Constructors

Handlers

$obj->compileClient(%options)

 -Option   --Default
  async      <false>
  decode     <required>
  encode     <required>
  kind       request-response
  name       <undef>
  soap       'SOAP11'
  transport  <required>
    

   my $call = $wsdl->compileClient('myOp');
   my ($answer, $trace) = $call->(@params);
   #do something with $answer

   my $call = $wsdl->compileClient('myOp', async => 1);
   sub cb
   {  my ($answer, $trace) = @_;
      #do something with $answer
   };
   $call->(@params, _callback => \&cb);

Client side SOAP

First, you compile the call either via a WSDL file (see XML::Compile::WSDL11), or in a few manual steps (which are described in the next section). In either way, you end-up with a CODE references which can be called multiple times.

    # compile once
    my $call   = $soap->compileClient(...);

    # and call often
    my $answer = $call->(%request);  # list of pairs
    my $answer = $call->(\%request); # same, but HASH
    my $answer = $call->(\%request, 'UTF-8');  # same

    # or with trace details, see XML::Compile::SOAP::Trace
    my ($answer, $trace) = $call->...

But what is the structure of %request and $answer? Well, there are various syntaxes possible: from structurally perfect, to user-friendly.

First, find out which data structures can be present: when you compiled your messages explicitly, you have picked your own names. When the call was initiated from a WSDL file, then you have to find the names of the message parts which can be used: the part names for header blocks, body blocks, headerfaults, and (body) faults. Do not worry to much, you will get (hopefully understandable) run-time error messages when the structure is incorrect.

Let's say that the WSDL defines this (ignoring all name-space issues)

 <definitions xmlns:xx="MYNS"
   <message name="GetLastTradePriceInput">
    <part name="count" type="int" />
    <part name="request" element="xx:TradePriceRequest"/>
   </message>

   <message name="GetLastTradePriceOutput">
    <part name="answer" element="xx:TradePrice"/>
   </message>

   <binding
    <operation
     <input>
      <soap:header message="GetLastTradePriceInput" part="count"
      <soap:body message="GetLastTradePriceInput" parts="request"
     <output>
      <soap:body message="GetLastTradePriceOutput"

The input message needs explicitly named parts in this case, where the output message simply uses all defined in the body. So, the input message has one header part "count", and one body part "request". The output message only has one part named "answer", which is all defined for the message and therefore its name can be omitted.

Then, the definitions of the blocks:

 <schema targetNamespace="MYNS"
   <element name="TradePriceRequest">
    <complexType>
     <all>
      <element name="tickerSymbol" type="string"/>

   <element name="TradePrice">
    <complexType>
     <all>
      <element name="price" type="float"/>
 </schema>

Now, calling the compiled function can be done like this:

  my $got
     = $call->(  count => 5, request => {tickerSymbol => 'IBM'}  );
     = $call->({ count => 5, request => {tickerSymbol => 'IBM'} });
     = $call->({ count => 5, request => {tickerSymbol => 'IBM'} }
        , 'UTF-8');

If the first arguments for the code ref is a HASH, then there may be a second which specifies the required character-set. The default is "UTF-8", which is very much advised.

Parameter unpacking (Document Style)

In the example situation of previous section, you may simplify the call even further. To understand how, we need to understand the parameter unpacking algorithm.

The structure which we need to end up with, looks like this

  $call->(\%data, $charset);
  %data = ( Header => {count => 5}
          , Body   =>
             { request => {tickerSymbol => 'IBM'} }
          );

The structure of the SOAP message is directly mapped on this nested complex HASH. But is inconvenient to write each call like this, therefore the $call parameters are transformed into the required structure according to the following rules:

1.

if called with a LIST, then that will become a HASH

2.

when a "Header" and/or "Body" are found in the HASH, those are used

3.

if there are more parameters in the HASH, then those with names of known header and headerfault message parts are moved to the "Header" sub-structure. Body and fault message parts are moved to the "Body" sub-structure.

4.

If the "Body" sub-structure is empty, and there is only one body part expected, then all remaining parameters are put in a HASH for that part. This also happens if there are not parameters: it will result in an empty HASH for that block.

So, in our case this will also do, because "count" is a known part, and "request" gets all left-overs, being the only body part.

 my $got = $call->(count => 5, tickerSymbol => 'IBM');

This does not work if the block element is a simple type. In most existing Document style SOAP schemas, this simplification probably is possible.

Understanding the output (Document style)

The $got is a HASH, which will not be simplified automatically: it may change with future extensions of the interface. The return is a complex nested structure, and Data::Dumper is your friend.

 $got = { answer => { price => 16.3 } }

To access the value use

 printf "%.2f US\$\n", $got->{answer}->{price};
 printf "%.2f US\$\n", $got->{answer}{price};   # same

or

 my $answer = $got->{answer};
 printf "%.2f US\$\n", $answer->{price};

Calling the server (SOAP-RPC style literal)

SOAP-RPC style messages which have "<use=literal"> cannot be used without a little help. However, one extra definition per procedure call suffices.

This a complete code example, although you need to fill in some specifics about your environment. If you have a WSDL file, then it will be a little simpler, see XML::Compile::WSDL11::compileClient().

 # You probably need these
 use XML::Compile::SOAP11::Client;
 use XML::Compile::Transport::SOAPHTTP;
 use XML::Compile::Util  qw/pack_type/;

 # Literal style RPC
 my $outtype = pack_type $MYNS, 'myFunction';
 my $intype  = pack_type $MYNS, 'myFunctionResponse';

 # Encoded style RPC (see next section on these functions)
 my $outtype = \&my_pack_params;
 my $intype  = \&my_unpack_params;

 # For all RPC calls, you need this only once (or have a WSDL):
 my $transp  = XML::Compile::Transport::SOAPHTTP->new(...);
 my $http    = $transp->compileClient(...);
 my $soap    = XML::Compile::SOAP11::Client->new(...);
 my $send    = $soap->compileMessage('SENDER',   style => $style, ...);
 my $get     = $soap->compileMessage('RECEIVER', style => $style, ...);

 # Per RPC procedure
 my $myproc = $soap->compileClient
   ( name   => 'MyProc'
   , encode => $send, decode => $get, transport => $http
   );

 my $answer = $myproc->(@parameters);   # as document style

Actually, the @paramers are slightly less flexible as in document style SOAP. If you use header blocks, then the called CODE reference will not be able to distinguish between parameters for the RPC block and parameters for the header blocks.

  my $answer = $trade_price
    ->( {symbol => 'IBM'}    # the RPC package implicit
      , transaction => 5     # in the header
      );

When the number of arguments is odd, the first is indicating the RPC element, and the other pairs refer to header blocks.

The $answer structure may contain a "Fault" entry, or a decoded datastructure with the results of your query. One call using Data::Dumper will show you more than I can explain in a few hundred words.

Calling the server (SOAP-RPC style, encoded)

SOAP-RPC is a simplification of the interface description: basically, the interface is not described at all, but left to good communication between the client and server authors. In strongly typed languages, this is quite simple to enforce: the client side and server side use the same method prototypes. However, in Perl we are blessed to go without these strongly typed prototypes.

The approach of SOAP::Lite, is to guess the types of the passed parameters. For instance, "42" will get passed as Integer. This may lead to nasty problems: a float parameter "2.0" will get passed as integer "2", or a string representing a house number "8" is passed as an number. This may not be accepted by the SOAP server.

So, using SOAP-RPC in XML::Compile::SOAP will ask a little more effort from you: you have to state parameter types explicitly. In the examples/namesservice/ directory, you find a detailed example. You have to create a CODE ref which produces the message, using methods defined provided by XML::Compile::SOAP11::Encoding.

Faults (Document and RPC style)

Faults and headerfaults are a slightly different story: the type which is specified with them is not of the fault XML node itself, but of the "detail" sub-element within the standard fault structure.

When producing the data for faults, you must be aware of the fact that the structure is different for SOAP1.1 and SOAP1.2. When interpreting faults, the same problems are present, although the implementation tries to help you by hiding the differences.

Check whether SOAP1.1 or SOAP1.2 is used by looking for a "faultcode" (SOAP1.1) or a "Code" (SOAP1.2) field in the data:

  if(my $fault = $got->{Fault})
  {  if($fault->{faultcode}) { ... SOAP1.1 ... }
     elsif($fault->{Code})   { ... SOAP1.2 ... }
     else { die }
  }

In either protocol case, the following will get you at a compatible structure in two steps:

  if(my $fault = $got->{Fault})
  {   my $decoded = fault->{_NAME}};
      print $got->{$decoded}->{code};
      ...
  }

See the respective manuals XML::Compile::SOAP11 and XML::Compile::SOAP12 for the hairy details. But one thing can be said: when the fault is declared formally, then the "_NAME" will be the name of that part.

SOAP without WSDL (Document style)

When using SOAP without WSDL file, it gets a little bit more complicate to use: you need to describe the content of the messages yourself. The following example is used as test-case "t/10soap11.t", directly taken from the SOAP11 specs section 1.3 example 1.

 # for simplification
 my $TestNS   = 'http://test-types';
 use XML::Compile::Util qw/SCHEMA2001/;
 my $SchemaNS = SCHEMA2001;

First, the schema (hopefully someone else created for you, because they can be quite hard to create correctly) is in file "myschema.xsd"

 <schema targetNamespace="$TestNS"
   xmlns="$SchemaNS">

 <element name="GetLastTradePrice">
   <complexType>
      <all>
        <element name="symbol" type="string"/>
      </all>
   </complexType>
 </element>

 <element name="GetLastTradePriceResponse">
   <complexType>
      <all>
         <element name="price" type="float"/>
      </all>
   </complexType>
 </element>

 <element name="Transaction" type="int"/>
 </schema>

Ok, now the program you create the request:

 use XML::Compile::SOAP11;
 use XML::Compile::Util  qw/pack_type/;

 my $soap   = XML::Compile::SOAP11->new;
 $soap->schemas->importDefinitions('myschema.xsd');

 my $get_price = $soap->compileMessage
   ( 'SENDER'
   , header =>
      [ transaction => pack_type($TestNS, 'Transaction') ]
   , body  =>
      [ request => pack_type($TestNS, 'GetLastTradePrice') ]
   , mustUnderstand => 'transaction'
   , destination    => [ transaction => 'NEXT http://actor' ]
   );

"INPUT" is used in the WSDL terminology, indicating this message is an input message for the server. This $get_price is a WRITER. Above is done only once in the initialization phase of your program.

At run-time, you have to call the CODE reference with a data-structure which is compatible with the schema structure. (See XML::Compile::Schema::template() if you have no clue how it should look) So: let's send this:

 # insert your data
 my %data_in = (transaction => 5, request => {symbol => 'DIS'});
 my %data_in = (transaction => 5, symbol => 'DIS'); # alternative

 # create a XML::LibXML tree
 my $xml  = $get_price->(\%data_in, 'UTF-8');
 print $xml->toString;

And the output is:

 <SOAP-ENV:Envelope
    xmlns:x0="http://test-types"
    xmlns:SOAP-ENV="http://schemas.xmlsoap.org/soap/envelope/">
   <SOAP-ENV:Header>
     <x0:Transaction
       mustUnderstand="1"
       actor="http://schemas.xmlsoap.org/soap/actor/next http://actor">
         5
     </x0:Transaction>
   </SOAP-ENV:Header>
   <SOAP-ENV:Body>
     <x0:GetLastTradePrice>
       <symbol>DIS</symbol>
     </x0:GetLastTradePrice>
   </SOAP-ENV:Body>
 </SOAP-ENV:Envelope>

Some transport protocol will sent this data from the client to the server. See XML::Compile::Transport::SOAPHTTP, as one example.

On the SOAP server side, we will parse the message. The string $soap contains the XML. The program looks like this:

 my $server = $soap->compileMessage # create once
  ( 'RECEIVER'
  , header => [ transaction => pack_type($TestNS, 'Transaction') ]
  , body   => [ request => pack_type($TestNS, 'GetLastTradePrice') ]
  );

 my $data_out = $server->($soap);   # call often

Now, the $data_out reference on the server, is stucturally exactly equivalent to the %data_in from the client.