cansecurity
Advanced tools
Authentication, authorization, session manager, single-sign-on (SSO) and security framework for node applications
cansecurity is a security library for user authentication management and authorization in node (and specifically expressjs) apps.
cansecurity will manage your user authentication, handle stateless sessions, using both native express sessions and its own session mechanism, as you request. The cansecurity stateless session mechanism means that, with a shared secret, a user can be authenticated automatically across nodejs instances, essentially creating single-sign-on.
cansecurity also provides middleware to handle authorization in your requests. It is structured in a manner similar to the expressjs examples restrictTo*.
Installation is fairly straightforward, just install the npm module:
npm install cansecurity
Usage involves two key steps: initialization and validation.
To initialize cansecurity, you must first require() it, and then init() it, which will return the middleware you can use:
var cs = require('cansecurity');
var cansec = cs.init({});
In initialization, you set four key authentication parameters as properties of the config object passed to cs.init():
Validation is straightforward. Once you have set up cansecurity properly, it functions as standard expressjs middleware:
server.use(cansec.validate);
This should generally be done before your router.
If the user is successfully authenticated, then the user object will be placed in two locations:
req["X-CS-Auth"];
req.session["X-CS-AUTH"].user; // only if expressjs sessions have been enabled
validatePass: validatePass must have the following signature
validatePass(username,password,callback);
The validatePass function is expected to validate a given password for a given username, and indicate to the callback if it succeeded or failed. The signature and expected parameters to the callback are as follows:
callback(user,message,pass);
Where: user = the actual user object. This can be a function, a JavaScript object, anything you want. It will be placed inside the session and the request for you to use later. If validation was successful, this must not be null/undefined. message = the error message in case of validation failure. This can be anything you want, and will be passed along with the 401 unauthenticated response. pass = the user's password or any other unique per-user string, not easily guessable. Commonly, this would be the hash of a password.
getUser: getUser returns a user object for a given user. This is only called when we do not need/use the password to validate the user. If we did, we would already have the user from validatePass(). Rather, we use getUser to get a user object and generate a new set of auth credentials for user who we authenticated using either a local expressjs session or a CS session.
getUser has the following signature:
getUser(username,success,failure);
getUser is expected to know what to do with a given username, and get the user object to pass to the callback. The signature and expected parameters of the success callback are as follows:
success(user,login,password);
Where: user = the actual user object. This can be a function, a JavaScript object, anything you want. It will be placed inside the session and the request for you to use later. If such a user is found, this must not be null/undefined. pass = the user's password or any other unique per-user string, not easily guessable. Commonly, this would be the hash of a password.
The signature and expected parameters of the failure callback are as follows:
failure(error);
cansecurity will never directly return errors. It will authenticate a user, or fail to do so, set request["X-CS-Auth"], and request.session["X-CS-Auth"] if sessions are enabled, and then call next() to jump to the next middleware.
cansecurity will call next(error) in only the following case: If the user has provided HTTP Basic Authentication credentials in the form of username/password and the authentication fails. In that case, cansecurity will call
next({status: 401, message:"some message"});
It is up to you to make sure that you use expressjs's server.error() handler to correctly handle this error.
getUser() and validatePassword() both require the calling program to return a password. Although this is never passed out, why is the password necessary?
In reality, this can be any unique string at all, as long as it is consistent for the same user. Normally, this would be a hashed password. This is used, along with the secret session key, to create the authtoken for CS sessions. Without using the password or some other unique, non-guessable string, it would be theoretically possible to use one login to spoof another. With the unique non-guessable user string (hashed password or similar) as part of the hash input, this risk is mitigated. PLEASE PLEASE PLEASE do not pass cleartext passwords here. In reality, your app should never know cleartext passwords, rather storing them as SHA1 or similar hashes.
Thus, to create a unique authentication token that is useful for single-sign-on and cannot be spoofed to another user, we include the unique user string (e.g. a hashed password) as part of the input to the authentication token.
With each request, the following algorithm is followed:
Note that failing to get an authentication for all of the above steps does not necessarily indicate that a 401 or 403 should be sent back. It is entirely possible that the user is accessing a resource that does not require authentication! This part of the cansecurity library is entirely about authentication; authorization is a different topic.
The X-CS-Auth header contains error responses or success tokens. If authentication was successful, by any means, then a new header is generated with each request. This header is of the following format:
success=sha1hash:username:expiry
Where: sha1hash = a sha1 hash of the username, the expiry, the secret session key and the user's unique string (likely itself a hashed password). username = the user's username expiry = when this auth token will expire, as a JavaScript (Unix) millisecond timestamp, provided by Date().getTime().
Essentially, we are using a message validation algorithm to validate that the username and expiry are, indeed, valid.
Because the auth header is created anew with each request, the expiry window is rolling - x minutes from the last valid request.
Extensive performance testing has not been done. However, all of the algorithms are symmetric, which are very high-performance. The expensive part is getUser() and validatePassword(), which may require your app to look in a data source or database. However, since the majority of requests will simply hit the local session, the user will be stored locally, and it is not an issue. The hit will only be for the first authentication for each user, as well as when a user switches between nodejs servers using SSO.
For a good example, see the test suite in test/test.js, specifically the section beginning cansec.init. It is reproduced below:
var cs = require('cansecurity'), cansec,
// static database for testing
user = {name:"john",pass:"1234",age:25};
cansec = cs.init({
getUser: function(login,success,failure){
if (user.name === login) {
success(user,user.name,user.pass);
} else {
failure();
}
},
validatePassword: function(login,pass,cb){
var p = null, message, resuser = null;
if (user.name !== login) {
message = "invaliduser";
} else if (user.pass !== pass) {
message = "invalidpass";
} else {
message = null;
resuser = user;
p = pass;
}
cb(resuser,message,p);
},
sessionKey: SESSIONKEY
});
// create our express server
server = express.createServer();
server.configure(function(){
server.use(express.cookieParser());
server.use(express.session({secret: "agf67dchkQ!"}));
server.use(cansec.validate);
server.use(function(req,res,next){
// send a 200
sendResponse(req,res,200);
});
});
server.error(function(err,req,res,next){
var data;
if (err && err.status) {
// one of ours
data = err.message ? {message: err.message} : null;
sendResponse(req,res,err.status,data);
} else if (err && err.type && err.type === "unexpected_token") {
// malformed data
sendResponse(req,res,{message:err.type},400);
} else {
sendResponse(req,res,500);
}
});
server.listen(PORT);
Authorization is structured as route-specific middleware. Once authentication has (or has not) been performed, it is possible to restrict access to a particular route.
As of release 0.3.0, authorization also includes conditionals, allowing the authorization to be applied only if certain parameters are met.
The easiest way to demonstrate this is with an example, following which we will describe all of the options and APIs.
express = require('express'),
cansec = require('cansecurity').init({});
server = express.createServer();
// do lots of server initialization
server.get("/some/route/:user",cansec.restrictToLoggedIn,routeHandler);
server.get("/some/route",cansec.ifParam("private","true").restrictToRoles("admin"),routeHandler);
Usage of cansecurity authorization is only possible if you are using cansecurity authentication as well. To use authorization, you do two steps: initialization and middleware.
The authorization component of cansecurity is initialized at the same time as the authentication component:
var cs = require('cansecurity'), cansec;
cansec = cs.init({});
or more simply:
var cansec = require('cansecurity').init({});
In initialization, you set two key authorization parameters as properties of the config object passed to cs.init(). Both are objects and both are optional.
Initialization returns the object that has the restrictTo* middleware.
As in the example above, once you have authentication and authorization set up and initialized, you may use authorization middleware:
server.get("/some/route/:user",cansec.restrictToLoggedIn,routeHandler);
server.get("/my/data/:user",cansec.restrictToSelf,routeHandler);
server.get("/admin/:user",cansec.restrictToRoles("admin"),routeHandler);
server.get("/user",cansec.ifParam("secret","true").restrictToRoles("admin"),routeHandler);
cansecurity authorization will never directly return errors. If a restrictTo* middleware is called, and authorization fails, it will call next(error). The error will always be structured as follows:
next({status: 403, message:"unauthorized"});
Obviously, authentication comes before authorization, and if the user fails to authenticated, you may get a 401 from the authentication section without ever reaching authorization.
It is up to you to make sure that you use expressjs's server.error() handler to correctly handle this error.
The following authorization middleware methods are available. Each one is followed by an example. There are two sections
Regular API interfaces are used to restrict access, each example is given below.
server.get("/some/route/:user",cansec.restrictToLoggedIn,routeHandler);
var cs = require('cansecurity');
cansec = cs.init({
fields: {id: "userid"},
params: {id: "user"}
});
// only allow a person to see their own stuff
server.get("/some/route/:user",cansec.restrictToSelf,routeHandler);
// note that the param in the route is ":user", which matches the params.id:"user" in cansec.init()
var cansec = require('cansecurity').init({
fields: {id: "userid", roles:"roles"},
params: {id: "user"}
});
server.get("/api/admin",cansec.restrictToRoles("admin"),routeHandler);
server.get("/api/siteadmin",cansec.restrictToRoles(["admin","superadmin"]),routeHandler);
// will require the User object from authentication to have a property "roles", which is an array of strings. If one of the strings is "admin", then /api/admin will be allowed. If one of the roles is "admin" or "superadmin", then /api/siteadmin will be allowed.
var cansec = require('cansecurity').init({
fields: {id: "userid", roles:"roles"},
params: {id: "user"}
});
server.put("/api/user/:user",cansec.restrictToSelfOrRoles("admin"),routeHandler);
/*
* Will work if one of the following is true:
* 1) The logged in user has a property "userid" and it matches exactly the value of the param ":user"; OR
* 2) The logged in user has a property "roles" which is an array of strings, one of which is "admin"
*/
var cansec = require('cansecurity').init({
fields: {id: "userid", roles:"roles"},
params: {id: "user"}
});
server.put("/api/user/search",cansec.restrictToParam("searchParam"),routeHandler);
/*
* Will work if the logged in user has a property "userid" (since init() set fields.id to "userid"), and the value of that property matches req.param("searchParam").
* Useful for using parameters in searches.
*/
var cansec = require('cansecurity').init({
fields: {id: "userid", roles:"roles"},
params: {id: "user"}
});
server.put("/api/user/search",cansec.restrictToParamOrRoles("searchParam",["admin","superadmin"]),routeHandler);
server.put("/api/address/search",cansec.restrictToParamOrRoles(["searchParam","addParam"],"admin"),routeHandler);
/*
* Will work if one of the following is true:
* 1) the logged in user has a property "userid" (since init() set fields.id to "userid"), and the value of that property matches req.param("searchParam"), or, in the second example, one of "searchParam" or "addParam".
* 2) The logged in user has the role, as one of the array of strings of the property "roles", set to "admin" or "superadmin" (for the first example), or "admin" (for the second example).
*/
var cansec = require('cansecurity').init({
fields: {id: "userid", roles:"roles"},
params: {id: "user"}
});
server.get("/api/user/search",cansec.restrictToField("owner",getObjectFn),routeHandler);
/*
* Will call getObjectFn(req,res) to get a regular JavaScript object, and then try to match the requested fields, in the above example "owner", to the ID of the User from authentication. The ID from authentication is user.userid, as given in the init() for fields.id.
*/
var cansec = require('cansecurity').init({
fields: {id: "userid", roles:"roles"},
params: {id: "user"}
});
server.get("/api/user/search",cansec.restrictToFieldOrRoles("owner","admin",getObjectFn),routeHandler);
/*
* Will call getObjectFn(req,res) to get a regular JavaScript object, and then try to match the requested fields, in the above example "owner", to the ID of the User from authentication. The ID from authentication is user.userid, as given in the init() for fields.id.
*/
server.get("/api/user/search",cansec.restrictToFieldOrRoles(["owner","recipient"],["admin","superadmin"],getOwnerFn),routeHandler);
A typical use case for restrictToField and its variant restrictToFieldOrRoles is that you may load an object, and want to restrict its access to the owner of an object. For example, let us assume that you are loading an employee record. For that, restrictToSelf would be fine, since the User ID from authentication is likely to match the ID for requesting the employee record. The following example shows this use case:
var cansec = require('cansecurity').init({
fields: {id: "userid", roles:"roles"},
params: {id: "user"}
});
server.get("/api/employee/:user",cansec.restrictToSelfOrRoles("admin"),sendDataFn);
However, what if you are retrieving a record whose authorization requirements are not known until it is loaded. For example, you are loading a paystub, whose URL is /api/paystubs/34567. Until you load the paystub, you don't actually know who the owner is. Of course, you might make it accessible only via a more detailed API as /api/employee/12345/paystubs/34567, but let us assume that you need to do it directly, with the first simplified API. Until you load the actual paystubs object, and see that the employee is, indeed, 12345, the one who logged in, you don't know whether or not to show it. The following example describes how to simply implement this use case:
var cansec = require('cansecurity').init({
fields: {id: "userid", roles:"roles"},
params: {id: "user"}
});
server.get("/api/employee/:user",cansec.restrictToSelfOrRoles("admin"),sendDataFn);
server.get("/api/paystub/:payid",payStubLoad,cansec.restrictToField("employee",getObjectFn),sendDataFn);
In this example, we load the paystub, but do not send it. The paystub object retrieved by payStubLoad looks like this:
{
id: "34567",
employee: "12345",
date: "2011-01-31",
amount: "$100"
}
This is then stored in the request object. Now getObjectFn can return the same object, which has employee as "12345". This is then matched to User.userid, which will allow it to proceed.
The conditional API simply creates conditions under which the regular restrictions are applied. If the conditions are not met, then restrictions are not applied.
What if you have a resource that is normally accessible by all, but if certain parameters are applied, e.g. ?secret=true, then it should be restricted to an admin?
server.get("/api/employee",cansec.ifParameter("secret","true").restrictToRoles("admin"),sendDataFn);
In the above example, anyone can do a GET on /api/employee, but if they pass the parameter ?secret=true, then they will have to be logged in and have the role "admin" defined.
In our example, sendDataFn also checks for that parameter. If it is not set, then it sends public data about the employee list; if it is set, it sends public and private data, trusting that cansecurity prevented someone from getting in with ?secret=true unless they are authorized.
A common pattern, as shown in the last example above, is to retrieve an object, check it against the user, and then determine whether or not to allow the request to proceed. cansecurity would love to be able to just do the check directly. It even knows which field/property to check: "employee". It has two problems, however:
FAQs
cansecurity is your all-in-one security library for user authentication, authorization and management in node expressjs apps
The npm package cansecurity receives a total of 53 weekly downloads. As such, cansecurity popularity was classified as not popular.
We found that cansecurity demonstrated a not healthy version release cadence and project activity because the last version was released a year ago. It has 1 open source maintainer collaborating on the project.
Did you know?
Socket for GitHub automatically highlights issues in each pull request and monitors the health of all your open source dependencies. Discover the contents of your packages and block harmful activity before you install or update your dependencies.
Security News
Research
A malicious npm package disguised as a WhatsApp client is exploiting authentication flows with a remote kill switch to exfiltrate data and destroy files.
Security News
PyPI now supports digital attestations, enhancing security and trust by allowing package maintainers to verify the authenticity of Python packages.
Security News
GitHub removed 27 malicious pull requests attempting to inject harmful code across multiple open source repositories, in another round of low-effort attacks.