anonymizer

anonymizer

Let's say you have to put a dataset on your website for your own use, but don't really want to share it to everyone. If you upload the raw, nice dataset with all the field names, people can just take it for their own use easily. You want to make it a bit harder for them to interpret the data without shooting yourself in the foot.

This anonymizer was built for that purpose:

• Given a schema, it can easily strip out all the field names.
• It works with nested schema, e.g., Array of Objects, Array of Objects that contains Array, and so on.
• It can also take all categorical values and map them into integers, which are more compact and help you anonymize the data.

The library is also very lightweight (raw 3KB, minified 1KB) and supports CommonJS, AMD (RequireJS) as well as browser use.

1. Installation

Install with npm:

npm install --save anonymizer

or bower

bower install --save anonymizer

2. Usage

var anonymizer = new Anonymizer();

2.1 Basic

anonymizer.encode(data, schema)

anonymizer.encode(
  {a: 1, b: true, c: 'test'},
  {a: 'Number', b: 'Boolean', c: 'String'}
);
=> [1, true, 'test']

anonymizer.encode(
  [{a: 1, b: true, c: 'test'}, {a: 2, b: false, c: 'test2'}],
  [{a: 'Number', b: 'Boolean', c: 'String'}]
);
=> [ [1, true, 'test'], [2, false, 'test2'] ]

anonymizer.decode(data, schema)

anonymizer.decode(
  [1, true, 'test'],
  {a: 'Number', b: 'Boolean', c: 'String'}
);
=> {a: 1, b: true, c: 'test'}

anonymizer.decode(
  [ [1, true, 'test'], [2, false, 'test2'] ],
  [{a: 'Number', b: 'Boolean', c: 'String'}]
);
=> [{a: 1, b: true, c: 'test'}, {a: 2, b: false, c: 'test2'}]

2.2 Categorical values

Anonymizer can map categorical values to/from integers.

anonymizer.encode('test', 'Category');
=> 1

anonymizer.encode(['test', 'test2', 'test'], ['Category']);
=> [1, 2, 1]

anonymizer.getCategories();
=> ['test', 'test2']

// Create a new anonymizer and pass known categories
var anonymizer2 = new Anonymizer(anonymizer.getCategories());

anonymizer2.decode(1);
=> 'test'

anonymizer2.decode([1, 2, 1]);
=> ['test', 'test2', 'test']

3. Schema definition

Primitive values Use string

2
=> schema = 'Number'
'test value'
=> 'String'
true
=> schema = 'Boolean'

Categorical values Use string "Categorical"

'category1'
=> schema = 'Categorical'

Object Use curly braces and contain all the keys that you want to export. The value for each key is the type for that field.

{a: 1, b:'test', c: true}
=> schema = {a: 'Number', b: 'String', c: 'Boolean'}

Array Use square brackets and contains the type of an element. This library assume that all children of an array are of the same type, so you will define the schema for the child only once.

[1, 2, 3]
=> schema = ['Number']
['test1', 'test2']
=> schema = ['String']
['category1', 'category2']
=> schema = ['Category']

Nesting You can combine the syntax above to define any schema for you data.

Array of objects

[
  {a: 1, b: true, c: 'test'},
  {a: 2, b: false, c: 'test2'}
]
=> schema = [{a: 'Number', b: 'Boolean', c: 'String'}]

Array of arrays

[
  [1, 2, 3],
  ['a', 'b', 'c']
]
=> schema = [['Number'], ['String']]

Object that contains an array

{
  a: [1, 2, 3],
  b: ['a','b','c'],
  c: 'test'
}
=> schema = {a: ['Number'], b: ['String'], c: 'String'}

4. Testing

From the repo root:

npm install
npm test