Conversion of XML schema design styles with StyleVolution

• Date: 24 August 2019
• Published date: 24 August 2019
• DOI: https://doi.org/10.1108/IJWIS-05-2019-0022
• Pages: 23-64
• Author: Zouhaier Brahmia,Fabio Grandi,Rafik Bouaziz
• Subject Matter: Information & knowledge management

Conversion of XML schema design

styles with StyleVolution

Zouhaier Brahmia

University of Sfax, Sfax, Tunisia

Fabio Grandi

DISI, University of Bologna, Bologna, Italy, and

Rafik Bouaziz

University of Sfax, Sfax, Tunisia

Abstract

Purpose –Any XML schema definition can be organizedaccording to one of the following design styles:

“Russian Doll”,“Salami Slice”,“Venetian Blind”and “Garden of Eden”(with the additional “Bologna”style

actually representingabsence of style). Conversion from a design style to another can facilitatethe reuse and

exchange of schema specifications encodedusing the XML schema language. Without any computer-aided

engineering support,style conversions must be performed very carefullyas they are difficult and error-prone

operations.The purpose of this paper is to efficiently deal withsuch XML schema design style conversions.

Design/methodology/approach –A general approach, named StyleVolution, for automatic

management of XML schemadesign style conversions, is proposed. StyleVolutionis equipped with a suite of

seven procedures: four for converting a valid XML schema from any other design style to the “Garden of

Eden”style, which has been chosenas a normalized XML schema format, and three for converting from the

“Gardenof Eden”styleto any of the other desired design styles.

Findings –Procedures, algorithms andmethods for XML schema design style conversions are presented.

The feasibility of the approach has been shown throughthe encoding (using the XQuery language) and the

testing (with the Altova XMLSpy 2019 tool) of a suite of seven ready-to-use procedures.Moreover, four test

proceduresare provided for checking the conformanceof a given input XML schema to a schema design style.

Originality/value –The proposed approach implements a new technique for efficiently managing XML

schema design style conversions, which can be used to make any given XML schema file to conform to a

desired designstyle.

Keywords Schema change, Schema evolution, XML database, XML Schema,

XML schema design style, XML schema translation

Paper type Research paper

1. Introduction

The eXtensible Markup Language XML (W3C,2008) along with the XML Schema language

(W3C, 2004;Van der Vlist, 2011) is likely the description and specification formalism that

has had the most significant impact on the development of Web-related technologies and

applications. In particular, although XML has been designed as a general-purpose data

storage and exchange format, it has been widely exploited for the representation and

management of (semistructured) data in Web-based information systems, for which XML

schema is the elective data modeling formalism (Abiteboul et al.,2011;Aiken and Allen,

2004;Chaudhri et al.,2003).

Nowadays, XML and all languagesbased on it, like XML schema, XQuery (W3C, 2014),

XQuery Update Facility,XPath and XSLT, continue to be of great interest for the developers

Conversion of

XML schema

design styles

23

Received2 May 2019

Revised28 June 2019

Accepted4 July 2019

InternationalJournal of Web

InformationSystems

Vol.16 No. 1, 2020

pp. 23-64

© Emerald Publishing Limited

1744-0084

DOI 10.1108/IJWIS-05-2019-0022

The current issue and full text archive of this journal is available on Emerald Insight at:

https://www.emerald.com/insight/1744-0084.htm

of modern XML-based applications (e.g. Web services, cloud computingapplications, social

network applications, e-commerce systems) and designers/administrators of XML

repositories or databases (Bourret,2005, 2010). Among the most important requirements of

these actors, we find those concerning application maintenance (application source code

correction, applicationextension, XML schema changes, etc.). In general, changesperformed

on XML schema files are error-prone and time consuming tasks (Klímek et al.,2015), since

they should be accomplished manually, as there is often no technical support (e.g. stand-

alone or IDE-embeddedcomputer-aided engineering tools) for performing them.

In the present work, we continue our research work on changes to the overall XML

schema design patterns, which already covered changes to XML namespaces in (Brahmia

et al., 2016a,2016b). In particular, we focushere on changes involving XML schema design

styles. In fact, any XML schema could be designed according to one of the following five

styles (Maler, 2002;McBeath et al., 2004;Lämmel et al.,2005;Darr et al.,2011;Range

Commanders Council (RCC), 2015;xFront, 2018): “Russian Doll”,“Salami Slice”,“Venetian

Blind”,“Garden of Eden”, and “Bologna”. They differ on the way they define, globally or

locally, XML schema components: element declarations, attribute declarations, simple type

definitions, and complex type definitions. A global component is an immediate child of the

root <xsd:schema>element; it is also automatically associated with the target namespace of

the XML schema, and therefore it couldbe re-used in other XML schemas. However, a local

component is not an immediate child of the <xsd:schema> element and, thus, it is not

visible from the outside of the schema definition. The “Russian Doll”design style means

having only one global element declaration that nests all other possible element/attribute

declarations and simple/complex type definitions. In the “Salami Slice”style all element/

attribute declarations are globally defined and all simple/complex type definitions are

locally defined. In the “VenetianBlind”style all element/attribute declarations are defined as

local components and all simple/complextype definitions are defined as global components.

The “Garden of Eden”style defines all element/attribute declarations and all possible

simple/complex type definitions as global components. Otherwise, an XML schema can be

by default considered as conforming to the “Bologna”design style that is actually not a

defined style[1].

Considering an XML database (DB),that is a repository of XML-encoded documents each

one conforming to an XML schema specification, the XML DB designer could need to

translate an XML schema from its current design style to another one or putting an XML

schema, whose style is undefinedor unknown, into any desired style.

The reasons behind a style conversionmay be different and involve the features of global

versus local definitions as described in the following. On the one hand, the use of local

definitions emphasizesdecoupling (i.e. definitions are self-contained, without dependenceon

other components) and cohesion (i.e. related data are grouped together into self-contained

definitions) of specifications, and supports namespace complexity hiding (according to the

value of the elementFormDefault). Notice that complex schema definition details can be

deliberately maintainedhidden via local definitions to namespace users, also to preservethe

intellectual property of the design (e.g. by adopting a “Russian Doll”style). On the other

hand, the use of global definitions improves the sharing and re-use of specifications, as the

definition of some subcomponent can be referenced (without code duplication and, thus,

reducing the verbosity of schema definitions) by several component specifications in the

same schema or made available to the designers of other schemas. With global definitions,

the full complexity of namespaces can be exposed and the collaborative and incremental

definition of even more complex schemas and namespaces can be supported.Reusability of

components can be made available at different levels, according to the adopted style: the

IJWIS

16,1

24

“Salami Slice”style only allows the re-use of element/attribute declarations, the “Venetian

Blind”style only allows the re-use of simple/complex type definitions, whereas the “Garden

of Eden”style allows the re-use of both element/attribute declarations and simple/complex

type definitions.

Nevertheless, schema translation operations are not straightforward and should be

performed carefully, since,from one hand, there are no available tools that allow performing

such style design changes,and, from the other hand, some style changes can be difficultand

error-prone to manually apply on large schemas and may also have side effects on the

underlying XML instance documents(in such a case, they modify not only the XML schema

presentation but also the XML schema specification).Therefore, our purpose is to allow the

XML DB designers/administratorsto automatically:

change the design style of any XML schema to another design style;

put an XML schema whose style is undefined or unknown into any desired style;

correct design style errors in some existing XML schemas supposed to be in a given

design style but not completely conformant (e.g. because developed by non-expert

designers); and

make available for re-use definitions that are local in a schema, via transformation

into global definitions in a new schema.

This should be done by means of automatic design style conversion tools, which should

reduce the intervention of the schema designer to the minimum and generate, in a

transparent manner, a new version of a given schema, semantically equivalent to it but

conformant to the target style.

To this purpose, in this paper, we propose StyleVolution,a suite of seven procedures for

efficiently managing design styles of XML schemas:putting an XML schema whose design

style is undefined or unknown into a desired style, or converting an XML schema that is

designed accordingto a given style into a different design style.

To define such procedures, we started by choosing the “Garden of Eden”design style as

a normalized style, thanks to the fact that it maximizes reusability of specifications, since it

globally exposes all element/attribute declarations and all simple/complex type definitions.

After that, we have definedseven translation procedures that will be presented in thispaper:

four procedures, named RD2GE, SS2GE, VB2GE and BO2GE,for translating any “Russian

Doll”,“Salami Slice”,“Venetian Blind”or “Bologna”XML schema to an equivalent schema

designed according to the normalized style, respectively; and three procedures, named

GE2RD, GE2SS, and GE2VB, for translating a “Garden of Eden”XML schema to an

equivalent schema in any (desired) one of the three other non-defaultdesign styles, namely

“Russian Doll”,“SalamiSlice”or “Venetian Blind”, respectively.

Any design style conversion could then be performed with the direct use of either one

procedure (when the target or the source design style is “Garden of Eden”), or with a

combination of two procedures, one from the first subset {RD2GE, SS2GE, VB2GE,

BO2GE} and the other from the second subset {GE2RD, GE2SS, GE2VB} (i.e. passing

through the “Garden of Eden”styleas an intermediate step of the conversion).

The rest of this paper is structuredas follows. Section 2 describes the main XML schema

design styles found in the literature and used in XML-based application development.

Section 3 introduces the procedures that we propose for converting any XML schema,

having a defined (“Russian Doll”,“Salami Slice”,“Venetian Blind”,“Garden of Eden”)ora

non-defined (“Bologna”) design style, into any other desired defined style. Section 4 deals

with effects of changes to XML schema design styles on underlying XML document

Conversion of

XML schema

design styles

25