anonymize

Anonymize a SQL database (currently only postgresql)

simple usage

./anonymize.sh [psql_command] [config_filename]

see anonymize.sh for implementation details.

how it works

anonymize will analyze your sql schema to generate a set of update statements and apply to the given database. All nullable fields become null, all character based fields become CONCAT(table_name, column_name, id), all numerics become 1, etc. primary key fields and relationships will remain intact. there is only one update statement per table (i.e. bulk update). defaults and specific table/column values can be customized.

configuration

anonymize requires a config file to run (example provided). the config is a javascript object that consists of 3 parts:

1. _default_type_values format: [ {types: [string], value: string}, ... ]. types is a list of sql data types, value is a sql expression.
2. _skip_all format: {tables: [string], types: [string], columns: [string]}. skips anonymizing specified tables, types, or columns.
3. custom table and column values: format: table_name: {column_name: value, ...}. value must be a sql expression that will apply to all rows in the column. optional: table_name: {_skip: [string]} to skip specified columns.

when working with the config file, keep in mind you can and should use raw sql statements on the value properties. for example: {email: "CONCAT('user', id, '@domain.com')"} will be per row, whereas {email: 'user${id}@domain.com'} will be per table (and id is undefined).

debugging

anonymize.sh is a three step process:

1. generate the schema template json.
2. generate the update statements.
3. apply the update statements.

these can be broken down during initial setup and debugging:

1. psql -d my_database -t -c "$(node anonymize.js -gen postgres)" > _schema.json should create a recognizable json file of your schema
2. node anonymize.js _schema.json ./config.js > _updates.sql should generate a list of update statements.
3. psql -d my_database -c "$(cat _updates.sql)" should run the updates.