import subprocess
import json
import time

notebook_id = "825ad1eb-2ca9-4576-8dd3-aadf468e5578"
nlm_bin = "/opt/ai-os/products/ceo/integrations/notebooklm-mcp-cli/.venv/bin/nlm"

res = subprocess.run([nlm_bin, "source", "list", notebook_id, "--json"], capture_output=True, text=True)
sources = json.loads(res.stdout)

to_delete = []

for s in sources:
    t = s.get("title", "")
    sid = s.get("id", "")
    # 1. tmpxxxx.pdf files
    if t.startswith("tmp") and t.endswith(".pdf"):
        to_delete.append((sid, t))
    # 2. Krugman international economics duplicate files
    elif "Krugman, Obstfeld, and Melitz, 2015, International Economics_ Theory and Policy.pdf" in t:
        to_delete.append((sid, t))

print(f"Total sources identified for deletion: {len(to_delete)}")

# Delete in batches of 10
batch_size = 10
deleted_count = 0
for i in range(0, len(to_delete), batch_size):
    batch = to_delete[i:i + batch_size]
    ids = [item[0] for item in batch]
    print(f"Deleting batch {i//batch_size + 1}: {len(ids)} items...")
    cmd = [nlm_bin, "source", "delete"] + ids + ["--confirm"]
    del_res = subprocess.run(cmd, capture_output=True, text=True)
    if del_res.returncode == 0:
        deleted_count += len(ids)
        print(f"  ✓ Deleted successfully: {del_res.stdout.strip()}")
    else:
        print(f"  ✗ Batch delete failed: {del_res.stderr.strip()}")
    time.sleep(0.5)

print(f"\nCompleted! Deleted {deleted_count} sources.")

# Verify remaining count
res_after = subprocess.run([nlm_bin, "source", "list", notebook_id, "--json"], capture_output=True, text=True)
sources_after = json.loads(res_after.stdout)
print(f"Initial sources: {len(sources)} -> Remaining sources: {len(sources_after)}")