Guruprasad J Rao
Guruprasad J Rao

Reputation: 29683

Firebase rules validation does not validate data

In continuation to my previous question I have below design for my application.

Design

Here is how I add the data through my android application.

FirebaseInstance mFirebaseInstance = FirebaseDatabase.getInstance();
FirebaseDatabase mFirebaseDatabase = mFirebaseInstance.getReference("tbl-customers").child(FirebaseAuth.getInstance().getCurrentUser().getUid());
//This will create or fetch user id node under tbl-customers.

btnSave.setOnClickListener(new View.OnClickListener() {
       @Override
       public void onClick(View view) {
            String name = inputName.getText().toString();
            String email = inputCode.getText().toString();
            String limit= inputLimit.getText().toString();
            createUser(name, email,limit);
       }
});


private void createUser(String name, String email,String limit) {
    userId = mFirebaseDatabase.push().getKey();
    Customer customer = new Customer(name, email,limit);

    mFirebaseDatabase.child(userId).setValue(customer);
}

That's it.. In Database it is somehow represented as below:

tbl-customers
|___loggedInUserId1
    |___customerID1
        |___customerName
        |___customerCode
        |___customerLimit
    |___customerID2
        |___customerName
        |___customerCode
        |___customerLimit
    |___customerID3
        |___customerName
        |___customerCode
        |___customerLimit
|___loggedInUserId2
    |___customerID4
        |___customerName
        |___customerCode
        |___customerLimit

and I have rules defined as

{
    "rules": {
        "tbl-customers": {
            ".read": "auth != null",
            ".write": "auth != null",
            "$custId": {
                "customerName": {
                    ".validate": "newData.isString() && newData.val().length < 100 && newData.val().length > 8"
                },
                "customerCode": {
                    ".validate": "newData.isString() && newData.val().length<4 && !newData.exists() && newData.val().length>1"
                },
                "customerLimit": {}
            }
        }
    }
}

Unfortunately, the data is inserted without considering any of the validation written for each properties. Even the empty data gets inserted.

I started thinking whether the rule written is of proper structure, because if I see the data inserted then it has 3 levels - tbl-customer-->loggedInUserId-->customerId but rules have been only written for tbl-customer-->customerId.

So I changed the rules as below.

{
    "rules": {
        "tbl-customers": {
            ".read": "auth != null",
            ".write": "auth != null",
            "$user_id": {
                ".validate": "auth.uid===$user_id",
                "$custId": {
                    "customerName": {
                        ".validate": "newData.isString() && newData.val().length < 100"
                    },
                    "customerCode": {
                        ".validate": "newData.isString() && newData.val().length<4 && !newData.exists()"
                    },
                    "customerLimit": {}
                }
            }

        }
    }
}

extending it to one more level by including,

"$user_id": {
             ".validate": "auth.uid===$user_id"
           ...
}

But now this throws Permission Denied Exception. Am out of ideas at this point of time. Could someone guide me in the right direction? I have referred lot of posts from my previous question but to my bad, I couldn't grab much information from it. Hope to find some clear explanation as on why rules validation are failing and why data gets inserted with above mentioned first rule.

Upvotes: 0

Views: 1260

Answers (1)

Richard Ansell
Richard Ansell

Reputation: 926

I might be wrong here, but the .read and .write rules will always cascade down your tree regardless here, which the line ".write": "auth != null" is most likely where you're facing the problems.

It would be best to move this rule to where the ".validate": "auth.uid===$user_id" rule is and remove the validate rule. You would then change it to something like "auth.uid == $user_id". The 'validate' rule I believe is just for taking in a written input at that specific location and then accepting or rejecting.

This is how I believe it should look:

{
    "rules": {
        "tbl-customers": {
            ".read": "auth != null"
            "$user_id": {
                ".write": "auth.uid === $user_id"
                "$custId": {
                    "customerName": {
                        ".validate": "newData.isString() && newData.val().length < 100"
                    },
                    "customerCode": {
                        ".validate": "newData.isString() && newData.val().length<4 && !newData.exists()"
                    },
                    "customerLimit": {}
                }
            }

        }
    }
}

This structure will allow all logged in users to read the data, which you could restrict further to the $user_id location if you wish, similar to the write rule currently in place. This would then only allow users who match the $user_id read and write privileges.

CASE STUDY

To add to this, this is a structure I set up sometime ago that would allow an administrator the ability to add users:

{
  "rules": {
      "Administrator": {
          ".read": "auth != null"
      },
      "Users": {
        "$user_id": {
          ".write": "$user_id === auth.uid",
          ".read": "$user_id === auth.uid"
        }
      }
  }
}

Upvotes: 2

Related Questions