cyclop/openai_compatible_inference_bot.py

import importlib
import json
import os
import logging
import inspect
import re
from abc import abstractmethod
from openai import OpenAI
from tools.base_tool import BaseTool
from telegram_helper import TelegramHelper
import argparse
from inference_bot import InferenceBot
import tiktoken # Added this import

class OpenAICompatibleInferenceBot(InferenceBot):
    def __init__(
        self,
        api_key: str | None = None,
        base_url: str | None = None,
        small_model_name: str | None = None,
        small_model_max_tokens: str | None = None,
        large_model_name: str | None = None,
        large_model_max_tokens: str | None = None,
        allowed_function_tags: list[str] | None = None,
        system_prompt_path: str | None = None,
        use_large_model: bool = False
    ):
        self.model_config = {
            "small_model_name": small_model_name,
            "small_model_max_tokens": small_model_max_tokens,
            "large_model_name": large_model_name,
            "large_model_max_tokens": large_model_max_tokens
        }
        self.allowed_function_tags = allowed_function_tags if allowed_function_tags else None
        self.conversation_history = {}
        self._processing_status = {}
        self.system_prompt_path = system_prompt_path
        self.system_prompt = self.load_system_prompt(
            file_path=system_prompt_path
        )
        self.tools, self.functions = self.load_functions()
        self.client = OpenAI(api_key=api_key, base_url=base_url)
        log_msg = f"Initialized OpenAI compatible client. Target URL: {base_url if base_url else 'OpenAI default'}."
        logging.info(log_msg)

        # Load inference token limits (defaults: small=16k, large=32k)
        self.small_model_max_inference_tokens = int(os.getenv("_SMALL_MODEL_MAX_INFERENCE_TOKENS", "16384"))
        self.large_model_max_inference_tokens = int(os.getenv("_LARGE_MODEL_MAX_INFERENCE_TOKENS", "32768"))

        # Configure the actual model name and max_tokens for API calls
        if use_large_model:
            self._configure_model_and_tokens(
                self.model_config["large_model_name"], 
                self.model_config["large_model_max_tokens"]
            )
        else:
            self._configure_model_and_tokens(
                self.model_config["small_model_name"], 
                self.model_config["small_model_max_tokens"]
            )
    
    @property
    def processing_status(self):
        return self._processing_status
    
    def clear_conversation_history(self, user_id):
        if user_id in self.conversation_history:
            del self.conversation_history[user_id]
        
        for tool in self.tools:
            tool.clear()

    def _configure_model_and_tokens(self, model_name: str | None, max_tokens_str: str | None):
        self.model = model_name
        try:
            if max_tokens_str and max_tokens_str.lower() not in ["none", "", "null"]:
                self.max_tokens = int(max_tokens_str)
            else:
                self.max_tokens = None
        except ValueError:
            logging.warning(f"Invalid value for max_tokens: {max_tokens_str}. Using API default (None)")
            self.max_tokens = None
        
        logging.info(f"Configured to use model: {self.model} with max_tokens: {self.max_tokens if self.max_tokens is not None else 'API default'}")

    def get_llm_description(self) -> str:
        client_type = type(self.client).__name__
        return f"Client: {client_type}, LLM: {self.model}, Max Tokens: {self.max_tokens if self.max_tokens is not None else 'API default'}"

    def _encoding_for_model(self, model: str | None):
        try:
            return tiktoken.encoding_for_model(model) if model else tiktoken.get_encoding("cl100k_base")
        except KeyError:
            logging.warning(f"Warning: model {model} not found. Using cl100k_base encoding.")
            return tiktoken.get_encoding("cl100k_base")

    def _normalize_messages(self, messages):
        """Return a list of plain dict chat messages acceptable by the API.
        - Converts OpenAI SDK message objects into dicts
        - Preserves tool_calls structure where present
        """
        normalized = []
        for m in messages:
            if isinstance(m, dict):
                # Ensure only known keys are present; copy shallowly
                entry = {k: v for k, v in m.items() if k in {"role", "content", "name", "tool_call_id", "tool_calls"}}
                normalized.append(entry)
            else:
                # Likely an OpenAI message object
                role = getattr(m, "role", None)
                content = getattr(m, "content", None)
                name = getattr(m, "name", None)
                tool_calls = []
                tc_list = getattr(m, "tool_calls", None)
                if tc_list:
                    for tc in tc_list:
                        try:
                            tool_calls.append({
                                "id": getattr(tc, "id", None),
                                "type": getattr(tc, "type", "function"),
                                "function": {
                                    "name": getattr(getattr(tc, "function", None), "name", None),
                                    "arguments": getattr(getattr(tc, "function", None), "arguments", "{}"),
                                }
                            })
                        except Exception:
                            # Best-effort fallback
                            tool_calls.append({"id": None, "type": "function", "function": {"name": "unknown", "arguments": "{}"}})
                entry = {"role": role, "content": content}
                if name:
                    entry["name"] = name
                if tool_calls:
                    entry["tool_calls"] = tool_calls
                normalized.append(entry)
        return normalized

    def _estimate_tokens(self, messages):
        """Estimate tokens for messages with tiktoken, including tool_calls arguments.
        Based on OpenAI's chat token counting rules approximation.
        """
        enc = self._encoding_for_model(self.model)
        num_tokens = 0
        for m in messages:
            num_tokens += 4  # per-message overhead
            if not isinstance(m, dict):
                continue
            # role/content
            for key in ("role", "name", "content"):
                v = m.get(key)
                if isinstance(v, str):
                    num_tokens += len(enc.encode(v))
            # tool calls request portion (arguments)
            tcs = m.get("tool_calls")
            if tcs and isinstance(tcs, list):
                # approximate cost of the tool_calls JSON the model sees
                for tc in tcs:
                    fn = tc.get("function", {}) if isinstance(tc, dict) else {}
                    fname = fn.get("name")
                    fargs = fn.get("arguments")
                    if isinstance(fname, str):
                        num_tokens += len(enc.encode(fname))
                    if isinstance(fargs, str):
                        num_tokens += len(enc.encode(fargs))
        num_tokens += 2  # assistant priming
        return num_tokens

    def _get_inference_limit(self):
        current_model_is_small = self.model == self.model_config["small_model_name"]
        current_model_is_large = self.model == self.model_config["large_model_name"]
        if current_model_is_small:
            return self.small_model_max_inference_tokens
        if current_model_is_large:
            return self.large_model_max_inference_tokens
        logging.warning(f"Could not determine inference token limit for model: {self.model}. Proceeding without check.")
        return None

    def _summarize_tool_args(self, args_str: str, max_chars: int = 512) -> str:
        """Summarize tool-call request arguments without altering tool responses.
        - If JSON, keep keys and short previews of string values.
        - If plain string, truncate with an indicator.
        """
        try:
            parsed = json.loads(args_str)
            if isinstance(parsed, dict):
                summary = {}
                for k, v in parsed.items():
                    if isinstance(v, str):
                        if len(v) > 160:
                            summary[k] = v[:120] + f"... [len={len(v)}]"
                        else:
                            summary[k] = v
                    elif isinstance(v, (list, dict)):
                        # structural summary only
                        summary[k] = f"<{type(v).__name__} size={len(v)}>"
                    else:
                        summary[k] = v
                s = json.dumps(summary, ensure_ascii=False)
                if len(s) > max_chars:
                    s = s[: max_chars - 20] + "... [summarized]"
                return s
        except Exception:
            pass
        # Fallback: truncate raw string
        return (args_str[: max_chars - 20] + "... [summarized]") if len(args_str) > max_chars else args_str

    def _summarize_tool_call_requests_in_messages(self, messages):
        changed = False
        for m in messages:
            if isinstance(m, dict) and m.get("tool_calls"):
                new_tool_calls = []
                for tc in m["tool_calls"]:
                    if not isinstance(tc, dict):
                        new_tool_calls.append(tc)
                        continue
                    fn = tc.get("function", {})
                    args = fn.get("arguments")
                    if isinstance(args, str) and args and len(args) > 700:
                        # summarize long request arguments only
                        fn = dict(fn)
                        fn["arguments"] = self._summarize_tool_args(args)
                        tc = dict(tc)
                        tc["function"] = fn
                        changed = True
                    new_tool_calls.append(tc)
                if changed:
                    m["tool_calls"] = new_tool_calls
        return changed

    def _elide_redundant_code_blocks(self, messages):
        """As a last resort, remove large code blocks from older assistant messages.
        Keep the latest assistant message intact.
        """
        changed = False
        # Identify indices of assistant messages
        assistant_indices = [i for i, m in enumerate(messages) if isinstance(m, dict) and m.get("role") == "assistant" and m.get("content")]
        if len(assistant_indices) <= 1:
            return changed
        # Protect the last assistant message
        for i in assistant_indices[:-1]:
            m = messages[i]
            content = m.get("content")
            if not isinstance(content, str):
                continue
            if "```" in content or "\n    " in content:
                # Replace code blocks fenced by ``` with succinct markers
                orig = content
                content = re.sub(r"```[\s\S]*?```", "[code block omitted]", content)
                # Also collapse long indented blocks
                content = re.sub(r"(?:\n\s{4,}.+)+", "\n[long block omitted]", content)
                if content != orig:
                    m["content"] = content
                    changed = True
        return changed

    def _enforce_budget(self, messages):
        """Normalize and enforce token budget by summarizing only tool-call requests first,
        then eliding redundant code blocks if still too large. Returns normalized messages.
        """
        normalized = self._normalize_messages(messages)
        limit = self._get_inference_limit()
        if not limit:
            return normalized
        # Reserve space for completion tokens
        reserve = self.max_tokens if isinstance(self.max_tokens, int) else 1024
        budget = max(1024, limit - reserve)

        tokens = self._estimate_tokens(normalized)
        if tokens <= budget:
            return normalized

        # Step 1: summarize only tool-call request arguments
        if self._summarize_tool_call_requests_in_messages(normalized):
            tokens = self._estimate_tokens(normalized)
            logging.info(f"Applied tool-call request summarization. tokens={tokens}/{budget}")
            if tokens <= budget:
                return normalized

        # Step 2: elide redundant code blocks from older assistant messages
        if self._elide_redundant_code_blocks(normalized):
            tokens = self._estimate_tokens(normalized)
            logging.info(f"Elided redundant code blocks. tokens={tokens}/{budget}")
            if tokens <= budget:
                return normalized

        # If still over, log and proceed; the API may still reject; caller may choose to abort
        logging.warning(f"Projected tokens still exceed budget after optimizations: {tokens}/{budget}")
        return normalized

    def get_chat_response(self, messages):
        if not self.client:
            logging.error("OpenAI client not initialized before get_chat_response.")
            raise ValueError("OpenAI client not initialized.")
        try:
            cleaned_tools = None
            if hasattr(self, 'functions') and self.functions:
                cleaned_tools = []
                for func in self.functions:
                    include_function = False
                    
                    if not hasattr(self, 'allowed_function_tags') or self.allowed_function_tags is None:
                        include_function = True
                    else:
                        tags = func.get("_tags", [])
                        if any(tag in self.allowed_function_tags for tag in tags):
                            include_function = True
                    
                    if include_function:
                        func_copy = {k: v for k, v in func.items() if k != "_tags"}
                        cleaned_tools.append(func_copy)

            # Enforce token budget prior to API call
            messages_for_api = self._enforce_budget(messages)

            response = self.client.chat.completions.create(
                model=self.model, 
                messages=messages_for_api,
                tools=cleaned_tools,
                tool_choice="auto" if cleaned_tools else None,
                max_tokens=self.max_tokens,
                
            )
            return response
        except Exception as e:
            logging.error(f"API call to model {self.model} failed: {e}")
            raise

    def get_bot_status(self):
        """
        Returns a message with the currently enabled model and the system prompt path being used.
        """
        model_name = self.model if hasattr(self, 'model') else None
        prompt_path = self.system_prompt_path or os.getenv("SYSTEM_PROMPT_PATH") or "(default prompt in use)"
        return f"Current model: {model_name}\nSystem prompt path: {prompt_path}"

    async def handle_message(self, user_id, user_message):
        if user_id not in self.conversation_history or not self.conversation_history[user_id]:
            self.conversation_history[user_id] = []
            if self.system_prompt:
                 self.conversation_history[user_id].append({"role": "system", "content": self.system_prompt})

        self.conversation_history[user_id].append({"role": "user", "content": user_message})
        messages = list(self.conversation_history[user_id])
        
        # Pre-inference token limit check with budgeted optimizations
        limit = self._get_inference_limit()
        if limit is not None:
            # Estimate on normalized messages after applying request-only summarization if needed
            provisional = self._enforce_budget(messages)
            token_count = self._estimate_tokens(provisional)
            reserve = self.max_tokens if isinstance(self.max_tokens, int) else 1024
            budget = max(1024, limit - reserve)
            if token_count > budget:
                logging.warning(f"Request for user {user_id} exceeds inference token budget even after optimizations ({token_count}/{budget}).")
                # Do not persist this message in history as it was not processed by LLM
                if self.conversation_history[user_id] and self.conversation_history[user_id][-1]["role"] == "user" and self.conversation_history[user_id][-1]["content"] == user_message:
                    self.conversation_history[user_id].pop()
                return "Request exceeds inference token limit after optimization. Please shorten your request, use /clear, or implement RAG in your application."

        response = self.get_chat_response(messages)
        
        if not (response.choices and response.choices[0].message):
            logging.error("No valid response choice message from LLM.")
            self.conversation_history[user_id] = messages 
            return "Error: Could not get a valid response from the LLM."
        
        assistant_message = response.choices[0].message
        messages.append(assistant_message) 

        tool_calls_from_response = list(assistant_message.tool_calls) if assistant_message.tool_calls else []
        
        tool_use_count = 0
        MAX_TOOL_ITERATIONS = 200

        while tool_calls_from_response and tool_use_count < MAX_TOOL_ITERATIONS:
            tool_results_for_model = [] 

            for tool_call in tool_calls_from_response:
                tool_call_id = tool_call.id
                function_to_call = tool_call.function
                function_name = function_to_call.name
                function_args_str = function_to_call.arguments
                
                logging.info(f"Attempting to call tool: {function_name} with args: [request summarized if large]")
                if function_name not in [f["function"]["name"] for f in self.functions]:
                    logging.warning(f"Tool function {function_name} not found in available functions.")
                    tool_results_for_model.append({
                        "role": "tool",
                        "tool_call_id": tool_call_id,
                        "name": function_name,
                        "content": f"Error: Tool function {function_name} not found."
                    })
                    continue

                try:
                    tool_response_content = self.call_tool(function_name, function_args_str)
                    if not isinstance(tool_response_content, str):
                        tool_response_content = json.dumps(tool_response_content) 
                except Exception as e:
                    logging.error(f"Error calling tool {function_name}: {e}")
                    tool_response_content = f"Error executing tool {function_name}: {str(e)}"
                
                tool_results_for_model.append({
                    "role": "tool",
                    "tool_call_id": tool_call_id,
                    "name": function_name,
                    "content": tool_response_content
                })

            messages.extend(tool_results_for_model)

            # Enforce budget before next LLM call (summarize request portion only; preserve tool responses)
            response = self.get_chat_response(messages)
            if not (response.choices and response.choices[0].message):
                logging.error("No valid response choice message from LLM after tool call.")
                self.conversation_history[user_id] = messages
                return "Error: Could not get a valid response from the LLM after tool call."

            assistant_message = response.choices[0].message
            messages.append(assistant_message)

            tool_calls_from_response = list(assistant_message.tool_calls) if assistant_message.tool_calls else []
            
            tool_use_count += 1
            if tool_use_count >= MAX_TOOL_ITERATIONS and tool_calls_from_response:
                logging.warning(f"Max tool iterations ({MAX_TOOL_ITERATIONS}) reached. Returning last assistant message.")
                break 

        self.conversation_history[user_id] = messages

        final_assistant_message = messages[-1]
        return final_assistant_message.content if getattr(final_assistant_message, "role", None) == "assistant" and getattr(final_assistant_message, "content", None) is not None else (final_assistant_message.get("content") if isinstance(final_assistant_message, dict) else "Assistant did not provide a textual response.")

    async def start(self):
        logging.info(f"{self.__class__.__name__} (Model: {self.model}) started.")

    async def abort_processing(self, user_id):
        if user_id in self.processing_status:
            self.clear_processing_status(user_id)
            logging.info(f"Processing aborted for user {user_id}.")
            return "Processing aborted. You can send a new message or /clear the conversation."
        else:
            return "No active processing found to abort. If you wish, /clear the conversation history."
    
    def load_functions(self):
        tools = []
        functions = []
        tools_dir = os.path.join(os.path.dirname(__file__), 'tools')
        if not os.path.exists(tools_dir):
            logging.warning(f"Tools directory not found: {tools_dir}")
            return [], []

        for filename in os.listdir(tools_dir):
            if filename.endswith('.py') and filename != '__init__.py' and filename != 'base_tool.py':
                module_name = f'tools.{filename[:-3]}'
                try:
                    module = importlib.import_module(module_name)
                    for name, obj in inspect.getmembers(module):
                        if inspect.isclass(obj) and issubclass(obj, BaseTool) and obj != BaseTool:
                            try:
                                tools.append(obj()) # This instantiation might be an issue for tools needing config
                            except Exception as e:
                                logging.error(f"Error instantiating tool {name} from {filename}: {e}")
                except Exception as e:
                    logging.error(f"Error importing module {module_name}: {e}")
        
        for tool in tools:
            functions.extend(tool.get_functions())
        return tools, functions

    def load_system_prompt(self, direct_content: str | None = None, file_path: str | None = None) -> str:
        default_prompt = "You are a helpful AI assistant."

        if direct_content:
            logging.info("Using direct content for system prompt.")
            return direct_content.strip()

        prompt_path_to_try = file_path or os.getenv("SYSTEM_PROMPT_PATH")

        if prompt_path_to_try:
            if os.path.isfile(prompt_path_to_try):
                try:
                    with open(prompt_path_to_try, "r", encoding="utf-8") as file:
                        content = file.read().strip()
                        logging.info(f"Successfully loaded system prompt from {prompt_path_to_try}.")
                        return content
                except IOError as e:
                    logging.warning(f"Could not read system prompt file {prompt_path_to_try}: {e}. Using default.")
                    return default_prompt
            else:
                logging.warning(f"System prompt file {prompt_path_to_try} not found. Using default system prompt.")
                return default_prompt
        else:
            logging.info("No system prompt path provided (argument or ENV) or direct content. Using default system prompt.")
            return default_prompt

    def set_processing_status(self, user_id: int, message_id: int):
        self.processing_status[user_id] = {"processing": True, "message_id": message_id}

    def clear_processing_status(self, user_id: int):
        if user_id in self.processing_status:
            del self.processing_status[user_id]

    def call_tool(self, function_call_name, function_call_arguments):
        function_name = function_call_name
        function_args = None
        if isinstance(function_call_arguments, dict):
            function_args = function_call_arguments
        elif isinstance(function_call_arguments, str):
            try:
                function_args = json.loads(function_call_arguments)
            except json.JSONDecodeError as e:
                logging.error(f"Error decoding function call arguments (string) for {function_call_name}: {e}. Arguments: {function_call_arguments}")
                return f"Error: Malformed arguments for tool call: {e}"
        else: 
            if function_call_arguments is None:
                function_args = {} 
            else:
                logging.error(f"Unexpected type for function_call_arguments for {function_name}: {type(function_call_arguments)}. Arguments: {function_call_arguments}")
                return f"Error: Invalid argument type for tool call: {type(function_call_arguments)}"

        for tool in self.tools:
            for function in tool.get_functions():
                if function["function"]["name"] == function_name:
                    try:
                        if not isinstance(function_args, dict):
                             logging.error(f"Internal error: function_args not a dict for {function_name} before execution. Args: {function_args}")
                             return f"Internal error preparing arguments for tool {function_name}."
                        return tool.execute(function_name, **function_args)
                    except Exception as e:
                        logging.error(f"Error executing tool {function_name} with args {function_args}: {e}")
                        return f"Error executing tool {function_name}: {e}"
        logging.warning(f"Tool function {function_name} not found.")
        return f"Error: Tool function {function_name} not found."
    
    async def switch_model(self):
        if not self.model_config["small_model_name"] or not self.model_config["large_model_name"]:
            logging.warning("Small or Large model names are not defined. Cannot switch model.")
            return f"Model switching not fully configured. Currently using {self.model}."

        current_is_small = self.model == self.model_config["small_model_name"]
        current_is_large = self.model == self.model_config["large_model_name"]

        if current_is_large:
            target_model = self.model_config["small_model_name"]
            target_max_tokens_str = self.model_config["small_model_max_tokens"]
        elif current_is_small:
            target_model = self.model_config["large_model_name"]
            target_max_tokens_str = self.model_config["large_model_max_tokens"]
        else:
            logging.warning(f"Current model {self.model} is unrecognized. Switching to default small model: {self.model_config['small_model_name']}.")
            target_model = self.model_config["small_model_name"]
            target_max_tokens_str = self.model_config["small_model_max_tokens"]

        self._configure_model_and_tokens(target_model, target_max_tokens_str)
        return f"Switched model to {self.model}. Max tokens set to {self.max_tokens if self.max_tokens is not None else 'API default'}."

def main():
    logging.basicConfig(level=logging.INFO, format='%(asctime)s - %(name)s - %(levelname)s - %(message)s')
    
    bot = None

    try:
        parser = argparse.ArgumentParser(description='OpenAI Compatible Inference Bot')
        parser.add_argument('--config', type=str, help='Configuration Prepend (i.e. gemini, openai, etc)', default="Telegram")
        parser.add_argument('--messenger', type=str, help='Messenger type (i.e. telegram)', required=True)
        parser.add_argument('--persona', type=str, help='Path to system prompt file', required=False)
        parser.add_argument('--tools', nargs='+', help='List of allowed function tags', required=False)
        parser.add_argument('--use-large-model', action='store_true', help='Use the large model instead of the small model')
        # Add these to launch.json arguments if you want to limit the toolset available: "--tools", "read", "communicate"
        # Parse command line arguments
        args = parser.parse_args()
        if args.persona:
            logging.info(f"Using custom persona from: {args.persona}")


        system_prompt_path=args.persona if args.persona else None
        allowed_function_tags=args.tools if args.tools else None
        config_prepend = args.config if args.config else None
        messenger = args.messenger if args.messenger else None
        use_large_model = args.use_large_model

        # Initialize model and max tokens based on the config prepend
        if config_prepend:
            api_key = os.environ.get(f"{config_prepend.upper()}_API_KEY")
            baseurl = os.environ.get(f"{config_prepend.upper()}_API_BASE_URL", "")
            small_model_name = os.environ.get(f"{config_prepend.upper()}_SMALL_MODEL")
            large_model_name = os.environ.get(f"{config_prepend.upper()}_LARGE_MODEL")
            small_model_max_tokens = os.environ.get(f"{config_prepend.upper()}_SMALL_MODEL_MAX_TOKENS")
            large_model_max_tokens = os.environ.get(f"{config_prepend.upper()}_LARGE_MODEL_MAX_TOKENS")

            bot = OpenAICompatibleInferenceBot(
                api_key=api_key,
                base_url=baseurl,
                small_model_name=small_model_name,
                small_model_max_tokens=small_model_max_tokens,
                large_model_name=large_model_name,
                large_model_max_tokens=large_model_max_tokens,
                system_prompt_path=system_prompt_path,
                allowed_function_tags=allowed_function_tags,
                use_large_model=use_large_model
            )
            full_code_file = importlib.import_module(f'{messenger.lower()}_helper')
            messenger_helper_class_name = f"{messenger.capitalize()}Helper"
            if not hasattr(full_code_file, messenger_helper_class_name):
                messenger_helper_class_name = f"{messenger.upper()}Helper"
                if not hasattr(full_code_file, messenger_helper_class_name):
                    raise ValueError(f"Messenger helper class {messenger_helper_class_name} not found in {full_code_file.__name__}.")
            helper_class = getattr(full_code_file, messenger_helper_class_name)
            
            helper = helper_class(bot)
            helper.run()
    except ValueError as e:
        logging.error(f"FATAL: {e}")
        return
    except Exception as e: # Catch any other init errors
        logging.error(f"An unexpected error occurred during bot initialization: {e}")
        return

if __name__ == '__main__':
    main()