_path, idf_path, stop_word_path});
 }
 
-extern "C" char **cut(const char *str, bool hmm) {
+extern "C" char **jieba_cut(jieba jieba, const char *str, bool hmm) {
   vector<string> words;
-  vector<cppjieba::Word> jiebawords;
+  vector<Word> jiebawords;
   string s = str;
 
-  (*jieba).Cut(s, words, hmm);
+  reinterpret_cast<Jieba *>(jieba)->Cut(s, words, hmm);
   char **results = (char **)malloc(sizeof(char *) * DEFAULT_BUFFER_SIZE);
   char **p = results;
   for (auto word : words)
@@ -27,7 +26,6 @@ extern "C" char **cut(const char *str, bool hmm) {
   return results;
 }
 
-extern "C" void deinit() {
-  delete jieba;
-  jieba = nullptr;
+extern "C" void jieba_delete(jieba jieba) {
+  delete reinterpret_cast<Jieba *>(jieba);
 }

jieba.h

@@ -4,16 +4,11 @@
 #include <sys/cdefs.h>
 __BEGIN_DECLS
 
-struct jieba_path {
-  const char *dict_path;
-  const char *model_path;
-  const char *user_dict_path;
-  const char *idf_path;
-  const char *stop_word_path;
-};
-void init(struct jieba_path);
-char **cut(const char *str, bool hmm);
-void deinit();
+typedef void *jieba;
+void *jieba_new(char *dict_path, char *model_path, char *user_dict_path,
+                char *idf_path, char *stop_word_path);
+char **jieba_cut(jieba jieba, const char *str, bool hmm);
+void jieba_delete(jieba jieba);
 
 __END_DECLS
 #endif /* jieba.h */

jieba.nobj.lua

@@ -0,0 +1,21 @@
+-- luacheck: ignore 113
+---@diagnostic disable: undefined-global
+c_module "jieba" {
+    use_globals = true,
+    include "jieba.h",
+    object "jieba" {
+        constructor {
+            c_call "void *" "jieba_new" { "char *", "dict_path",
+                "char *", "model_path", "char *", "user_dict_path",
+                "char *", "idf_path", "char *", "stop_word_path" },
+        },
+        destructor "delete" {
+            c_method_call "void" "jieba_delete" {}
+        },
+        method "cut" {
+            c_method_call "char **" "jieba_cut" {
+                "const char *", "str", "bool", "hmm"
+            }
+        }
+    },
+}

lua/jieba/nvim/config.lua

@@ -0,0 +1,17 @@
+-- luacheck: ignore 112 113
+---@diagnostic disable: undefined-global
+local dict_dir = vim.fs.joinpath(
+    vim.fs.dirname(vim.fs.dirname(debug.getinfo(1).source:match("@?(.*)"))),
+    "dict"
+)
+
+return {
+    hmm = true,
+    paths = {
+        vim.fs.joinpath(dict_dir, "jieba.dict.utf8"),
+        vim.fs.joinpath(dict_dir, "hmm_model.utf8"),
+        vim.fn.has("win32") == 0 and "/dev/null" or "nul",
+        vim.fs.joinpath(dict_dir, "idf.utf8"),
+        vim.fs.joinpath(dict_dir, "stop_words.utf8"),
+    }
+}

lua/jieba/nvim/init.lua

@@ -21,14 +21,10 @@
 -- SOFTWARE.
 
 local M = {}
-local jieba = require("jieba.nvim.utils")
+local config = require("jieba.nvim.config")
 local ut = require("jieba.utils")
 
-if not jieba.is_init then
-	require("jieba").init(jieba.jieba_path)
-	jieba.is_init = true
-end
-
+local jieba = require("jieba").jieba(unpack(config.paths))
 local str_match = string.match
 local sub = ut.sub
 local len = vim.api.nvim_strwidth
@@ -412,7 +408,7 @@ local function navigate(primary_index_func, secondary_index_func, backward, buff
 	-- unwrap the row and col from the cursor position
 	local row, col = cursor_pos[1], cursor_pos[2]
 	if row == sentinel_row then
-		pt = parse_tokens(jieba.cut(buffer[row], jieba.hmm))
+		pt = parse_tokens(jieba.cut(buffer[row], config.hmm))
 		pt = stack_merge(pt)
 		col = primary_index_func(pt, col)
 
@@ -435,23 +431,23 @@ local function navigate(primary_index_func, secondary_index_func, backward, buff
 		return { row, col }
 	end
 	-- similar steps for when row is not the sentinel_row
-	pt = parse_tokens(jieba.cut(buffer[row], jieba.hmm))
+	pt = parse_tokens(jieba.cut(buffer[row], config.hmm))
 	pt = stack_merge(pt)
 	col = primary_index_func(pt, col)
 	if col ~= nil then
 		return { row, col }
 	end
 	row = row + row_step
 	while row ~= sentinel_row do
-		pt = parse_tokens(jieba.cut(buffer[row], jieba.hmm))
+		pt = parse_tokens(jieba.cut(buffer[row], config.hmm))
 		pt = stack_merge(pt)
 		col = secondary_index_func(pt)
 		if col ~= nil then
 			return { row, col }
 		end
 		row = row + row_step
 	end
-	pt = parse_tokens(jieba.cut(buffer[row], jieba.hmm))
+	pt = parse_tokens(jieba.cut(buffer[row], config.hmm))
 	pt = stack_merge(pt)
 	col = secondary_index_func(pt)
 	if col == nil then
@@ -549,7 +545,7 @@ end
 M.select_w = function()
 	local cursor_pos = vim.api.nvim_win_get_cursor(0)
 	local current_line = Lines[cursor_pos[1]]
-	local line = parse_tokens(jieba.cut(current_line, jieba.hmm))
+	local line = parse_tokens(jieba.cut(current_line, config.hmm))
 	print(line)
 	line = stack_merge(line)
 	local _, start, row = index_tokens(line, cursor_pos[2])
@@ -573,7 +569,7 @@ end
 
 -- TODO: 高亮当前光标下的词
 -- local function hightlight_under_curosr()
--- 	local line = parse_tokens(jieba.cut(vim.api.nvim_get_current_line(), jieba.hmm))
+-- 	local line = parse_tokens(jieba.cut(vim.api.nvim_get_current_line(), config.hmm))
 -- 	line = stack_merge(line)
 -- 	local cursor_pos = vim.api.nvim_win_get_cursor(0)
 -- 	local _, start, row = index_tokens(line, cursor_pos[2] + 1)

xmake.lua

@@ -1,12 +1,39 @@
--- luacheck: ignore 111 113
+-- luacheck: ignore 111 113 143
 ---@diagnostic disable: undefined-global
+---@diagnostic disable: undefined-field
 add_rules("mode.debug", "mode.release")
 
+rule("lua-native-object")
+do
+    set_extensions(".nobj.lua")
+    before_buildcmd_file(function(target, batchcmds, sourcefile, opt)
+        -- get c source file for lua-native-object
+        local dirname = path.join(target:autogendir(), "rules", "lua-native-object")
+        local sourcefile_c = path.join(dirname, path.basename(sourcefile) .. ".c")
+
+        -- add objectfile
+        local objectfile = target:objectfile(sourcefile_c)
+        table.insert(target:objectfiles(), objectfile)
+
+        -- add commands
+        batchcmds:show_progress(opt.progress, "${color.build.object}compiling.nobj.lua %s", sourcefile)
+        batchcmds:mkdir(path.directory(sourcefile_c))
+        batchcmds:vrunv("native_objects.lua",
+            { "-outpath", dirname, "-gen", "lua", path(sourcefile) })
+        batchcmds:compile(sourcefile_c, objectfile)
+
+        -- add deps
+        batchcmds:add_depfiles(sourcefile)
+        batchcmds:set_depmtime(os.mtime(objectfile))
+        batchcmds:set_depcache(target:dependfile(objectfile))
+    end)
+end
+
 target("jieba")
 do
-    add_rules("luarocks.module")
-    add_files("*.c", "*.cpp")
-    add_includedirs("cppjieba/include", "cppjieba/deps/limonp/include")
+    add_includedirs(".", "cppjieba/include", "cppjieba/deps/limonp/include")
+    add_rules("luarocks.module", "lua-native-object")
+    add_files("*.cpp", "*.nobj.lua")
     add_links("stdc++")
     before_build(
         -- luacheck: ignore 212/target