VTD-XML is a non-validating,
"non-extractive" XML processing software API implementing Virtual
Token Descriptor. Currently it only supports built-in entity references(" &s; ' > <).
Designed
to be random-access capable, VTD-XML attempts to be both high-performance and
memory-efficient. In addition, because VTD is inherent persistent, it can be
saved on disk alongside of the XML document to achieve "parse-once, use many
times." Moreover, the "non-extractive" nature of VTD means that incremental,
dynamic update doesn't require re-serialization of irrelevant parts of the original document.
On a 1.5gHz Athlon
PC, VTD-XML consistently outperforms SAX parsers with NULL content handler,
delivering 25~35 MB/sec sustained throughput, without sacrifice random
access. Its memory usage is typical between 1.3x~1.5 x the size of the XML
document, with 1 being the XML itself.
A good starting point
to understand how VTD-XML works is to view the
demo, which previews the XPath
features of VTD-XML version 1.0.
The latest version is Version 0.9 which
can be downloaded
here.