Permission is hereby granted, free of charge, to any person obtaining a copy of this software and associated documentation files (the "Software"), to deal in the Software without restriction, including without limitation the rights to use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies of the Software, and to permit persons to whom the Software is furnished to do so, subject to the following conditions:
The above copyright notice and this permission notice shall be included in all copies or substantial portions of the Software.
THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
Version 3 adds several new, frequently requested features. To do so, it introduces a few breaking changes. We've worked to keep these as minimal as possible. This guide explains the breaking changes and how you can quickly update your code.
### `Token.Claims` is now an interface type
The most requested feature from the 2.0 verison of this library was the ability to provide a custom type to the JSON parser for claims. This was implemented by introducing a new interface, `Claims`, to replace `map[string]interface{}`. We also included two concrete implementations of `Claims`: `MapClaims` and `StandardClaims`.
`MapClaims` is an alias for `map[string]interface{}` with built in validation behavior. It is the default claims type when using `Parse`. The usage is unchanged except you must type cast the claims property.
The old example for parsing a token looked like this..
fmt.Printf("Token for user %v expires %v",claims["user"],claims["exp"])
}
```
`StandardClaims` is designed to be embedded in your custom type. You can supply a custom claims type with the new `ParseWithClaims` function. Here's an example of using a custom claims type.
fmt.Printf("Token for user %v expires %v",claims.User,claims.StandardClaims.ExpiresAt)
}
```
### `ParseFromRequest` has been moved
To keep this library focused on the tokens without becoming overburdened with complex request processing logic, `ParseFromRequest` and its new companion `ParseFromRequestWithClaims` have been moved to a subpackage, `request`. The method signatues have also been augmented to receive a new argument: `Extractor`.
`Extractors` do the work of picking the token string out of a request. The interface is simple and composable.
fmt.Printf("Token for user %v expires %v",claims["user"],claims["exp"])
}
```
There are several concrete `Extractor` types provided for your convenience:
*`HeaderExtractor` will search a list of headers until one contains content.
*`ArgumentExtractor` will search a list of keys in request query and form arguments until one contains content.
*`MultiExtractor` will try a list of `Extractors` in order until one returns content.
*`AuthorizationHeaderExtractor` will look in the `Authorization` header for a `Bearer` token.
*`OAuth2Extractor` searches the places an OAuth2 token would be specified (per the spec): `Authorization` header and `access_token` argument
*`PostExtractionFilter` wraps an `Extractor`, allowing you to process the content before it's parsed. A simple example is stripping the `Bearer ` text from a header
### RSA signing methods no longer accept `[]byte` keys
Due to a [critical vulnerability](https://auth0.com/blog/2015/03/31/critical-vulnerabilities-in-json-web-token-libraries/), we've decided the convenience of accepting `[]byte` instead of `rsa.PublicKey` or `rsa.PrivateKey` isn't worth the risk of misuse.
To replace this behavior, we've added two helper methods: `ParseRSAPrivateKeyFromPEM(key []byte) (*rsa.PrivateKey, error)` and `ParseRSAPublicKeyFromPEM(key []byte) (*rsa.PublicKey, error)`. These are just simple helpers for unpacking PEM encoded PKCS1 and PKCS8 keys. If your keys are encoded any other way, all you need to do is convert them to the `crypto/rsa` package's types.
```go
funckeyLookupFunc(*Token)(interface{},error){
// Don't forget to validate the alg is what you expect:
A [go](http://www.golang.org)(or'golang' for search engine friendliness) implementation of [JSON Web Tokens](http://self-issued.info/docs/draft-ietf-oauth-json-web-token.html)
**BREAKING CHANGES:*** Version 3.0.0 is here. It includes _a lot_ of changes including a few that break the API. We've tried to break as few things as possible, so there should just be a few type signature changes. A full list of breaking changes is available in `VERSION_HISTORY.md`. See `MIGRATION_GUIDE.md` for more information on updating your code.
**NOTICE:** It's important that you [validate the `alg` presented is what you expect](https://auth0.com/blog/2015/03/31/critical-vulnerabilities-in-json-web-token-libraries/). This library attempts to make it easy to do the right thing by requiring key types match the expected alg, but you should take the extra step to verify it in your usage. See the examples provided.
## What the heck is a JWT?
JWT.io has [a great introduction](https://jwt.io/introduction) to JSON Web Tokens.
In short, it's a signed JSON object that does something useful (for example, authentication). It's commonly used for `Bearer` tokens in Oauth 2. A token is made of three parts, separated by `.`'s. The first two parts are JSON objects, that have been [base64url](http://tools.ietf.org/html/rfc4648) encoded. The last part is the signature, encoded the same way.
The first part is called the header. It contains the necessary information for verifying the last part, the signature. For example, which encryption method was used for signing and what key was used.
The part in the middle is the interesting bit. It's called the Claims and contains the actual stuff you care about. Refer to [the RFC](http://self-issued.info/docs/draft-jones-json-web-token.html) for information about reserved keys and the proper way to add your own.
## What's in the box?
This library supports the parsing and verification as well as the generation and signing of JWTs. Current supported signing algorithms are HMAC SHA, RSA, RSA-PSS, and ECDSA, though hooks are present for adding your own.
## Examples
See [the project documentation](https://godoc.org/github.com/dgrijalva/jwt-go) for examples of usage:
*[Simple example of parsing and validating a token](https://godoc.org/github.com/dgrijalva/jwt-go#example-Parse--Hmac)
*[Simple example of building and signing a token](https://godoc.org/github.com/dgrijalva/jwt-go#example-New--Hmac)
*[Directory of Examples](https://godoc.org/github.com/dgrijalva/jwt-go#pkg-examples)
## Extensions
This library publishes all the necessary components for adding your own signing methods. Simply implement the `SigningMethod` interface and register a factory method using `RegisterSigningMethod`.
Here's an example of an extension that integrates with the Google App Engine signing tools: https://github.com/someone1/gcp-jwt-go
## Compliance
This library was last reviewed to comply with [RTF 7519](http://www.rfc-editor.org/info/rfc7519) dated May 2015 with a few notable differences:
* In order to protect against accidental use of [Unsecured JWTs](http://self-issued.info/docs/draft-ietf-oauth-json-web-token.html#UnsecuredJWT), tokens using `alg=none` will only be accepted if the constant `jwt.UnsafeAllowNoneSignatureType` is provided as the key.
## Project Status & Versioning
This library is considered production ready. Feedback and feature requests are appreciated. The API should be considered stable. There should be very few backwards-incompatible changes outside of major version updates (and only with good reason).
This project uses [Semantic Versioning 2.0.0](http://semver.org). Accepted pull requests will land on `master`. Periodically, versions will be tagged from `master`. You can find all the releases on [the project releases page](https://github.com/dgrijalva/jwt-go/releases).
While we try to make it obvious when we make breaking changes, there isn't a great mechanism for pushing announcements out to users. You may want to use this alternative package include: `gopkg.in/dgrijalva/jwt-go.v2`. It will do the right thing WRT semantic versioning.
## Usage Tips
### Signing vs Encryption
A token is simply a JSON object that is signed by its author. this tells you exactly two things about the data:
* The author of the token was in the possession of the signing secret
* The data has not been modified since it was signed
It's important to know that JWT does not provide encryption, which means anyone who has access to the token can read its contents. If you need to protect (encrypt) the data, there is a companion spec, `JWE`, that provides this functionality. JWE is currently outside the scope of this library.
### Choosing a Signing Method
There are several signing methods available, and you should probably take the time to learn about the various options before choosing one. The principal design decision is most likely going to be symmetric vs asymmetric.
Symmetric signing methods, such as HSA, use only a single secret. This is probably the simplest signing method to use since any `[]byte` can be used as a valid secret. They are also slightly computationally faster to use, though this rarely is enough to matter. Symmetric signing methods work the best when both producers and consumers of tokens are trusted, or even the same system. Since the same secret is used to both sign and validate tokens, you can't easily distribute the key for validation.
Asymmetric signing methods, such as RSA, use different keys for signing and verifying tokens. This makes it possible to produce tokens with a private key, and allow any consumer to access the public key for verification.
### JWT and OAuth
It's worth mentioning that OAuth and JWT are not the same thing. A JWT token is simply a signed JSON object. It can be used anywhere such a thing is useful. There is some confusion, though, as JWT is the most common type of bearer token used in OAuth2 authentication.
Without going too far down the rabbit hole, here's a description of the interaction of these technologies:
* OAuth is a protocol for allowing an identity provider to be separate from the service a user is logging in to. For example, whenever you use Facebook to log into a different service (Yelp, Spotify, etc), you are using OAuth.
* OAuth defines several options for passing around authentication data. One popular method is called a "bearer token". A bearer token is simply a string that _should_ only be held by an authenticated user. Thus, simply presenting this token proves your identity. You can probably derive from here why a JWT might make a good bearer token.
* Because bearer tokens are used for authentication, it's important they're kept secret. This is why transactions that use bearer tokens typically happen over SSL.
## More
Documentation can be found [on godoc.org](http://godoc.org/github.com/dgrijalva/jwt-go).
The command line utility included in this project (cmd/jwt) provides a straightforward example of token creation and parsing as well as a useful tool for debugging your own integration. You'll also find several implementation examples in the documentation.
