The guiding force behind the OSGi Specifications is a reusable component model. The OSGi Core Release 8 provides a solid foundation for such a component model by providing a component collaboration framework with a comprehensive management model. The service specifications provide the abstract APIs to allow many different collaborations between components. This Repository Service Specification provides the capability to manage the external access to components and other resources.
Though the Repository service can be used as a standalone service to search and retrieve general binary artifacts, called resources, it is intended to be used in conjunction with the [6] Resolver Service Specification.
The model of the Repository is based on the generic Requirement-Capability model defined in [3] Resource API Specification, this chapter relies on the definitions of the generic model.
-
External - Provide access to external components and resources.
-
Resolve - The Repository API must be closely aligned with the Resolver API since they are intended to be used in conjunction.
-
Searching - Support general queries.
-
Metadata - Allow resources to provide content information.
-
Retrieval - Allow the retrieval of Resources from remote locations.
-
Batching - Repositories must be able to batch queries.
-
Distribution - Allow Repositories to be defined with a simple storage scheme such that Repositories can be distributed on a removable media like a CD/DVD.
-
Mirroring - Repositories must be able to support selecting a remote site based on the local situation.
-
Repository - A facade to a (remote) set of resources described by capabilities.
-
Resource - An artifact that has requirements that must be satisfied before it is available but provides capabilities when it becomes available.
-
Requirement - An expression that asserts a capability.
-
Capability - Describes a feature of the resource so that it can be required by a requirement.
-
Resource Content - Provides access to the underlying bytes of the resource in the default format.
There are many different repositories available on the Internet or on fixed media. A repository can be made available to bundles by providing a Repository service. If such a bundle, for example a Management Agent performing a provisioning operation, finds that it has an unmatched requirement then it can query the repository services to find matching capabilities. The Repository service can implement the query in many different ways. It can ship the requirement to a remote side to be processed or it can process the query locally.
This specification also provides an XML schema that can be used to describe a Repository. Instances of this schema can be downloaded from a remote repository for local indexing or they can be stored for example on a DVD together with the resources.
The Repository service provides an abstraction to a, potentially remote, set of resources. In the generic Capability-Requirement model, resources are modeled to declare capabilities and requirements. The primary purpose of a Repository is to enable a management agent that uses the Resolver API to leverage a wide array of repositories. This Repository service specification allows different Repository providers to be installed as bundles, and each bundle can register multiple Repository services. The Repository is sufficiently abstract to allow many different implementations.
Repository services are identified by a number of service properties:
-
service.pid
- A mandatory unique identity for this Repository service. -
service.description
- An optional human readable name for this Repository. -
repository.url
- Optional URLs to landing pages of the repository, if they exist.
In general, the users of the Repository service should aggregate all services in the service registry. This strategy allows the deployer to control the available Repositories. The following example, using Declarative Service annotations to show the dependencies on the service registry, shows how to aggregate the different Repository services.
List<Repository> repos = new CopyOnWriteArrayList<Repository>();
@Reference(
cardinality = ReferenceCardinality.MULTIPLE,
policy = ReferencePolicy.DYNAMIC)
void addRepository( Repository repo ) { repos.add(repo); }
void removeRepository( Repository repo ) { repos.remove(repo); }
To access a resource in a Repository service it is necessary to construct a requirement, pass this to the Repository service, and then use the returned capabilities to satisfy the resolver or to get the resource from the capability. The Repository then returns all matching capabilities. The requirement matches the capability if their namespaces match and the requirement's filter is absent or matches the attributes.
The findProviders(Collection) method takes a Collection of requirements. The reason for this collection is that it allows the caller to specify multiple requirements simultaneously so that Repositories can batch requests, the requirements in this collection are further unrelated. That is, they do not form an expression in any way. Multiple requirements as the parameter means that the result must be a map so that the caller can find out what requirement matched what capabilities. For example:
List<Capability> find( Requirement r ){
List<Capability> result = new ArrayList<Capability>();
for ( Repository repo : repos ) {
Map<Requirement,Collection<Capability>> answer =
repo.findProviders( Collections.singleton( r ) );
result.addAll( answer.get( r ) );
}
return result;
}
Access to resources is indirect since the Repository returns
capabilities. Each capability is declared in a resource and the
getResource()
method provides access to the underlying
resource. Since each resource declares an osgi.identity
capability it is possible to retrieve a resource from a repository if the
identity name, type, and version are known. For example, to get a bundle
resource:
Resource getResource( String type, String name, Version version ) {
String filter = String.format(
"(&(type=%s)(osgi.identity=%s)(version=%s))",
type,
name,
version );
RequirementBuilder builder = repo.newRequirementBuilder("osgi.identity");
builder.addDirective("filter", filter);
Requirement r = builder.build();
List<Capability> capabilities = find( r );
if ( capabilities.isEmpty() )
return null;
return capabilities.get( 0 ).getResource();
}
Resources that originate from Repository services must implement the RepositoryContent interface, this interface provides stream access to the default storage format. It is therefore possible to get the content with the following code.
InputStream getContent( String type, String name, Version version ) {
Resource r = getResource( type, name, version );
if ( r == null )
return null;
return ((RepositoryContent)r).getContent();
}
The getContent() method returns an Input Stream in the default
format for that resource type. Resources from a Repository should also
have one or more osgi.content
capabilities that advertise the
same resource in the same or different formats. The
osgi.content
capability has a number of attributes that
provide information about the resource's download format:
-
osgi.content
- A unique SHA-256 for the content as read from the URL. -
url
- A URL to the content. -
mime
- An IANA MIME type for the content. -
size
- Size in bytes of the content.
It is therefore possible to search for a specific MIME type and download that format. For example:
String getURL( String type, String name, Version version, String mime )
throws Exception {
Resource r = getResource( type, name, version );
for ( Capability cap : r.getCapabilities( "osgi.content") ) {
Map<String,Object> attrs = cap.getAttributes();
String actual = (String) attrs.get("mime");
if ( actual!=null && mime.equalsIgnoreCase( actual) ) {
String url = (String) attrs.get( "url" );
if ( url != null )
return url;
}
}
return null;
}
Since the osgi.content
capability contains the SHA-256
digest as the osgi.content
attribute it is possible to verify
the download that it was correct.
Every resource has an osgi.identity
capability. This
namespace defines, in [2] Framework Namespaces, the possibility to add
related resources, for example javadoc or
sources. A resource then has informational
requirements to osgi.identity
capabilities; these
requirements are marked with a classifier
directive that
holds the type of relation. The following example
shows how it would be possible to find such a related resource:
InputStream getRelated(Resource resource,String classifier)
throws Exception {
for ( Requirement r : resource.getRequirements( "osgi.identity") ) {
if ( classifier.equals( r.getDirectives().get( "classifier") ) ) {
Collection<Capability> capabilities =
repository.findProviders( Collections.singleton( r )).get( r );
if ( capabilities.isEmpty())
continue;
Capability c = capabilities.iterator().next();
Resource related = c.getResource();
return ((RepositoryContent)related).getContent();
}
}
return null;
}
In some cases it may be useful to find resources in the repository that satisfy criteria across multiple namespaces.
A simple Requirement object can contain a filter that makes
assertions about capability attributes within a single namespace. So for
example, a single requirement can state that a package
org.example.mypkg
must be exported in a version between
3.1
inclusive and 4.0
exclusive:
RequirementBuilder rb = repo.newRequirementBuilder("osgi.wiring.package");
String rf = "(&(osgi.wiring.package=org.example.mypkg)"
+ "(version>=3.1)(!(version>=4.0)))";
rb.addDirective("filter", rf);
Requirement r = rb.build();
This requirement contains three conditions on the
osgi.wiring.package
capability.
In some situations it may be needed to specify requirements that
cover multiple namespaces. For example a bundle might be needed that
exports the above package, but the bundle must also have the Apache
License, Version 2.0 license. A resource's license is available as an
attribute on the osgi.identity
namespace. Constructing a
constraint that combines requirements from multiple namespaces can be
done by using an Expression Combiner, which can be obtained from the
Repository service. The Repository service provides a findProviders(RequirementExpression) overload that can take a requirement expression
and returns a Promise to a collection of matching resources.
RequirementBuilder lb = repo.newRequirementBuilder("osgi.identity");
String lf = "(license=http://opensource.org/licenses/Apache-2.0)";
lb.addDirective("filter", lf);
RequirementExpression expr = repo.getExpressionCombiner().and(
lb.buildExpression(), rb.buildExpression());
Promise<Collection<Resource>> p = repo.findProviders(expr);
// Let findProviders() do its work async and update a ui component
// once the result is available
p.then(new Success<Collection<Resource>, Void>() {
public Promise<Void> call(Promise<Collection<Resource>> resolved)
throws Exception {
ui.update(resolved.getValue());
return null;
}
});
// Instead of the async chain above its also possiblye to
// wait for the promise value synchronously:
// Collection<Resource> resources = p.getValue();
For more details on OSGi Promises, see the Promises Specification.
A Repository service provides access to capabilities that satisfy a given requirement. A Repository can be the facade of a remote server containing a large amount of resources, a repository on removable media, or even a collection of bundles inside a ZIP file. A Repository communicates in terms of requirements and capabilities as defined in [3] Resource API Specification. This model is closely aligned with the [6] Resolver Service Specification.
A Repository service must be registered with the service properties given in the following table.
Table 132.1 Repository Service Properties
Attribute | Opt | Type | Description |
---|---|---|---|
service.pid |
mandatory |
String |
A globally unique identifier for this Repository. |
service.description |
optional |
String |
The Repository Name |
repository.url |
optional |
String+ |
URLs related to this Repository. |
The Repository implements the following methods:
-
findProviders(Collection) - For each requirement find all the capabilities that match that requirement and return them as a
Map<Requirement,Collection<Capability>>
. -
findProviders(RequirementExpression) - Find all resources that match the requirement expression. The requirement expression is used to combine multiple requirements using the
and
,or
andnot
operators. -
getExpressionCombiner() - Obtain an expression combiner. This expression combiner is used to produce requirement expressions from simple requirements or other requirement expressions.
-
newRequirementBuilder(String) - Obtain a convenience builder for Requirement objects.
A Repository must not perform any namespace specific actions or matching. The Repository must therefore match a requirement to a capability with the following rules:
-
The namespace must be identical, and
-
The requirement's filter is absent or it must match the capability's attributes.
Resources originating from a Repository service must additionally:
-
Implement the RepositoryContent interfaces, see Repository Content.
-
Provide at least one
osgi.content
Capability, see osgi.content Namespace.
Resources originating from a Repository must implement the RepositoryContent interface. The purpose of this interface is to allow users of the Repositories access to an Input Stream that provides access to the resource.
The RepositoryContent interface provides a single method:
-
getContent() - Return an Input Stream for the resource, if more than one
osgi.content
capability is present the content associated with the first capability is returned.
A resource is a logical concept, to install a resource in an environment it is necessary to get access to its contents. A resource can be formatted in different ways. It is possible to deliver a bundle as a JAR file, a Pack200 file, or some other format. In general, the RepositoryContent interface provides access to the default format.
The Repository can advertise the different formats with
osgi.content
capabilities. Each of those capabilities is
identified with a unique SHA-256 checksum and has a URL for the resource
in the specified format. The size
and mime
attributes provide information the download format, this can be used for
selection. If more than one osgi.content
capability is
associated with a resource, the first capability must represent the
default format. If the resource has a standard or widely used format
(e.g., JAR for bundles and ESA for subsystems), and that format is
provided as part of the repository, then that format should be the default
format.
The osgi.content
Namespace supports the attributes
defined in the following table and ContentNamespace.
Table 132.2 osgi.content definition
Name | Kind | M/O | Type | Syntax | Description |
---|---|---|---|---|---|
osgi.content |
CA |
M |
String |
[0-9a-fA-F]{64} |
The SHA-256 hex encoded digest for this resource |
url |
CA |
M |
String |
<url> |
The URL to the bytes. This must be an absolute URL. |
size |
CA |
M |
Long |
[0-9]+ |
The size of the resource in bytes as it will be read from the URL. |
mime |
CA |
M |
String |
<mime type> |
An IANA defined MIME type for the format of this content. |
This is an optional part of the specification since the Repository interface does not provide access how the Repository obtains its information. However, the purpose of this part of the specification is to provide a commonly recognized format for interchanging Repository metadata.
This section therefore describes an XML schema to represent Repository content. It is expected that Internet based Repositories can provide such an XML file to clients. A Repository XML file can be used as a common interchange format between multiple Repository implementations.
The Repository XML describes a number of resources with their capabilities and requirements. Additionally the XML can refer to other Repository XML files. The XML Schema can be found at its XML namespace, see XML Repository Schema. The XML structure, which closely follows the Requirement-Capability model, is depicted in Figure 132.2.
The different elements are discussed in the following sections. All types are derived from the XML Schema types, see [4] XML Schema Part 2: Data types Second Edition. Any relative URIs in a Repository XML file must be resolved as specified in [5] XML Base (Second Edition), Resolving Relative URIs.
The repository
element is the root of the document.
The repository
element has the following child
elements:
-
referral*
- Referrals to other repositories for a federated model, see Referral Element. -
resource*
- Resource definitions, see Resource Element.
The repository
element has the attributes defined in
the following table.
Table 132.3 repository element attributes
Attribute | Type | Description |
---|---|---|
name |
NCName |
The name of this Repository. For informational purposes. |
increment |
long |
Counter which increments every time the repository is changed. Can be used by clients to check for changes. The counter is not required to increase monotonically. |
The purpose of the referral
element is to allow a
Repository to refer to other Repositories, allowing for federated
Repositories. Referrals are applied recursively. However, this is not
always desired. It is therefore possible to limit the depth of
referrals. If the depth
attribute is >= 1, the referred
repository must be included but it must not follow any referrals from
the referred repository. If the depth
attribute is more
than one, referrals must be included up to the given depth. Depths of
referred repositories must also be obeyed, where referred repositories
may reduce the effective depth but not increase it. For example if a top
repository specifies a depth of 5 and a level 3 repository has a depth
of 1 then the repository on level 5 must not be used. If not specified
then there is no limit to the depth. Referrals that have cycles must be
ignored, a resource of a given Repository must only occur once in a
Repository.
The referral
element has the attributes defined in
the following table.
Table 132.4 referral element attributes
Attribute | Type | Description |
---|---|---|
depth |
int |
The max depth of referrals |
url |
anyURI |
A URL to where the referred repository XML can be found. The URL can be absolute or relative to the URI of the current XML resource. |
The resource
element defines a Resource. The
resource
element has the following child elements:
-
requirement*
- The requirements of this resource, see Requirement Element. -
capability
* - The capabilities of this resource, see Capability Element.
The Resource element has no attributes.
The capability
element maps to a capability, it holds
the attributes and directives. The capability
element has
the following child elements:
-
directive*
- The directives for the capability, see Directive Element. -
attribute*
- The attributes for the capability, see Attribute Element.
The capability
element has the attributes defined in
the following table.
Table 132.5 capability element attributes
Attribute | Type | Description |
---|---|---|
namespace |
token |
The namespace of this capability |
The requirement
element maps to a requirement, it
holds the attributes and directives. The requirement
element has the following child elements:
-
directive*
- The directives for the requirement, see Directive Element. -
attribute*
- The attributes for the requirement, see Attribute Element.
The requirement
element has the attributes defined in
the following table.
Table 132.6 requirement element attributes
Attribute | Type | Description |
---|---|---|
namespace |
token |
The namespace of this requirement |
An attribute
element describes an attribute of a
capability or requirement. Attributes are used to convey information
about the Capability-Requirement. Attributes for the capability are used
for matching the requirement's filter. The meaning of attributes is
described with the documentation of the namespace in which they
reside.
Attributes are optionally typed according to the [1] Framework Module Layer specification. The default type is
String
, the value of the value
attribute.
However, if a type
attribute is specified and it is not
String
then the value attribute must be converted according
to the type attribute specifier. The syntax of the type attribute is as
follows:
type ::= list | scalar
list ::= 'List<' scalar '>' // no spaces between terminals
scalar ::= 'String' | 'Version' | 'Long' | 'Double'
A list conversion requires the value to be broken in tokens
separated by comma (',' \u002C
). Whitespace around the list
and around commas must be trimmed for non-String types. Each token must
then be converted to the given type according to the scalar
type specifier. The exact rules for the comma separated lists are
defined in [1] Framework Module Layer, see Bundle Capability
Attributes.
The conversion of value s
, when scalar
,
must take place with the following methods:
-
String
- No conversion, uses
-
Version
-Version.parseVersion(s)
-
Long
- After trimming whitespace,Long.parseLong(s)
-
Double
- After trimming whitespace,Double.parseDouble(s)
The attribute
element has the attributes defined in
the following table.
Table 132.7 attribute element attributes
Attribute | Type | Description |
---|---|---|
name |
token |
The name of the attribute |
value |
string |
The value of the attribute. |
type |
The type of the attribute, the syntax is outlined in the previous paragraphs. |
A directive
element describes a directive of a
capability or a requirement. Directives are used to convey information
about the Capability-Requirement. The meaning of directives is described
with the documentation of the namespace in which they reside.
The directive
element has the attributes defined in
the following table.
Table 132.8 directive element attributes
Attribute | Type | Description |
---|---|---|
name |
token |
The name of the attribute |
value |
string |
The value of the attribute. |
The following example shows a very small XML file. The file contains one resource.
<repository name='OSGiRepository'
increment='13582741'
xmlns='http://www.osgi.org/xmlns/repository/v1.0.0'>
<resource>
<requirement namespace='osgi.wiring.package'>
<directive name='filter' value=
'(&(osgi.wiring.package=org.apache.commons.pool)(version>=1.5.6))'/>
</requirement>
<requirement namespace='osgi.identity'>
<directive name='effective' value='meta'/>
<directive name='resolution' value='optional'/>
<directive name='filter' value=
'(&(version=1.5.6)(osgi.identity=org.acme.pool-src))'
<directive name='classifier' value='sources'/>
</requirement>
<capability namespace='osgi.identity'>
<attribute name='osgi.identity' value='org.acme.pool'/>
<attribute name='version'type='Version' value='1.5.6'/>
<attribute name='type' value='osgi.bundle'/>
</capability>
<capability namespace='osgi.content'>
<attribute name='osgi.content' value='e3b0c44298fc1c149afbf4c8996fb92427ae41e4649b934ca495991b7852b855'
<attribute name='url' value='http://www.acme.com/repository/org/acme/pool/org.acme.pool-1.5.6.jar'/>
<attribute name='size' type='Long' value='4405'/>
<attribute name='mime' value='application/vnd.osgi.bundle'/>
</capability>
<capability namespace='osgi.wiring.bundle'>
<attribute name='osgi.wiring.bundle' value='org.acme.pool'/>
<attribute name='bundle-version' type='Version' value='1.5.6'/>
</capability>
<capability namespace='osgi.wiring.package'>
<attribute name='osgi.wiring.package' value='org.acme.pool'/>
<attribute name='version' type='Version' value='1.1.2'/>
<attribute name='bundle-version' type='Version' value='1.5.6'/>
<attribute name='bundle-symbolic-name' value='org.acme.pool'/>
<directive name='uses' value='org.acme.pool,org.acme.util'/>
</capability>
</resource>
</repository>
The namespace of this schema is:
http://www.osgi.org/xmlns/repository/v1.0.0
The schema for this namespace can be found at the location implied
in its name. The recommended prefix for this namespace is
repo
.
<schema xmlns="http://www.w3.org/2001/XMLSchema"
xmlns:repo="http://www.osgi.org/xmlns/repository/v1.0.0"
targetNamespace="http://www.osgi.org/xmlns/repository/v1.0.0"
elementFormDefault="unqualified"
attributeFormDefault="unqualified"
version="1.0.1">
<element name="repository" type="repo:Trepository" />
<complexType name="Trepository">
<sequence>
<choice minOccurs="0" maxOccurs="unbounded">
<element name="resource" type="repo:Tresource" />
<element name="referral" type="repo:Treferral" />
</choice>
<!-- It is non-deterministic, per W3C XML Schema 1.0:
http://www.w3.org/TR/xmlschema-1/#cos-nonambig
to use name space="##any" below. -->
<any namespace="##other" processContents="lax" minOccurs="0"
maxOccurs="unbounded" />
</sequence>
<attribute name="name" type="string">
<annotation>
<documentation xml:lang="en">
The name of the repository. The name may contain
spaces and punctuation.
</documentation>
</annotation>
</attribute>
<attribute name="increment" type="long">
<annotation>
<documentation xml:lang="en">
An indication of when the repository was last changed. Client's can
check if a
repository has been updated by checking this increment value.
</documentation>
</annotation>
</attribute>
<anyAttribute processContents="lax" />
</complexType>
<complexType name="Tresource">
<annotation>
<documentation xml:lang="en">
Describes a general resource with
requirements and capabilities.
</documentation>
</annotation>
<sequence>
<element name="requirement" type="repo:Trequirement" minOccurs="0" maxOccurs="unbounded"/>
<element name="capability" type="repo:Tcapability" minOccurs="1" maxOccurs="unbounded"/>
<!-- It is non-deterministic, per W3C XML Schema 1.0:
http://www.w3.org/TR/xmlschema-1/#cos-nonambig
to use name space="##any" below. -->
<any namespace="##other" processContents="lax" minOccurs="0"
maxOccurs="unbounded" />
</sequence>
<anyAttribute processContents="lax" />
</complexType>
<complexType name="Treferral">
<annotation>
<documentation xml:lang="en">
A referral points to another repository XML file. The
purpose of this element is to create a federation of
repositories that can be accessed as a single
repository.
</documentation>
</annotation>
<attribute name="depth" type="int" use="optional">
<annotation>
<documentation xml:lang="en">
The depth of referrals this repository acknowledges.
</documentation>
</annotation>
</attribute>
<attribute name="url" type="anyURI" use="required">
<annotation>
<documentation xml:lang="en">
The URL to the referred repository. The URL can be
absolute or relative from the given repository's
URL.
</documentation>
</annotation>
</attribute>
<anyAttribute processContents="lax" />
</complexType>
<complexType name="Tcapability">
<annotation>
<documentation xml:lang="en">
A named set of type attributes and directives. A capability can be
used to resolve a requirement if the resource is included.
</documentation>
</annotation>
<sequence>
<choice minOccurs="0" maxOccurs="unbounded">
<element name="directive" type="repo:Tdirective" />
<element name="attribute" type="repo:Tattribute" />
</choice>
<!-- It is non-deterministic, per W3C XML Schema 1.0:
http://www.w3.org/TR/xmlschema-1/#cos-nonambig
to use name space="##any" below. -->
<any namespace="##other" processContents="lax" minOccurs="0"
maxOccurs="unbounded" />
</sequence>
<attribute name="namespace" type="string">
<annotation>
<documentation xml:lang="en">
Name space of the capability. Only requirements with the
same name space must be able to match this capability.
</documentation>
</annotation>
</attribute>
<anyAttribute processContents="lax" />
</complexType>
<complexType name="Trequirement">
<annotation>
<documentation xml:lang="en">
A filter on a named set of capability attributes.
</documentation>
</annotation>
<sequence>
<choice minOccurs="0" maxOccurs="unbounded">
<element name="directive" type="repo:Tdirective" />
<element name="attribute" type="repo:Tattribute" />
</choice>
<!-- It is non-deterministic, per W3C XML Schema 1.0:
http://www.w3.org/TR/xmlschema-1/#cos-nonambig
to use name space="##any" below. -->
<any namespace="##other" processContents="lax" minOccurs="0"
maxOccurs="unbounded" />
</sequence>
<attribute name="namespace" type="string">
<annotation>
<documentation xml:lang="en">
Name space of the requirement. Only capabilities within the
same name space must be able to match this requirement.
</documentation>
</annotation>
</attribute>
<anyAttribute processContents="lax" />
</complexType>
<complexType name="Tattribute">
<annotation>
<documentation xml:lang="en">
A named value with an optional type that decorates
a requirement or capability.
</documentation>
</annotation>
<sequence>
<any namespace="##any" processContents="lax" minOccurs="0"
maxOccurs="unbounded" />
</sequence>
<attribute name="name" type="string">
<annotation>
<documentation xml:lang="en">
The name of the attribute.
</documentation>
</annotation>
</attribute>
<attribute name="value" type="string">
<annotation>
<documentation xml:lang="en">
The value of the attribute.
</documentation>
</annotation>
</attribute>
<attribute name="type" type="repo:TpropertyType" default="String">
<annotation>
<documentation xml:lang="en">
The type of the attribute.
</documentation>
</annotation>
</attribute>
<anyAttribute processContents="lax" />
</complexType>
<complexType name="Tdirective">
<annotation>
<documentation xml:lang="en">
A named value of type string that instructs a resolver
how to process a requirement or capability.
</documentation>
</annotation>
<sequence>
<any namespace="##any" processContents="lax" minOccurs="0"
maxOccurs="unbounded" />
</sequence>
<attribute name="name" type="string">
<annotation>
<documentation xml:lang="en">
The name of the directive.
</documentation>
</annotation>
</attribute>
<attribute name="value" type="string">
<annotation>
<documentation xml:lang="en">
The value of the directive.
</documentation>
</annotation>
</attribute>
<anyAttribute processContents="lax" />
</complexType>
<simpleType name="TpropertyType">
<restriction base="string">
<enumeration value="String" />
<enumeration value="Version" />
<enumeration value="Long" />
<enumeration value="Double" />
<enumeration value="List<String>" />
<enumeration value="List<Version>" />
<enumeration value="List<Long>" />
<enumeration value="List<Double>" />
</restriction>
</simpleType>
<attribute name="must-understand" type="boolean" default="false">
<annotation>
<documentation xml:lang="en">
This attribute should be used by extensions to documents to require that
the document consumer understand the extension. This attribute must be
qualified when used.
</documentation>
</annotation>
</attribute>
</schema>
Implementations of the Repository Service specification must provide the capabilities listed in this section.
The Repository Service implementation bundle must provide the
osgi.implementation
capability with name osgi.repository
. This capability can
be used by provisioning tools and during resolution to ensure that a
Repository Service implementation is present. The capability must also
declare a uses constraint for the
org.osgi.service.repository
package and provide the version
of this specification:
Provide-Capability: osgi.implementation;
osgi.implementation="osgi.repository";
uses:="org.osgi.service.repository";
version:Version="1.1"
This capability must follow the rules defined for the osgi.implementation Namespace.
The Repository Service implementation must provide a capability in
the osgi.service
namespace representing the Repository
service. This capability must also declare a uses constraint for the
org.osgi.service.repository
package. For example:
Provide-Capability: osgi.service;
objectClass:List<String>="org.osgi.service.repository.Repository";
uses:="org.osgi.service.repository"
This capability must follow the rules defined for the osgi.service Namespace.
Repositories in general will get their metadata and artifacts from
an external source, which makes them an attack vector for a malevolent
Bundle that needs unauthorized external access. Since a Bundle using a
Repository has no knowledge of what sources the Repository will access
it will be necessary for the Repository to implement the external access
in a doPrivileged
block. Implementations must ensure that
callers cannot influence/modify the metadata in such a way that the
getContent() method could provide access to arbitrary
Internet resources. This could for example happen if:
-
The implementation relies on the
osgi.content
namespace to hold the URL -
The attributes Map from the
osgi.content
Capability is modifiable
If the malevolent Bundle could change the osgi.content attribute it could change it to arbitrary URLs. This example should make it clear that Repository implementations must be very careful.
Implementations of this specification will need the following minimum permissions.
ServicePermission[...Repository, REGISTER ]
SocketPermission[ ... carefully restrict external access...]
Users of this specification will need the following minimum permissions.
ServicePermission[...Repository, GET ]
Repository Service Package Version 1.1.
Bundles wishing to use this package must list the package in the Import-Package header of the bundle's manifest. This package has two types of users: the consumers that use the API in this package and the providers that implement the API in this package.
Example import for consumers using the API in this package:
Import-Package: org.osgi.service.repository; version="[1.1,2.0)"
Example import for providers implementing the API in this package:
Import-Package: org.osgi.service.repository; version="[1.1,1.2)"
-
AndExpression
- A RequirementExpression representing theand
of a number of requirement expressions. -
ContentNamespace
- Content Capability and Requirement Namespace. -
ExpressionCombiner
- AnExpressionCombiner
can be used to combine requirement expressions into a single complex requirement expression using theand
,or
andnot
operators. -
IdentityExpression
- A RequirementExpression representing a requirement. -
NotExpression
- A RequirementExpression representing thenot
(negation) of a requirement expression. -
OrExpression
- A RequirementExpression representing theor
of a number of requirement expressions. -
Repository
- A repository service that contains resources. -
RepositoryContent
- An accessor for the content of a resource. -
RequirementBuilder
- A builder for requirements. -
RequirementExpression
- The super interface for all requirement expressions.
A RequirementExpression representing the and
of a number of
requirement expressions.
1.1
Thread-safe
Consumers of this API must not implement this type
Return the requirement expressions that are combined by this
AndExpression
.
An unmodifiable list of requirement expressions that are combined
by this AndExpression
. The list contains the requirement
expressions in the order they were specified when this
requirement expression was created.
Content Capability and Requirement Namespace.
This class defines the names for the attributes and directives for this
namespace. All unspecified capability attributes are of type String
and are used as arbitrary matching attributes for the capability. The values
associated with the specified directive and attribute keys are of type
String
, unless otherwise indicated.
Immutable
The capability attribute that defines the IANA MIME Type/Format for this content.
The capability attribute that contains the size, in bytes, of the
content. The value of this attribute must be of type Long
.
The capability attribute that contains the URL to the content.
An ExpressionCombiner
can be used to combine requirement expressions
into a single complex requirement expression using the and
,
or
and not
operators.
1.1
Thread-safe
Consumers of this API must not implement this type
The first requirement expression to combine into the returned requirement expression.
The second requirement expression to combine into the returned requirement expression
Combine two RequirementExpressions into a requirement expression
using the and
operator.
An AndExpression representing an and
of the
specified requirement expressions.
The first requirement expression to combine into the returned requirement expression.
The second requirement expression to combine into the returned requirement expression
Optional, additional requirement expressions to combine into the returned requirement expression.
Combine multiple RequirementExpressions into a requirement
expression using the and
operator.
An AndExpression representing an and
of the
specified requirement expressions.
The requirement to wrap in a requirement expression.
Wrap a Requirement in an IdentityExpression. This can be
useful when working with a combination of Requirement
s and
RequirementExpresion
s.
An IdentityExpression representing the specified requirement.
The requirement expression to negate.
Return the negation of a RequirementExpression.
A NotExpression representing the not
of the
specified requirement expression.
The first requirement expression to combine into the returned requirement expression.
The second requirement expression to combine into the returned requirement expression
Combine two RequirementExpressions into a requirement expression
using the or
operator.
An OrExpression representing an or
of the
specified requirement expressions.
The first requirement expression to combine into the returned requirement expression.
The second requirement expression to combine into the returned requirement expression
Optional, additional requirement expressions to combine into the returned requirement expression.
Combine multiple RequirementExpressions into a requirement
expression using the or
operator.
An OrExpression representing an or
of the
specified requirement expressions.
A RequirementExpression representing a requirement.
1.1
Thread-safe
Consumers of this API must not implement this type
A RequirementExpression representing the not
(negation) of a
requirement expression.
1.1
Thread-safe
Consumers of this API must not implement this type
A RequirementExpression representing the or
of a number of
requirement expressions.
1.1
Thread-safe
Consumers of this API must not implement this type
Return the requirement expressions that are combined by this
OrExpression
.
An unmodifiable list of requirement expressions that are combined
by this OrExpression
. The list contains the requirement
expressions in the order they were specified when this
requirement expression was created.
A repository service that contains resources.
Repositories may be registered as services and may be used as by a resolve context during resolver operations.
Repositories registered as services may be filtered using standard service properties.
Thread-safe
Consumers of this API must not implement this type
Service property to provide URLs related to this repository.
The value of this property must be of type String
,
String[]
, or Collection<String>
.
The requirements for which matching capabilities
should be returned. Must not be null
.
Find the capabilities that match the specified requirements.
A map of matching capabilities for the specified requirements.
Each specified requirement must appear as a key in the map. If
there are no matching capabilities for a specified requirement,
then the value in the map for the specified requirement must be
an empty collection. The returned map is the property of the
caller and can be modified by the caller. The returned map may be
lazily populated, so calling size()
may result in a long
running operation.
The RequirementExpression
for which matching
capabilities should be returned. Must not be null
.
Find the resources that match the specified requirement expression.
A promise to a collection of matching Resource
s. If there
are no matching resources, an empty collection is returned. The
returned collection is the property of the caller and can be
modified by the caller. The returned collection may be lazily
populated, so calling size()
may result in a long running
operation.
1.1
Return an expression combiner. An expression combiner can be used to combine multiple requirement expressions into more complex requirement expressions using and, or and not operators.
An ExpressionCombiner
.
1.1
The namespace for the requirement to be created.
Return a new RequirementBuilder
which provides a convenient way
to create a requirement.
For example:
Requirement myReq = repository.newRequirementBuilder("org.foo.ns1").
addDirective("filter", "(org.foo.ns1=val1)").
addDirective("cardinality", "multiple").build();
A new requirement builder for a requirement in the specified namespace.
1.1
An accessor for the content of a resource.
All Resource objects which represent resources in a
Repository must implement this interface. A user of the resource can
then cast the Resource object to this type and then obtain an
InputStream
to the content of the resource.
Thread-safe
Consumers of this API must not implement this type
Returns a new input stream to the content of this resource. The content
is represented on the resource through the osgi.content
capability. If more than one such capability is associated with the
resource, the first such capability is returned.
A new input stream for associated content.
A builder for requirements.
1.1
Consumers of this API must not implement this type
The attribute name.
The attribute value.
Add an attribute to the set of attributes.
This requirement builder.
The directive name.
The directive value.
Add a directive to the set of directives.
This requirement builder.
Create a requirement based upon the values set in this requirement builder.
A requirement created based upon the values set in this requirement builder.
Create a requirement expression for a requirement based upon the values set in this requirement builder.
A requirement expression created for a requirement based upon the values set in this requirement builder.
The map of attributes.
Replace all attributes with the attributes in the specified map.
This requirement builder.
The map of directives.
Replace all directives with the directives in the specified map.
This requirement builder.
[4]XML Schema Part 2: Data types Second Editionhttp://www.w3.org/TR/xmlschema-2/
[5]XML Base (Second Edition), Resolving Relative URIshttps://www.w3.org/TR/xmlbase/#resolution