sanity_checks.py 603 B

123456789101112131415161718192021222324
  1. import pandas as pd
  2. import tqdm
  3. from read_di_unfccc import UNFCCCApiReader
  4. def main():
  5. reader = UNFCCCApiReader()
  6. assert len(reader.parties) >= 190
  7. for party in tqdm.tqdm(reader.parties["code"]):
  8. try:
  9. df = pd.read_parquet(f"data/annexI/{party}.parquet")
  10. except FileNotFoundError:
  11. df = pd.read_parquet(f"data/non-annexI/{party}.parquet")
  12. assert not df.duplicated(
  13. ["party", "category", "classification", "measure", "gas", "unit", "year"]
  14. ).any()
  15. print("no obvious problems")
  16. if __name__ == "__main__":
  17. main()