MISC: GroqCloud support & Commands refactor (#104)

* add GroqCloud support

* refactor commands

* refactor textgen_webui commands

* move removing prompt command prefix to controller

* add support for command decorators

* refactor OpenAI commands

* move command logic from providers

* unite 2 approaches of making system message

* minor fixes

* update logging for commands

* move error handling logic to controller

* update README

* add groq model settings

* fix config

Author: dborodin836

README.md

@@ -387,7 +387,7 @@ Hi chatGPT, you are going to pretend to be MEDIC from Team Fortress 2. You can d
 ## That's all?
 
 If you want to know more here are listed some things that were left unexplained, and some tips and tricks:
-[unexplained_explained.md](docs/unexplained_explained.md)
+[unexplained_explained.md](docs/unexplained_explained.md) or at project [Wiki](https://github.com/dborodin836/TF2-GPTChatBot/wiki).
 
 ## Screenshots
 

config.ini

@@ -1,7 +1,19 @@
 [GENERAL]
-TF2_LOGFILE_PATH=H:\Programs\Steam\steamapps\common\Team Fortress 2\tf\console.log
+TF2_LOGFILE_PATH=C:\Program Files (x86)\Steam\steamapps\common\Team Fortress 2\tf\console.log
 OPENAI_API_KEY=
 
+[GROQ]
+GROQ_ENABLE=False
+GROQ_API_KEY=
+; Available models
+; https://console.groq.com/docs/models
+GROQ_MODEL=
+GROQ_COMMAND=!g
+GROQ_CHAT_COMMAND=!gc
+GROQ_PRIVATE_CHAT=!gpc
+; Example {"max_tokens": 2}
+GROQ_SETTINGS=
+
 [COMMANDS]
 ENABLE_OPENAI_COMMANDS=True
 GPT_COMMAND=!gpt3

config.py

@@ -85,6 +85,14 @@ class Config(BaseModel):
 
     CUSTOM_MODEL_SETTINGS: Optional[str | dict]
 
+    GROQ_API_KEY: str
+    GROQ_COMMAND: str
+    GROQ_CHAT_COMMAND: str
+    GROQ_PRIVATE_CHAT: str
+    GROQ_MODEL: str
+    GROQ_ENABLE: bool
+    GROQ_SETTINGS: Optional[str | dict]
+
     @validator("OPENAI_API_KEY")
     def api_key_pattern_match(cls, v):
         if not re.fullmatch(OPENAI_API_KEY_RE_PATTERN, v):
@@ -158,6 +166,7 @@ def init_config():
             for key, value in configparser_config.items(section)
         }
         global config
+
         try:
             if config_dict.get("CUSTOM_MODEL_SETTINGS") != "":
                 config_dict["CUSTOM_MODEL_SETTINGS"] = json.loads(
@@ -168,9 +177,19 @@ def init_config():
                 f"CUSTOM_MODEL_SETTINGS is not dict [{e}].", "BOTH", level="ERROR"
             )
 
+        try:
+            if config_dict.get("GROQ_SETTINGS") != "":
+                config_dict["GROQ_SETTINGS"] = json.loads(
+                    config_dict.get("GROQ_SETTINGS")
+                )
+        except Exception as e:
+            buffered_fail_message(
+                f"GROQ_SETTINGS is not dict [{e}].", "BOTH", level="ERROR"
+            )
+
         config = Config(**config_dict)
 
-        if not config.ENABLE_OPENAI_COMMANDS and not config.ENABLE_CUSTOM_MODEL:
+        if not config.ENABLE_OPENAI_COMMANDS and not config.ENABLE_CUSTOM_MODEL and not config.GROQ_ENABLE:
             buffered_message("You haven't enabled any AI related commands.")
 
     except (pydantic.ValidationError, Exception) as e:

modules/api/base.py

@@ -0,0 +1,11 @@
+from abc import ABC, abstractmethod
+
+from modules.typing import MessageHistory
+
+
+class LLMProvider(ABC):
+
+    @staticmethod
+    @abstractmethod
+    def get_completion_text(message_array: MessageHistory, username: str, model: str) -> str:
+        ...

modules/api/groq.py

@@ -0,0 +1,35 @@
+import hashlib
+
+import groq
+
+from config import config
+from modules.api.base import LLMProvider
+from modules.utils.text import remove_hashtags
+
+
+class GroqCloudLLMProvider(LLMProvider):
+
+    @staticmethod
+    def get_completion_text(message_array, username, model):
+        client = groq.Groq(
+            max_retries=0,
+            api_key=config.GROQ_API_KEY
+        )
+
+        if isinstance(config.GROQ_SETTINGS, dict):
+            completion = client.chat.completions.create(
+                model=model,
+                messages=message_array,
+                user=hashlib.md5(username.encode()).hexdigest(),
+                **config.GROQ_SETTINGS
+            )
+        else:
+            completion = client.chat.completions.create(
+                model=model,
+                messages=message_array,
+                user=hashlib.md5(username.encode()).hexdigest(),
+            )
+
+        response_text = completion.choices[0].message.content.strip()
+        filtered_response = remove_hashtags(response_text)
+        return filtered_response

modules/api/openai.py

@@ -1,20 +1,32 @@
 import hashlib
-import time
 
 import openai
 
 from config import config
-from modules.conversation_history import ConversationHistory
-from modules.logs import get_logger, log_gui_general_message, log_gui_model_message
-from modules.servers.tf2 import send_say_command_to_tf2
-from modules.typing import Message, MessageHistory
-from modules.utils.text import get_system_message, remove_args, remove_hashtags
+from modules.api.base import LLMProvider
+from modules.logs import get_logger
 
 main_logger = get_logger("main")
 gui_logger = get_logger("gui")
 
 
-def is_violated_tos(message: str) -> bool:
+class OpenAILLMProvider(LLMProvider):
+
+    @staticmethod
+    def get_completion_text(conversation_history, username, model):
+        openai.api_key = config.OPENAI_API_KEY
+
+        completion = openai.ChatCompletion.create(
+            model=model,
+            messages=conversation_history,
+            user=hashlib.md5(username.encode()).hexdigest(),
+        )
+
+        response_text = completion.choices[0].message["content"].strip()
+        return response_text
+
+
+def is_flagged(message: str) -> bool:
     openai.api_key = config.OPENAI_API_KEY
     try:
         response = openai.Moderation.create(
@@ -28,123 +40,3 @@ def is_violated_tos(message: str) -> bool:
         return True
 
     return response.results[0]["flagged"]
-
-
-def send_gpt_completion_request(
-    conversation_history: MessageHistory, username: str, model: str
-) -> str:
-    openai.api_key = config.OPENAI_API_KEY
-
-    completion = openai.ChatCompletion.create(
-        model=model,
-        messages=conversation_history,
-        user=hashlib.md5(username.encode()).hexdigest(),
-    )
-
-    response_text = completion.choices[0].message["content"].strip()
-    return response_text
-
-
-def handle_cgpt_request(
-    username: str,
-    user_prompt: str,
-    conversation_history: ConversationHistory,
-    model,
-    is_team: bool = False,
-) -> ConversationHistory:
-    """
-    This function is called when the user wants to send a message to the AI chatbot. It logs the
-    user's message, and sends a request to generate a response.
-    """
-    log_gui_model_message(model, username, user_prompt)
-
-    user_message = remove_args(user_prompt)
-    if (
-        not config.TOS_VIOLATION
-        and is_violated_tos(user_message)
-        and config.HOST_USERNAME != username
-    ):
-        gui_logger.error(f"Request '{user_prompt}' violates OPENAI TOS. Skipping...")
-        return conversation_history
-
-    conversation_history.add_user_message_from_prompt(user_prompt)
-
-    response = get_response(conversation_history.get_messages_array(), username, model)
-
-    if response:
-        conversation_history.add_assistant_message(Message(role="assistant", content=response))
-        log_gui_model_message(model, username, " ".join(response.split()))
-        send_say_command_to_tf2(response, username, is_team)
-
-    return conversation_history
-
-
-def handle_gpt_request(
-    username: str, user_prompt: str, model: str, is_team_chat: bool = False
-) -> None:
-    """
-    This function is called when the user wants to send a message to the AI chatbot. It logs the
-    user's message, and sends a request to GPT-3 to generate a response. Finally, the function
-    sends the generated response to the TF2 game.
-    """
-    log_gui_model_message(model, username, user_prompt)
-
-    user_message = remove_args(user_prompt)
-    sys_message = get_system_message(user_prompt)
-
-    if (
-        not config.TOS_VIOLATION
-        and is_violated_tos(user_message)
-        and config.HOST_USERNAME != username
-    ):
-        gui_logger.warning(
-            f"Request '{user_prompt}' by user {username} violates OPENAI TOS. Skipping..."
-        )
-        return
-
-    payload = [
-        sys_message,
-        Message(role="assistant", content=config.GREETING),
-        Message(role="user", content=user_message),
-    ]
-
-    response = get_response(payload, username, model)
-
-    if response:
-        main_logger.info(
-            f"Got response for user {username}. Response: {' '.join(response.split())}"
-        )
-        log_gui_model_message(model, username, " ".join(response.split()))
-        send_say_command_to_tf2(response, username, is_team_chat)
-
-
-def get_response(conversation_history: MessageHistory, username: str, model) -> str | None:
-    attempts = 0
-    max_attempts = 2
-
-    while attempts < max_attempts:
-        try:
-            response = send_gpt_completion_request(conversation_history, username, model=model)
-            filtered_response = remove_hashtags(response)
-            return filtered_response
-        except openai.error.RateLimitError:
-            log_gui_general_message("Rate limited! Trying again...")
-            main_logger(f"User is rate limited.")
-            time.sleep(2)
-            attempts += 1
-        except openai.error.APIError as e:
-            log_gui_general_message(f"Wasn't able to connect to OpenAI API. Cancelling...")
-            main_logger.error(f"APIError happened. [{e}]")
-            return
-        except openai.error.AuthenticationError:
-            log_gui_general_message("Your OpenAI api key is invalid.")
-            main_logger.error("OpenAI API key is invalid.")
-            return
-        except Exception as e:
-            log_gui_general_message(f"Unhandled error happened! Cancelling ({e})")
-            main_logger.error(f"Unhandled error happened! Cancelling ({e})")
-            return
-
-    if attempts == max_attempts:
-        log_gui_general_message("Max number of attempts reached! Try again later!")
-        main_logger(f"Max number of attempts reached. [{max_attempts}/{max_attempts}]")

modules/api/textgen_webui.py

@@ -1,39 +1,25 @@
 import requests
 
 from config import config
+from modules.api.base import LLMProvider
 from modules.logs import get_logger
-from modules.typing import Message
 
 main_logger = get_logger("main")
 combo_logger = get_logger("combo")
 
 
-def get_custom_model_response(conversation_history: list[Message]) -> str | None:
-    uri = f"http://{config.CUSTOM_MODEL_HOST}/v1/chat/completions"
+class TextGenerationWebUILLMProvider(LLMProvider):
 
-    headers = {"Content-Type": "application/json"}
+    @staticmethod
+    def get_completion_text(conversation_history, username, model):
+        uri = f"http://{config.CUSTOM_MODEL_HOST}/v1/chat/completions"
+        headers = {"Content-Type": "application/json"}
 
-    data = {"mode": "chat", "messages": conversation_history}
+        data = {"mode": "chat", "messages": conversation_history}
+        data.update(config.CUSTOM_MODEL_SETTINGS)
 
-    data.update(config.CUSTOM_MODEL_SETTINGS)
-
-    try:
         response = requests.post(uri, headers=headers, json=data, verify=False)
-    except Exception as e:
-        combo_logger.error(f"Failed to get response from the text-generation-webui server. [{e}]")
-        return
-
-    if response.status_code == 200:
-        try:
-            data = response.json()["choices"][0]["message"]["content"]
-            return data
-        except Exception as e:
-            combo_logger.error(f"Failed to parse data from server [{e}].")
-    elif response.status_code == 500:
-        combo_logger.error(f"There's error on the text-generation-webui server. [HTTP 500]")
-    else:
-        main_logger.error(
-            f"Got non-200 status code from the text-generation-webui server. [HTTP {response.status_code}]"
-        )
-
-    return None
+        if response.status_code == 500:
+            raise Exception('HTTP 500')
+        data = response.json()["choices"][0]["message"]["content"]
+        return data

modules/chat.py

@@ -1,13 +1,16 @@
-from config import config
+from config import config, RTDModes
 from modules.api.github import check_for_updates
 from modules.bans import bans_manager
 from modules.bot_state import state_manager
 from modules.command_controllers import CommandController, InitializerConfig
 from modules.commands.clear_chat import handle_clear
 from modules.commands.github import handle_gh_command
-from modules.commands.openai import handle_user_chat, handle_gpt3, handle_gpt4, handle_gpt4l, handle_global_chat
+from modules.commands.groq import GroqQuickQueryCommand, GroqGlobalChatCommand, GroqPrivateChatCommand
+from modules.commands.openai import OpenAIGlobalChatCommand, OpenAIPrivateChatCommand, OpenAIGPT3QuickQueryCommand, \
+    OpenAIGPT4QuickQueryCommand, OpenAIGPT4LQuickQueryCommand
 from modules.commands.rtd import handle_rtd
-from modules.commands.textgen_webui import handle_custom_user_chat, handle_custom_model, handle_custom_global_chat
+from modules.commands.textgen_webui import TextgenWebUIGlobalChatCommand, TextgenWebUIPrivateChatCommand, \
+    TextgenWebUIQuickQueryCommand
 from modules.logs import get_logger
 from modules.message_queueing import messaging_queue_service
 from modules.servers.tf2 import check_connection, set_host_username
@@ -53,18 +56,23 @@ def parse_console_logs_and_build_conversation_history() -> None:
 
     # Commands
     controller.register_command("!gh", handle_gh_command)
-    controller.register_command(config.RTD_COMMAND, handle_rtd)
     controller.register_command(config.CLEAR_CHAT_COMMAND, handle_clear)
+    if config.RTD_MODE != RTDModes.DISABLED:
+        controller.register_command(config.RTD_COMMAND, handle_rtd)
     if config.ENABLE_OPENAI_COMMANDS:
-        controller.register_command(config.GPT4_COMMAND, handle_gpt4)
-        controller.register_command(config.GPT4_LEGACY_COMMAND, handle_gpt4l)
-        controller.register_command(config.CHATGPT_COMMAND, handle_user_chat)
-        controller.register_command(config.GLOBAL_CHAT_COMMAND, handle_global_chat)
-        controller.register_command(config.GPT_COMMAND, handle_gpt3)
+        controller.register_command(config.GPT4_COMMAND, OpenAIGPT4QuickQueryCommand.as_command())
+        controller.register_command(config.GPT4_LEGACY_COMMAND, OpenAIGPT4LQuickQueryCommand.as_command())
+        controller.register_command(config.CHATGPT_COMMAND, OpenAIPrivateChatCommand.as_command())
+        controller.register_command(config.GLOBAL_CHAT_COMMAND, OpenAIGlobalChatCommand.as_command())
+        controller.register_command(config.GPT_COMMAND, OpenAIGPT3QuickQueryCommand.as_command())
     if config.ENABLE_CUSTOM_MODEL:
-        controller.register_command(config.CUSTOM_MODEL_COMMAND, handle_custom_model)
-        controller.register_command(config.CUSTOM_MODEL_CHAT_COMMAND, handle_custom_user_chat)
-        controller.register_command(config.GLOBAL_CUSTOM_CHAT_COMMAND, handle_custom_global_chat)
+        controller.register_command(config.CUSTOM_MODEL_COMMAND, TextgenWebUIQuickQueryCommand.as_command())
+        controller.register_command(config.CUSTOM_MODEL_CHAT_COMMAND, TextgenWebUIPrivateChatCommand.as_command())
+        controller.register_command(config.GLOBAL_CUSTOM_CHAT_COMMAND, TextgenWebUIGlobalChatCommand.as_command())
+    if config.GROQ_ENABLE:
+        controller.register_command(config.GROQ_COMMAND, GroqQuickQueryCommand.as_command())
+        controller.register_command(config.GROQ_CHAT_COMMAND, GroqGlobalChatCommand.as_command())
+        controller.register_command(config.GROQ_PRIVATE_CHAT, GroqPrivateChatCommand.as_command())
 
     # Services
     controller.register_service(messaging_queue_service)