in src/translation/scripts/hive/extract_hive_ddls_manual.py [0:0]
def get_table_format(tbl, hive_db, spark):
"""
Get table format
"""
df = spark.sql(f"describe formatted {hive_db}.{tbl}")
format_str = (
df.filter("col_name == 'InputFormat'").select("data_type").first()[0].upper()
)
if "AVRO" in format_str:
return "AVRO"
elif "PARQUET" in format_str:
return "PARQUET"
elif "ORC" in format_str:
return "ORC"
elif "TEXT" in format_str:
return "CSV"
else:
return "OTHER"