SOAP::Serializer - the means by which the toolkit manages the expression of data as XML

The SOAP::Serializer class is the means by which the toolkit manages the expression of data as XML. The object that a SOAP::Lite instance uses by default is generally enough for the task, with no need for the application to create its own. The main purpose of this class is to provide a place for applications to extend the serializer by defining additional methods for handling new datatypes.

new(optional key/value pairs)

    $serialize = SOAP::Serializer->new( );

This is the constructor method for the class. In addition to creating a basic object and initializing it with default values, the constructor can also take names and values for most of the accessor methods that the class supports.

envelope(method, data arguments)

    $serialize->envelope(fault => $fault_obj);

Provides the core purpose for the SOAP::Serializer class. It creates the full SOAP envelope based on the input passed in to it. The data arguments passed in the list of parameters to the method are divided into two sublists: any parameters that are SOAP::Header objects or derivatives of go into one list, while the remainder go into the other. The nonheader objects are used as the content for the message body, with the body itself being largely dependent on the value of the first argument in the list. This argument is expected to be a string and should be one of the following:

context

    $serialize->context->packager();

This provides access to the calling context of CWSOAP::Serializer. In a client side context the often means a reference to an instance of SOAP::Lite. In a server side context this means a reference to a SOAP::Server instance.

method

The envelope is being created to encapsulate a RPC-style method call.

response

The message being created is that of a response stemming from a RPC-style method call.

fault

For this specifier, the envelope being created is to transmit a fault.

freeform

This identifier is used as a general-case encoding style for messages that don't fit into any of the previous cases. The arguments are encoded into the envelope's Body tag without any sort of context sensitivity. Any value other than these four results in an error.

envprefix(optional value)

    $serialize->envprefix('env');

Gets or sets the prefix that labels the SOAP envelope namespace. This defaults to SOAP-ENV.

encprefix(optional value)

    $serialize->envprefix('enc');

Gets or sets the prefix that labels the SOAP encoding namespace. Defaults to SOAP-ENC.

soapversion(optional value)

    $serial->soapversion('1.2');

If no parameter is given, returns the current version of SOAP that is being used as the basis for serializing messages. If a parameter is given, attempts to set that as the version of SOAP being used. The value should be either 1.1 or 1.2. When the SOAP version is being set, the package selects new URNs for envelope and encoding spaces and also calls the xmlschema method to set the appropriate schema definition.

xmlschema(optional value)

    $serial->xmlschema($xml_schema_1999);

Gets or sets the URN for the schema being used to express the structure of the XML generated by the serializer. If setting the value, the input must be the full URN for the new schema and is checked against the list of known SOAP schemas.

register_ns

The register_ns subroutine allows users to register a global namespace with the SOAP Envelope. The first parameter is the namespace, the second parameter to this subroutine is an optional prefix. If a prefix is not provided, one will be generated automatically for you. All namespaces registered with the serializer get declared in the <soap:Envelope /> element.

find_prefix

The find_prefix subroutine takes a namespace as a parameter and returns the assigned prefix to that namespace. This eliminates the need to declare and redeclare namespaces within an envelope. This subroutine is especially helpful in determining the proper prefix when assigning a type to a SOAP::Data element. A good example of how this might be used is as follows:

    SOAP::Data->name("foo" => $inputParams{'foo'})
              ->type($client->serializer->find_prefix('urn:Foo').':Foo');

When serializing an object, or blessed hash reference, into XML, CWSOAP::Serializer first checks to see if a subroutine has been defined for the corresponding class name. For example, in the code below, CWSOAP::Serializer will check to see if a subroutine called CWas_MyModule__MyPackage has been defined. If so, then it will pass CW$foo to that subroutine along with other data known about the CWSOAP::Data element being encoded.

   $foo = MyModule::MyPackage->new;
   my $client = SOAP::Lite
      ->uri($NS)
      ->proxy($HOST);
   $som = $client->someMethod(SOAP::Data->name("foo" => $foo));

Naming Convention

The subroutine should always be prepended with CWas_ followed by the type's name. The type's name must have all colons (':') substituded with an underscore ('_').

Input

The input to CWas_TypeName will have at least one parameter, and at most four parameters. The first parameter will always be the value or the object to be encoded. The following three parameters depend upon the context of the value/object being encoded. If the value/object being encoded was part of a CWSOAP::Data object (as in the above example), then the second, third and fourth parameter will be the CWSOAP::Data element's name, type, and attribute set respectively. If on the other hand, the value/object being encoded is not part of a CWSOAP::Data object, as in the code below:

   $foo = MyModule::MyPackage->new;
   my $client = SOAP::Lite
      ->uri($NS)
      ->proxy($HOST);
   $som = $client->someMethod($foo);

Then the second and third parameters will be the class name of the value/object being encoded (e.g. MyModule::MyPackage in the example above), and the fourth parameter will be an empty hash.

Output

The encoding subroutine must return an array containing three elements: 1) the name of the XML element, 2) a hash containing the attributes to be placed into the element, and 3) the value of the element.

When the type of an element has not been declared explicitly, SOAP::Lite must guess at the object's type. That is due to the fact that the only form of introspection that Perl provides (through the use of the CWref subroutine) does not provide enough information to CWSOAP::Serializer to allow SOAP::Lite to determine the exact type of an element being serialized.

To work around this limitation, the CWSOAP::Serializer::typelookup hash was created. This hash is populated with all the data types that the current CWSOAP::Serializer can auto detect. Users and developers are free to modify the contents of this hash allowing them to register new data types with the system.

When CWSOAP::Serializer is asked to encode an object into XML, it goes through the following steps. First, CWSOAP::Serializer checks to see if a type has been explicitly stated for the current object. If a type has been provided CWSOAP::Serializer checks to see if an CWas_TypeName subroutine as been defined for that type. If such a subroutine exists, then CWSOAP::Serializer passes the object to it to be encoded. If the subroutine does not exist, or the type has not been provided, then CWSOAP::Serializer must attempt to guess the type of the object being serialized.

To do so, CWSOAP::Serializer runs in sequence a set of tests stored in the CWSOAP::Serializer::typelookup hash. CWSOAP::Serializer continues to run each test until one of the tests returns true, indicating that the type of the object has been detected. When the type of the object has been detected, then CWSOAP::Serializer passes the object to the encoding subroutine that corresponds with the test that was passed. If all the tests fail, and the type was not determined, then CWSOAP::Serializer will as a last resort encode the object based on one of the four basic data types known to Perl: REF, SCALAR, ARRAY and HASH.

The following table contains the set of data types detectable by CWSOAP::Lite by default and the order in which their corresponding test subroutine will be run, according to their precedence value.

  Table 1 - Autotyping Precedence

  TYPENAME    PRECEDENCE VALUE
  ----------------------------
  base64      10
  int         20
  long        25 
  float       30
  gMonth      35
  gDay        40
  gYear       45
  gMonthDay   50
  gYearMonth  55
  date        60
  time        70
  dateTime    75
  duration    80
  boolean     90
  anyURI      95
  string      100

To register a new data type that can be automatically detected by CWSOAP::Lite and then serialized into XML, the developer must provide the following four things:

•

The name of the new data type.

•

A subroutine that is capable of detecting whether a value passed to it is of the corresponding data type.

•

A number representing the test subroutine's precedence relative to all the other types' test subroutinestypes. See Table 1 - Autotyping Precedence.

•

A subroutine that is capable of providing CWSOAP::Serializer with the information necessary to serialize an object of the corresponding data type into XML.

EXAMPLE 1

If, for example, you wish to create a new datatype called CWuriReference for which you would like Perl values to be automatically detected and serialized into, then you follow these steps.

Step 1: Write a Test Subroutine

The test subroutine will have passed to it by CWSOAP::Serializer a value to be tested. The test subroutine must return 1 if the value passed to it is of the corresponding type, or else it must return 0.

    sub SOAP::Serializer::uriReferenceTest {
      my ($value) = @_;
      return 1 if ($value =~ m!^http://!);
      return 0;
    }

Step 2: Write an Encoding Subroutine

The encoding subroutine provides CWSOAP::Serializer with the data necessary to encode the value passed to it into XML. The encoding subroutine name's should be of the following format: CWas_<Type Name>.

The encoding subroutine will have passed to it by CWSOAP::Serializer four parameters: the value to be encoded, the name of the element being encoded, the assumed type of the element being encoded, and a reference to a hash containing the attributes of the element being encoded. The encoding subroutine must return an array representing the encoded datatype. CWSOAP::Serializer will use the contents of this array to generate the corresponding XML of the value being encoded, or serialized. This array contains the following 3 elements: the name of the XML element, a hash containing the attributes to be placed into the element, and the value of the element.

  sub SOAP::Serializer::as_uriReference {
    my $self = shift;
    my($value, $name, $type, $attr) = @_;
    return [$name, {'xsi:type' => 'xsd:uriReference', %$attr}, $value];
  }

Step 3: Register the New Data Type

To register the new data type, simply add the type to the CWSOAP::Serializer::typelookup hash using the type name as the key, and an array containing the precedence value, the test subroutine, and the encoding subroutine.

  $s->typelookup->{uriReference} 
      = [11, \&uriReferenceTest, 'as_uriReference'];

Tip: As a short hand, you could just as easily use an anonymous test subroutine when registering the new datatype in place of the CIurlReferenceTest subroutine above. For example:

  $s->typelookup->{uriReference} 
      = [11, sub { $_[0] =~ m!^http://! }, 'as_uriReference'];

Once complete, CWSOAP::Serializer will be able to serialize the following CWSOAP::Data object into XML:

  $elem = SOAP::Data->name("someUri" => 'http://yahoo.com')->type('uriReference');

CWSOAP::Serializer will also be able to automatically determine and serialize the following untyped CWSOAP::Data object into XML:

  $elem = SOAP::Data->name("someUri" => 'http://yahoo.com');

Special thanks to O'Reilly publishing which has graciously allowed SOAP::Lite to republish and redistribute large excerpts from Programming Web Services with Perl, mainly the SOAP::Lite reference found in Appendix B.

Copyright (C) 2000-2004 Paul Kulchenko. All rights reserved.

This library is free software; you can redistribute it and/or modify it under the same terms as Perl itself.

Paul Kulchenko (paulclinger@yahoo.com)

Randy J. Ray (rjray@blackperl.com)

Byrne Reese (byrne@majordojo.com)