Xord: An Implementation Framework for Efficient XSLT Processing

Abstract

Summary. We introduce an implementation of Xord -an XSLT processing framework which enables us to design and implement efficient algorithms for clearly characterized classes of XSLT transformations with known memory requirements. Within the framework, we design and implement a streaming algorithm using stack of the size proportional to the depth of the input document and associate it with the class of simple order-preserving and branch-disjoint transformations. The framework provides an unified interface to the underlying algorithms and acts as a standard XSLT processor.

Using Input Buffers for Streaming XSLT Processing

“…The transformer takes two inputs: an XSLT stylesheet xsl, and an XML document xml valid with respect to xsd. It is based on the non-buffering SSXT streaming algorithm [5] which is extended by a possibility to store parts of the input temporarily in memory buffers. The decisions on when to start buffering and when to process buffer content are taken according to the information stored in the fragments.…”

“…In case a final state appears in the new sequence, the transformer checks whether buffering is needed by examining all fragment items (6). If some of them contains start-of-buffer node for 5 Note that call1 may be the beginning of the template and call2 may be the end of the template. the current context, a new buffer is filled with the content of the current tag.…”

“…During the previous work on the Xord project In [5,6], the Xord framework for the streaming processing of XSLT transformations was designed and implemented. The framework is intended to contain several streaming engines for processing XSLT.…”

“…Within the framework, the SSXT 1 engine was implemented. The SSXT transformer [5] processes a subset of top-down XSLT using stack of the size proportional to the depth of the input XML document. The SSXT analyzer [6] takes a schema and a topdown XSLT stylesheet, and it determines whether transformation 1 SSXT stands for simple streaming XML transducer.…”

Visibly Pushdown Transducers

“…They however do not tackle the problem of efficiently performing the transformation. A second trend of related researches looks for an efficient model for performing XML transformations [DZ09,DZ08,FN07]. Clearly, first and foremost we would have to assess the exact expressiveness of macro VPT with regards to all those existing models.…”