Update update_docx_with_pdf.py

update_docx_with_pdf.py

@@ -25,9 +25,11 @@ def update_json_with_pdf(word_json_file, pdf_txt_file, output_file):
 
     # --- Build prompt ---
     user_prompt = f"""Here is a JSON template. It contains only the fields that need updating:
+
 {word_json}
 
 Here is the extracted text from a PDF:
+
 {pdf_txt}
 
 Instructions:
@@ -35,9 +37,14 @@ Instructions:
 - DO NOT add any extra fields, and do not change the JSON structure.
 - Update ALL nested sections properly (like "Operator Declaration" with its "Print Name" and "Position Title")
 - Make sure to update both the main sections AND the flattened keys (like "Operator Declaration.Print Name")
+- For Operator Declaration specifically:
+  * Print Name should be the actual person's name (e.g., "Jeff Nitschke")
+  * Position Title should be their job role (e.g., "Director", "Manager", "Owner")
+- Pay special attention to signatures and declarations - extract the person's name and their position/title
+- Look for patterns like "Name - Position" or "Name, Position" in signature areas
 - Output ONLY the updated JSON, as raw JSON (no markdown, no extra text, no greetings).
 - Make sure the JSON is valid and ready to use.
-- Pay special attention to updating operator names, auditor names, and all personal details consistently throughout all sections."""
+- Update operator names, auditor names, and all personal details consistently throughout all sections."""
 
     # --- Call OpenAI API ---
     api_key = os.environ.get("OPENAI_API_KEY")