diff --git a/doc/pic/juliaset.png b/doc/pic/juliaset.png
new file mode 100644
index 0000000..4ff8b2a
Binary files /dev/null and b/doc/pic/juliaset.png differ
diff --git a/module/json.cpp b/module/json.cpp
new file mode 100644
index 0000000..3771206
--- /dev/null
+++ b/module/json.cpp
@@ -0,0 +1,349 @@
+#include "../src/nasal.h"
+#include "../src/nasal_type.h"
+#include "../src/nasal_gc.h"
+
+#include <iostream>
+#include <cstring>
+#include <sstream>
+#include <vector>
+
+namespace nasal {
+
+enum class token_type {
+    tok_eof,
+    tok_lbrace,
+    tok_rbrace,
+    tok_lbrkt,
+    tok_rbrkt,
+    tok_comma,
+    tok_colon,
+    tok_str,
+    tok_num,
+    tok_id,
+    tok_bool
+};
+
+std::string get_content(token_type type) {
+    switch(type) {
+        case token_type::tok_eof: return "eof";
+        case token_type::tok_lbrace: return "`{`";
+        case token_type::tok_rbrace: return "`}`";
+        case token_type::tok_lbrkt: return "`[`";
+        case token_type::tok_rbrkt: return "`]`";
+        case token_type::tok_comma: return "`,`";
+        case token_type::tok_colon: return "`:`";
+        case token_type::tok_str: return "string";
+        case token_type::tok_num: return "number";
+        case token_type::tok_id: return "identifier";
+        case token_type::tok_bool: return "boolean";
+    }
+}
+
+struct token {
+    token_type type;
+    std::string content;
+};
+
+class json {
+private:
+    usize parse_error = 0;
+    std::string text = "";
+    usize line = 1;
+    usize ptr = 0;
+    token this_token;
+
+private:
+    std::string var_generate(var&);
+    std::string vector_generate(nas_vec&);
+    std::string hash_generate(nas_hash&);
+    std::string map_generate(nas_map&);
+
+private:
+    bool is_num(char c) {
+        return std::isdigit(c);
+    }
+    bool is_id(char c) {
+        return std::isalpha(c) || c=='_';
+    }
+    bool check(char c) {
+        return c=='{' || c=='}' || c=='[' || c==']' ||
+               c==':' || c==',' || c=='"' || c=='\'' ||
+               is_num(c) || is_id(c);
+    }
+    void next();
+    void match(token_type);
+    var vector_member(gc*);
+    var vector_object_generate(gc*);
+    void hash_member(nas_hash&, gc*);
+    var hash_object_generate(gc*);
+
+public:
+    std::string stringify(var&);
+    var parse(const std::string&, gc*);
+    bool has_error() const { return parse_error; }
+};
+
+std::string json::var_generate(var& value) {
+    switch(value.type) {
+        case vm_type::vm_num: {
+            std::stringstream out;
+            out << value.num();
+            return out.str();
+        }
+        case vm_type::vm_str: return "\"" + value.str() + "\"";
+        case vm_type::vm_vec: return vector_generate(value.vec());
+        case vm_type::vm_hash: return hash_generate(value.hash());
+        default: break;
+    }
+    return "\"undefined\"";
+}
+
+std::string json::vector_generate(nas_vec& vect) {
+    std::string out = "[";
+    for(auto& i : vect.elems) {
+        out += var_generate(i) + ",";
+    }
+    if (out.back()==',') {
+        out.pop_back();
+    }
+    out += "]";
+    return out;
+}
+
+std::string json::hash_generate(nas_hash& hash) {
+    std::string out = "{";
+    for(auto& i : hash.elems) {
+        out += "\"" + i.first + "\":";
+        out += var_generate(i.second) + ",";
+    }
+    if (out.back()==',') {
+        out.pop_back();
+    }
+    out += "}";
+    return out;
+}
+
+std::string json::map_generate(nas_map& nmap) {
+    std::string out = "{";
+    for(auto& i : nmap.mapper) {
+        out += "\"" + i.first + "\":";
+        out += var_generate(*i.second) + ",";
+    }
+    if (out.back()==',') {
+        out.pop_back();
+    }
+    out += "}";
+    return out;
+}
+
+std::string json::stringify(var& object) {
+    if (object.is_vec()) {
+        return vector_generate(object.vec());
+    } else if (object.is_hash()) {
+        return hash_generate(object.hash());
+    } else if (object.is_map()) {
+        return map_generate(object.map());
+    }
+    return "[]";
+}
+
+void json::next() {
+    while(ptr<text.length() && !check(text[ptr])) {
+        if (text[ptr]=='\n') {
+            ++line;
+        }
+        ++ptr;
+    }
+    if (ptr>=text.length()) {
+        this_token = {token_type::tok_eof, "eof"};
+        return;
+    }
+    auto c = text[ptr];
+    switch(c) {
+        case '{': this_token = {token_type::tok_lbrace, "{"}; ++ptr; return;
+        case '}': this_token = {token_type::tok_rbrace, "}"}; ++ptr; return;
+        case '[': this_token = {token_type::tok_lbrkt, "["}; ++ptr; return;
+        case ']': this_token = {token_type::tok_rbrkt, "]"}; ++ptr; return;
+        case ',': this_token = {token_type::tok_comma, ","}; ++ptr; return;
+        case ':': this_token = {token_type::tok_colon, ":"}; ++ptr; return;
+        default: break;
+    }
+    if (is_num(c)) {
+        auto temp = std::string(1, c);
+        ++ptr;
+        while(ptr<text.length() && (
+            is_num(text[ptr]) ||
+            text[ptr]=='.' ||
+            text[ptr]=='e' ||
+            text[ptr]=='-' ||
+            text[ptr]=='+')) {
+            temp += text[ptr];
+            ++ptr;
+        }
+        --ptr;
+        this_token = {token_type::tok_num, temp};
+    } else if (is_id(c)) {
+        auto temp = std::string(1, c);
+        ++ptr;
+        while(ptr<text.length() && (is_id(text[ptr]) || is_num(text[ptr]))) {
+            temp += text[ptr];
+            ++ptr;
+        }
+        --ptr;
+        if (temp=="true" || temp=="false") {
+            this_token = {token_type::tok_bool, temp};
+        } else {
+            this_token = {token_type::tok_id, temp};
+        }
+    } else if (c=='"' || c=='\'') {
+        auto begin = c;
+        auto temp = std::string("");
+        ++ptr;
+        while(ptr<text.length() && text[ptr]!=begin) {
+            temp += text[ptr];
+            ++ptr;
+            if (text[ptr-1]=='\\' && ptr<text.length()) {
+                temp += text[ptr];
+                ++ptr;
+            }
+        }
+        this_token = {token_type::tok_str, temp};
+    }
+    ++ptr;
+    return;
+}
+
+void json::match(token_type type) {
+    if (this_token.type!=type) {
+        std::cerr << "json::parse: line " << line << ": expect ";
+        std::cerr << get_content(type) << " but get `";
+        std::cerr << this_token.content << "`.\n";
+        ++parse_error;
+    }
+    next();
+    return;
+}
+
+var json::vector_member(gc* ngc) {
+    auto result = nil;
+    if (this_token.type==token_type::tok_lbrace) {
+        result = hash_object_generate(ngc);   
+    } else if (this_token.type==token_type::tok_lbrkt) {
+        result = vector_object_generate(ngc);
+    } else if (this_token.type==token_type::tok_str) {
+        result = ngc->newstr(this_token.content);
+        next();
+    } else if (this_token.type==token_type::tok_num) {
+        result = var::num(str2num(this_token.content.c_str()));
+        next();
+    }
+    return result;
+}
+
+var json::vector_object_generate(gc* ngc) {
+    auto vect_object = ngc->alloc(vm_type::vm_vec);
+    auto& vec = vect_object.vec().elems;
+    ngc->temp = vect_object;
+    match(token_type::tok_lbrkt);
+    vec.push_back(vector_member(ngc));
+    while(this_token.type==token_type::tok_comma) {
+        match(token_type::tok_comma);
+        vec.push_back(vector_member(ngc));
+    }
+    match(token_type::tok_rbrkt);
+    ngc->temp = nil;
+    return vect_object;
+}
+
+void json::hash_member(nas_hash& hash, gc* ngc) {
+    const auto name = this_token.content;
+    if (this_token.type==token_type::tok_rbrace) {
+        return;
+    }
+    if (this_token.type==token_type::tok_str) {
+        match(token_type::tok_str);
+    } else {
+        match(token_type::tok_id);
+    }
+    match(token_type::tok_colon);
+    if (this_token.type==token_type::tok_lbrace) {
+        hash.elems.insert({name, hash_object_generate(ngc)});   
+    } else if (this_token.type==token_type::tok_lbrkt) {
+        hash.elems.insert({name, vector_object_generate(ngc)});
+    } else if (this_token.type==token_type::tok_str ||
+        this_token.type==token_type::tok_bool) {
+        hash.elems.insert({name, ngc->newstr(this_token.content)});
+        next();
+    } else if (this_token.type==token_type::tok_num) {
+        hash.elems.insert({name, var::num(str2num(this_token.content.c_str()))});
+        next();
+    }
+}
+
+var json::hash_object_generate(gc* ngc) {
+    auto hash_object = ngc->alloc(vm_type::vm_hash);
+    ngc->temp = hash_object; // cause problem
+    match(token_type::tok_lbrace);
+    hash_member(hash_object.hash(), ngc);
+    while(this_token.type==token_type::tok_comma) {
+        match(token_type::tok_comma);
+        hash_member(hash_object.hash(), ngc);
+    }
+    match(token_type::tok_rbrace);
+    ngc->temp = nil;
+    return hash_object;
+}
+
+var json::parse(const std::string& input, gc* ngc) {
+    usize parse_error = 0;
+    usize line = 1;
+    usize ptr = 0;
+    this_token = {token_type::tok_eof, ""};
+
+    if (input.empty()) {
+        std::cerr << "json::parse: empty string.\n";
+        ++parse_error;
+        return nil;
+    }
+    text = input;
+    next();
+    if (this_token.type==token_type::tok_lbrkt) {
+        return vector_object_generate(ngc);
+    } else {
+        return hash_object_generate(ngc);
+    }
+    return nil;
+}
+
+var stringify(var* args, usize size, gc* ngc) {
+    auto object = args[0];
+    if (!object.is_vec() && !object.is_hash() && !object.is_map()) {
+        return nas_err("json::stringify", "must use hashmap or vector");
+    }
+    return ngc->newstr(json().stringify(object));
+}
+
+var parse(var* args, usize size, gc* ngc) {
+    auto input = args[0];
+    if (!input.is_str()) {
+        return nas_err("json::parse", "must use string");
+    }
+    json instance;
+    auto result = instance.parse(input.str(), ngc);
+    if (instance.has_error()) {
+        return nas_err("json::parse", "parse error");
+    }
+    return result;
+}
+
+module_func_info func_tbl[] = {
+    {"stringify", stringify},
+    {"parse", parse},
+    {nullptr, nullptr}
+};
+
+extern "C" module_func_info* get() {
+    return func_tbl;
+}
+
+}
\ No newline at end of file
diff --git a/module/libjson.nas b/module/libjson.nas
new file mode 100644
index 0000000..4eb91ba
--- /dev/null
+++ b/module/libjson.nas
@@ -0,0 +1,17 @@
+# module json
+# 2023/11/27 ValKmjolnir
+
+use std.dylib;
+
+var _dynamic_lib = dylib.dlopen("libjson."~(os.platform()=="windows"?"dll":"so"));
+var _stringify = _dynamic_lib.stringify;
+var _parse = _dynamic_lib.parse;
+var _call = dylib.limitcall(1);
+
+var stringify = func(object) {
+    return _call(_stringify, object);
+}
+
+var parse = func(input_string) {
+    return _call(_parse, input_string);
+}
\ No newline at end of file