NAME

Net::ISBNDB::Agent - Base class for data-retrieval agents

SYNOPSIS

package Net::ISBNDB::Agent::REST;

use strict;
use warnings;
use base 'Net::ISBNDB::Agent';

DESCRIPTION

The Net::ISBNDB::Agent class is a base class for all the classes that provide actual communication protocol support for the Net::ISBNDB::API module. Unlike the API class, this class is not usable on its own except as a factory to create instances of classes that derive from it.

The agent classes are responsible for actually setting up the web requests to retrieve data, parsing the results of those calls, and returning the data in formats usable by the API classes.

This class (and all sub-classes of it) are based on the Class::Std inside-out objects pattern. See Class::Std for more detail.

All error conditions in the methods of this class are handled using the exception model provided by the Error module. Most errors are thrown in the form of Error::Simple exception objects. See Error for more detail.

METHODS

The following methods are implemented (or in some cases, stubbed) in this base class. In some cases, the method requires that an implementation class override it in order to work. This is noted when it applies.

Constructor

The constructor for this class should only ever be called to act as a factory constructor. Instantiating this class directly is not permitted, and will cause an exception to be thrown.

new($PROTO [ , $ARGS ]): Create a new object of the specified protocol and return a referent to it. If $args is passed, it is passed along to the protocol class' constructor. If the protocol referred to by $PROTO is unknown, or if $PROTO is not passed, an exception will be thrown.

Accessors

The accessor methods are used to set and retrieve the attributes (instance data) stored on the object. While a few of them have special behavior, most operate as simple get or set accessors as described in Class::Std. The attributes for this class are:

agent: The user-agent (an instance or derivative of LWP::UserAgent) used to make all the HTTP requests to the service. Unless explicitly provided by the user, this value is not initialized until the first request made to it. The next attribute allows the user to specify arguments to the constructor when the object is finally instantiated.
agent_args: A hash reference of parameters to be passed to the constructor of the agent when it is created. The LWP::UserAgent constructor takes ordinary key/value pairs as arguments, not a single hash reference like classes derived from Class::Std. This value will be "flattened" when the constructor is called. Value within it, however, will not be. You must make certain that the values for any keys specified match the expected format within LWP::UserAgent.

The following accessor methods are provided by this class:

get_agent: Retrieve the user-agent this object uses for HTTP communication. The creation of this object is delayed until the first request to fetch it (unless the user has explicitly set the agent, or provided an agent in the construction of the Net::ISBNDB::Agent-derived object).
set_agent($AGENT): Explicitly set the user-agent for this object to use. The new value must be an instance of LWP::UserAgent, or an object of a class that is derived from that one. If it isn't, an exception will be thrown.
get_agent_args: Get the current value of the arguments used in the creation of a user-agent instance. If none have been set, the value returned will be undef.
set_agent_args($ARGS): Set a new hash reference of arguments to be used when the user-agent is instantiated. Note that the object will instantiate the user-agent at most once, so setting this after the first call to get_agent (or after explicitly setting the agent attribute) will have no effect. The value of $ARGS must be a hash reference, or an exception will be thrown.

Managing Protocols

For communication protocols, the only built-in protocol is:

REST: The web services protocol known as REpresentational State Transfer, this protocol uses the URL exclusively for data-fetch operations (which, since isbndb.com is a read-only source, is all this module does). All parameters for searches and data retrieval are passed as query parameters in the URL in the request.

All protocol names are treated as upper-case strings. The values are forced to upper-case within the following methods:

add_protocol($PROTO, $CLASS): Add a mapping of the new protocol specified by $PROTO to the class given as $CLASS. As with types, you can use this to override the class that will be instantiated for any of the built-in protocols. You cannot delete mappings for any core protocols, so if you wish to temporarily override the class, you must save the existing map value (with class_for_protocol, below) and re-assign it yourself.
class_for_protocol($PROTO): Returns the class-name for the given protocol. If $PROTO is not know, then an exception (of type Error::Simple ) is thrown.
remove_protocol($PROTO): Removes the mapping for $PROTO from the internal table. You cannot remove the mapping for a core protocol (an exception will be thrown if you try to). You can only override it with another call to add_protocol.
protocol([$PROTO]): Return the protocol this object implements, or test a given string to see if to matches the implemented protocol. If $PROTO is passed in, the method will return either a true or false value, depending on whether the value matches the protocol of this object. If $PROTO is not passed in, the return value is a string representation of the module's protocol.

The protocol-oriented methods are intended for the future, if/when isbndb.com should offer other methods besides REST. These class methods can be used by separate modules to register their protocols with this class.

As with the type-map methods in Net::ISBNDB::API, all of the protocol-map methods may be called as static methods.

Making Requests

The role of the agent classes is to make the requests for data from the isbndb.com service, parse the body of the response and convert that data to objects from the Net::ISBNDB::API hierarchy. To do this, this base class provides methods for making the requests, which themselves are composed of several methods restricted to the Net::ISBNDB::Agent hierarchy.

The methods are:

request_single($OBJ, $ARGS)

Make a request of the service, returning a single object as a result. $OBJ controls the type of object returned, as well as the type of request sent. $OBJ can be one of three types of values:

Net::ISBNDB::API-derived object: If the value is an object from one of the API classes (excluding Net::ISBNDB::API itself), it is used not only to control the type of request, but it is also overwritten with the result of the request. It is also the return value of the call when successful.
Type name: If the value is a type recognized by the Net::ISBNDB::API class, the class itself is retrieved via the class_for_type method. That class is used to provide the type-specific data that would otherwise be retrieved through an existing object, and it is used to instantiate the new object with the data returned by the request.
Class name: If the value is a full class name, it is first tested to see that the class is a decendant of Net::ISBNDB::API. If so, it is used in the same way as the class derived from the previous case.

The $ARGS parameter provides the arguments used in making the specific request. It is a hash reference, whose keys and values are dependent on the specific implementation class.

If the request returns more than one value, the first one is taken and the rest are discarded. If the request returns no data, undef is returned. If there is an error of any sort, an exception is thrown.

request_all($OBJ, $ARGS)

This method sends the request, and returns all the resulting records from the service. The arguments and behavior are identical to that of request_single, except that the return value includes all records returned by the query.

Presently, the return value for a successful query is an array reference containing the objects representing the matched records. This reference may contain only one object, or even none, depending on whether the query returned any data. In future versions, the return value will be an iterator that manages the list internally, for faster response time and better memory usage.

resolve_obj($OBJ) (R)

Resolves the disposition of the argument $OBJ. This is what gets called by request_single and request_all to determine how to interpret the first argument.