Home

Master`s Thesis

1. Supports transactions and user authentication E2 E3 Supports binary data and Unicode E4 Does not support server side XSLT processing No portability between platforms due to the ActiveX technology Reduced performance because of the use of DOM S1 Services U2 5 4 3 Native HTTP Client API Last but not least the database can simply be accessed by utilizing the HTTP protocol This requires a prior installation of an HTTP server on the server computer and an additional module which connects the server to the database As with the 54 CHAPTER 5 DESIGN Tamino API for C the native interface communicates directly with the X Machine and thus shares the same functionality Furthermore the HTTP API can be used in conjunction with the Tamino Passthru Servlet a Java component which runs on the server and supports the execution of XSLT stylesheets The Passthru Servlet in turn can only be executed in a proper system environment namely a java servlet engine The Pros and cons of this interface are Portable between platforms P1 Good performance due to direct access to the X Machine S1 Supports server side XSLT processing Does not require additional software on the client U2 Supports transactions and user authentication E2 E3 Supports binary data and Unicode E4 Services U2 5 4 4 Conclusion As one can easily see the best choice out of these three is the last one the Native HTTP i
2. www mydomain com xml person xmlns computer http www mydomain com xml computer gt lt person address gt New York Broadway 173 lt person address gt lt computer address gt 192 168 0 10 lt computer address gt lt form gt 16 CHAPTER 2 RELATED WORK The root tag form contains the declaration of two namespaces person and computer which start with xmlns followed by and a shortcut for the namespace called namespace prefix The string which is assigned to these attributes is used to identify the namespace and is usually a worldwide unique URI Universal Resource Identifier The namespace prefix is then utilized in the remaining document to refer to these declarations It is written preceding to the tag followed by and the name of the tag This way the two address tags can be distinguished from each other See 8 for a comprehensive description of XML namespaces 2 1 4 Selecting data with XPath XPath is a standardized addressing scheme for selecting a node or a set of nodes from XML data The selected nodes can then be further processed by an application or by other languages like XSLT or XQuery In fact XPath is specified as part of XSLT the W3C standard for XML transformation which is described later in this chapter XPath allows to specify a location path which defines the context of the nodes to find and a condition the selected nodes must satisfy These path expressions
3. 3 4 Mntrod ction ad a a Ta XME Database SEVA A Klient interface a a cae Sap ear eh ee des ds SE a RS Data representation in form of SOXML 1 11 11 11 12 15 16 17 18 20 21 23 23 24 25 25 25 26 26 27 28 28 30 CONTENTS 3 5 Specification of a test application Gar area 35 3 6 Implementation E O a a A ia 39 Choosing a database product 37 A ci ee A Glad We eae A Guts a 37 A ANAIS ei tod El ly Ale Ge Seok da Oe itn a A 37 ADA installation a ee do Re e lema 38 4 22 Applications s aa a e ea A Ee a A 38 4 2 3 Database features y Po E Er En rel 39 AZAL CAPI A A E a N a a an an 39 4 2 5 Documentation 43 vate ded bak AA AAA OO 39 A a un he Be ee a ne 40 4 3 1 Installation Loy ey An Bere Bist Hoe amp ii Be 40 4 3 2 APPUCAMORE ui aia ee A A when 41 4 3 3 Database features me na ae ehr ara 42 AE AS A A Da N PO re N a sel 42 4 3 5 Documentation li at a e es 43 AA EXIS E AAA e A A 43 TAL Installatien et tios rs y e Bea eI 43 4 4 2 PPC LOS ai o Ge DI e A 43 4 4 3 Database features ta be eee kw ra oa 43 4144 APLS IAEA e Re Boe ee ee Ss A 44 4A Documentation r p s Toe O A a ee 44 ADs Conclusion 2 415 5 eaea ar A tes 44 Design 47 Jell COVEI EW ES a a EE ee E aA 47 5 2 Environment analysis 2 32 2 43 22 22 28 Ar 4 He ei 47 Rt Design iss es let 20 gs ee da Aes 48 aL Portability B ah 5 Sethe sand e a Bent e Bees Na AE 49 5 3 2 Modularity and Reusability M oi ke bo oe be
4. Done manually in Tamino Manager XIndice No Yes but no support for XPath queries across collections Yes user indexes only Yes by programming extensions for the servlet engine No No Unclear Logically at node level and physically at document level No eXist No Yes Yes automatic full text index Yes by programming extensions for the servlet engine No Yes Unix like access permissions for users groups at collection and document level Yes At node level Yes Done via Shutdown Copy amp Delete Command Line or Gui complete database Client SAX Simple API for XML DOM Document Object Model XML DB API Data update JDBC for XML JAXP Java API for XML Processing JDOM Java Document Object Model Other Programmatic APIs Tamino XIndice Language independent API s Yes as part of Tamino API for Java HTTP Client API for ActiveX C Visual Basic Java JScript Tamino API for Java DOM2 Yes implemented in Java Yes implemented in Java Java only API s No Yes Core Level 1 implementation XUpdateQueryService CollectionManagementServ ice Yes Yes No API for the X Machine programming language XinCJ a subset of the XML DB interface for C eXist Yes implemented in Java as part of the XML DB API Yes implemented in Java as part of the XML DB API Yes Core Level 1 implementation UserManagementSer
5. Each well formed XML document has to start with a declaration which specifies the XML version and the character encoding used in the document the so called XML Declaration It has to be the foremost item of a document The following example specifies to use XML version 1 0 and character encoding UTF 8 lt xml version 1 0 encoding UTF 8 gt This declaration tells an application which processes the document how to interpret the remainder of the XML document Although the W3C standard defines it as obligatory it is often omitted in practice Tags Tags are used to define the logical structure of an XML document A tag is a text which is enclosed by angle brackets lt gt There are always pairs of tags A start 2 1 XML 13 tag and an end tag which enclose a part of a document and thus give it a certain meaning The name of the end tag is the name of the start tag preceded by a Such a pair of tags is also referred to as XML element The following example shows an XML element which marks a text as a caption lt caption gt The building blocks of XML lt caption gt Start and an end tags without any text between them are called empty elements Since it is quite usual to use empty elements the W3C standard defines an abbreviated form for that Instead of writing lt caption gt lt caption gt one can also write lt caption gt Normally an XML document will contain more than one XML element
6. In order to run eXist in servlet mode the distribution already contains the servlet engine Jetty so there is no need to install one For starting and stopping the server and client applications there are some batch files supplied We had to edit these batch files slightly to get them to work 4 4 2 Applications eXist comes with an integrated combined graphical and command line Java application called Client Shell It offers functions for managing collections querying and uploading data and for creating database backups This application is similar to the previously mentioned XMLdbGUI which can be used with eXist too Furthermore eXist comes with a local copy of its web site which includes several simple forms for upload query and a web administration interface All in all to adjust database specific features one still has to edit the configuration file of eXist with the help of a text editor but the daily used functions are covered sufficiently by the Client Shell and by XMLdbG UI 4 43 Database features eXist currently supports XPath and XUpdate and support for XQuery is on its way Though on the one hand XPath is implemented with extensions offering additional 44 CHAPTER 4 CHOOSING A DATABASE PRODUCT functions to speed up query times its implementation is not fully W3C compliant because it has not been finished yet To improve query speed eXist makes use of an automated full text index which can also be turned off if
7. TransactionService Lycos Inheritance gt setContent mo getParentCollection Association gt rollback Figure 2 3 Collaboration diagram of XML DB API database classes thus has been designed to provide as much flexibility as possible For this reason it is also the most complex interface of those examined here We will only discuss the most important aspects of this class library here For the interested readers we recommend to read the introduction written by Kimbro Staken 31 and to have a look at the API specification 30 Figure 2 3 on page 29 shows a collaboration diagram of the main classes of this API In the middle of this figure one finds the Collection class which is the counterpart to the classes TDatabase and CDatabase of the previous sections It represents an XML Database collection which generally contains a set of XML documents and further sub collections A collection is the data source of queries and the destination of updates which are represented by the classes X PathQueryService and X Update Query Service The resulting XML documents or document fragments of a query are stored in an instance of ResourceSet whose items can be iterated by Resourcelterator These items are stored in objects of the XML Resource and BinaryResource type not depicted here because of the lack of space which are derived from the abstract base class Resource A bit misleading is the name of the class Database which
8. string variable xmlText and then written to the XML file configuration xml 6 7 Finally the reference counter of the root object needs to be decreased in order to delete it Appendix C BAUMLBrowser manual This manual describes the BAUMLBrowser application a database browsing tool which is used to browse and manipulate geometric information stored in an XML Database BAUML the Building AUgmentation Markup Language is an XML language to store geometric information of buildings and parts of those buildings like walls floors and corridors The recursive definition of the language allows to create a tree structure of spatial objects where objects are composed of a number of smaller objects The BAUMLBrowser application is capable of viewing and browsing such a BAUML tree and offers functions to insert update and delete tree nodes On the one hand it is an example application that demonstrates the usage of the Studierstube XML Database API and on the other hand it provides a reusable layer on top of this API to process BAUML documents The following sections describe the user interface of the application and the reusable core component C 1 User Interface Guide BAUMLBrowser is a dialog based application written with Microsoft Visual C utilizing the Microsoft Foundation Classes MFC The following section describes the meaning of the various control elements of this dialog and how you can use it to edit aBAUML docum
9. stylesheet of the class Query which is used in the example of the previous section This parameter is a URL identifying the location of the stylesheet The stylesheet might be retrieved from file store or directly from the XML server If the value of the parameter is an absolute URL the stylesheet is retrieved using that URL If the value of the parameter is a relative URL the stylesheet is retrieved from the server as a named document relative to the collection of the current query See the following examples Assume that the queries are run on the collection meetings in the database staff When the query of the first example is executed it processes the result of this query using the stylesheet getresult xslt found in collection stylesheets of the database staff query stylesheet stylesheets getresult xslt The second example obtains the stylesheet from the file store of the web server not from the XML server Here the stylesheet getresult xslt is located in the directory stylesheets of the local web server query stylesheet http localhost stylesheets getresult xslt B 1 DATABASE ACCESS MODULE 97 Updating documents The W3C consortium has already passed a standard for the XML query language XQuery But the current version 1 0 does not comprise any update operations The Tamino XQuery language enhances this standard by supporting additional language constructs for insertin
10. 2 1 Server components As outlined in the design chapter an essential portion of the database server is made of third party software Since we decided to employ the Tamino Passthru Servlet in our project we also needed to choose a servlet engine and a web server in order to properly run the servlet Figure 6 4 on page 67 depicts the logical structure of the server including the most important configuration scripts In the following sections we only discuss the function of these components see the appendix of this work for a detailed description of how to install them Apache Web Server Apart from the fact that several other components of the Tamino XML server require the installation of a web server the Tamino Passthru Servlet needs one in order to query database Basically this task can be carried out by any web server which is capable of forwarding requests to particular web locations Since the Tamino setup 68 CHAPTER 6 IMPLEMENTATION procedure directly supports the use of the Microsoft IIS and that of the Apache web server it is easiest to employ one of these two As Apache is in contrast to the Microsoft IIS free of any charges and we only need it to implement the database access we decided to employ Apache Furthermore Apache is available for several operating systems including Windows and Linux design request P1 and at the time of writing the most popular web server on the Internet and thus a well tested piece of softwa
11. 31 2001 3 Nigel Bevan Jurek Kirakowski and Jonathan Maissel What is Usability In Proceedings of the 4th International Conference on HCI Stuttgart September 1991 4 Steven Clarke Measuring API Usability Dr Dobb s Journal Special Windows NET Supplement May 2004 5 Gerhard Reitmayr On Software Design for Augmented Reality Dissertation at Technical University of Vienna 2004 6 Helmut Erlenk tter XML Extensible Markup Language von Anfang an Rowohlt Taschenbuch Verlag September 2003 ISBN 3 499 61209 7 David Gulbransen et al Using XML Second Edition Que Publishing June 2002 ISBN 0 7897 2748 x 8 W3C Consortium Extensible Markup Language XML http www w3 org XML 1996 2003 9 W3Schools XPath Tutorial http www w3schools com xpath 10 Miloslav Nic Jiri Jirat XPath Tutorial http www zvon org xxl XPathTutorial General examples html 2000 11 W3Schools XSLT Tutorial http www w3schools com xsl 12 W3Schools XQuery Tutorial http www w3schools com xquery 13 W3Consortium XQuery 1 0 An XML Query language 125 126 LIST OF FIGURES http www w3 org TR xquery 14 W3Consortium XML Syntax for XQuery 1 0 XQueryX http www w3 org TR xqueryx 15 W3Consortium XML Schema http www w3c org XML Schema 16 W3Schools XML Schema Tutorial http www w3schools com schema 17 Roger L Coste
12. Product name Tamino Xindice eXist http xml apache org xi http exist sourceforge n Home URL http www tamino com TE ai a v3 1 2 1 v1 1b3 dev v0 9 2 Version Sept 2002 Dec 2003 Aug 2003 License Commercial Open Source Open Source Windows NT Implemented in Java Implemented in Java Windows 2000 platform independent platform independent Y Sun Solaris 7 and 8 IBM AIX 4 3 3 Tested on Tested on Y HP UX 11 0 MS Windows Windows 2000 Supported Platforms a Sn a a Unix Linux Windows XP OS 390 mainframe Linux SuSE 7 1 Mandrake 9 1 Y Solaris 8 Apache Web Server Sun Java SDK Sun Java SDK Version gt 1 3 24 Version 1 3 or higher 1 4 Required Software Microsoft Java Virtual Y Tomcat Servlet before installation Machine gt 3805 Container Version hard to get 4 1 12 or higher a Test system Intel Pentium III 800Mhz with 256 MB RAM MS Windows 2000 Professional German with Service Pack 4 Internet Explorer 6 Service Pack 4 Windows Update Security update amp Java VM 3810 MS Office 97 Word Excel Power Point Access German Frontpage 2000 German MS Visual C 6 English Sun Java 2 SDK v1 4 2_02 Apache v1 3 24 Web server Tamino XML Server 3 1 2 Patch Level 1 Tomcat Servlet Container v4 1 12 Xindice v1 0 and v1 1b3 dev eXist vO 92 XPath Path like data query XQuery SQL like data query XSD XML Schema Definition XSL XSLT Data
13. Ronald Bourret gt This paper gives a high level overview of how to use XML with databases It describes how lt article gt lt article title eXist An Open Source Native XML Database author Wolfgang Meier gt eXist is an Open Source effort to develop a native XML database system which lt article gt Basically it does not matter whether to use a pair of tags or an attribute to store a value as long as it can be expressed as a single line string The biggest difference between tags and attributes is that attributes cannot be further structured like tags Attributes are well suited to store atomic values whereas tags are used when storing structured or big sized data CData Some characters like angle brackets commas and apostrophes have a special meaning in XML They cannot be used directly in regular text sections Though there is a way to include them by using special expressions sometimes a text must be kept in its original form This is where the CData section comes into play It allows to store arbitrary text in an XML document The following example shows how you can use this language element to include special characters in an article lt article gt lt CDATA Here can come any text including special characters like lt gt and amp gt lt article gt A CData section is always enclosed by the strings lt CDATA and gt Refer to 7 for a more comprehensive description of the CData
14. Tamino database utilizing the proposed client interface Rectangles framed by a bold line denote planned components which have to be implemented by us whereas existing applications and components are marked by a thin frame Arrows between these components indicate the direction of the data flow that is caused by the query Solid arrows show the data flow of the client request and dotted arrows the response of the database server The dashed line in the middle of the diagram divides the system into client and server which are discussed separately in the following sections 5 5 3 Client The left side of figure 5 2 depicts the client which is designed according to our own approach instead of implementing the XML DB API All topics discussed in this section refer to this part of the diagram The components with a bold frame together form the Studierstube XML Database API which can be utilized by Studierstube applications to query the database and to process XML data Each of these components 5 5 SYSTEM ARCHITECTURE 57 can be thought of as an instance of a C class offering various methods The most important part of this diagram is the Query Database pair which is used to execute database queries This concept is adopted from the Borland VCL class library because of its simplicity and effectiveness The basic idea is to have a dedicated passive container class which subsumes all parameters related to a database query in a single componen
15. The worldwide proliferation of PC systems and the resulting deterioration on prices for computer memory enabled the distribution of this technology Strictly speaking the additional bytes needed by XML to store data are not really wasted but provide information about the structure of the data This way it is possible to write generic applications which can process many kinds of different document formats This advantage can not be achieved when using a proprietary binary data format Nowadays the ever growing pool of Open Source Software offers a vast amount of freely available applications and utilities which support the processing of XML data As with many new technologies universities and research labs are the first ones 10 CHAPTER 1 INTRODUCTION employing them like the Interactive Media Systems Group at Vienna University of Technology did with XML Meanwhile a large number of different XML documents have been accumulated as single files stored on workstation computers of staff members Thus the file system is not an adequate storage medium any more and another solution is needed to manage the data Since databases have proven to be a good technology to store and query data in a scalable manner the idea was born to employ an XML database as a replacement for the file system At the time of writing XML Databases are a relatively new technology The experiences in their robustness and usefulness are quite limited compared to those wi
16. There are two ways to align multiple elements The first way is to align them serially like in the following example which represents a list of two captions lt caption gt The building blocks of XML lt caption gt lt caption gt Using namespaces lt caption gt The second way is by nesting them This enables to structure a document hierarchically The following example shows how previous captions can be defined as part of an article element lt article gt lt caption gt The building blocks of XML lt caption gt lt caption gt Using namespaces lt caption gt lt article gt There is one exception where it is not allowed to align XML elements serially This is the root tag of the document The W3C standard specifies that a well formed XML document must have only one topmost tag and all other tags have to be part of it 14 CHAPTER 2 RELATED WORK Attributes Beside the way to store data between a pair of tags it is also possible to store data in form of attributes An attribute is a named parameter which can be attached to a start tag Each start tag can own an arbitrary number of attributes This is accomplished by adding the name of the attribute followed by a sign and the value of the attribute enclosed in apostrophes The following example shows a list of two articles where the title of the article and the name of the author are stored in attributes lt article title XML and Databases author
17. Transformation Server side XUpdate Data update SiXDML Simple XML Data Manipulation Language DTD Document Type Definition XPointer Data Selection XLink Links across document boundaries XInclude Including document fragments XML standards Tamino Yes in an adopted form called X Query or XQL No Yes in a form called Tamino schema not fully compliant with the standard Yes by using X Tension and a third party XSLT processor No No Yes by importing into Tamino Schema Editor Yes No XIndice Yes implemented by Xalan No No No Yes using the Lexus implementation Yes by sixdml at sourceforge No No No Yes partially eXist Yes Yes but not complete yet Yes when using HTTP Yes but not complete yet Yes Yes partially No Yes Schema support Collection support Indexes Server extensions Transaction support Authorization mechanism Multi user access concurrency Update granularity Backup Restore strategy Database features Tamino Yes but not mandatory Yes but no support for nesting Yes based on schemas X Tension service Java C C Yes Yes at different object levels Yes designed for high concurrence The smallest unit of XML that Tamino can process is a document You can use NodeLevelUpdate if you use Microsoft Internet Information Server IIS as your web gateway to Tamino Yes
18. a Node object has been selected to which you want to add a new child object 2 3 Using the function readFile the child object is loaded from disk and stored in a string variable 4 Finally the content of the file is added to the child list of the BAUML object by calling insertObject C 2 7 Special functions Beside the standard operations for querying and updating the database the BAUMLBrowser class provides two test functions for object intersection and point inclusion The intersect method tests if the representation part of two BAUML objects share a common region in space It uses the SolntersectionDetection Action class of Coin to implement this functionality Due to this the test is a bit limited meaning that an intersection is found only when the planes of the surface of two objects do intersect So for example if an object is completely contained in the other one the intersection test will fail although it should deliver a positive result The second operation containsPoint tests whether a certain three dimensional point is contained in the representation part of a BAUML object It does this by intersecting the object with a ray which starts at the test point and counting the intersection points In case of an odd quantity of intersection points the test point lies inside the object otherwise outside The following example tests whether the three dimensional point x 1 y 2 2 3 is in
19. around the Xalan C library and tries to hide the quite flexible but sometimes really complex interface of Xalan The following example uses this class to extract the result set from the database response of the previous section For this purpose we need an appropriate stylesheet which implements the extraction The following stylesheet is copied from the BAUMLBrowser demo application and does exactly what we want lt xsl stylesheet version 1 0 xmlns xsl http www w3 org 1999 XSL Transform xmlns stbxml http www studierstube org xml xmldb response gt lt xsl output method xml indent yes omit xml declaration yes gt lt xsl template match gt lt xsl copy of select stbxml response stbxml result x gt lt xsl template gt lt xsl stylesheet gt The stylesheet copies the content of the path stbxml response stbxml result to the output when the root of the input document is reached Note that we had to declare the Studierstube namespace in line two to be able to access the right tags Assuming that this stylesheet is stored in the file getresult xslt the following code example shows how to apply the transformation to the server response of the previous section It extends the previous source code sample by adding a transformation before the output is printed to the screen added 1 we additionally need the Transformer header file include lt transformer h gt 5 Run the query on
20. contained in a database As mentioned before BAUMLBrowser uses the Node class to represent BAUML objects A Node instance does not contain the XML object itself but the information needed to build up a tree control illustrating the structure of the BAUML tree Furthermore it holds location information which is required to identify the associated BAUML object The following code is a simplified version of the update function of the BAUMLBrowser application It adds child items to an existing tree control item void CBAUMLBrowserDlg AddChildItems HTREEITEM hItem 1 get the associated node of the tree view item BAUMLBrowser Node node BAUMLBrowser Node m_TreeCtrl GetItemData hItem 2 get all child items of the node from the database BAUMLBrowser NodeVector childList m_Browser getNodeChildren node childList 3 for each node found in the database insert a new treeview item for int i 0 i lt childList size i TVINSERTSTRUCT tvlnsert BAUMLBrowser Node childNode childlist i the new items are children of the current item tvInsert hParent hlten 4 set the name of the new item tvInsert item pszText _T childNode gt getName c_str 5 add a plus sign if the Node has children tvInsert item cChildren childNode gt hasChildren 1 store the Node object in the treeview item tvInsert item 1Param LPARAM childNode insert the item into the treevieu m_TreeCtrl InsertItem amp
21. especially to store XML documents Their internal logical model is based on XML but is not a question of the physical storage 26 CHAPTER 2 RELATED WORK model The physical backend of this database type can still be a Relational Database like the first version of eXist showed One important characteristic of Native XML Databases is that they preserve important properties of XML documents like element order and XML specific elements like processing instructions and comments Which means that what one puts in will also come out A further characteristic is the support for XML query languages like XPath or XQuery Naturally Native XML Databases should also offer common database features like transactions and multi user access Native XML Databases are best used when the data is semi structured The structure varies and a mapping to a Relational Database is difficult to achieve For instance object oriented tree structures which make use of derivation would need an additional table for each derived class Native XML Databases can handle both categories of XML documents data centric as well as document centric Finally depending on the database architecture they use Native XML Databases can be further divided into text based and model based databases Text based databases store XML documents as text thus storing an identical copy of the data which is put in This gives text based XML Databases a tremendous speed advantage when retri
22. gt addAttribute bg white color gt addAttribute fg black appSettings gt addChild color 5 Serialize the DOM tree using a parser object 104 APPENDIX B DATABASE API MANUAL Parser parser string xmlText parser serialize appSettings 6 Write the XML document to a file ofstream os configuration xml os lt lt xmlText 7 unref the root object appSettings gt unref 1 After including all necessary header files and defining the namespaces to use we create the root tag of the document This is accomplished by using a special constructor of the SoXML class which takes the name of the tag as parameter 2 Since the SoXML class is derived from SoGroup we also need to increment the reference counter in order to be able to access the object 3 Then we create the child tag language add a text object to its child list holding the value of the tag and add it to the child list of the appSettings tag 4 Next the color tag is created and the attributes storing the background and foreground color are added to it by calling the method addAttribute which takes the name and the value of the new attribute Having done this we have finished constructing the DOM model and are ready to convert it to text format 5 6 We instantiate a Parser object call the serialize method and pass the root object of the DOM model to it The result of the conversion is buffered in the
23. hierarchical tree structure of the scene graph can be used as a base for an XML DOM model This way Studierstube applications will be able to access XML data in the same way as graphical objects In order to implement this feature a new class which is derived from the Coin base class SoGroup can be defined as the following structure 34 CHAPTER 3 PROBLEM DESCRIPTION SoXML SoSFString name SoMFString attributes SoMFString values SoSFString content SoSF String and SoMFString are Coin classes for the representation of single strings S and string arrays M The tag name of an XML element is hold by name its text content is stored in the member variable content Attribute names are stored in attributes attribute values in values Nested elements are put into the child list of SoGroup Example Assume the following XML fragment lt ell attri A gt lt el2 attr2 1 1 el3 B gt lt el3 attr4 true gt lt el4 gt Hello lt el4 gt lt el2 gt lt el1 gt This fragment can be converted by an XSLT stylesheet into following SoXML representation XMLElement name eli attributes attri values A XMLElement name el2 attributes attr2 attr3 values 1 1 B XMLElement name el3 attributes attr4 values true XMLElement name el4 content hello 3 5 SPECIFICATION OF A TEST APPLICATION 35 In case the SoXML model is represented as text in m
24. is actually not an interface for an XML Database as one could suppose but a vendor specific implementation of a database driver These implementations are managed by the class DatabaseManager which is also the starting point for an application using the API DatabaseManager 30 CHAPTER 2 RELATED WORK provides access to particular XML Databases by providing instances of the class Collection In order to be modular and extensible non basic features are implemented as services by deriving new classes from the abstract base class Service So for instance the collection management features CollectionManagementService and transactions TransactionService have been implemented this way In order to get a version of this API which complies with the lowest level of the standard core level 0 one has to support the basic framework and thus has to implement most of the classes in figure 2 3 At the time of writing even vendors like Software AG the maker of the Tamino XML Database have not completely implemented this standard API for their database product Hopefully future versions will 2 35 XinCJ Xindice C API The XinCJ class library 32 is an attempt to implement a programming interface for the XML Database Xindice It is a C Java Native Interface wrapper for Xindice implementing a subset of the XML DB interface and thus allows to access the Xindice XML Database from C It tries to match the original Java class structure as cl
25. is not the latest version available which is version v4 1 4 at the time of writing Unfortunately we did not have a possibility to upgrade to the latest version and so we had to stick to the older one Therefore everything said here does not apply to the 37 38 CHAPTER 4 CHOOSING A DATABASE PRODUCT latest version which has undergone major improvements Most notably the added support for XQuery 12 and the XML DB API 31 4 2 1 Installation Installing Tamino was not as easy as one would expect The first attempt to run the installation program failed because of Tamino s licensing system The version we have does not install on Windows XP because it is restricted to be set up on Windows 9x or Windows 2000 Using Windows 2000 instead of XP helped At first the installation wizard asks to setup an HTTP server preferably Apache which is included on the CD This server is needed because some of the supplied applications in particular the main administration application Tamino Manager are implemented as web applications Done this and starting the installation program again the next thing we had to do was to update Internet Explorer and the Microsoft Java VM Using Internet Explorer as client side browser application one needs to have a more recent version of Java VM than the one that comes with Windows 2000 Already a bit exasperated by the length of the installation procedure we had to discover that Microsoft is not supporting the Ja
26. minimum 450 MHz e 256 MB RAM minimum e approximately 600 MB free disk space For an up to date list of prerequisites see the Tamino XML Starterkit web page www xmlstarterkit com D 1 2 Software prerequisites The following software packages and documents are needed for a complete installation of the Studierstube XML database e Windows 2000 Professional and Server or Windows XP Professional 117 118 APPENDIX D INSTALLATION AND CONFIGURATION GUIDE Java 2 SDK Version 1 4 1_05 Apache Web Server 2 0 43 included in the Tamino setup Apache Tomcat 4 1 29 mod_jk_1 2 5_2 0 47 dll Tamino Passthru Servlet Studierstube Passthru Servlet e Server side stylesheet D 1 3 Installation procedure Install a compatible Operating System The current version of Tamino XML server supports the following Windows versions e Windows 2000 Professional and Server e Windows XP Professional Hint With Windows 2000 you will need to update Internet Explorer to a recent version 6 0 Install Java Get Java 2 SDK Version 1 4 1_05 by downloading it from Sun s Java developer web page java sun com Install it to the default installation directory C j2sdk1 4 1_05 using the default settings Hint This SDK already includes the Java versions of Apache Xerces and Xalan therefore these components need not to be installed separately Install Apache Web server Install Apache Web Server 2 0 43 This web server version should be incl
27. network Staff members and students work on Desktop PCs running different operating systems like Microsoft Windows or Linux The TCP IP network which they are connected to enables them to share common resources like printers and file servers Additional to the local workstations the resources can be accessed from outside of the LAN through a firewall These are important details because we want our database to be accessible in the same ways Recent projects of the Institute deal with navigation systems and will use our database as primary storage medium In the context of their work the collaborators of these projects often have to leave the building and thus have to access the network from outside We have to consider that when we choose one of the database interfaces provided by Tamino The technology used in this network to access resources is the widely adopted Client Server architecture The Tamino database system is designed following the paradigms of this technology It provides a database server and several interfaces which can be used on the client side to access the database Thus it perfectly fits into the network of the Institute and we can benefit from utilizing the Client Server model and get enhanced flexibility centralized management of data lower total cost and so forth 5 3 Design issues Generally developers tend to concentrate on the functional features of their architectures and seldom address the ways in which the
28. of XPath and XSLT in one comprehensive query language Although we recommend to use XQuery because this will be the future standard for XML Databases it is up to you to choose the 89 90 APPENDIX B DATABASE API MANUAL Query result n Figure B 1 Collaboration diagram of database access classes Query execute starting point query Database Server Transformation result end point appropriate language for your database application Both alternatives have their own advantages and disadvantages If you for example already own a set of XSLT stylesheets matching your needs you will prefer to use the combination of XPath and XSLT Whereas it will be a good idea to choose XQuery if you start a new project from scratch The following sections assume that you are familiar with the basics of Tamino XQuery The documentation of Tamino 24 includes a comprehensive guide about it Read it first it will help you to understand this manual B 1 1 Getting started This section describes the Database access module which is part of the Studierstube XML Database API The Database access module is a small set of C classes providing access to a Tamino XML Database and for executing XSLT transformations It consists of the three classes Database Query and Transformer The class Database manages a connection to an XML Database server handling the transport of database queries and query results over a TCP IP network It a
29. often referred to as the SQL for XML One big advantage to XSLT is due to its non XML syntax its improved readability for humans and thus it is much easier to learn 2 1 XML 19 The XML Query language has more than one syntax In order to store XQuery statements in an XML document using a syntax that is compliant with the requirements of XML the W3C Consortium has specified another syntax called XQueryX Since there is a bijective relation between XQuery and XQueryX they can be exchanged for each other according to the current application Another advantage to the combination of XPath XSLT is the improved support for datatypes in XQuery The datatypes of XQuery are based on those of XML Schema see next section which provides a really extensive range Moreover it even allows to extend these in order to define and customize own types Hereby the introduction of XQuery also eliminated the biggest deficiency of XML query languages which was the poor support of datatypes Though there is a common sense that update queries are an important part of a query language due to time reasons they have not been included in the first version of XQuery So utilizing a W3C compliant implementation of XQuery we still have to use alternative ways to update documents like XUpdate Other vendors like Software AG the maker of Tamino have forgone to be compliant with the standard and released an extended version of XQuery which supports update operati
30. particular candidate Thus the first step of this work will be to install these databases on a computer in order to gain first experiences and to check out the features they support Thereby the focus of the investigation should be directed to programming interfaces and to administration applications which are available for each candidate Especially a tool is needed to maintain the data preferably one providing a graphical user interface Furthermore the databases should support an XML query language like XPath or XQuery and an update language like XUpdate Advanced database features like multi user access transactions and backup strategies are not that important but cannot be completely neglected since this work is part of a long running project In the case of a successful outcome it will be the base for several other works Results of this work step e Test installation of the database candidates on a test server e A report which is describing and comparing their features with focus on programming interfaces 3 3 Client interface In order to connect Studierstube applications to the database server a C library should be developed on top of the database API mainly for reading access In the following text this library is also referred as Studierstube XML Database API This library should be designed in a way which optimizes the usability for Studierstube applications A database query should then support the following fea
31. point of time E1 The client should be completely independent of the particular database used in order that the database server can be transparently replaced E2 A subsequent integration of transactions should be considered E3 User authentication should be considered E4 Support for binary data and Unicode should be considered 5 3 6 Scalability Availability and Security Scalability Availability and Security are important aspects of system design in many projects especially when they deal with databases However we do not consider these characteristics to be relevant to our project Anyhow the former two are mainly a matter of the database server and the latter is partially accomplished by our demand for authentication E3 5 3 7 Cost Since this is a non commercial project we do not have a big budget at our disposal that can be spent on resources Therefore we will have to rely on royalty free software preferably Open Source Software when we are in need of additional third party products 52 CHAPTER 5 DESIGN 5 4 Selecting an interface Preceding to the system design we have to select the interface we want to use at the client side Each interface needs additional components in order to be able to work properly Therefore the decision for a particular interface has a great impact on the structure of our system especially to that of the server Tamino provides several ways to access the database among them are i
32. section lists the most important variables used and intends to facilitate the understanding of the doGet method fixedStylesheet A member variable storing the path to the fixed stylesheet it is initialized by the method init xslstr A local string variable storing the path of the stylesheet contained in the xsisrc HTTP parameter xmlSource A string variable storing the intermediate results of the stylesheet processing xmlResult A string variable storing the final result of the stylesheet processing 6 2 SERVER 71 Compiling the servlet The command line script build cmd in the bin directory of the Passthru Servlet compiles the source code and creates the Java library passthru jar which can then be copied to Tomcat s servlet folder For a successful compilation we needed the following prerequisites Sun Java SDK 1 3 x The Passthru Servlet was developed by using version 1 3 x of the Java SDK Though it is not a problem to execute it under a later version it did fail to successfully compile using Java SDK 1 4 1 JSDK 2 0 The Java Servlet development kit JAXP 1 1 The Java API library for XML processing 6 2 3 Performance problems During the system test we stumbled over a strange problem regarding the performance of the Tamino database server While the query response time was usually within the range of a few seconds when using a certain test data set it abruptly became worse aft
33. selected object with the content of the xml file Whereas updating from an iv file replaces only the representation part of the current BAUML object C 1 7 Deleting objects The delete operation removes objects from the BAUML tree Press the button Delete to remove the currently selected object from the database Consider that it is not possible to delete top level objects from the database with BAUMLBrowser Use the Tamino Xplorer utility to do that C 1 8 Saving objects The save operation stores a BAUML object or its representation to a file Press the button Save to open a file dialog and choose a directory and a filename You can either save the currently selected object to an xml file in BAUML format or you can save only the representation part of the object as an Open Inventor scene graph to an iv file C 2 CORE COMPONENT 109 C 1 9 Object intersection test The intersection operation tests whether the representation parts of two BAUML objects share a common region in space In order to be able to use this function you need to select two objects from the tree view at first Selecting two objects is done by clicking the first object with the left mouse button and clicking on a second object while holding the Ctrl key This enables the Do intersect button at the bottom of the graphics viewer in figure C 1 Press the button to start the test The application then presents the result of the test in a m
34. server Click on it and go to Tamino Databases At the bottom left side of the Explorer window you will find a button labeled Create database Click on it and provide a name for your database Push the button Finish to create the database In order to be able to access a database from a client application the database has to be started first When selecting your database from the database list you will find a button labeled Startup database A click on it will start the server process which manages your database Hint You can also automate the startup process at server boot time by editing the server properties of your database and setting the parameter autostart to yes D 2 2 Creating a collection Since XML Databases are organized in form of collections you will need to create a collection too Start the Tamino X plorer application and login to your server by clicking on the server name with the right mouse button and select connect from the local menu Provide an administrator s user name and password The first time D 3 CONFIGURING THE FIXED STYLESHEET 121 logging in you will also need to provide the name of your database Right click at your database in the tree window and select New Collection Provide a name for your collection and push create D 2 3 Providing a database scheme In order to work with your database collection you will also need to define a schema of
35. sized documents originates from the storage and indexing technique used by Xindice 36 Documents are stored as a whole into the database and each time a part of the document needs to be read or written it must be completely loaded into the memory Furthermore indexes do not store positional information of tags in documents but are solely used to find candidate documents for queries and updates 4 3 4 APTVs Xindice supports the XML DB API for Java and a language independent XML RPC API which takes the place of the CORBA API of version 1 0 Server side XSLT processing is currently not available 4 4 EXIST 43 4 3 5 Documentation There is a documentation which is dedicated to the latest version of Xindice and reflects the changes made to the database While it can be said to be accurate currently it lacks of a comprehensive installation guide and a description of the XML RPC API which is needed when working with programming languages other than Java 4 4 eXist eXist 35 38 is another Open Source project founded by Wolfgang Meier Like Xindice eXist is completely implemented in Java 4 4 1 Installation Installing eXist is simply accomplished by copying and unpacking the supplied zip or war file to a proper location of a hard disk and by editing an environment variable The eXist database server supports three modes of operation Standalone servlet and embedded mode each of which has its own advantages and disadvantages
36. the XML documents it should contain You do this by right clicking your database in Tamino X plorer and selecting Define Schema As a starting point we have provided a simple schema here which defines an element named MyElement which is able to hold arbitrary content You will need to rename the string MyElement according to the name of your XML root tag lt xml version 1 0 encoding UTF 8 gt lt xs schema xmlns xs http www w3 org 2001 XMLSchema xmlns tsd http namespaces softwareag com tamino TaminoSchemaDefinition gt lt xs annotation gt lt xs appinfo gt lt tsd schemaInfo name MyElement gt lt tsd collection name MyCollection gt lt tsd doctype name MyElement gt lt tsd logical gt lt tsd content gt closed lt tsd content gt lt tsd logical gt lt tsd doctype gt lt tsd collection gt lt tsd schemalnfo gt lt xs appinfo gt lt xs annotation gt lt xs element name MyElement gt lt xs schema gt D 2 4 Inserting documents Since root level documents can not be inserted using the Studierstube XML Database APT you need Tamino X plorer to accomplish this After login right click on your database in the tree window and select Insert instance from the local menu This will open up a dialog which let s you point to an XML document for insertion D 3 Configuring the fixed stylesheet The modified version of the Tamino Passthru servlet th
37. to set up the directory where the stylesheet files reside Normally this is the directory path of the BAUMLBrowser executeable Consider that you have to specify an absolute path here so that the browser component is able to locate the stylesheets in the case the current working directory is changed during program execution 3 Next you should verify that all stylesheet files exist which are needed by the browser This is done by calling the method verifyStylesheet In the case a stylesheet file is missing an error message is returned by the function which can be reported to the user 4 Finally the fourth and last step loads application options from an XML configuration file See the section about configuration for further details on application parameters C 2 3 Establishing a connection After successful initialization a connection to the database must be established before you can start browsing it At this point you need to know the name of the database and collection you want to edit Use the open method to accomplish this BAUMLBrowser browser browser open MyDatabase MyCollection Keep in mind that the current implementation of this function does not check for the existence of the database or collection Thus take care of supplying correct names to it 112 APPENDIX C BAUMLBROWSER MANUAL C 2 4 Reading nodes from the database This section shows how to gain information about BAUML objects
38. tvInsert 1 Each tree view item is associated with a Node object stored in the ItemData field The first line of the code example above gets this object by calling the GetItemData method of the tree control This is Windows specific code but other operating systems will have similar methods to store user data in a tree control C 2 CORE COMPONENT 113 2 The next step reads out all child items of this node from the database into a variable of type NodeVector which is a container for dynamically created Node objects 3 5 The resulting list of child nodes is then processed and for each child node a new item is added to the tree view Properties like the name of the node 4 and its parenthood 5 determine the appearance and the state of the new tree view item Using this function a tree control can be filled step by step with node information starting at the root node by passing NULL as the first parameter to getNodeChildren up to the leave nodes of a BAUML document The structure of the tree view items then reflects the parent child relationship of these BAUML objects C 2 5 Getting object data While the previous section showed how to gain structural information from the database this section will teach you how to get data about specific objects The following code example is taken from the BAUMLBrowser application It is part of a function which displays the representation part of a BAUML objec
39. written in Java Although there is a JAXP and JDOM implementation this version of Tamino does not implement the de facto standard API for Java the XML DB API Support for C is given in form of an Active X control which communicates to the server over HTTP Server side XSLT processing can be implemented using X Tension the interface for enhancing server functionality 4 2 5 Documentation If there is one thing outstanding in Tamino then it is the documentation and help system There are about 3000 pages describing Tamino s features and applications including pictures graphics and many practical examples and about 1500 additional pages documenting SQL and ODBC features Although the HTML search function 40 CHAPTER 4 CHOOSING A DATABASE PRODUCT is not as comfortable as the one within the native Windows help system it is still useful and helps to cut through this vast amount of information 4 3 Xindice Xindice 34 formerly known as dbXML is part of the Apache XML project It is an Open Source XML database developed by volunteers under the Apache license In spite of the fact that the first official version version 1 0 of Xindice was released more than a year ago it is currently undergoing major changes As a result of this modifications there are actual two different versions of Xindice now each of it providing different features Although the newer version v1 1 is still in beta state it is the one discussed here because it r
40. 002 01 09 xmldb_api html January 09 2002 32 Hauke von Bremen XinCJ Xinidice C Database APT http www codexperts com download html 33 Software AG Websites for the XML Database Tamino http www softwareag com tamino http www xmilstarterkit com 34 Apache group Website for the XML Database Xindice http xml apache org xindice 35 Wolfgang Meier Website for the XML Database eXist http exist db org 36 James Bates Kevin O Neill Xindice internals http xml apache org xindice dev guide internals html 37 Hendrik Seffler XML Datenbank Exist http www informatik hu berlin de seffler 2 ausarbeitung pdf HU Berlin 38 Wolfgang Meier eXist An Open Source Native XML Database http exist db org webdb pdf Darmstadt University of Technology
41. 1 webapps This directory contains the Tamino Passthru Servlet and the Studierstube Passthru Servlet Configure Tomcat Copy the files from the Tomcat configuration from Studierstube Subversion server directory to C Program Files Apache Group Tomcat 4 1 conf and adjust the path settings as necessary Add the server side stylesheet Copy the contents of the Apache htdocs stylesheets from Studierstube Subversion server directory to C Program Files Apache Group Apache2 htdocs stylesheets 120 APPENDIX D INSTALLATION AND CONFIGURATION GUIDE Configure Apache Webserver Add the line Include C Program Files Apache Group Tomcat4 1 conf apache connector conf to the Apache configuration file at C Program Files Apache Group Apache2 conf httpd conf D 2 Creating databases and collections This section is a quick guide to database and collection management It is meant solely as a starting point for working with Tamino databases For a comprehensive description of this topic read the documentation that comes with Tamino D 2 1 Creating a database Probably the first thing you want to do after installation is to set up a database This is accomplished by using the web application Tamino Manager After login by providing an administrator s user name and password you see a list of managed hosts at the top left of the Explorer window Normally this list will contain only one item the name of your database
42. 49 633 Usability and Acceptability U eve sr ew ser 50 5 3 4 Performance S ee ee te Boe oe BS et 50 5 3 9 Extensibility ass 2 da ne a a a ee 51 5 3 6 Scalability Availability and Security 51 De OS o ee ne bd ut Re aos Diner 51 5 4 Selecting an interface ra Da OS ee SS ee ee 52 SAL Tamno APT ORC orara a ee ea WES Ge a 52 5 4 2 HTTP Client API for ActiveX 53 CONTENTS 5 4 3 Native HTTP Client API 2 inks ee ea 5AA O o ent eis 5 5 System architecture ee er ee A n gt toe Ge ep ahh 55 9 XMEDBRPL a ay eS gate hn is ne EEE wer ha a all 9 9 2 SOSSE MOCEL ss a ne a a See a Se SG ee Es 55 Client alas Ma Ra eh ok IA es a ds Del o o A a a er ee A 5 5 5 Workflow of a query rn a Mae nr 5 6 API Classes A A a e e ee Sieh 50b y TER ONE r ISI ee Ae A 5 6 2 String as query result type creciera ss Se Conclusion AA E ale a ben ud en a 6 Implementation De Mole us a RE O a cies bob a ie 6 1 1 Database access classes 2 2 2 2 2 a a 6 1 2 Transformation Classes u 22 2 2 2 a ar a 6 1 3 SoXxXML DOM Model classes vr as VE as re 6 1 4 Optimizing the API 2 de E nie DB VAS o o A 5 ee A A a a AIM 6 2 1 Server components ker ie A 6 2 2 Studierstube Passthru Servlet 6 2 3 Performance problems 26 2 ia 24 SB A e E 7 Sample application TI Introduch n gs A Stk a A A AAA 52 BAUME Lansmage al ls A A 7 3 BAUMLBrowser Application o o a fol TRES er
43. Database TableName AnsiString RecordCount integer FieldCount integer Fields TField Legend Figure 2 1 Collaboration diagram of VCL database classes Due to the fact that there have not been many XML Database APT s released until now only two we know of we will also have a look at two SQL Database APT s Basically they follow the same approach as their XML counterpart They provide classes to query databases and have to handle the resulting data in some way In principle they only differ in the syntax of query scripts and in the type of query results Therefore analyzing the design of this interfaces makes sense t00 2 3 2 Borland VCL Borland VCL 29 the Visual Component Library is a class library originating from Borland Delphi Object Pascal which has been made available for C This library includes a set of classes which are designed for database access mainly for Relational Databases which support SQL The collaboration diagram in figure 2 1 on page 27 shows the most important classes of this library which are involved in a database query TDatabase represents a physical database It supports operations to connect Open and disconnect Close to a particular database and methods for transactions StartTransaction Commit Rollback A database query is represented by the class TQuery which holds a SQL script and additional connection parameters When calling the Open method of TQuery the SQ
44. In order to seamlessly integrate the XML database technology into Studierstube applications we have designed and implemented a programming interface which provides access to the database using XML query languages like XPath and XQuery Finally the implementation of the system was tested by means of a sample application which processes XML documents stored in the database CONTENTS T Kurzfassung Der Einsatz von XML dem standardisierten Dokument und Kommunikationsformat erfordert ein geeignetes Organisationssystem wenn eine gro e Anzahl von Dokumenten verwaltet werden muss Datenbanken eignen sich im allgemeinen am besten um diese Aufgabe zu bewerkstelligen Im Kontext von XML ist dies am effizientesten durch den Einsatz einer XML Datenbank zu erreichen die direkte Unterst tzung f r diese Technologie bietet Diese Diplomarbeit implementiert ein derartiges Datenbanksystem f r die Interactive Media System Group der Technischen Universit t Wien Ausgehend von einer Auswahl von drei Kandidat Datenbanken haben wir diejenige ausgesucht die am besten den Anforderungen des Instituts entspricht Gem Client Server Architektur wurde von uns ein zentraler Datenbankserver installiert der als gemeinsam genutzter Speicher dient Die Client Anwendungen die auf diesen Server zugreifen sind Teil des Augmented Reality Systems Studierstube dem aktuellen Forschungsprojekt der Gruppe Um die XML Datenbank Technologie in Studierstube Anwendungen
45. L query is executed against a database which is passed in the member variable Database In consequence T Database runs the query and stores the result in an instance of T Table The query result can then 28 CHAPTER 2 RELATED WORK CDatabase CRecordSet Open SQL char Close GetRecordCount int CommitTrans GetField alue name char Rollback value CDBYVariant amp Figure 2 2 Collaboration diagram of MFC ODBC database classes be accessed in the Fields member variable of TTable Though there are more classes which can be utilized for other tasks like managing sessions or running batch jobs the previous paragraph describes the basic scenario for querying databases in VCL This quite simple but efficient and easy to use approach is worth to be considered in our design 2 3 3 Microsoft MFC MFC the Microsoft Foundation Classes 28 is a C class library for developing MS Windows based applications It supports database operations by providing different sets of classes for different types of databases Figure 2 2 on page 28 depicts a collaboration diagram of the classes for ODBC database access The approach used by the classes in figure 2 2 is quite similar to that of Borland s VCL though with different class names There is a CDatabase class which supports operations for connection and transaction and a class CRecordSet for storing the result set of database queries The main differe
46. PI for XML 19 uses an event oriented strategy to pass the content of an XML document to an application When the parser reads in the document it generates events for each XML entity found in the document An application can then process these events by deriving a new class from a special interface class and overwriting its virtual callback methods This way an application can selectively process certain document entities and leave out the remaining parts Compared to DOM this method needs much less memory but as disadvantage it requires to store and manage context information if needed Other techniques Additionally to DOM and SAX there are further techniques to deal with XML documents These will not be discussed here For the sake of completeness we have compiled the following list See the literature references in order to read more about alternative ways to DOM and SAX e XML Data Binding Bindings of XML data to programming language structures 23 JAXB Java Architecture for XML Binding 25 JAXP The Java API for XML Processing 20 JDOM The Java Document Object Model 21 e DOM4J Document Object Model for Java 22 2 2 XML DATABASES 23 2 2 XML Databases 2 2 1 Introduction The conventional way to manage XML documents is to store them as files in a directory structure on a computer drive This approach follows a natural way of keeping data as documents since XML is a document format and documents are us
47. Technische Universit t Wien Diplomarbeit XML Databases for Augmented Reality ausgef hrt am Institut f r Softwaretechnik und Interaktive Systeme der Technischen Universit t Wien unter Anleitung von Ao Univ Prof DI Dr Dieter Schmalstieg und DI Dr Gerhard Reitmayr als verantwortlich mitwirkenden Assistenten durch Werner Frieb Treustra e 57 1200 Wien Wien 02 Oktober 2004 Contents 1 Introduction 2 Related work 2 2 2 3 RME N A AR ER ER aa a aa 2 12 Tntrod ctioi aie a rer rer ea ee ed 2 1 2 The building blocks Of XML 2 3 2 2224 4 ee EOE Dh 2 1 3 Using namespaces re ar a a ae eh 2 1 4 Selecting data with XPath sr 2er ar Dr 2 1 5 Applying transformations using XSLT 2 1 6 Querying documents using XQuery 2 1 7 Defining languages using XML Schema 2 1 8 Working with XML documents se oe ee Es ee Re XML Databases O er IN A pe et os 2 2 1 ntroduction a octet ets e a os e o ds a 2 2 2 Data centric vs Document centric XML data 2 2 3 Relational Databases 2 8122 Sarah 2 2 4 XML enabled databases 2 2 222 a 008 2 2 5 Native XML Databases 0 cr a sat Database interfaces ie adi a Wera Mein Ba ae Sas ie 2 3 1 Introduction Ta ea ee er re 232 Borland VOL a gors Bete o O tm Ge Se Be go ty Ae nk ie a A ts MEC Ze eet ge ne RE ent 234 AMO AP ers noes ee rare Pass 2 3 5 XinCJ Xindice C APL bao a 3 Problem Description 3 1 3 2 3 3
48. a C 2 6 Updating objects Sehr a seiten rn C 2 7 Special functions 22 a 24 as D Installation and Configuration guide D 1 D 2 Installati n s 6 2 46 20 0 2 OS Bee D D 1 1 Hardware prerequisites D 1 2 Software prerequisites D 1 3 Installation procedure Creating databases and collections D 2 1 Creating a database not rate D 2 2 Creating a collection se at ae rest D 2 3 Providing a database scheme CONTENTS CONTENTS D 2 4 Inserting documents D 3 Configuring the fixed stylesheet 6 CONTENTS Abstract The application of XML an upcoming storage and communication format requires an appropriate organisational system if a certain quantity of document entities is exceeded Generally this can be best accomplished by employing a database system which is targeted towards this technology namely an XML Database This master thesis implements such a system for the Interactive Media Systems Group at Vienna University of Technology Starting from the choice of three candidate databases we have selected the one which is best complying with the requirements of the Institute According to the Client Server architecture we have installed and set up a central database server as a common storage system The client applications accessing this database are part of Studierstube an Augmented Reality system which is the current research project of the group
49. ain memory it can be easily converted to a corresponding scene graph Results of this work step e Specification of SoXML Eventually namespaces should be also regarded in the design Therefore the definition of XMLElement must be extended as in the following example SoXML SoSFString name SoSFString namespace SoMFString attributes SoMFString values SoMFString attributeNs SoSFString content Alternatively the namespace could be stored as a prefix in the string variables name and attribute like ns name But probably this might not comfortable enough for the users of the library because in this way the string needs to be parsed first 3 5 Specification of a test application An application should be designed to test the implemented components with real and large data This application has to use SoXML for data representation in order to demonstrate the connection to Coin Results of this work step e Specification of a test application 3 6 Implementation The last step of this work is to implement and test all of previously described features Results of this work step e Implementation of the Studierstube XML Database API e Implementation of the SoXML component e Implementation of the test sample application 36 CHAPTER 3 PROBLEM DESCRIPTION e Documentation of the components and applications in form of manuals and web pages Chapter 4 Choosing a database product 4 1 Introduction The
50. an A A ee A es 7 3 2 Core component oaoa a aa e due add e 13 3 a AA ae or ge okt x e SE 7 3 4 Implementation gt ea a ae 8 Summary A Database Comparison Charts B Database API manual B 1 Database access module 00000 a BALL Getting st rt ed ser du dra phase a is G 93 54 54 59 59 56 97 58 59 59 60 60 61 61 61 64 64 65 66 67 69 71 73 73 73 75 75 76 76 76 79 81 B 2 B 1 2 Sample database 4 ev ara ara ea B 1 3 Retrieving objects from the database B 1 4 Transforming query results 2 2 2 2 B 1 5 Advanced topics ooo aa men SoXML DOM Model module B 2 1 Getting started separe alee A B 2 2 Parsing an XML document B 2 3 Constructing an XML document C BAUMLBrowser manual C 1 C 2 User Interface Guide 2454408514484 A C 1 1 Tree control A eh Oe es Gh wend 6 ca a C 1 2 Graphics window 2 2 4 0 Gee 4 ido C 1 3 Setting application options C 1 4 Connecting toadatabase C 1 5 Inserting new objects C 1 6 Updating objects 5 cack ala C 1 7 Deleting objects 32 2234 2 22 3 8 244 C 1 8 Saving A ae C 1 9 Object intersection test 2222020 C 1 10 Has point operation Core Component alero raise aa ride HB ASICNCONCE DEY y A EP al ne Ss C 2 2 Class initialization ei C 2 3 Establishing a connection C 2 4 Reading nodes from the database 0 2 5 Getting object dat
51. ause selects the nodes which should be updated and allows to define variables and the do clause specifies how the data should be changed Read the Tamino documentation 24 about XQuery for further details 98 APPENDIX B DATABASE API MANUAL B 1 5 Advanced topics This section deals with topics which did not fit into one of the previous sections while still being important enough to be mentioned when working with Tamino in the real world Limitations of Tamino XQuery Though Tamino XQuery enhances the W3C standard with update operations it lacks some of the features mentioned The following list mentions the most important drawbacks and limitations when working with Tamino e XQuery does not support the unabbreviated syntax of location paths using named axes For example it is not possible to use the ancestor axis in XQuery This should be changed in the next release of Tamino e Not all of the functions specified by the W3C consortium have been implemented yet Asa workaround they can be implemented using Tamino server extensions e It is not possible to insert or delete a document of a collection using XQuery Which means that you have to use an external tool like Tamino XPlorer to insert and delete top level nodes The document identifier ino id Normally a query is working on all documents contained in the current collection Though sometimes it is necessary to restrict the scope of a query to one specific document Si
52. bjects embedded in a type hierarchy and each type defines a different aspect of the BAUML language The most basic types are listed in the following text ObjectType is the basic object type It defines an annotation tag and an id attribute All other object types are directly or indirectly derived from it SpatialObjectType is derived from ObjectType It is the basic type for all spatial objects and defines a pose tag and a representation tag SpatialContainerType is derived from SpatialObject Type It is the basic type for all spatial objects containing child objects It defines a children tag The geometric information mentioned above vertices and surfaces is stored in the representation tag of the SpatialObjectType All spatial objects must be derived from this type and they must utilize this tag in order to specify their representation The same counts for the children tag of the SpatialContainerType which holds a list of child objects and thus describes the parent object in more detail These are all abstract types forming the base of the BAUML language Abstract means that there must not be instances of these types but instead they are used to derive further types from them In addition to these abstract types BAUML also defines concrete elements like SpatialObject SpatialContainer Room Building ARToolkitMarker and Waypoint which are all derived from the above types Each of these element
53. borators an appropriate storage system i e a database is needed The research data which needs to be managed by the database system varies strongly in its structure from flat relational tables to deeply structured recursive trees Since XML is best suitable to handle a wide spectrum of document formats the decision was made to store future data as XML documents In order to share these documents a central XML Database should then serve as a common storage system Essentially the task is now to select an appropriate database product to install the server application on a computer of the Institute as well as to enable the connection of Studierstube applications to the server 31 32 CHAPTER 3 PROBLEM DESCRIPTION 3 2 XML Database server Since the implementation of a fully featured database server is out of the scope of this work we need to employ one of the products which are available on the market Currently the number of alternatives is still a bit limited Thus the choice of candidates can be narrowed down to the following three databases e Tamino 33 a commercial product of Software AG e Xindice 34 an Open Source project of the Apache group and e eXist 35 another Open Source database founded by Wolfgang Meier At the time of writing XML Databases are a relatively new technology They cannot be expected to be fully matured yet Since that we need more information about them to facilitate our decision for a
54. cations and thus save time and money The following sections give an overview of the language features and related technologies like XSLT XPath and XML Schema Since a comprehensive description of XML is far out of the scope of this work we will only discuss the basic features and refer to literature when needed 2 1 2 The building blocks of XML XML is a standardized meta language or also meta document format which is purely based on Unicode text It is used to describe elements and structures of documents The W3C XML standard defines a set of rules which specifies the building blocks and syntax of a well formed XML document As the term markup language already reveals the most characteristic entities of XML are marks which are called tags A tag is a text which is enclosed by angle brackets lt gt Tags are used to label and structure the content of an XML document When defining a new document format which is based on XML one has to specify his own set of tags according to the type of data he wants to handle The W3C standard only specifies the syntax rules for these tags but does not say anything about their meaning This is left to the creator of this new document format Additional to these tags the W3C standard specifies further entities which can be used to form a document The following sections give an overview of these entities which an XML document can or respectively has to be composed of XML Declaration
55. ch was defined in the previous section this section shows how to open the database and how to run a simple query on it For the ease of reading assume that all operations of the following code example succeed so that we do not have to add any error handling code to it Anyway there is a special section about error handling later in this manual 1 include header files tinclude lt stbxml Database h gt include lt stbxml Query h gt include lt string gt 2 use namespaces using namespace std using namespace stbxml 3 Define and open the database Database database database open 192 168 0 1 stuff 4 Setup a Query object Query query query language Query XQUERY query query input meetings meeting title 5 Run the query on the database string response database execute query meetings response 6 Print the result printf s response c_str 7 Close the database database close 1 First of all you need to include the correct header files The ones related to the database API reside in the folder stbxml which should be in the search path of your compiler This example uses the classes Database Query and the standard C class string to store the response of the database server B 1 DATABASE ACCESS MODULE 93 2 All classes of the database API are defined in the namespace stbxml in order to avoid name conflicts It is a good idea to specify the namespace i
56. d Transport Services which is part of the Tamino setup The Pros and cons of this interface are Portable between platforms P1 5 4 SELECTING AN INTERFACE 53 Good performance due to direct access to the X Machine S1 Supports transactions and user authentication E2 E3 Supports binary data and Unicode E4 Does not support server side XSLT processing Needs a local installation of eXtended Transport Services U2 5 4 2 HTTP Client API for ActiveX The HTTP Client API for ActiveX consists of two controls which communicate with the Tamino XML Server at HTTP protocol level One is for accessing and manipulating XML documents and the other for non XML documents including binary data It is a more convenient way of communicating with the server than using the native HTTP protocol Since it uses ActiveX technology it is only available for the Microsoft Windows operating systems The API offers a number of properties and methods to manipulate data in Tamino via document names perform queries local and distributed transactions and for lock management The API s methods return a DOM object as a result or require a DOM object as input The DOM model supported by the ActiveX API is the Microsoft DOM In order to use the API from within an application you need the Microsoft Foundation Classes library and Microsoft Internet Explorer version 5 x or 6 x to be installed on the system The Pros and cons of this interface are
57. d by using an ampersand amp The objects returned by this query are enclosed in the HTTP response body inside an XML document By default the document consists of elements and attributes defined in the Tamino namespace http namespaces softwareag com tamino response2 using the prefix ino See the section Database access module of the developer manual in order to get an explanation of how this document is mapped to our own namespace Limitations of the Apache web server Some web servers restrict the length of the URLs that they can process In the case of the Apache web server this limit is around 8KB for the total sum of all query parameters Especially when using the insert and update features of XQuery where whole XML objects have to be passed to the X Machine this limit can be exceeded with ease Since we plan to process big sized objects which are magnitudes bigger than the limit we had to overcome this problem by utilizing the HTTP POST method instead of the GET method This way we were able to shift the query scripts which are stored in the parameter _xquery to the body of the HTTP request and so reduce the size of the URL The remaining parameters are still transmitted in the search part of the URL The servlet path The Database class provides a static member variable called servletpath This string variable holds the path which points to the location of the Passthru servlet By default i
58. e Studierstube Passthru Servlet offers a feature to apply a server side stylesheet to each query result It is 122 APPENDIX D INSTALLATION AND CONFIGURATION GUIDE meant to translate Tamino namespaces to namespaces which are independent from the manufacturer The name of the stylesheet used by the servlet can be configured by editing the web xml configuration file which is located in the webapps stbxml WEB INF directory of Tomcat The servlet tag of this XML file contains a list of initialization parameters which are passed to the servlet at startup time Assuming that the stylesheet is located in the web directory stylesheets the parameter which points to the fixed stylesheet would look like the following example lt init param gt lt param name gt fixedStylesheet lt param name gt lt param value gt http localhost stylesheets stbresult xsl lt param value gt lt init param gt The tag param name is the name of the parameter which we have added and must be set to fixedStylesheet The tag param value contains an URL pointing to the stylesheet In the case of an absolute URL this URL is used to retrieve the stylesheet In the case of a relative URL for example stylehsheet stbresult xsl the stylesheet is loaded directly from the Tamino database where the first part of the path refers to the collection containing the stylesheet In this example the name of the collection would be
59. e XML data where parse means to convert XML text to a SoXML DOM Model and serialize stands for generating text from a DOM tree Parser uses the SAX interface of the C Xerces parser in order to convert an XML document to a DOM model whereas the opposite way is accomplished by utilizing the tree traversing mechanisms of Coin When parsing an XML document the parser class creates a tree which is composed of instances of the classes SoX ML and SoXMLText As mentioned before the model supports only a limited set of XML elements or to be more precise it supports XML tags attributes and text data Any other elements like XML comments and processing instructions though they are allowed to be present are discarded during the parsing process An XML tag is represented by the class SoXML which also stores the attributes of the tag and a list of its child elements This child list can contain an arbitrary number of SoXML and SoXMLText objects in any order Due to the fact that an XML document owns at least one tag which is called root or document node and that the SoXML DOM model does not provide any document class the root node of a SoXML tree is always an instance of the class SoXML Whereas the third class of this module SoXMLText which holds the text fragments of an XML document is always a leaf node Figure B 2 on page 101 illustrates the collaboration of aforementioned classes Following the solid arrows the XML document on
60. e decided to employ Tamino a commercial database system of Software AG Now we will elaborate a system concept that meets our needs and integrates the features offered by Tamino Starting from the specification we first analyze the future operating environment of our project Then we work out a number of common design issues like portability and reusability in order to ensure and improve the quality of our work The topics outlined here serve as a basis for the next chapters where we select an appropriate programming interface for the client and develop a model of the planned system architecture Then we split up the model into two parts namely server and client Each of these parts is then discussed and refined separately Finally we evaluate the end result and try to point out the pros and cons of our work 5 2 Environment analysis A system can hardly ever be built completely on its own Most of the time you will have to consider the environment in which the system has to operate Just like in our case where the operating environment is given by the local area network of the Institute Hence we need to have a look at this network in order to properly 47 48 CHAPTER 5 DESIGN Modem AB K Notebook Windows Desktop PC Linux Desktop PC Firewall Printer Print server Figure 5 1 Local area network of the Institute integrate the database system Figure 5 1 on page 48 illustrates the basic structure of the
61. e needed storage technology The following sections discuss the different techniques with regard to the structure of the documents they can handle 2 2 3 Relational Databases This type of XML Database uses a table based mapping to store particular XML documents in a Relational Database It models XML documents as a single table or as a set of tables The model is directly derived from the schema of the document complex elements are mapped to tables and simple elements and attributes are mapped to columns Since there is a direct structural relationship between the database tables and the XML document it cannot be used for any XML documents that do not match the format The table based mapping is most suitable to handle data centric documents when transferring data between two relational databases 2 2 4 XML enabled databases XML enabled databases use an object relational mapping Therefore XML documents are modeled as a tree of objects which are specific to the data in the documents The model is then mapped to Relational Databases using object relational mapping techniques or SQL 3 object views What means that classes are mapped to tables scalar properties are mapped to columns and object valued properties are mapped to primary key foreign key values This technique can handle data centric and document centric XML documents though it does not support mixed content well 2 2 5 Native XML Databases Native XML Databases are designed
62. e using a Web browser All in all the impression of the delivered applications is quite good with pleasing user interfaces and useful features None of them ever crashed or had troubles connecting to the database during the test 4 2 3 Database features Since this is an older version of Tamino it lacks of support for a query language capable of doing updates As a consequence partial updates of documents in the database are not available It is only possible to update documents as a whole More mature are the features for querying the database W3C s XPath query language has been implemented in an extended form and in order to cause confusion is called X Query Note the dash between the X and the Q On the one hand X Query enhances XPath with additional functions and operators and offers the use of variables as well but on the other hand there are some W3C functions left out which means that it is not fully W3C compliant Tamino also allows to define a schema for data structures Using W3C s schema definition language as basis Tamino s schema language covers in addition to the standardized features the definition of database indexes and the use of data types for elements and attributes The feature to define data types is unique among the tested candidates and is especially useful when working with a lot of data centric information 4 2 4 API s As usual within the XML domain most of the programming language API s are
63. each of them can be compiled and used independently In the following we discuss each of these groups separately 6 1 1 Database access classes The database access classes provide the database query features of the client API As outlined in the design chapter we use the HTTP protocol to access the Tamino database Thus we need an appropriate portable software package which enables us to communicate through a TCP IP network Since the Studierstube system uses the ACE package to implement networking functions and ACE is a well designed highly portable and approved C network library we chose to use ACE as well However the drawback of utilizing ACE is although it supplies the basic networking components for building client applications it does not provide direct support for the HTTP protocol Therefore we had to implement the protocol by using the ACE 61 62 CHAPTER 6 IMPLEMENTATION mean Aggregation lt Z gt Association _ Figure 6 1 UML class diagram of the Database access classes socket classes as depicted in figure 6 1 on page 62 The HTTPConnection class encapsulates the implementation of the HTTP protocol in a way that it is easily reusable by other applications and classes Hence the class Database can utilize an instance of HTTPConnection to transmit the query parameters as corresponding HTTP requests The next section shows how this translation is accomplished Accessing Tamino using HTTP One of t
64. easy to learn and easy to handle API U2 An easy to use setup procedure for the client 5 3 4 Performance S The tests of the Xindice server revealed see previous chapter that we can not simply act on the assumption that the database system delivers an acceptable performance Depending on the structure and size of the retrieved data the response time of a query can vary heavily Additional to the time needed by the database server to execute a query and deliver it to the client some time will be required to accomplish several post processing steps on the client An overall response time for a typical 8 3 DESIGN ISSUES ol query beyond a few seconds will frustrate users and could even demolish our work completely Therefore we demand that S1 The response time of a typical query including post processing should be kept within the range of a few seconds at maximum 5 seconds Since we can hardly impact the performance of the server apart from adjusting its configuration performance tuning will mainly be an issue of optimizing the source code of the client 5 3 5 Extensibility E This section discusses topics regarding possible future extensions to the database system This mainly covers typical database related features which are standard in most database products but not explicitly required by our specification Nevertheless we want to consider these features in our design that an easy integration is possible at a later
65. eate your own XML language Originating from SGML the Standard Generalized Markup Language XML was standardized in its first version by the World Wide Web Consortium in 1998 Since then it has achieved a great acceptance in the worldwide computer industry and is supported by many companies including the big ones like Microsoft Sun and IBM XML was originally intended as a universal data format in order to facilitate the exchange of information between applications But due to its popularity it has entered many different fields of the information technology 6 Until today a number of languages have been defined which are based on XML Among them there are standardized languages like XHTML for World Wide Web applications WML for WAP phones MATHML for mathematical expressions and XMLRPC for interprocess communication In the course of the introduction of XML Databases XML has even begun to be a partial replacement for Relational Databases Employing XML for storing data has many advantages over the usage of a proprietary binary data format XML is platform independent readable by humans supports localization and is based on an international standard Moreover the common syntax of XML languages enables its users to share a wide range of tools applications and related technologies like editors parsers and XML processors This way projects 11 12 CHAPTER 2 RELATED WORK utilizing XML can benefit from a big common pool of software appli
66. element 2 1 XML 15 Processing instruction The processing instruction PI is a convenient way to include additional information for applications which process the XML document A PI starts with lt followed by the name of the application It can hold any number of attributes and is ended by the string gt The following example shows a PI which defines two debugging attributes lt myapp debug yes output print gt Comment The comment tag is used to include annotations in an XML document which are not part of the document content Everything between the strings lt and gt is treated as comment The following example shows an annotation containing the name of the author lt Created by Tom Jones gt 2 1 3 Using namespaces When working with XML it happens that one wants to use the same name for two different XML tags For example think of the tag address which could be used to hold a persons home address as well as the network address of a computer Using the same name for different purposes would lead to a name conflict because we would not be able to distinguish between these tags This name conflict can be resolved by the use of namespaces which can be specified in any start tag The following example shows how aforementioned address tags can be used side by side in the same XML document lt xml version 1 0 encoding UTF 8 gt lt form xmlns person http
67. ements in a certain order to restrict the number of allowed occurrences of an element to define default values and to declare an element as optional just to name the most important Structures can be built by nesting elements and declaring arrays and lists of elements One of the biggest advantages that comes along with the introduction of XML Schema is the improved support for datatypes in XML As mentioned in the section about XQuery XML Schema provides an extensive range of datatypes we can choose from Among them are numeric types like byte int and float string types like string token ID date and time types and types for storing binary data likes hexBinary or base64Binary This enriches the capabilities of defining XML languages a lot and at the same time reduces the chances of error prone type conversions which would be necessary otherwise Furthermore by utilizing this type system one is able to exactly specify the content of a valid XML document Another powerful feature of XML Schema is the possibility to declare user defined types and elements In the manner of object oriented languages XML Schema allows to derive new types and elements from existing ones Depending on the kind of inheritance used which can be restriction or extension the derived types support a reduced or extended range of allowed elements and values Several predefined types 2 1 XML 21 of XML Schema are defined by utilizing this inheritance scheme thu
68. ent C 1 1 Tree control Figure C 1 on page 106 shows a screenshot of the BAUMLBrowser application dialog On the left side of the dialog you see a tree control displaying the structure of the BAUML objects which are stored in the database collection Each node in the tree control matches a BAUML object in the database and each child node of a tree node corresponds to a child object of a BAUML object Other element tags like 105 106 APPENDIX C BAUMLBROWSER MANUAL de BAUML XML Database Browser Options About Database stb Collection bauml Disconnect Bauml Collection Room HE0446 SpatialContainer Vienna SpatialContainer csbuilding Building csbuilding Room YARD1 Room FLOORO Room HED446 Room HE0435 Room HG0401 Room HEO406A Room HE0401 Room HF0405 Room HFO405x Room HE04194 Room HD0407 Room HDO405 Room HD0403 Room LIFT3 Room HDO401B a RotX RotY Hi Insert Update Delete Save Has point x jo Figure C 1 BAUMLBrowser user interface representation and pose tags which can also be part of a BAUML object are not shown by the browser The name of such a tree node is composed of the object type name i e the name of the XML element that represents the object and the value of its id attribute if available The plus sign you can see on the left side of some of the t
69. ents Until now we have seen how XML documents are structured and discussed related technologies to query transform and describe XML data This section deals with the question how one can work with XML data within programming languages In the course of the introduction of XML a number of tools and libraries have been created to facilitate the processing of XML data within software applications each of them embarking on a different strategy All of these libraries have one thing in common They serve as an interface between an application and XML data Meanwhile two of these interfaces have evolved into a standard namely DOM and SAX 22 CHAPTER 2 RELATED WORK DOM DOM the Document Object Model standardized by the W3C Consortium 18 defines a set of classes and methods to access the elements of an XML document When using DOM to process an XML document a parser reads in the XML text and creates a tree of objects which represents the elements contained in the document The advantage of this method is that the document is read in as a whole thus enabling an application to randomly access and manipulate all of the elements Whereas the disadvantage is given by high memory requirements and the delay resulting from the parsing process which an application has to wait for until it can access the document content SAX Alternatively to the DOM Model there exists another de facto standard for accessing XML data SAX the Simple A
70. epresents the future of Xindice Differences to the older version are mentioned when it s relevant 4 3 1 Installation Due to the very short installation documentation that comes with Xindice it took us several hours to find out how to get a working system with a running server get the command line tools working and the XMLdbGUI graphical client application connecting properly In order to share my experience Pll try to give a short installation guide including some tips on installing Xindice On the download page 34 you can see that Xindice 1 1 is delivered in form of three zipped files The two files named xml xindice xxx war zip and xml xindice xxx jar zip contain the binary distribution and xml xindice xxx src zip contains the source distribution The source distribution is needed when you want to contribute to the project or when you want to use the included Jetty servlet engine We use the binary distribution instead Because support for the standalone server version has been dropped in the current release the database server of Xindice is delivered as a servlet and hence needs a servlet engine like Apache Tomcat to be installed first Installation of Tomcat is straightforward just download the setup program and follow the instructions Next copy the war file which is included in xml xindice xxx war zip into the webapps folder of Tomcat and restart the Tomcat application It will then unpack the
71. er Parser amp SoXML M1 We have developed a fairly simple solution for the problem U1 which can be installed simply by adding the modules to the Studierstube setup U2 Due to direct access to the X Machine we can expect a good performance S1 The usage of HTTP facilitates the exchange of the underlying database El Transactions can be supported by adding appropriate methods to the Database class E2 User Authentication has been considered E3 Cons Support for binary data and Unicode was dropped for simplicity E4 Chapter 6 Implementation In this chapter we present the implementation of our project We give an in depth explanation of the system components and show how they work together In this regard we investigate the classes and applications involved in the implementation and discuss topics concerning the performance of the system Furthermore at this point we replace all remaining abstract components of the design with appropriate real software packages As in the previous chapter we discuss client and server separately 6 1 Client The client side of the system consists of a C class library and several third party software packages The classes of this library can be divided into three logical groups The database access classes the transformation classes and the classes for generating the SoXML DOM Model The standard C class string serves as an interface between these groups Thus
72. er new XML objects were added to this set Depending on the size of these objects the response time was even extended up to two minutes Luckily with the help of a service member of Software AG we were able to solve this problem The internal cache size of the XQuery processor is set to a very low value by default and thus creates a bottleneck in the query engine We were advised to insert the string parameter XQuery_document_cache size type REG_SZ into the Windows registry to the key HKEY_LOCAL_MACHINE SOFTWARE SoftwareAG Tamino servers lt database name gt and set its value to 100 Megabytes After that the original performance of the server was restored Even if we doubled the size of the test data then it did not have a noteworthy impact on the query speed 72 CHAPTER 6 IMPLEMENTATION Chapter 7 Sample application 7 1 Introduction Having completed the implementation of the client API and the modifications on the server we wanted to test the software by means of a small sample application In order to obtain something more useful than a simple demonstration application we searched for a proper task in the scope of the Studierstube project We wanted to run the software in a real world environment and at the same time achieve an application which can be utilized later too At last the SignPost subproject offered what we were looking for SignPost is a Studierstube application that is able to guide a person th
73. esentation part Inserting an object to a node means to append it to the child list of the corresponding object in the database while update and delete operations process the object itself Keep in mind that data manipulation functions like insert and delete change the relative position of all sibling objects following the processed object Thus the position information of these siblings which is stored in Node objects becomes invalid BAUMLBrowser corrects this simply by rereading the entire child list of the parent object This has proven to be a very straightforward and reliable way to overcome this effect Furthermore a restriction you should know of is that you cannot insert and delete top level objects usingthe BAUMLBrowser class This is caused by limitations of the Tamino XQuery language which is not able to act at this level Instead use Tamino Xplorer as a workaround for this task The following example reads aBAUML object from a file on disk and adds it to the child list of an object in the database 1 Browser and node object BAUMLBrowser browser select an existing node BAUMLBrowser Node node 2 Path of the file to be inserted string filePath NewBAUMLObject xml 3 read the file into a string string s readFile filePath 4 append the object to the child list of lt node gt browser insertObject node s C 2 CORE COMPONENT 115 1 Analogue to the previous example assume that
74. essage box showing either the string YES when an intersection was found or NO otherwise Keep in mind that this test is a bit limited due to the implementation of the SolntersectionDetectionAction class of Coin Which means that an intersection is only found when the planes of the surface of two objects do intersect So for example if an object is completely contained in the other one the intersection test will show a negative result C 1 10 Has point operation The Has Point operation tests whether a certain three dimensional point is contained in the representation part of a BAUML object On the bottom right of the application dialog you find three edit controls named X Y and Z Here you enter the Cartesian world coordinates of your point to be tested This function accepts floating point values if necessary Start the test by pressing the Has point button on the left of the edit controls The application then presents the test result in a message box showing either the string YES when the point is inside the object or NO otherwise C 2 Core Component This section describes the BAUMLBrowser C class the core component of the BAUMLBrowser application This class encapsulates the parts of the application which are independent of the Windows operating system and the Microsoft MFC class library Thus it facilitates porting the Windows application to another operating sys
75. et of user defined parameters which can be used to configure the servlet In the case of the Studierstube Servlet one of these parameters points to the fixed server side stylesheet stbresult xsl 6 2 2 Studierstube Passthru Servlet The Studierstube Passthru Servlet also referred to as Studierstube Servlet is a modification of the Tamino Passthru Servlet It enhances the original Tamino servlet by adding support for the execution of an additional fixed stylesheet This stylesheet is executed prior to the user defined stylesheet which can be specified by client applications using the stylesheet member variable of the Query class Since there is no common standard for result sets of XML Databases so far each database manufacturer is defining its own format The purpose of the enhancement to the Tamino Passthru Servlet is to transform the query results of the Tamino database server into a form which is specified by us Without this transformation the code of the client applications would highly depend on Tamino and thus make it very difficult to replace the database server at a later point of time As mentioned in the previous section the stylesheet defining the fixed transformation can be configured 70 CHAPTER 6 IMPLEMENTATION in the web xml script of the servlet Read the developer manual of the Client API for more details about the usage of this servlet Modifying the servlet Due to a lack of source code documentat
76. ether in order to query the database Therefore we discuss the scenario depicted in figure 5 2 The numbers on the arrows in this diagram indicate the order of the performed steps In the following we will refer to these numbers 1 If we want to query the database we have to use an instance of the class Query and pass the query script and the path of the server side stylesheet to it 2 Then we execute the query on a Database object which creates a corresponding HTTP request 3 The Database object sends the request over the network to the web server by utilizing the HTTPConnection component 4 The server forwards it to the servlet engine and thus to the Tamino Passthru Servlet 5 8 Then the servlet analyses the HTTP request and extracts the part which contains the path of the stylesheet The remaining HTTP request is then sent back to the web server not depicted here which executes the query on the database 5a When the database server returns the servlet applies the stylesheet to the query result 5b and delivers the processed data back to the Database object 6 8 9 12 The Database object passes the query result to the application 9 once the HTTP response is validated and thus completes the database query The query result can now be post processed using the classes Transformer 11 and Parser 12 5 6 API CLASSES HTTPConnection Parser open hostname string port intege
77. eving entire documents or document fragments compared to the solutions mentioned Though they also can encounter performance problems when retrieving data which is different from the predefined hierarchy Model based databases build an internal object model of the XML document and store this model Some databases store the model in Relational Databases others in object oriented databases or they store the model as persistent DOM like the recent version of eXist They have the same performance characteristics like text based databases Whether they are faster or slower than text based systems is not clear Future applications will show which approach is the better one 2 3 Database interfaces 2 3 1 Introduction In order to benefit from the long running experience of approved database interfaces this section investigates the design of several programming interfaces for databases All of these interfaces are implemented using an object oriented language preferably C because our implementation language is C too Thereby the focus of our attention is directed mainly to class collaboration and derivation We will analyze how these class libraries work and try to pick out the approaches which fit our needs best 2 3 DATABASE INTERFACES 27 TDatabase DatabaseName AnsiString Database TDatabase SQL AnsiString Params AnsiString StartTransaction Committ Rollback Open TTable Database T
78. first step of this work was to evaluate and select an appropriate XML Database for this project Since the data to be processed concerns differently structured XML documents the choice was quickly narrowed down to one of the following three Native XML Databases An older version of Tamino 33 a commercial product developed by Software AG Xindice 34 an Open Source XML database by the Apache Group and eXist 35 an Open Source XML database founded by Wolfgang Meier at the Technical University Darmstadt Germany There are a plenty of other commercial alternatives which did not come into consideration just because they were not available to us and there are further Open Source solutions which simply did not fit our requirements We were looking for a Native XML Database system supporting XPath 9 and XUpdate which comes with a set of usable and working administration applications preferably with support for sever side XSLT processing and a C API for easy integration into our system This chapter tries to evaluate and compare aforementioned systems showing the strengths and weaknesses of each in order to have a basis to select one for our task Additionally appendix A of this work contains several charts which list and compare the features we were interested in 4 2 Tamino Tamino 33 is a commercial Native Database system developed by Software AG First of all it has to be mentioned that the version used for this evaluation v3 1 2
79. g updating deleting and renaming parts of XML documents All operations work on the node level which means you are able to update all parts of a document which can be addressed by an XPath expression even at attribute level All update operations begin with the keyword update The following example utilizes the update functions of XQuery to insert an additional person to the second meeting record of our sample collection meetings For this reason we substitute the query definition 4 of our source example with a query that performs the update 4 Insert a new person Query query query language Query XQUERY query query update for a in input meetings meeting where a number 2 do insert lt person gt lt firstname gt David lt firstname gt lt lastname gt Johnson lt lastname gt lt person gt into a people As a result the document will contain the additional element and the database server will respond with a document which indicates whether the operation was successful or not Apart from into there are two other keywords that you can use when inserting element nodes Using preceding the element nodes will be inserted as preceding siblings to the update node Using following the element nodes will be inserted as siblings following the update node The delete update and rename operations are specified using a similar language construct The For Let Where cl
80. gt lt lastname gt Bauer lt lastname gt lt assuranceid gt 1234 56678 lt assuranceid gt lt street gt 123 Main street lt street gt lt city gt Vienna lt city gt lt country gt Austria lt country gt lt patient gt lt patient gt lt firstname gt Fred lt firstname gt lt lastname gt Mayer lt lastname gt lt assuranceid gt 432 83474 lt assuranceid gt lt street gt 124 Haydn street lt street gt lt city gt Eisenstadt lt city gt lt country gt Austria lt country gt lt patient gt lt patients gt Document centric documents The second category document centric documents are characterized by a less regular or irregular structure with larger grained data and a lots of mixed content These are documents which are usually designed for human consumption and typically do not fit well into a relational database scheme For example to stay within the hospital domain it could be a diagnostic description of a doctor like the following 2 2 XML DATABASES 25 lt xml version 1 0 encoding UTF 8 gt lt diagnosis gt The patient lt firstname gt Otmar lt firstname gt lt lastname gt Bauer lt lastname gt was admitted to the lt ward gt surgery lt ward gt because of a lt admission_diagnose gt broken arm lt admission_diagnose gt lt diagnosis gt Though both categories of XML documents data centric and document centric follow the XML specification there is a big difference between them regarding th
81. he central parts of the Tamino XML Database is the XML engine also referred to as X Machine This engine provides an HTTP based interface which offers various operations to access the XML data store The X Machine interface allows to send special commands to the engine either encoded in URLs when using the HTTP GET method or as multipart form data when using the HTTP POST method If the GET method is used the commands are provided in the URL as keyword value pairs in the search part of the URL The part that is separated by a question mark This way the X Machine is accessible through a wide range of operations Those utilized by the Database class are described in the following table 6 1 CLIENT 63 Command Meaning _xql Retrieves one or more objects using the XPath query language _xquery Specifies a query using the XQuery query language _xslsre Executes a server side stylesheet additionally supported by the Tamino Passthru Servlet Assume that we want to get a list of all surnames of our staff members and their personal data is contained in the database addresses and the collection staff Using the query language XPath the corresponding HTTP request would look like this http hostname tamino addresses staff _xql person surname Where hostname stands for the IP address of the database server and person for the XML root tag of the data Multiple keyword value pairs can be connecte
82. he schema of this part will be the one which you defined in your query and is more or less manufacturer independent You will need to extract the result set from the database response in order to be able to further process the information The easiest way to achieve this is by using the class Transformer which has not been discussed so far It is shown in form of an example in the next section B 1 4 Transforming query results In addition to the query operations which were introduced in the previous section the Database access module also offers ways to process information using XSLT stylesheets There are two possibilities which we will discuss separately You can choose between client side stylesheets and server side stylesheets or even use both alternatives when needed While the server uses the Java version of the Xalan transformer the client transformation is implemented using the C version Due to the fact that these are independent projects the output of both alternatives sometimes shows minor differences when running the same stylesheet Keep that in mind when you decide to change the side of transformation at a later point You will have to adjust your stylesheet accordingly But don t worry this can be done with little effort B 1 DATABASE ACCESS MODULE 95 Client side transformation Client side transformation is supported by the class Transformer which is part of the Database access module Transformer is a wrapper class
83. ias Kramer who encouraged me to proceed with my work in the most difficult phase Chapter 1 Introduction XML was seen as a miracle drug for the software industry when it was introduced and standardized by the W3C consortium in 1998 Rumors and wild speculations were spread It was said that this technology would revolutionize and completely change computer technology especially the Internet Now a few years later these thoughts have turned out to be a hype The revolution did not take place and COBOL based computer systems are still used by several banks Instead XML is going to be established as a standard in a slower but quite steady way The advantages of a common readable data format are increasingly outweighing the fears of management staff to fail at employing a new technology The idea of describing data with the help of a meta language is by no means new XML is originating from SGML the Standard Generalized Markup Language which was conceived in the 1960s 1970s and standardized by the ISO organization in 1986 So one can ask why did it take so long for a good idea to be employed by a bigger community SGML is compared to XML much more customizable and thus more expensive to implement Furthermore at that time computer memory was much more expensive than it is today It was seen as a waste of resources and money to use several bytes of computer memory in order to store a single byte of information as it is common with XML
84. ing the API you can query and update XML documents in a database transform query results using XSLT and build a DOM Model for accessing the data The API is logically divided into two modules which can be used independently of each other e Database access module Provides classes for database access and data transformation e SoXML DOM Model module Provides classes to build a DOM model based on Open inventor nodes B 1 Database access module Tamino XML Server is a data management system for storing XML documents It helps with finding and managing XML data and to search effectively the information A single Tamino server is able to run several databases which can be accessed by name and is able to supply many clients A Tamino database is organized in form of collections A collection is the largest unit of information within a database and can contain multiple XML documents You can think of it as a folder in a file system with the difference that it stores XML documents instead of files In order to search for information within a collection Tamino supports two query languages XQuery and XPath Since XQuery is the standard query language in Tamino and a superset of XPath the following sections will focus on it Beside the ability to retrieve objects or parts of an object Tamino XQuery is capable of inserting updating and deleting objects and also to compose the query result using constructors It more or less combines the features
85. ion it was quite difficult to modify the servlet and to integrate the extension In order to facilitate further enhancements of the servlet this section gives an overview of the modifications made The complete Java source code of the servlet can be found in the Java directory com softwareag tamino passthru Included in this directory is a class called TaminoFilter which is derived from the class HttpServlet This is the only class which needed to be modified In order to preserve the original source code the first step was to copy it to a new class called StudierstubeFilter Appropriate comments are given for all modifications being made At startup time of the servlet the method init reads several parameters from the configuration file web xml We added an additional parameter called fixedStylesheet to it which contains the URL of the fixed stylesheet The processing of the stylesheets is implemented in the doGet method which is called upon each HTTP GET POST request We had to restructure this method in order to add the extension and moved a portion of its code to a new method cacheFile which caches the compiled version of stylesheets Prior to the user defined stylesheet which is stored in the HTTP parameter _xslsrc we apply the fixed stylesheet and write the transformation result back to the variable xmlSource which is then used by the remaining part as input SOUTCE The following
86. ir architectures support quality concerns such as portability modularity and usability 1 We will try to overcome this deficiency by defining a set of design criteria according to the requirements of our project 8 3 DESIGN ISSUES 49 The functional aspects of the system have already been outlined in the specification In this chapter we will work out the characteristics of a solution to our problem Neglecting the constraining facts of the real working environment at first and thinking about an ideal system helps to get the most out of our work and allows to obtain a broader view of the problem domain Moreover by clarifying things that remain informal otherwise we get some kind of checklist that serves as a basis for design decisions The considerations made here will influence the overall system design and have a major impact on which components we chose In the following sections we investigate software characteristics like portability modularity and reusability and evaluate whether they are important to us or not For those considered to be important we will outline what has to be done to achieve them This also includes characteristics which are regarded as negligible at the moment but could possibly turn out to be essential for future versions of the project 5 3 1 Portability P As today s PC systems become more and more heterogeneous designing portable software becomes an important issue Studierstube is available for seve
87. laceable M1 By employing the Client Server architecture for our project we are able to split up the system into two major blocks namely client and server which are interconnected by an interface based on the HTTP protocol Hence we can treat them as separate units and design them almost independently Each of these blocks can then be further divided into several components We identify these components by analyzing the requirements of the specification and by applying the paradigms of object oriented design most notably encapsulation and 56 CHAPTER 5 DESIGN Client l Seryer Studierstube Application XSLT Processor Tamino XML i Database 12 Create H A 1 xPath an an Apply seed DOM madel E 5a Query 6 Return result Sb Transform F Servlet Engine Transformer Parser Tamino Passthru Servlet 7 Return result 4 Pass to servlet 2 Execute 9 Return result h i h Database Web Server m mE 8 HTTP se se sasae se i Response HTTPConnection 3 HTTP Request Legend Reqgust e Planned Etting TS Response Seems Component Component Figure 5 2 Dataflow diagram of the system architecture inheritance Using object oriented terminology we can also speak of objects instead of components Figure 5 2 on page 56 illustrates the system architecture by means of a dataflow diagram It shows a typical scenario of the system where a Studierstube application queries the
88. ld be quite time and memory consuming Instead only the information needed at the moment should be requested As a tradeoff between reading all at once and getting the information node by node BAUMLBrowser provides a method to get information about all child elements of a BAUML object This way it is possible to traverse the BAUML tree beginning at the root node down to the leaf nodes while keeping time and memory requirements within a reasonable limit Moreover this is exactly what you need to successively fill a tree control C 2 2 Class initialization Before you are able to connect to a database several initialization steps have to be performed The following code example showcases the correct calling sequence of the methods used to setup a browser instance BAUMLBrowser browser 1 init the browser class browser init 2 set the Stylesheet directory to the current working directory browser setStylesheetDirectory GetCurrentDirectory 3 verify that all stylesheet files exist string msg if browser verifyStylesheets msg print msg c_str exit 1 4 load application options BAUMLBrowser Options options C 2 CORE COMPONENT 111 if options read browser setOptions options 1 First of all the BAUMLBrowser class needs to be initialized by calling the init method This initializes the third party libraries used by the browser like Parser and Transformer 2 The next step is
89. lize this new technology in the context of Augmented Reality Though there are still features missing like full XML Schema support and insert delete and update of root level documents the installed database product proved to be a useful storage medium for a certain spectrum of XML documents Furthermore we are expecting that one of the next releases of Tamino will also remove those last deficiencies On top of the database we have implemented an API for Studierstube applications the Studierstube XML Database API which enables to seamlessly integrate XML Database functionality Thus Augmented Reality applications can be realized which work with big data sets much bigger than it is possible with single XML documents on a file basis So for instance it is possible to store and retrieve detailed comprehensive graphical information which originates from urban geographical information systems Another conceivable application is to employ the database for storing the state of distributed Studierstube applications and utilize it to exchange information between them The potentials of applications are virtually unlimited The API can be easily extended to support advanced database features like transactions and sub collections In order to fully benefit from employing an XML Database system we recommend to implement these features in a future project 79 80 CHAPTER 8 SUMMARY Appendix A Database Comparison Charts 81 General
90. llo XML Schema Tutorial http www xfront com xml schema html 2001 18 W3Consortium Document Object Model DOM http www w3c org DOM 19 Official website for SAX http www saxproject org 20 Sun Microsystems Java API for XML Processing http java sun com xml jaxp 21 Open Source Project JDOM Java Document Object Model http www jdom org 22 Open Source Project DOM4J Document Object Model for Java http dom4j org 23 Ronald Bourret XML Data Binding Resources http www rpbourret com xml XMLDataBinding htm 2001 2004 24 Software AG Tamino XML Database Documentation Tamino Version 4 1 4 1 2004 25 Ed Ort Bhakti Mehta Java Architecture for XML Binding JAXB http java sun com de March 2003 26 INCITS H2 and ISO IEC JTC1 SC32 WG3 standards groups SQLX SQL amp XML Working Together http www sqlx org 27 Ronald Bourret XML and Databases http www rpbourret com xml XMLAndDatabases htm July 2003 28 Microsoft Corp MFC Reference Library ODBC classes http msdn microsoft com library en us velib html _mfc_odbc_classes asp 29 Borland C Builder 6 User manual 30 XML DB Initiative Application Programming Interface for XML Databases http xmldb org sourceforge net xapi 2000 2003 LIST OF FIGURES 127 31 Kimbro Staken An Introduction to the XML DB API http www xml com pub a 2
91. look very much like directory paths when working with a computer file system But XPath path expressions support several relationships which are called axis For example there is a child parent sibling or ancestor axis XPath also defines a library of standard functions for working with strings numbers and Boolean expressions These functions can be called in the conditional part of an XPath statement in order to specify further properties of the selected nodes Assume a library which stores scientific articles in form of XML documents and that the following XML example represents a section of this library lt library gt lt article title XML and Databases author Ronald Bourret gt This paper gives a high level overview of how to use XML with databases It describes how lt article gt lt article title eXist An Open Source Native XML Database author Wolfgang Meier gt eXist is an Open Source effort to develop a native XML database system which lt article gt lt library gt Assume further that we want to select all articles written by Ronald Bourret from the library Using XPath the statement doing this would look like the following line library article author Ronald Bourret 2 1 XML 17 The first part of the statement the path expression library article selects all articles of the library The subsequent conditional part which is always enclosed in brackets then reduces the selected set to
92. lso provides a method to execute queries in form of instances of the class Query which holds a query script and other query related parameters Query results can then be further processed with XSLT stylesheets using the Transformer class See figure B 1 on page 90 for an illustration of how these classes collaborate The discussed classes are marked by a bold frame The following sections will take you step by step through some typical basic operations you will probably want to do They will teach you how to use the classes to connect to a database server query and update XML documents how to B 1 DATABASE ACCESS MODULE 91 interpret database responses and how to process query results with XSLT stylesheets Furthermore a section that discusses advanced database topics is included It is recommended to read this guide from the beginning to the end so that you will have a complete overview after you have finished reading B 1 2 Sample database The following sections rely on code examples which work on a database containing information about staff meetings Assume that we have created an XML Database stuff and a collection meetings An XML document of this collection stores information about our staff members meetings Further assume that the database already contains the following document lt meetings gt lt meeting number 1 gt lt title gt XML for Augmented Reality lt title gt lt date gt 1 3 2004 lt date gt lt pe
93. ma The XML specification only gives a general description of the building blocks of a well formed XML document But it does not include a method to define rules for a specific user defined XML language Therefore the W3C Consortium standardized a special language which accomplishes this namely XML Schema XML Schema which is itself an XML language provides ways to specify the valid elements and structures of an XML document The fact that XML Schema is formally expressed in XML allows applications to process schemes in the same way as any XML document This enables them to share a common code base and to save a lot of work and time Actually many of the XML parsers support the use of XML Schema thus providing a way to automatically check whether a document follows certain rules or not The specification of the sophisticated mechanisms of XML schema 15 is quite extensive and a comprehensive explanation of them fills whole books Thus we can only give an overview of the most important features and refer to further literature for interested readers As usual in this chapter we will also give a short example which showcases the syntax of this language Since the basic building blocks of XML are tags and attributes XML Schema provides means to specify which of these entities are allowed in a certain context In addition to defining datatypes for element content and attribute values XML Schema includes mechanisms to define sequences of el
94. n language to English and the tag color contains the attributes bg and fg which sets the background color to white and the foreground color to black Assume that this script is stored in a file named configuration xml The following code example parses this file and prints the contained settings to the screen 1 Include header files include lt stbxml Parser h gt include lt stbxml SoXML h gt include lt stbxml SoXMLText h gt 102 APPENDIX B DATABASE API MANUAL 2 Use the DB API namespace using namespace stbxml 3 Parse the configuration file Parser parser SoXML appSettings parser parseFile configuration xml 4 Get child tags SoXML language appSettings gt getChildByName language SoXML color appSettings gt getChildByName color 5 Print the content of the tags printf language s n language gt getText getString printf background s n color gt getAttributeByName bg getString printf foreground s n color gt getAttributeByName fg getString 1 The database API header files are located in the folder stbxml which should be included in the search path of your compiler First of all we need to include the header files of the SoXML module 2 In order to be able to use the SoXML classes without specifying any namespace prefix we declare the namespace stbxml 3 Then we use the class Parser to parse the configuration file a
95. n order to work with simpler class names 3 These lines declare a Database object and open a connection to the database stuff on an XML server with the following IP address 192 168 0 1 By the way the open method also supports the usage of a host name instead of an IP address Keep in mind that the current implementation of the open method does not check for the existence of the database or collection Thus take care of supplying correct names to it 4 Since the Query class supports two query languages namely XQuery and XPath you have to specify which language should be used when defining the query Here we use the language XQuery to specify a query command which retrieves a list of all meeting titles contained in the database 5 6 Then we run the query on the database collection meetings by calling the method execute of the database object The response of the database server is stored in the string variable response and printed 6 to standard output 7 The final step closes the connection to the database Next we will analyze the answer of the database server which should contain the result set of the query As you probably expect the database answers in form of an XML document When all operations have been successfully executed the output of the sample program should look like the following text lt xml version 1 0 encoding UTF 8 gt lt stbxml response xmlns s
96. nce there is no standard way to achieve this each manufacturer has implemented it s own solution In Tamino each document gets a unique identifier namely ino id which can be used to refer to this document in a query The following XQuery example shows how to retrieve the document with the ino id 100 Note that you have to specify the namespace for Tamino functions because the function getInold which delivers the document ID of any node is specific to Tamino declare namespace tf http namespaces softwareag com tamino TaminoFunction for a in input where tf getInoId a 100 return a If the the where clause was omitted in this query it would change the result set to all documents of the collection instead of the document with ino id 100 B 2 SOXML DOM MODEL MODULE 99 Nested collections Although this is a de facto standard of XML Databases Tamino does not yet provide any support for nested collections It means that you are not able to divide a collection into further sub collections If you are in need of utilizing sub collections for your project you will have to map the tree structure of your design to the flat collection structure of Tamino by using an appropriate naming scheme Error handling When working with the Database access module you are utilizing a number of different components each of which can fail to operate successfully The error messages produced by these c
97. nce to the VCL approach is that there is no separate class for queries SQL queries are executed by calling the Open method of CRecordSet which thereby covers both tasks querying and storing result sets As long as there is only a single query parameter to keep track of here in form of an SQL string this approach seems to be sufficient Whereas the VCL solution seems to be better when a query supports several parameters because in this way they can be reused more easily in another context 2 3 4 XML DB API The XML DB API 30 is the standard programming interface for XML Databases It is targeted towards all object oriented programming languages but has only been implemented for Java so far The class library is intended to be vendor neutral and 2 3 DATABASE INTERFACES 29 Database I getName Service Configurable A getProperty setProperty a getName setCollection A CollectionManagementServic getCollection registers DatabaseManager Collection A getCollection getName createCollection registerDatabase getParentCollection removeCollecton getChildCollection getService mE E _ _v ee gr pm gg hasMoreResources r getSize query XPath string nextResouce getResouce getiterator XUpdateQueryService returns update Commands string Resource Legend
98. nd create a SoXML DOM model by calling the parseFile method of the class which takes a file path as input parameter Other input sources like streams and URLs are also supported by this class After successfully parsing the tree s root node gets stored in the variable appSettings In the case of an error the parse method would return NULL to indicate that something was going wrong You can then call the getLastError method of the class to obtain further details about the error Let us assume that the operation succeeds though 4 Starting from the root node we access the child tags of the document by calling the method getChildByName which is provided by the class SoXML This is much more convenient than accessing the tags by using the member functions of the base class SoGroup Knowing that Whitespace characters like carriage return and space also generate SoXMLText objects which are stored in the child list we cannot access the child tags simply by specifying a constant index This means that the first object in the child list of the appSettings element in the example above will be an SoXMLText object containing An and not the child tag language as one would probably expect Thus utilizing the getChildByName method really makes sense B 2 SOXML DOM MODEL MODULE 103 5 Finally the last part of the example prints the content of the tags to the screen The language token is ext
99. nd vice versa Since the Xerces C parser is already part of the Studierstube project we employed it to implement the translation In order to feed the parser with an XML document which is contained in a Standard C stream we had to implement an appropriate class providing streams as input source We accomplished this by deriving the classes StreamInputStream and StreamInputSource from the Xerces base classes BinInputStream and InputSource as depicted in figure 6 3 on page 65 6 1 CLIENT 65 Xerces C InputSource BininputStream DefaultHandler A N i Sax2Handler StreaminputStream SoXML Text 4 1 creates 1 L StreaminputSource returns 1 Inheritance gt Association ae a USOS Figure 6 3 UML class diagram of the SOXML DOM Model classes In order to be able to use the SAX interface of Xerces we had to derive a new class from Defaulthandler namely Sax2Handler The Sax2Handler class implements the actual translation process by overwriting the virtual methods of it s base class The output of this translation process is a tree which is composed of instances of the classes SoXML and SoXMLText In order to access this tree with the scene graph operations of Coin2 we derived these classes from the Coin base class SoNode and SoGroup Read the section SoXML DOM Model module of the developer manual in order to get additional details about these classes 6 1 4 Optimi
100. nd xml The fourth line defines a template element the pattern matching part of the stylesheet which selects all articles from the source document by passing the XPath expression library article in the attribute match Within the template element we extract the title and author attributes from the current XML fragment and add a comma between them The template element is called for each article of the source document and adds the selected attributes to the output list The result of applying the stylesheet to the source document is then a list separated by comma like the following XML and Databases Ronald Bourret eXist An Open Source Native XML Database Wolfgang Meier In order to learn more about the XSLT language we recommend to read the XSLT tutorial by W3Schools 11 2 1 6 Querying documents using XQuery In addition to the languages XPath and XSLT the W3C Consortium is currently working on a draft of a new XML query language called XQuery Though some details may still be a subject of change the specification of this language already seems to have reached its final phase at the time of writing The current version has already been implemented in a number of projects including the XML Databases Tamino and eXist XQuery is a combination of the features of XPath and XSLT and to some extent can be seen as a replacement for these two XQuery syntax has some similarities to the syntax of SQL 7 and is therefore
101. ndice Xindice Webadmin a graphical web application in servlet mode only Y Xindice Webadmin a graphical web application in servlet mode only Y YAB is an Explorer like browser for Xindice Java XDataFinder vo 6 is an application for browsing and querying XML files and native XML databases Java Y XinCon a web amp webdav administration interface XIndice Browser vo 85 a tool to browse a local Xindice 1 0 database Java 4 eXist eXist Client Shell an integrated and combined Command Line and Gui Java Client based on XML DB API eXist Client Shell an integrated and combined graphical and command line Java application interface a simple administrative web interface XMLdbGUI v1 3 1 is an application that allows the user to browse and modify databases conforming to the XML DB API specification Java Y Xindice command line client Export import collections to from files in a directory hierarchy a eXist Client Shell using the upload and backup feature Pros amp Cons Tamino XIndice eXist Good and extensive Supports standard API s Supports standard API s documentation Additional embedded mode Good administration Good administration applications applications Automated index generation Support for schema s and data Additional embedded mode types Sophisticated database engine No support for XUpdate it s not Does not support querying No manual indexes
102. necessary Furthermore the user is able to adjust the nesting level of the index which means that solely element tags down to a distinct hierarchy level are covered by the index This indexing strategy seems to be a good tradeoff between user comfort and index size Hereby the database user does not have to think too much about indexes but query speed and storage costs can be kept within an acceptable range In contrast to Xindice eXist uses a B Tree data model to organize XML documents Additionally document fragments are stored as persistent DOM 37 Due to this eXist can be expected to handle big sized documents better than Xindice 4 4 4 API s eXist supports a number of different API s depending on the server mode used First of all is the XML DB API the de facto standard Java API for XML Databases The feature to query the database from a web browser is provided by the HTTP API This API also supports server side XSLT processing Furthermore there are two language independent APIs namely XML RPC and SOAP The second one is implemented as a Java servlet and therefore only available when running eXist within Tomcat or any other servlet container 4 4 5 Documentation Though eXist has the smallest documentation set out of the three databases it is still useable and informative 4 5 Conclusion Taking into account that the latest version of Tamino has undergone major improvements and hopefully also got a less complicated ins
103. nterface Apart from the mentioned properties there are further arguments to choose this interface First the HTTP protocol is designed for transporting text data Since XML documents are completely made of plain text lines this is quite suitable for that Second most other vendors also provide an HTTP interface for their XML Database Thus migrating to a different database if needed is much easier than it would be when we would use the proprietary API for C Third server side XSLT processing is only supported by the Tamino Passthru Servlet which implies the use of the native interface This way we can save work and must not develop a server process which implements the XSLT processing Fourth the usage of the HTTP protocol facilitates the routing of database queries over firewalls and proxies 5 5 System architecture This section discusses the architecture of the system and the components it is composed of Before proceeding to the description of the system model a few words have to be said about the standard XML Database API 5 5 SYSTEM ARCHITECTURE 59 5 5 1 XML DB API Since the XML DB Initiative already defined an open standard for accessing XML Databases namely the XML DB API one could argue that this is the right way to implement the client interface Although we considered to follow this standard at first we decided to use our own approach for the following reasons e Due to its architecture the XML DB API is q
104. nterfaces for Java Java Script and Microsoft NET C These three options are out of question to us because we have to use the implementation language of Studierstube namely C When reducing the available possibilities on the basis of programming languages the following interfaces are left e Tamino API for C e HTTP Client API for ActiveX e Native HTTP Client API The following sections investigate these interfaces in order to have a basis for our decision If appropriate we refer to the previously outlined design characteristics by their code 5 4 1 Tamino API for C The Tamino API for C allows client applications to access a Tamino XML Server in a direct way without going through a web server like most other Tamino APls Since this API is entirely written in the programming language C it is very well suited for client applications written in C or C The Tamino API for C is available for Windows as well as for UNIX platforms It provides roughly the same functionality as the X Machine programming interface The X Machine interface is a low level interface to Tamino which offers a set of commands for storing retrieving and deleting database objects creating or erasing collections or schemas performing transaction processing and diagnostic testing In order to run an existing application that uses the Tamino API for C one needs the library for webserverless access and a local installation of the Software AG product eXtende
105. o a database and to edit the contents of a BAUML document The appendix of this work contains a user interface guide which explains the meaning of the various widgets and describes how one can use the application to edit a BAUML document Read it to see some screenshots of the application and to get further information about its user interface 7 3 4 Implementation During the implementation of the browser we stumbled over a strange performance problem This topic has already been treated in the implementation chapter and thus will not be discussed here any further 7 3 BAUMLBROWSER APPLICATION TT Figure 7 1 Simple room model dr BAUML XML Database Browser Options About Database stb Collection bauml Disconnect Bauml Collection Room HEO446 SpatialContainer Vienna SpatialContainer csbuilding Building csbuilding Room YARDT Room FLOORO Room HE0435 Room HG0401 Room HED4064 Room HEO401 Room HF0405 Room HFO405x Room HED4194 Room HD0407 Room HD0405 Room HD0403 Room LIFT3 Room HD0401B comic aac A AA Se gt RotX Roty a Insert Update Delete Save Has point x fo Figure 7 2 Screenshot of the BAUMLBrowser application 78 CHAPTER 7 SAMPLE APPLICATION Chapter 8 Summary This work is the base for employing an XML database system for Studierstube applications It shows that it is possible to uti
106. odel can be depicted as in figure 7 1 on page 77 This example only demonstrates the most basic features of BAUML language Read chapter 5 of 5 in order to get a comprehensive description of the language concepts and more advanced features 7 3 BAUMLBrowser Application BAUMLBrowser is our test application for the Studierstube XML Database API It was written to test the functionality and usability of the API and was meant to gain first experiences utilizing an XML Database in a real world application Therefore it covers all of the features offered by the API and thus can be seen as a reference for implementing Studierstube XML Database clients Apart from that it is a handy tool when dealing with BAUML documents which are stored in an XML Database 7 3 1 Features BAUMLBrowser is a Microsoft Windows application which is capable of browsing and editing BAUML documents stored in an XML Database It covers features for viewing the structure of aBAUML document displaying a graphical representation of BAUML objects converting BAUML data from and to Coin2 scripts and editing the nodes of a BAUML document The editing capabilities include functions for inserting updating and deleting BAUML objects In addition to that the application offers two special functions A function for testing object intersection and one for testing point inclusion 76 CHAPTER 7 SAMPLE APPLICATION 7 3 2 Core component In order to be able to reuse the feature
107. of XMLdbGUI as a comfortable user interface seems to be the crucial factor for the usability of Xindice 4 3 3 Database features Xindice is currently supporting XPath XUpdate and additionally SIXDML a simplified non standard query language The XPath implementation is taken from the Apache Xalan project which is said to be fully W3C compliant But this W3C specification does not cover XPath queries across collections and recursive queries of sub collections which are currently not supported by Xindice Database indexes are important to speed up queries Therefore Xindice offers manual indexes which means that for each element one wants to be included in the index one has to add an entry in the Xindice configuration file The note in the Xindice FAQ 34 that the database does not process big sized documents made us to do a simple test We took the sample phonebook files that come with Tamino they are stepwise sized from 1000 entries up to 8000 entries which matches 470KB to 3 7MB tried to insert them into the database and then run some queries Using Xindice out of the box with the default maximum main memory setting of 64MB resulted in memory exceptions and wrong or empty query results This malfunction could be eliminated by increasing this memory limit to 256MB Though we then got correct query results we noticed that inserting more big sized documents into the database dramatically decreased query speed The problem with big
108. omponents which contain details about the error reason are delivered in quite different formats Some components use plain text others HTML and the third one XML in order to help the user solve the problem To be able to subsume these different formats into one the Database access module uses the most common of these the text format Most of the module s methods return a Boolean value indicating whether the operation was successful or not In the case of a failure you can call the getLastError method of the object to find out what was going wrong The format of this message depends on the component which caused the error As mentioned before the result of a query operation can also contain an error method returns true The Database message even when the Database execute class does not analyze the server response because it cannot distinguish between an error message and a valid database response When using server side stylesheets one can produce any imaginable output which also includes error messages Thus it is left to the user to decide whether a server response is a valid query result or not B 2 SoXML DOM Model module This section describes the SoXML DOM Model module which is part of the Studierstube Database API The SoXML DOM Model provides a simple document object model to represent and process XML data When using this model you are able to access and update the content and structure of XML doc
109. ons for inserting updating and deleting data Since Software AG is a member of the W3C XML working group we can expect that similar update operations will be introduced in one of the future versions of XQuery The following example demonstrates how to use XQuery in order to implement the task of the previous section We presented an XSLT stylesheet which extracts the title and the author from a list of articles stored in an XML document In this example we use the For Let Where Return clause spelled FLOWER of XQuery to implement this FOR a IN input library article WHERE a author Ronald Bourret RETURN lt result gt a Otitle a author lt result gt The FOR statement specifies the working set of the query which in our example is the list of all articles contained in the XML document Furthermore in this statement the items of the set are bound to the variable a The WHERE clause then reduces this set to the articles written by Ronald Bourret by testing the author attribute Finally the RETURN clause is called for each matching instance of the set and thus returns the list of articles As one can see compared to the XSLT code of the previous example the solution using XQuery is rather short and much more readable If you want to learn more about this forthcoming standard we recommend to read 12 and to have a look at 13 and 14 20 CHAPTER 2 RELATED WORK 2 1 7 Defining languages using XML Sche
110. ople gt lt person gt lt firstname gt Edward lt firstname gt lt lastname gt Samson lt lastname gt lt person gt lt person gt lt firstname gt Ernestine lt firstname gt lt lastname gt Johnson lt lastname gt lt person gt lt person gt lt firstname gt Betty lt firstname gt lt lastname gt Richardson lt lastname gt lt person gt lt people gt lt meeting gt lt meeting number 2 gt lt title gt XML databases lt title gt lt date gt 31 5 2004 lt date gt lt people gt lt person gt lt firstname gt Ernestine lt firstname gt lt lastname gt Johnson lt lastname gt lt person gt lt person gt lt firstname gt Betty lt firstname gt lt lastname gt Richardson lt lastname gt lt person gt lt people gt lt meeting gt lt meetings gt The root tag lt meetings gt of this sample document contains a list of the staff meetings which have been held Currently there are two meetings in the list A meeting 92 APPENDIX B DATABASE API MANUAL enclosed by the tag lt meeting gt is identified by a unique number attribute stores a title the date when the meeting was and a list of people participating For each person of the people list there are two child tags containing the persons first name and last name This will be the basic data for the source code examples of the next sections B 1 3 Retrieving objects from the database Using the database whi
111. osely as possible Unfortunately the development of this class library has been suspended in an early state and currently it does not seem that it will ever be finished Nevertheless we had a look at this implementation We were once more disappointed to find it poorly documented and a bit quirky Maybe the increasing demand for XML applications will motivate someone to continue this work at a later time Chapter 3 Problem Description 3 1 Introduction This chapter specifies the conceptual formulation of this project or in other words describes the task which was consigned to the author of this thesis It defines the goal of the project by describing the basic features of the applications and the content of the documents which should be included in the final result We have also included some of the details that have only been verbally discussed later in the project Studierstube is a scientific platform for contriving and developing Augmented Reality applications Naturally in the context of research a large number of documents and a lot of resulting data has to be managed Up to now the collaborators of this platform solely utilized the file system of their workstations to persistently store and organize the accruing data As one can imagine at a certain point this form of organization will come across its limits When a certain amount of data is exceeded or the data needs to be interconnected and easily shared between the colla
112. ple pattern matching transformation statements which are executed successively Since an XSLT document is just plain text a special application is needed to interpret the included statements and apply them to an XML document Such an application is called XSLT processor There are a number of freely available XSLT processors which can be downloaded from the Internet like Xalan from the Apache project or Saxon which was written by Michael Kay In the following example we will compile a list of all articles contained in the library document of the previous section The resulting document should contain the article s title and the name of the author separated by a comma lt xml version 1 0 encoding UTF 8 gt lt xsl stylesheet version 1 0 xmlns xsl http www w3 org 1999 XSL Transform gt 18 CHAPTER 2 RELATED WORK lt xsl output method text gt lt xsl template match library article gt lt xsl value of select title gt lt xsl text gt lt xsl text gt lt xsl value of select author gt lt xsl template gt lt xsl stylesheet gt Since an XSLT stylesheet is an XML document the first line has to contain the XML declaration The second line defines the root element xsl stylesheet of the stylesheet and declares the namespace of the XSLT language The subsequent xsl output statement then sets the output method to plain text The XSLT output statement supports the methods text html a
113. possible to update documents across collections and XPath and XUpdate partially subcollections implementation not complete yet Less support for standard API s Inappropriate storage technology Still in beta state Needs MS Java VM gt 3805 Only manual indexes which is hard to get Is said to be retired Expensive Other free available XML Databases Product URL Comment dbXML Bio en Seems that the Open Source Version is a performance reduced lite Ze version ofthe commercial product http ashpool sourcefor Uses standard SQL92 syntax to query add update and delete XML Ashpool ge net documents 4Suite SR Suite Ayaan 4Suite is a collection of Python tools for XML processing and object ae z database management Berkeley http www sleepycat co Berkeley DB XML is supplied as a library that links directly into the DBXML m products xml shtml application s address space No support for XUpdate DBDOM A Is an implementation of the DOM over a relational database ge net Be tion Mean ane A Java based object oriented database management system No support for XPath or XUpdate http sourceforge net p XDBM Is based upon the DOM standard rojects xdbm XDB hip zvon org indexp Built on arelational database hp nav_id 61 Appendix B Database API manual The Studierstube XML Database API is a programming interface which provides access to a Tamino XML Database Us
114. r username string password string bool close bool get path string result string bool Database parseString inputString string SoXML open hostname string database string username string password string bool close bool execute query Query collection string result string bool Query query string stylesheet string tagName SoSFString tagPrefix SoSFString attributes SoMF String attributePrefix SOMF String values SoMF String SoXML Text cdata SoSFString Transformer applyStylesheet xslFilePath string xmiStr string bool Legend Aggregation gt 99 Figure 5 3 UML class diagram of the Studierstube XML Database API 5 6 API Classes This section presents the design of the Studierstube XML Database API classes in form of a UML class diagram see figure 5 3 on page 59 For the ease of reading we included only the most important attributes and operations Most classes are self explanatory because we have already discussed their features in the previous chapter Therefore we will merely discuss topics which are in need of further explanation 5 6 1 HTTPConnection Normally HTTP is a stateless protocol and thus would not require a dedicated start and endpoint as provided by the open close methods of the HTTP Connection class But the Keep Alive extension to HTTP as defined by
115. racted by using the method getText which concatenates all text objects contained in the tag language and the background and foreground attributes of the color tag are accessed with the help of the getAttributeBy Name method which delivers the value of an attribute by supplying its name The output of the sample program should then look like this language eng background white foreground black B 2 3 Constructing an XML document In contrast to the previous section this section shows how to construct and serialize an XML document from scratch by utilizing the SoXML classes First we will build a SoXML DOM model in memory which represents the XML document then we serialize the document using the class Parser and finally we write the result of this process to disk This way for instance you are able to create a default configuration file solely from within your application The following example constructs the XML document of the previous section and stores it to the file configuration xml Include header files include lt string gt include lt fstream gt 1 Create the root tag SoXML appSettings new SoXML appSettings 2 ref the root object appSettings gt ref 3 Add the language tag SoXML language new SoXML language language gt addChild new SoXMLText eng appSettings gt addChild language 4 Add the color tag SoXML color new SoXML color color
116. ral operating systems Linux IRIX Windows Most important of all are Windows and Linux which run on the desktop computers of Institute s staff members Hence it is required that we consider this when designing our software At least the client side of our system should be portable between these operating systems Generally speaking it will not be necessary to demand such a flexibility from the server side Since Tamino is also available for above operating systems except for IRIX we should try to accomplish this portability for the server too Thus we demand P1 The entire system including the server should be portable between different operating systems We can realize this by selecting appropriate components and programming languages which are also available for mentioned operating systems 5 3 2 Modularity and Reusability M Though the importance of modularity seems to be widely appreciated the enormous increases in the quantity of software production in the past decades have not yet led to a similar increase in the quality of software Modularity is the key to improve software quality and to build robust applications 2 Sticking to this paradigm we 50 CHAPTER 5 DESIGN can acquire reliable and reusable components and reduce the effort of testing the implementation Furthermore a modular design enables us to easily exchange parts of the system at a later point of time if necessary Since these are worthwhile feature
117. re Tomcat servlet engine Every Java servlet needs a proper operating environment in order to be executed In principle the Tamino Passthru Servlet can be used with any Java servlet container that supports the Java servlet specification version 1 2 Since the main environment used by the Software AG developer team was Tomcat 4 and the Tamino manual gives detailed instructions on installation and configuration we decided to employ this version of Tomcat Furthermore it is freely available from the Apache website completely written in Java design request P1 and used in the official reference implementation for the Java Servlet technologies Note There is also a later version of Tomcat available version 5 x but this one is not compatible with the Tamino Passthru Servlet because it does not support the JSP 1 2 specifications anymore MOD_JK module The MOD_JK module is a plug in to Apache that handles the communication between the Apache web server and Tomcat It is of course also available for UNIX platforms Studierstube Passthru Servlet The Studierstube Passthru Servlet is an enhanced version of the Tamino Passthru Servlet It supports the execution of an additional fixed server side stylesheet This is discussed in detail later in this chapter Xalan XSLT processor Since we want to execute stylesheets on the server we need an XSLT processor to accomplish this The Tamino Passthru Servlet will work with any XSLT processor
118. ree nodes indicates that the corresponding BAUML object contains some child objects Click on it with the left mouse button to open the list of child nodes C 1 2 Graphics window Right to the tree control of figure C 1 you can see a window showing the graphical representation of a room object An embedded Coin viewer is used to display the representation part of the BAUML objects You can view an object simply by selecting it with the mouse in the tree control The title bar on top of the viewer window then shows the name of the object viewed If an object does not have a representation part the text no representation is displayed instead Consider that the representation of an object does not include the representation parts of its child objects C 1 USER INTERFACE GUIDE 107 Options Host localhost Port 80 Cancel Servlet path stbxml servlet transform tamino Reconnect to the database in order to use the new settings Figure C 2 BAUMLBrowser options dialog C 1 3 Setting application options The first thing you need to do when working with the browser application is to set up the basic application options Select the Options item from the program menu to open the dialog which is depicted in figure C 2 on page 107 Enter the correct host name or IP address of your XML Database server into the Host field The Port and Serlvet path fields depend on the server installation and default to 80 for
119. rough an unfamiliar building It is an Augmented Reality navigation system where the person wears a mobile equipment consisting of a head mounted display a camera and a tracking system SignPost relies on an XML based data structure called Building AUgmentation Markup Language BAUML that holds geometric information of rooms and buildings as well as the placement of markers which allow to determine the current position of the user Since one of the future application fields of our XML Database is to provide BAUML data we decided to write a small application which is capable of browsing and manipulating BAUML documents stored in a database The following sections present this application called BA UML Browser after giving a basic introduction to the BAUML language 7 2 BAUML Language BAUML is an XML language for the representation of geometric information It allows to describe any object or set of objects by specifying the position of its vertices and surfaces in a three dimensional coordinate system Some BAUML objects can be further refined and described in more detail by adding a set of child objects The 73 74 CHAPTER 7 SAMPLE APPLICATION resulting data structure is a tree of BAUML objects where each additional tree level refines the associated objects above Due to its recursive definition BAUML allows an infinite nesting of objects and thus enables to specify an arbitrary level of detail There are several types of BAUML o
120. s in general we demand that M1 The system should be built of components which can be exchanged and reused later In order to reach this goal we apply the techniques of object oriented design information hiding and modularization of both data and processing We will split up the system into small reasonable units with tight interfaces between them 5 3 3 Usability and Acceptability U Usability is a function of the ease of use and the acceptability of a product where ease of use affects the user performance and satisfaction and acceptability affects whether or not the product is used 3 Usability related questions usually deal with the design of graphical user interfaces in order to improve the productivity and acceptance of an application Likewise the same characteristics can be applied to programming interfaces which are the user interfaces for developers According to 4 the usability of the languages and libraries that developers use even have a significant impact on their ability to successfully complete a set of development tasks Serious usability work includes tasks like investigating the target user group creating usage scenarios running usability studies and so forth Since this is out of the scope of our work we will take an easier approach and try to improve the usability of the client by keeping it as simple as possible while still providing all of the required features We should aim to get U1 A simple
121. s which represent the elements and tags of the corresponding XML document 5 5 4 Server The right side of figure 5 2 shows the server which is completely composed of existing components This means that we do not have to develop any new components for this part of the system All we need to do is install configure and optimize the components of the server in order to make them work properly together Since we want to perform server side stylesheet processing we need an XSLT processor on the server which can be used by the Tamino Passthru Servlet to transform the query results As with the client this feature will be implemented by integrating one of the freely available software packages The Tamino Passthru Servlet in turn needs a proper operating environment in 58 CHAPTER 5 DESIGN order to be executed For this purpose we have to install a Java servlet engine and embed the Passthru Servlet into it Normally an additional web server would not be required to run a servlet but in this implementation the Passthru Servlet utilizes the web server to access the database by forwarding the query request to it For the ease of reading we have omitted this step in the diagram of figure 5 2 Furthermore the web server is needed to run Tamino Manager the administration application of the database system 5 5 5 Workflow of a query Having introduced all of the components of the client and the server this section explains how they work tog
122. s establishing a hierarchy analog to class hierarchies of object oriented programming languages The technique of creating new types by derivation is one of the key features of XML Schema since it has already proven its usefulness in many software projects for years In the following example we define an XML Schema for the article element which we used in the previous sections lt xml version 1 0 encoding UTF 8 gt lt xs schema xmlns xs http www w3 org 2001 XMLSchema gt lt xs element name article gt lt xs complexType gt lt xs attribute name title type xs string gt lt xs attribute name author type xs string gt lt xs complexType gt lt xs element gt lt xs schema gt Above XML Schema starts with the XML declaration followed by the root element xs schema which is always the topmost element Since the XML Schema specification defines a namespace we need to declare it here too Next comes the definition of the article element which is of complex type because it owns two attributes The attributes for the title and the author of the article are then declared as child elements of the xs complexType tag using the xs attribute element Finally the datatype of both attributes is specified as simple string because they should be able to store any text For further reading we recommend to study the tutorials at 16 and 17 and the W3C specification at 15 2 1 8 Working with XML docum
123. s extends its basic type with additional tags and attributes specific to its needs In the following example we use the BAUML language to define a model of a simple rectangular room width 5 depth 10 height 3 consisting of a floor and four walls lt SpatialObject id Simple room baseType Spatial0bjectType gt lt annotation gt Simple room model for demonstration purposes lt annotation gt lt representation gt lt Vertex position 0 0 0 gt lt Vertex position 5 0 0 gt lt Vertex position 5 10 0 gt lt Vertex position 0 10 0 gt lt Vertex position 0 0 3 gt 7 3 BAUMLBROWSER APPLICATION 75 lt Vertex position 5 0 3 gt lt Vertex position 5 10 3 gt lt Vertex position 0 10 3 gt lt Polygon type floor vertices 0 1 2 3 gt lt Polygon type wall vertices 0 1 5 4 gt lt Polygon type wall vertices 1 2 6 5 gt lt Polygon type wall vertices 2 3 7 6 gt lt Polygon type wall vertices 0 3 7 4 gt lt representation gt lt Spatial0bject gt Since this room has no doors or windows we use the generic SpatialObject type for that We define the model by specifying the eight vertices of the cuboid and five polygons a floor and four walls For this aim the vertices attribute of a polygon tag contains a list of numbers indexes of the cuboid vertices Furthermore we have annotated the object in order to document its use If we use the BAUMLBrowser application this m
124. s of the browser in other applications we have separated the parts which are independent of the operating system All functions which are not related to the Microsoft Windows system have been encapsulated in a class called BAUMLBrowser Mostly this concerns functions which are not dealing with the user interface of the application The BAUML Browser class builds a layer on top of the database API and offers functions to access and process BAUML documents This way the browser can be easily ported to other operating systems like Linux The appendix of this work contains a developer manual which gives a comprehensive description of this class in order to implement a database browser 7 3 3 User interface Since BAUML is a data format which is based on a tree structure we decided to give the application an Explorer like user interface BAUMLBrowser is a so called dialog based application which means that its user interface basically consists of a single window containing all the widgets needed for operation At the left side of the dialog see figure 7 2 on page 77 one finds a tree view displaying the structure and the nodes of a BAUML document It can be used to browse through a document and to select certain BAUML objects for manipulation Right to the tree view is a Coin viewer window which shows a graphical representation of the currently selected BAUML object Furthermore the dialog contains various widgets which are used to connect t
125. se only consists of the system collections at the very beginning Next we install XMLdbGUI a Java application with a graphical user interface used to browse and edit database content Since the full version of XMLdbGUI currently only supports Xindice 1 0 and the included Java libraries are already out of date it is recommended to download the light version and copy the required files as described in the documentation Furthermore you will need to edit the configuration file conf xml which is also explained in the documentation that comes with XMLdbGUI Hint Use the URI xmldb xindice localhost 8080 if you do not get a connection to the database 4 3 2 Applications On the Internet one finds a number of client applications written for Xindice although most of them did not leave beta states until today Most outstanding are Xindice Webadmin a simple web interface for creating collections querying and uploading data and XMLdbG UI a Java application with similar features Unfortunately due to the changes made to the new database version applications like Xindice Browser which access the database directly do not work with Xindice anymore All in all the user interface provided by the applications does not cover all features of Xindice In order to be able to adjust and fine tune database specific settings 42 CHAPTER 4 CHOOSING A DATABASE PRODUCT one still has to edit configuration files by hand The availability
126. side an object and writes the result of the test to standard out 1 Browser and node object BAUMLBrowser browser select an existing node BAUMLBrowser Node node 2 Test if the point 1 2 3 is inside the object if browser containsPoint node 1 0 2 0 3 0 printf Yes The point is inside else printf No the object does not contain the point 1 The first lines of this example instantiate a BAUMLBrowser class and select a Node object which is then used in the test 116 APPENDIX C BAUMLBROWSER MANUAL 2 Then the representation part of the BAUML object which is represented by the variable node is tested wether it includes the point 1 2 3 by calling the browsers containsPoint method Finally depending on the test result an appropriate message is printed Appendix D Installation and Configuration guide This appendix lists the software needed and describes the necessary steps to install and configure an XML server for Studierstube The installation procedure also includes the server side part of the Studierstube XML Database API It is strongly recommended to use the same versions of the software packages as specified here since these are the versions the software was developed and tested for D 1 Installation D 1 1 Hardware prerequisites The required hardware components are mainly determined by prerequisites of the Tamino XML server which currently are e Intel Pentium III at
127. stylesheets List of Figures 2 1 2 2 2 3 5 1 5 2 5 3 6 1 6 2 6 3 6 4 7 1 7 2 B 1 B 2 C 1 C 2 Collaboration diagram of VCL database classes 2 27 Collaboration diagram of MFC ODBC database classes 28 Collaboration diagram of XML DB API database classes 29 Local area network of the Institute 48 Dataflow diagram of the system architecture 56 UML class diagram of the Studierstube XML Database API 59 UML class diagram of the Database access classes 2 2 2 2 62 UML class diagram of the Transformation classes 64 UML class diagram of the SOoXML DOM Model classes 65 Diagram of server components and scripts 67 Simple room model 21d aa ee AAA es 77 Screenshot of the BAUMLBrowser application red Collaboration diagram of database access classes 2 2 2 2 90 Collaboration diagram of SOXML classes 4 101 BAUMLBrowser user interface 2 oa a 106 BAUMLBrowser options dialog ooa 107 123 124 LIST OF FIGURES Bibliography 1 Rick Kazman Leonard J Bass Mike Webb and Gregory D Abowd SAAM A method for analyzing the properties of software architectures In International Conference on Software Engineering pages 81 90 ICSE 1994 2 Karl Crary Robert Harper Peter Lee Frank Pfenning Modularity Matters Most Carnegie Mellon University Pittsburgh October
128. t 1 various objects used in this example create a Coin viewer object BAUMLBrowser browser SoWinExaminerViewer viewer select an existing node to be displayed BAUMLBrowser Node node 2 get the representation part of the current node string ivData browser getRepresentation node 3 setup an input buffer for the SoDB readAll function Solnput in in setBuffer void ivData c_str ivData length 4 make a scenegraph from the text representation SoSeparator root NULL root SoDB readAll gin 5 show the scene graph in the viewer viewer gt setSceneGraph root viewer gt show 1 In order to keep this example short assume that a Coin viewer object has already been created and the BAUMLBrowser node to be viewed has been selected too 114 APPENDIX C BAUMLBROWSER MANUAL 2 The BAUMLBrowser class provides various methods for getting object data Probably the most interesting of these is the getRepresentation function It reads the representation part of a BAUML object from the database and converts it to a Coin script 3 4 Using an input buffer the Coin script is delivered to the readAll function 4 which creates a Coin scene graph 5 This scene graph can then be displayed by the viewer object C 2 6 Updating objects This section introduces functions for inserting updating and deleting whole BAUML objects and functions to insert and update their repr
129. t This set of parameters which is represented by the class Query includes the XPath or XQuery script and the path to the server side stylesheet Due to this design the same query can be consecutively executed on different databases The class Database is an abstraction of a database instance available on the server Remember Tamino is capable of serving multiple databases at the same time The job of this class is to store network related parameters manage an HTTP connection to the server and to translate query parameters to HTTP requests This way we are also getting a layer making it transparent which interface is used to access the database According to our demand for reusability M1 we provide the HTTP connection in form of a dedicated class encapsulating the parameters and features which are associated with this protocol The remaining components implement the post processing features of the API which can be optionally applied to a query response The class Transformer offers methods to execute a locally hosted stylesheet Since there are freely available XML processors which can be downloaded from the Internet we use one of these packages to implement this feature The same applies to the class Parser which uses an XML Parser to create a SoXML DOM model from an XML document and vice versa For the lack of space the SoX MI DOM model is depicted only by a single symbol in figure 5 2 Usually this model consists of a tree of node
130. t is set to the value ino servlet transform tamino and only depends on the configuration of the database server It has to be updated when the servlet is stored in a different location 64 CHAPTER 6 IMPLEMENTATION XSLTInputSource XalanTransformer XSLTResultTarget Xalan C A ER TR class class library Figure 6 2 UML class diagram of the Transformation classes 6 1 2 Transformation classes The second group of the client interface is represented by the transformation classes It supports the execution of XSLT stylesheets in order to transform XML documents Thus we needed to find an appropriate XSLT processor which can be used to implement this feature The current Studierstube setup already includes a recent version of the XML parser Xerces C which is developed by the Apache group Since the members of this group are usually known for delivering well done products and they also feature a portable XSLT processor namely Xalan C it was the candidate of first choice for our project The Transformer class of the client API is utilizing this XSLT processor library see figure 6 2 on page 64 In order to facilitate the handling of this library Transformer builds a simple wrapper around the Xalan library and offers an interface which is much easier to learn and understand 6 1 3 SoXML DOM Model classes The SoXML DOM Model classes cover the features to transform an XML document to a simple DOM model a
131. t this operation does not check the contents of the file It is the responsibility of the user to ensure the validity of the inserted data The second alternative creates a new SpatialContainer object with a representation part that complies with the contents of the iv file Since the BAUML format represents all geometric data in form of polygons Inventor shapes like cubes and spheres need to be converted to an appropriate form With the exception of IndexedFaceSet all shapes in the iv file are transformed to a triangle representation using the triangle callback action of Coin Whereas all polygons contained in an IndexedFaceSet shape are added to the representation as they are During conversion any transformations defined in the Coin script are applied to the data The newly created object is then inserted at the end of the child list of the currently selected object and the tree view is updated to reflect the changes Consider that is not possible to insert top level objects into the database with BAUMLBrowser Use the Tamino Xplorer utility to accomplish that C 1 6 Updating objects The update operation supports the same file types as the insert operation Everything said about file validation and conversion applies to the update operation as well Press the button Update to open a file dialog and select an xml or iv file which xml is used to update the currently selected BAUML object Updating from an file replaces the
132. tallation procedure it will be the first choice if one does not mind its costs It convinces by providing a well done set of client applications and last but not least a really extensive help system If Tamino is too expensive and if one can temporarily live with an incomplete implementation of XPath eXist is the way to go Though it is still in beta state it seems to be stable enough to be employed for a project It provides sufficient and accurate documentation in order to implement own client applications 4 5 CONCLUSION 45 Next choice would be the elder version of Tamino But then one will have to live without the feature to update documents with XUpdate what can become cumb ersome very SOON Unfortunately the current version of Xindice can not be recommended by us to be utilized for this project The missing XML RPC documentation the problems with big sized documents the lack of support for server side XSLT processing and last but not least the recurring rumor that it should be retired give it the lowest rank out of the three candidates But future will show what happens to it 46 CHAPTER 4 CHOOSING A DATABASE PRODUCT Chapter 5 Design Everything should be made as simple as possible but not simpler Albert Einstein 5 1 Overview This chapter discusses the design of the Studierstube XML Database system Having analyzed and compared the usability of the three candidate databases see previous chapter w
133. tbxml http www studierstube org xml xmldb response gt lt stbxml message gt lt ino message ino returnvalue 0 gt lt ino messageline gt XQuery Request processing lt ino messageline gt lt ino message gt lt ino message ino returnvalue 0 gt lt ino messageline gt XQuery Request processed lt ino messageline gt lt ino message gt lt stbxml message gt lt stbxml result gt lt title gt XML for Augmented Reality lt title gt 94 APPENDIX B DATABASE API MANUAL lt title gt XML databases lt title gt lt stbxml result gt lt stbxml response gt Since the schema of an XML Database result has not been standardized until now and each database server uses a different format we decided to define our own format in order to be independent of the manufacturer The root element stbxml response of the result document declares a namespace stbxml which is used to identify responses of the Studierstube database server It is composed of the sub elements message and result which are also declared in this namespace These sub elements are mandatory and thus always present in a response document The message part contains manufacturer specific information about the query most notably error messages You will find this information useful when debugging your application Whereas the result part holds the actual result set of the query the data you are interested in When using XQuery t
134. tem like Linux The class basically defines a layer on top of the Studierstube XML Database API which provides operations to deal with BAUML documents contained in an XML Database collection This covers functions to insert update and delete BAUML objects and their representation as well as transformations of this representations to and from Coin scene graph scripts In addition to that two test functions have been implemented for checking object intersection and point position The following sections show the basic ideas behind this component and illustrate how to utilize it to implement a database browser 110 APPENDIX C BAUMLBROWSER MANUAL C 2 1 Basic concept A BAUML document can be basically seen as a tree of BAUML objects Each node of this tree is a target for manipulation through the user Thus a mechanism is needed to uniquely identify each node This is the job of the Node class A Node object stores the location of a BAUML object and additional information like object type and name which can be used to label nodes of a tree control The Node class is a convenient way to work with BAUML objects without having to deal with XML related implementation details All database related methods make use of this class When working with these methods you first get information about existing BAUML objects in form of Nodes and then use these Nodes to identify the BAUML object you want to update Reading the entire content of a database at once wou
135. th Relational Databases Thus it was not even clear whether they are matured enough to be utilized for a project like that Therefore this thesis can also be seen as an experiment to check out the current state of XML Database technology And to anticipate it it turned out to be an experiment with a successful outcome Reading the problem statement of this thesis one may think that this is an easy task which can be accomplished in no time as the author admittedly was when he started his work Due to the youth of this technology several problems needed to be solved which came across us mainly during the implementation phase Nevertheless the author does not regret having started this project since he learned much about the upcoming XML technology Hopefully this work is a contribution to support the development of the Studierstube system and will serve as a base for further projects Chapter 2 Related work 2 1 XML 2 1 1 Introduction XML is like many other names in the information technology an abbreviation and stands for the term Extensible Markup Language It is a text based meta language for the definition of computer languages which describe the syntax and structure of data As a markup language the syntax of XML is based on tags and attributes and thus looks quite similar to HTML But in contrast to the fixed language constructs of HTML XML allows to define your own tags and attributes In this way it is possible to cr
136. that implements the Java API defined in the JAXP 1 1 specification There are several freely available processors we can choose from for example Saxon XSLTC jd xsit and Xalan J 6 2 SERVER 69 In order to get transformation results which are as close as possible to that of the client when using the same stylesheet we decided to employ the Java version of Xalan Xalan J This facilitates to shift the execution of a stylesheet from the client to the server and vice versa without the need to modify it Configuration scripts The configuration scripts depicted in figure 6 4 are an important part of the database server because they define how the components are interconnected In the following section we give a short description for each of the scripts httpd conf is the main configuration file of the Apache web server The Tamino setup modifies this file in order to enable HTTP access to the database and for the execution of its web based services Furthermore it includes a link to the configuration script of Tomcat apache connector conf is responsible for loading the MOD_JK module and defines the mount points of the servlets Furthermore it includes a link to the workers properties configuration script workers properties gives Tomcat information about where to listen for requests from the Apache web server web xml is the main configuration script of a Tomcat servlet It defines the mapping of URLs to servlet code and provides a s
137. the HTTP 1 1 draft allows persistent connections These long lived HTTP sessions allow multiple requests to be sent over the same TCP connection and as a consequence will speed up database queries Moreover adding the open close pair to these classes facilitates to replace the HTTP protocol by any connection oriented protocol at a later point of time if necessary M1 60 CHAPTER 5 DESIGN In order to support authentication E3 we also added the parameters username and password to the open methods of the classes HTTPConnection and Database 5 6 2 String as query result type Examining the execute method of the class Database reveals that the data type of a query result is a string This is in contrast to our demand for the support of binary data and Unicode E4 Here we preferred simplicity U1 over extensibility E4 Our first draft of the API provided a resource class to store arbitrary query results but this turned out to be impractical Anyway binary data can be encoded in strings for example in form of Uuencoded text if really needed 5 7 Conclusion This section resumes the system design and tries to point out the pros and cons of our approach with the help of the following list Pros Depending on the components we choose we can achieve a completely portable system P1 The client system can even be divided into three independent parts Query amp Database Transform
138. the database string response database execute query meetings response added 2 Extract the query result Transformer transformer transformer applyStylesheet getresult xslt response 6 Print the result printf s response c_str 96 APPENDIX B DATABASE API MANUAL After including the header file added 1 we add an instance of Transformer and call the method applyStylesheet added 2 to transform the server response by applying the stylesheet file getresult xslt The processed result set and thus the program output will then look like the following text lt title gt XML for Augmented Reality lt title gt lt title gt XML databases lt title gt Server side transformation Beside the possibility to run client side transformations the database access module also supports the execution of stylesheets on the server You will prefer this method if you want to keep network bandwidth low or use some basic stylesheets in many different applications for example The downside of this method is that the server load increases with the execution of each stylesheet and that the server s error messages are discarded before the result reaches the client This might be an annoying fact when it comes to debug your application You will have to temporarily disable the execution of the stylesheet in your source code in order to find out what is wrong The transformation is specified by setting the parameter
139. the left side of figure B 2 is fed to the Parser by calling its method parse which creates the SoXML tree on the right side The root object root of this tree contains two child objects One SoXMLText object storing the text Hello world and one SoXML object holding the content of the tag element The dotted line shows the opposite way where the B 2 SOXML DOM MODEL MODULE 101 XML document root gt Hello world lt element gt Parser serialize lt jelement gt root gt SoXML Text Hello world SoXML element Figure B 2 Collaboration diagram of SoXML classes SoX ML tree is serialized by the parser and generates an XML document B 2 2 Parsing an XML document This section shows how to parse an XML file and how to access the tags and attributes of the resulting SoX ML tree Imagine that we are writing an application which uses a configuration file based on XML These application settings should cover the language of the application and the background and foreground color of the main window For this purpose we create an XML file with the following content lt appSettings gt lt language gt eng lt language gt lt color bg white fg black gt lt appSettings gt The root element appSettings of this document embeds two child tags which define the configuration data for our application The tag language encloses a token setting the applicatio
140. the port and stbxml servlet transform tamino for the servlet path Leave them unchanged if not told otherwise Close the dialog by clicking on the OK Button This saves the application options in the application configuration file Consider that you need to close an open database connection and reconnect to apply the new settings C 1 4 Connecting to a database The next step after configuration is to open a connection to an existing XML Database server On top of the dialog in figure C 1 you see two edit fields named Database and Collection Enter the name of the database and collection you want to browse and press the button Connect After establishing a connection the application reads in the top level objects and the text of the Connect button changes to Disconnect C 1 5 Inserting new objects On the bottom left side of the application dialog in figure C 1 you find a number of buttons that offer operations to edit the current BAUML document Located on the left is the button for inserting new objects Press the button Insert to open a file dialog which allows you to select a file to be inserted into the database This operation supports two file types You can either insert a BAUML object from an xml file or from an Coin iv script 108 APPENDIX C BAUMLBROWSER MANUAL The first alternative creates a new BAUML object which is identical to the contents of the xml file Consider tha
141. those articles which contain an author attribute with the value Ronald Bourret For further information concerning XPath see W3Schools XPath Tutorial 9 and 10 for a comprehensive set of practical examples explaining the various ways of selecting data with XPath 2 1 5 Applying transformations using XSLT XSLT which stands for the term Extensible Stylesheet Language Transformations is a programming language especially developed for the transformation of XML data An XSLT document also called stylesheet contains rules and statements which define how to transform the data Using this language an XML document can be transformed to any text format including HTML and XML Since XSLT is defined using XML an XSLT program is also an XML document it follows the syntactical rules of XML too The basic principle of how an XSLT stylesheet works is to select certain parts of an XML document and apply a transformation to them It works a bit like a pattern matching mechanism where the patterns are specified by XPath expressions see previous section The XPath expression selects the parts of the documents which should be transformed in a certain way The transformation is then defined by special XSLT statements which access the data stored in the XML fragment and copy transform or rearrange them in a new way The transformation is applied for each XML fragment matching the XPath expression In general an XSLT stylesheet contains multi
142. tures 3 4 DATA REPRESENTATION IN FORM OF SOXML 33 Specification of the source database Network address of the server and database Specification of the wanted data by the query language XPath Query results in the following forms XML document as text in the computer memory XML DOM as SoXML model see below Specification of an XSLT Transformation which is applied to the query result on the server side Specification of an XSLT Transformation which is applied to the query result on the client side For the processing of XSLT a library like Xalan should be selected and linked to the library Furthermore query parameters like XSLT style sheets and XPath statements should be easily reusable within an application A suitable object oriented container class has to be designed to achieve this The client API should also be designed in a modular manner that is possible to utilize only parts of the API In particular the transformation features should be usable also by Non Studierstube applications Results of this work step e A specification of the Client API 3 4 Data representation in form of SoX ML Studierstube is built on top of Coin2 formerly Open Inventor a toolkit for the interaction and rendering of 3D graphics Coin uses a scene graph model where graphical objects are represented as nodes of a tree In order to integrate XML data into this model a connection between XML and Coin is needed The
143. ually kept in files So for instance applications that need to manage structured configuration data can do this conveniently by utilizing XML as storage medium in form of files Due to the popularity of XML the amount of data that is organized using XML has enormously increased in the past few years The more data a system has to manage the harder it is to keep it consistent Handling a large number of XML documents in form of single files and using the directory structure to organize them can become very impractical soon And if the number goes into the thousands this can become an impossible mission and an appropriate storage system usually a database is needed Apart from managing large amounts of data there are several more reasons that speak for employing a database in a project So for instance by utilizing indexes a database usually provides a better performance than a file system Furthermore a database is capable of managing concurrent access of multiple users offers data integrity functions transactions triggers and a lot more In regard to XML such a database system is called XML Database The term XML Database subsumes all available database systems which are capable of storing and retrieving XML documents This can be accomplished either by utilizing conventional Relational Databases or by databases which are especially designed for the storage of XML documents Ronald Bourret 27 distinguishes between the follo
144. uded with Tamino In the case it is not included you can download it from the Apache Web Server page www apache org Choose Typical installation from the menu and follow the installation instructions D 1 INSTALLATION 119 Install Tamino XML Server Install Tamino XML Server 4 1 4 Choose Complete setup from the menu and use the default settings You should have an installation CD for the Tamino software Otherwise you can get a 30 day trial version at the XML Starterkit page of Software AG www xmlstarterkit com You will also need to register to get a valid license file via e mail Hint You will need to reboot the system after installation Install Apache Tomcat Install Apache Tomcat 4 1 29 This servlet container is needed to run the software which executes the server side XSLT transformation Download it from the Apache Jakarta Project page jakarta apache org During installation you will be asked to point to a directory containing a Java SDK choose the one you have installed before and install Tomcat as NT Service Install MOD_JK DLL Install mod_jk_1 2 5_2 0 47 dll This DLL is needed to connect Tomcat and the Apache Web Server Get it from the Apache web page www apache org and copy it to C Program Files Apache Group Apache2 modules Add the Passthru Servlets Copy the contents of the Tomcat webapps directory from Studierstube Subversion server to C Program Files Apache Group Tomcat 4
145. uite flexible to use but relatively complicated to handle This is in contrast to our demand for usability U1 e In order to be as flexible as possible the class framework of the XML DB API covers a lot of features like a driver concept a service mechanism collection management and several more While these are worthwhile features in general they are not needed for our solution Thus we would have to invest a lot of work in order to implement the basic framework just to comply with the standard e The XML DB API neither supports the query language XQuery nor the execution of server side stylesheets Thus we would have to implement a new service extension for instance like XQueryService WithStylesheets which does not comply with the standard Which in turn would foil the utilization of the standard e Though the XML DB API is defined using IDL an implementation in C can be quite quirky See the XinCJ implementation to get an impression of what that means Thus it is clear why a simple but efficient implementation of the client interface looks much more reasonable 5 5 2 System model After completing all preparatory work for the design we are now ready to create a model of our system It is meant to describe the logical structure of the system and should outline all of the major components the system is composed of Furthermore each component should get a well defined task and a well defined interface in order to be easily rep
146. uments It offers operations to convert the text representation of XML documents to a simple tree based object model and vice versa The nodes of this tree model are made of objects which are derived from Coin base classes like SoNode and SoGroup The parent child relationship of the tree nodes is directly mapped to the model of Coin and implemented by utilizing the data structures of SoGroup That way XML data can be seamlessly integrated into a Coin scene graph and being processed by using standard Coin methods like tree 100 APPENDIX B DATABASE API MANUAL parsing and callback actions The advantage of deploying the SoXML DOM Model in your application is that it is easy to learn and use and its direct integration into Coin while the drawbacks are that it does not support all kinds of XML tags only provides simple methods to deal with namespaces and does not feature any data types Probably the most useful application of the SoXML DOM Model is for reading and writing simply structured configuration data that does not require sophisticated XML processing The following sections show how you can utilize the classes of this module to read and write XML based configuration data and how to access and modify its content B 2 1 Getting started The SoXML DOM Model module consists of three classes namely the classes Parser SoXML and SoXMLText The Parser class can be seen as the active part of the module It provides methods to parse and serializ
147. va VM anymore After an odyssey to find a way to get this update we succeeded by running the Windows update function and installed Java VM 3810 and in the end Tamino 4 2 2 Applications The Tamino database system comes with a number of applications The following sections will try to give a short description of each program Tamino Manager implemented as web application lets tune and configure the database server from a remote location manage databases and run the backup restore function Tamino Schema Editor a Java application is needed to edit Tamino s own XML Schema an adopted version of the W3C schema On the one hand it enhances this schema by adding logical and physical properties which describe storage and indexing information On the other hand it lacks of support for the more advanced features of XML Schema like derivation recursion and complex basic datatypes These limitations seriously restrict the area of application of this language especially when one needs to work with recursive data structures Tamino Explorer another Java application is used to browse the database through an explorer like user interface Database contents can be viewed and edited quite comfortably The Query windows allows to commit XPath queries to the database server and to view and inspect the result set 4 2 TAMINO 39 The last application to mention is Tamino Interactive Interface a simple Web form for querying and updating the databas
148. vice DatabaseInstanceManager IndexQueryService Available formats Size Quality Search functions HTTP Query the database with a web browser XML RPC Remote procedure calls using XML over HTTP WebDAV Web based Distributed Authoring and Versioning SOAP Simple Object Access Protocol Documentation Tamino XIndice eXist Y HTML Y HTML Y HTML Y PDF Y PDF 3000 printable pages 80 printable pages 50 printable pages Very good with many Good but missing a Small but accurate examples although comprehensive sometimes a bit installation guide and digressing XML RPC documentation Full text HTML search i None None engine Supported network protocols Tamino XIndice eXist No just document Yes in standalone Yes retrieval by Xindice mode HTTP Yes by Xindice XML Yes by Apache XML No RPC module written in RPC Java Yes by XinCon and Yes by XinCon in Yes Xindice Webadmin in servlet mode servlet mode No No Yes in servlet mode System configuration amp administration DTD amp Schema editors Data editors Data import export Applications Tamino Tamino Manager a graphical web application Tamino Schema Editor a graphical Java application editing a schema as a tree Tamino X Plorer a graphical Java application Tamino Interactive Interface a simple form based web interface Y Tamino Data Loader Command line Java Loader Command line XI
149. war file and install the servlet The war file can be safely deleted then Note When using Tomcat 4 1 as servlet engine you have to rename the war file to xindice war before installation otherwise you will not be able to connect to the database server Furthermore it is recommended to make more memory available to Xindice than 4 3 XINDICE 41 64MB which is the default configuration Add the parameter Xmx999m to the startup line of the Tomcat server where 999 stands for the maximum size of memory in megabytes used for Tomcat You should then check the installation by browsing to http localhost 8080 xindice If the installation was successful a web page for browsing the database called The ugly debug tool will show up Next we install the command line client tools which are included in xml xindice xxx jar zip Unpack the contents of the zip file to C Xindice If you use this directory you will not have to edit the batch file xindice bat which starts the command line programs Hint If the batch file is not working properly this is most likely because of missing quotation marks around environment variables Attention This does not apply to the environment variable CL Check the installation of the command line tools by starting the following line from a command prompt Xindice lc c xmldb xindice localhost 8080 db This should list the current database content which of cour
150. wing three types of XML Databases e Relational databases supporting XML e XML enabled databases e Native XML databases The key to choose the right one for a particular application lies in the structure of the data which needs to be processed Thus we have to have a look on the different forms of XML data first The subsequent sections then discuss the different properties and features of these database types 24 CHAPTER 2 RELATED WORK 2 2 2 Data centric vs Document centric XML data According to their structure XML documents can be roughly divided into two categories Data centric XML documents and document centric XML documents 27 Though it is not always clear to which category a document should be added to it makes sense to distinguish between these two categories It helps to choose an appropriate storage technology Data centric documents Data centric documents are characterized by a fairly regular structure with fine grained data small independent units of data at element and attribute level and with little or no mixed content XML tags mixed with text These are documents which have a structure similar to that of a relational database table and could possibly originate from it The following XML document showcases an example for a data centric document which stores patient data of a hospital lt xml version 1 0 encoding UTF 8 gt lt patients gt lt patient gt lt firstname gt Otmar lt firstname
151. y by replacing the stream library we were able to significantly speed up the processing which is shown in the following table Processing times after optimization Operation Time Time seconds percentage getRepresentation 5 9 sec 100 querying the database 1 9 sec 32 applying the stylesheet getResult xslt 2 3 sec 38 applying the stylesheet bauml2iv xslt 1 7 sec 28 The processing time of each operation is now about one third of the entire task and the overall performance gain is about 300 6 2 Server This section discusses the implementation of the database server Apart from modifying the Passthru Servlet the work that had to be done was mainly a matter of installation and configuration The following section describes which components were chosen in order to implement the server and how they are configured to work together Then we discuss why the Passthru Serlvet had to be adopted and the modifications which were made to it The last section deals with some unexpected and strange performance problems which came up when the server was tested with big sized XML documents 6 2 SERVER 67 SI Xalan Tamino XSLT Processor XML Database stbresult xsl loads apache workers connector conf Loads properties Studierstube Legend Aggregation gt component Assoziation e Figure 6 4 Diagram of server components and scripts 6
152. zing the API While developing the demonstration application BAUMLBrowser we found that the performance of the client API was somewhat poor It took about 20 seconds to query and process the mid sized BAUML object Karlskirche_bd what we found way too long The basic steps involved in this operation are to query the database apply a stylesheet in order to extract the result set getResult xslt and apply a second stylesheet bauml2iv xslt to transform it into a Coin script After we profiled the concerned BAUMLBrowser method getRepresentation we got the following 66 CHAPTER 6 IMPLEMENTATION surprising results Processing times before optimization Operation Time Time seconds percentage getRepresentation 19 3 sec 100 querying the database 1 9 sec 10 applying the stylesheet getResult xslt 11 7 sec 60 applying the stylesheet bauml2iv xslt 5 9 sec 30 From the table above one can see that applying the two stylesheets took about 90 of the entire operation while the query itself only took 10 Further investigation and profiling revealed the reason for this performance problem The stringstream template classes lt sstream gt of the Microsoft C runtime library are implemented in a way that slows down string stream processing Fortunately this library contains another string stream implementation namely lt strstream gt which performs much better Simpl
153. zu integrieren haben wir eine Programmierschnittstelle entworfen und implementiert die den Zugriff auf die Datenbank mit Hilfe der XML Abfragesprachen XPath und XQuery erm glicht Abschlie end haben wir die Implementierung des Systems mit Hilfe einer Beispielanwen dung getestet die XML Dokumente verarbeitet welche in der Datenbank gespeichert sind 8 CONTENTS Acknowledgements First of all I would like to thank my parents who made it possible for me to start studying computer sciences Hopefully now I can give something back and my mother has a reason to be a bit proud of her son A big Thank You also goes to my professor DI Dr techn Dieter Schmalstieg who offered me this project and always provided me with resources and people in no time thereby avoiding unnecessary delays in the progress of my work Furthermore I would like to thank Dr techn Gerhard Reitmayr for supporting me with ideas and suggestions that helped me designing the software and to solve a lot of arising implementation problems And in particular for his patience and understanding with my situation when I was indignant and impatient Special thanks and two big kisses for Magistra Elisabeth Lahner Altmann and Gudrun Wakolbinger for proofreading my work and helping me to get rid of many errors And last but not least a Thank You to all of my friends who supported me mentally during this work especially in hard times Special thanks go to Matth

Master`s Thesis

Contents

Download Pdf Manuals

Related Search

Related Contents