A comprehensive XQuery to SQL translation using dynamic interval encoding

Abstract

The W3C XQuery language recommendation, based on a hierarchical and ordered document model, supports a wide variety of constructs and use cases. There is a diversity of approaches and strategies for evaluating XQuery expressions, in many cases only dealing with limited subsets of the language. In this paper we describe an implementation approach that handles XQuery with arbitrarily-nested FLWR expressions, element constructors and built-in functions (including structural comparisons). Our proposal maps an XQuery expression to a single equivalent SQL query using a novel dynamic interval encoding of a collection of XML documents as relations, augmented with information tied to the query evaluation environment. The dynamic interval technique enables (suitably enhanced) relational engines to produce predictably good query plans that do not restrict the use of sort-merge join query operators. The benefits are realized despite the challenges presented by intermediate results that create arbitrary documents and the need to preserve document order as prescribed by semantics of XQuery. Finally, our experimental results demonstrate that (native or relational) XML systems can benefit from the above technique to avoid a quadratic scale up penalty that effectively prevents the evaluation of nested FLWR expressions for large documents.

XML-to-SQL Query Translation Literature: The State of the Art and Open Problems

“…In [7], all XML data is stored in a single table containing a tuple for each element, attribute and text node. For an element, the element name and an interval representing the region covered by the element is stored.…”

“…The dynamic intervals approach [7] deals with a larger fragment of XQuery with arbitrarily nested FLWR expressions, element constructors and built-in functions including structural comparisons. The core idea is to begin with static intervals for each element and construct dynamic intervals for XML elements constructed in the query.…”

“…The only published work that we are aware of that deals with more general XQuery queries is [7]. The main focus of the paper is on issues such as structural equality in FLWR where clauses, full compositionality of XML query expressions (in particular, the possibility of nesting FLWR expressions within functions), and the need for constructed XML documents representing intermediate query results.…”

“…As an example, using their technique, the path expression /site/people is translated to an SQL query involving five temporary relations created using the With clause in SQL99, three of which involve correlated subqueries. To conclude, excepting [7], all prior work has been on translating path expression queries into SQL. Using the approach proposed by [7], we observe that functionality-wise, a large fragment of XQuery can be handled using dynamic intervals in a schema-oblivious fashion.…”

“…It is our hope that this paper will stimulate others to refine our classification and, more importantly, to improve the state of the art and to address and solve the open problems that the classification reveals. [7] [ 24,32] XS/SB SS recursive - [2,16,19,21,27] XS/SB SS tree -XP/GAV: XML Publishing, Global-as-view XP/LAV: XML Publishing, Local-as-view XS/SO: XML Storage, schema-oblivious XS/SB: XML Storage, schema-based QT: Query Translation VD: View Definition SS: Storage scheme restricted XPath 1 : child and attribute axes restricted XPath 2 : child, attribute, self and parent axes The interaction between XML and RDBMS can be broadly classified as shown in Figure 1. The main scenarios are:…”