uPortal
uPortal copied to clipboard
fix(deps): update json web token to v0.12.6
This PR contains the following updates:
Package | Change | Age | Adoption | Passing | Confidence |
---|---|---|---|---|---|
io.jsonwebtoken:jjwt-jackson | 0.11.5 -> 0.12.6 |
||||
io.jsonwebtoken:jjwt-impl | 0.11.5 -> 0.12.6 |
||||
io.jsonwebtoken:jjwt-api | 0.11.5 -> 0.12.6 |
[!WARNING] Some dependencies could not be looked up. Check the Dependency Dashboard for more information.
Release Notes
jwtk/jjwt (io.jsonwebtoken:jjwt-jackson)
v0.12.6
This patch release:
- Ensures that after successful JWS signature verification, an application-configured Base64Url
Decoder
output is used to construct aJws
instance (instead of JJWT's default decoder). See Issue 947. - Fixes a decompression memory leak in concurrent/multi-threaded environments introduced in 0.12.0 when decompressing JWTs with a
zip
header ofGZIP
. See Issue 949. - Upgrades BouncyCastle to 1.78 via PR 941.
v0.12.5
This patch release:
-
Ensures that builders'
NestedCollection
changes are applied to the collection immediately as mutation methods are called, no longer requiring application developers to call.and()
to 'commit' or apply a change. For example, prior to this release, the following code did not apply changes:JwtBuilder builder = Jwts.builder(); builder.audience().add("an-audience"); // no .and() call builder.compact(); // would not keep 'an-audience'
Now this code works as expected and all other
NestedCollection
instances like it apply changes immediately (e.g. when calling.add(value)
).However, standard fluent builder chains are still recommended for readability when feasible, e.g.
Jwts.builder() .audience().add("an-audience").and() // allows fluent chaining .subject("Joe") // etc... .compact()
See Issue 916.
v0.12.4
This patch release includes various changes listed below.
Jackson Default Parsing Behavior
This release makes two behavioral changes to JJWT's default Jackson ObjectMapper
parsing settings:
-
In the interest of having stronger standards to reject potentially malformed/malicious/accidental JSON that could have undesirable effects on an application, JJWT's default
ObjectMapper
is now configured to explicitly reject/fail parsing JSON (JWT headers and/or Claims) if/when that JSON contains duplicate JSON member names.For example, now the following JSON, if parsed, would fail (be rejected) by default:
{ "hello": "world", "thisWillFail": 42, "thisWillFail": "test" }
Technically, the JWT RFCs do allow duplicate named fields as long as the last parsed member is the one used (see JWS RFC 7515, Section 4), so this is allowed. However, because JWTs often reflect security concepts, it's usually better to be defensive and reject these unexpected scenarios by default. The RFC later supports this position/preference in Section 10.12:
Ambiguous and potentially exploitable situations could arise if the JSON parser used does not enforce the uniqueness of member names or returns an unpredictable value for duplicate member names.
Finally, this is just a default, and the RFC does indeed allow duplicate member names if the last value is used, so applications that require duplicates to be allowed can simply configure their own
ObjectMapper
and use that with JJWT instead of assuming this (new) JJWT default. See Issue #877 for more. -
If using JJWT's support to use Jackson to parse Custom Claim Types (for example, a Claim that should be unmarshalled into a POJO), and the JSON for that POJO contained a member that is not represented in the specified class, Jackson would fail parsing by default. Because POJOs and JSON data models can sometimes be out of sync due to different class versions, the default behavior has been changed to ignore these unknown JSON members instead of failing (i.e. the
ObjectMapper
'sDeserializationFeature.FAIL_ON_UNKNOWN_PROPERTIES
is now set tofalse
) by default.Again, if you prefer the stricter behavior of rejecting JSON with extra or unknown properties, you can configure
true
on your ownObjectMapper
instance and use that instance with theJwts.parser()
builder.
Additional Changes
This release also:
- Fixes a thread-safety issue when using
java.util.ServiceLoader
to dynamically lookup/instantiate pluggable implementations of JJWT interfaces (e.g. JSON parsers, etc). See Issue #873 and its documented fix in PR #893. - Ensures Android environments and older
org.json
library usages can parse JSON from aJwtBuilder
-providedjava.io.Reader
instance. Issue 882. - Ensures a single string
aud
(Audience) claim is retained (without converting it to aSet
) when copying/applying a source Claims instance to a destination Claims builder. Issue 890. - Ensures P-256, P-384 and P-521 Elliptic Curve JWKs zero-pad their field element (
x
,y
, andd
) byte array values if necessary before Base64Url-encoding per RFC 7518, Sections 6.2.1.2, 6.2.1.3, and 6.2.2.1, respectively. Issue 901. - Ensures that Secret JWKs for HMAC-SHA algorithms with
k
sizes larger than the algorithm minimum can be parsed/used as expected. See Issue #905 - Ensures there is an upper bound (maximum) iterations enforced for PBES2 decryption to help mitigate potential DoS attacks. Many thanks to Jingcheng Yang and Jianjun Chen from Sichuan University and Zhongguancun Lab for their work on this. See PR 911.
- Fixes various typos in documentation and JavaDoc. Thanks to those contributing pull requests for these!
v0.12.3
This patch release:
- Upgrades the
org.json
dependency to20231013
to address that library's CVE-2023-5072 vulnerability. - (Re-)enables empty values for custom claims, which was the behavior in <= 0.11.5. Issue 858.
v0.12.2
This is a follow-up release to finalize the work in 0.12.1 that tried to fix a reflection scope problem
on >= JDK 17. The 0.12.1 fix worked, but only if the importing project or application did not have its own
module-info.java
file.
This release removes that reflection code entirely in favor of a JJWT-native implementation, eliminating JPMS
module (scope) problems on >= JDK 17. As such, --add-opens
flags are no longer required to use JJWT.
The fix has been tested up through JDK 21 in a separate application environment (out of JJWT's codebase) to assert
expected functionality in a 'clean room' environment in a project both with and without module-info.java
usage.
v0.12.1
Enabled reflective access on JDK 17+ to java.io.ByteArrayInputStream
and sun.security.util.KeyUtil
for
jjwt-impl.jar
v0.12.0
This is a big release! JJWT now fully supports Encrypted JSON Web Tokens (JWE), JSON Web Keys (JWK) and more! See the sections below enumerating all new features as well as important notes on breaking changes or backwards-incompatible changes made in preparation for the upcoming 1.0 release.
Because breaking changes are being introduced, it is strongly recommended to wait until the upcoming 1.0 release where you can address breaking changes one time only.
Those that need immediate JWE encryption and JWK key support however will likely want to upgrade now and deal with the smaller subset of breaking changes in the 1.0 release.
Simplified Starter Jar
Those upgrading to new modular JJWT versions from old single-jar versions will transparently obtain everything they need in their Maven, Gradle or Android projects.
JJWT's early releases had one and only one .jar: jjwt.jar
. Later releases moved to a modular design with 'api' and
'impl' jars including 'plugin' jars for Jackson, GSON, org.json, etc. Some users upgrading from the earlier single
jar to JJWT's later versions have been frustrated by being forced to learn how to configure the more modular .jars.
This release re-introduces the jjwt.jar
artifact again, but this time it is simply an empty .jar with Maven
metadata that will automatically transitively download the following into a project, retaining the old single-jar
behavior:
-
jjwt-api.jar
-
jjwt-impl.jar
-
jjwt-jackson.jar
Naturally, developers are still encouraged to configure the modular .jars as described in JJWT's documentation for greater control and to enable their preferred JSON parser, but this stop-gap should help those unaware when upgrading.
JSON Web Encryption (JWE) Support!
This has been a long-awaited feature for JJWT, years in the making, and it is quite extensive - so many encryption
algorithms and key management algorithms are defined by the JWA specification, and new API concepts had to be
introduced for all of them, as well as extensive testing with RFC-defined test vectors. The wait is over!
All JWA-defined encryption algorithms and key management algorithms are fully implemented and supported and
available immediately. For example:
AeadAlgorithm enc = Jwts.ENC.A256GCM;
SecretKey key = enc.key().build();
String compact = Jwts.builder().setSubject("Joe").encryptWith(key, enc).compact();
Jwe<Claims> jwe = Jwts.parser().decryptWith(key).build().parseEncryptedClaims(compact);
Many other RSA and Elliptic Curve examples are in the full README documentation.
JSON Web Key (JWK) Support!
Representing cryptographic keys - SecretKeys, RSA Public and Private Keys, Elliptic Curve Public and
Private keys - as fully encoded JSON objects according to the JWK specification - is now fully implemented and
supported. The new Jwks
utility class exists to create JWK builders and parsers as desired. For example:
SecretKey key = Jwts.SIG.HS256.key().build();
SecretJwk jwk = Jwks.builder().forKey(key).build();
assert key.equals(jwk.toKey());
// or if receiving a JWK string:
Jwk<?> parsedJwk = Jwks.parser().build().parse(jwkString);
assert jwk.equals(parsedJwk);
assert key.equals(parsedJwk.toKey());
Many JJWT users won't need to use JWKs explicitly, but some JWA Key Management Algorithms (and lots of RFC test vectors) utilize JWKs when transmitting JWEs. As this was required by JWE, it is now implemented in full for JWE use as well as general-purpose JWK support.
JWK Thumbprint and JWK Thumbprint URI support
The JWK Thumbprint and JWK Thumbprint URI RFC specifications are now fully supported. Please see the README.md file's corresponding named sections for both for full documentation and usage examples.
JWS Unencoded Payload Option (b64
) support
The JSON Web Signature (JWS) Unencoded Payload Option RFC specification is now fully supported. Please see the README.md corresponding named section for documentation and usage examples.
Better PKCS11 and Hardware Security Module (HSM) support
Previous versions of JJWT enforced that Private Keys implemented the RSAKey
and ECKey
interfaces to enforce key
length requirements. With this release, JJWT will still perform those checks when those data types are available,
but if not, as is common with keys from PKCS11 and HSM KeyStores, JJWT will still allow those Keys to be used,
expecting the underlying Security Provider to enforce any key requirements. This should reduce or eliminate any
custom code previously written to extend JJWT to use keys from those KeyStores or Providers.
Additionally, PKCS11/HSM tests using SoftHSMv2 are run on every build with every JWS MAC and Signature algorithm and every JWE Key algorithm to ensure continued stable support with Android and Sun PKCS11 implementations and spec-compliant Hardware Security Modules that use the PKCS11 interface (such as YubiKey, etc.)
Custom Signature Algorithms
The io.jsonwebtoken.SignatureAlgorithm
enum has been deprecated in favor of new
io.jsonwebtoken.security.SecureDigestAlgorithm
, io.jsonwebtoken.security.MacAlgorithm
, and
io.jsonwebtoken.security.SignatureAlgorithm
interfaces to allow custom algorithm implementations. The new nested
Jwts.SIG
static inner class is a registry of all standard JWS algorithms as expected, exactly like the
old enum. This change was made because enums are a static concept by design and cannot
support custom values: those who wanted to use custom signature algorithms could not do so until now. The new
interfaces now allow anyone to plug in and support custom algorithms with JJWT as desired.
KeyBuilder and KeyPairBuilder
Because the io.jsonwebtoken.security.Keys#secretKeyFor
and io.jsonwebtoken.security.Keys#keyPairFor
methods
accepted the now-deprecated io.jsonwebtoken.SignatureAlgorithm
enum, they have also been deprecated in favor of
calling new key()
or keyPair()
builder methods on MacAlgorithm
and SignatureAlgorithm
instances directly.
For example:
SecretKey key = Jwts.SIG.HS256.key().build();
KeyPair pair = Jwts.SIG.RS256.keyPair().build();
The builders allow for customization of the JCA Provider
and SecureRandom
during Key or KeyPair generation if desired, whereas
the old enum-based static utility methods did not.
Preparation for 1.0
Now that the JWE and JWK specifications are implemented, only a few things remain for JJWT to be considered at version 1.0. We have been waiting to apply the 1.0 release version number until the entire set of JWT specifications are fully supported and we drop JDK 7 support (to allow users to use JDK 8 APIs). To that end, we have had to deprecate some concepts, or in some cases, completely break backwards compatibility to ensure the transition to 1.0 (and JDK 8 APIs) are possible. Most backwards-incompatible changes are listed in the next section below.
Backwards Compatibility Breaking Changes, Warnings and Deprecations
-
io.jsonwebtoken.Jwt
'sgetBody()
method has been deprecated in favor of a newgetPayload()
method to reflect correct JWT specification nomenclature/taxonomy. -
io.jsonwebtoken.Jws
'sgetSignature()
method has been deprecated in favor of a newgetDigest()
method to support expected congruent behavior withJwe
instances (both have digests). -
io.jsonwebtoken.JwtParser
'sparseContentJwt
,parseClaimsJwt
,parseContentJws
, andparseClaimsJws
methods have been deprecated in favor of more intuitive respectiveparseUnsecuredContent
,parseUnsecuredClaims
,parseSignedContent
andparseSignedClaims
methods. -
io.jsonwebtoken.CompressionCodec
is now deprecated in favor of the newio.jsonwebtoken.io.CompressionAlgorithm
interface. This is to guarantee API congruence with all other JWT-identifiable algorithm IDs that can be set as a header value. -
io.jsonwebtoken.CompressionCodecResolver
has been deprecated in favor of the newJwtParserBuilder#addCompressionAlgorithms
method.
Breaking Changes
-
io.jsonwebtoken.Claims
andio.jsonwebtoken.Header
instances are now immutable to enhance security and thread safety. Creation and mutation are supported with newly introducedClaimsBuilder
andHeaderBuilder
concepts. Even though mutation methods have migrated, there are a couple that have been removed entirely:-
io.jsonwebtoken.JwsHeader#setAlgorithm
has been removed - theJwtBuilder
will always set the appropriatealg
header automatically based on builder state. -
io.jsonwebtoken.Header#setCompressionAlgorithm
has been removed - theJwtBuilder
will always set the appropriatezip
header automatically based on builder state.
-
-
io.jsonwebtoken.Jwts
'sheader(Map)
,jwsHeader()
andjwsHeader(Map)
methods have been removed in favor of the newheader()
method that returns aHeaderBuilder
to support method chaining and dynamicHeader
type creation. TheHeaderBuilder
will dynamically create aHeader
,JwsHeader
orJweHeader
automatically based on builder state. -
Similarly,
io.jsonwebtoken.Jwts
'sclaims()
static method has been changed to return aClaimsBuilder
instead of aClaims
instance. -
JWTs that do not contain JSON Claims now have a payload type of
byte[]
instead ofString
(that is,Jwt<byte[]>
instead ofJwt<String>
). This is because JWTs, especially when used with thecty
(Content Type) header, are capable of handling any type of payload, not just Strings. The previous JJWT releases didn't account for this, and now the API accurately reflects the JWT RFC specification payload capabilities. Additionally, the name ofplaintext
has been changed tocontent
in method names and JavaDoc to reflect this taxonomy. This change has impacted the following JJWT APIs:-
The
JwtBuilder
'ssetPayload(String)
method has been deprecated in favor of two new methods:-
setContent(byte[])
, and -
setContent(byte[], String contentType)
These new methods allow any kind of content within a JWT, not just Strings. The existing
setPayload(String)
method implementation has been changed to delegate to this newsetContent(byte[])
method with the argument's UTF-8 bytes, for examplesetContent(payloadString.getBytes(StandardCharsets.UTF_8))
. -
-
The
JwtParser
'sJwt<Header, String> parsePlaintextJwt(String plaintextJwt)
andJws<String> parsePlaintextJws(String plaintextJws)
methods have been changed toJwt<Header, byte[]> parseContentJwt(String plaintextJwt)
andJws<byte[]> parseContentJws(String plaintextJws)
respectively. -
JwtHandler
'sonPlaintextJwt(String)
andonPlaintextJws(String)
methods have been changed toonContentJwt(byte[])
andonContentJws(byte[])
respectively. -
io.jsonwebtoken.JwtHandlerAdapter
has been changed to reflect the above-mentioned name andString
-to-byte[]
argument changes, as well adding theabstract
modifier. This class was never intended to be instantiated directly, and is provided for subclassing only. The missing modifier has been added to ensure the class is used as it had always been intended. -
io.jsonwebtoken.SigningKeyResolver
'sresolveSigningKey(JwsHeader, String)
method has been changed toresolveSigningKey(JwsHeader, byte[])
.
-
-
io.jsonwebtoken.JwtParser
is now immutable. All mutation/modification methods (setters, etc) deprecated 4 years ago have been removed. All parser configuration requires using theJwtParserBuilder
. -
Similarly,
io.jsonwebtoken.Jwts
'sparser()
method deprecated 4 years ago has been changed to now return aJwtParserBuilder
instead of a directJwtParser
instance. The previousJwts.parserBuilder()
method has been removed as it is now redundant. -
The
JwtParserBuilder
no longer supportsPrivateKey
s for signature verification. This was an old legacy behavior scheduled for removal years ago, and that change is now complete. For various cryptographic/security reasons, asymmetric public/private key signatures should always be created withPrivateKey
s and verified withPublicKey
s. -
io.jsonwebtoken.CompressionCodec
implementations are no longer discoverable viajava.util.ServiceLoader
due to runtime performance problems with the JDK'sServiceLoader
implementation per https://github.com/jwtk/jjwt/issues/648/648. Custom implementations should be made available to theJwtParser
via the newJwtParserBuilder#addCompressionAlgorithms
method. -
Prior to this release, if there was a serialization problem when serializing the JWT Header, an
IllegalStateException
was thrown. If there was a problem when serializing the JWT claims, anIllegalArgumentException
was thrown. This has been changed up to ensure consistency: any serialization error with either headers or claims will now throw aio.jsonwebtoken.io.SerializationException
. -
Parsing of unsecured JWTs (
alg
header ofnone
) are now disabled by default as mandated by RFC 7518, Section 3.6. If you require parsing of unsecured JWTs, you must call theJwtParserBuilder#enableUnsecured()
method, but note the security implications mentioned in that method's JavaDoc before doing so. -
io.jsonwebtoken.gson.io.GsonSerializer
now requiresGson
instances that have a registeredGsonSupplierSerializer
type adapter, for example:new GsonBuilder() .registerTypeHierarchyAdapter(io.jsonwebtoken.lang.Supplier.class, GsonSupplierSerializer.INSTANCE) .disableHtmlEscaping().create();
This is to ensure JWKs have
toString()
and application log safety (do not print secure material), but still serialize to JSON correctly. -
io.jsonwebtoken.InvalidClaimException
and its two subclasses (IncorrectClaimException
andMissingClaimException
) were previously mutable, allowing the corresponding claim name and claim value to be set on the exception after creation. These should have always been immutable without those setters (just getters), and this was a previous implementation oversight. This release has ensured they are immutable without the setters.
Configuration
📅 Schedule: Branch creation - At any time (no schedule defined), Automerge - At any time (no schedule defined).
🚦 Automerge: Disabled by config. Please merge this manually once you are satisfied.
♻ Rebasing: Whenever PR becomes conflicted, or you tick the rebase/retry checkbox.
🔕 Ignore: Close this PR and you won't be reminded about these updates again.
- [ ] If you want to rebase/retry this PR, check this box
This PR was generated by Mend Renovate. View the repository job log.
This project does no follow SemVer. There are code and package changes that are not backwards compatible.