SYSTEMS AND METHODS FOR IMPLEMENTING AN XML QUERY LANGUAGE CLAIM OF PRIORITY This application claims priority to U. S. Provisional Patent Application No.
60/450, 381 filed February 27,2003, entitled"SYSTEMS AND METHODS FOR IMPLEMENTING AN XML QUERY LANGUAGE"Attomey Docket No.
BEAS-01447USO) ; and U. S. Patent Application No. filed February 26,2004, entitled "SYSTEMS AND METHODS FOR IMPLEMENTING AN XML QUERY LANGUAGE"Attorney Docket No. BEAS-01447US1), which are hereby incorporated herein by reference.
COPYRIGHT NOTICE A portion of the disclosure of this patent document contains material which is subject to copyright protection. The copyright owner has no objection to the facsimile reproduction by anyone of the patent document of the patent disclosure, as it appears in the Patent and Trademark Office patent file or records, but otherwise reserves all copyright rights whatsoever.
FIELD OF THE INVENTION The present invention relates to the querying of XML data.
BACKGROUND XQuery is a standard query language, published by the W3C (World Wide Web Consortium), that uses XML (Extensible Markup Language) notation to define query requests and handle query results. XQuery is designed to be an efficient language in which queries are concise and easily understood. XQuery is flexible enough to query a broad spectrum of data sources, including relational databases, XML documents, Web services, packaged applications, and legacy systems.
The XQuery language is derived from various sources, including SQL. It even provides a For-Let-Where-Return (FLWR-pronounced"flower") expression. Developers familiar with SQL will find XQuery very easy to learn.
BRIEF DESCRIPTION OF THE DRAWINGS Figure 1 is a diagram of exemplary datatype hierarchy showing complex data types directly by their structure, eventually reducing to a structured composition of simple types and their derivatives.
DETAILED DESCRIPTION Systems and methods in accordance with the present invention provide for an implementation of the XQuery standard, and such an implementation will be referred to herein as an XQRL (XML Query language) implementation. An XQRL implementation can be based on a token stream representation, which allows large XML documents to be processed without having to first read the entire document. An XQRL implementation can also utilize a streaming XQuery engine. A set of algorithms and a set of rewrite rules can be used for the implementation. Given a query, a set of rules can be applied to that query to transform the query into another query. Some of these rules allow the resulting query to be streamable. In other words, a standard query that would otherwise require an XML document to be read into memory can now be transformed into a query that can work on an XML stream.
In many cases, such a transformation can be applied to a query such that the entire query process is streamable. For example, a user could generate a query where data is to be sorted in a particular some order. Later, that query might be used in a larger context where the order is not important. An XQRL implementation can analyze the query, and can determine that it would be an unnecessary use of resources to sort the later result. Sorting typically requires all the data to be read into memory and stored until all elements are known and can be sorted accordingly. If the sorting step is eliminated from the later query, it is possible to simply process the XML on an XML stream.
A user or developer can add extensions or annotations to the XQuery language in order to address certain aspects that are not yet covered in the draft specification. A graphical tool, or graphical user interface, can be used with an XQRL implementation that allows a user to build data transformations. The transformation tools in the interface in at least one embodiment are sitting on top of the XQuery engines.
Token Stream A token stream (TS) is an XML representation that can be used throughout an XQuery language implementation, such as XQRL. A token stream can be compact and efficient. The tokenized format can represent the structure of an XML document with a minimal memory footprint, even in presence of repetitive elements. Although a token stream can be designed for the internal data representation of a processor, the stream can also be a convenient XML format for application interchange. A token stream can be used to easily manage XML fragments, and is easily serializable on both network and disk. If an application uses XQuery as a principal means for XML processing, intermediate results may be better kept in a format directly available for data processing.
An XQuery processor such as XQRL can process XML data in a streaming fashion. A data transformation operation, or query, in XQuery can be decomposed in a series of successive elementary operations on the data itself. An XQuery expression can be transformed by the XQRL processor in a sequence of elementary operators, with each operation applying a transformation on its input and feeding with its output the next operator.
Instead of completely consuming input before performing the transformation, each operator can consume a minimal amount of data and can immediately begin to produce results, which can be directly available to the next operator.
Using such an approach, an operator may not need to bufferize or materialize large amounts of data, thereby reducing memory consumption and processing time. Also, the result of such a query can be consumed and made available as soon as possible, often well before the input data has been completely consumed.
A streaming implementation can be consumer directed, such as may utilize a'pull'mode. A consumer of the result of the query can demand data tokens from the outmost operator composing the query. The outmost operator can, in turn, ask for data from the previous queries, and so on, up to the input source.
To make such a flow of data possible, the data can be represented in a format that is granular enough for this processing, such as the token stream. As the name implies, a token stream is a tokenized representation of an XML data fragment. Repeating structural elements can be replaced by tokens, which can retain the semantics of the original data element, discarding any umecessary infonnation accompanying those elements.
An XQRL processor can be implemented in the Java programming language, which can have tokens represented by objects, or the addresses of those objects. Repeating elements in the stream can be repetitions of the same pointer to a static object.
Tokens in a token stream can be immutable objects. Data transformations can be performed on the same input tokens coming from an XML parser or tokenizer. The output of the query can be made of the same exact tokens as the input, and possibly some generated by element constructors. Such an approach can keep the rate of object creation to a bare minimum during all the different phases of data transformation.
For example, consider the following XML data fragment: Schema : <xsd : complexType name="vote"> <xsd : simpleContent> <xsd : extension base="xsd : float"> <xsd : attribute name="index"type="xsd : int"/> </xsd : extension> </xsd : simpleContent> </xsd : complexType> <xsd : element name="judgement"type="vote"/> Data Instance : judgement index="11">43. 5</judgement> The tokenizer and the schema validator can transform the previous data fragment in the following sequence of tokens : [Begin Element] QName : [j udgement (http ://www. xqrl. com/examplel] QName : [vote@http ://www. xqrl. com/examplel] [Begin Attribute] QName : [index] QName : [int@http ://www. w3. org/2001/XMLSchema] int : 11 [End Attribute] [Begin Text] CharData :"43. 5" [End Text] float : 43. 5 [End Element] In the example, the first token [Begin Element] marks the beginning of the element judgement. Following is a QName token indicating the fully qualified name for the element, and another QName specifying the type of the element as it is defined in the schema. [Begin Attribute] marks the beginning of the attribute, following the QName with its name and type. A typed value token (represented by int: 11) gives the value of the attribute as an integer. [End Attribute] marks the end of the attribute.
According to the XQuery specification, the original lexical representation for each element should be available for processing, thereby the [Begin Text] CharData:"43. 5" [End Text] tokens, followed by the value token carrying the typed float representation of the element's data float: 43.5 finally the [End Element] token marks the end of the element judgment.
The type system of the XQuery language can be based on XML Schema.
XML Schema has a number of built-in (more or less) basic types (elements in the Figure 1 below : duration, dateTime, time, date, gYearMonth, gYear, gMonthDay, gDay, gMonth, Boolean, base64Binary, hexBinary, float, double, anyURI, QName, NOTATION, string, decimal) which exist as typed tokens in the Token Stream representation. Each basic type can have derived types, which are represented by typed tokens inheriting from the base type and having an additional field specifying the name of the derived type. The physical data representation and the default behavior is instead shared with the parent, the derivation being essentially a restriction of the parent type.
As seen in the previous example, complex data types can be represented directly by their structure, eventually reducing to a structured composition of simple types and their derivatives. An exemplary datatype hierarchy is given by Figure 1.
A second example shows an entire XML document and its tokenized representation. In the token stream the [Begin Document] and [End Document] tokens can be seen, as well as the name space declarations and a few additional data types.
The Schema : <? xml version="1. 0"encoding="UTF-8"? > <xsd : schema xmlns : xsd='http ://www. w3. org/2001/XMLSchema' xmlns="http ://www. xqrl. com/examplel" targetNamespace="http ://www. xqrl. com/examplel" elementFormDefault="qualified"> <xsd : complexType name="rootType"> <xsd : sequence> <xsd : element name="judgement"type="vote" minOccurs="I"maxOccurs="5"/> </xsd : sequence> </xsd : complexType> <xsd : complexType name="vote"> <xsd : simpleContent> <xsd : extension base="grade"> <xsd : attribute name="index" type="xsd : int" default="0"/> </xsd : extension> </xsd : simpleContent> </xsd : complexType> <xsd : simpleType name="grade"> <xsd : unions <xsd : simpleType> <xsd : restriction base="xsd : date"/> </xsd : simpleType> <xsd : simpleType> <xsd : restriction base="xsd : int"/> </xsd : simpleType> <xsd : simpleType> <xsd : restriction base="xsd : float"/> </xsd : simpleType> <xsd : simpleType> <xsd : restriction base="xsd : double"/> </xsd : simpleType> <xsd : simpleType> <xsd : restriction base="xsd : string"/> </xsd : simpleType> </xsd : union> </xsd : simpleType> <xsd : element name="root"type="rootType"/> </xsd : schema> A Data Instance : <? xml version="1. 0" encoding="UTF-8" ? > <root xmlns="http ://www. xqrl. com/examplel" xmlns : xsi="http ://www. w3. org/2001/XMLSchema-instance" xsi : schemaLocation="http ://www. xqrl. com/examplel tests/schema/examplel. xsd" > judgement index="11">43. 5</judgement> <judgement>3. 14</judgement> <judgement>pizza</judgement> <judgement>2002-06-14</judgement> </root> Tlze token stream : [Begin Document] [Begin Element] QName : [root@http ://www. xqrl. com/examplel] QName : [rootType@http ://www. xqrl. com/examplel] [Begin Namespace] QName : [@http ://www. xqrl. com/examplel] [End Namespace] [Begin Namespace] QName : [xsi@http ://www. w3. oi'g/2001/3CMLSchema-instance] [End Namespace] [Begin Attribute] QName : [schemaLocation@http ://www. w3. org/2001/XMLSchema- instance] QName : [anySimpleType@http ://www. w3. org/2001/XMLSchema] CharData :"http ://www. xqrl. com/examplel tests/schema/examplel. xsd" [End Attribute] [Begin Text] CharData :"\n\n" [End Text] [Begin Element] QName : [judgement@http ://www. xqrl. com/examplel] QName : [vote@http ://www. xqrl. com/examplel] [Begin Attribute] QName : [index] QName : [int@http ://www. w3. org/2001/XMLSchema] int : 11 [End Attribute] [Begin Text] CharData :"43. 5" [End Text] float : 43. 5 [End Element] [Begin Text] CharData :"\n" [End Text] [Begin Element] QName : [judgement@http ://www. xqrl. com/examplel] QName : [vote@http ://www. xqrl. com/examplel] [Begin Attribute] QName : [index] QName : [int@http ://www. w3. org/2001/XMLSchema] int : 0 [End Attribute] [Begin Text] CharData :"3. 14" [End Text] float : 3. 14 [End Element] [Begin Text] CharData :"\n" [End Text] [Begin Element] QName : [judgement@http ://www. xqrl. com/examplel] QName : [vote@http ://www. xqrl. com/examplel] [Begin Attribute] QName : [index] QName : [int@http ://www. w3. org/2001/XMLSchema] int : 0 [End Attribute] [Begin Text] CharData :"pizza" [End Text] string : pizza [End Element] [Begin Text] CharData :"\n" [End Text] [Begin Element] QName : [judgement@http ://www. xqrl. com/examplel] QName : [vote@http ://www. xqrl. com/examplel] [Begin Attribute] QName : [index] QName : [int@http ://www. w3. org/2001/XMLSchema] int : 0 [End Attribute] [Begin Text] CharData :"2002-06-14" [End Text] date : 2002-6-14TO : 0 : 0. 00 [End Element] [Begin Text] CharData :"\n" [End Text] [End Element] [End Document] Context XQuery programs can execute in a Context. The context can holds the variable bindings, namespaces, operators, and default parameters. Furthermore, the context can hold (static) references to an ID manager (IdManager) in charge of creating node Ds in XML documents, a type manager used by the static type inference engine and run-time type validation, and a schema manager or database of XML schemas.
The context can be entirely defined at the query compilation or preparation time, and may never be modified by the query execution. The compiler can also resolve all the elements of the context to direct pointer references in the compiled query. Consequently, a query execution engine may never need to resolve variables by their name at run time.
To implement the semantics of FLWR expressions and LET statements, implicit variables defined in [] xpath expressions, and sortby dots, a new context can be created holding the newly (implicitly or explicitly) scoped variables. This new context can be linked to its parent. All variables of the parent can be accessible from the context, possibly except for those that are redefined by the new context shadowing their parent counterpart.
To factorize things and to reduce start-up time, a base context can be used to hold the default environment, which can include standard namespaces, built-in functions and operators, and default behaviors.
Queries can reuse pre-initialized contexts, but they may not be able to directly modify those contexts. A new scratch context can always be created for query execution, which can inherit directly from the base context or from another explicitly-given context. Special queries can be used to pre-initialize contexts to be reutilized by subsequent queries, reducing the cost of context creation.
Given to their critical nature, execution contexts may not be directly accessible to the user. Execution objects can be opaque objects embedded in Connection objects, which are provided in one embodiment by an XDBC interface.
A Connection object can be created through an interface such as a DriverManager getConnection interface, and can be used to prepare statements which are subsequently populated using a special query prologue, such as a populateContext method of Connection.
Contexts can be saved for later use associating them to a URI, such as by using a publish method of a DriverManager, such that other contexts can be subsequently created inheriting from the saved ones.
Examples //derive a connection from the Base Context (default) Connection c = DriverManager. getConnection () ; //populate it with someXQuery prologue c. populateContext (prologue); //publish it! DriverManager. publish ("www. xqrl. com/contexts/stuffed", c); //some other time, some other place,... //derive a Connection from the previously saved one Connection c = DriverManager. getConmection ( "www. xqrl. com/contexts/stuffed", Connection. Type. INHERITED) ; //use it to execute your favorite query ! PreparedStatement ds = (PreparedStatement) c. prepareStatement (dataQuery) ; TokenIterator doc = ds. executeQuery () ; The foregoing description of preferred embodiments of the present invention has been provided for the purposes of illustration and description. It is not intended to be exhaustive or to limit the invention to the precise forms disclosed. Many modifications and variations will be apparent to one of ordinary skill in the art. The embodiments were chosen and described in order to best explain the principles of the invention and its practical application, thereby enabling others skilled in the art to understand the invention for various embodiments and with various modifications that are suited to the particular use contemplated. It is intended that the scope of the invention be defined by the following claims and their equivalence.