|
|
SOAP::Lite - Perl's Web Services Toolkit
SOAP::Lite is a collection of Perl modules which provides a simple and lightweight interface to the Simple Object Access Protocol (SOAP) both on client and server side.
the SOAP::Lite manpage - Main class provides all logic
the SOAP::Transport manpage - Supports transport architecture
the SOAP::Data manpage - Provides extensions for serialization architecture
the SOAP::Header manpage - Provides extensions for header serialization
SOAP::Parser - Parses XML file into object tree
the SOAP::Serializer manpage - Serializes data structures to SOAP package
SOAP::Deserializer - Deserializes results of SOAP::Parser into objects
the SOAP::SOM manpage - Provides access to deserialized object tree
the SOAP::Constants manpage - Provides access to common constants
the SOAP::Trace manpage - Provides tracing facilities
the SOAP::Schema manpage - Provides access and stub(s) for schema(s)
SOAP::Schema::WSDL - WSDL implementation for SOAP::Schema
the SOAP::Server manpage - Handles requests on server side
SOAP::Server::Object - Handles objects-by-reference
the SOAP::Fault manpage - Provides support for Faults on server side
the SOAP::Utils manpage - A set of private and public utility subroutines
the SOAP::Packager manpage - Provides an abstract class for implementing custom packagers.
SOAP::Packager::MIME - Provides MIME support to SOAP::Lite
SOAP::Packager::DIME - Provides DIME support to SOAP::Lite
SOAP::Transport::HTTP::Client - Client interface to HTTP transport
SOAP::Transport::HTTP::Server - Server interface to HTTP transport
SOAP::Transport::HTTP::CGI - CGI implementation of server interface
SOAP::Transport::HTTP::Daemon - Daemon implementation of server interface
SOAP::Transport::HTTP::Apache - mod_perl implementation of server interface
SOAP::Transport::POP3::Server - Server interface to POP3 protocol
SOAP::Transport::MAILTO::Client - Client interface to SMTP/sendmail
SOAP::Transport::LOCAL::Client - Client interface to local transport
SOAP::Transport::TCP::Server - Server interface to TCP protocol
SOAP::Transport::TCP::Client - Client interface to TCP protocol
SOAP::Transport::IO::Server - Server interface to IO transport
The first group of methods presented are the constructor and the accessor methods. All accessor methods share the trait of returning the current appropriate value when called with no arguments, while returning the object reference itself when called with a new value for the field in question. This allows the set-attribute calls to be chained together.
$client = SOAP::Lite->new(proxy => $endpoint)
This is the constructor of the class. Many of the accessor methods defined here may be initialized at creation by providing their name as a key, followed by the desired value. The example provides the value for the proxy element of the client.
$transp = $client->transport( );
Provides access to the transport object that the client has allocated to manage the communication layer operations. You can set this by passing a new object that derives from SOAP::Transport, but this is generally not needed or recommended. Several of the following methods are shortcuts to this object's accessors.
$serial = $client->serializer( )
Provides access to the SOAP::Serializer object that the client uses to transform the elements and data of a request into an XML document for the sake of transport. As with transport, this may be set by providing a new object reference, but it is generally not needed.
$packager = $client->packager( )
Provides access to the SOAP::Packager object that the client uses to manage the use of attachments. The default packager is a MIME packager, but unless you specify parts to send, no MIME formatting will be done.
See also: the SOAP::Packager manpage.
$client->proxy('http://soap.xml.info/ endPoint');
The proxy is the server or endpoint to which the client is going to connect. It shouldn't be confused with the uri method discussed later, which refers to a different element of the conversation. This method allows the setting of the endpoint, along with any extra information that the transport object may need when communicating the request. Indeed, this method is actually an alias to the proxy method of the SOAP::Transport manpage. It is the same as typing:
$client->transport( )->proxy(...arguments);
When extra information is needed, it is also passed in the call to this method. Connecting to a server that uses browser cookies for authentication can be done by creating an instance of the HTTP::Cookies class (from the LWP package) and passing it as the value following a key of cookie_jar. The value for sockettime-outs may also be set this way. The full range of options vary by transport method. One common theme is that the endpoint string is always the first argument, with all additional arguments following it.
The following is a list of optional arguments that may be passed to proxy():
This is for SOAP::Lite specific options. Currently the only option one is allowed to set is the compress_threshold option. See COMPRESSION in the HTTP::Transport manpage.
For example, if you wish to set the HTTP timeout for a SOAP::Lite client to 5 seconds, use the following code:
my $soap = SOAP::Lite ->uri($uri) ->proxy($proxyUrl, timeout => 5 );
See the LWP::UserAgent manpage.
$client->endpoint('http://soap.xml.info/ newPoint')
It may be preferable to set a new endpoint without the additional work of examining the new address for protocol information and checking to ensure the support code is loaded and available. This method allows the caller to change the endpoint that the client is currently set to connect to, without reloading the relevant transport code. Note that the proxy method must have already been called before this method is used.
$client->service('http://svc.perl.org/Svc.wsdl');
SOAP::Lite offers some support for creating method stubs from service descriptions. At present, only WSDL support is in place. This method loads the specified WSDL schema and uses it as the basis for generating stubs.
outputxml(boolean)
$client->outputxml('true');
Controls whether the returned information from a remote method call is the raw XML from the server. The default is to process the data from the server and present it to the caller as an object of the the SOAP::SOM manpage class. If the application prefers to use a different parser or do something else entirely with the results, this method may be used to inhibit the parsing of the returned information.
autotype(boolean)
$client->autotype(0);
This method is a shortcut for:
$client->serializer->autotype(boolean);
By default, the serializer tries to automatically deduce types for the data being sent in a message. Setting a false value with this method disables the behavior.
readable(boolean)
$client->readable(1);
This method is a shortcut for:
$client->serializer->readable(boolean);
When this is used to set a true value for this property, the generated XML sent to the endpoint has extra characters (spaces and new lines) added in to make the XML itself more readable to human eyes (presumably for debugging). The default is to not send any additional characters.
default_ns($uri)
Sets the default namespace for the request to the specified uri. This overrides any previous namespace declaration that may have been set using a previous call to ns() or default_ns(). Setting the default namespace causes elements to be serialized without a namespace prefix, like so:
<soap:Envelope>
<soap:Body>
<myMethod xmlns="http://www.someuri.com">
<foo />
</myMethod>
</soap:Body>
</soap:Envelope>
ns($uri,$prefix=undef)
Sets the namespace uri and optionally the namespace prefix for the request to the specified values. This overrides any previous namespace declaration that may have been set using a previous call to ns() or default_ns(). If a prefix is not specified, one will be generated for you automatically. Setting the namespace causes elements to be serialized with a declared namespace prefix, like so:
<soap:Envelope>
<soap:Body>
<my:myMethod xmlns:my="http://www.someuri.com">
<my:foo />
</my:myMethod>
</soap:Body>
</soap:Envelope>
use_prefix(boolean)
Deprecated - the use_prefix() subroutine has been deprecated in order to provide a more intuitive naming scheme for subroutines that set namespaces. use_prefix() was originally added to allow users to turn on or off the use of a default namespace. This functionality is being replaced by ns() and default_ns().
Shortcut for serializer->use_prefix(). This lets you turn on/off the use of a namespace prefix for the children of the /Envelope/Body element. Default is 'true'. (This was introduced in 0.61 for better .NET compatibility)
When use_prefix is set to 'true', serialized XML will look like this:
<SOAP-ENV:Envelope ...attributes skipped>
<SOAP-ENV:Body>
<namesp1:mymethod xmlns:namesp1="urn:MyURI" />
</SOAP-ENV:Body>
</SOAP-ENV:Envelope>
When use_prefix is set to 'true', serialized XML will look like this:
<SOAP-ENV:Envelope ...attributes skipped>
<SOAP-ENV:Body>
<mymethod xmlns="urn:MyURI" />
</SOAP-ENV:Body>
</SOAP-ENV:Envelope>
$client->soapversion('1.2');
If no parameter is given, returns the current version of SOAP that is being used by the client object to encode requests. If a parameter is given, the method attempts to set that as the version of SOAP being used. The value should be either 1.1 or 1.2.
envprefix(QName)
$client->envprefix('env');
This method is a shortcut for:
$client->serializer->envprefix(QName);
The namespace label used for the main SOAP namespace elements (such as Envelope, Body, and the attributes) defaults to SOAP-ENV. As has been discussed in earlier chapters, the label itself isn't important. But applications that wish to explicitly choose a different one (such as env to denote a SOAP 1.2 message) may do so with this method.
encprefix(QName)
$client->encprefix('enc');
This method is a shortcut for:
$client->serializer->encprefix(QName);
As with the envprefix method, this gets or sets the label used for the namespace of the encoding rules. The default value is SOAP-ENC, as is generally used in SOAP 1.1 messages, though the label itself has no actual meaning.
While it may seem to be an unnecessary operation to set a value that isn't relevant to the message, such as the namespace labels for the envelope and encoding URNs, the ability to set these labels explicitly can prove to be a great aid in distinguishing and debugging messages on the server side of operations.
$client->encoding($soap_12_encoding_URN);
This method is a shortcut for:
$client->serializer->encoding(args);
Where the earlier method dealt with the label used for the attributes related to the SOAP encoding scheme, this method actually sets the URN to be specified as the encoding scheme for the message. The default is to specify the encoding for SOAP 1.1, so this is handy for applications that need to encode according to SOAP 1.2 rules.
$client->typelookup;
This method is a shortcut for:
$client->serializer->typelookup;
Gives the application access to the type-lookup table from the serializer object. See the section on the SOAP::Serializer manpage.
Deprecated - the uri subroutine has been deprecated in order to provide a more intuitive naming scheme for subroutines that set namespaces. In the future, you will be required to use either the ns() or default_ns() subroutines in lieu of uri().
$client->uri($service_uri);
This method is a shortcut for:
$client->serializer->uri(service);
The URI associated with this accessor on a client object is the service-specifier for the request, often encoded for HTTP-based requests as the SOAPAction header. While the names may seem confusing, this method doesn't specify the endpoint itself. In most circumstances, the uri refers to the namespace used for the request.
Often times, the value may look like a valid URL. Despite this, it doesn't have to point to an existing resource (and often doesn't). This method sets and retrieves this value from the object. Note that no transport code is triggered by this because it has no direct effect on the transport of the object.
multirefinplace(boolean)
$client->multirefinplace(1);
This method is a shortcut for:
$client->serializer->multirefinplace(boolean);
Controls how the serializer handles values that have multiple references to them. Recall from previous SOAP chapters that a value may be tagged with an identifier, then referred to in several places. When this is the case for a value, the serializer defaults to putting the data element towards the top of the message, right after the opening tag of the method-specification. It is serialized as a standalone entity with an ID that is then referenced at the relevant places later on. If this method is used to set a true value, the behavior is different. When the multirefinplace attribute is true, the data is serialized at the first place that references it, rather than as a separate element higher up in the body. This is more compact but may be harder to read or trace in a debugging environment.
Used to specify an array of the MIME::Entity manpage's to be attached to the transmitted SOAP message. Attachments that are returned in a response can be accessed by SOAP::SOM::parts().
$ref = SOAP::Lite->self;
Returns an object reference to the default global object the SOAP::Lite package maintains. This is the object that processes many of the arguments when provided on the use line.
The following method isn't an accessor style of method but neither does it fit with the group that immediately follows it:
call(arguments)
$client->call($method => @arguments);
As has been illustrated in previous chapters, the SOAP::Lite client objects can manage remote calls with auto-dispatching using some of Perl's more elaborate features. call is used when the application wants a greater degree of control over the details of the call itself. The method may be built up from a the SOAP::Data manpage object, so as to allow full control over the namespace associated with the tag, as well as other attributes like encoding. This is also important for calling methods that contain characters not allowable in Perl function names, such as A.B.C.
The next four methods used in the SOAP::Lite class are geared towards handling the types of events than can occur during the message lifecycle. Each of these sets up a callback for the event in question:
on_action(callback)
$client->on_action(sub { qq("$_[0]") });
Triggered when the transport object sets up the SOAPAction header for an HTTP-based call. The default is to set the header to the string, uri#method, in which URI is the value set by the uri method described earlier, and method is the name of the method being called. When called, the routine referenced (or the closure, if specified as in the example) is given two arguments, uri and method, in that order.
on_fault(callback)
$client->on_fault(sub { popup_dialog($_[1]) });
Triggered when a method call results in a fault response from the server. When it is called, the argument list is first the client object itself, followed by the object that encapsulates the fault. In the example, the fault object is passed (without the client object) to a hypothetical GUI function that presents an error dialog with the text of fault extracted from the object (which is covered shortly under the the SOAP::SOM manpage methods).
on_nonserialized(callback)
$client->on_nonserialized(sub { die "$_[0]?!?" });
Occasionally, the serializer may be given data it can't turn into SOAP-savvy XML; for example, if a program bug results in a code reference or something similar being passed in as a parameter to method call. When that happens, this callback is activated, with one argument. That argument is the data item that could not be understood. It will be the only argument. If the routine returns, the return value is pasted into the message as the serialization. Generally, an error is in order, and this callback allows for control over signaling that error.
on_debug(callback)
$client->on_debug(sub { print @_ });
This is kept for backwards-compatibility with earlier versions of the toolkit. Each method has a trace step built in, which is called at routine entry. This specifies a callback to be used when these trace statements are reached. Because this is deprecated, it is recommended that applications use the +debug and +trace facilities described later under the SOAP::Trace manpage. Note also that debugging isn't handled on a per-object basis; if this method is used on a given object, it sets debugging behavior for all objects of the class.
TODO - soap client example
See the SOAP::Server manpage, or the SOAP::Transport manpage.
SOAP::Lite features support for the SOAP with Attachments specification.
Currently, SOAP::Lite only supports MIME based attachments. DIME based attachments
are yet to be fully functional.
SOAP::Lite clients can specify attachments to be sent along with a request by using the SOAP::Lite::parts() method, which takes as an argument an ARRAY of MIME::Entity's.
use SOAP::Lite;
use MIME::Entity;
my $ent = build MIME::Entity
Type => "image/gif",
Encoding => "base64",
Path => "somefile.gif",
Filename => "saveme.gif",
Disposition => "attachment";
my $som = SOAP::Lite
->uri($SOME_NAMESPACE)
->parts([ $ent ])
->proxy($SOME_HOST)
->some_method(SOAP::Data->name("foo" => "bar"));
A client accessing attachments that were returned in a response by using the SOAP::SOM::parts() accessor.
use SOAP::Lite;
use MIME::Entity;
my $soap = SOAP::Lite
->uri($NS)
->proxy($HOST);
my $som = $soap->foo();
foreach my $part (${$som->parts}) {
print $part->stringify;
}
Servers, like clients, use the SOAP::SOM module to access attachments trasmitted to it.
package Attachment;
use SOAP::Lite;
use MIME::Entity;
use strict;
use vars qw(@ISA);
@ISA = qw(SOAP::Server::Parameters);
sub someMethod {
my $self = shift;
my $envelope = pop;
foreach my $part (@{$envelope->parts}) {
print "AttachmentService: attachment found! (".ref($part).")\n";
}
# do something
}
Servers wishing to return an attachment to the calling client need only return MIME::Entity objects along with SOAP::Data elements, or any other data intended for the response.
package Attachment;
use SOAP::Lite;
use MIME::Entity;
use strict;
use vars qw(@ISA);
@ISA = qw(SOAP::Server::Parameters);
sub someMethod {
my $self = shift;
my $envelope = pop;
my $ent = build MIME::Entity
'Id' => "<1234>",
'Type' => "text/xml",
'Path' => "some.xml",
'Filename' => "some.xml",
'Disposition' => "attachment";
return SOAP::Data->name("foo" => "blah blah blah"),$ent;
}
Though this feature looks similar to autodispatch they have (almost) nothing in common. This capability allows you specify default settings so that all objects created after that will be initialized with the proper default settings.
If you wish to provide common proxy() or uri() settings for all SOAP::Lite objects in your application you may do:
use SOAP::Lite
proxy => 'http://localhost/cgi-bin/soap.cgi',
uri => 'http://my.own.com/My/Examples';
my $soap1 = new SOAP::Lite; # will get the same proxy()/uri() as above print $soap1->getStateName(1)->result;
my $soap2 = SOAP::Lite->new; # same thing as above print $soap2->getStateName(2)->result;
# or you may override any settings you want
my $soap3 = SOAP::Lite->proxy('http://localhost/');
print $soap3->getStateName(1)->result;
Any SOAP::Lite properties can be propagated this way. Changes in object copies will not affect global settings and you may still change global settings with SOAP::Lite->self call which returns reference to global object. Provided parameter will update this object and you can even set it to undef:
SOAP::Lite->self(undef);
The use SOAP::Lite syntax also lets you specify default event handlers for your code. If you have different SOAP objects and want to share the same on_action() (or on_fault() for that matter) handler. You can specify on_action() during initialization for every object, but you may also do:
use SOAP::Lite
on_action => sub {sprintf '%s#%s', @_};
and this handler will be the default handler for all your SOAP objects. You can override it if you specify a handler for a particular object. See t/*.t for example of on_fault() handler.
Be warned, that since use ... is executed at compile time all use statements will be executed before script execution that can make unexpected results. Consider code:
use SOAP::Lite proxy => 'http://localhost/'; print SOAP::Lite->getStateName(1)->result;
use SOAP::Lite proxy => 'http://localhost/cgi-bin/soap.cgi'; print SOAP::Lite->getStateName(1)->result;
Both SOAP calls will go to 'http://localhost/cgi-bin/soap.cgi'. If you want to execute use at run-time, put it in eval:
eval "use SOAP::Lite proxy => 'http://localhost/cgi-bin/soap.cgi'; 1" or die;
Or alternatively,
SOAP::Lite->self->proxy('http://localhost/cgi-bin/soap.cgi');
One feature of SOAP::Lite is the ability to control the maximum size of a message a SOAP::Lite server will be allowed to process. To control this feature simply define $SOAP::Constants::MAX_CONTENT_SIZE in your code like so:
use SOAP::Transport::HTTP;
use MIME::Entity;
$SOAP::Constants::MAX_CONTENT_SIZE = 10000;
SOAP::Transport::HTTP::CGI
->dispatch_to('TemperatureService')
->handle;
SOAP::Lite gives you access to all parameters (both in/out and out) and also does some additional work for you. Lets consider following example:
<mehodResponse>
<res1>name1</res1>
<res2>name2</res2>
<res3>name3</res3>
</mehodResponse>
In that case:
$result = $r->result; # gives you 'name1' $paramout1 = $r->paramsout; # gives you 'name2', because of scalar context $paramout1 = ($r->paramsout)[0]; # gives you 'name2' also $paramout2 = ($r->paramsout)[1]; # gives you 'name3'
or
@paramsout = $r->paramsout; # gives you ARRAY of out parameters $paramout1 = $paramsout[0]; # gives you 'res2', same as ($r->paramsout)[0] $paramout2 = $paramsout[1]; # gives you 'res3', same as ($r->paramsout)[1]
Generally, if server returns return (1,2,3) you will get 1 as the result and 2 and 3 as out parameters.
If the server returns return [1,2,3] you will get an ARRAY from result() and undef from paramsout().
Results can be arbitrary complex: they can be an array, they can be objects, they can be anything and still be returned by result() . If only one parameter is returned, paramsout() will return undef.
Furthermore, if you have in your output parameters a parameter with the same signature (name+type) as in the input parameters this parameter will be mapped into your input automatically. For example:
Server Code:
sub mymethod {
shift; # object/class reference
my $param1 = shift;
my $param2 = SOAP::Data->name('myparam' => shift() * 2);
return $param1, $param2;
}
Client Code:
$a = 10;
$b = SOAP::Data->name('myparam' => 12);
$result = $soap->mymethod($a, $b);
After that, $result == 10 and $b->value == 24! Magic? Sort of.
Autobinding gives it to you. That will work with objects also with one difference: you do not need to worry about the name and the type of object parameter. Consider the PingPong example (examples/My/PingPong.pm and examples/pingpong.pl):
Server Code:
package My::PingPong;
sub new {
my $self = shift;
my $class = ref($self) || $self;
bless {_num=>shift} => $class;
}
sub next {
my $self = shift;
$self->{_num}++;
}
Client Code:
use SOAP::Lite +autodispatch =>
uri => 'urn:',
proxy => 'http://localhost/';
my $p = My::PingPong->new(10); # $p->{_num} is 10 now, real object returned
print $p->next, "\n"; # $p->{_num} is 11 now!, object autobinded
Let us scrutinize the deployment process. When designing your SOAP server you can consider two kind of deployment: static and dynamic. For both, static and dynamic, you should specify MODULE, MODULE::method, method or PATH/ when creating useing the SOAP::Lite module. The difference between static and dynamic deployment is that in case of 'dynamic', any module which is not present will be loaded on demand. See the SECURITY section for detailed description.
use SOAP::Transport::HTTP; use My::Examples; # module is preloaded
SOAP::Transport::HTTP::CGI
# deployed module should be present here or client will get 'access denied'
-> dispatch_to('My::Examples')
-> handle;
For static deployment you should specify the MODULE name directly.
You should also use static binding when you have several different classes in one file and want to make them available for SOAP calls.
use SOAP::Transport::HTTP; # name is unknown, module will be loaded on demand
SOAP::Transport::HTTP::CGI
# deployed module should be present here or client will get 'access denied'
-> dispatch_to('/Your/Path/To/Deployed/Modules', 'My::Examples')
-> handle;
For dynamic deployment you can specify the name either directly (in that case it will be required without any restriction) or indirectly, with a PATH. In that case, the ONLY path that will be available will be the PATH given to the dispatch_to() method). For information how to handle this situation see SECURITY section.
dispatch_to(
# dynamic dispatch that allows access to ALL modules in specified directory
PATH/TO/MODULES
# 1. specifies directory
# -- AND --
# 2. gives access to ALL modules in this directory without limits
# static dispatch that allows access to ALL methods in particular MODULE
MODULE
# 1. gives access to particular module (all available methods)
# PREREQUISITES:
# module should be loaded manually (for example with 'use ...')
# -- OR --
# you can still specify it in PATH/TO/MODULES
# static dispatch that allows access to particular method ONLY
MODULE::method
# same as MODULE, but gives access to ONLY particular method,
# so there is not much sense to use both MODULE and MODULE::method
# for the same MODULE
)
In addition to this SOAP::Lite also supports an experimental syntax that allows you to bind a specific URL or SOAPAction to a CLASS/MODULE or object. For example:
dispatch_with({
URI => MODULE, # 'http://www.soaplite.com/' => 'My::Class',
SOAPAction => MODULE, # 'http://www.soaplite.com/method' => 'Another::Class',
URI => object, # 'http://www.soaplite.com/obj' => My::Class->new,
})
URI is checked before SOAPAction. You may use both the dispatch_to() and dispatch_with() methods in the same server, but note that dispatch_with() has a higher order of precedence. dispatch_to() will be checked only after URI and SOAPAction has been checked. See t/03-server.t for more information and examples.
See also: EXAMPLE APACHE::REGISTRY USAGE, SECURITY
SOAP::Lite provides you option to enable transparent compression over the wire. Compression can be enabled by specifying a threshold value (in the form of kilobytes) for compression on both the client and server sides:
Note: Compression currently only works for HTTP based servers and clients.
Client Code
print SOAP::Lite
->uri('http://localhost/My/Parameters')
->proxy('http://localhost/', options => {compress_threshold => 10000})
->echo(1 x 10000)
->result;
Server Code
my $server = SOAP::Transport::HTTP::CGI
->dispatch_to('My::Parameters')
->options({compress_threshold => 10000})
->handle;
For more information see COMPRESSION in the HTTP::Transport manpage.
For security reasons, the exisiting path for Perl modules (@INC) will be disabled once you have chosen dynamic deployment and specified your own PATH/. If you wish to access other modules in your included package you have several options:
Switch to static linking:
use MODULE;
$server->dispatch_to('MODULE');
Which can also be useful when you want to import something specific from the deployed modules:
use MODULE qw(import_list);
Change use to require. The path is only unavailable during the initialization phase. It is available once more during execution. Therefore, if you utilize require somewhere in your package, it will work.
Wrap use in an eval block:
eval 'use MODULE qw(import_list)'; die if $@;
Set your include path in your package and then specify use. Don't forget to put @INC in a BEGIN{} block or it won't work. For example,
BEGIN { @INC = qw(my_directory); use MODULE }
In order to use a .NET client with a SOAP::Lite server, be sure you use fully qualified names for your return values. For example:
return SOAP::Data->name('myname')
->type('string')
->uri($MY_NAMESPACE)
->value($output);
In addition see comment about default incoding in .NET Web Services below.
If experiencing problems when using a SOAP::Lite client to call a .NET Web service, it is recommended you check, or adhere to all of the following recommendations:
For example, use on_action( sub { 'http://www.myuri.com/WebService.aspx#someMethod'; } ).
Some users have said that Microsoft .NET prefers the value of the Content-type header to be a mimetype exclusively, but SOAP::Lite specifies a character set in addition to the mimetype. This results in an error similar to:
Server found request content type to be 'text/xml; charset=utf-8', but expected 'text/xml'
To turn off this behavior specify use the following code:
use SOAP::Lite; $SOAP::Constants::DO_NOT_USE_CHARSET = 1; # The rest of your code
For example, the following code is preferred:
SOAP::Data->name(Query => 'biztalk')
->uri('http://tempuri.org/')
As opposed to:
SOAP::Data->name('Query' => 'biztalk')
For example, the following code is preferred:
my $method = SOAP::Data->name('add')
->attr({xmlns => 'http://tempuri.org/'});
my @rc = $soap->call($method => @parms)->result;
As opposed to:
my @rc = $soap->call(add => @parms)->result; # -- OR -- my @rc = $soap->add(@parms)->result;
Some user's have reported that .NET will simply not parse messages that use namespace prefixes on anything but SOAP elements themselves. For example, the following XML would not be parsed:
<SOAP-ENV:Envelope ...attributes skipped>
<SOAP-ENV:Body>
<namesp1:mymethod xmlns:namesp1="urn:MyURI" />
</SOAP-ENV:Body>
</SOAP-ENV:Envelope>
SOAP::Lite allows users to disable the use of explicit namespaces through the use_prefix() method. For example, the following code:
$som = SOAP::Lite->uri('urn:MyURI')
->proxy($HOST)
->use_prefix(0)
->myMethod();
Will result in the following XML, which is more pallatable by .NET:
<SOAP-ENV:Envelope ...attributes skipped>
<SOAP-ENV:Body>
<mymethod xmlns="urn:MyURI" />
</SOAP-ENV:Body>
</SOAP-ENV:Envelope>
Stefan Pharies <stefanph@microsoft.com>:
SOAP::Lite uses the SOAP encoding (section 5 of the soap 1.1 spec), and the default for .NET Web Services is to use a literal encoding. So elements in the request are unqualified, but your service expects them to be qualified. .Net Web Services has a way for you to change the expected message format, which should allow you to get your interop working. At the top of your class in the asmx, add this attribute (for Beta 1):
[SoapService(Style=SoapServiceStyle.RPC)]
Another source said it might be this attribute (for Beta 2):
[SoapRpcService]
Full Web Service text may look like:
<%@ WebService Language="C#" Class="Test" %> using System; using System.Web.Services; using System.Xml.Serialization;
[SoapService(Style=SoapServiceStyle.RPC)]
public class Test : WebService {
[WebMethod]
public int add(int a, int b) {
return a + b;
}
}
Another example from Kirill Gavrylyuk <kirillg@microsoft.com>:
``You can insert [SoapRpcService()] attribute either on your class or on operation level''.
<%@ WebService Language=CS class="DataType.StringTest"%>
namespace DataType {
using System;
using System.Web.Services;
using System.Web.Services.Protocols;
using System.Web.Services.Description;
[SoapRpcService()]
public class StringTest: WebService {
[WebMethod]
[SoapRpcMethod()]
public string RetString(string x) {
return(x);
}
}
}
Example from Yann Christensen <yannc@microsoft.com>:
using System; using System.Web.Services; using System.Web.Services.Protocols;
namespace Currency {
[WebService(Namespace="http://www.yourdomain.com/example";)]
[SoapRpcService]
public class Exchange {
[WebMethod]
public double getRate(String country, String country2) {
return 122.69;
}
}
}
Special thanks goes to the following people for providing the above description and details on .NET interoperability issues:
Petr Janata <petr.janata@i.cz>,
Stefan Pharies <stefanph@microsoft.com>,
Brian Jepson <bjepson@jepstone.net>, and others
Because Perl is loosely typed, there is no 100% reliable way to predict what the *intended* type of a variable is. So SOAP::Lite has a system of guessing what a type is. But it is not 100% reliable. You do however, have the control to override this autotyping behavior with your own.
Suppose you wanted to case every element of an array as a string. Then the following code will change the precedence of SOAP::Seriailizer's is-a-string test to be the first test run. Because the is-a-string test always returns true every element will be determined to be a string.
my @list = qw(-1 45 foo bar 3838);
my $proxy = SOAP::Lite->uri($uri)->proxy($proxyUrl);
$proxy->serializer->typelookup->{string}->[0] = 0;
$proxy->myMethod(\@list);
Alternatively, you can build the XML data structure manually...
See SOAP::Serializer.
+autodispatch doesn't work in Perl 5.8
There is a bug in Perl 5.8's UNIVERSAL::AUTOLOAD functionality that prevents the +autodispatch functionality from working properly. The workaround is to use dispatch_from instead. Where you might normally do something like this:
use Some::Module;
use SOAP::Lite +autodispatch =>
uri => 'urn:Foo'
proxy => 'http://...';
You would do something like this:
use SOAP::Lite dispatch_from(Some::Module) =>
uri => 'urn:Foo'
proxy => 'http://...';
You probably did not register Lite.dll using regsvr32 Lite.dll
It is likely that you have install Perl in two different locations and the location of ActiveState's Perl is not the first instance of Perl specified in your PATH. To rectify, rename the directory in which the non-ActiveState Perl is installed, or be sure the path to ActiveState's Perl is specified prior to any other instance of Perl in your PATH.
If you are using the Apache web server, and you are seeing something like the following in your webserver log file:
Can't load '/usr/local/lib/perl5/site_perl/.../XML/Parser/Expat/Expat.so'
for module XML::Parser::Expat: dynamic linker: /usr/local/bin/perl:
libexpat.so.0 is NEEDED, but object does not exist at
/usr/local/lib/perl5/.../DynaLoader.pm line 200.
Then try placing the following into your httpd.conf file and see if it fixes your problem.
<IfModule mod_env.c>
PassEnv LD_LIBRARY_PATH
</IfModule>
Using SOAP::Lite (or the XML::Parser::Expat manpage) in combination with mod_perl causes random segmentation faults in httpd processes. To fix, try configuring Apache with the following:
RULE_EXPAT=no
If you are using Apache 1.3.20 and later, try configuring Apache with the following option:
./configure --disable-rule=EXPAT
See http://archive.covalent.net/modperl/2000/04/0185.xml for more details and lot of thanks to Robert Barta <rho@bigpond.net.au> for explaining this weird behavior.
If this doesn't address the problem, you may wish to try -Uusemymalloc, or a similar option in order to instruct Perl to use the system's own malloc.
Thanks to Tim Bunce <Tim.Bunce@pobox.com>.
CGI scripts may not work under IIS unless scripts use the .pl extension, opposed to .cgi.
In some cases SOAP messages created by SOAP::Lite may not be parsed properly by a SAX2/Java XML parser. This is due to a known bug in org.xml.sax.helpers.ParserAdapter. This bug manifests itself when an attribute in an XML element occurs prior to the XML namespace declaration on which it depends. However, according to the XML specification, the order of these attributes is not significant.
Thanks to Steve Alpert (Steve_Alpert@idx.com) for pointing on it.
SOAP::Lite is based on the XML::Parser manpage which is basically wrapper around James
Clark's expat parser. Expat's behavior for parsing XML encoded string can
affect processing messages that have lot of encoded entities, like XML
fragments, encoded as strings. Providing low-level details, parser will call
char() callback for every portion of processed stream, but individually for
every processed entity or newline. It can lead to lot of calls and additional
memory manager expenses even for small messages. By contrast, XML messages
which are encoded as base64Binary, don't have this problem and difference in
processing time can be significant. For XML encoded string that has about 20
lines and 30 tags, number of call could be about 100 instead of one for
the same string encoded as base64Binary.
Since it is parser's feature there is NO fix for this behavior (let me know if you find one), especially because you need to parse message you already got (and you cannot control content of this message), however, if your are in charge for both ends of processing you can switch encoding to base64 on sender's side. It will definitely work with SOAP::Lite and it may work with other toolkits/implementations also, but obviously I cannot guarantee that.
If you want to encode specific string as base64, just do
SOAP::Data->type(base64 => $string) either on client or on server
side. If you want change behavior for specific instance of SOAP::Lite, you
may subclass SOAP::Serializer, override as_string() method that is
responsible for string encoding (take a look into as_base64Binary()) and
specify new serializer class for your SOAP::Lite object with:
my $soap = new SOAP::Lite
serializer => My::Serializer->new,
..... other parameters
or on server side:
my $server = new SOAP::Transport::HTTP::Daemon # or any other server
serializer => My::Serializer->new,
..... other parameters
If you want to change this behavior for all instances of SOAP::Lite, just
substitute as_string() method with as_base64Binary() somewhere in your
code after use SOAP::Lite and before actual processing/sending:
*SOAP::Serializer::as_string = \&SOAP::Serializer::as_base64Binary;
Be warned that last two methods will affect all strings and convert them into base64 encoded. It doesn't make any difference for SOAP::Lite, but it may make a difference for other toolkits.
No support for multidimensional, partially transmitted and sparse arrays (however arrays of arrays are supported, as well as any other data structures, and you can add your own implementation with SOAP::Data).
Limited support for WSDL schema.
XML::Parser::Lite relies on Unicode support in Perl and doesn't do entity decoding.
Limited support for mustUnderstand and Actor attributes.
Information about XML::Parser for MacPerl could be found here:
http://bumppo.net/lists/macperl-modules/1999/07/msg00047.html
Compiled XML::Parser for MacOS could be found here:
http://www.perl.com/CPAN-local/authors/id/A/AS/ASANDSTRM/XML-Parser-2.27-bin-1-MacOS.tgz
You can download the latest version SOAP::Lite for Unix or SOAP::Lite for Win32 from the following sources:
* SOAP::Lite Homepage: http://soaplite.com/ * CPAN: http://search.cpan.org/search?dist=SOAP-Lite * Sourceforge: http://sourceforge.net/projects/soaplite/
You are welcome to send e-mail to the maintainers of SOAP::Lite with your with your comments, suggestions, bug reports and complaints.
Special thanks to Randy J. Ray, author of Programming Web Services with Perl, who has contributed greatly to the documentation effort of SOAP::Lite.
Special thanks to O'Reilly publishing which has graciously allowed SOAP::Lite to republish and redistribute the SOAP::Lite reference manual found in Appendix B of Programming Web Services with Perl.
And special gratitude to all the developers who have contributed patches, ideas, time, energy, and help in a million different forms to the development of this software.
Please report all suspected SOAP::Lite bugs using Sourceforge. This ensures proper tracking of the issue and allows you the reporter to know when something gets fixed.
http://sourceforge.net/tracker/?group_id=66000&atid=513017
If under dire circumstances you need immediate assistance with the resolution of an issue, you are welcome to contact Byrne Reese at <byrne at majordojo dot com>.
Copyright (C) 2000-2005 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.
This text and all associated documentation for this library is made available under the Creative Commons Attribution-NoDerivs 2.0 license. http://creativecommons.org/licenses/by-nd/2.0/
Paul Kulchenko (paulclinger@yahoo.com)
Randy J. Ray (rjray@blackperl.com)
Byrne Reese (byrne@majordojo.com)