Update file_utils.py

file_utils.py

@@ -5,9 +5,16 @@ import os
 
 # === GPT-4 Client Setup ===
 client = OpenAI(api_key=os.getenv("OPENAI_API_KEY"))
+MAX_TOKENS = 7000  # Safety buffer below GPT-4's 8192 limit
 
 # === Centralized LLM Caller ===
 def call_llm(prompt, system_message="You are a helpful assistant.", temperature=0.7):
+    # Rough token approximation: 1 token ≈ 4 characters (English average)
+    approx_tokens = len(prompt) // 4
+    if approx_tokens > MAX_TOKENS:
+        prompt = prompt[:MAX_TOKENS * 4]  # truncate by character length
+        prompt += "\n[NOTE: Truncated due to length limit.]"
+
     response = client.chat.completions.create(
         model="gpt-4",
         messages=[
@@ -102,4 +109,4 @@ Content:
 Question:
 {question}
 """
-    return call_llm(prompt, system_message="You are a helpful and accurate tutor that stays grounded in provided content.")
+    return call_llm(prompt, system_message="You are a helpful and accurate tutor that stays grounded in provided content.")