The Shibboleth IdP V4 software has reached its End of Life and is no longer supported. This documentation is available for historical purposes only. See the IDP5 wiki space for current documentation on the supported version.
HTTPConnector
Namespace: urn:mace:shibboleth:2.0:resolver
Schema: http://shibboleth.net/schema/idp/shibboleth-attribute-resolver.xsd
The HTTP
data connector generates multiple attributes resulting from calling a web service. The connector itself is primarily just a framework for constructing a request to issue using the Apache HttpClient library and processing a response. There is no attempt to standardize a web service interface, but the design is geared towards making simple, templated GET requests and processing XML or JSON result sets using a script. More advanced use cases can be supported by means of pluggable Java interfaces or more advanced scripting.
General Configuration
The connector marries three essential pieces of configuration to be supplied by the deployer:
an HttpClient bean and various security and networking parameters necessary to invoke the web service safely
request creation
response processing
Two options for request creation exist, one for simple GET requests and one for more complex POST requests with a body. Both rely on a <URLTemplate> element to produce the URL, which can be populated with dependency data much like an LDAP search filter or database query. The POST option also supports a <BodyTemplate> element that can similarly generate a request body based on dependency data, typically JSON or XML. With POST, you also can control caching of results (if a results cache is used) by generating a cache entry key using a <CacheKeyTemplate> element.
The only supplied implementation for response processing is script-based, using the <ResponseMapping> element to supply a script to produce attribute data from the response body. For efficiency, the implementation assumes that the response processing script can consume the results in real time, and leaves any buffering of the data to the script, should that be necessary. Support is built-in for limiting the size of the data, allowed content types, and HTTP status codes accepted, to address the most common sanity checks.
At minimum, an httpClientRef
attribute must be supplied to provide the client runtime bean to use. This will usually if not always be accompanied by an httpClientSecurityParametersRef
attribute to supply security settings, although a few shortcut settings are available for certificate authentication use cases. A complete summary and examples can be found on the HttpClientConfiguration page.
Spring beans may be defined either in additionally loaded Spring resources configured into the Attribute Resolver service's resource collection, or in a central location such as global.xml
Caching Behavior
The built-in caching support (if enabled) keys the cached data for a GET request by the fully populated template URL generated, typically including information about the subject in the URL. For POST requests, you may provide a template for producing an appropriate cache key, or omit it to disable caching.
Customized implementations of the ExecutableSearchBuilder<HTTPSearch> interface MUST implement appropriate cache key semantics, particularly if they inherit from that base class.
Examples
Any examples here omit the Spring beans that define the HTTP client and security parameters to use. Examples of these can be found in the HttpClientConfiguration topic.
Reference
Spring Configuration
The springResource
or springResourceRef
attributes are DEPRECATED in V4.3 and will not work in V5
If the springResource
or springResourceRef
attributes are specified, then the configuration of the data connector bean is delegated to the supplied resources. The system will create a factory for an HTTPDataConnector object, and look for beans in the Spring resource(s) supplied that match the types of properties supported by that type and its parent classes. Note that since these are not public, but implementation classes, they are subject to change, which creates some risk during non-patch upgrades, so you must take additional precautions to use this feature.
In practice, the HTTP Data Connector may be supplied with beans of the following types:
ExecutableSearchBuilder<HTTPSearch>
Note that there are no classes available which implement this interface
com.google.common.cache.Cache<String,Map<String,IdPAttribute>>
The only available class which implements this Interface is NonFailFastValidator
Note that there are no classes available which implement this interface
In addition native bean IDs can be injected as follows:
The HttpClient instance and its security settings are injected via the
httpClientRef
andhttpClientSecurityParametersRef
attributes.The builder for the request can be specified as an externally defined bean via the
executableSearchBuilderRef
attribute (as a replacement for the<URLTemplate>
element and related elements). This allows for complete generality of the request-building process.The processing of the response can be specified with an externally defined bean via the
mappingStrategyRef
attribute (as a replacement for the<ResponseMapping>
element).The caching of results can be specified as an externally defined bean via the
<ResultCacheBean>
element (as a replacement for the<ResultCache>
element).A Validator can be specifier as as an externally defined bean via the
validatorRef
attribute. The only available class which implements this Interface is NonFailFastValidatorRarely, a non-default Velocity engine can be injected via the
templateEngine
attribute.