FEAT+FIX: Greeting and minor fixes

* FEAT: Greeting feature (like in text-generation webui)
* FIX: !ai and !chat prefixes removed from the user's message
**Backstory of the PR:**
One day... I was just sitting and scrolling through the text generation webui's API, looking for the settings that could improve the chatbot. My eyes catched the "greeting" string, and I tried to use it. I added the setting into CUSTOM_MODEL_SETTINGS and reloaded the program, after typing "!ai Hello World!" in the chat I discovered that this setting does not work, or at least not the way I think it would. Filled with sadge I decided to code a simple solution myself.

Author: teufortressIndustries

config.ini

@@ -71,7 +71,7 @@ HARD_COMPLETION_LIMIT=300
 DISABLE_KEYBOARD_BINDINGS=False
 
 TOS_VIOLATION = 0
-; Adds a custom prompt after the SOFT_COMPLETION_LIMIT message.
+; Adds a custom prompt after the SOFT_COMPLETION_LIMIT message. This affects !gpt3, !gpt4, !gpt4l, !cgpt commands
 ; So it will look something like:
 ; <prompt goes here> Answer in less than 128 chars! <custom prompt goes here>
 ; Type: string
@@ -99,6 +99,12 @@ ENABLE_SOFT_LIMIT_FOR_CUSTOM_MODEL=True
 ENABLE_CUSTOM_MODEL=False
 CUSTOM_MODEL_COMMAND=!ai
 CUSTOM_MODEL_CHAT_COMMAND=!chat
+; Adds the first message on behalf of AI,
+; does the same thing as the "greeting" in text generation webui.
+; Should help with the way you want the AI to talk.
+; Example:
+; GREETING=Hello fellow human. *Bzzzt!* *Steam pump sounds* Don't mind it, it's my circuits working on burning money, so I can work a bit longer. So, tell me, do you want to buy hats? These precious... magnificent... works of art? I'm sure you can't withstand their beauty! I give you hats, you give me money, I live longer and give you more hats. What do you think?
+GREETING=
 
 ; 127.0.0.1:5000 or your-uri-here.trycloudflare.com
 CUSTOM_MODEL_HOST=127.0.0.1:5000

config.py

@@ -74,6 +74,7 @@ class Config(BaseModel):
     CUSTOM_MODEL_HOST: str
     CUSTOM_MODEL_COMMAND: str
     CUSTOM_MODEL_CHAT_COMMAND: str
+    GREETING: str
 
     CONFIRMABLE_QUEUE: bool
 

modules/commands/textgen_webui.py

@@ -11,7 +11,7 @@
 def handle_custom_model(logline: LogLine, shared_dict: dict):
     main_logger.info(
         f"'{config.CUSTOM_MODEL_COMMAND}' command from user '{logline.username}'. "
-        f"Message: '{logline.prompt.removeprefix(config.GPT_COMMAND).strip()}'"
+        f"Message: '{logline.prompt.removeprefix(config.CUSTOM_MODEL_COMMAND).strip()}'"
     )
     log_gui_model_message(
         "CUSTOM",
@@ -23,8 +23,11 @@ def handle_custom_model(logline: LogLine, shared_dict: dict):
         logline.prompt, enable_soft_limit=config.ENABLE_SOFT_LIMIT_FOR_CUSTOM_MODEL
     )
 
+    message = message.replace("!ai", "")
+
     response = get_custom_model_response(
         [
+            {"role": "assistant", "content": config.GREETING},
             {"role": "user", "content": message},
         ]
     )
@@ -44,6 +47,8 @@ def handle_custom_chat(logline: LogLine, shared_dict: dict):
     )
 
     message = add_prompts_by_flags(logline.prompt)
+    message = message.replace("!chat", "")
+    conversation_history.append({"role": "assistant", "content": config.GREETING})
     conversation_history.append({"role": "user", "content": message})
     response = get_custom_model_response(conversation_history)