Java^TM Certification Path API Specification

Public Review Draft Specification

Public Review Draft #1

Author: Sean Mullan (sean.mullan@sun.com)

Last Modified: 08 Nov 2000

Changes

Overview

Acknowledgments
Who Should Read This Document
Related Documentation
Introduction

Core Classes and Interfaces

Basic Classes

• The CertPath Class

• The CertificateFactory Class

• The CertPathParameters Interface

Certification Path Validation Classes

• The CertPathValidator Class

• The CertPathValidatorResult Interface

Certification Path Building Classes

• The CertPathBuilder Class

• The CertPathBuilderResult Interface

Certificate/CRL Storage Classes

• The CertStore Class

• The CertStoreParameters Interface

  • The LDAPCertStoreParameters Class

  • The CollectionCertStoreParameters Class

• The CertSelector and CRLSelector Interfaces

  • The X509CertSelector Class

  • The X509CRLSelector Class

PKIX Classes

• The PKIXParameters Class

• The PKIXCertPathValidatorResult Class

• The PolicyNode and PolicyQualifierInfo Classes

• The PKIXBuilderParameters Class

• The PKIXCertPathBuilderResult Class

• The PKIXCertPathChecker Class

Implementing a Service Provider

Appendix A: Standard Names

Changes

• Enhanced PKIX classes to provide support for the PKIX validation algorithm defined in http://www.ietf.org/internet-drafts/draft-ietf-pkix-new-part1-02.txt. This is the third draft of a specification based upon RFC 2459. When complete, this specification will obsolete RFC 2459. In particular, the changes include:
  • Added the setAnyPolicyInhibited and isAnyPolicyInhibited methods to the PKIXParameters class to support the initial-any-policy-inhibit input flag.
  • Added the PolicyNode interface, representing a node of the policy tree as defined in the PKIX draft.
  • Removed the getQualifiers and getPolicies methods from the PKIXCertPathValidatorResult class, and replaced them with the getPolicyTree method which returns a PolicyNode object representing the root of the policy tree as returned by the PKIX validation algorithm. Also added the getPublicKey method which returns the subject public key (including any inherited parameters if applicable) resulting from the validation algorithm.
  • Removed the getCertificateIndex method and the index constructor parameter from the PolicyQualifierInfo class. This functionality is replaced by the getDepth method of the PolicyNode interface.
• Changed the format of String DN parameters (used in methods and constructors of the X509CertSelector, X509CRLSelector, PKIXParameters and PKIXBuilderParameters classes) from RFC 1779 to RFC 2253 (2253 obsoletes 1779).
• Improved class description of the PolicyQualifierInfo class.
• Removed the setCertPath and setIndex methods of CertPathValidatorException and added equivalent parameters to the constructors.
• Changed the getEncodings method of the CertPath class to return an Iterator instead of an Enumeration. Also applies to the getCertPathEncodings method of CertificateFactory and the engineGetCertPathEncodings method of CertificateFactorySpi.
• Changed the LDAPCertStoreParameters class to not implement Serializable.
• Changed constructors and setTrustedCerts method of the PKIXParameters and PKIXBuilderParameters classes to throw ClassCastException when the elements of the Set are not of type X509Certificate. Also changed the setCertPathCheckers method to throw ClassCastException when the elements of the List are not of type PKIXCertPathChecker.
• Changed the CertStoreParameters interface to extend Cloneable and defined a clone method that throws no exceptions. This allows all CertStoreParameters to be cloned.
• Added the certificateEquals criterion and the setCertificate and getCertificate methods to the X509CertSelector class. This is useful for specifying target certificate constraints with a PKIXBuilderParameters object when a relying party possesses the target certificate.
• In the X509CertSelector class description, described which criteria combinations would likely select a unique certificate.
• Added the setMatchAllSubjectAltNames and getMatchAllSubjectAltNames methods to the X509CertSelector class. These methods allow a caller to choose to match on at least one or all subject alternative names in an X509Certificate.
• Documented cases when the getCertificates and getCRLs methods of CertStore may throw a CertStoreException because the caller has supplied insufficient information for the CertStore to find the requested certificates or CRLs.

Overview

The Java Certification Path API consists of classes and interfaces for handling certification paths (also known as "certificate chains"). A certification path is an ordered list of certificates. If a certification path meets certain validation rules, it may be used to securely establish the mapping of a public key to a subject.

This API defines interfaces and abstract classes for creating, building, and validating certification paths.  Implementations may be plugged in using a provider-based interface. The API is based on the Cryptographic Service Provider architecture, described in the Java Cryptography Architecture Document.

The API also includes algorithm specific classes for building and validating X.509 certification paths according to the PKIX standards. The PKIX standards are developed by the IETF PKIX working group. Section 6 of http://www.ietf.org/internet-drafts/draft-ietf-pkix-new-part1-02.txt defines an algorithm for validating chains of X.509 certificates.

This API is being reviewed using the Java Community Process program. This is the Public Review Draft version of the CertPath API. Comments from experts and members of the Java Community should be sent to the email address noted at the end of this specification.

Acknowledgments

The author would like to thank the individuals who contributed to the Certification Path API and provided useful comments and technical advice. Special thanks to the members of the Sun Microsystems Laboratories team who designed and developed the Certification Path API which provided the basis of the work for the Java Community Process. This team includes Anne Anderson, Yassir Elley, Geoff Goodell, Steve Hanna, Sean Mullan, Radia Perlman, and Seth Proctor.

The expert group helped improve and refine the API using the Java Community Process and includes the following members:

• Peter Hesse, CygnaCom Solutions, An Entrust Technologies Company
• Sean Mullan, Sun Microsystems (Specification Lead)
• Anthony Nadalin, IBM
• Bob Naugle, Bluestone Software
• Hemma Prafullchandra, Critical Path
• Ming Yung, DSTC

Maxine Erlund, Steve Hanna, Phil Rosenzweig and Bob Sproull of Sun Microsystems provided leadership and vision. The members of the Java Security and Networking Group of Sun Microsystems contributed invaluable comments and support, especially Sharon Liu, Jeff Nisewanger, and Gary Ellison. Helpful comments and advice were received from many in the technical community, especially Edward Dobner, Tom Gindin, Jan Luehe, David Kuehr-McLaren, and Alexei Semidetnov.

Who Should Read This Document

This document is intended for two classes of experienced developers:

1. those who want to design secure applications that create, build, or validate certification paths.

2. those who want to write a service provider implementation for a particular certification path algorithm, type, or storage repository.

Related Documentation

This document assumes you have already read the following documents (available online at http://java.sun.com/j2se/1.3/docs/guide/security/index.html):

• X.509 Certificates and Certificate Revocation Lists (CRLs)

• Java Cryptography Architecture

• How to Implement a Provider for the Java Cryptography Architecture

Introduction

Users of public key applications and systems must be confident that a subject's public key is genuine, i.e. that the associated private key is owned by the subject. Public key certificates are used to establish this trust. A public key (or identity) certificate is a binding of a public key to an identity, which is digitally signed by the private key of another entity, often called a Certification Authority (CA). We shall use the term CA to refer to an entity that signs a certificate for the remainder of this section.

If the user does not have a trusted copy of the public key of the CA that signed the subject's public key certificate, then another public key certificate vouching for the signing CA is required. This logic can be applied recursively, until a chain of certificates (or a certification path) is discovered from a trust anchor or a most-trusted CA to the subject. The most-trusted CA is usually specified by a certificate issued to a CA that the user directly trusts. In general, a certification path is an ordered list of certificates, usually comprised of the subject's public key certificate and zero or more additional certificates. A certification path typically has one or more encodings, allowing it to be safely transmitted across networks and to different operating system architectures.

Figure 1 illustrates a certification path from a most-trusted CA's public key (CA 1) to the subject (Alice). The certification path establishes trust in Alice's public key through an intermediate CA named CA2.

Figure 1: Certification Path

A certification path must be validated before it can be relied on to establish trust in a subject's public key. Validation can consist of various checks on the certificates contained in the certification path, such as verifying the signatures and checking that each certificate has not been revoked. The PKIX standards define an algorithm for validating certification paths consisting of X.509 certificates.

Often a user may not have a certification path from a most-trusted CA to the subject. Providing services to build or discover certification paths is an important feature of public key enabled systems. RFC 2587 defines an LDAP schema definition which facilitates the discovery of X.509 certification paths using the LDAP directory service protocol.

Building and validating certification paths is an important part of many standard security protocols such as SSL/TLS, S/MIME, and IPSEC.  The Java Certification Path API provides a set of classes and interfaces to Java developers who need to integrate this functionality into their applications. This API benefits two types of developers: those who need to write service provider implementations for a specific certification path algorithm; and those who need to access standard algorithms for creating, building, and validating certification paths in an implementation independent manner.

Core Classes and Interfaces

The core classes of the Java Certification Path API consist of interfaces and classes that support certification path functionality in an algorithm and implementation independent manner. The API also includes a set of algorithm-specific classes for the PKIX standards and these are discussed in the section titled PKIX Classes. The API builds on and extends the existing J2SE (Java 2 Standard Edition) java.security.cert package for handling certificates. The core classes can be broken up into 4 class categories: Basic, Validation, Building, and Storage:

• Basic Certification Path Classes

  • CertPath, CertificateFactory, CertPathParameters

• Certification Path Validation Classes

  • CertPathValidator, CertPathValidatorResult

• Certification Path Building Classes

  • CertPathBuilder, CertPathBuilderResult

• Certificate/CRL Storage Classes

  • CertStore, CertStoreParameters, CertSelector, CRLSelector

The following sections describe the most commonly used methods of each class and interface. Usage examples for some of the classes are interspersed throughout the guide. The complete reference documentation for the relevant Certification Path API classes can be found in:

• java.security.cert package summary

Unless otherwise specified, the classes and interfaces in the CertPath API are not thread-safe. Multiple threads that need to access a single object concurrently should synchronize amongst themselves and provide the necessary locking. Multiple threads each manipulating separate objects need not synchronize.

Basic Certification Path Classes

The basic certification path classes are those that provide fundamental functionality for creating certification paths. The key basic class in the Java Certification Path API is CertPath, which encapsulates the universal aspects shared by all types of certification paths. An application uses an instance of the CertificateFactory class to create a CertPath object.

The CertPath Class

The CertPath class is an abstract class for certification paths. It defines the functionality shared by all certification path objects. Various certification path types can be implemented by subclassing the CertPath class, even though they may have different contents and ordering schemes. All CertPath objects are immutable and thread-safe and share the following characteristics:

• A type
  
  This corresponds to the type of the certificates in the certification path, for example: X.509. The type of a CertPath is obtained using the method:
  
  

      public String getType()

• A list of certificates
  
  The getCertificates method returns the list of certificates in the certification path:  
  
  

      public abstract List getCertificates()

  This method returns a List of zero or more Certificate objects. The returned List and the Certificates contained within it are immutable, in order to protect the contents of the CertPath object.
  The ordering of the certificates returned depends on the type. By convention, the certificates in a CertPath object of type X.509 are ordered from target subject to trust anchor. That is, the issuer of one certificate is the subject of the following one. However, unvalidated X.509 CertPaths may not follow this convention. PKIX CertPathValidators will detect any departure from this convention and throw a CertPathValidatorException.

  In addition, an X.509 CertPath may not contain the most-trusted CA certificate used to anchor the path. However, the getTrustedCert method of a PKIXCertPathValidatorResult returns the most-trusted CA certificate (or null if not applicable) used by a PKIX CertPathValidator to anchor the path. The PKIX classes are discussed more in the PKIX Classes section.
  
  

• One or more encodings
  
  Each CertPath object supports one or more encodings. These are external encoded forms for the certification path, used when a standard representation of the path is needed outside the Java Virtual Machine (as when transmitting the path over a network to some other party). Each path can be encoded in a default format, the bytes of which are returned using the method:
  
  

      public abstract byte[] getEncoded()

  The getEncoded(String) method returns a supported encoding by specifying the encoding format as a String (ex: "PKCS7"). A list of standard encoding formats are defined in Appendix A.
  
  

      public abstract byte[] getEncoded(String encoding)

  The getEncodings method returns an iterator over the supported encoding format Strings (the default encoding format is returned first):
  
  

      public abstract Iterator getEncodings()

CertPath objects are generated from an encoded byte array or list of Certificates using a CertificateFactory. Alternatively a CertPathBuilder may be used to try to find a CertPath from a most-trusted CA to a particular subject. Once a CertPath object has been created, it may be validated by passing it to the validate method of CertPathValidator. Each of these concepts are explained in more detail in subsequent sections.

The CertificateFactory Class

The CertificateFactory class is an existing engine class that defines the functionality of a certificate factory. This class has been enhanced to generate certification path (CertPath) objects as well as Certificate and CRL objects. A CertificateFactory should not be confused with a CertPathBuilder. A CertPathBuilder (discussed later) is used to discover or find a certification path when one does not exist. In contrast, a CertificateFactory is used when a certification path has already been discovered and the caller needs to instantiate a CertPath object from its contents which exist in a different form such as an encoded PKCS#7 byte array.

Creating a CertificateFactory Object

See the CertificateFactory section in the Java Cryptography Architecture for details on creating a CertificateFactory object.

Generating CertPath Objects

A CertificateFactory instance generates CertPath objects from a List of Certificate objects or from an InputStream that contains the encoded form of a CertPath. Just like a CertPath, each CertificateFactory supports a default encoding format for certification paths (ex: PKCS#7). To generate a CertPath object and initialize it with the data read from an input stream (in the default encoding format), use the generateCertPath method:

    public final CertPath generateCertPath(InputStream inStream)

or from a particular encoding format:

    public final CertPath generateCertPath(InputStream inStream, 
    					   String encoding)

To find out what encoding formats are supported, use the getCertPathEncodings method (the default encoding is returned first):

    public final Iterator getCertPathEncodings()

To generate a certification path object from a List of Certificate objects, use the following method:

    public final CertPath generateCertPath(List certificates)

A CertificateFactory always returns CertPath objects that consist of Certificates that are of the same type as the factory. For example, a CertificateFactory of type X.509 returns CertPath objects consisting of certificates that are an instance of java.security.cert.X509Certificate.

The following code sample illustrates generating a certification path from a PKCS#7 encoded certificate reply stored in a file:

// open an input stream to the file
FileInputStream fis = new FileInputStream(filename);
// instantiate a CertificateFactory for X.509
CertificateFactory cf = CertificateFactory.getInstance("X.509");
// extract the certification path from
// the PKCS7 SignedData structure
CertPath cp = cf.generateCertPath(fis, "PKCS7");
// print each certificate in the path
List certs = cp.getCertificates();
Iterator i = certs.iterator();
while (i.hasNext()) {
    X509Certificate cert = (X509Certificate) i.next();
    System.out.println(cert);
}

Note that there is an existing method in CertificateFactory named generateCertificates that parses a sequence of Certificates. For encodings consisting of multiple certificates, use generateCertificates when you want to parse a collection of possibly unrelated certificates. Otherwise, use generateCertPath when you want to generate a CertPath and subsequently validate it with a CertPathValidator (discussed later).

The CertPathParameters Interface

The CertPathParameters interface is a transparent representation of the set of parameters used with a particular certification path builder or validation algorithm. Its main purpose is to group (and provide type safety for) all certification path parameter specifications. The CertPathParameters interface extends the Cloneable interface and defines a clone() method that does not throw an exception. All concrete implementations of this interface should implement and override the Object.clone() method, if necessary. This allows applications to clone any CertPathParameters object.

Objects implementing the CertPathParameters interface are passed as arguments to methods of the CertPathValidator and CertPathBuilder classes.

Certification Path Validation Classes

The Java Certification Path API includes classes and interfaces for validating certification paths. An application uses an instance of the CertPathValidator class to validate a CertPath object. If successful, the result of the validation algorithm is returned in an object implementing the CertPathValidatorResult interface.

The CertPathValidator Class

The CertPathValidator class is an engine class used to validate a certification path.

Creating a CertPathValidator Object

As with all engine classes, the way to get a CertPathValidator object for a particular validation algorithm is to call the getInstance static factory method on the CertPathValidator class:

	public static CertPathValidator getInstance(String algorithm)
    	public static CertPathValidator getInstance(String algorithm, 
						    String provider)

Validating a Certification Path

Once a CertPathValidator object is created, paths can be validated by calling the validate method, passing it the certification path to be validated and a set of algorithm-specific parameters (see PKIXParameters for an example):

	public final CertPathValidatorResult 
	    	validate(CertPath certPath, CertPathParameters params)
		throws CertPathValidatorException, 
		       InvalidAlgorithmParametersException

If the validation algorithm is successful, the result is returned in an object implementing the CertPathValidatorResult interface. Otherwise, a CertPathValidatorException is thrown. The CertPathValidatorException contains methods that return the CertPath, and if relevant, the index of the certificate that caused the algorithm to fail and the root exception or cause of the failure.

The CertPath object passed to the validate method must be of a type that is supported by the validation algorithm. For example, a CertPathValidator instance that implements the PKIX algorithm validates CertPath objects of type X.509.

The CertPathValidatorResult Interface

The CertPathValidatorResult interface is a transparent representation of the result or output of a certification path validation algorithm. Its main purpose is to group (and provide type safety for) all validation results. Like the CertPathParameters interface, CertPathValidatorResult extends Cloneable and defines a clone() method that does not throw an exception. This allows applications to clone any CertPathValidatorResult object.

Objects implementing the CertPathValidatorResult interface are returned by the validate method of CertPathValidator.

Certification Path Building Classes

The Java Certification Path API includes classes for building (or discovering) certification paths. An application uses an instance of the CertPathBuilder class to build a CertPath object. If successful, the result of the build is returned in an object implementing the CertPathBuilderResult interface.

The CertPathBuilder Class

The CertPathBuilder class is an engine class used to build a certification path.

Creating a CertPathBuilder Object

As with all engine classes, the way to get a CertPathBuilder object for a particular build algorithm is to call the getInstance static factory method on the CertPathBuilder class:

	public static CertPathBuilder getInstance(String algorithm)
	public static CertPathBuilder getInstance(String algorithm, 
						  String provider)

Building a Certification Path

Once a CertPathBuilder object is created, paths can be constructed by calling the build method, passing it an algorithm-specific parameter specification (see PKIXBuilderParameters for an example):

	public final CertPathBuilderResult build(CertPathParameters params)
		throws CertPathBuilderException, 
		       InvalidAlgorithmParametersException

If the build algorithm is successful, the result is returned in an object implementing the CertPathBuilderResult interface. Otherwise, a CertPathBuilderException is thrown containing information about the failure; for example, the underlying exception (if any) and an error message.

The CertPathBuilderResult Interface

The CertPathBuilderResult interface is a transparent representation of the result or output of a certification path builder algorithm. This interface contains a method to return the certification path that has been successfully built:

	public CertPath getCertPath()

The purpose of the CertPathBuilderResult interface is to group (and provide type safety for) all build results. Like the CertPathValidatorResult interface, CertPathBuilderResult extends Cloneable and defines a clone() method that does not throw an exception. This allows applications to clone any CertPathBuilderResult object.

Objects implementing the CertPathBuilderResult interface are returned by the build method of CertPathBuilder.

Certificate/CRL Storage Classes

The Java Certification Path API also includes the CertStore class for retrieving certificates and CRLs from a repository. This is useful because it allows a caller to specify the repository a CertPathValidator or CertPathBuilder implementation should use to find certificates and CRLs (see the addCertStores method of PKIXParameters for an example).

A CertPathValidator or CertPathBuilder implementation may use the methods of the CertStore class as a callback mechanism to select certificates and CRLs from a repository. Selection criteria for choosing specific certificates and CRLs are specified in objects implementing the CertSelector and CRLSelector interfaces. CertStore instances may need to be initialized with repository specific input parameters before usage.

The CertStore Class

The CertStore class is an engine class used to provide the functionality of a certificate and certificate revocation list (CRL) repository. It can be used by CertPathBuilder and CertPathValidator implementations to find certificates and CRLs or as a general purpose certificate and CRL retrieval mechanism.

Unlike the java.security.KeyStore class, which provides access to a cache of private keys and trusted certificates, a CertStore is designed to provide access to a potentially vast repository of untrusted certificates and CRLs. For example, an LDAP implementation of CertStore provides access to certificates and CRLs stored in one or more directories using the LDAP protocol.

Creating a CertStore Object

As with all engine classes, the way to get a CertStore object for a particular repository type is to call the getInstance static factory method on the CertStore class:

	public static CertStore getInstance(String type)
	public static CertStore getInstance(String type, String provider)

Initializing a CertStore Object

A CertStore object may need to be initialized before it can be used. The initialization is done by calling the init method and passing it a repository specific set of parameters:

	public final void init(CertStoreParameters params)
		throws InvalidAlgorithmParametersException

The initialization parameters are specific to the repository type. For example, the initialization parameters for a server based repository may include the hostname and the port of the server. An InvalidAlgorithmParametersException is thrown if the parameters are invalid for this CertStore type.

Retrieving Certificates

Once you have created and initialized a CertStore object, you can retrieve certificates from the repository using the getCertificates method. This method takes a CertSelector object as an argument, which specifies a set of selection criteria for determining which certificates should be returned:

	public final Collection getCertificates(CertSelector selector) 
		throws CertStoreException

This method returns a Collection of java.security.cert.Certificate objects that satisfy the selection criteria. An empty Collection is returned if there are no matches. A CertStoreException is usually thrown if an unexpected error condition is encountered, such as a communications failure with a remote repository.

For some CertStore implementations, it may not be feasible to search the entire repository for certificates or CRLs that match the specified selection criteria. In these instances, the CertStore implementation may use information that is specified in the selectors to locate certificates and CRLs. For instance, an LDAP CertStore may not search all entries in the directory. Instead, it may just search entries that are likely to contain the certificates it is looking for. If the CertSelector provided does not provide enough information for the LDAP CertStore to determine which entries it should look in, the LDAP CertStore may throw a CertStoreException.

Retrieving CRLs

You can also retrieve CRLs from the repository using the getCRLs method. This method takes a CRLSelector object as an argument, which specifies a set of selection criteria for determining which CRLs should be returned:

	public final Collection getCRLs(CRLSelector selector) 
		throws CertStoreException

This method returns a Collection of java.security.cert.CRL objects that satisfy the selection criteria. An empty Collection is returned if there are no matches.

The getCertificates and getCRLs methods of all CertStore objects must be thread-safe. That is, multiple threads may concurrently invoke these methods on a single CertStore object (or more than one) with no ill effects. This allows a CertPathBuilder to search for a CRL while simultaneously searching for further certificates, for instance. The static methods of this class are also guaranteed to be thread-safe. Multiple threads may concurrently invoke the static methods defined in this class with no ill effects. The init method should not be called concurrently with any other method nor should multiple threads invoke the init method concurrently on a single CertStore object. Otherwise, the results are undefined.

The CertStoreParameters Interface

The CertStoreParameters interface is a transparent representation of the set of parameters used with a particular certificate and CRL repository (CertStore). Its main purpose is to group (and provide type safety for) all certificate storage parameter specifications. The CertStoreParameters interface extends the Cloneable interface and defines a clone method that does not throw an exception. Implementations of this interface should implement and override the Object.clone() method, if necessary. This allows applications to clone any CertStoreParameters object.

Objects implementing the CertStoreParameters interface are passed as arguments to the init method of the CertStore class. Two classes implementing the CertStoreParameters interface are defined in this API: the LDAPCertStoreParameters and the CollectionCertStoreParameters classes.

The LDAPCertStoreParameters Class

The LDAPCertStoreParameters class is an implementation of the CertStoreParameters interface and holds a set of minimal initialization parameters (host and port number of the directory server) for retrieving certificates and CRLs from an LDAP directory. It is passed to the init method of CertStore instances of type LDAP.

Please refer to the LDAPCertStoreParameters API documentation for more detailed information on this class.

The CollectionCertStoreParameters Class

The CollectionCertStoreParameters class is an implementation of the CertStoreParameters interface and holds a set of initialization parameters for retrieving certificates and CRLs from a Collection. It is passed to the init method of CertStore instances of type Collection.

Please refer to the CollectionCertStoreParameters API documentation for more detailed information on this class.

The CertSelector and CRLSelector Interfaces

The CertSelector and CRLSelector interfaces are a specification of the set of criteria for selecting certificates and CRLs from a collection or large group of certificates and CRLs. The interfaces group and provide type safety for all selector specifications. Each selector interface extends Cloneable and defines a clone() method that does not throw an exception. This allows applications to clone any CertSelector or CRLSelector object.

The CertSelector and CRLSelector interfaces each define a method named match. The match method takes a Certificate or CRL object as an argument and returns true if the object satisfies the selection criteria. Otherwise, it returns false. The match method for the CertSelector interface is defined as follows:

	public boolean match(Certificate cert)

and for the CRLSelector interface:

	public boolean match(CRL crl)

Objects implementing these interfaces are passed as parameters to the getCertificates and getCRLs methods of the CertStore class. These methods return a Collection of Certificates or CRLs from the CertStore repository that match the specified selection criteria.

The X509CertSelector Class

The X509CertSelector class is an implementation of the CertSelector interface that defines a set of criteria for selecting X.509 certificates. An X509Certificate object must match all of the specified criteria to be selected by the match method. The selection criteria are designed to be used by a CertPathBuilder implementation to discover potential certificates as it builds an X.509 certification path.

For example, the setPathToName method of X509CertSelector allows a PKIX CertPathBuilder to filter out X509Certificates that contain a name constraints extension that would prohibit building a path to the requested subject name(s). By setting this and other criteria in an X509CertSelector object, it allows the CertPathBuilder to discard non-relevant certificates and more easily find an X.509 certification path that meets the requirements specified in the CertPathParameters object.

Please refer to http://www.ietf.org/internet-drafts/draft-ietf-pkix-new-part1-02.txt for definitions of the X.509 certificate extensions mentioned in this section.

Creating an X509CertSelector Object

An X509CertSelector object is created by calling the default constructor:

	public X509CertSelector()

No criteria are initially set (any X509Certificate will match).

Setting Selection Criteria

The selection criteria allow a caller to match on different components of an X.509 certificate. A few of the methods for setting selection criteria are described here, please refer to the X509CertSelector API documentation for details on the other methods.

The setIssuer methods set the issuer criterion:

	public void setIssuer(String issuerDN)
	public void setIssuer(byte[] issuerDN)

The specified distinguished name (in RFC 2253 String or ASN.1 DER encoded form) must match the issuer distinguished name in the certificate. If null, any issuer distinguished name will do.

Similarly, the setSubject methods set the subject criterion:

	public void setSubject(String subjectDN)
	public void setSubject(byte[] subjectDN)

The setSerialNumber method sets the serialNumber criterion:

	public void setSerialNumber(BigInteger serial)

The specified serial number must match the certificate serial number in the certificate. If null, any certificate serial number will do.

The setAuthorityKeyIdentifier method sets the authorityKeyIdentifier criterion:

	public void setAuthorityKeyIdentifier(byte[] authorityKeyID)

The certificate must contain an Authority Key Identifier extension matching the specified value. If null, no check will be done on the authorityKeyIdentifier criterion.

The setCertificateValid method sets the certificateValid criterion:

	public void setCertificateValid(Date certValid)

The specified date must fall within the certificate validity period for the certificate. If null, any date is valid.

The setKeyUsage method sets the keyUsage criterion:

	public void setKeyUsage(boolean[] keyUsage)

The certificate's Key Usage Extension must allow the specified key usage values (those which are set to true). If null, no keyUsage check will be done.

Getting Selection Criteria

The current values for each of the selection criteria can be retrieved using an appropriate get method. Please refer to the X509CertSelector API documentation for further details on these methods.

Example

This is an example of retrieving X.509 certificates from an LDAP CertStore with the X509CertSelector class.

First, create the CertStore object, as in the following statement:

	CertStore cs = CertStore.getInstance("LDAP");

This call creates a CertStore object that retrieves certificates and CRLs from an LDAP repository using the schema defined in RFC 2587. See Appendix A for a complete discussion of standard names and algorithms.

The next step is to initialize the certificate storage object with the hostname and port of the LDAP server. To do that, we create an LDAPCertStoreParameters object and pass it to the init method:

	LDAPCertStoreParameters csp = new 
		LDAPCertStoreParameters("ldap.sun.com", 389);
	cs.init(csp);

The following block of code establishes an X509CertSelector to retrieve all unexpired (as of the current date and time) end-entity certificates issued to a particular subject with 1) a key usage that allows digital signatures, and 2) a subject alternative name with a specific email address:

	X509CertSelector xcs = new X509CertSelector();
	// select only unexpired certificates
	xcs.setCertificateValid(new Date());
	// select only certificates issued to
	// 'CN=mullan, O=sun, C=us'
	xcs.setSubject("CN=mullan, O=sun, C=us");
	// select only end-entity certificates
	xcs.setBasicConstraints(-2);
	// select only certificates with a digitalSignature
	// keyUsage bit set (set the first entry in the
	// boolean array to true)
	boolean[] keyUsage = {true};
	xcs.setKeyUsage(keyUsage);
	// select only certificates with a subjectAltName of
	// 'sean.mullan@sun.com' (1 is the integer value of 
	// an RFC822Name)
	xcs.addSubjectAlternativeName(1, "sean.mullan@sun.com");

Then we pass the selector to the getCertificates method of our CertStore object that we previously created:

	Collection certs = cs.getCertificates(xcs);

The X509CRLSelector Class

The X509CRLSelector class is an implementation of the CRLSelector interface that defines a set of criteria for selecting X.509 CRLs. An X509CRL object must match all of the specified criteria to be selected by the match method. The selection criteria are designed to be useful to a CertPathValidator or CertPathBuilder implementation that must retrieve CRLs from a repository to check the revocation status of certificates in an X.509 certification path.

For example, the setDateAndTime method of X509CRLSelector allows a PKIX CertPathValidator to filter out X509CRLs that have been issued after or expire before the time indicated. By setting this and other criteria in an X509CRLSelector object, it allows the CertPathValidator to discard non-relevant CRLs and more easily check if a certificate has been revoked.

Please refer to http://www.ietf.org/internet-drafts/draft-ietf-pkix-new-part1-02.txt for definitions of the X.509 CRL fields and extensions mentioned in this section.

Creating an X509CRLSelector Object

An X509CRLSelector object is created by calling the default constructor:

	public X509CRLSelector()

No criteria are initially set (any X509CRL will match).

Setting Selection Criteria

The selection criteria allow a caller to match on different components of an X.509 CRL. A few of the methods for setting selection criteria are described here, please refer to the X509CRLSelector API documentation for details on the other methods.

The setIssuerNames method sets the issuerNames criterion:

	public void setIssuerNames(Collection names)

The issuer distinguished name in the CRL must match at least one of the specified distinguished names. Each entry of the names argument is either a String or a byte array (representing the name, in RFC 2253 or ASN.1 DER encoded form, respectively). If null, any issuer distinguished name will do.

The setMinCRLNumber and setMaxCRLNumber methods set the minCRLNumber and maxCRLNumber criterion:

	public void setMinCRLNumber(BigInteger minCRL)
	public void setMaxCRLNumber(BigInteger maxCRL)

The CRL must have a CRL Number extension whose value is greater than or equal to the specified value if the setMinCRLNumber method is called, and less than or equal to the specified value if the setMaxCRLNumber method is called. If the value passed to one of these methods is null, the corresponding check is not done.

The setDateAndTime method sets the dateAndTime criterion:

	public void setDateAndTime(Date dateAndTime)

The specified date must be equal to or later than the value of the thisUpdate component of the CRL and earlier than the value of the nextUpdate component. If null, no dateAndTime check will be done.

The setCertificateChecking method sets the certificate whose revocation status is being checked:

	public void setCertificateChecking(X509Certificate cert)

This is not a criterion. Rather, it is optional information that may help a CertStore find CRLs that would be relevant when checking revocation for the specified certificate. If null is specified, then no such optional information is provided. An application should always call this method when checking revocation for a particular certificate, as it may provide the CertStore with more information for finding the correct CRLs and filtering out non-relevant ones.

Getting Selection Criteria

The current values for each of the selection criteria can be retrieved using an appropriate get method. Please refer to the X509CRLSelector API documentation for further details on these methods.

Example

Creating an X509CRLSelector to retrieve CRLs from an LDAP repository is similar to the X509CertSelector example. Suppose we want to retrieve all current (as of the current date and time) CRLs issued by a specific CA and with a minimum CRL number. First, we create an X509CRLSelector object and call the appropriate methods to set the selection criteria:

	X509CRLSelector xcrls = new X509CRLSelector();
	// select CRLs satisfying current date and time
	xcrls.setDateAndTime(new Date());
	// select CRLs issued by 'O=sun, C=us'
	xcrls.addIssuerName("O=sun, C=us");
	// select only CRLs with a CRL number at least '2'
	xcrls.setMinCRLNumber(new BigInteger("2"));

Then we pass the selector to the getCRLs method of our CertStore object (created in the X509CertSelector example):

	Collection crls = cs.getCRLs(xcrls);

PKIX Classes

The Java Certification Path API also includes a set of algorithm specific classes for use with the PKIX certification path validation algorithm defined in http://www.ietf.org/internet-drafts/draft-ietf-pkix-new-part1-02.txt. This is the third draft of a specification based upon RFC 2459. When complete, this specification will obsolete RFC 2459.

The PKIXParameters Class

This class (which implements the CertPathParameters interface) specifies the set of input parameters defined by the PKIX certification path validation algorithm. It also includes a few additional useful parameters.

An X.509 CertPath object and a PKIXParameters object are passed as arguments to the validate method of a CertPathValidator instance implementing the PKIX algorithm. The CertPathValidator uses the parameters to initialize the PKIX certification path validation algorithm.

Creating a PKIXParameters Object

To instantiate a PKIXParameters object, a caller must specify "the most-trusted CA(s)" as defined by the PKIX validation algorithm. The most-trusted CA can be specified in one of several ways. First, the caller can specify the public key, name, and optionally the unique identifier of the most-trusted CA. Alternatively, if more than one CA is trusted, a caller can specify a Set of trusted X509Certificates. Finally, a caller can specify a KeyStore instance containing trusted certificate entries, each of which will be considered as a most-trusted CA. Use one of the following constructors to create a PKIXParameters object:

	public PKIXParameters(PublicKey pubKey, String caName)
	public PKIXParameters(PublicKey pubKey, String caName, 
			      boolean[] uniqueID)
	public PKIXParameters(Set trustedCerts)
	public PKIXParameters(KeyStore keystore)

Setting Parameter Values

Once a PKIXParameters object has been created, a caller can set (or replace the current value of) various parameters. A few of the methods for setting parameters are described here, please refer to the PKIXParameters API documentation for details on the other methods.

The setInitialPolicies method sets the initial policy identifiers, as specified by the PKIX validation algorithm. The elements of the Set are object identifiers (OIDs) represented as a String. If the initialPolicies parameter is null or not set, any policy is acceptable:

	public void setInitialPolicies(Set initialPolicies)

The setDate method sets the time for which the validity of the path should be determined. If the date parameter is not set or is null, the current date is used:

	public void setDate(Date date)

The setPolicyMappingInhibited method sets the value of the policy mapping inhibited flag. The default value for the flag, if not specified, is false:

	public void setPolicyMappingInhibited(boolean val)

The setExplicitPolicyRequired method sets the value of the explicit policy required flag. The default value for the flag, if not specified, is false:

	public void setExplicitPolicyRequired(boolean val)

The setAnyPolicyInhibited method sets the value of the any policy inhibited flag. The default value for the flag, if not specified, is false:

	public void setAnyPolicyInhibited(boolean val)

The setTargetCertConstraints method allows the caller to set constraints on the target certificate. For example, the caller can specify that the target certificate must contain a specific subject name. The constraints are specified as a CertSelector object. If the selector parameter is null or not set, no constraints are defined on the target certificate:

	public void setTargetCertConstraints(CertSelector selector)

The setCertStores method allows a caller to specify a List of CertStore objects that will be used by a PKIX implementation of CertPathBuilder or CertPathValidator to find certificates and CRLs for path construction and validation. This provides an extensible mechanism for specifying where to locate certificates and CRLs. The setCertStores method takes a List of CertStore objects as a parameter. The first CertStores in the list may be preferred to those that appear later.

	public void setCertStores(List stores)

The setCertPathCheckers method allows a caller to extend the PKIX validation algorithm by creating implementation-specific certification path checkers. For example, this mechanism can be used to process private certificate extensions. The setCertPathCheckers method takes a list of PKIXCertPathChecker (discussed later) objects as a parameter:

	public void setCertPathCheckers(List checkers)

The setRevocationEnabled method allows a caller to disable revocation checking. Revocation checking is enabled by default, since it is a required check of the PKIX validation algorithm. However, PKIX does not define how revocation should be checked. An implementation may use CRLs or OCSP, for example. This method allows the caller to disable the implementation's default revocation checking mechanism if it is not appropriate. A different revocation checking mechanism can then be specified by calling the setCertPathCheckers method, and passing it a PKIXCertPathChecker that implements the alternate mechanism.

	public void setRevocationEnabled(boolean val)

The setPolicyQualifiersRejected method allows a caller to enable or disable policy qualifier processing. When a PKIXParameters object is created, this flag is set to true. This setting reflects the most common (and simplest) strategy for processing policy qualifiers. Applications that want to use a more sophisticated policy must set this flag to false.

	public void setPolicyQualifiersRejected(boolean qualifiersRejected)

Getting Parameter Values

The current values for each of the parameters can be retrieved using an appropriate get method. Please refer to the PKIXParameters API documentation for further details on these methods.

The PKIXCertPathValidatorResult Class

This class (which implements the CertPathValidatorResult interface) represents the result of the PKIX certification path validation algorithm.  It holds the valid policy tree and subject public key resulting from the validation algorithm, and includes methods (getPolicyTree() and getPublicKey()) for returning them. Instances of PKIXCertPathValidatorResult are returned by the validate method of CertPathValidator objects implementing the PKIX algorithm. The getTrustedCert method returns the Certificate of the most-trusted CA (if applicable) that was selected to anchor the path.

Please refer to the PKIXCertPathValidatorResult API documentation for more detailed information on this class.

The PolicyNode and PolicyQualifierInfo Classes

The PolicyNode interface represents a node of a valid policy tree resulting from a successful execution of the PKIX certification path validation. An application can obtain the root of a valid policy tree using the getPolicyTree method of PKIXCertPathValidatorResult. Policy Trees are discussed in more detail in the PKIX Certificate and CRL profile (http://www.ietf.org/internet-drafts/draft-ietf-pkix-new-part1-02.txt).

The getPolicyQualifiers method of PolicyNode returns a Set of PolicyQualifierInfo objects, each of which represents a policy qualifier contained in the Certificate Policies extension of the relevant certificate that this policy applies to.

Please refer to the PolicyNode and PolicyQualifierInfo API documentation for more detailed information on these classes.

Example of Validating a Certification Path using the PKIX algorithm

This is an example of validating a certification path with the PKIX validation algorithm. First, create the CertPathValidator, as in the following line:

    CertPathValidator cpv = CertPathValidator.getInstance("PKIX");

This call creates a CertPathValidator object that implements the PKIX algorithm defined in http://www.ietf.org/internet-drafts/draft-ietf-pkix-new-part1-02.txt. See Appendix A for a complete discussion of standard names and algorithms.

The next step is to create a PKIXParameters object. This will be used to populate the parameters used by the PKIX algorithm. In this example, the most-trusted CA is specified as a public key and name:

    PKIXParameters params = new PKIXParameters(pubKey, "O=sun, C=us");

Next, we populate the parameters object with constraints or other parameters used by the validation algorithm. In this example, we enable the explicitPolicyRequired flag and specify a set of initial policy OIDs:

    // set other PKIX parameters here
    params.setExplicitPolicyRequired(true);
    params.setInitialPolicies(policyIds);

The final step is to validate the certification path using the input parameter set we have created:

    PKIXCertPathValidatorResult result =
        (PKIXCertPathValidatorResult) cpv.validate(certPath, params);

If the validation algorithm is successful, the policy tree and subject public key resulting from the validation algorithm can be obtained from the result using these methods:

    PolicyNode policyTree = result.getPolicyTree();
    PublicKey subjectPublicKey = result.getPublicKey();

Otherwise, a CertPathValidatorException is thrown and the caller can catch the exception (not shown) and print the error message and which certificate caused the failure as follows:

    System.out.println("Validation failure, cert["
        + e.getIndex() + "] " + e.getMessage());

The PKIXBuilderParameters Class

This class (which extends the PKIXParameters class) specifies the set of parameters to be used with CertPathBuilders that build certification paths validated against the PKIX certification path validation algorithm.

A PKIXBuilderParameters object is passed as an argument to the build method of a CertPathBuilder instance implementing the PKIX algorithm. All PKIX CertPathBuilders must return certification paths which have been validated according to the PKIX certification path validation algorithm.

Creating a PKIXBuilderParameters Object

Creating a PKIXBuilderParameters object is similar to creating a PKIXParameters object. However, a caller must specify constraints on the target certificate when creating a PKIXBuilderParameters object. These constraints should provide the CertPathBuilder with enough information to find the target certificate. The constraints are specified as a CertSelector object. Use one of the following constructors to create a PKIXBuilderParameters object:

	public PKIXBuilderParameters(PublicKey pubKey, String caName, 
	       	CertSelector targetConstraints)
	public PKIXBuilderParameters(PublicKey pubKey, String caName, 
		boolean[] uniqueID, CertSelector targetConstraints)
	public PKIXBuilderParameters(Set trustedCerts, 
		CertSelector targetConstraints)
	public PKIXBuilderParameters(KeyStore keystore, 
		CertSelector targetConstraints)

Getting/Setting Parameter Values

The PKIXBuilderParameters class inherits all of the parameters that can be set in the PKIXParameters class. In addition, the setMaxPathLength method can be called to place a limit on the maximum number of certificates in a certification path:

	public void setMaxPathLength(int maxPathLength)

The maxPathLength parameter specifies the maximum number of CA certificates that can exist in a certification path. A CertPathBuilder instance implementing the PKIX algorithm must not build paths longer than the length specified. If the value is 0, the path can only contain a single end-entity certificate. If the value is -1, the path length is unconstrained (i.e. there is no maximum). The default maximum path length, if not specified, is 5. This method is useful to prevent the builder from spending resources and time constructing long paths that may or may not meet the caller's requirements.

If any of the CA certificates in the path contain a Basic Constraints extension, the value of the pathLenConstraint component of the extension overrides the value of the  maxPathLength parameter whenever the result is a certification path of smaller length.

The PKIXCertPathBuilderResult Class

This class (which extends the PKIXCertPathValidatorResult class and implements the CertPathBuilderResult interface) represents the result of the PKIX certification path construction algorithm.  Instances of PKIXCertPathBuilderResult are returned by the build method of CertPathBuilder objects implementing the PKIX algorithm.

The getCertPath method of a PKIXCertPathBuilderResult instance always returns a CertPath object validated using the PKIX certification path validation algorithm. The returned CertPath object does not include the most-trusted CA certificate that may have been used to anchor the path. Instead, use the getTrustedCert method to get the Certificate of the most-trusted CA.

Please refer to the PKIXCertPathBuilderResult API documentation for more detailed information on this class.

Example of Building a Certification Path using the PKIX algorithm

This is an example of of building a certification path validated against the PKIX algorithm. First, create the CertPathBuilder, as in the following example:

    CertPathBuilder cpb = CertPathBuilder.getInstance("PKIX");

This call creates a CertPathBuilder object that returns paths validated against the PKIX algorithm. See Appendix A for a complete discussion of standard names and algorithms.

The next step is to create a PKIXBuilderParameters object. This will be used to populate the PKIX parameters used by the CertPathBuilder:

    // Create parameters object, passing it a Set of
    // trusted certs for anchoring the path
    // and a target subject DN.
    X509CertSelector targetConstraints = new X509CertSelector();
    targetConstraints.setSubject("CN=mullan, O=sun, C=us");
    PKIXBuilderParameters params = 
        new PKIXBuilderParameters(trustedCerts, targetConstraints);

The next step is to specify the CertStore that the CertPathBuilder will use to look for certificates and CRLs. For this example, we will populate a Collection CertStore with the certificates and CRLs:

    CertStore store = CertStore.getInstance("Collection");
    CollectionCertStoreParameters p = new CollectionCertStoreParameters(certs);
    store.init(p);
    params.addCertStore(store);

The next step is to build the certification path using the input parameter set we have created:

    PKIXCertPathBuilderResult result = 
	(PKIXCertPathBuilderResult) cpb.build(params);

If the builder cannot build a path that meets the supplied parameters it will throw a CertPathBuilderException. Otherwise, the validated certification path can be obtained from the PKIXCertPathBuilderResult as follows:

    CertPath cp = result.getCertPath();

The PKIXCertPathChecker Class

This section describes a powerful class that allows a user to extend a PKIX CertPathValidator or CertPathBuilder implementation. This is an advanced feature that most users will not need to understand. However, anyone implementing a PKIX service provider should read this section.

The PKIXCertPathChecker class is an abstract class that executes one or more checks on an X.509 certificate. Developers should create concrete implementations of the PKIXCertPathChecker class when it is necessary to dynamically extend a PKIX CertPathValidator or CertPathBuilder implementation at runtime. The following are a few examples of when a PKIXCertPathChecker implementation is useful:

• If the revocation mechanism supplied by a PKIX CertPathValidator or CertPathBuilder implementation is not adequate. For example, a developer might implement a PKIXCertPathChecker that uses OCSP (RFC 2560) to check that certificates have not been revoked.

• If the user wants to recognize certificates containing a critical private extension. Since the extension is private, it will not be recognized by the PKIX CertPathValidator or CertPathBuilder implementation and a CertPathValidatorException will be thrown. In this case, a developer can implement a PKIXCertPathChecker that recognizes and processes the critical private extension.

• If the developer wants to record information about each certificate processed for debugging or display purposes.

• If the user wants to reject certificates with certain policy qualifiers.

The setCertPathCheckers method of the PKIXParameters class allows a user to pass a List of PKIXCertPathChecker objects to a PKIX CertPathValidator or CertPathBuilder implementation. Each of the PKIXCertPathChecker objects will be called in turn, for each certificate processed by the PKIX CertPathValidator or CertPathBuilder implementation.

Creating and using a PKIXCertPathChecker Object

The PKIXCertPathChecker class does not have a public constructor. This is intentional, since creating an instance of PKIXCertPathChecker is an implementation specific issue. For example, the constructor for a PKIXCertPathChecker implementation that uses OCSP to check a certificate's revocation status may require the hostname and port of the OCSP server:

	PKIXCertPathChecker checker = new OCSPChecker("ocsp.sun.com", 1321);

Once the checker has been instantiated, it can be added as a parameter using the addCertPathChecker method of the PKIXParameters class:

	params.addCertPathChecker(checker);

Alternatively, a List of checkers can be added using the setCertPathCheckers method of the PKIXParameters class.

Implementing a PKIXCertPathChecker Object

The PKIXCertPathChecker class is abstract. It has four methods (check, getSupportedExtensions, init, and isForwardCheckingSupported) that all concrete subclasses must implement.

Implementing a PKIXCertPathChecker may be trivial or complex. A PKIXCertPathChecker implementation can be stateless or stateful. A stateless implementation does not maintain state between successive calls of the check method. For example, a PKIXCertPathChecker that checks that each certificate contains a particular policy qualifier is stateless. In contrast, a stateful implementation does maintain state between successive calls of the check method. The check method of a stateful implementation usually depends on the contents of prior certificates in the certification path. For example, a PKIXCertPathChecker that processes the NameConstraints extension is stateful.

Also, the order in which the certificates processed by a service provider implementation are presented (passed) to a PKIXCertPathChecker is very important, especially if the implementation is stateful. Depending on the algorithm used by the service provider, the certificates may be presented in reverse or forward order. A reverse ordering means that the certificates are ordered from the most trusted CA (if present) to the target subject, whereas a forward ordering means that the certificates are ordered from the target subject to the most trusted CA. The order must be made known to the PKIXCertPathChecker implementation, so that it knows how to process consecutive certificates.

Initializing a PKIXCertPathChecker Object

The init method initializes the internal state of the checker:

	public abstract void init(boolean forward)

All stateful implementations should clear or initialize any internal state in the checker. This prevents a service provider implementation from calling a checker that is in an uninitialized state. It also allows stateful checkers to be reused in subsequent operations without reinstantiating them. The forward parameter indicates the order of the certificates presented to the PKIXCertPathChecker. If forward is true, the certificates are presented from target to trust anchor; if false, from trust anchor to target.

Supported Extensions

The getSupportedExtensions method returns an immutable Set of OID Strings for the X.509 extensions that the PKIXCertPathChecker implementation supports (i.e. recognizes, is able to process):

	public abstract Set getSupportedExtensions()

The method should return null if no extensions are processed. All implementations should return the Set of OID Strings that the check method may process.

Forward Checking

The isForwardCheckingSupported method returns a boolean that indicates if the PKIXCertPathChecker supports forward checking:

	public abstract boolean isForwardCheckingSupported()

All PKIXCertPathChecker implementations must support reverse checking. A PKIXCertPathChecker implementation may support forward checking.

Executing the Check

The following method executes a check on the certificate:

	public abstract void 
	        check(Certificate cert, Collection unresolvedCritExts)
        	throws CertPathValidatorException

The unresolvedCritExts parameter contains a collection of OIDs as Strings. These OIDs represent the set of critical extensions in the certificate that have not yet been resolved by the certification path validation algorithm. Concrete implementations of the check method should remove any critical extensions that it processes from the unresolvedCritExts parameter.

If the certificate does not pass the check(s), a CertPathValidatorException should be thrown.

Cloning a PKIXCertPathChecker

The PKIXCertPathChecker class implements the Cloneable interface. All stateful PKIXCertPathChecker implementations must override the clone method if necessary. The default implementation of the clone method calls the Object.clone method, which performs a simple clone by copying all fields of the original object to the new object. A stateless implementation should not override the clone method. However, all stateful implementations must ensure that the default clone method is correct, and override it if necessary. For example, a PKIXCertPathChecker that stores state in an array must override the clone method to make a copy of the array, rather than just a reference to the array.

The reason that PKIXCertPathChecker objects are Cloneable is to allow a PKIX CertPathBuilder implementation to efficiently backtrack and try another path when a potential certification path reaches a dead end or point of failure. In this case, the implementation is able to restore prior path validation states by restoring the cloned objects.

Example

This is an example of a stateless PKIXCertPathChecker implementation. It checks if a private extension exists in a certificate and processes it according to some rules.

	import java.security.cert.Certificate;
	import java.security.cert.X509Certificate;
	import java.util.Collection;
	import java.util.Collections;
	import java.util.Set;
	import java.security.cert.PKIXCertPathChecker;
	import java.security.cert.CertPathValidatorException;

	public class MyChecker extends PKIXCertPathChecker {
	    private static Set supportedExtensions =
	        Collections.singleton("2.16.840.1.113730.1.1");

	    /*
 	     * Initialize checker
	     */
	    public void init(boolean forward) 
		throws CertPathValidatorException {
		// nothing to initialize
	    }

	    public Set getSupportedExtensions() {        
		return supportedExtensions;
	    }

	    public boolean isForwardCheckingSupported() {
		return true;
	    }

	    /*
	     * Check certificate for presence of Netscape's
	     * private extension
	     * with OID "2.16.840.1.113730.1.1"
	     */
	    public void check(Certificate cert, Collection unresolvedCritExts)
		throws CertPathValidatorException 
	    {
		X509Certificate xcert = (X509Certificate) cert;
	        byte[] ext = 
		    xcert.getExtensionValue("2.16.840.1.113730.1.1");
		if (ext == null)
		    return;

		//
	        // process private extension according to some 
		// rules - if check fails, throw a 
		// CertPathValidatorException ...
		// {insert code here}

		// remove extension from collection of unresolved 
		// extensions (if it exists)
		if (unresolvedCritExts != null)
		    unresolvedCritExts.remove("2.16.840.1.113730.1.1");
	    }
	}

How a PKIX Service Provider implementation should use a PKIXCertPathChecker

Each PKIXCertPathChecker object must be initialized by a service provider implementation before commencing the build or validation algorithm, for example:

	List checkers = params.getCertPathCheckers();
	ListIterator li = checkers.listIterator();
	while (li.hasNext()) {
	    PKIXCertPathChecker checker = (PKIXCertPathChecker) li.next();
	    checker.init(false);
	}

For each certificate that it validates, the service provider implementation must call the check method of each PKIXCertPathChecker object in turn, passing it the certificate and any remaining unresolved critical extensions:

	ListIterator li = checkers.listIterator();
	while (li.hasNext()) {   
	    PKIXCertPathChecker checker = (PKIXCertPathChecker) li.next();
	    checker.check(cert, unresolvedCritExts);
	}

If any of the checks throw a CertPathValidatorException, the service provider implementation should terminate the validation procedure. However, a CertPathBuilder implementation may simply log the failure and continue to search for other potential paths. If all of the checks are successful, the service provider implementation should check that all critical extensions have been resolved and if not, throw an exception. For example:

	if (unresolvedCritExts != null &&
	    !unresolvedCritExts.isEmpty())
	{
	    throw new 
		CertPathValidatorException("Unrecognized Critical Extension");
	}

As discussed in the previous section, a CertPathBuilder implementation may need to backtrack when a potential certification path reaches a dead end or point of failure. Backtracking in this context implies returning to the previous certificate in the path and checking for other potential paths. If the CertPathBuilder implementation is validating the path as it is building it, it will need to restore the previous state of each PKIXCertPathChecker. It can do this by making clones of the PKIXCertPathChecker objects before each certificate is processed, for example:

	/* clone checkers */
	List newList = new ArrayList(checkers);
	ListIterator li = newList.listIterator();
	while (li.hasNext()) {   
	    PKIXCertPathChecker checker = (PKIXCertPathChecker) li.next();
	    li.set(checker.clone());
	}