Update utils.py

utils.py

@@ -43,7 +43,7 @@ def call_llm_with_retry(groq_client, **payload):
     """
     Wraps groq_client.chat.completions.create(**payload) in a retry loop
     to catch 429 rate-limit errors. If we see “try again in XXs,” we parse
-    that wait time, sleep, then retry. We also do a short sleep (0.3s) 
+    that wait time, sleep, then retry. We also do a short sleep (0.3s)
     after each successful call to spread usage.
     """
     max_retries = 3
@@ -127,8 +127,8 @@ def generate_script(
     sponsor_provided=None
 ):
     """
-    If you do a single call to generate the entire script. 
-    Uses DEEPSEEK_R1. Just ensure you parse the JSON. 
+    If you do a single call to generate the entire script.
+    Uses DEEPSEEK_R1. Just ensure you parse the JSON.
     """
     print("[LOG] Generating script with tone:", tone, "and length:", target_length)
 
@@ -590,7 +590,7 @@ include key data points and context:
         merged_input = "\n\n".join(truncated_summaries)
 
         final_prompt = f"""
-IMPORTANT: Do NOT include chain-of-thought or hidden planning. 
+IMPORTANT: Do NOT include any chain-of-thought, internal planning, or hidden reasoning in the final output.
 Produce a long, academic-style research report with the following structure:
 - Title Page (concise descriptive title)
 - Table of Contents
@@ -612,7 +612,7 @@ Partial Summaries:
 {merged_input}
 References (URLs):
 {references_text}
-Now, merge these partial summaries into one thoroughly expanded research paper:
+Now, merge these partial summaries into one thoroughly expanded research report:
 """
         final_data = {
             "model": MODEL_COMBINATION,
@@ -623,6 +623,12 @@ Now, merge these partial summaries into one thoroughly expanded research paper:
         final_response = call_llm_with_retry(groq_client, **final_data)
         final_text = final_response.choices[0].message.content.strip()
 
+        # --- NEW POST-PROCESSING STEP ---
+        # Remove any lingering chain-of-thought content (anything between <think> and </think> tags)
+        final_text = re.sub(r"<think>.*?</think>", "", final_text, flags=re.DOTALL)
+        final_text = final_text.strip()
+        # ------------------------------
+
         # Step 6: PDF generation
         print("[LOG] Step 6: Generating final PDF from the merged text.")
         final_report = generate_report(final_text)