XML Pipelines Part4

Home > Books > Tutorials and Training Guides > Pipeline Processing > XML Pipelines > XML Pipelines Part4

Procedural pipelines and XML object model independence

In the previous section we saw how to create interactive pipelines by passing parameter arguments on the active URI. We also saw that pipelines can be decomposed into fine grained units and then invoked using a language runtime. In this section we demonstrate the same pipeline written in a procedural style using Beanshell (scripted Java). We also show how an individual stage of a pipeline may choose it's preferred XML object model without effecting the higher level scheduling of the pipeline.

Pipeline

Below is the same interactive pipeline process we saw in the last section. Here it is written in Beanshell and uses the NetKernel Foundation API in the form of the context object.


/**************************
* Beanshell script to perform multistage
* XQuery into XSLT.
***************************/

void main()
{   //Run previous DPML pipeline at Stage 3 - Produces all Gloucester's speeches
    req=context.createSubRequest();
    req.setURI("active:dpml");
    req.addArgument("operand","../pipeline2/stage3.idoc");
    output=context.issueSubRequest(req);

    //Handle the parameter argument 
    param=null;
    try
    {   param=context.source("this:param:param");
    }
    catch(Exception e)
    {   //Use default
        param=context.source("default.xml");
    }

    //Extract all speeches with lines containing the parameter argument
    req=context.createSubRequest();
    req.setURI("active:xquery");
    req.addArgument("input", output);
    req.addArgument("param", param);
    req.addArgument("operator","../pipeline3/xq4.xq");
    output=context.issueSubRequest(req);
    
    //Apply XSLT
    req=context.createSubRequest();
    req.setURI("active:xslt");
    req.addArgument("operand", output);
    req.addArgument("param", param);
    req.addArgument("operator","../pipeline3/style1.xsl");
    output=context.issueSubRequest(req);
    
    //Finally return response
    response=context.createResponseFrom(output);
    //response.setMimeType("text/plain");
    context.setResponse(response);
}

try it!

On inspection you will see that the Beanshell script is quite similar to the previous DPML version. In this case we see a little more closely that each stage of the pipeline consists of building an active URI request, adding arguments and then issuing the request. The results of a given request is passed as an argument on a subsequent request, forming the pipeline. We also note that familiar Java exception handling is used to set a default value if the this:param:param argument is not supplied.

Heterogeneous XML Object Models

So far all of our examples have worked without any consideration of the underlying XML object model (DOM, SAX, JDOM etc etc) that is being used. In fact this is almost always the case when sequencing pipelines. Each individual stage may use it's own preferred XML object model, NetKernel transparently manages the transreption (isomorphic transformation) from one form to another. To illustrate, lets introduce a new stage which will apply an UpperCase transform on all element text and add this into the pipeline after the last XQuery and before the XSLT transform.

Here's a script which performs the transform. We can see that it requests it's input argument with a SAXAspect it then connects a to Upper SAX Filter transform (a simple SAX content handler which uppercases all element text) to the SAX stream and returns SAXFilterAspect.


/**************************
* Beanshell script - SAX Example
* Use SAX filter to Upper case all text
***************************/
import org.ten60.netkernel.xml.representation.*;
import org.ten60.netkernel.demo.sax.ToUpperSAXFilter;

void main()
{   
    //Do some work with SAX
    IAspectSAX asp=(IAspectSAX)context.sourceAspect("this:param:input", IAspectSAX.class);
    
    filter = new ToUpperSAXFilter();
    sfa=new SAXFilterAspect(asp, filter);

    //Finally return response
    response=context.createResponseFrom(sfa);
    context.setResponse(response);
}

Below you will see the toUpper filter inserted into the pipeline at stage 4b prior to the final XSLT transform.


/**************************
* Beanshell script to perform multistage
* XQuery into XSLT.
***************************/

void main()
{   //Run previous DPML pipeline at Stage 3 - Produces all Gloucester's speeches
    req=context.createSubRequest();
    req.setURI("active:dpml");
    req.addArgument("operand","../pipeline2/stage3.idoc");
    output=context.issueSubRequest(req);

    //Handle the parameter argument 
    param=null;
    try
    {   param=context.source("this:param:param");
    }
    catch(Exception e)
    {   //Use default
        param=context.source("default.xml");
    }

    //Extract all speeches with lines containing the parameter argument
    req=context.createSubRequest();
    req.setURI("active:xquery");
    req.addArgument("input", output);
    req.addArgument("param", param);
    req.addArgument("operator","../pipeline3/xq4.xq");
    output=context.issueSubRequest(req);
    
    //Upper case all text
    req=context.createSubRequest();
    req.setURI("active:beanshell");
    req.addArgument("input", output);
    req.addArgument("operator","toUpper.bsh");
    output=context.issueSubRequest(req);

    //Apply XSLT
    req=context.createSubRequest();
    req.setURI("active:xslt");
    req.addArgument("operand", output);
    req.addArgument("param", param);
    req.addArgument("operator","../pipeline3/style1.xsl");
    output=context.issueSubRequest(req);
    
    //Finally return response
    response=context.createResponseFrom(output);
    //response.setMimeType("text/plain");
    context.setResponse(response);
}

try it!

You will see that the pipeline does not have any code to convert object models. All object model conversion is handled transparently within the pipeline by the NetKernel transreptor architecture. This leads to highly maintainable pipelines since each stage is truely an encapsulated service and the pipeline is in fact simply a sequencing of the services.