Skip to content

Latest commit

 

History

History
1642 lines (1158 loc) · 63.5 KB

README.md

File metadata and controls

1642 lines (1158 loc) · 63.5 KB
Raw

Build Status Coverage Status Gitter

Java JWT: JSON Web Token for Java and Android

JJWT aims to be the easiest to use and understand library for creating and verifying JSON Web Tokens (JWTs) on the JVM and Android.

JJWT is a pure Java implementation based exclusively on the JWT, JWS, JWE, JWK and JWA RFC specifications and open source under the terms of the Apache 2.0 License.

The library was created by Okta's Senior Architect, Les Hazlewood and is supported and maintained by a community of contributors.

Okta is a complete authentication and user management API for developers.

We've also added some convenience extensions that are not part of the specification, such as JWS compression and claim enforcement.

Table of Contents

Features

  • Fully functional on all JDKs and Android

  • Automatic security best practices and assertions

  • Easy to learn and read API

  • Convenient and readable fluent interfaces, great for IDE auto-completion to write code quickly

  • Fully RFC specification compliant on all implemented functionality, tested against RFC-specified test vectors

  • Stable implementation with enforced 100% test code coverage. Literally every single method, statement and conditional branch variant in the entire codebase is tested and required to pass on every build.

  • Creating, parsing and verifying digitally signed compact JWTs (aka JWSs) with all standard JWS algorithms:

    • HS256: HMAC using SHA-256
    • HS384: HMAC using SHA-384
    • HS512: HMAC using SHA-512
    • ES256: ECDSA using P-256 and SHA-256
    • ES384: ECDSA using P-384 and SHA-384
    • ES512: ECDSA using P-521 and SHA-512
    • RS256: RSASSA-PKCS-v1_5 using SHA-256
    • RS384: RSASSA-PKCS-v1_5 using SHA-384
    • RS512: RSASSA-PKCS-v1_5 using SHA-512
    • PS256: RSASSA-PSS using SHA-256 and MGF1 with SHA-2561
    • PS384: RSASSA-PSS using SHA-384 and MGF1 with SHA-3841
    • PS512: RSASSA-PSS using SHA-512 and MGF1 with SHA-5121

    1. Requires JDK 11 or a compatible JCA Provider (like BouncyCastle) in the runtime classpath.

  • Convenience enhancements beyond the specification such as

    • Body compression for any large JWT, not just JWEs
    • Claims assertions (requiring specific values)
    • Claim POJO marshaling and unmarshaling when using a compatible JSON parser (e.g. Jackson)
    • Secure Key generation based on desired JWA algorithms
    • and more...

Currently Unsupported Features

  • Non-compact serialization and parsing.
  • JWE (Encryption for JWT)

These features will be implemented in a future release. Community contributions are welcome!

Community

Getting Help

If you have trouble using JJWT, please first read the documentation on this page before asking questions. We try very hard to ensure JJWT's documentation is robust, categorized with a table of contents, and up to date for each release.

Questions

If the documentation or the API JavaDoc isn't sufficient, and you either have usability questions or are confused about something, please ask your question here.

After asking your question, you may wish to join our Slack or Gittr chat rooms, but note that they may not always be attended. You will usually have a better chance of having your question answered by asking your question here.

If you believe you have found a bug or would like to suggest a feature enhancement, please create a new GitHub issue, however:

Please do not create a GitHub issue to ask a question.

We use GitHub Issues to track actionable work that requires changes to JJWT's design and/or codebase. If you have a usability question, instead please ask your question here, or try Slack or Gittr as described above.

If a GitHub Issue is created that does not represent actionable work for JJWT's codebase, it will be promptly closed.

Bugs and Feature Requests

If you do not have a usability question and believe you have a legitimate bug or feature request, please do create a new JJWT issue.

If you feel like you'd like to help fix a bug or implement the new feature yourself, please read the Contributing section next before starting any work.

Contributing

Pull Requests

Simple Pull Requests that fix anything other than JJWT core code (documentation, JavaDoc, typos, test cases, etc) are always appreciated and have a high likelihood of being merged quickly. Please send them!

However, if you want or feel the need to change JJWT's functionality or core code, please do not issue a pull request without creating a new JJWT issue and discussing your desired changes first, before you start working on it.

It would be a shame to reject your earnest and genuinely appreciated pull request if it might not not align with the project's goals, design expectations or planned functionality. We've sadly had to reject large PRs in the past because they were out of sync with project or design expectations - all because the PR author didn't first check in with the team first before working on a solution.

So, please create a new JJWT issue first to discuss, and then we can see if (or how) a PR is warranted. Thank you!

Help Wanted

If you would like to help, but don't know where to start, please visit the Help Wanted Issues page and pick any of the ones there, and we'll be happy to discuss and answer questions in the issue comments.

If any of those don't appeal to you, no worries! Any help you would like to offer would be appreciated based on the above caveats concerning contributing pull reqeuests. Feel free to discuss or ask questions first if you're not sure. :)

What is a JSON Web Token?

Don't know what a JSON Web Token is? Read on. Otherwise, jump on down to the Installation section.

JWT is a means of transmitting information between two parties in a compact, verifiable form.

The bits of information encoded in the body of a JWT are called claims. The expanded form of the JWT is in a JSON format, so each claim is a key in the JSON object.

JWTs can be cryptographically signed (making it a JWS) or encrypted (making it a JWE).

This adds a powerful layer of verifiability to the user of JWTs. The receiver has a high degree of confidence that the JWT has not been tampered with by verifying the signature, for instance.

The compact representation of a signed JWT is a string that has three parts, each separated by a .:

eyJhbGciOiJIUzI1NiJ9.eyJzdWIiOiJKb2UifQ.ipevRNuRP6HflG8cFKnmUPtypruRC4fb1DWtoLL62SY

Each part is Base64URL-encoded. The first part is the header, which at a minimum needs to specify the algorithm used to sign the JWT. The second part is the body. This part has all the claims of this JWT encoded in it. The final part is the signature. It's computed by passing a combination of the header and body through the algorithm specified in the header.

If you pass the first two parts through a base 64 url decoder, you'll get the following (formatting added for clarity):

header

{
  "alg": "HS256"
}

body

{
  "sub": "Joe"
}

In this case, the information we have is that the HMAC using SHA-256 algorithm was used to sign the JWT. And, the body has a single claim, sub with value Joe.

There are a number of standard claims, called Registered Claims, in the specification and sub (for subject) is one of them.

To compute the signature, you need a secret key to sign it. We'll cover keys and algorithms later.

Installation

Use your favorite Maven-compatible build tool to pull the dependencies from Maven Central.

The dependencies could differ slightly if you are working with a JDK project or an Android project.

JDK Projects

If you're building a (non-Android) JDK project, you will want to define the following dependencies:

Maven

<dependency>
    <groupId>io.jsonwebtoken</groupId>
    <artifactId>jjwt-api</artifactId>
    <version>0.11.2</version>
</dependency>
<dependency>
    <groupId>io.jsonwebtoken</groupId>
    <artifactId>jjwt-impl</artifactId>
    <version>0.11.2</version>
    <scope>runtime</scope>
</dependency>
<dependency>
    <groupId>io.jsonwebtoken</groupId>
    <artifactId>jjwt-jackson</artifactId> <!-- or jjwt-gson if Gson is preferred -->
    <version>0.11.2</version>
    <scope>runtime</scope>
</dependency>
<!-- Uncomment this next dependency if you are using JDK 10 or earlier and you also want to use 
     RSASSA-PSS (PS256, PS384, PS512) algorithms.  JDK 11 or later does not require it for those algorithms:
<dependency>
    <groupId>org.bouncycastle</groupId>
    <artifactId>bcprov-jdk15on</artifactId>
    <version>1.60</version>
    <scope>runtime</scope>
</dependency>
-->

Gradle

dependencies {
    compile 'io.jsonwebtoken:jjwt-api:0.11.2'
    runtime 'io.jsonwebtoken:jjwt-impl:0.11.2',
            // Uncomment the next line if you want to use RSASSA-PSS (PS256, PS384, PS512) algorithms:
            //'org.bouncycastle:bcprov-jdk15on:1.60',
            'io.jsonwebtoken:jjwt-jackson:0.11.2' // or 'io.jsonwebtoken:jjwt-gson:0.11.2' for gson
}

Android Projects

Android projects will want to define the following dependencies and Proguard exclusions:

Dependencies

Add the dependencies to your project:

dependencies {
    api 'io.jsonwebtoken:jjwt-api:0.11.2'
    runtimeOnly 'io.jsonwebtoken:jjwt-impl:0.11.2' 
    runtimeOnly('io.jsonwebtoken:jjwt-orgjson:0.11.2') {
        exclude group: 'org.json', module: 'json' //provided by Android natively
    }
    // Uncomment the next line if you want to use RSASSA-PSS (PS256, PS384, PS512) algorithms:
    //runtimeOnly 'org.bouncycastle:bcprov-jdk15on:1.60'
}

Proguard

You can use the following Android Proguard exclusion rules:

-keepattributes InnerClasses

-keep class io.jsonwebtoken.** { *; }
-keepnames class io.jsonwebtoken.* { *; }
-keepnames interface io.jsonwebtoken.* { *; }

-keep class org.bouncycastle.** { *; }
-keepnames class org.bouncycastle.** { *; }
-dontwarn org.bouncycastle.**

Understanding JJWT Dependencies

Notice the above dependency declarations all have only one compile-time dependency and the rest are declared as runtime dependencies.

This is because JJWT is designed so you only depend on the APIs that are explicitly designed for you to use in your applications and all other internal implementation details - that can change without warning - are relegated to runtime-only dependencies. This is an extremely important point if you want to ensure stable JJWT usage and upgrades over time:

JJWT guarantees semantic versioning compatibility for all of its artifacts except the jjwt-impl .jar. No such guarantee is made for the jjwt-impl .jar and internal changes in that .jar can happen at any time. Never add the jjwt-impl .jar to your project with compile scope - always declare it with runtime scope.

This is done to benefit you: great care goes into curating the jjwt-api .jar and ensuring it contains what you need and remains backwards compatible as much as is possible so you can depend on that safely with compile scope. The runtime jjwt-impl .jar strategy affords the JJWT developers the flexibility to change the internal packages and implementations whenever and however necessary. This helps us implement features, fix bugs, and ship new releases to you more quickly and efficiently.

Quickstart

Most complexity is hidden behind a convenient and readable builder-based fluent interface, great for relying on IDE auto-completion to write code quickly. Here's an example:

import io.jsonwebtoken.Jwts;
import io.jsonwebtoken.SignatureAlgorithm;
import io.jsonwebtoken.security.Keys;
import java.security.Key;

// We need a signing key, so we'll create one just for this example. Usually
// the key would be read from your application configuration instead.
Key key = Keys.secretKeyFor(SignatureAlgorithm.HS256);

String jws = Jwts.builder().setSubject("Joe").signWith(key).compact();

How easy was that!?

In this case, we are:

  1. building a JWT that will have the registered claim sub (subject) set to Joe. We are then
  2. signing the JWT using a key suitable for the HMAC-SHA-256 algorithm. Finally, we are
  3. compacting it into its final String form. A signed JWT is called a 'JWS'.

The resultant jws String looks like this:

eyJhbGciOiJIUzI1NiJ9.eyJzdWIiOiJKb2UifQ.1KP0SsvENi7Uz1oQc07aXTL7kpQG5jBNIybqr60AlD4

Now let's verify the JWT (you should always discard JWTs that don't match an expected signature):

assert Jwts.parserBuilder().setSigningKey(key).build().parseClaimsJws(jws).getBody().getSubject().equals("Joe");

NOTE: Ensure you call the parseClaimsJws method (since there are many similar methods available). You will get an UnsupportedJwtException if you parse your JWT with wrong method.

There are two things going on here. The key from before is being used to validate the signature of the JWT. If it fails to verify the JWT, a SignatureException (which extends from JwtException) is thrown. Assuming the JWT is validated, we parse out the claims and assert that that subject is set to Joe.

You have to love code one-liners that pack a punch!

But what if parsing or signature validation failed? You can catch JwtException and react accordingly:

try {

    Jwts.parserBuilder().setSigningKey(key).build().parseClaimsJws(compactJws);

    //OK, we can trust this JWT

} catch (JwtException e) {

    //don't trust the JWT!
}

Signed JWTs

The JWT specification provides for the ability to cryptographically sign a JWT. Signing a JWT:

  1. guarantees the JWT was created by someone we know (it is authentic) as well as
  2. guarantees that no-one has manipulated or changed the JWT after it was created (its integrity is maintained).

These two properties - authenticity and integrity - assure us that a JWT contains information we can trust. If a JWT fails authenticity or integrity checks, we should always reject that JWT because we can't trust it.

So how is a JWT signed? Let's walk through it with some easy-to-read pseudocode:

  1. Assume we have a JWT with a JSON header and body (aka 'Claims') as follows:

    header

    {
  "alg": "HS256"
}

    body

    {
  "sub": "Joe"
}

  2. Remove all unnecessary whitespace in the JSON:

    String header = '{"alg":"HS256"}'
String claims = '{"sub":"Joe"}'

  3. Get the UTF-8 bytes and Base64URL-encode each:

    String encodedHeader = base64URLEncode( header.getBytes("UTF-8") )
String encodedClaims = base64URLEncode( claims.getBytes("UTF-8") )

  4. Concatenate the encoded header and claims with a period character between them:

    String concatenated = encodedHeader + '.' + encodedClaims

  5. Use a sufficiently-strong cryptographic secret or private key, along with a signing algorithm of your choice (we'll use HMAC-SHA-256 here), and sign the concatenated string:

    Key key = getMySecretKey()
byte[] signature = hmacSha256( concatenated, key )

  6. Because signatures are always byte arrays, Base64URL-encode the signature and append a period character '.' and it to the concatenated string:

    String jws = concatenated + '.' + base64URLEncode( signature )

And there you have it, the final jws String looks like this:

eyJhbGciOiJIUzI1NiJ9.eyJzdWIiOiJKb2UifQ.1KP0SsvENi7Uz1oQc07aXTL7kpQG5jBNIybqr60AlD4

This is called a 'JWS' - short for signed JWT.

Of course, no one would want to do this manually in code, and worse, if you get anything wrong, you could cause security problems or weaknesses. As a result, JJWT was created to handle all of this for you: JJWT completely automates both the creation of JWSs as well as the parsing and verification of JWSs for you.

But before we dig in to showing you how to create a JWS using JJWT, let's briefly discuss Signature Algorithms and Keys, specifically as they relate to the JWT specifications. Understanding them is critical to being able to create a JWS properly.

Signature Algorithms Keys

The JWT specification identifies 12 standard signature algorithms - 3 secret key algorithms and 9 asymmetric key algorithms - identified by the following names:

  • HS256: HMAC using SHA-256
  • HS384: HMAC using SHA-384
  • HS512: HMAC using SHA-512
  • ES256: ECDSA using P-256 and SHA-256
  • ES384: ECDSA using P-384 and SHA-384
  • ES512: ECDSA using P-521 and SHA-512
  • RS256: RSASSA-PKCS-v1_5 using SHA-256
  • RS384: RSASSA-PKCS-v1_5 using SHA-384
  • RS512: RSASSA-PKCS-v1_5 using SHA-512
  • PS256: RSASSA-PSS using SHA-256 and MGF1 with SHA-256
  • PS384: RSASSA-PSS using SHA-384 and MGF1 with SHA-384
  • PS512: RSASSA-PSS using SHA-512 and MGF1 with SHA-512

These are all represented in the io.jsonwebtoken.SignatureAlgorithm enum.

What's really important about these algorithms - other than their security properties - is that the JWT specification RFC 7518, Sections 3.2 through 3.5 requires (mandates) that you MUST use keys that are sufficiently strong for a chosen algorithm.

This means that JJWT - a specification-compliant library - will also enforce that you use sufficiently strong keys for the algorithms you choose. If you provide a weak key for a given algorithm, JJWT will reject it and throw an exception.

This is not because we want to make your life difficult, we promise! The reason why the JWT specification, and consequently JJWT, mandates key lengths is that the security model of a particular algorithm can completely break down if you don't adhere to the mandatory key properties of the algorithm, effectively having no security at all. No one wants completely insecure JWTs, right? Neither would we.

So what are the requirements?

HMAC-SHA

JWT HMAC-SHA signature algorithms HS256, HS384, and HS512 require a secret key that is at least as many bits as the algorithm's signature (digest) length per RFC 7512 Section 3.2. This means:

  • HS256 is HMAC-SHA-256, and that produces digests that are 256 bits (32 bytes) long, so HS256 requires that you use a secret key that is at least 32 bytes long.

  • HS384 is HMAC-SHA-384, and that produces digests that are 384 bits (48 bytes) long, so HS384 requires that you use a secret key that is at least 48 bytes long.

  • HS512 is HMAC-SHA-512, and that produces digests that are 512 bits (64 bytes) long, so HS512 requires that you use a secret key that is at least 64 bytes long.

RSA

JWT RSA signature algorithms RS256, RS384, RS512, PS256, PS384 and PS512 all require a minimum key length (aka an RSA modulus bit length) of 2048 bits per RFC 7512 Sections 3.3 and 3.5. Anything smaller than this (such as 1024 bits) will be rejected with an InvalidKeyException.

That said, in keeping with best practices and increasing key lengths for security longevity, JJWT recoommends that you use:

  • at least 2048 bit keys with RS256 and PS256
  • at least 3072 bit keys with RS384 and PS384
  • at least 4096 bit keys with RS512 and PS512

These are only JJWT suggestions and not requirements. JJWT only enforces JWT specification requirements and for any RSA key, the requirement is the RSA key (modulus) length in bits MUST be >= 2048 bits.

Elliptic Curve

JWT Elliptic Curve signature algorithms ES256, ES384, and ES512 all require a minimum key length (aka an Elliptic Curve order bit length) that is at least as many bits as the algorithm signature's individual R and S components per RFC 7512 Section 3.4. This means:

  • ES256 requires that you use a private key that is at least 256 bits (32 bytes) long.

  • ES384 requires that you use a private key that is at least 384 bits (48 bytes) long.

  • ES512 requires that you use a private key that is at least 512 bits (64 bytes) long.

Creating Safe Keys

If you don't want to think about bit length requirements or just want to make your life easier, JJWT has provided the io.jsonwebtoken.security.Keys utility class that can generate sufficiently secure keys for any given JWT signature algorithm you might want to use.

Secret Keys

If you want to generate a sufficiently strong SecretKey for use with the JWT HMAC-SHA algorithms, use the Keys.secretKeyFor(SignatureAlgorithm) helper method:

SecretKey key = Keys.secretKeyFor(SignatureAlgorithm.HS256); //or HS384 or HS512

Under the hood, JJWT uses the JCA provider's KeyGenerator to create a secure-random key with the correct minimum length for the given algorithm.

If you need to save this new SecretKey, you can Base64 (or Base64URL) encode it:

String secretString = Encoders.BASE64.encode(key.getEncoded());

Ensure you save the resulting secretString somewhere safe - Base64-encoding is not encryption, so it's still considered sensitive information. You can further encrypt it, etc, before saving to disk (for example).

Asymmetric Keys

If you want to generate sufficiently strong Elliptic Curve or RSA asymmetric key pairs for use with JWT ECDSA or RSA algorithms, use the Keys.keyPairFor(SignatureAlgorithm) helper method:

KeyPair keyPair = Keys.keyPairFor(SignatureAlgorithm.RS256); //or RS384, RS512, PS256, PS384, PS512, ES256, ES384, ES512

You use the private key (keyPair.getPrivate()) to create a JWS and the public key (keyPair.getPublic()) to parse/verify a JWS.

NOTE: The PS256, PS384, and PS512 algorithms require JDK 11 or a compatible JCA Provider (like BouncyCastle) in the runtime classpath. If you are using JDK 10 or earlier and you want to use them, see the Installation section to see how to enable BouncyCastle. All other algorithms are natively supported by the JDK.

Creating a JWS

You create a JWS as follows:

  1. Use the Jwts.builder() method to create a JwtBuilder instance.
  2. Call JwtBuilder methods to add header parameters and claims as desired.
  3. Specify the SecretKey or asymmetric PrivateKey you want to use to sign the JWT.
  4. Finally, call the compact() method to compact and sign, producing the final jws.

For example:

String jws = Jwts.builder() // (1)

    .setSubject("Bob")      // (2) 

    .signWith(key)          // (3)
     
    .compact();             // (4)

Header Parameters

A JWT Header provides metadata about the contents, format and cryptographic operations relevant to the JWT's Claims.

If you need to set one or more JWT header parameters, such as the kid (Key ID) header parameter, you can simply call JwtBuilder setHeaderParam one or more times as needed:

String jws = Jwts.builder()

    .setHeaderParam("kid", "myKeyId")
    
    // ... etc ...

Each time setHeaderParam is called, it simply appends the key-value pair to an internal Header instance, potentially overwriting any existing identically-named key/value pair.

NOTE: You do not need to set the alg or zip header parameters as JJWT will set them automatically depending on the signature algorithm or compression algorithm used.

Header Instance

If you want to specify the entire header at once, you can use the Jwts.header() method and build up the header paramters with it:

Header header = Jwts.header();

populate(header); //implement me

String jws = Jwts.builder()

    .setHeader(header)
    
    // ... etc ...

NOTE: Calling setHeader will overwrite any existing header name/value pairs with the same names that might have already been set. In all cases however, JJWT will still set (and overwrite) any alg and zip headers regardless if those are in the specified header object or not.

Header Map

If you want to specify the entire header at once and you don't want to use Jwts.header(), you can use JwtBuilder setHeader(Map) method instead:

Map<String,Object> header = getMyHeaderMap(); //implement me

String jws = Jwts.builder()

    .setHeader(header)
    
    // ... etc ...

NOTE: Calling setHeader will overwrite any existing header name/value pairs with the same names that might have already been set. In all cases however, JJWT will still set (and overwrite) any alg and zip headers regardless if those are in the specified header object or not.

Claims

Claims are a JWT's 'body' and contain the information that the JWT creator wishes to present to the JWT recipient(s).

Standard Claims

The JwtBuilder provides convenient setter methods for standard registered Claim names defined in the JWT specification. They are:

For example:

String jws = Jwts.builder()

    .setIssuer("me")
    .setSubject("Bob")
    .setAudience("you")
    .setExpiration(expiration) //a java.util.Date
    .setNotBefore(notBefore) //a java.util.Date 
    .setIssuedAt(new Date()) // for example, now
    .setId(UUID.randomUUID()) //just an example id
    
    /// ... etc ...

Custom Claims

If you need to set one or more custom claims that don't match the standard setter method claims shown above, you can simply call JwtBuilder claim one or more times as needed:

String jws = Jwts.builder()

    .claim("hello", "world")
    
    // ... etc ...

Each time claim is called, it simply appends the key-value pair to an internal Claims instance, potentially overwriting any existing identically-named key/value pair.

Obviously, you do not need to call claim for any standard claim name and it is recommended instead to call the standard respective setter method as this enhances readability.

Claims Instance

If you want to specify all claims at once, you can use the Jwts.claims() method and build up the claims with it:

Claims claims = Jwts.claims();

populate(claims); //implement me

String jws = Jwts.builder()

    .setClaims(claims)
    
    // ... etc ...

NOTE: Calling setClaims will overwrite any existing claim name/value pairs with the same names that might have already been set.

Claims Map

If you want to specify all claims at once and you don't want to use Jwts.claims(), you can use JwtBuilder setClaims(Map) method instead:

Map<String,Object> claims = getMyClaimsMap(); //implement me

String jws = Jwts.builder()

    .setClaims(claims)
    
    // ... etc ...

NOTE: Calling setClaims will overwrite any existing claim name/value pairs with the same names that might have already been set.

Signing Key

It is recommended that you specify the signing key by calling call the JwtBuilder's signWith method and let JJWT determine the most secure algorithm allowed for the specified key.:

String jws = Jwts.builder()

   // ... etc ...
   
   .signWith(key) // <---
   
   .compact();

For example, if you call signWith with a SecretKey that is 256 bits (32 bytes) long, it is not strong enough for HS384 or HS512, so JJWT will automatically sign the JWT using HS256.

When using signWith JJWT will also automatically set the required alg header with the associated algorithm identifier.

Similarly, if you called signWith with an RSA PrivateKey that was 4096 bits long, JJWT will use the RS512 algorithm and automatically set the alg header to RS512.

The same selection logic applies for Elliptic Curve PrivateKeys.

NOTE: You cannot sign JWTs with PublicKeys as this is always insecure. JJWT will reject any specified PublicKey for signing with an InvalidKeyException.

SecretKey Formats

If you want to sign a JWS using HMAC-SHA algorithms and you have a secret key String or encoded byte array, you will need to convert it into a SecretKey instance to use as the signWith method argument.

If your secret key is:

  • An encoded byte array:

    SecretKey key = Keys.hmacShaKeyFor(encodedKeyBytes);

  • A Base64-encoded string:

    SecretKey key = Keys.hmacShaKeyFor(Decoders.BASE64.decode(secretString));

  • A Base64URL-encoded string:

    SecretKey key = Keys.hmacShaKeyFor(Decoders.BASE64URL.decode(secretString));

  • A raw (non-encoded) string (e.g. a password String):

    SecretKey key = Keys.hmacShaKeyFor(secretString.getBytes(StandardCharsets.UTF_8));

    It is always incorrect to call secretString.getBytes() (without providing a charset).

    However, raw password strings like this, e.g. correcthorsebatterystaple should be avoided whenever possible because they can inevitably result in weak or susceptible keys. Secure-random keys are almost always stronger. If you are able, prefer creating a new secure-random secret key instead.

SignatureAlgorithm Override

In some specific cases, you might want to override JJWT's default selected algorithm for a given key.

For example, if you have an RSA PrivateKey that is 2048 bits, JJWT would automatically choose the RS256 algorithm. If you wanted to use RS384 or RS512 instead, you could manually specify it with the overloaded signWith method that accepts the SignatureAlgorithm as an additional parameter:

   .signWith(privateKey, SignatureAlgorithm.RS512) // <---
   
   .compact();

This is allowed because the JWT specification allows any RSA algorithm strength for any RSA key >= 2048 bits. JJWT just prefers RS512 for keys >= 4096 bits, followed by RS384 for keys >= 3072 bits and finally RS256 for keys >= 2048 bits.

In all cases however, regardless of your chosen algorithms, JJWT will assert that the specified key is allowed to be used for that algorithm according to the JWT specification requirements.

JWS Compression

If your JWT claims set is large (contains a lot of data), and you are certain that JJWT will also be the same library that reads/parses your JWS, you might want to compress the JWS to reduce its size. Note that this is not a standard feature for JWS and is not likely to be supported by other JWT libraries.

Please see the main Compression section to see how to compress and decompress JWTs.

Reading a JWS

You read (parse) a JWS as follows:

  1. Use the Jwts.parserBuilder() method to create a JwtParserBuilder instance.
  2. Specify the SecretKey or asymmetric PublicKey you want to use to verify the JWS signature.1
  3. Call the build() method on the JwtParserBuilder to return a thread-safe JwtParser.
  4. Finally, call the parseClaimsJws(String) method with your jws String, producing the original JWS.
  5. The entire call is wrapped in a try/catch block in case parsing or signature validation fails. We'll cover exceptions and causes for failure later.

1. If you don't know which key to use at the time of parsing, you can look up the key using a SigningKeyResolver which we'll cover later.

For example:

Jws<Claims> jws;

try {
    jws = Jwts.parserBuilder()  // (1)
    .setSigningKey(key)         // (2)
    .build()                    // (3)
    .parseClaimsJws(jwsString); // (4)
    
    // we can safely trust the JWT
     
catch (JwtException ex) {       // (5)
    
    // we *cannot* use the JWT as intended by its creator
}

NOTE: If you are expecting a JWS, always call JwtParser's parseClaimsJws method (and not one of the other similar methods available) as this guarantees the correct security model for parsing signed JWTs.

Verification Key

The most important thing to do when reading a JWS is to specify the key to use to verify the JWS's cryptographic signature. If signature verification fails, the JWT cannot be safely trusted and should be discarded.

So which key do we use for verification?

  • If the jws was signed with a SecretKey, the same SecretKey should be specified on the JwtParserBuilder. For example:

    Jwts.parserBuilder()
    
  .setSigningKey(secretKey) // <----
  
  .build()
  .parseClaimsJws(jwsString);

  • If the jws was signed with a PrivateKey, that key's corresponding PublicKey (not the PrivateKey) should be specified on the JwtParserBuilder. For example:

    Jwts.parserBuilder()
    
  .setSigningKey(publicKey) // <---- publicKey, not privateKey
  
  .build()
  .parseClaimsJws(jwsString);

But you might have noticed something - what if your application doesn't use just a single SecretKey or KeyPair? What if JWSs can be created with different SecretKeys or public/private keys, or a combination of both? How do you know which key to specify if you can't inspect the JWT first?

In these cases, you can't call the JwtParserBuilder's setSigningKey method with a single key - instead, you'll need to use a SigningKeyResolver, covered next.

Signing Key Resolver

If your application expects JWSs that can be signed with different keys, you won't call the setSigningKey method. Instead, you'll need to implement the SigningKeyResolver interface and specify an instance on the JwtParserBuilder via the setSigningKeyResolver method.
For example:

SigningKeyResolver signingKeyResolver = getMySigningKeyResolver();

Jwts.parserBuilder()

    .setSigningKeyResolver(signingKeyResolver) // <----
    
    .build()
    .parseClaimsJws(jwsString);

You can simplify things a little by extending from the SigningKeyResolverAdapter and implementing the resolveSigningKey(JwsHeader, Claims) method. For example:

public class MySigningKeyResolver extends SigningKeyResolverAdapter {
    
    @Override
    public Key resolveSigningKey(JwsHeader jwsHeader, Claims claims) {
        // implement me
    }
}

The JwtParser will invoke the resolveSigningKey method after parsing the JWS JSON, but before verifying the jws signature. This allows you to inspect the JwsHeader and Claims arguments for any information that can help you look up the Key to use for verifying that specific jws. This is very powerful for applications with more complex security models that might use different keys at different times or for different users or customers.

Which data might you inspect?

The JWT specification's supported way to do this is to set a kid (Key ID) field in the JWS header when the JWS is being created, for example:

Key signingKey = getSigningKey();

String keyId = getKeyId(signingKey); //any mechanism you have to associate a key with an ID is fine

String jws = Jwts.builder()
    
    .setHeaderParam(JwsHeader.KEY_ID, keyId) // 1
    
    .signWith(signingKey)                    // 2
    
    .compact();

Then during parsing, your SigningKeyResolver can inspect the JwsHeader to get the kid and then use that value to look up the key from somewhere, like a database. For example:

public class MySigningKeyResolver extends SigningKeyResolverAdapter {
    
    @Override
    public Key resolveSigningKey(JwsHeader jwsHeader, Claims claims) {
        
        //inspect the header or claims, lookup and return the signing key
        
        String keyId = jwsHeader.getKeyId(); //or any other field that you need to inspect
        
        Key key = lookupVerificationKey(keyId); //implement me
        
        return key;
    }
}

Note that inspecting the jwsHeader.getKeyId() is just the most common approach to look up a key - you could inspect any number of header fields or claims to determine how to lookup the verification key. It is all based on how the JWS was created.

Finally remember that for HMAC algorithms, the returned verification key should be a SecretKey, and for asymmetric algorithms, the key returned should be a PublicKey (not a PrivateKey).

Claim Assertions

You can enforce that the JWS you are parsing conforms to expectations that you require and are important for your application.

For example, let's say that you require that the JWS you are parsing has a specific sub (subject) value, otherwise you may not trust the token. You can do that by using one of the various require* methods on the JwtParserBuilder:

try {
    Jwts.parserBuilder().requireSubject("jsmith").setSigningKey(key).build().parseClaimsJws(s);
} catch(InvalidClaimException ice) {
    // the sub field was missing or did not have a 'jsmith' value
}

If it is important to react to a missing vs an incorrect value, instead of catching InvalidClaimException, you can catch either MissingClaimException or IncorrectClaimException:

try {
    Jwts.parserBuilder().requireSubject("jsmith").setSigningKey(key).build().parseClaimsJws(s);
} catch(MissingClaimException mce) {
    // the parsed JWT did not have the sub field
} catch(IncorrectClaimException ice) {
    // the parsed JWT had a sub field, but its value was not equal to 'jsmith'
}

You can also require custom fields by using the require(fieldName, requiredFieldValue) method - for example:

try {
    Jwts.parserBuilder().require("myfield", "myRequiredValue").setSigningKey(key).build().parseClaimsJws(s);
} catch(InvalidClaimException ice) {
    // the 'myfield' field was missing or did not have a 'myRequiredValue' value
}

(or, again, you could catch either MissingClaimException or IncorrectClaimException instead).

Please see the JwtParserBuilder class and/or JavaDoc for a full list of the various require* methods you may use for claims assertions.

Accounting for Clock Skew

When parsing a JWT, you might find that exp or nbf claim assertions fail (throw exceptions) because the clock on the parsing machine is not perfectly in sync with the clock on the machine that created the JWT. This can cause obvious problems since exp and nbf are time-based assertions, and clock times need to be reliably in sync for shared assertions.

You can account for these differences (usually no more than a few minutes) when parsing using the JwtParserBuilder's setAllowedClockSkewSeconds. For example:

long seconds = 3 * 60; //3 minutes

Jwts.parserBuilder()
    
    .setAllowedClockSkewSeconds(seconds) // <----
    
    // ... etc ...
    .build()
    .parseClaimsJws(jwt);

This ensures that clock differences between the machines can be ignored. Two or three minutes should be more than enough; it would be fairly strange if a production machine's clock was more than 5 minutes difference from most atomic clocks around the world.

Custom Clock Support

If the above setAllowedClockSkewSeconds isn't sufficient for your needs, the timestamps created during parsing for timestamp comparisons can be obtained via a custom time source. Call the JwtParserBuilder's setClock method with an implementation of the io.jsonwebtoken.Clock interface. For example:

Clock clock = new MyClock();

Jwts.parserBuilder().setClock(myClock) //... etc ...

The JwtParser's default Clock implementation simply returns new Date() to reflect the time when parsing occurs, as most would expect. However, supplying your own clock could be useful, especially when writing test cases to guarantee deterministic behavior.

JWS Decompression

If you used JJWT to compress a JWS and you used a custom compression algorithm, you will need to tell the JwtParserBuilder how to resolve your CompressionCodec to decompress the JWT.

Please see the Compression section below to see how to decompress JWTs during parsing.

Compression

The JWT specification only standardizes this feature for JWEs (Encrypted JWTs) and not JWSs (Signed JWTs), however JJWT supports both. If you are positive that a JWS you create with JJWT will also be parsed with JJWT, you can use this feature with JWSs, otherwise it is best to only use it for JWEs.

If a JWT's Claims set is sufficiently large - that is, it contains a lot of name/value pairs, or individual values are very large or verbose - you can reduce the size of the created JWS by compressing the claims body.

This might be important to you if the resulting JWS is used in a URL for example, since URLs are best kept under 4096 characters due to browser, user mail agent, or HTTP gateway compatibility issues. Smaller JWTs also help reduce bandwidth utilization, which may or may not be important depending on your application's volume or needs.

If you want to compress your JWT, you can use the JwtBuilder's compressWith(CompressionAlgorithm) method. For example:

   Jwts.builder()
   
   .compressWith(CompressionCodecs.DEFLATE) // or CompressionCodecs.GZIP
   
   // .. etc ...

If you use the DEFLATE or GZIP Compression Codecs - that's it, you're done. You don't have to do anything during parsing or configure the JwtParser for compression - JJWT will automatically decompress the body as expected.

Custom Compression Codec

If however, you used your own custom compression codec when creating the JWT (via JwtBuilder compressWith), then you need to supply the codec to the JwtParserBuilder using the setCompressionCodecResolver method. For example:

CompressionCodecResolver ccr = new MyCompressionCodecResolver();

Jwts.parserBuilder()

    .setCompressionCodecResolver(ccr) // <----
    
    // .. etc ...

Typically a CompressionCodecResolver implementation will inspect the zip header to find out what algorithm was used and then return a codec instance that supports that algorithm. For example:

public class MyCompressionCodecResolver implements CompressionCodecResolver {
        
    @Override
    public CompressionCodec resolveCompressionCodec(Header header) throws CompressionException {
        
        String alg = header.getCompressionAlgorithm();
            
        CompressionCodec codec = getCompressionCodec(alg); //implement me
            
        return codec;
    }
}

JSON Support

A JwtBuilder will serialize the Header and Claims maps (and potentially any Java objects they contain) to JSON with a Serializer<Map<String, ?>> instance. Similarly, a JwtParser will deserialize JSON into the Header and Claims using a Deserializer<Map<String, ?>> instance.

If you don't explicitly configure a JwtBuilder's Serializer or a JwtParserBuilder's Deserializer, JJWT will automatically attempt to discover and use the following JSON implementations if found in the runtime classpath.
They are checked in order, and the first one found is used:

  1. Jackson: This will automatically be used if you specify io.jsonwebtoken:jjwt-jackson as a project runtime dependency. Jackson supports POJOs as claims with full marshaling/unmarshaling as necessary.

  2. Gson: This will automatically be used if you specify io.jsonwebtoken:jjwt-gson as a project runtime dependency. Gson also supports POJOs as claims with full marshaling/unmarshaling as necessary.

  3. JSON-Java (org.json): This will be used automatically if you specify io.jsonwebtoken:jjwt-orgjson as a project runtime dependency.

    NOTE: org.json APIs are natively enabled in Android environments so this is the recommended JSON processor for Android applications unless you want to use POJOs as claims. The org.json library supports simple Object-to-JSON marshaling, but it does not support JSON-to-Object unmarshalling.

If you want to use POJOs as claim values, use either the io.jsonwebtoken:jjwt-jackson or io.jsonwebtoken:jjwt-gson dependency (or implement your own Serializer and Deserializer if desired). But beware, Jackson will force a sizable (> 1 MB) dependency to an Android application thus increasing the app download size for mobile users.

Custom JSON Processor

If you don't want to use JJWT's runtime dependency approach, or just want to customize how JSON serialization and deserialization works, you can implement the Serializer and Deserializer interfaces and specify instances of them on the JwtBuilder and JwtParserBuilder respectively. For example:

When creating a JWT:

Serializer<Map<String,?>> serializer = getMySerializer(); //implement me

Jwts.builder()

    .serializeToJsonWith(serializer)
    
    // ... etc ...

When reading a JWT:

Deserializer<Map<String,?>> deserializer = getMyDeserializer(); //implement me

Jwts.parserBuilder()

    .deserializeJsonWith(deserializer)
    
    // ... etc ...

Jackson JSON Processor

If you want to use Jackson for JSON processing, just including the io.jsonwebtoken:jjwt-jackson dependency as a runtime dependency is all that is necessary in most projects, since Gradle and Maven will automatically pull in the necessary Jackson dependencies as well.

After including this dependency, JJWT will automatically find the Jackson implementation on the runtime classpath and use it internally for JSON parsing. There is nothing else you need to do - JJWT will automatically create a new Jackson ObjectMapper for its needs as required.

However, if you have an application-wide Jackson ObjectMapper (as is typically recommended for most applications), you can configure JJWT to use your own ObjectMapper instead.

You do this by declaring the io.jsonwebtoken:jjwt-jackson dependency with compile scope (not runtime scope which is the typical JJWT default). That is:

Maven

<dependency>
    <groupId>io.jsonwebtoken</groupId>
    <artifactId>jjwt-jackson</artifactId>
    <version>0.11.2</version>
    <scope>compile</scope> <!-- Not runtime -->
</dependency>

Gradle or Android

dependencies {
    compile 'io.jsonwebtoken:jjwt-jackson:0.11.2'
}

And then you can specify the JacksonSerializer using your own ObjectMapper on the JwtBuilder:

ObjectMapper objectMapper = getMyObjectMapper(); //implement me

String jws = Jwts.builder()

    .serializeToJsonWith(new JacksonSerializer(objectMapper))
    
    // ... etc ...

and the JacksonDeserializer using your ObjectMapper on the JwtParserBuilder:

ObjectMapper objectMapper = getMyObjectMapper(); //implement me

Jwts.parserBuilder()

    .deserializeJsonWith(new JacksonDeserializer(objectMapper))
    
    // ... etc ...

Parsing of Custom Claim Types

By default JJWT will only convert simple claim types: String, Date, Long, Integer, Short and Byte. If you need to deserialize other types you can configure the JacksonDeserializer by passing a Map of claim names to types in through a constructor. For example:

new JacksonDeserializer(Maps.of("user", User.class).build())

This would trigger the value in the user claim to be deserialized into the custom type of User. Given the claims body of:

{
    "issuer": "https://example.com/issuer",
    "user": {
      "firstName": "Jill",
      "lastName": "Coder"
    }
}

The User object could be retrieved from the user claim with the following code:

Jwts.parserBuilder()

    .deserializeJsonWith(new JacksonDeserializer(Maps.of("user", User.class).build())) // <-----

    .build()

    .parseClaimsJwt(aJwtString)

    .getBody()
    
    .get("user", User.class) // <-----

NOTE: Using this constructor is mutually exclusive with the JacksonDeserializer(ObjectMapper) constructor described above. This is because JJWT configures an ObjectMapper directly and could have negative consequences for a shared ObjectMapper instance. This should work for most applications, if you need a more advanced parsing options, configure the mapper directly.

Gson JSON Processor

If you want to use Gson for JSON processing, just including the io.jsonwebtoken:jjwt-gson dependency as a runtime dependency is all that is necessary in most projects, since Gradle and Maven will automatically pull in the necessary Gson dependencies as well.

After including this dependency, JJWT will automatically find the Gson implementation on the runtime classpath and use it internally for JSON parsing. There is nothing else you need to do - just declaring the dependency is all that is required, no code or config is necessary.

If you're curious, JJWT will automatically create an internal default Gson instance for its own needs as follows:

new GsonBuilder().disableHtmlEscaping().create();

However, if you prefer to use a different Gson instance instead of JJWT's default, you can configure JJWT to use your own.

You do this by declaring the io.jsonwebtoken:jjwt-gson dependency with compile scope (not runtime scope which is the typical JJWT default). That is:

Maven

<dependency>
    <groupId>io.jsonwebtoken</groupId>
    <artifactId>jjwt-gson</artifactId>
    <version>0.11.2</version>
    <scope>compile</scope> <!-- Not runtime -->
</dependency>

Gradle or Android

dependencies {
    compile 'io.jsonwebtoken:jjwt-gson:0.11.2'
}

And then you can specify the GsonSerializer using your own Gson instance on the JwtBuilder:

Gson gson = getGson(); //implement me

String jws = Jwts.builder()

    .serializeToJsonWith(new GsonSerializer(gson))
    
    // ... etc ...

and the GsonDeserializer using your Gson instance on the JwtParser:

Gson gson = getGson(); //implement me

Jwts.parser()

    .deserializeJsonWith(new GsonDeserializer(gson))
    
    // ... etc ...

Base64 Support

JJWT uses a very fast pure-Java Base64 codec for Base64 and Base64Url encoding and decoding that is guaranteed to work deterministically in all JDK and Android environments.

You can access JJWT's encoders and decoders using the io.jsonwebtoken.io.Encoders and io.jsonwebtoken.io.Decoders utility classes.

io.jsonwebtoken.io.Encoders:

  • BASE64 is an RFC 4648 Base64 encoder
  • BASE64URL is an RFC 4648 Base64URL encoder

io.jsonwebtoken.io.Decoders:

  • BASE64 is an RFC 4648 Base64 decoder
  • BASE64URL is an RFC 4648 Base64URL decoder

Understanding Base64 in Security Contexts

All cryptographic operations, like encryption and message digest calculations, result in binary data - raw byte arrays.

Because raw byte arrays cannot be represented natively in JSON, the JWT specifications employ the Base64URL encoding scheme to represent these raw byte values in JSON documents or compound structures like a JWT.

This means that the Base64 and Base64URL algorithms take a raw byte array and converts the bytes into a string suitable to use in text documents and protocols like HTTP. These algorithms can also convert these strings back into the original raw byte arrays for decryption or signature verification as necessary.

That's nice and convenient, but there are two very important properties of Base64 (and Base64URL) text strings that are critical to remember when they are used in security scenarios like with JWTs:

Base64 is not encryption

Base64-encoded text is not encrypted.

While a byte array representation can be converted to text with the Base64 algorithms, anyone in the world can take Base64-encoded text, decode it with any standard Base64 decoder, and obtain the underlying raw byte array data. No key or secret is required to decode Base64 text - anyone can do it.

Based on this, when encoding sensitive byte data with Base64 - like a shared or private key - the resulting string NOT is safe to expose publicly.

A base64-encoded key is still sensitive information and must be kept as secret and as safe as the original thing you got the bytes from (e.g. a Java PrivateKey or SecretKey instance).

After Base64-encoding data into a string, it is possible to then encrypt the string to keep it safe from prying eyes if desired, but this is different. Encryption is not encoding. They are separate concepts.

Changing Base64 Characters

In an effort to see if signatures or encryption is truly validated correctly, some try to edit a JWT string - particularly the Base64-encoded signature part - to see if the edited string fails security validations.

This conceptually makes sense: change the signature string, you would assume that signature validation would fail.

But this doesn't always work. Changing base64 characters is an invalid test.

Why?

Because of the way the Base64 algorithm works, there are multiple Base64 strings that can represent the same raw byte array.

Going into the details of the Base64 algorithm is out of scope for this documentation, but there are many good Stackoverflow answers and JJWT issue comments that explain this in detail.
Here's one good answer:

Remember that Base64 encodes each 8 bit entity into 6 bit chars. The resulting string then needs exactly 11 * 8 / 6 bytes, or 14 2/3 chars. But you can't write partial characters. Only the first 4 bits (or 2/3 of the last char) are significant. The last two bits are not decoded. Thus all of:

dGVzdCBzdHJpbmo
dGVzdCBzdHJpbmp
dGVzdCBzdHJpbmq
dGVzdCBzdHJpbmr

All decode to the same 11 bytes (116, 101, 115, 116, 32, 115, 116, 114, 105, 110, 106).

As you can see by the above 4 examples, they all decode to the same exact 11 bytes. So just changing one or two characters at the end of a Base64 string may not work and can often result in an invalid test.

Adding Invalid Characters

JJWT's default Base64/Base64URL decoders automatically ignore illegal Base64 characters located in the beginning and end of an encoded string. Therefore prepending or appending invalid characters like { or ] or similar will also not fail JJWT's signature checks either. Why?

Because such edits - whether changing a trailing character or two, or appending invalid characters - do not actually change the real signature, which in cryptographic contexts, is always a byte array. Instead, tests like these change a text encoding of the byte array, and as we covered above, they are different things.

So JJWT 'cares' more about the real byte array and less about its text encoding because that is what actually matters in cryptographic operations. In this sense, JJWT follows the Robustness Principle in being slightly lenient on what is accepted per the rules of Base64, but if anything in the real underlying byte array is changed, then yes, JJWT's cryptographic assertions will definitely fail.

To help understand JJWT's approach, we have to remember why signatures exist. From our documentation above on signing JWTs:

  • guarantees it was created by someone we know (it is authentic), as well as
  • guarantees that no-one has manipulated or changed it after it was created (its integrity is maintained).

Just prepending or appending invalid text to try to 'trick' the algorithm doesn't change the integrity of the underlying claims or signature byte arrays, nor the authenticity of the claims byte array, because those byte arrays are still obtained intact.

Please see JJWT Issue #518 and its referenced issues and links for more information.

Custom Base64

If for some reason you want to specify your own Base64Url encoder and decoder, you can use the JwtBuilder base64UrlEncodeWith method to set the encoder:

Encoder<byte[], String> base64UrlEncoder = getMyBase64UrlEncoder(); //implement me

String jws = Jwts.builder()

    .base64UrlEncodeWith(base64UrlEncoder)
    
    // ... etc ...

and the JwtParserBuilder's base64UrlDecodeWith method to set the decoder:

Decoder<String, byte[]> base64UrlDecoder = getMyBase64UrlDecoder(); //implement me

Jwts.parserBuilder()

    .base64UrlDecodeWith(base64UrlEncoder)
    
    // ... etc ...

Learn More

Author

Maintained by Les Hazlewood & Okta

License

This project is open-source via the Apache 2.0 License.