This solution seems to have been answered by an AI.
I need to know how create a SELECT for create this tables: USERS, GROUPS and GROUP_USERS.

---

@Sardenberg wrote:

This solution seems to have been answered by an AI.
I need to know how create a SELECT for create this tables: USERS, GROUPS and GROUP_USERS.

---

# Databricks notebook source
from pyspark.sql.functions import lit

# COMMAND ----------

df_groups = spark.sql("""SHOW GROUPS""")

# COMMAND ----------

df_users = spark.sql("""SHOW USERS""")

# COMMAND ----------


result = spark.createDataFrame([], "name: string, directGroup: boolean, user: string")
for user in df_users.collect():
    df_user_groups = spark.sql(f"""SHOW GROUPS WITH USER `{user.name}`""").withColumn("user", lit(user.name))
    result = result.unionAll(df_user_groups)

# COMMAND ----------

display(result) # Here you can create temp view and select data

It will work fine if groups added to workspace.

I don't think it's possible yet. Unfortunately, I look in all system tables and command and didn't found this kind of things.

But with a Python notebook, like what did the AI, you can reconstruct it:

first you list all the users with

SHOW USERS

then you loop for all users to get its groups with:

SHOW GROUPS WITH USER `{user.name}`

And you can reconstruct the list of users in each groups like that, and put it in a delta table with a create table.

With this table you can easily list all users in a specific group in a SQL request. You "just" need to run your python notebook regularly to update your user-group table.

Due to the special type of command that this is, you can't (to my knowledge) do that kind of thing directly in SQL, (can't "SELECT * FROM SHOW ..." for example), and there is no specific command for that, so it's a little laborious, especially if you have a lot of users and groups.