Priority Model
Casbin supports loading policies with priority.
Load Policy with Implicit Priority
It's quite simple: the order determines the priority; policies that appear earlier have higher priority.
model.conf:
[policy_effect]
e = priority(p.eft) || deny
Load Policy with Explicit Priority
Also see: casbin#550
A smaller priority value indicates a higher priority. If there's a non-numerical character in the priority, it will be placed last instead of throwing an error.
The conventionally used priority token name in the policy definition is "priority". To use a custom one, you need to invoke e.SetFieldIndex()
and reload the policies (see the full example on TestCustomizedFieldIndex).
model.conf:
[policy_definition]
p = customized_priority, sub, obj, act, eft
Golang code example:
e, _ := NewEnforcer("./example/priority_model_explicit_customized.conf",
"./example/priority_policy_explicit_customized.csv")
// Due to the customized priority token, the enforcer fails to handle the priority.
ok, err := e.Enforce("bob", "data2", "read") // the result will be `true, nil`
// Set PriorityIndex and reload
e.SetFieldIndex("p", constant.PriorityIndex, 0)
err := e.LoadPolicy()
if err != nil {
log.Fatalf("LoadPolicy: %v", err)
}
ok, err := e.Enforce("bob", "data2", "read") // the result will be `false, nil`
Currently, explicit priority only supports AddPolicy
& AddPolicies
. If UpdatePolicy
has been called, you shouldn't change the priority attribute.
model.conf:
[request_definition]
r = sub, obj, act
[policy_definition]
p = priority, sub, obj, act, eft
[role_definition]
g = _, _
[policy_effect]
e = priority(p.eft) || deny
[matchers]
m = g(r.sub, p.sub) && r.obj == p.obj && r.act == p.act
policy.csv
p, 10, data1_deny_group, data1, read, deny
p, 10, data1_deny_group, data1, write, deny
p, 10, data2_allow_group, data2, read, allow
p, 10, data2_allow_group, data2, write, allow
p, 1, alice, data1, write, allow
p, 1, alice, data1, read, allow
p, 1, bob, data2, read, deny
g, bob, data2_allow_group
g, alice, data1_deny_group
request:
alice, data1, write --> true // because `p, 1, alice, data1, write, allow` has the highest priority
bob, data2, read --> false
bob, data2, write --> true // because bob has the role of `data2_allow_group` which has the right to write data2, and there's no deny policy with higher priority
Load Policy with Priority Based on Role and User Hierarchy
The inherited structure of roles and users can only be multiple trees, not graphs. If a user has multiple roles, you have to make sure the user has the same level in different trees. If two roles have the same level, the policy (associated with the role) that appeared earlier has higher priority. For more details, also see casbin#833 and casbin#831.
model.conf:
[request_definition]
r = sub, obj, act
[policy_definition]
p = sub, obj, act, eft
[role_definition]
g = _, _
[policy_effect]
e = subjectPriority(p.eft) || deny
[matchers]
m = g(r.sub, p.sub) && r.obj == p.obj && r.act == p.act
policy.csv
p, root, data1, read, deny
p, admin, data1, read, deny
p, editor, data1, read, deny
p, subscriber, data1, read, deny
p, jane, data1, read, allow
p, alice, data1, read, allow
g, admin, root
g, editor, admin
g, subscriber, admin
g, jane, editor
g, alice, subscriber
Request:
jane, data1, read --> true // because jane is at the bottom, her priority is higher than that of editor, admin, and root
alice, data1, read --> true
The role hierarchy looks like this:
role: root
└─ role: admin
├─ role editor
│ └─ user: jane
│
└─ role: subscriber
└─ user: alice
The priority automatically looks like this:
role: root # auto priority: 30
└─ role: admin # auto priority: 20
├─ role: editor # auto priority: 10
└─ role: subscriber # auto priority: 10