diff --git a/.gitignore b/.gitignore
index 9374cdb..7e2a35c 100644
--- a/.gitignore
+++ b/.gitignore
@@ -24,4 +24,8 @@
 /players/bot.zip
 /common/proxinject
 /log
-/account
\ No newline at end of file
+/account
+/config.json
+/mjai/bot/mortal.pth
+/mjai/bot/libriichi.*
+/mjai/bot_3p
\ No newline at end of file
diff --git a/README.md b/README.md
index 38b8504..b8abcc1 100644
--- a/README.md
+++ b/README.md
@@ -35,65 +35,46 @@ https://github.com/shinkuan/RandomStuff/assets/35415788/ce1b598d-b1d7-49fe-a175-
 
 ## Setup
 
-### Installation, I suggest you watch the video.
+### Installation.
 
-### Some content are outdated:
-- You don't need to install mjai now.
-- You need to install chromium if you want to use playwright: `playwright install chromium`
-
-[YouTube Video for you to follow.](https://youtu.be/ZN3qbMyZeus)
+[YouTube Video for you to follow.](https://youtu.be/NdAULJgBJ-U)
 
 ### You will need:
-1. Docker
-2. Proxifier or something that can redirect connection to mitmproxy.
-   1. __Remember to trust mitmproxy's Certificate!!!__ Don't know how? Google it.
-3. A `bot.zip` that contains the AI. [Examples](https://github.com/smly/mjai.app/tree/main/examples)
-4. (Optional, Recommend) Use Windows Terminal to open client.py for a nice looking TUI.
-
-__Don't know how to make a bot.zip yourself? Don't worry!__
-
-__Get bot.zip at [Discord](https://discord.gg/Z2wjXUK8bN)__
-
-### Docker:
-1. Install Docker on your PC
-2. `docker pull smly/mjai-client:v3`
+1. A `mortal.pth`. [(Get one from Discord server if you don't have one.)](https://discord.gg/Z2wjXUK8bN)
+2. A `libriichi` that match your system. [(Get it here)](https://github.com/shinkuan/Akagi/releases/tag/v0.1.0-libriichi)
+3. (Optional, Recommend) Use Windows Terminal to open client.py for a nice looking TUI.
+4. (Optional) If you want to use Steam, Majsoul Plus, or anything other client, proxy the client using tools like proxifier.
+
+__Get mortal.pth at [Discord](https://discord.gg/Z2wjXUK8bN)__
+1. Go to #verify and click the ✅ reaction.
+2. Go to #bot-zip
+3. Download a bot you like.
+4. Extract the zip.
+5. And mortal.pth is there.
 
 ### Akagi:
 
 Download `install_akagi.ps1` at [Release](https://github.com/shinkuan/Akagi/releases/latest)
 
-1. Place `install_akagi.ps1` at the folder you want to install Akagi.
-2. Open Powershell as admin.
-3. Go to the folder.
-4. To prevent your PC restrict from executing powershell script, run:
-  - `Set-ExecutionPolicy -Scope Process -ExecutionPolicy Bypass`
-5. Run `install_akagi.ps1`
-6. put `bot.zip` into ./players folder
-
-### if `install_akagi.ps1` not working for you:
-
-Follow the [video](https://youtu.be/ZN3qbMyZeus).
-
-1. `git clone this`
-2. `cd into this repo`
-3. Create a Python Virtual Env `python -m venv venv`
-4. Activate it. `.\venv\Scripts\activate`
-5. `pip install -r requirement.txt`
-6. `playwright install chromium`
-7. put `bot.zip` into ./players folder
-
-### Run
-
-After you __activate the venv__:
-1. Configure your setting at `setting.json`
-2. `python mitm_playwright.py` or `python mitm.py`
-3. `python client.py`
+1. Put `install_akagi.ps1` at the location you want to install Akagi.
+2. Open **Powershell** as **Administrator**
+3. cd in to the directory
+4. Run: `Set-ExecutionPolicy -Scope Process -ExecutionPolicy Bypass`
+5. Run: `install_akagi.ps1`
+6. Open mitmproxy if this is your first time using it.
+7. Close it.
+8. Go to your user home directory `~/.mitmproxy`
+9. Install the certificate.
+10. Put `libriichi` into `./Akagi/mjai/bot` and rename it as `libriichi`
+11. Put `mortal.pth` into `./Akagi/mjai/bot`
 
 ### settings.json
 
  - `Unlocker`: Decide to use [MajsoulUnlocker](https://github.com/shinkuan/MajsoulUnlocker) or not.
  - `v10`: If your Majsoul client in still in v0.10.x and you want to use MajsoulUnlocker, set it to true.
  - `Autoplay`: Autoplay.
+ - `Helper`: To use [mahjong-helper](https://github.com/EndlessCheng/mahjong-helper) or not
+ - `Autohu`: Auto Ron.
  - `Port`:
    - `MITM`: The MITM Port, you should redirect Majsoul connection to this port.
    - `XMLRPC`: The XMLRPC Port.
@@ -126,7 +107,7 @@ On top right is the MJAI Messages, this is the message our bot sent back to us,
 
 Then below is our tehai, it is composed using unicode characters.
 
-Bottom left is the settings. (WIP, currently you should change settings via settings.json)
+Bottom left is the settings. 
 
 Bottom right is the bot's action.
 
@@ -145,6 +126,8 @@ Following guide can minimum the probility of account suspension.
 ### There is no way to guarantee 100% no account suspension currently.
 
 # TODO
+ - [x] 3 Player Mahjong
+   - Already done, but not planned to release yet.
  - [x] Change Setting inside application.
  - [x] Autoplay
    - [ ] Auto use stickers to make opponent think we are not a bot.
@@ -179,7 +162,7 @@ Following guide can minimum the probility of account suspension.
 
 ETH Mainnet: 0x83095C4355E43bDFe9cEf2e439F371900664D41F
 
-Paypal or Others: Maybe? Contact me.
+Paypal or Others: Contact me on Discord.
 
 You can find me at [Discord](https://discord.gg/Z2wjXUK8bN).
 
@@ -204,8 +187,12 @@ Next, you can find me on Discord, where I will assign you a donor role.
 
 # Special Thanks
 
+[Equim-chan/Mortal](https://github.com/Equim-chan/Mortal)
+
 [Majsoul Mod Plus](https://github.com/Avenshy/majsoul_mod_plus)
 
+[mahjong-helper](https://github.com/EndlessCheng/mahjong-helper)
+
 [MahjongRepository/mahjong_soul_api](https://github.com/MahjongRepository/mahjong_soul_api)
 
 [smly/mjai.app](https://github.com/smly/mjai.app)
diff --git a/README_CH.md b/README_CH.md
index 1cf297d..86b3f28 100644
--- a/README_CH.md
+++ b/README_CH.md
@@ -35,64 +35,43 @@ https://github.com/shinkuan/RandomStuff/assets/35415788/ce1b598d-b1d7-49fe-a175-
 
 ### 安裝
 
-[點我到Youtube觀看安裝影片](https://youtu.be/ZN3qbMyZeus)
+[點我到Youtube觀看安裝影片](https://youtu.be/NdAULJgBJ-U)
 
 在開始前，你需要以下東西:
-1. Docker
-2. Proxifier或任何可以將雀魂連線導向到MITM Proxy的工具。
-   1. __記得要信任mitmproxy的憑證!!!__ 不要問我怎麼做。去問Google。
-3. 一個`bot.zip`。 [Examples](https://github.com/smly/mjai.app/tree/main/examples)
-4. (Optional) 推薦使用Windows Terminal，以獲得預期中的UI效果。
+1. `mortal.pth` [(如果你沒有的話，到Discord去下載)](https://discord.gg/Z2wjXUK8bN)
+2. `libriichi` [(到這邊根據你的平台下載對應版本)](https://github.com/shinkuan/Akagi/releases/tag/v0.1.0-libriichi)
+3. (Optional) 推薦使用Windows Terminal，以獲得預期中的UI效果。
+4. (Optional) 如果你要使用Steam或Majsoul Plus之類的，請使用類似Proxifier的軟體將連線導向至MITM
 
-__不知道如何製作bot.zip嗎？不用擔心！__
+__到[Discord](https://discord.gg/Z2wjXUK8bN)下載我提供的mortal.pth__
+1. 到 #verify 頻道點擊 ✅ 驗證身分.
+2. 到 #bot-zip
+3. 選一個你喜歡的bot下載
+4. 解壓縮
 
-__到[Discord](https://discord.gg/Z2wjXUK8bN)下載我提供的bot.zip__
-
-Docker:
-1. 安裝Docker Desktop
-2. 在cmd輸入`docker pull smly/mjai-client:v3`
-
-Akagi:
-
-### 影片部分內容已過時:
-- 現在已經有自動化安裝 `install_akagi.ps1`
-- 現在你不再需要安裝mjai
-- 如果你想使用playwright記得安裝chromium: `playwright install chromium`
+### Akagi:
 
 到[Release](https://github.com/shinkuan/Akagi/releases/latest)下載`install_akagi.ps1` 
 
-1. 把`install_akagi.ps1`放到你想安裝的位置
-2. 以管理員開啟Powershell
-3. cd到安裝位置
-4. 有些電腦會限制Powershell腳本的執行，這也是為什麼我們需要管理員權限，執行以下指令以讓腳本能夠運行：
-  - `Set-ExecutionPolicy -Scope Process -ExecutionPolicy Bypass`
-5. 執行 `install_akagi.ps1`
-6. 將`bot.zip`放入 ./players
-
-### 如果 `install_akagi.ps1` 有問題：
-
-建議你跟著[影片](https://youtu.be/ZN3qbMyZeus)中的教學安裝。
-
-1. 下載這個Repo到你電腦內，或在cmd輸入`git clone <repo>`
-2. 在cmd輸入`cd Akagi`進到資料夾
-3. 創建一個Python venv `python -m venv venv`
-4. 啟動它. `.\venv\Scripts\activate`
-5. `pip install -r requirement.txt`
-6. `playwright install chromium`
-7. put `bot.zip` into ./players folder
-
-### 執行
-
-在**啟動venv**後:
-1. 在`setting.json`調整設定
-2. `python mitm_playwright.py` or `python mitm.py`
-3. `python client.py`
+1. 將 `install_akagi.ps1` 放在您想安裝Akagi的位置。
+2. 以**管理員**身份打開 **Powershell**
+3. cd 到該目錄
+4. 執行： `Set-ExecutionPolicy -Scope Process -ExecutionPolicy Bypass`
+5. 執行： `install_akagi.ps1`
+6. 如果這是您第一次使用mitmproxy，請打開它。
+7. 關閉它。
+8. cd 到用戶主目錄 `~/.mitmproxy`
+9. 安裝證書。
+10. 將 `libriichi` 放入 `./Akagi/mjai/bot` 並將其重新命名為 `libriichi`
+11. 將 `mortal.pth` 放入 `./Akagi/mjai/bot`
 
 ### settings.json
 
  - `Unlocker`: 使用 [MajsoulUnlocker](https://github.com/shinkuan/MajsoulUnlocker)
  - `v10`: 如果你的客戶端還在v0.10.x版本而且你想要使用MajsoulUnlocker，設為true
  - `Autoplay`: 自動打牌.
+ - `Helper`: [mahjong-helper](https://github.com/EndlessCheng/mahjong-helper)
+ - `Autohu`: Auto Ron.
  - `Port`:
    - `MITM`: MITM Port, 你應該將雀魂連線導向到這個Port.
    - `XMLRPC`: The XMLRPC Port.
@@ -125,7 +104,7 @@ LiqiProto訊息隨後被轉錄為mjai格式並發送給機器人。
 
 然後下方是我們的手牌，它是使用Unicode字符組成的。
 
-左下角是設置。（目前還未完成，你應該通過settings.json來更改設置）
+左下角是設置。
 
 右下角是機器人的動作。
 
@@ -146,9 +125,10 @@ LiqiProto訊息隨後被轉錄為mjai格式並發送給機器人。
 ### 目前沒有任何辦法保證完全不封號。
 
 # TODO
-
+ - [x] 三麻模式
+   - 已完成，但尚未決定公布。
  - [x] 在應用程式內更改Setting。
- - [x] 自動打牌 - （能運作，但 __不__ 穩定）
+ - [x] 自動打牌
    - [ ] 自動使用貼圖，讓對手認為我們不是機器人。
    - [ ] 在settings.json中添加隨機時間，讓用戶選擇他們想要的時間。
  - [ ] 混合多個AI的決策，讓我們看起來更像人類，而不是完美的機器人。
@@ -181,17 +161,11 @@ LiqiProto訊息隨後被轉錄為mjai格式並發送給機器人。
 
 ETH Mainnet: 0x83095C4355E43bDFe9cEf2e439F371900664D41F
 
-Paypal或其他: Maybe? Contact me.
+Paypal或其他: 到Discord找我
 
 You can find me at [Discord](https://discord.gg/Z2wjXUK8bN).
 
-### 斗內後我能得到什麼？
-
-首先，非常感謝您願意支持作者。
-
-我會優先考慮捐款者的意見，比如功能要求和錯誤修復。
-
-接下來，您可以在Discord上找到我，我會給您指定一個捐款者身分組。
+![image](https://github.com/shinkuan/RandomStuff/assets/35415788/2b500d5c-21cd-422c-9684-50477835fe13)
 
 <!-- 未來計劃：
 
@@ -207,8 +181,12 @@ You can find me at [Discord](https://discord.gg/Z2wjXUK8bN).
 
 # Special Thanks
 
+[Equim-chan/Mortal](https://github.com/Equim-chan/Mortal)
+
 [Majsoul Mod Plus](https://github.com/Avenshy/majsoul_mod_plus)
 
+[mahjong-helper](https://github.com/EndlessCheng/mahjong-helper)
+
 [MahjongRepository/mahjong_soul_api](https://github.com/MahjongRepository/mahjong_soul_api)
 
 [smly/mjai.app](https://github.com/smly/mjai.app)
diff --git a/action.py b/action.py
index 8f6a128..80b725b 100644
--- a/action.py
+++ b/action.py
@@ -34,6 +34,9 @@
         (10.875, 5.9),           #   2   1   0
         (8.6375, 5.9),           #
         (6.4   , 5.9),
+        (10.875, 4.8),           # Not used
+        (8.6375, 4.8),           # Not used
+        (6.4   , 4.8),           # Not used
     ],
     "candidates": [
         (3.6625,  6.3),         # (-(len/2)+idx+0.5)*2+5
@@ -71,7 +74,8 @@
     2,  # Reach     # Self Discard
     1,  # Zimo      # Self Discard
     1,  # Rong      # Opponent Discard
-    4,  # Ryukyoku  # Self Discard
+    5,  # Ryukyoku  # Self Discard
+    4,  # Nukidora  # Self Discard
 ]
 
 ACTION2TYPE = {
@@ -82,6 +86,7 @@
     "hora": 9,
     #^^^^^^^^^^^^^^^^Opponent Discard^^^^^^^^^^^^^^^^
     "ryukyoku": 10,
+    "nukidora": 11,
     "ankan": 4,
     "kakan": 6,
     "reach": 7,
@@ -110,8 +115,8 @@ def do_autohu(self):
 
     def decide_random_time(self):
         if self.isNewRound:
-            return random.uniform(2.3, 2.5)
-        return random.uniform(0.3, 1.2)
+            return random.uniform(3.5, 4.5)
+        return random.uniform(1.0, 3.2)
 
 
     def click_chiponkan(self, mjai_msg: dict | None, tehai: list[str], tsumohai: str | None):
@@ -144,6 +149,8 @@ def click_chiponkan(self, mjai_msg: dict | None, tehai: list[str], tsumohai: str
         for idx, operation in enumerate(latest_operation_list_temp):
             if operation['type'] == ACTION2TYPE[mjai_msg['type']]:
                 self.page_clicker(LOCATION['actions'][idx])
+                self.do_autohu()
+                self.isNewRound = False
                 break
 
         if mjai_msg['type'] == 'reach':
@@ -280,7 +287,7 @@ def mjai2action(self, mjai_msg: dict | None, tehai: list[str], tsumohai: str | N
             time.sleep(dahai_delay)
             self.click_dahai(mjai_msg, tehai, tsumohai)
             return
-        if mjai_msg['type'] in ['none', 'chi', 'pon', 'daiminkan', 'ankan', 'kakan', 'hora', 'reach', 'ryukyoku']:
+        if mjai_msg['type'] in ['none', 'chi', 'pon', 'daiminkan', 'ankan', 'kakan', 'hora', 'reach', 'ryukyoku', 'nukidora']:
             time.sleep(2)
             self.click_chiponkan(mjai_msg, tehai, tsumohai)
             # kan can have multiple candidates too! ex: tehai=1111m 1111p 111s 11z, tsumohai=1s
diff --git a/client.py b/client.py
index 44334de..2222463 100644
--- a/client.py
+++ b/client.py
@@ -6,7 +6,6 @@
 
 from typing import Any, Coroutine
 from xmlrpc.client import ServerProxy
-import docker
 import json
 from loguru import logger
 
@@ -37,18 +36,6 @@
     AUTOPLAY = settings["Autoplay"]
     ENABLE_PLAYWRIGHT = settings["Playwright"]["enable"]
 
-def get_container_ports():
-    client = docker.from_env()
-    containers = client.containers.list()
-    used_port_list = []
-    for container in containers:
-        ports = container.ports
-        for _, bindings in ports.items():
-            if bindings is not None:
-                used_port_list.append(bindings[0]['HostPort'])
-    used_port_list = [int(p) for p in used_port_list]
-    return used_port_list
-
 
 class FlowScreen(Screen):
 
@@ -154,6 +141,7 @@ def refresh_log(self) -> None:
                                 self.action.latest_operation_list = liqi_msg['data']['data']['operation']['operationList']
                         if liqi_msg['data']['name'] == 'ActionDiscardTile':
                             self.action.isNewRound = False
+                            pass
                         if liqi_msg['data']['name'] == 'ActionNewRound':
                             self.action.isNewRound = True
                             self.action.reached = False
@@ -380,7 +368,6 @@ class Akagi(App):
     def __init__(self, rpc_server, *args, **kwargs) -> None:
         super().__init__(*args, **kwargs)
         self.rpc_server = rpc_server
-        self.mjai_client: dict[str, MjaiPlayerClient]={}
         self.liqi: dict[str, LiqiProto]={}
         self.bridge: dict[str, MajsoulBridge]={}
         self.active_flows = []
@@ -390,25 +377,7 @@ def __init__(self, rpc_server, *args, **kwargs) -> None:
         self.akagi_log_dict= dict() # flow.id -> List[akagi_log]
         self.loguru_log = [] # List[loguru_log]
 
-        self.four_mjai_client = []
-        used_port = get_container_ports()
-        port_num = PORT_NUM
-        four_port_num = []
-        for i in range(4):
-            while port_num in used_port:
-                port_num+=1
-            four_port_num.append(port_num)
-            used_port.append(port_num)
             
-        with ThreadPoolExecutor(max_workers=4) as executor:
-            futures = {executor.submit(self.launch_client, i, four_port_num[i], submission) for i in range(4)}
-            self.four_mjai_client = [future.result() for future in futures]
-        self.four_mjai_client.sort(key=lambda x: x.player_id)
-        
-    def launch_client(self, i, port_num, submission):
-        client = MjaiPlayerClient(submission, timeout=15, port_num=port_num)
-        client.launch_container(i)
-        return client
 
     def on_mount(self) -> None:
         self.update_flow = self.set_interval(1, self.refresh_flow)
@@ -427,7 +396,6 @@ def refresh_flow(self) -> None:
                 self.liqi_msg_dict.pop(flow_id)
                 self.mjai_msg_dict.pop(flow_id)
                 self.akagi_log_dict.pop(flow_id)
-                self.mjai_client[flow_id].delete_container()
         for flow_id in flows:
             try:
                 self.query_one("#FlowContainer")
@@ -442,15 +410,6 @@ def refresh_flow(self) -> None:
                 self.liqi_msg_dict[flow_id] = []
                 self.mjai_msg_dict[flow_id] = []
                 self.akagi_log_dict[flow_id] = []
-                used_port = get_container_ports()
-                port_num = PORT_NUM
-                while port_num in used_port:
-                    port_num+=1
-                self.mjai_client[flow_id] = MjaiPlayerClient(
-                    submission,
-                    timeout=15,
-                    port_num=port_num
-                )
                 self.liqi[flow_id] = LiqiProto()
                 self.bridge[flow_id] = MajsoulBridge()
 
@@ -469,7 +428,7 @@ def get_messages(self):
                     if liqi_msg['method'] == '.lq.FastTest.authGame' and liqi_msg['type'] == MsgType.Req:
                         self.app.push_screen(FlowScreen(flow_id))
                         pass
-                    mjai_msg = self.bridge[flow_id].input(self.four_mjai_client, liqi_msg)
+                    mjai_msg = self.bridge[flow_id].input(liqi_msg)
                     if mjai_msg is not None:
                         if self.bridge[flow_id].reach and mjai_msg["type"] == "dahai":
                             mjai_msg["type"] = "reach"
@@ -492,11 +451,7 @@ def my_sink(self, message) -> None:
     def action_quit(self) -> None:
         self.update_flow.stop()
         self.get_messages_flow.stop()
-        self.rpc_server.reset_message_idx()
-        for flow_id in self.active_flows:
-            self.mjai_client[flow_id].delete_container()
-        for mjai_client in self.four_mjai_client:
-            mjai_client.delete_container()
+        self.rpc_server.reset_message_idx() 
         self.exit()
 
     def action_settings(self) -> None:
@@ -504,11 +459,6 @@ def action_settings(self) -> None:
         pass
 
 def exit_handler():
-    containers = docker.from_env().containers.list()
-    for container in containers:
-        if container.image.tags[0] == 'smly/mjai-client:v3':
-            container.stop()
-            container.remove()
     pass
 
 if __name__ == '__main__':
@@ -529,4 +479,4 @@ def exit_handler():
     except Exception as e:
         exit_handler()
         raise e
-    
\ No newline at end of file
+    
diff --git a/client.tcss b/client.tcss
index 216c4e9..104ab94 100644
--- a/client.tcss
+++ b/client.tcss
@@ -249,6 +249,33 @@ Button.action_ryukyoku.-active {
     border-top: tall $ryukyoku-darken-3;
 }
 
+$nukidora: #d5508d;
+$nukidora-lighten-1: #df79a8;
+$nukidora-lighten-2: #e9a2c3;
+$nukidora-lighten-3: #f3cbdd;
+$nukidora-darken-1: #c32f73;
+$nukidora-darken-2: #9a255b;
+$nukidora-darken-3: #711b42;
+
+Button.action_nukidora {
+    background: $nukidora;
+    color: $text;
+    border-top: tall $nukidora-lighten-3;
+    border-bottom: tall $nukidora-darken-3;
+}
+
+Button.action_nukidora:hover {
+    background: $nukidora-darken-2;
+    color: $text;
+    border-top: tall $nukidora;
+}
+
+Button.action_nukidora.-active {
+    background: $nukidora;
+    border-bottom: tall $nukidora-lighten-3;
+    border-top: tall $nukidora-darken-3;
+}
+
 #bottom_container {
     width: 1fr;
     height: 8;
diff --git a/majsoul2mjai.py b/majsoul2mjai.py
index 6824ed3..6272646 100644
--- a/majsoul2mjai.py
+++ b/majsoul2mjai.py
@@ -6,6 +6,7 @@
 from convert import MS_TILE_2_MJAI_TILE, MJAI_TILE_2_MS_TILE
 from liqi import LiqiProto
 from functools import cmp_to_key
+from loguru import logger
 
 class Operation:
     NoEffect = 0
@@ -37,6 +38,7 @@ def __init__(self) -> None:
         self.mjai_message = []
         self.lastDiscard = None
         self.reach = False
+        self.accept_reach = None
         self.operation = {}
         self.AllReady = False
         self.temp = {}
@@ -44,16 +46,19 @@ def __init__(self) -> None:
         self.my_tehais = ["?"]*13
         self.my_tsumohai = "?"
         self.syncing = False
+
+        self.mjai_client = MjaiPlayerClient()
+        self.is_3p = False
         pass
 
-    def input(self, mjai_client: list[MjaiPlayerClient], parse_msg: dict) -> dict | None:
+    def input(self, parse_msg: dict) -> dict | None:
         # TODO SyncGame
         if parse_msg['method'] == '.lq.FastTest.syncGame' or parse_msg['method'] == '.lq.FastTest.enterGame':
             self.syncing = True
             syncGame_msgs = LiqiProto().parse_syncGame(parse_msg)
             reacts = []
             for msg in syncGame_msgs:
-                reacts.append(self.input(mjai_client, msg))
+                reacts.append(self.input(msg))
             if len(reacts)>=1:
                 self.syncing = False
                 return reacts[-1]
@@ -70,16 +75,18 @@ def input(self, mjai_client: list[MjaiPlayerClient], parse_msg: dict) -> dict |
             self.__init__()
             self.accountId = parse_msg['data']['accountId']
         if parse_msg['method'] == '.lq.FastTest.authGame' and parse_msg['type'] == MsgType.Res:
+            self.is_3p = len(parse_msg['data']['seatList']) == 3
+
             seatList = parse_msg['data']['seatList']
             self.seat = seatList.index(self.accountId)
-            # mjai_client.launch_container(self.seat)
+            self.mjai_client.launch_bot(self.seat, self.is_3p)
             self.mjai_message.append(
                 {
                     'type': 'start_game',
                     'id': self.seat
                 }
             )
-            self.react(mjai_client[self.seat])
+            self.react(self.mjai_client)
             return None
         if parse_msg['method'] == '.lq.ActionPrototype':
             # start_kyoku
@@ -93,6 +100,8 @@ def input(self, mjai_client: list[MjaiPlayerClient], parse_msg: dict) -> dict |
                 kyoku = oya + 1
                 kyotaku = parse_msg['data']['data']['liqibang']
                 scores = parse_msg['data']['data']['scores']
+                if self.is_3p:
+                    scores = scores + [0]
                 tehais = [['?']*13]*4
                 my_tehais = ['?']*13
                 for hai in range(13):
@@ -141,6 +150,10 @@ def input(self, mjai_client: list[MjaiPlayerClient], parse_msg: dict) -> dict |
                 else:
                     raise
 
+            if self.accept_reach is not None:
+                self.mjai_message.append(self.accept_reach)
+                self.accept_reach = None
+
             # According to mjai.app, in the case of an ankan, the dora event comes first, followed by the tsumo event.
             if 'data' in parse_msg['data']:
                 if 'doras' in parse_msg['data']['data']:
@@ -193,12 +206,10 @@ def input(self, mjai_client: list[MjaiPlayerClient], parse_msg: dict) -> dict |
                     }
                 )
                 if parse_msg['data']['data']['isLiqi']:
-                    self.mjai_message.append(
-                        {
-                            'type': 'reach_accepted',
-                            'actor': actor
-                        }
-                    )
+                    self.accept_reach = {
+                                            'type': 'reach_accepted',
+                                            'actor': actor
+                                        }
                 if actor == self.seat:
                     if self.my_tsumohai != "?":
                         self.my_tehais.append(self.my_tsumohai)
@@ -274,7 +285,7 @@ def input(self, mjai_client: list[MjaiPlayerClient], parse_msg: dict) -> dict |
                 actor = parse_msg['data']['data']['seat']
                 match parse_msg['data']['data']['type']:
                     case OperationAnGangAddGang.AnGang:
-                        consumed = [MS_TILE_2_MJAI_TILE[parse_msg['data']['data']['tiles']]]*4
+                        consumed = [MS_TILE_2_MJAI_TILE[parse_msg['data']['data']['tiles']].replace("r","")]*4
                         if parse_msg['data']['data']['tiles'][0] == '5' and parse_msg['data']['data']['tiles'][1] != 'z':
                             consumed[0] += 'r'
                         self.mjai_message.append(
@@ -316,6 +327,25 @@ def input(self, mjai_client: list[MjaiPlayerClient], parse_msg: dict) -> dict |
                                 self.my_tehais.append("?")
                             self.my_tehais.remove(pai)
                             self.my_tehais = sorted(self.my_tehais, key=cmp_to_key(compare_pai))
+
+            if parse_msg['data']['name'] == 'ActionBaBei':
+                actor = parse_msg['data']['data']['seat']
+                self.mjai_message.append(
+                    {
+                        'type': 'nukidora',
+                        'actor': actor,
+                        'pai': 'N'
+                    }
+                )
+                if actor == self.seat:
+                    if self.my_tsumohai != "?":
+                        self.my_tehais.append(self.my_tsumohai)
+                        self.my_tsumohai = "?"
+                    else:
+                        self.my_tehais.append("?")
+                    self.my_tehais.remove("N")
+                    self.my_tehais = sorted(self.my_tehais, key=cmp_to_key(compare_pai))
+
             # hora
             if parse_msg['data']['name'] == 'ActionHule':
                 # actor = parse_msg['data']['hules']['seat']
@@ -339,7 +369,7 @@ def input(self, mjai_client: list[MjaiPlayerClient], parse_msg: dict) -> dict |
                 )
                 self.my_tehais = ["?"]*13
                 self.my_tsumohai = "?"
-                self.react(mjai_client[self.seat])
+                self.react(self.mjai_client)
                 return None
             # notile
             if parse_msg['data']['name'] == 'ActionNoTile':
@@ -351,7 +381,7 @@ def input(self, mjai_client: list[MjaiPlayerClient], parse_msg: dict) -> dict |
                 )
                 self.my_tehais = ["?"]*13
                 self.my_tsumohai = "?"
-                self.react(mjai_client[self.seat])
+                self.react(self.mjai_client)
                 return None
             # ryukyoku
             if parse_msg['data']['name'] == 'ActionLiuJu':
@@ -368,13 +398,13 @@ def input(self, mjai_client: list[MjaiPlayerClient], parse_msg: dict) -> dict |
                 )
                 self.my_tehais = ["?"]*13
                 self.my_tsumohai = "?"
-                self.react(mjai_client[self.seat])
+                self.react(self.mjai_client)
                 return None
             
             if 'data' in parse_msg['data']:
                 if 'operation' in parse_msg['data']['data']:
                     self.operation = parse_msg['data']['data']['operation']
-                    return self.react(mjai_client[self.seat])
+                    return self.react(self.mjai_client)
         # end_game
         if parse_msg['method'] == '.lq.NotifyGameEndResult' or parse_msg['method'] == '.lq.NotifyGameTerminate':
             self.mjai_message.append(
@@ -384,8 +414,8 @@ def input(self, mjai_client: list[MjaiPlayerClient], parse_msg: dict) -> dict |
             )
             self.my_tehais = ["?"]*13
             self.my_tsumohai = "?"
-            self.react(mjai_client[self.seat])
-            mjai_client[self.seat].restart_container(self.seat)
+            self.react(self.mjai_client)
+            self.mjai_client.restart_bot(self.seat)
             return None
         return None
 
diff --git a/mhmp.json b/mhmp.json
index 7b6fe59..3c3fe78 100644
--- a/mhmp.json
+++ b/mhmp.json
@@ -16,7 +16,9 @@
     },
     "mitmdump": {
         "http2": false,
-        "mode": "regular@7878"
+        "mode": [
+            "regular@7878"
+        ]
     },
     "proxinject": {
         "name": "jantama_mahjongsoul",
diff --git a/mitm.py b/mitm.py
index d911af3..b2e57a4 100644
--- a/mitm.py
+++ b/mitm.py
@@ -9,6 +9,7 @@
 import mitmproxy.log
 import mitmproxy.tcp
 import mitmproxy.websocket
+import mhm
 from pathlib import Path
 from optparse import OptionParser
 from mitmproxy import proxy, options, ctx
@@ -60,7 +61,7 @@ def request(self, flow: mitmproxy.http.HTTPFlow):
             elif re.search(r'^https://mahjongsoul\.game\.yo-star\.com/v[0-9\.]+\.w/code\.js$', flow.request.url):
                 flow.request.url = "http://cdn.jsdelivr.net/gh/Avenshy/majsoul_mod_plus/safe_code.js"
 
-async def start_proxy(host, port, enable_unlocker, v10):
+async def start_proxy(host, port, enable_unlocker):
     opts = options.Options(listen_host=host, listen_port=port)
 
     master = DumpMaster(
@@ -71,10 +72,7 @@ async def start_proxy(host, port, enable_unlocker, v10):
     master.addons.add(ClientWebSocket())
     master.addons.add(ClientHTTP())
     if enable_unlocker:
-        if v10:
-            from unlocker_v10 import Unlocker
-        else:
-            from mhm.addons import WebSocketAddon as Unlocker
+        from mhm.addons import WebSocketAddon as Unlocker
         master.addons.add(Unlocker())
     await master.run()
     return master
@@ -129,11 +127,11 @@ def serve_forever(self):
         mitm_port = settings["Port"]["MITM"]
         rpc_port = settings["Port"]["XMLRPC"]
         enable_unlocker = settings["Unlocker"]
-        v10 = settings["v10"]
         enable_helper = settings["Helper"]
         enable_playwright = settings["Playwright"]["enable"]
         playwright_width = settings["Playwright"]["width"]
         playwright_height = settings["Playwright"]["height"]
+        autohu = settings["Autohu"]
         scale = playwright_width / 16
 
     mitm_host="127.0.0.1"
@@ -145,7 +143,6 @@ def serve_forever(self):
     p.add_option("--rpc-host", default=None)
     p.add_option("--rpc-port", default=None)
     p.add_option("--unlocker", default=None)
-    p.add_option("--v10", default=None)
     opts, arguments = p.parse_args()
     if opts.mitm_host is not None:
         mitm_host = opts.mitm_host
@@ -157,17 +154,18 @@ def serve_forever(self):
         rpc_port = int(opts.rpc_port)
     if opts.unlocker is not None:
         enable_unlocker = bool(opts.unlocker)
-    if opts.v10 is not None:
-        v10 = bool(opts.v10)
+
+    print("fetching resver...")
+    mhm.fetch_resver()
 
     with open("mhmp.json", "r") as f:
         mhmp = json.load(f)
-        mhmp["mitmdump"]["mode"] = f"regular@{mitm_port}"
+        mhmp["mitmdump"]["mode"] = [f"regular@{mitm_port}"]
         mhmp["hook"]["enable_aider"] = enable_helper
     with open("mhmp.json", "w") as f:
         json.dump(mhmp, f, indent=4)
     # Create and start the proxy server thread
-    proxy_thread = threading.Thread(target=lambda: asyncio.run(start_proxy(mitm_host, mitm_port, enable_unlocker, v10)))
+    proxy_thread = threading.Thread(target=lambda: asyncio.run(start_proxy(mitm_host, mitm_port, enable_unlocker)))
     proxy_thread.start()
 
     liqiServer = LiqiServer(rpc_host, rpc_port)
@@ -208,7 +206,7 @@ def serve_forever(self):
                     page.mouse.move(x=xy_scale["x"], y=xy_scale["y"])
                     time.sleep(0.1)
                     page.mouse.click(x=xy_scale["x"], y=xy_scale["y"], delay=100)
-                if do_autohu:
+                if do_autohu and autohu:
                     print(f"do_autohu")
                     page.evaluate("() => view.DesktopMgr.Inst.setAutoHule(true)")
                     # page.locator("#layaCanvas").click(position=xy_scale)
@@ -228,12 +226,8 @@ def serve_forever(self):
     mitm_port = settings["Port"]["MITM"]
     rpc_port = settings["Port"]["XMLRPC"]
     enable_unlocker = settings["Unlocker"]
-    v10 = settings["v10"]
 if enable_unlocker:
-    if v10:
-        from unlocker_v10 import Unlocker
-    else:
-        from mhm.addons import WebSocketAddon as Unlocker
+    from mhm.addons import WebSocketAddon as Unlocker
     addons = [ClientWebSocket(), Unlocker()]
 else:
     addons = [ClientWebSocket()]
diff --git a/mjai/bot/PUT mortal.pth HERE b/mjai/bot/PUT mortal.pth HERE
new file mode 100644
index 0000000..e69de29
diff --git a/mjai/bot/bot.py b/mjai/bot/bot.py
new file mode 100644
index 0000000..faaa9a9
--- /dev/null
+++ b/mjai/bot/bot.py
@@ -0,0 +1,44 @@
+import json
+import sys
+
+from loguru import logger
+
+from . import model
+
+
+class Bot:
+    def __init__(self, player_id: int):
+        self.player_id = player_id
+        self.model = model.load_model(player_id)
+
+    def react(self, events: str) -> str:
+        events = json.loads(events)
+
+        # logger.info("hi")
+        return_action = None
+        for e in events:
+            return_action = self.model.react(json.dumps(e, separators=(",", ":")))
+
+        if return_action is None:
+            return json.dumps({"type":"none"}, separators=(",", ":"))
+        else:
+            raw_data = json.loads(return_action)
+            del raw_data["meta"]
+            return json.dumps(raw_data, separators=(",", ":"))
+
+
+def main():
+    player_id = int(sys.argv[1])
+    assert player_id in range(4)
+    bot = Bot(player_id)
+
+    while True:
+        line = sys.stdin.readline().strip()
+        resp = bot.react(line)
+        sys.stdout.write(resp + "\n")
+        sys.stdout.flush()
+
+
+if __name__ == "__main__":
+    # debug()
+    main()
\ No newline at end of file
diff --git a/mjai/bot/model.py b/mjai/bot/model.py
new file mode 100644
index 0000000..cb96791
--- /dev/null
+++ b/mjai/bot/model.py
@@ -0,0 +1,360 @@
+import numpy as np
+import torch
+import pathlib
+from torch import nn, Tensor
+from torch.nn import init, functional as F
+from torch.nn.utils.rnn import pack_padded_sequence, pad_sequence
+from torch.distributions import Normal, Categorical
+from typing import *
+from functools import partial
+from itertools import permutations
+from .libriichi.mjai import Bot
+from .libriichi.consts import obs_shape, oracle_obs_shape, ACTION_SPACE, GRP_SIZE
+
+class ChannelAttention(nn.Module):
+    def __init__(self, channels, ratio=16, actv_builder=nn.ReLU, bias=True):
+        super().__init__()
+        self.shared_mlp = nn.Sequential(
+            nn.Linear(channels, channels // ratio, bias=bias),
+            actv_builder(),
+            nn.Linear(channels // ratio, channels, bias=bias),
+        )
+        if bias:
+            for mod in self.modules():
+                if isinstance(mod, nn.Linear):
+                    nn.init.constant_(mod.bias, 0)
+
+    def forward(self, x):
+        avg_out = self.shared_mlp(x.mean(-1))
+        max_out = self.shared_mlp(x.amax(-1))
+        weight = (avg_out + max_out).sigmoid()
+        x = weight.unsqueeze(-1) * x
+        return x
+
+class ResBlock(nn.Module):
+    def __init__(
+        self,
+        channels,
+        *,
+        norm_builder = nn.Identity,
+        actv_builder = nn.ReLU,
+        pre_actv = False,
+    ):
+        super().__init__()
+        self.pre_actv = pre_actv
+
+        if pre_actv:
+            self.res_unit = nn.Sequential(
+                norm_builder(),
+                actv_builder(),
+                nn.Conv1d(channels, channels, kernel_size=3, padding=1, bias=False),
+                norm_builder(),
+                actv_builder(),
+                nn.Conv1d(channels, channels, kernel_size=3, padding=1, bias=False),
+            )
+        else:
+            self.res_unit = nn.Sequential(
+                nn.Conv1d(channels, channels, kernel_size=3, padding=1, bias=False),
+                norm_builder(),
+                actv_builder(),
+                nn.Conv1d(channels, channels, kernel_size=3, padding=1, bias=False),
+                norm_builder(),
+            )
+            self.actv = actv_builder()
+        self.ca = ChannelAttention(channels, actv_builder=actv_builder, bias=True)
+
+    def forward(self, x):
+        out = self.res_unit(x)
+        out = self.ca(out)
+        out = out + x
+        if not self.pre_actv:
+            out = self.actv(out)
+        return out
+
+class ResNet(nn.Module):
+    def __init__(
+        self,
+        in_channels,
+        conv_channels,
+        num_blocks,
+        *,
+        norm_builder = nn.Identity,
+        actv_builder = nn.ReLU,
+        pre_actv = False,
+    ):
+        super().__init__()
+
+        blocks = []
+        for _ in range(num_blocks):
+            blocks.append(ResBlock(
+                conv_channels,
+                norm_builder = norm_builder,
+                actv_builder = actv_builder,
+                pre_actv = pre_actv,
+            ))
+
+        layers = [nn.Conv1d(in_channels, conv_channels, kernel_size=3, padding=1, bias=False)]
+        if pre_actv:
+            layers += [*blocks, norm_builder(), actv_builder()]
+        else:
+            layers += [norm_builder(), actv_builder(), *blocks]
+        layers += [
+            nn.Conv1d(conv_channels, 32, kernel_size=3, padding=1),
+            actv_builder(),
+            nn.Flatten(),
+            nn.Linear(32 * 34, 1024),
+        ]
+        self.net = nn.Sequential(*layers)
+
+    def forward(self, x):
+        return self.net(x)
+
+class Brain(nn.Module):
+    def __init__(self, *, conv_channels, num_blocks, is_oracle=False, version=1):
+        super().__init__()
+        self.is_oracle = is_oracle
+        self.version = version
+
+        in_channels = obs_shape(version)[0]
+        if is_oracle:
+            in_channels += oracle_obs_shape(version)[0]
+
+        norm_builder = partial(nn.BatchNorm1d, conv_channels, momentum=0.01)
+        actv_builder = partial(nn.Mish, inplace=True)
+        pre_actv = True
+
+        match version:
+            case 1:
+                actv_builder = partial(nn.ReLU, inplace=True)
+                pre_actv = False
+                self.latent_net = nn.Sequential(
+                    nn.Linear(1024, 512),
+                    nn.ReLU(inplace=True),
+                )
+                self.mu_head = nn.Linear(512, 512)
+                self.logsig_head = nn.Linear(512, 512)
+            case 2:
+                pass
+            case 3 | 4:
+                norm_builder = partial(nn.BatchNorm1d, conv_channels, momentum=0.01, eps=1e-3)
+            case _:
+                raise ValueError(f'Unexpected version {self.version}')
+
+        self.encoder = ResNet(
+            in_channels = in_channels,
+            conv_channels = conv_channels,
+            num_blocks = num_blocks,
+            norm_builder = norm_builder,
+            actv_builder = actv_builder,
+            pre_actv = pre_actv,
+        )
+        self.actv = actv_builder()
+
+        # always use EMA or CMA when True
+        self._freeze_bn = False
+
+    def forward(self, obs, invisible_obs: Optional[Tensor] = None) -> Union[Tuple[Tensor, Tensor], Tensor]:
+        if self.is_oracle:
+            assert invisible_obs is not None
+            obs = torch.cat((obs, invisible_obs), dim=1)
+        phi = self.encoder(obs)
+
+        match self.version:
+            case 1:
+                latent_out = self.latent_net(phi)
+                mu = self.mu_head(latent_out)
+                logsig = self.logsig_head(latent_out)
+                return mu, logsig
+            case 2 | 3 | 4:
+                return self.actv(phi)
+            case _:
+                raise ValueError(f'Unexpected version {self.version}')
+
+    def train(self, mode=True):
+        super().train(mode)
+        if self._freeze_bn:
+            for mod in self.modules():
+                if isinstance(mod, nn.BatchNorm1d):
+                    mod.eval()
+                    # I don't think this benefits
+                    # module.requires_grad_(False)
+        return self
+
+    def reset_running_stats(self):
+        for mod in self.modules():
+            if isinstance(mod, nn.BatchNorm1d):
+                mod.reset_running_stats()
+
+    def freeze_bn(self, value: bool):
+        self._freeze_bn = value
+        return self.train(self.training)
+
+class AuxNet(nn.Module):
+    def __init__(self, dims=None):
+        super().__init__()
+        self.dims = dims
+        self.net = nn.Linear(1024, sum(dims), bias=False)
+
+    def forward(self, x):
+        return self.net(x).split(self.dims, dim=-1)
+
+class DQN(nn.Module):
+    def __init__(self, *, version=1):
+        super().__init__()
+        self.version = version
+        match version:
+            case 1:
+                self.v_head = nn.Linear(512, 1)
+                self.a_head = nn.Linear(512, ACTION_SPACE)
+            case 2 | 3:
+                hidden_size = 512 if version == 2 else 256
+                self.v_head = nn.Sequential(
+                    nn.Linear(1024, hidden_size),
+                    nn.Mish(inplace=True),
+                    nn.Linear(hidden_size, 1),
+                )
+                self.a_head = nn.Sequential(
+                    nn.Linear(1024, hidden_size),
+                    nn.Mish(inplace=True),
+                    nn.Linear(hidden_size, ACTION_SPACE),
+                )
+            case 4:
+                self.net = nn.Linear(1024, 1 + ACTION_SPACE)
+                nn.init.constant_(self.net.bias, 0)
+
+    def forward(self, phi, mask):
+        if self.version == 4:
+            v, a = self.net(phi).split((1, ACTION_SPACE), dim=-1)
+        else:
+            v = self.v_head(phi)
+            a = self.a_head(phi)
+        a_sum = a.masked_fill(~mask, 0.).sum(-1, keepdim=True)
+        mask_sum = mask.sum(-1, keepdim=True)
+        a_mean = a_sum / mask_sum
+        q = (v + a - a_mean).masked_fill(~mask, -torch.inf)
+        return q
+    
+
+class MortalEngine:
+    def __init__(
+        self,
+        brain,
+        dqn,
+        is_oracle,
+        version,
+        device = None,
+        stochastic_latent = False,
+        enable_amp = False,
+        enable_quick_eval = True,
+        enable_rule_based_agari_guard = False,
+        name = 'NoName',
+        boltzmann_epsilon = 0,
+        boltzmann_temp = 1,
+        top_p = 1,
+    ):
+        self.engine_type = 'mortal'
+        self.device = device or torch.device('cpu')
+        assert isinstance(self.device, torch.device)
+        self.brain = brain.to(self.device).eval()
+        self.dqn = dqn.to(self.device).eval()
+        self.is_oracle = is_oracle
+        self.version = version
+        self.stochastic_latent = stochastic_latent
+
+        self.enable_amp = enable_amp
+        self.enable_quick_eval = enable_quick_eval
+        self.enable_rule_based_agari_guard = enable_rule_based_agari_guard
+        self.name = name
+
+        self.boltzmann_epsilon = boltzmann_epsilon
+        self.boltzmann_temp = boltzmann_temp
+        self.top_p = top_p
+
+    def react_batch(self, obs, masks, invisible_obs):
+        with (
+            torch.autocast(self.device.type, enabled=self.enable_amp),
+            torch.no_grad(),
+        ):
+            return self._react_batch(obs, masks, invisible_obs)
+
+    def _react_batch(self, obs, masks, invisible_obs):
+        obs = torch.as_tensor(np.stack(obs, axis=0), device=self.device)
+        masks = torch.as_tensor(np.stack(masks, axis=0), device=self.device)
+        invisible_obs = None
+        if self.is_oracle:
+            invisible_obs = torch.as_tensor(np.stack(invisible_obs, axis=0), device=self.device)
+        batch_size = obs.shape[0]
+
+        match self.version:
+            case 1:
+                mu, logsig = self.brain(obs, invisible_obs)
+                if self.stochastic_latent:
+                    latent = Normal(mu, logsig.exp() + 1e-6).sample()
+                else:
+                    latent = mu
+                q_out = self.dqn(latent, masks)
+            case 2 | 3 | 4:
+                phi = self.brain(obs)
+                q_out = self.dqn(phi, masks)
+
+        if self.boltzmann_epsilon > 0:
+            is_greedy = torch.full((batch_size,), 1-self.boltzmann_epsilon, device=self.device).bernoulli().to(torch.bool)
+            logits = (q_out / self.boltzmann_temp).masked_fill(~masks, -torch.inf)
+            sampled = sample_top_p(logits, self.top_p)
+            actions = torch.where(is_greedy, q_out.argmax(-1), sampled)
+        else:
+            is_greedy = torch.ones(batch_size, dtype=torch.bool, device=self.device)
+            actions = q_out.argmax(-1)
+
+        return actions.tolist(), q_out.tolist(), masks.tolist(), is_greedy.tolist()
+
+def sample_top_p(logits, p):
+    if p >= 1:
+        return Categorical(logits=logits).sample()
+    if p <= 0:
+        return logits.argmax(-1)
+    probs = logits.softmax(-1)
+    probs_sort, probs_idx = probs.sort(-1, descending=True)
+    probs_sum = probs_sort.cumsum(-1)
+    mask = probs_sum - probs_sort > p
+    probs_sort[mask] = 0.
+    sampled = probs_idx.gather(-1, probs_sort.multinomial(1)).squeeze(-1)
+    return sampled
+
+def load_model(seat: int) -> Bot:
+
+    # check if GPU is available
+    if torch.cuda.is_available():
+        device = torch.device('cuda')
+    else:
+        device = torch.device('cpu')
+
+    # control_state_file = "./mortal_offline_v6_510k.pth"
+
+    # latest binary model
+    control_state_file = "./mortal.pth"
+
+    # Get the path of control_state_file = current directory / control_state_file
+    control_state_file = pathlib.Path(__file__).parent / control_state_file
+
+
+    state = torch.load(control_state_file, map_location=device)
+    mortal = Brain(version=state['config']['control']['version'], conv_channels=state['config']['resnet']['conv_channels'], num_blocks=state['config']['resnet']['num_blocks']).eval()
+    dqn = DQN(version=state['config']['control']['version']).eval()
+    mortal.load_state_dict(state['mortal'])
+    dqn.load_state_dict(state['current_dqn'])
+
+    engine = MortalEngine(
+        mortal,
+        dqn,
+        is_oracle = False,
+        device = device,
+        enable_amp = False,
+        enable_quick_eval = False,
+        enable_rule_based_agari_guard = True,
+        name = 'mortal',
+        version= state['config']['control']['version']
+    )
+
+    bot = Bot(engine, seat)
+    return bot
\ No newline at end of file
diff --git a/mjai/player.py b/mjai/player.py
index 4e7a39e..6cedf20 100644
--- a/mjai/player.py
+++ b/mjai/player.py
@@ -14,6 +14,13 @@
 import requests
 from loguru import logger
 from mjai.exceptions import EngineRuntimeError, TimeoutExpired
+from .bot.bot import Bot
+try:
+    from .bot_3p.bot import Bot as Bot3p
+    three_player = True
+except Exception:
+    three_player = False
+    pass
 
 # MEMORY_SIZE = "1G"
 # CPU_CORES = "1"
@@ -22,181 +29,36 @@
 class MjaiPlayerClient:
     def __init__(
         self,
-        submission_path: Path | str,
-        docker_image_name: str = "docker.io/smly/mjai-client:v3",
-        port_num: int = 28088,
-        timeout: float = 100,
     ) -> None:
-        self.submission_path: Path = Path(submission_path)
-        self.docker_image_name = docker_image_name
-        self.container_name: str | None = None
-        self.proc: subprocess.Popen | None = None
-        self.port_num = port_num
-        self.timeout = timeout
         self.player_id = 0
 
-    def check_docker_image_available(self) -> bool:
-        if shutil.which("docker") is None:
-            return False
-
-        command_pull = [
-            "docker",
-            "pull",
-            "docker.io/smly/mjai-client:v3",
-        ]
-        proc = subprocess.Popen(
-            command_pull,
-            bufsize=0,
-            stdin=subprocess.DEVNULL,
-            stdout=subprocess.DEVNULL,
-            stderr=subprocess.DEVNULL,
-        )
-        proc.wait()
-
-        command_images = [
-            "docker",
-            "images",
-            "--format",
-            "{{.Repository}}",
-            "smly/mjai-client",
-        ]
-        proc = subprocess.Popen(
-            command_images,
-            bufsize=0,
-            stdin=subprocess.DEVNULL,
-            stdout=subprocess.PIPE,
-            stderr=subprocess.DEVNULL,
-        )
-        proc.wait()
-
-        if proc.stdout is None:
-            return False
-
-        if proc.stdout.readline().strip().decode("utf8") != "smly/mjai-client":
-            return False
-
-        return True
-
-    def launch_container(self, player_id: int) -> None:
-        if self.proc is not None:
-            # raise ValueError("Container is already running")
-            return
+        self.bot = None
 
+    def launch_bot(self, player_id: int, is_3p = False) -> None:
         self.player_id = player_id
-        logger.info(f"Start docker container for player{player_id}")
-        command = [
-            "docker",
-            "run",
-            "-d",
-            # "--rm",
-            "--platform",
-            "linux/x86_64",
-            "-w",
-            "/workspace",
-            "-p",
-            f"{self.port_num}:3000",
-            # "--memory",
-            # MEMORY_SIZE,
-            # "--cpus",
-            # CPU_CORES,
-            "--mount",
-            f"type=bind,src={self.submission_path.resolve()},dst=/bot.zip,readonly",  # noqa: E501
-            "--mount",
-            f"type=bind,src={Path(__file__).parent / 'http_server/server.py'},dst=/workspace/00__server__.py,readonly",  # noqa: E501
-            self.docker_image_name,
-            "/workspace/.pyenv/shims/python",
-            "-u",
-            "/workspace/00__server__.py",
-            f"{player_id}",
-        ]
-        logger.info("cmd: " + " ".join(command))
-        proc = subprocess.Popen(
-            command,
-            bufsize=0,
-            stdin=subprocess.PIPE,
-            stdout=subprocess.PIPE,
-            stderr=subprocess.PIPE,
-        )
-        assert proc.stdout is not None
-        self.container_name = proc.stdout.readline().strip().decode("utf8")
-        logger.info(f"Started. Container ID: {self.container_name}")
-
-        # Wait until the server is ready
-        logger.info(f"Wait until the player {player_id} is ready")
-        # Workaround: http server が立ち上がる前にリクエストが飛ぶとエラーになる
-        time.sleep(1.0)
-
-        wait_start_ts = time.time()
-        while True:
-            try:
-                resp = requests.get(
-                    f"http://localhost:{self.port_num}", timeout=100
-                )
-                if resp.status_code == 200:
-                    break
-            except requests.exceptions.ConnectionError:
-                time.sleep(0.1)
-                if time.time() - wait_start_ts > 100.0:
-                    raise ValueError(
-                        "Failed to receive response from http server: timeout"
-                    )
-                continue
-        logger.info(f"Done. Player {player_id} is ready!")
-
-        self.proc = proc
-
-        if proc.stdout is None:
-            raise ValueError("Failed to launch container: stdout is None")
-        if proc.stderr is None:
-            raise ValueError("Failed to launch container: stderr is None")
-
-    def delete_container(self):
-        if self.container_name is None:
-            return
-
-        command = [
-            "docker",
-            "rm",
-            "-f",
-            self.container_name,
-        ]
-        proc = subprocess.Popen(
-            command,
-            bufsize=0,
-            stdin=subprocess.DEVNULL,
-            stdout=subprocess.DEVNULL,
-            stderr=subprocess.DEVNULL,
-        )
-        proc.wait()
-        self.proc = None
-
-    def restart_container(self, player_id: int) -> None:
-        self.delete_container()
-        self.launch_container(player_id)
+        if is_3p:
+            if not three_player:
+                raise NotImplementedError("3p bot is not implemented")
+            self.bot = Bot3p(player_id)
+        else:
+            self.bot = Bot(player_id)
+
+    def delete_bot(self):
+        self.bot = None
+
+    def restart_bot(self, player_id: int) -> None:
+        self.delete_bot()
+        self.launch_bot(player_id)
 
     def react(self, events: str) -> str:
-        if self.proc is None:
-            raise ValueError("Container is not running (3)")
+        if self.bot is None:
+            raise ValueError("bot is not running (3)")
 
         try:
             input_data = events.encode("utf8")
 
-            timeout_per_action = self.timeout
-            if json.loads(events)[0]["type"] in ["start_game", "start_kyoku"]:
-                # Workaround: To avoid timeouts
-                # Some bots time out when loading models with start_game.
-                timeout_per_action = 100.0
-                if json.loads(events)[0]["type"] == "start_game":
-                    self.player_id = json.loads(events)[0]["id"]
-
             logger.debug(f"{self.player_id} <- {input_data}")
-            resp = requests.post(
-                f"http://localhost:{self.port_num}",
-                data=input_data,
-                timeout=timeout_per_action,
-            )
-            assert resp.status_code == 200
-            outs = resp.content.decode("utf8")
+            outs = self.bot.react(input_data)
             logger.debug(f"{self.player_id} -> {outs}")
 
             if (
diff --git a/requirement.txt b/requirement.txt
index ea0e371..8b27039 100644
--- a/requirement.txt
+++ b/requirement.txt
@@ -6,4 +6,5 @@ Pillow==10.1.0
 protobuf==4.25.1
 rich==13.7.0
 textual==0.46.0
-playwright==1.41.0
\ No newline at end of file
+playwright==1.41.0
+torch
\ No newline at end of file
diff --git a/resver.json b/resver.json
index 0c75a6c..fe4ad3c 100644
--- a/resver.json
+++ b/resver.json
@@ -1 +1 @@
-{"version": "0.11.12.w", "emotes": {"200001": [10, 11, 12, 14, 15, 16, 17, 18, 888, 966, 983, 984, 996, 997], "200002": [10, 11, 12, 14, 15, 16, 17, 18, 888, 995, 998], "200003": [10, 11, 12, 14, 15, 16, 17, 18, 888, 959, 979, 999], "200004": [10, 11, 12, 14, 15, 16, 17, 18, 888, 957, 986, 994, 999], "200005": [10, 11, 12, 14, 15, 16, 17, 18, 888, 968, 989, 997], "200006": [10, 11, 12, 14, 15, 16, 17, 18, 888, 961, 977, 997], "200007": [10, 11, 12, 14, 15, 16, 17, 18, 888, 961, 970, 976, 993, 999], "200008": [10, 11, 12, 14, 15, 16, 17, 18, 888, 972, 991, 997], "200009": [10, 11, 12, 14, 15, 16, 17, 18, 888, 966, 994, 998], "200010": [10, 11, 12, 14, 15, 16, 17, 18, 888, 965, 982], "200011": [10, 11, 12, 14, 15, 16, 17, 18, 888, 970, 977, 998], "200012": [10, 11, 12, 14, 15, 16, 17, 18, 888, 959, 998], "200013": [10, 11, 12, 14, 15, 16, 17, 18, 888, 966, 988, 999], "200014": [10, 11, 12, 14, 15, 16, 17, 18, 888, 969, 982, 993], "200015": [10, 11, 12, 14, 15, 16, 17, 18, 888, 994], "200016": [10, 11, 12, 14, 15, 16, 17, 18, 888, 961, 970, 988, 996], "200017": [10, 11, 12, 14, 15, 16, 17, 18, 888, 962, 982, 995], "200018": [9, 10, 11, 12, 14, 15, 16, 17, 18, 888, 962, 994], "200019": [9, 10, 11, 12, 14, 15, 16, 17, 18, 888, 972, 981, 986], "200020": [10, 11, 12, 14, 888, 967, 988, 994], "200021": [10, 11, 12, 14, 888, 981, 985], "200022": [10, 11, 12, 14, 888, 991], "200023": [10, 11, 12, 14, 888, 955, 981], "200024": [10, 11, 12, 14, 888, 960, 976, 992], "200025": [10, 11, 12, 14, 888, 957, 982], "200026": [10, 11, 12, 14, 888, 973, 989], "200027": [10, 11, 12, 14, 888, 968, 988], "200028": [10, 11, 12, 14, 888, 971, 991], "200029": [10, 11, 12, 14, 888, 964, 976, 992], "200030": [10, 11, 12, 14, 888, 955, 979], "200031": [10, 11, 12, 14, 888, 959, 978], "200032": [10, 11, 12, 14, 888, 957, 976, 985], "200033": [10, 11, 12, 14, 888, 969], "200034": [10, 11, 12, 14, 969, 990], "200035": [10, 11, 12, 14, 969, 990], "200036": [10, 11, 12, 14, 969, 990], "200037": [10, 11, 12, 14, 969, 990], "200038": [10, 11, 12, 14, 888, 959, 967, 974], "200039": [10, 11, 12, 14, 888, 961, 970], "200040": [10, 11, 12, 14, 987], "200041": [10, 11, 12, 14, 987], "200042": [10, 11, 12, 14, 987], "200043": [10, 11, 12, 14, 987], "200044": [10, 11, 12, 14, 888, 960, 978], "200045": [10, 11, 12, 14, 888, 973, 981], "200046": [10, 11, 12, 14, 888, 967], "200047": [10, 11, 12, 14, 888, 957, 967], "200048": [10, 11, 12, 14, 888, 965, 971], "200049": [10, 11, 12, 14, 888, 964], "200050": [10, 11, 12, 14, 980], "200051": [10, 11, 12, 14, 980], "200052": [10, 11, 12, 14, 888], "200053": [10, 11, 12, 14, 888, 958], "200054": [10, 11, 12, 14, 888, 974], "200055": [10, 11, 12, 14, 975], "200056": [10, 11, 12, 14, 975], "200057": [10, 11, 12, 14, 975], "200058": [10, 11, 12, 14, 975], "200059": [10, 11, 12, 14, 888, 955, 960], "200060": [10, 11, 12, 14, 888, 958], "200061": [10, 11, 12, 14, 888], "200062": [10, 11, 12, 14, 969], "200063": [10, 11, 12, 14, 969], "200064": [10, 11, 12, 14, 969], "200065": [10, 11, 12, 14, 969], "200066": [10, 11, 12, 14, 888, 955, 962], "200067": [10, 11, 12, 14, 888], "200068": [10, 11, 12, 14, 955], "200069": [10, 11, 12, 14], "200070": [10, 11, 12, 14, 963], "200071": [10, 11, 12, 14, 963], "200072": [10, 11, 12, 14, 963], "200073": [10, 11, 12, 14, 963], "200074": [10, 11, 12, 14], "200075": [10, 11, 12, 14], "200076": [10, 11, 12, 14], "200077": [10, 11, 12, 14], "200078": [10, 11, 12, 14], "200079": [10, 11, 12, 956], "200080": [10, 11, 12, 956], "200081": [10, 11, 12, 956], "200082": [10, 11, 12, 956], "200083": [10, 11, 12], "200090": [10, 11, 12]}}
\ No newline at end of file
+{"version": "0.11.21.w", "emotes": {"200001": [10, 11, 12, 14, 15, 16, 17, 18, 888, 966, 983, 984, 996, 997], "200002": [10, 11, 12, 14, 15, 16, 17, 18, 888, 995, 998], "200003": [10, 11, 12, 14, 15, 16, 17, 18, 888, 959, 979, 999], "200004": [10, 11, 12, 14, 15, 16, 17, 18, 888, 957, 986, 994, 999], "200005": [10, 11, 12, 14, 15, 16, 17, 18, 888, 968, 989, 997], "200006": [10, 11, 12, 14, 15, 16, 17, 18, 888, 961, 977, 997], "200007": [10, 11, 12, 14, 15, 16, 17, 18, 888, 961, 970, 976, 993, 999], "200008": [10, 11, 12, 14, 15, 16, 17, 18, 888, 972, 991, 997], "200009": [10, 11, 12, 14, 15, 16, 17, 18, 888, 966, 994, 998], "200010": [10, 11, 12, 14, 15, 16, 17, 18, 888, 965, 982], "200011": [10, 11, 12, 14, 15, 16, 17, 18, 888, 970, 977, 998], "200012": [10, 11, 12, 14, 15, 16, 17, 18, 888, 959, 998], "200013": [10, 11, 12, 14, 15, 16, 17, 18, 888, 966, 988, 999], "200014": [10, 11, 12, 14, 15, 16, 17, 18, 888, 969, 982, 993], "200015": [10, 11, 12, 14, 15, 16, 17, 18, 888, 994], "200016": [10, 11, 12, 14, 15, 16, 17, 18, 888, 961, 970, 988, 996], "200017": [10, 11, 12, 14, 15, 16, 17, 18, 888, 962, 982, 995], "200018": [9, 10, 11, 12, 14, 15, 16, 17, 18, 888, 962, 994], "200019": [9, 10, 11, 12, 14, 15, 16, 17, 18, 888, 972, 981, 986], "200020": [10, 11, 12, 14, 888, 967, 988, 994], "200021": [10, 11, 12, 14, 888, 954, 981, 985], "200022": [10, 11, 12, 14, 888, 991], "200023": [10, 11, 12, 14, 888, 955, 981], "200024": [10, 11, 12, 14, 888, 960, 976, 992], "200025": [10, 11, 12, 14, 888, 957, 982], "200026": [10, 11, 12, 14, 888, 973, 989], "200027": [10, 11, 12, 14, 888, 968, 988], "200028": [10, 11, 12, 14, 888, 971, 991], "200029": [10, 11, 12, 14, 888, 964, 976, 992], "200030": [10, 11, 12, 14, 888, 955, 979], "200031": [10, 11, 12, 14, 888, 959, 978], "200032": [10, 11, 12, 14, 888, 957, 976, 985], "200033": [10, 11, 12, 14, 888, 969], "200034": [10, 11, 12, 14, 969, 990], "200035": [10, 11, 12, 14, 969, 990], "200036": [10, 11, 12, 14, 969, 990], "200037": [10, 11, 12, 14, 969, 990], "200038": [10, 11, 12, 14, 888, 959, 967, 974], "200039": [10, 11, 12, 14, 888, 961, 970], "200040": [10, 11, 12, 14, 987], "200041": [10, 11, 12, 14, 987], "200042": [10, 11, 12, 14, 987], "200043": [10, 11, 12, 14, 987], "200044": [10, 11, 12, 14, 888, 960, 978], "200045": [10, 11, 12, 14, 888, 973, 981], "200046": [10, 11, 12, 14, 888, 967], "200047": [10, 11, 12, 14, 888, 957, 967], "200048": [10, 11, 12, 14, 888, 965, 971], "200049": [10, 11, 12, 14, 888, 954, 964], "200050": [10, 11, 12, 14, 980], "200051": [10, 11, 12, 14, 980], "200052": [10, 11, 12, 14, 888], "200053": [10, 11, 12, 14, 888, 958], "200054": [10, 11, 12, 14, 888, 974], "200055": [10, 11, 12, 14, 975], "200056": [10, 11, 12, 14, 975], "200057": [10, 11, 12, 14, 975], "200058": [10, 11, 12, 14, 975], "200059": [10, 11, 12, 14, 888, 955, 960], "200060": [10, 11, 12, 14, 888, 958], "200061": [10, 11, 12, 14, 888], "200062": [10, 11, 12, 14, 969], "200063": [10, 11, 12, 14, 969], "200064": [10, 11, 12, 14, 969], "200065": [10, 11, 12, 14, 969], "200066": [10, 11, 12, 14, 888, 955, 962], "200067": [10, 11, 12, 14, 888, 954], "200068": [10, 11, 12, 14, 888, 955], "200069": [10, 11, 12, 14, 888], "200070": [10, 11, 12, 14, 963], "200071": [10, 11, 12, 14, 963], "200072": [10, 11, 12, 14, 963], "200073": [10, 11, 12, 14, 963], "200074": [10, 11, 12, 14, 888], "200075": [10, 11, 12, 14, 888], "200076": [10, 11, 12, 14, 888], "200077": [10, 11, 12, 14, 888], "200078": [10, 11, 12, 14, 888], "200079": [10, 11, 12, 956], "200080": [10, 11, 12, 956], "200081": [10, 11, 12, 956], "200082": [10, 11, 12, 956], "200083": [10, 11, 12, 888], "200084": [10, 11, 12, 888], "200085": [10, 11, 12, 888], "200090": [10, 11, 12, 888]}}
\ No newline at end of file
diff --git a/settings.json b/settings.json
index e8ff2c4..5f7df7b 100644
--- a/settings.json
+++ b/settings.json
@@ -3,6 +3,7 @@
     "v10": false,
     "Autoplay": true,
     "Helper": true,
+    "Autohu": true,
     "Port": {
         "MITM": 7878,
         "XMLRPC": 7879,
diff --git a/unlocker.py b/unlocker.py
deleted file mode 100644
index fc4497c..0000000
--- a/unlocker.py
+++ /dev/null
@@ -1,390 +0,0 @@
-import mitmproxy.addonmanager
-import mitmproxy.http
-import mitmproxy.log
-import mitmproxy.tcp
-import mitmproxy.websocket
-from mitmproxy import ctx
-from mitmproxy import proxy, options
-from mitmproxy.tools.dump import DumpMaster
-from liqi import LiqiProto, MsgType
-from rich.console import Console
-from rich.table import Table
-from rich.text import Text
-from rich import box
-from rich.panel import Panel
-import json
-console = Console()
-
-activated_flows = [] # store all flow.id ([-1] is the recently opened)
-messages_dict = dict() # flow.id -> Queue[flow_msg]
-with open("settings.json", "r") as f:
-    settings = json.load(f)
-    unlock_avartar = settings["Unlocker"]
-
-class Unlocker:
-
-    def __init__(self):
-        self.liqi: dict[str, LiqiProto]={}
-        self.liqiModify: LiqiModify=LiqiModify(unlock_avartar=unlock_avartar)
-        pass
-
-    def websocket_start(self, flow: mitmproxy.http.HTTPFlow):
-        assert isinstance(flow.websocket, mitmproxy.websocket.WebSocketData)
-        global activated_flows,messages_dict
-        
-        activated_flows.append(flow.id)
-        messages_dict[flow.id]=[]
-        self.liqi[flow.id] = LiqiProto()
-
-    def websocket_message(self, flow: mitmproxy.http.HTTPFlow):
-        assert isinstance(flow.websocket, mitmproxy.websocket.WebSocketData)
-        global activated_flows,messages_dict
-        if flow.websocket.messages[-1].injected:
-            # This message is injected by us, ignore it
-            console.print("Injected!", style="bold red")
-            return
-        original_proto_msg = flow.websocket.messages[-1].content
-        messages_dict[flow.id].append(original_proto_msg)
-        original_liqi_msg = self.liqi[flow.id].parse(original_proto_msg)
-        action, modified_liqi_msg = self.liqiModify.modify(original_liqi_msg)
-        # Three kinds of action: drop, modify, fake_response | None
-        if action == 'drop':
-            console.log(original_liqi_msg, style="bold green")
-            console.print("Dropping Message!", style="bold red")
-            flow.websocket.messages[-1].drop()
-        elif action == 'modify':
-            console.log(original_liqi_msg, style="bold green")
-            console.print("Modifying Message!", style="bold red")
-            console.log(modified_liqi_msg, style="bold red")
-            modified_proto_msg = self.liqi[flow.id].compose(modified_liqi_msg, msg_id=original_liqi_msg['id'])
-            flow.websocket.messages[-1].content = modified_proto_msg
-        elif action == 'fake_response':
-            console.log(original_liqi_msg, style="bold green")
-            console.print("Faking Server Response!", style="bold red")
-            console.print(f"Target: {modified_liqi_msg['method']}", style="bold red")
-            fake_response = modified_liqi_msg['fake_response']
-            console.log(modified_liqi_msg, style="bold red")
-            console.log(fake_response, style="bold red")
-            modified_liqi_msg.pop('fake_response')
-            
-            flow.websocket.messages[-1].drop()
-            fake_proto_msg = self.liqi[flow.id].compose(fake_response, msg_id=fake_response['id'])
-            # Refer to mitmproxy/addons/proxyserver.py:inject_websocket
-            ctx.master.commands.call(
-                "inject.websocket", flow, True, fake_proto_msg, False
-            )
-        else:
-            console.log(original_liqi_msg)
-
-
-    def websocket_end(self, flow: mitmproxy.http.HTTPFlow):
-        global activated_flows,messages_dict
-        activated_flows.remove(flow.id)
-        messages_dict.pop(flow.id)
-
-
-async def start_proxy(host, port):
-    opts = options.Options(listen_host=host, listen_port=port)
-
-    master = DumpMaster(
-        opts,
-        with_termlog=False,
-        with_dumper=False,
-    )
-    master.addons.add(Unlocker())
-    
-    await master.run()
-    return master
-
-
-class LiqiModify:
-    def __init__(self, unlock_avartar: bool = False):
-        self.unlock_avartar = unlock_avartar
-        self.accountId = -1
-        self.current_character = -1
-        self.current_skin = -1
-        self.current_frame = None
-        self.current_title = 0
-        with open('./settings.json', 'r', encoding='utf-8') as f:
-            settings = json.load(f)
-            self.current_character = settings['main_character']
-            self.current_title = settings['title']
-            for view in settings['views']:
-                if view['slot'] == 5:
-                    self.current_frame = view['itemId']
-        with open('./id/CharacterId.json', 'r', encoding='utf-8') as f:
-            characterId = json.load(f)
-            self.current_skin = characterId[str(self.current_character)]['init_skin']
-        pass
-
-    def modify(self, msg: dict):
-        if msg is None:
-            return None, None
-        action = None
-        modify_msg = msg
-        if 'error' in msg['data']:
-            if msg['data']['error']['code'] != 0:
-                console.print("===Error==="*10, style="bold red")
-                console.log(msg, style="bold red")
-                console.print("===Error==="*10, style="bold red")
-        if self.unlock_avartar:
-            # Response from server, modify it
-            if msg['type'] == MsgType.Res:
-                if msg['method'] == '.lq.Lobby.login':
-                    action = 'modify'
-                    self.accountId = msg['data']['account']['accountId']
-                    modify_msg['data']['account']['avatarId'] = self.current_skin
-                    modify_msg['data']['account']['title'] = self.current_title
-                if msg['method'] == '.lq.Lobby.oauth2Login':
-                    action = 'modify'
-                    self.accountId = msg['data']['account']['accountId']
-                    modify_msg['data']['account']['avatarId'] = self.current_skin
-                    modify_msg['data']['account']['title'] = self.current_title
-
-                if msg['method'] == '.lq.Lobby.fetchInfo':
-                    action = 'modify'
-                    # Title
-                    modify_msg['data']['titleList']['titleList'] = []
-                    with open('./id/TitleId.json', 'r', encoding='utf-8') as f:
-                        titleId = json.load(f)
-                        for id in titleId:
-                            modify_msg['data']['titleList']['titleList'].append(titleId[id]['id'])
-                    # Items
-                    modify_msg['data']['bagInfo']['bag']['items'] = []
-                    with open('./id/ItemId.json', 'r', encoding='utf-8') as f:
-                        itemId = json.load(f)
-                        for id in itemId:
-                            if itemId[id]['is_unique'] == 1:
-                                modify_msg['data']['bagInfo']['bag']['items'].append({
-                                    "itemId": itemId[id]['id'],
-                                    "stack": 1
-                                })
-                            else:
-                                modify_msg['data']['bagInfo']['bag']['items'].append({
-                                    "itemId": itemId[id]['id'],
-                                    "stack": 999
-                                })
-                    # Characters
-                    modify_msg['data']['characterInfo']['characters'] = []
-                    modify_msg['data']['characterInfo']['characterSort'] = []
-                    with open('./id/CharacterId.json', 'r', encoding='utf-8') as f:
-                        characterId = json.load(f)
-                        for id in characterId:
-                            modify_msg['data']['characterInfo']['characters'].append({
-                                "charid": characterId[id]['id'],
-                                "level": 5,
-                                "exp": 0,
-                                "skin": characterId[id]['init_skin'],
-                                "views": [],
-                                "isUpgraded": True,
-                                "extraEmoji": [],
-                                "rewardedLevel": []
-                            })
-                            modify_msg['data']['characterInfo']['characterSort'].append(characterId[id]['id'])
-                    # Skins
-                    modify_msg['data']['characterInfo']['skins'] = []
-                    with open('./id/SkinId.json', 'r', encoding='utf-8') as f:
-                        skinId = json.load(f)
-                        for id in skinId:
-                            if id in ['400000', '400001']:
-                                continue
-                            modify_msg['data']['characterInfo']['skins'].append(skinId[id]['id'])
-                    modify_msg['data']['characterInfo']['mainCharacterId'] = self.current_character
-
-                    # Views
-                    with open('./settings.json', 'r', encoding='utf-8') as f:
-                        settings = json.load(f)
-                        modify_msg['data']['allCommonViews']['views'] = [{
-                            'values': settings['views'],
-                            "index": 0
-                        }]
-                        modify_msg['data']['allCommonViews']['use'] = 0
-                if msg['method'] == '.lq.Lobby.createRoom':
-                    action = 'modify'
-                    for person in modify_msg['data']['room']['persons']:
-                        if person['accountId'] == self.accountId:
-                            person['avatarId'] = self.current_skin
-                            person['character']['charid'] = self.current_character
-                            person['character']['skin'] = self.current_skin
-                            person['character']['level'] = 5
-                            person['character']['exp'] = 0
-                            person['character']['isUpgraded'] = True
-                            person['title'] = self.current_title
-                            if self.current_frame is not None:
-                                person['avatarFrame'] = self.current_frame
-                if msg['method'] == '.lq.Lobby.fetchAccountInfo':
-                    action = 'modify'
-                    if modify_msg['data']['account']['accountId'] == self.accountId:
-                        modify_msg['data']['account']['avatarId'] = self.current_skin
-                if msg['method'] == '.lq.FastTest.authGame':
-                    action = 'modify'
-                    for person in modify_msg['data']['players']:
-                        if person['accountId'] == self.accountId:
-                            person['avatarId'] = self.current_skin
-                            person['character']['charid'] = self.current_character
-                            person['character']['skin'] = self.current_skin
-                            person['character']['level'] = 5
-                            person['character']['exp'] = 0
-                            person['character']['isUpgraded'] = True
-                            person['title'] = self.current_title
-                            if self.current_frame is not None:
-                                person['avatarFrame'] = self.current_frame
-                            with open('./settings.json', 'r', encoding='utf-8') as f:
-                                settings = json.load(f)
-                                person['views'] = settings['views']
-
-                if msg['method'] == '.lq.Lobby.fetchRoom':
-                    # console.log(msg, style="bold red")
-                    action = 'modify'
-                    for person in modify_msg['data']['room']['persons']:
-                        if person['accountId'] == self.accountId:
-                            person['avatarId'] = self.current_skin
-                            person['character']['charid'] = self.current_character
-                            person['character']['skin'] = self.current_skin
-                            person['character']['level'] = 5
-                            person['character']['exp'] = 0
-                            person['character']['isUpgraded'] = True
-                            person['title'] = self.current_title
-                            if self.current_frame is not None:
-                                person['avatarFrame'] = self.current_frame
-
-            # Request from client, drop it or fake a response to avoid Majsoul found we cheat
-            if msg['type'] == MsgType.Req:
-                if msg['method'] == '.lq.Lobby.changeMainCharacter':
-                    action = 'drop'
-                    with open('./settings.json', 'r', encoding='utf-8') as f:
-                        settings = json.load(f)
-                        settings['main_character'] = msg['data']['characterId']
-                    with open('./settings.json', 'w', encoding='utf-8') as f:
-                        json.dump(settings, f, indent=4)
-                    self.current_character = msg['data']['characterId']
-                    with open('./id/CharacterId.json', 'r', encoding='utf-8') as f:
-                        characterId = json.load(f)
-                        self.current_skin = characterId[str(self.current_character)]['init_skin ']
-                if msg['method'] == '.lq.Lobby.changeCharacterSkin':
-                    action = 'drop'
-                    with open('./id/CharacterId.json', 'r', encoding='utf-8') as f:
-                        characterId = json.load(f)
-                        characterId[str(msg['data']['characterId'])]['init_skin'] = msg['data']['skin']
-                    with open('./id/CharacterId.json', 'w', encoding='utf-8') as f:
-                        json.dump(characterId, f, indent=4)
-                    self.current_skin = msg['data']['skin']
-                if msg['method'] == '.lq.Lobby.saveCommonViews':
-                    action = 'fake_response'
-                    modify_msg['fake_response'] = {
-                        'id': msg['id'],
-                        'type': MsgType.Res,
-                        'method': '.lq.Lobby.saveCommonViews',
-                        'data': {}
-                    }
-                    set_frame = False
-                    for view in msg['data']['views']:
-                        if view['slot'] == 5:
-                            self.current_frame = view['itemId']
-                            set_frame = True
-                    if not set_frame:
-                        self.current_frame = None
-                    with open('./settings.json', 'r', encoding='utf-8') as f:
-                        settings = json.load(f)
-                        settings['views'] = modify_msg['data']['views']
-                    with open('./settings.json', 'w', encoding='utf-8') as f:
-                        json.dump(settings, f, indent=4)
-                if msg['method'] == '.lq.Lobby.openAllRewardItem':
-                    action = 'drop'
-                if msg['method'] == '.lq.Lobby.useTitle':
-                    action = 'fake_response'
-                    with open('./settings.json', 'r', encoding='utf-8') as f:
-                        settings = json.load(f)
-                        settings['title'] = msg['data']['title']
-                        self.current_title = msg['data']['title']
-                    with open('./settings.json', 'w', encoding='utf-8') as f:
-                        json.dump(settings, f, indent=4)
-                    modify_msg['fake_response'] = {
-                        'id': msg['id'],
-                        'type': MsgType.Res,
-                        'method': '.lq.Lobby.useTitle',
-                        'data': {}
-                    }
-
-            if msg['type'] == MsgType.Notify:
-                if msg['method'] == '.lq.NotifyRoomPlayerUpdate':
-                    action = 'modify'
-                    for person in modify_msg['data']['updateList']:
-                        if person['accountId'] == self.accountId:
-                            person['avatarId'] = self.current_skin
-                            person['title'] = self.current_title
-                            if self.current_frame is not None:
-                                person['avatarFrame'] = self.current_frame
-                    for person in modify_msg['data']['playerList']:
-                        if person['accountId'] == self.accountId:
-                            person['avatarId'] = self.current_skin
-                            person['title'] = self.current_title
-                            if self.current_frame is not None:
-                                person['avatarFrame'] = self.current_frame
-
-        return action, modify_msg
-
-
-sponsor_message = Table.grid(padding=1)
-sponsor_message.add_column(style="green", justify="right")
-sponsor_message.add_column(no_wrap=True)
-
-sponsor_message.add_row(
-    "MajsoulUnlocker",
-    "[u blue link=https://github.com/shinkuan/MajsoulUnlocker]https://github.com/shinkuan/MajsoulUnlocker",
-)
-sponsor_message.add_row(
-    "RiichiCityUnlocker",
-    "[u blue link=https://github.com/shinkuan/RiichiCityUnlocker]https://github.com/shinkuan/RiichiCityUnlocker",
-)
-
-intro_message = Text.from_markup(
-        """\
-I hope you enjoy using the MajsoulUnlocker!
-
-MajsoulUnlocker is made by [link=https://github.com/shinkuan]shinkuan[/]
-
-[link=https://github.com/shinkuan/MajsoulUnlocker/issues]Open an issue[/] if you have any questions."""
-    )
-
-message = Table.grid(padding=2)
-message.add_column()
-message.add_column(no_wrap=True)
-message.add_row(intro_message, sponsor_message)
-
-console.print(
-    Panel.fit(
-        message,
-        box=box.ROUNDED,
-        padding=(1, 2),
-        title="[b cyan]Thanks for using MajsoulUnlocker!",
-        border_style="bright_blue",
-    ),
-    justify="center",
-)
-
-
-
-
-
-# addons = [
-#     Unlocker()
-# ]
-
-
-if __name__ == '__main__':
-    import asyncio
-    import argparse
-    parser = argparse.ArgumentParser()
-    parser.add_argument('--host', type=str, default='127.0.0.1')
-    parser.add_argument('--port', type=int, default=7878)
-    parser.add_argument('--passthrough', action='store_true')
-    args = parser.parse_args()
-    if args.passthrough:
-        unlock_avartar = False
-    else:
-        unlock_avartar = True
-    port = args.port
-    host = args.host
-    # start proxy
-    asyncio.run(start_proxy(host, port))
diff --git a/unlocker_v10.py b/unlocker_v10.py
deleted file mode 100644
index 15ba857..0000000
--- a/unlocker_v10.py
+++ /dev/null
@@ -1,424 +0,0 @@
-import mitmproxy.addonmanager
-import mitmproxy.http
-import mitmproxy.log
-import mitmproxy.tcp
-import mitmproxy.websocket
-from mitmproxy import ctx
-from mitmproxy import proxy, options
-from mitmproxy.tools.dump import DumpMaster
-from liqi import LiqiProto, MsgType
-from rich.console import Console
-from rich.table import Table
-from rich.text import Text
-from rich import box
-from rich.panel import Panel
-import json
-console = Console()
-
-activated_flows = [] # store all flow.id ([-1] is the recently opened)
-messages_dict = dict() # flow.id -> Queue[flow_msg]
-with open("settings.json", "r") as f:
-    settings = json.load(f)
-    unlock_avartar = settings["Unlocker"]
-
-class Unlocker:
-
-    def __init__(self):
-        self.liqi: dict[str, LiqiProto]={}
-        self.liqiModify: LiqiModify=LiqiModify(unlock_avartar=unlock_avartar)
-        pass
-
-    def websocket_start(self, flow: mitmproxy.http.HTTPFlow):
-        assert isinstance(flow.websocket, mitmproxy.websocket.WebSocketData)
-        global activated_flows,messages_dict
-        
-        activated_flows.append(flow.id)
-        messages_dict[flow.id]=[]
-        self.liqi[flow.id] = LiqiProto()
-
-    def websocket_message(self, flow: mitmproxy.http.HTTPFlow):
-        assert isinstance(flow.websocket, mitmproxy.websocket.WebSocketData)
-        global activated_flows,messages_dict
-        if flow.websocket.messages[-1].injected:
-            # This message is injected by us, ignore it
-            console.print("Injected!", style="bold red")
-            return
-        original_proto_msg = flow.websocket.messages[-1].content
-        messages_dict[flow.id].append(original_proto_msg)
-        original_liqi_msg = self.liqi[flow.id].parse(original_proto_msg)
-        action, modified_liqi_msg = self.liqiModify.modify(original_liqi_msg)
-        # Three kinds of action: drop, modify, fake_response | None
-        if action == 'drop':
-            console.log(original_liqi_msg, style="bold green")
-            console.print("Dropping Message!", style="bold red")
-            flow.websocket.messages[-1].drop()
-        elif action == 'modify':
-            console.log(original_liqi_msg, style="bold green")
-            console.print("Modifying Message!", style="bold red")
-            console.log(modified_liqi_msg, style="bold red")
-            modified_proto_msg = self.liqi[flow.id].compose(modified_liqi_msg, msg_id=original_liqi_msg['id'])
-            flow.websocket.messages[-1].content = modified_proto_msg
-        elif action == 'fake_response':
-            console.log(original_liqi_msg, style="bold green")
-            console.print("Faking Server Response!", style="bold red")
-            console.print(f"Target: {modified_liqi_msg['method']}", style="bold red")
-            fake_response = modified_liqi_msg['fake_response']
-            console.log(modified_liqi_msg, style="bold red")
-            console.log(fake_response, style="bold red")
-            modified_liqi_msg.pop('fake_response')
-            
-            flow.websocket.messages[-1].drop()
-            fake_proto_msg = self.liqi[flow.id].compose(fake_response, msg_id=fake_response['id'])
-            # Refer to mitmproxy/addons/proxyserver.py:inject_websocket
-            ctx.master.commands.call(
-                "inject.websocket", flow, True, fake_proto_msg, False
-            )
-        else:
-            console.log(original_liqi_msg)
-
-
-    def websocket_end(self, flow: mitmproxy.http.HTTPFlow):
-        global activated_flows,messages_dict
-        activated_flows.remove(flow.id)
-        messages_dict.pop(flow.id)
-
-
-async def start_proxy(host, port):
-    opts = options.Options(listen_host=host, listen_port=port)
-
-    master = DumpMaster(
-        opts,
-        with_termlog=False,
-        with_dumper=False,
-    )
-    master.addons.add(Unlocker())
-    
-    await master.run()
-    return master
-
-
-class LiqiModify:
-    def __init__(self, unlock_avartar: bool = False):
-        self.unlock_avartar = unlock_avartar
-        self.accountId = -1
-        self.current_character = -1
-        self.current_skin = -1
-        with open('./settings.json', 'r', encoding='utf-8') as f:
-            settings = json.load(f)
-            self.current_character = settings['main_character']
-            self.current_title = settings['title']
-            for view in settings['views']:
-                if view['slot'] == 5:
-                    self.current_frame = view['itemId']
-        with open('./id/CharacterId.json', 'r', encoding='utf-8') as f:
-            characterId = json.load(f)
-            self.current_skin = characterId[str(self.current_character)]['init_skin']
-        pass
-
-    def modify(self, msg: dict):
-        if msg is None:
-            return None, None
-        action = None
-        modify_msg = msg
-        if 'error' in msg['data']:
-            if msg['data']['error']['code'] != 0:
-                console.print("===Error==="*10, style="bold red")
-                console.log(msg, style="bold red")
-                console.print("===Error==="*10, style="bold red")
-        if self.unlock_avartar:
-            # Response from server, modify it
-            if msg['type'] == MsgType.Res:
-                if msg['method'] == '.lq.Lobby.login':
-                    action = 'modify'
-                    self.accountId = msg['data']['account']['accountId']
-                    modify_msg['data']['account']['avatarId'] = self.current_skin
-                    modify_msg['data']['account']['title'] = self.current_title
-                    # modify_msg['data']['account']['level'] = {'id': 10720, 'score': 0}
-                    # modify_msg['data']['account']['level3'] = {'id': 20720, 'score': 0}
-                if msg['method'] == '.lq.Lobby.oauth2Login':
-                    action = 'modify'
-                    self.accountId = msg['data']['account']['accountId']
-                    modify_msg['data']['account']['avatarId'] = self.current_skin
-                    modify_msg['data']['account']['title'] = self.current_title
-                    # modify_msg['data']['account']['level'] = {'id': 10720, 'score': 0}
-                    # modify_msg['data']['account']['level3'] = {'id': 20720, 'score': 0}
-
-                # =======================================================================
-
-                if msg['method'] == '.lq.Lobby.fetchTitleList':
-                    action = 'modify'
-                    modify_msg['data']['titleList'] = []
-                    with open('./id/TitleId.json', 'r', encoding='utf-8') as f:
-                        titleId = json.load(f)
-                        for id in titleId:
-                            modify_msg['data']['titleList'].append(titleId[id]['id'])
-
-                if msg['method'] == '.lq.Lobby.fetchBagInfo':
-                    action = 'modify'
-                    modify_msg['data']['bag']['items'] = []
-                    with open('./id/ItemId.json', 'r', encoding='utf-8') as f:
-                        itemId = json.load(f)
-                        for id in itemId:
-                            if itemId[id]['is_unique'] == 1:
-                                modify_msg['data']['bag']['items'].append({
-                                    "itemId": itemId[id]['id'],
-                                    "stack": 1
-                                })
-                            else:
-                                modify_msg['data']['bag']['items'].append({
-                                    "itemId": itemId[id]['id'],
-                                    "stack": 999
-                                })
-
-                if msg['method'] == '.lq.Lobby.fetchCharacterInfo':
-                    action = 'modify'
-                    modify_msg['data']['characters'] = []
-                    modify_msg['data']['characterSort'] = []
-                    with open('./id/CharacterId.json', 'r', encoding='utf-8') as f:
-                        characterId = json.load(f)
-                        for id in characterId:
-                            modify_msg['data']['characters'].append({
-                                "charid": characterId[id]['id'],
-                                "level": 5,
-                                "exp": 0,
-                                "skin": characterId[id]['init_skin'],
-                                "views": [],
-                                "isUpgraded": True,
-                                "extraEmoji": [],
-                                "rewardedLevel": []
-                            })
-                            modify_msg['data']['characterSort'].append(characterId[id]['id'])
-                    # Skins
-                    modify_msg['data']['skins'] = []
-                    with open('./id/SkinId.json', 'r', encoding='utf-8') as f:
-                        skinId = json.load(f)
-                        for id in skinId:
-                            if id in ['400000', '400001']:
-                                continue
-                            modify_msg['data']['skins'].append(skinId[id]['id'])
-                    modify_msg['data']['mainCharacterId'] = self.current_character
-
-                if msg['method'] == '.lq.Lobby.fetchAllCommonViews':
-                    action = 'modify'
-                    with open('./settings.json', 'r', encoding='utf-8') as f:
-                        settings = json.load(f)
-                        modify_msg['data']['views'] = [{
-                            'values': settings['views'],
-                            "index": 0
-                        }]
-                        modify_msg['data']['use'] = 0
-
-                if msg['method'] == '.lq.Lobby.createRoom':
-                    action = 'modify'
-                    for person in modify_msg['data']['room']['persons']:
-                        if person['accountId'] == self.accountId:
-                            person['avatarId'] = self.current_skin
-                            person['character']['charid'] = self.current_character
-                            person['character']['skin'] = self.current_skin
-                            person['character']['level'] = 5
-                            person['character']['exp'] = 0
-                            person['character']['isUpgraded'] = True
-                            person['title'] = self.current_title
-                            person['avatarFrame'] = self.current_frame
-                if msg['method'] == '.lq.Lobby.fetchAccountInfo':
-                    action = 'modify'
-                    if modify_msg['data']['account']['accountId'] == self.accountId:
-                        modify_msg['data']['account']['avatarId'] = self.current_skin
-                if msg['method'] == '.lq.FastTest.authGame':
-                    action = 'modify'
-                    for person in modify_msg['data']['players']:
-                        if person['accountId'] == self.accountId:
-                            person['avatarId'] = self.current_skin
-                            person['character']['charid'] = self.current_character
-                            person['character']['skin'] = self.current_skin
-                            person['character']['level'] = 5
-                            person['character']['exp'] = 0
-                            person['character']['isUpgraded'] = True
-                            person['title'] = self.current_title
-                            person['avatarFrame'] = self.current_frame
-                if msg['method'] == '.lq.Lobby.fetchRoom':
-                    action = 'modify'
-                    for person in modify_msg['data']['room']['persons']:
-                        if person['accountId'] == self.accountId:
-                            person['avatarId'] = self.current_skin
-                            person['character']['charid'] = self.current_character
-                            person['character']['skin'] = self.current_skin
-                            person['character']['level'] = 5
-                            person['character']['exp'] = 0
-                            person['character']['isUpgraded'] = True
-
-            # Request from client, drop it or fake a response to avoid Majsoul found we cheat
-            if msg['type'] == MsgType.Req:
-                if msg['method'] == '.lq.Lobby.changeMainCharacter':
-                    action = 'fake_response'
-                    with open('./settings.json', 'r', encoding='utf-8') as f:
-                        settings = json.load(f)
-                        settings['main_character'] = msg['data']['characterId']
-                    with open('./settings.json', 'w', encoding='utf-8') as f:
-                        json.dump(settings, f, indent=4)
-                    self.current_character = msg['data']['characterId']
-                    with open('./id/CharacterId.json', 'r', encoding='utf-8') as f:
-                        characterId = json.load(f)
-                        self.current_skin = characterId[str(self.current_character)]['init_skin']
-                    modify_msg['fake_response'] = {
-                        'id': msg['id'],
-                        'type': MsgType.Res,
-                        'method': '.lq.Lobby.changeMainCharacter',
-                        'data': {}
-                    }
-                if msg['method'] == '.lq.Lobby.changeCharacterSkin':
-                    action = 'fake_response'
-                    with open('./id/CharacterId.json', 'r', encoding='utf-8') as f:
-                        characterId = json.load(f)
-                        characterId[str(msg['data']['characterId'])]['init_skin'] = msg['data']['skin']
-                    with open('./id/CharacterId.json', 'w', encoding='utf-8') as f:
-                        json.dump(characterId, f, indent=4)
-                    self.current_skin = msg['data']['skin']
-                    modify_msg['fake_response'] = {
-                        'id': msg['id'],
-                        'type': MsgType.Res,
-                        'method': '.lq.Lobby.changeCharacterSkin',
-                        'data': {}
-                    }
-
-                    # modify_msg['fake_response'] = {
-                    #     'id': -1,
-                    #     'type': MsgType.Notify,
-                    #     'method': '.lq.NotifyAccountUpdate',
-                    #     'data': {
-                    #         'update': {
-                    #             'character': {
-                    #                 'characters': [
-                    #                     {
-                    #                         'charid': msg['data']['characterId'],
-                    #                         'skin': msg['data']['skin'],
-                    #                         'level': 5,
-                    #                         'exp': 0,
-                    #                         'views': [],
-                    #                         'isUpgraded': True,
-                    #                         'extraEmoji': [],
-                    #                         'rewardedLevel': []
-                    #                     }
-                    #                 ],
-                    #                 'skins': [],
-                    #                 'finishedEndings': [],
-                    #                 'rewardedEndings': []
-                    #             },
-                    #             'numerical': [],
-                    #             'newRechargedList': []
-                    #         }
-                    #     }
-                    # }
-                if msg['method'] == '.lq.Lobby.saveCommonViews':
-                    action = 'fake_response'
-                    modify_msg['fake_response'] = {
-                        'id': msg['id'],
-                        'type': MsgType.Res,
-                        'method': '.lq.Lobby.saveCommonViews',
-                        'data': {}
-                    }
-                    for view in msg['data']['views']:
-                        if view['slot'] == 5:
-                            self.current_frame = view['itemId']
-                    with open('./settings.json', 'r', encoding='utf-8') as f:
-                        settings = json.load(f)
-                        settings['views'] = modify_msg['data']['views']
-                    with open('./settings.json', 'w', encoding='utf-8') as f:
-                        json.dump(settings, f, indent=4)
-                if msg['method'] == '.lq.Lobby.openAllRewardItem':
-                    action = 'drop'
-                if msg['method'] == '.lq.Lobby.useTitle':
-                    action = 'fake_response'
-                    with open('./settings.json', 'r', encoding='utf-8') as f:
-                        settings = json.load(f)
-                        settings['title'] = msg['data']['title']
-                        self.current_title = msg['data']['title']
-                    with open('./settings.json', 'w', encoding='utf-8') as f:
-                        json.dump(settings, f, indent=4)
-                    modify_msg['fake_response'] = {
-                        'id': msg['id'],
-                        'type': MsgType.Res,
-                        'method': '.lq.Lobby.useTitle',
-                        'data': {}
-                    }
-
-            if msg['type'] == MsgType.Notify:
-                if msg['method'] == '.lq.NotifyRoomPlayerUpdate':
-                    action = 'modify'
-                    for person in modify_msg['data']['updateList']:
-                        if person['accountId'] == self.accountId:
-                            person['avatarId'] = self.current_skin
-                            person['title'] = self.current_title
-                            person['avatarFrame'] = self.current_frame
-                    for person in modify_msg['data']['playerList']:
-                        if person['accountId'] == self.accountId:
-                            person['avatarId'] = self.current_skin
-                            person['title'] = self.current_title
-                            person['avatarFrame'] = self.current_frame
-
-        return action, modify_msg
-
-
-sponsor_message = Table.grid(padding=1)
-sponsor_message.add_column(style="green", justify="right")
-sponsor_message.add_column(no_wrap=True)
-
-sponsor_message.add_row(
-    "MajsoulUnlocker",
-    "[u blue link=https://github.com/shinkuan/MajsoulUnlocker]https://github.com/shinkuan/MajsoulUnlocker",
-)
-sponsor_message.add_row(
-    "RiichiCityUnlocker",
-    "[u blue link=https://github.com/shinkuan/RiichiCityUnlocker]https://github.com/shinkuan/RiichiCityUnlocker",
-)
-
-intro_message = Text.from_markup(
-        """\
-I hope you enjoy using the MajsoulUnlocker!
-
-MajsoulUnlocker is made by [link=https://github.com/shinkuan]shinkuan[/]
-
-[link=https://github.com/shinkuan/MajsoulUnlocker/issues]Open an issue[/] if you have any questions."""
-    )
-
-message = Table.grid(padding=2)
-message.add_column()
-message.add_column(no_wrap=True)
-message.add_row(intro_message, sponsor_message)
-
-console.print(
-    Panel.fit(
-        message,
-        box=box.ROUNDED,
-        padding=(1, 2),
-        title="[b cyan]Thanks for using MajsoulUnlocker!",
-        border_style="bright_blue",
-    ),
-    justify="center",
-)
-
-
-
-
-
-addons = [
-    Unlocker()
-]
-
-
-if __name__ == '__main__':
-    import asyncio
-    import argparse
-    parser = argparse.ArgumentParser()
-    parser.add_argument('--host', type=str, default='127.0.0.1')
-    parser.add_argument('--port', type=int, default=7878)
-    parser.add_argument('--passthrough', action='store_true')
-    args = parser.parse_args()
-    if args.passthrough:
-        unlock_avartar = False
-    else:
-        unlock_avartar = True
-    port = args.port
-    host = args.host
-    # start proxy
-    asyncio.run(start_proxy(host, port))