gnu.xml.pipeline
Class PipelineFactory

java.lang.Object
  |
  +--gnu.xml.pipeline.PipelineFactory

public class PipelineFactory
extends java.lang.Object

This provides static factory methods for creating simple event pipelines. These pipelines are specified by strings, suitable for passing on command lines or embedding in element attributes. For example, one way to write a pipeline that restores namespace syntax, validates (stopping the pipeline on validity errors) and then writes valid data to standard output is this:

      nsfix | validate | write ( stdout )

In this syntax, the tokens are always separated by whitespace, and each stage of the pipeline may optionally have a parameter (which can be a pipeline) in parentheses. Interior stages are called filters, and the rightmost end of a pipeline is called a terminus.

Stages are usually implemented by a single class, which may not be able to act as both a filter and a terminus; but any terminus can be automatically turned into a filter, through use of a TeeConsumer. The stage identifiers are either class names, or are one of the following short identifiers built into this class. (Most of these identifiers are no more than aliases for classes.) The built-in identifiers include:

Stage Parameter Terminus Description
dom optional
Document classname
yes Applications code can access a DOM Document built from the input event stream. When used as a filter, this buffers data up to an endDocument call, and then uses a DOM parser to report everything that has been recorded (which can easily be less than what was reported to it).
nsfix none no This stage ensures that the XML element and attribute names in its output use namespace prefixes and declarations correctly. That is, so that they match the "Namespace plus LocalName" naming data with which each XML element and attribute is already associated.
null none yes This stage ignores all input event data.
server required
server URL
no Sends its input as XML request to a remote server, normally a web application server using the HTTP or HTTPS protocols. The output of this stage is the parsed response from that server.
tee required
first pipeline
no This sends its events down two paths; its parameter is a pipeline descriptor for the first path, and the second path is the output of this stage.
validate none yes This checks for validity errors, and reports them through its error handler. The input must include declaration events and some lexical events.
wf none yes This class provides some basic "well formedness" tests on the input event stream, and reports a fatal error if any of them fail. One example: start/end calls for elements must match. No SAX parser is permitted to produce malformed output, but other components can easily do so.
write required
"stdout", "stderr", or filename
yes Writes its input to the specified output, as pretty printed XML text encoded using UTF-8. Input events must be well formed and "namespace fixed", else the output won't be XML (or possibly namespace) conformant. The symbolic names represent System.out and System.err respectively; names must correspond to files which don't yet exist.
xhtml required
"stdout", "stderr", or filename
yes Like write (above), except that XHTML rules are followed. The XHTML 1.0 Transitional document type is declared, and only ASCII characters are written (for interoperability). Other characters are written as entity or character references; the text is pretty printed.

Note that EventFilter.bind(org.xml.sax.XMLReader, gnu.xml.pipeline.EventConsumer) can automatically eliminate some filters by setting SAX2 parser features appropriately. This means that you can routinely put filters like "nsfix", "validate", or "wf" at the front of a pipeline (for components that need inputs conditioned to match that level of correctness), and know that it won't actually be used unless it's absolutely necessary.

Version:
$Date: 2001/07/10 22:54:43 $
Author:
David Brownell

Method Summary
static EventConsumer createPipeline(java.lang.String description)
          Creates a simple pipeline according to the description string passed in.
static EventConsumer createPipeline(java.lang.String[] tokens, EventConsumer next)
          Extends an existing pipeline by prepending a pre-tokenized filter pipeline to the specified consumer.
static EventConsumer createPipeline(java.lang.String description, EventConsumer next)
          Extends an existing pipeline by prepending the filter pipeline to the specified consumer.
 
Methods inherited from class java.lang.Object
clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait
 

Method Detail

createPipeline

public static EventConsumer createPipeline(java.lang.String description)
                                    throws java.io.IOException
Creates a simple pipeline according to the description string passed in.

createPipeline

public static EventConsumer createPipeline(java.lang.String description,
                                           EventConsumer next)
                                    throws java.io.IOException
Extends an existing pipeline by prepending the filter pipeline to the specified consumer. Some pipelines need more customization than can be done through this simplified syntax. When they are set up with direct API calls, use this method to merge more complex pipeline segments with easily configured ones.

createPipeline

public static EventConsumer createPipeline(java.lang.String[] tokens,
                                           EventConsumer next)
                                    throws java.io.IOException
Extends an existing pipeline by prepending a pre-tokenized filter pipeline to the specified consumer. Tokens are class names (or the predefined aliases) left and right parenthesis, and the vertical bar.

Source code is GPL'd in the JAXP subproject at http://savannah.gnu.org/projects/classpathx
This documentation was derived from that source code on 2001-07-12.