This example shows how to compress an XML document during serialization 
and decompress it during parsing. The example uses the compression 
functionality provided by the zlib library[1] which needs to be installed
in order to build and run this example. It should also be fairly straight-
forward to modify the code in this example to use other compression 
libraries.

[1] http://www.zlib.net

The example consists of the following files:

library.xsd
  XML Schema which describes a library of books.

library.xml.gz
  Sample XML instance document compressed using the gzip format.

compressed-format-target.hxx
compressed-format-target.cxx
  Implementation of the Xerces-C++ XMLFormatTarget interface with the on-
  the-fly compression support. You can use it in your application to add
  XML compression.

compressed-input-source.hxx
compressed-input-source.cxx
  Implementation of the Xerces-C++ InputSource interface with the on-the-
  fly decompression support. You can use it in your application to add
  XML decompression.

library.hxx
library.cxx
  C++ types that represent the given vocabulary and a set of parsing
  functions that convert XML instance documents to a tree-like in-memory
  object model. These are generated by XSD from library.xsd.

driver.cxx
  Driver for the example. It first creates the compressed_input_source 
  object and passes it to one of the parsing functions that constructs 
  the object model from the compressed input file. It then prints the 
  content of the object model to STDERR. Finally, the driver creates the 
  compressed_format_target object and passes it to one of the serialization
  functions which converts the object model back to the compressed XML.

To run the example on the sample XML document simply execute:

$ ./driver library.xml.gz

The serialization output is written to the out.xml.gz file.