remove_repair_quotes.py

"""Remove repair quotes from existing claims (pre-adjudication cleanup)."""
import json
import glob
import os

claim_files = glob.glob("data/synthetic/CLM-*.json")
removed_count = 0

for claim_file in claim_files:
    with open(claim_file, 'r') as f:
        claim_data = json.load(f)
    
    original_count = len(claim_data.get("documents", []))
    
    # Remove repair quotes
    claim_data["documents"] = [
        doc for doc in claim_data["documents"] 
        if "Repair" not in doc.get("document_type", "") and "REPAIR" not in doc.get("document_type", "").upper()
    ]
    
    new_count = len(claim_data.get("documents", []))
    removed = original_count - new_count
    
    if removed > 0:
        removed_count += removed
        # Save updated claim
        with open(claim_file, 'w') as f:
            json.dump(claim_data, f, indent=2, default=str)
        print(f"Removed {removed} repair quote(s) from {os.path.basename(claim_file)}")

print(f"\nTotal repair quotes removed: {removed_count}")
print("Existing claims cleaned up for pre-adjudication stage.")