Kirby Schema Generator

Generates schema files that are used for describing data that is ingested by Kirby.

Usage

A sample data file is required for analyzing which will be used when generating the final schema.

$ python3 generate_schema.py <sample_data_file> --encryption_key_id <key_column_name> --personal_columns <column_names> 

sample_data_file: Path to the sample file to be analyzed.

encryption_key_id: If dataset is to be encrypted the column to use for keys, usually the user id.

personal_columns: Columns that contain personal information.

This will return the generated schema to stdout. To save it to a file:

$ python3 generate_schema.py <sample_data_file> [options] > schema.json

For a description on all options, use:

$ python3 generate_schema.py --help

Schema column description

Column descriptions have to be added to the schema manually after it has been generated.

Examples

From a CSV sample file

users.csv:

user_id, name, email, subscription_start, subscription_end
1000, Kirby Kirbysson, kirby@bonniernews.se, 2019-01-02, 2021-02-01
1001, Luigi Plumberson, luigi@bonniernews.se, 2019-03-01, 2019-04-01

$ python3 generate_schema.py users.csv --encryption_key_id user_id --personal_columns name email

From a JSON sample file

users.json:

{
  "user_id": 1000,
  "name": "Kirby Kirbysson",
  "email": "kirby@bonniernews.se",
  "subscription_start": "2019-01-02",
  "subscription_end": "2021-02-01"
}
{
  "user_id": 1001,
  "name": "Luigi Plumberson",
  "email": "luigi@bonniernews.se",
  "subscription_start": "2019-03-01",
  "subscription_end": "2019-04-01"
}

$ python3 generate_schema.py users.json --input_format json --encryption_key_id user_id --personal_columns name email

TODO

• Add support for nested JSON properties