Map Dataset to XML and XML to Dataset

Introduction

Since the arrival of the Internet, people have become more and more dependent on it for daily operations. The Web's (as well as computing applications') insatiable appetite for data has grown even more intense. Most websites have seen the interplay of relational database and XML. Websites rely on relational databases for storing a vast amount of data and performing fast retrieval and manipulation. For exchanging data, they opt for protocols such as SOAP or REST (what strange names!). In these cases, XML serves both as the foundation and the ultimate representation form. Let's briefly look at the origin of XML and make a skim comparison between a relational database and the XML data model.

A brief history of XML

Relational database was born out of the ideas of an IBM mathematician and computer scientist, E. B. Codd, in the early 1970s. Since then it has become the "IT" player in the database industry. In comparison, XML is the new kid in town and it is the result of a never-ending quest for a simpler yet more powerful language for data storage and communication.

In 1986, SGML or "Standard Generalized Markup Language" was issued as an international standard. It is a complex and somewhat rigid language that requires expensive authoring tools. SGML was used extensively in large government, legal and publishing industries. In 1990, a simpler and more flexible language called HTML was created and it soon became enormously popular. However, its shortcomings also became increasingly visible. In 1996, the World Wide Web Consortium (W3C) (underwritten by Sun and other outside organizations) formed a committee to explore ways of harnessing the strength of HTML and SGML while sidestepping its limitations. The result is the creation of XML.

Relational Database and XML data model

A relational database consists of a collection of two-dimensional tables of rows and columns, interconnected by keys shared across tables. Through relations (shared keys), tables can be joined, queried and rearranged in different ways. There is no hierarchy or sequence to speak of.XML on the other hand is all about hierarchy and sequence, as it is mirrored in such terms, parents, children and siblings (it might as well be superiors, subordinates and colleagues).

Since its debut, XML has wooed techies and laymen alike with its easiness and flexibility; and the fact that it is human-readable. However, as much as XML has gained enormous momentum and widespread usage, relational database has long been the backbone infrastructure of enterprise applications; and proven to be effective and efficient. In the foreseeable future, relational database technology will remain the dominant technology, with major adoptions with respect to XML.

Many relational database vendors have provided extended database features to incorporate XML technology. For instance, SQL server has XQuery support and has a new XML data type. Likewise, middleware vendors have also provided native support for XML, like ADO.NET.

In ADO.NET, the central data object - the DataSet - represents a hierarchical, disconnected data cache. Its design is based on XML, which makes the translation between relational data and XML easy. The following discussion will use some examples to illustrate the ways to map a DataSet from and to a XML file.

Load a DataSet from XML

To load a DataSet from an XML file, simply call the DataSet.ReadXml method, like so:

Figure 1 shows a screenshot of the output:

Figure 1: Screenshot of the output

In this example, a flat XML file with no hierarchy or nested element is selected. As a result, the DataSet contains only one table and it can be conveniently bound to a GridView.

Loading Schema Information from XML

The ReadXmlSchema or the InferXmlSchema methods of the DataSet allow you to load DataSet schema information from an XML document. Alternatively, you may use the overloaded ReadXml method with XmlReadMode.InferSchema as the second argument.

DataSet Schema information can be written to an .xsd file using the WriteXmlSchema method:

The following is the input XML file, books2.xml:

And here's the output file, books2.xsd:

The following figure shows a screenshot of the output table relationship:

Figure 2: Output table relationship

A somewhat more complicated XML file with a hierarchy of 3 tiers has been chosen for the example. It is worth noting how the relationship is constructed from the XML to DataSet transformation. The hierarchies are automatically translated into three tables and primary-foreign key pairs are inserted.

DataSet with a single table to XML

Want to map a DataSet containing a single table to XML? Piece of cake. The following code illustrates how to fork some information from an Access database into a table and write it to a XML file.

Here's a segment of the output XML file:

As you can see, the root element of the XML file takes the name of the DataSet. In .NET 2.0, many of the tasks that could previously only be performed by a DataSet can also be tackled by a DataTable; such as XML transformation.

DataSet with two tables in master-detail relationship

It gets a little complicated and code-heavy to map to an appropriately nested XML from a DataSet with master-detail tables. First, you need to specify the primary key of the master table. Second, you have to register the foreign key of the detail table. Finally, you must set to true the Nested property of the DataSet.