Skip to main content

ABAC

What is the ABAC model actually?

ABAC is Attribute-Based Access Control, meaning you can use the attributes (properties) of the subject, object or action instead of themselves (the string) to control the access. You may already hear of a complicated ABAC access control language named XACML. Compared to XACML, Casbin's ABAC is very simple: in ABAC, you can use structs (or class instances based on the programming language) instead of string for model elements.

Use the official ABAC example for example:

[request_definition]
r = sub, obj, act

[policy_definition]
p = sub, obj, act

[policy_effect]
e = some(where (p.eft == allow))

[matchers]
m = r.sub == r.obj.Owner

We use r.obj.Owner instead of r.obj in the matcher. The r.obj passed in the Enforce() function will be a struct or class instance instead of string. Casbin will use reflection to retrieve the obj member variable in that struct or class for you.

Here is a definition for the r.obj struct or class:

type testResource struct {
Name string
Owner string
}

How to use ABAC?

Simply speaking, to use ABAC, you need to do two things:

  1. Specify the attributes in the model matcher.
  2. Pass in the struct or class instance for the element as the argument in Casbin's Enforce() function.
danger

Currently, only request elements like r.sub, r.obj, r.act and so on support ABAC. You cannot use it on policy elements like p.sub, because there is no way to define a struct or class in Casbin's policy.

tip

You can use multiple ABAC attributes in a matcher, for example: m = r.sub.Domain == r.obj.Domain.

tip

If you need to use comma in policy which conflicts with csv's separator and we need to escape it. Casbin parses policy file through csv library, you could surround statement with quotation marks. For example, "keyMatch("bob", r.sub.Role)" will not be split.

Scaling the model for complex and large number of ABAC rules.

The above instance of ABAC implementation is at its core very simple, but oftentimes the authorization system needs a very complex and large number of ABAC rules. To fit this necessity the above implementation will increase the verbosity of the model to a large extent. So, it’s wise to add the rules in the policy instead of in the model. This is done by introducing a eval() functional construct. Below is the example instance to manage such ABAC models.

This is the definition of the CONF file used for defining the ABAC model.

[request_definition]
r = sub, obj, act

[policy_definition]
p = sub_rule, obj, act

[policy_effect]
e = some(where (p.eft == allow))

[matchers]
m = eval(p.sub_rule) && r.obj == p.obj && r.act == p.act

Here, p.sub_rule is of type struct or class(user-defined type) which consists of necessary attributes to be used in the policy.

This is the policy that is used against the model for Enforcement. Now, you can use the object instance which is passed to eval() as a parameter to define certain ABAC constraints.

p, r.sub.Age > 18, /data1, read
p, r.sub.Age < 60, /data2, write