README

Overview

This PHP library can anonymize a database dump file with random/fake data, making it particularly useful for anonymizing data when restoring a live database in a non-live environment. This helps mitigate the risk of exposing real customer data if the non-live database is stolen, for example.

How it works

It reads a file with the DB dump in the local storage (or from STDIN), then it produces a new DB dump file (or outputs in STDOUT) based on the instructions about which columns should be anonymized.

It is much faster than trying to run multiple UPDATE requests to a running DB, because the DB needs to handle with many things like: checking the uniqueness of some columns, indexing, and stuff like that.

Install

To install this library for usage, just run:

composer create-project printi/db-dump-file-anonymizer

Usage

This library exports an executable script called anonymize-db-dump. It is installed in the folder ./vendor/bin/ when the library is required.

The script can be called this way:

$ ./vendor/bin/anonymize-db-dump [OPTIONS]

These are the available options:

• -h or --help: to show the help
• -i FILE or --input=FILE: to inform the input file (dump file of MySQL)
• --stdin: to read input from STDIN
• -o FILE or --output=FILE: to inform the output file
• --stdout: to write the output in the STDOUT
• -m MODIFICATIONS or --modifications=MODIFICATIONS: to inform the JSON of expected modifications
• -f FILE or --modifications-file=FILE: to inform the JSON file with expected modifications
• -l LOCALE or --locale=LOCALE: to inform the locale to be used by Faker
• -q or --quiet: to ommit messages
• -r SIZE or --read-buffer-size=SIZE: to inform the read buffer size (example: 100, 1KB, 1MB, 1GB)
• -w SIZE or --write-buffer-size=SIZE: to inform the write buffer size (example: 100, 1KB, 1MB, 1GB)

This library uses Faker to generate the fake (anonymous) data. You will need to check the library to see the available options of formatters, arguments for formatters and locales.

Simple example:

$ php ./vendor/bin/anonymize-db-dump -i ./sample/dump.sql -o ./sample/dump.out.sql -f ./sample/modifications.json -l pt_BR

The example above will read the file ./sample/dump.sql, then produce a modified dump file ./sample/dump.out.sql using the modification instructions specified in the file ./sample/modifications.json and using the locale pt_BR (Brazilian Portuguese) to generate the fake data with localization.

If you want to generate the modified dump file in the STDOUT, you may replace -o ./sample/dump.out.sql by --stdout. This option is useful if you do not want to store the modified dump file and use the modified dump file directly to the mysql command to restore the data like the example bellow:

$ php ./vendor/bin/anonymize-db-dump -i ./sample/dump.sql --stdout -f ./sample/modifications.json -l pt_BR | mysql -uroot -proot -h localhost -D dbname

If you want to generate the modified dump file from the STDIN, you may replace -i ./sample/dump.sql by --stdin. This option is useful if you are getting the dump file from mysqldump directly like the example bellow:

$ mysqldump ... | php ./vendor/bin/anonymize-db-dump --stdin --o ./sample/dump.out.sql -f ./sample/modifications.json -l pt_BR

Specification of the JSON of modifications

The JSON to specify the modifications over the dump file uses this structure:

{
  "<TABLE_NAME>": {
    "<COLUMN_POSITION>": {
      "quote": <BOOLEAN_VALUE>,
      "format": <STRING_VALUE>,
      "args": <ARRAY_OF_ARGS>,
      "unique": <BOOLEAN_VALUE>,
      "optional": <BOOLEAN_VALUE>,
      "optional_weight": <FLOAT_VALUE>,
      "optional_default_value": <VALUE>
    }
  }
}

Basically, we need to specify the tables names, then the column positions of each table (starting from position 1), then the specification of the column.

The "quote" option is used to inform the value should be delimited by quotes (true) or not (false). For example, numeric values does not need to use quotes.

The "format" option is used to specify the format of the column according to the Faker formaters.

The "args" option is used to specify the array of arguments to be used when calling the specified format.

The "unique" option is used to specify the column is unique and should not repeat for that column of that table (it might repeat on other columns or other tables).

The "optional" option is used to specify whether the column is optional or not. When a column is optional, the library may select the default value for the field (if the default value is not specified, NULL is used).

The "optional_weight" is used to specify the probability to have an empty value. The value 0.0 indicates the default value will always be used, 1.0 indicates the default value will never be used, 0.2 indicates there is 20% of chance to use the default value.

The "optional_default_value" is used to specify the default value.

Example of modifications:

{
  "customers": {
    "2": {
      "quote": true,
      "format": "firstName"
    },
    "3": {
      "quote": true,
      "format": "lastName"
    },
    "4": {
      "quote": true,
      "format": "email",
      "unique": true
    },
    "5": {
      "quote": true,
      "format": "phoneNumber",
      "optional": true,
      "optional_weight": 0.9,
      "optional_default_value": null
    },
    "6": {
      "quote": true,
      "format": "lexify",
      "unique": true,
      "args": [
        "TEST?????????"
      ]
    }
  }
}