Removed max history length from openai calls

openai_compatible_inference_bot.py

@@ -6,7 +6,6 @@ from base_telegram_inference_bot import BaseTelegramInferenceBot
 from openai import OpenAI, AzureOpenAI # Import both
 
 class OpenAICompatibleInferenceBot(BaseTelegramInferenceBot):
-    DEFAULT_MAX_HISTORY_LENGTH = 20
     DEFAULT_MAX_TOKENS = 1000 # Default for _configure_model_and_tokens
 
     def __init__(
@@ -25,7 +24,6 @@ class OpenAICompatibleInferenceBot(BaseTelegramInferenceBot):
     ):
         super().__init__(system_prompt_content=system_prompt_content, system_prompt_path=system_prompt_path)
 
-        self.max_history_length = max_history_length if max_history_length is not None else self.DEFAULT_MAX_HISTORY_LENGTH
         self.client = client
 
         if not self.client:
@@ -194,16 +192,7 @@ class OpenAICompatibleInferenceBot(BaseTelegramInferenceBot):
                 # Ensure final content is returned even if max iterations hit with pending tool calls
                 break 
 
-        self.conversation_history[user_id] = messages # Persist the full exchange for this turn
-        # Apply history length limit
-        if len(self.conversation_history[user_id]) > self.max_history_length:
-            # Keep system prompt if present as the first message, then trim the rest
-            if self.conversation_history[user_id][0]["role"] == "system":
-                system_msg = [self.conversation_history[user_id][0]]
-                trimmed_history = self.conversation_history[user_id][-(self.max_history_length-1):]
-                self.conversation_history[user_id] = system_msg + trimmed_history
-            else:
-                self.conversation_history[user_id] = self.conversation_history[user_id][-self.max_history_length:]
+        self.conversation_history[user_id] = messages
 
         final_assistant_message = messages[-1]
         return final_assistant_message.content if final_assistant_message.role == "assistant" and final_assistant_message.content is not None else "Assistant did not provide a textual response."