databricks_catalog_access_report

databricks_catalog_access_report.py

from pyspark.sql.functions import *
from functools import reduce

df_grants = []

df_dbs = spark.sql("show databases")
dbs = df_dbs.select("databaseName").collect()
for db in dbs:
    database_name = db[0]
    df = spark.sql("SHOW GRANT ON DATABASE {}".format(database_name))
    df_grants.append(df)
    df_tables = spark.sql("show tables in {}".format(database_name)).filter(col("isTemporary") == False)
    tables = df_tables.collect()
    for table in tables:
        df = spark.sql("SHOW GRANT ON TABLE {}.{}".format(table.database, table.tableName))
        df_grants.append(df)
df_all = reduce(DataFrame.unionAll, df_grants)
display(df_all)