Leveraging eXist-db for Efficient TEI Document
Contents
1. in the WebDav portal to eXist db This example configuration assumes the eXist db collection is called LBT The current system uses Jetty as the web server but eXist db can be con figured to work with Apache if needed However eXist db and the website must be hosted together because of cross domain security concerns These can be alleviated by leveraging domain agnostic HTTP request libraries in Ajax or YAHOO 4 The Lucene module provided by the eXist db installation provides all the power of full text searching offered by Lucene This module replaces the legacy searching module within eXist db for the release used in our project The same type of configuration for Lucene in a standalone application can be created within eXist db anything from boosting fields to custom analyzers and ranking specific results that were queried from the database We will see that we can combine our Lucene queries uses XQuery to generate search results that are based on rank and show custom views and explanations as to why a particular document was chosen by Lucene The WebDav and WebApp modules of eXist db allow us to easily manage documents within a particular database There is no need to remote into a machine and edit documents on the server By using the RESTful end points offered by eXist db developers and researchers can access and modify remote documents in a local environment Additionally by hosting the website within eXist db we can executes XQuery s2. markup used by Professor Radcliffe is below Moreover his reasoning as to lt persName gt Mr Moore s lt persName gt conduct with regard to lt persName gt Lord Byron s lt persName gt lt name type title key LdByron Memoir gt Memoirs lt name gt Seems to us to be at once vague and inapplicable In the archives maintained by Professor Radcliffe much of the TEI markup is references to a person s name as well as references to written works and correspondence In the above example you may notice that Lord Byron s name is placed in a TEI tag as well as a tag for work a work entitled Memoir When the TEI is processed the first two persName tags will not be processed into some sort of reference as they contain no key However the name tag will be converted into a footnote as it contains a key referencing that particular document In the existing system XPath is used to generate these footnotes In the proposed LORD BYRON AND HIS CONTEMPORARIES Lord Byron and some of his Contemporaries with Recollections of the Author s Life and of his Visit to Italy By Leigu Hunt 1 vol 4to pp 513 London 1828 Coulburn It is for slaves to lie and for freemen to speak truth Montaigne ABOUT as much of this work as is devoted to Lord Byron is assigned to all the other literary men together who are noticed by the author Of these papers the most important relate to Mr Moore Mr Shelley Mr Keats Mr Cha
3. particular see example_byron js for scripting information 4 3 Client side Templating HTML CSS We have discussed both modifying the content of the HTML and styling on the client side but how exactly are we going to do this The former LBT system defined styles in line using the standard HTML styles attribute However this method is not the best design For example if you reuse a specific style for many different elements such as a navigation item if you wish to change the color or font size you have to go through every single entry and change the font size for every styles attribute Using CSS classes we can eliminate this bottleneck by 10 defining a specific type of styles class and then use the class attribute to assign a style to a particular element within the HTML document This decreases the complexity and increases the readability of our code without losing contextual information for a particular element Examples scripts and files can be retrieved from VTechWorks In particular see example_byron css for styling information 4 4 XQuery By migrating to eXist db from flat XML files and discontinuing the use of XSL stylesheets we are required to use XQuery scripts to process and generate XML data from eXist db XQuery in and of itself is fairly similar to XPath which is used within XSL XQuery takes advantage of the power behind eXist db by directly accessing the index and retrieving documents as quickly as po
4. 3 5 2 eXist db Notes 2 2 14 VTechWorks Inventory o o e 020000004 14 COMICS A a E E IES id 15 Abstract Professor David Radcliffe has created Lord Byron and his Times LBT a large digital archive of works surrounding Lord Byron and his contemporaries The original website was unusable slow due to the expensive XSLT transformations and XPath queries being applied to TEI documents By removing the reliance on XSL and using eXist db and XQuery while relying on ubiquitous and well documented CSS for client side styling we have been able to increase perfor mance dramatically without sacrificing features or functionality In this paper we go over an overview of the project including challenges and potential so lutions and performance metrics of the original system in section 1 Section 2 contains a user s manual detailing difference between the old and proposed systems Section 3 contains a developer s manual which contains overviews of various technologies that were being used in the system designed by Professor Radcliffe The fourth section describes technologies relevant to the proposed system Finally documentation and installation instructions are given in the fifth section The rest of the paper contains a VTechWorks inventory contacts for everyone involved with the LBT project and references 1 Project Details This section outlines the details of the project including a brief description of the project s contrac
5. As the client receives data it is automatically displayed on the client s browser such that they can begin exploring the site without having to wait for the entire page to load While they are browsing the system will continue to load new data that it has fetched 1 7 Future Deliverables This section aims to detail future deliverables for the LBT project with respect to migration maintenance and future development The major deliverables are listed below and further discussion of each deliverable is available in Developer s Manual 1 Monograph migration migrate the display of documents to the new sys tem 2 Styling simplification eliminated hardcoded styles in the TEI XML doc uments by using the defined CSS 6l classes 3 Index consolidation eliminate redundant and unused index by creating a single index per category i e a person and document index 2 User s Manual This user s manual aims to provide an overview of the current LBT website 2 1 Overview The former LBT system is fairly straight forward website with a navigation bar on the left side that is context sensitive and a content pane that provides details of the currently selected navigation item The current LBT system aims to mimic the look and feel of the former to provide a consistent and unnoticeable transition for users T here are minor changes in the display especially for batch loading large documents 2 2 Screenshots L
6. Leveraging eXist db for Efficient TEI Document Management Kyle Morgan Kyle Schutt knmorgan vt edu kschutt vt edu Final project report for CS 5604 Fall 2012 Project Byron In collaboration with Professor David Radcliffe Purdom Lindblad and Keith Battleson Department of Computer Science Virginia Tech 11 December 2012 Contents aD Project Details ati FAR ALG GSS a A tee 1 1 1 Overview ojo hb Bed Erb OA Se la Be ee g A a 1 12 Management comerse a o a 1 1 3 Challenges 4 20 2a a A A aA 2 1 4 SOLUTIONS 2D e A A A ie eat 2 1 5 Performance Metrics o o 2 1 6 Development a 3 1 7 Future Deliverables 3 User s Manual A toe BR ee a 4 2 1 OVERVIEW iia Beh BA oR Rood BB la as ute Sand 4 22 Sere nshots zs i ee bh OB ae a gh OAM 4 Developer s Manual 02 0000002 eee eee 6 3 1 EBI OWerview ee Boy Sok SY Boca den e ee oe 6 3 2 DEA A Se a We Bh ASR ths Dee Bape te On A 6 3 3 KSLA XSL ss Ain tes 6 ae bee te Ba Sees 7 3 4 Current Architecture 7 3 5 Proposed Architecture o o e 8 Technologiesin doses ed A DA A E A t 9 4 1 eXist db with Lucene 9 4 2 YAHOO Connection Manager and jQuery 10 4 3 Client side Templating HTML CSS 10 4 4 RXQUErY 2 SPN ES Pa AO ee 11 4 5 Example Migration o 11 Installation and Documentation 13 5 1 eXist db Installation 0 0 0004 1
7. Manager On the server side eXist db provides several points of access through its web portal and WebDav endpoints Additionally it provides a powerful XQuery processing engine that allows us to obtain data through the eXist db and Lucene indexes Lucene indexes the documents using the same collection definition that we used to index in eXist db Lucene provides us with a robust search engine that allows us to parse and index each individual TEI document based on specific parameters The new architecture allows us to perform operations asynchronously and in parallel to help speed up computation and retrieval of documents Web Browser vied Bary Index l eXist db wene Client side Apache HTTP Server Figure 7 Proposed architecture minimizing reliance on XSLT and XPath 4 Technologies 4 1 eXist db with Lucene Our implementation of eXist db and Lucene is available on VTechWorks see section below for more details The version used in our project is eXist db version 1 4 2 Build 20120416 on an Amazon EC2 Enterprise Red Hat Linux machine Small Instance l eXist db is an XML based database management system The current in stallation of eXist provides several different technologies straight of the box including Lucene WebDav XQuery processing WebApp development collec tion management and documentation The eXist db and Lucene indexes use the same collection definition at system config db LBT collection xconf
8. Nov Jetty WebServer Online http jetty codehaus org jetty SyncRO Soft SRL 2012 Nov XSLT Editor Online http oxygenxml com xml_editor xslt_editor html XML Path Language 2012 Oct XPath Documentation Online http www w3 org TR xpath The eXist Project 2012 Oct XQuery Documentation Online http exist db org exist xquery xml TEI Text Encoding Initiative 2012 Oct TEI Text Encoding Initiative Online http www tei c org index xml w3 org 2012 Oct The Extensible Stylesheet Language Family XSL Online http www w3 org Style XSL w3 org 2012 Oct XSL Transformations XSLT Online http www w3 org TR xslt SAXON 2012 Dec The XSLT and XQuery Processor Online http saxon sourceforge net The PHP Group 2012 PHP Hypertext Preprocessor Online http php net WSC 2012 W3C HTML Online http www w3 org html W3C 2012 Cascading Style Sheets Online http www w3 org Style CSS Center for Applied Technologies in the HUmanities 2012 Welcome to CATH Online http wiz2 cath vt edu David H Radcliffe 2012 The Life and Times of Lord Byron Online http lordbyron cath lib vt edu 16
9. ORD BYRON gt m TIMES Lor LORD BYRON m TIMES LORI d d Documents Biography Criticism Fetq Documents Biography Criticism INTRODUCTION Docume INTRODUCTION DOCUMEN CONTENTS aai CONTENTS SEARCH PERSONS mat Le dE Hi i ABOUT LBT SEARCH PERSONS Fiant Lei Clark d ADOUTLBT Hid LETTERS BY DATE a Clarke q DOCUMENTS BY DATE Choker LETTERS BY DATE wh SEARCH TITLE FILE a DOCUMENTS BY DATE Glen PERIODICALS Hunt Ld SEARCH TITLE FILE a ANONYMOUS WORKS sd af me PERIODICALS y BY THE NUMBERS sd Hunt Leig ANONYMOUS WORKS S Hunt Ld e al Figure 2 Before and after respectively of the Project Byron website Note the similarity in styling LORD BYRON wo TIMES SIR WALTER SCOTT BARONET Fetching page 9 of 29 1771 1832 HOME CORRESPONDENCE NAME AUTHORITIES REFERENCES TO PERSON LOC Scott Walter Sir 1771 1832 DDAN REFERENCES TO WORKS NRA Scott Sir Walter 1771 1832 1st Baronet poet and novelist nee eel SEARCH PERSONS DNB Scott Sir Walter 1771 1832 Ist Baronet poet and novelist ae LBT ID WaScott VIAF ID 95207079 LOC ID 278095541 SOURCES DNB Prance Companion to Charles Lamb 1983 Virtual International Authorit thePeerage com LOC Name ty File Scottish poet novelist antiquary biographer editor and sheriff 0909 CORRESPONDENCE OF SIR WALTER SCOTT BARONET Licensed under a Creative Walter Scott to William Clerk 1788 c in Memoirs of the Life of Commons Attribution Ciu Walt
10. cripts easily while taking advantage of eXist db s caching structure for commonly viewed pages and queries The eXist db installation can be retrieved from VTech Works 4 2 YAHOO Connection Manager and jQuery We have discussed a lot about batch loading and the caching system within eXist db but in order to access that system we must perform asynchronous HTTP requests to our XQuery scripts while modifying the HTML document The YAHOO Connection Manager YCM provides us the ability to perform asynchronous requests to any endpoint within our domain The main reasons we chose YCM over Ajax was because it is a smaller library more efficient with resources and executed slightly faster that basic Ajax request Additionally YCM has greater cross browser support and cross domain executions are avail able although unnecessary for this project YCM works within JavaScript and we use Query BI to make modification to the HTML DOM object when we receive new data By performing these modifications on the client side we can offload some of the computation and styling from the server side The for mer LBT system performed all modifications on the server side which tied up resources from performing necessary server side functions like processing ad ditional requests Modern web browsers have more than enough processing power to modify and apply styling and data to web pages on the fly Example scripts and files can be retrieved from VTechWorks In
11. d experience for Radcliffe s clients The remainder of this document aims to detail these goals the methods used and future deliverables for the project 1 2 Management The management team involves three types of personnel stakeholders soft ware engineers and collaborators The primary and solitary stakeholder of the project is Professor David Radcliffe curator of LBT under the Center for Ap plied Technologies in the Humanities CATH at the Virginia Polytechnic Institute and State University VT The software engineers are Kyle Morgan and Kyle Schutt both members of Dr Edward Fox s CS5604 Information Storage and Retrieval class in the Fall of 2012 Dr Fox Purdom Lindblad and Keith Battleson are the main collaborators who provided guidance in the areas of information design architectural design and encouragement We would like to thank all of those involved for their support throughout this project 1 3 Challenges The two main challenges of our project was both understanding and quanti fying the relationships between the current indexes and their corresponding documents and properly indexing existing documents such that the efficiency of Lucene and eXist db is maximized Some of the current indexes are redun dant and bloated with unnecessary text these indexes needed to be identified and minimized before being re indexed into eXist This re indexing and reclas sification of existing indexes also required changes i
12. ecty tei persNa G ms 2 36 1 gt Ah xsi for each select document AlTitesindex xm tei bibl tei author key xselectx 180 ms 1 45 1 b xst value of select selectPerson tei persName type LOC 88 ms 0 72 1 gt Ah xsh if test SselectPerson tel persame Gtype NRA 61 ms 0 49 1 b A xshapply templates select tei tite 57 ms 0 46 174 gt Ah div style margin eft 40px margin top Spx margin bottom 10px 45 ms 0 36 174 Qs test document allindex xm teieader descendant tei author Ckey xselectd 43 ms 0 35 1 Stack 8 Output Mapping Stack fy Trace Z Templates sx Nodes Values Set 4 Hotspots Figure 1 Profiling example of an XSL document using oXygen 1 6 Development With the results of the performance metrics in mind we were able to focus on specific areas that required tuning Re indexing and substituting a flat file system of an XML database system was not going to solve all the issues in performance While an XML database system provides certain improvements in terms of caching and in memory fetching for common documents it does not address the problem of transferring large amount of data to a client over the web Therefore we implemented an asynchronous batch loading system that stores data within an in memory session variable on the server Additionally the client side template system fetches the data asynchronously from the server
13. ed This gives the impression to the user that the website itself is slow and may be slow to the point that it is unusable even if it is working properly In Figure 4 we can see an example of a long running function of fetching data from a single document In this case we are fetching data from a very large XML document in a very large XSL document which has already been tuned In the modern era of high speed Internet connections a response of more than ten seconds from a web server is unacceptable With this in mind we aimed to tackle this problem on two fronts minimize the time it takes to display some content on the web page and minimize the time it takes to load all the data asynchronously 4 K 82 14 10218 ms 174 inv div style margin eft 40px margin top Spx margin bottom 1 4 K 82 14 10218 ms 174inv div style margin eft 80px margin top 15px margin botto 4 K 82 14 10218 ms 174 inv xsl for each select document AlTitiesIndex xml tei bib 4 K 82 14 10218 ms 174 inv xsl when test xchoosex PubRefs 4 K 82 14 10218 ms 174 inv xsl choose K 82 14 10218 ms 174 inv td colspan 2 Ah xsl variable name selectPerson select document persons xmi tei person xmisid xs J 697 ms 5 6 1 gt xshif test SselectPerson teispersName type LOCT i554 ms 4 45 1 gt A xshvalue of select document AllPersindex xml tei person Gxml id xsel
14. ing the 18th and 19th centuries The website is a great tool for illustrating connections between people When viewing a letter of correspondence between two people LBT generates footnotes for references to other people or works from within the document A user may click on a person s name for instance to view a footnote containing more information about that person One may search for references to a certain person in other documents or find references to works authored by that person The LBT website as it originally existed was slow for popular people That is people with a large number of references or who have authored a large number of works Finding references to Sir Walter Scott for instance takes roughly 15 seconds on the original LBT site This is simply too slow especially when considering scalability 3 2 TEI The Text Encoding Initiative TEI is a standard for maintaining digital archives of text l TEI is largely used in the humanities for maintenance of digital archives and is used extensively throughout the LBT project TEI is an XML format containing several hundred tags that can be used for marking up text with varying levels of granularity It should be noted that TEI is not a presen tational markup language That is it is not sufficient markup for presentation in say a web browser Some computation is required to translate a TEI XML format into XHTML which will be displayed within a browser An example of TEI
15. ion There are predefined functions that allow you to access certain types of data that are popular to ascertain Please see below for a full listing a The first function is simply fetching all the data returning a header and a results element tag and storing all the items in memory on the server b The first part of the function does not require a request parameter Once we have the data we can generate our header div with the class set to document_list as above This header will also contain the number of hits or pages that were generated there are 20 items per page Secondly we need to handle when a request comes in for a particular page number a We go out and fetch the page from memory and generate a list of the items stored for the page based on the details mentioned in Step 5 b We apply the document title style class to each item and return the XML document So far we have only worked on the server side let s take a look at the client side We need to execute the JavaScript function that will kick off our XQuery script function fetchDocumentsByDate loadData exist rest db LBT queries documents xq true false The JavaScript function loadData will take care of iterating over each page found and displaying in contents That is pretty much all you need to know about migrating from XSL to XQuery and JavaScript 12 LORD BYRON ms TIMES d Docume
16. lection Index Definition for Lucene and eXist db indexing example_documents xq XQuery for fetching documents by date example_index html Generic HTML file for displaying information about Project Byron This Document Final report for the Project Byron group 14 7 Contacts Name Email Expertise Role David Radcliffe drad vt edu Lord Byron Stakeholder Kyle Morgan knmorgan vt edu Software Engineering Graduate Student Kyle Schutt kschutt vt edu Software Engineering Graduate Student Purdom Lindblad purdom6 vt edu Library Services Collaborator Keith Battleson keith120vt edu Information Technology Collaborator Edward Fox fox vt edu Information Storage Retrieval Collaborator 15 References 1 10 11 12 13 14 15 16 17 18 eXist db 2012 Nov eXist db Open Source Native XML Database On line http exist db org exist index xml Michael McCandless Erik Hatcher and Otis Gospodnetic Lucene in Ac tion Safari Books Online Manning Publications 2010 The Query Foundation 2012 Nov Query The Write Less Do More JavaScript Library Online http jquery com YAHOO Developer Network 2012 Nov YUI 2 Connection Manager Online http developer yahoo com yui connection Amazon 2012 Nov Amazon Web Services LLC Online http aws amazon com ec2 Mort Bay Consulting 2012
17. n the underlying document retrieval system that uses XQuery instead of XPath 1 4 Solutions In order to address these challenges we focused on maintaining a working knowl edge of the current indexes and focused on modifying these indexes prior to implementing the retrieval system This reduced the maintenance and develop ment costs associated with refactoring existing code Indexing in Lucene and eXist is fairly consistent for the majority of time but it is possible to cause a massive increase in the index size when the definition is incorrect The failure to index properly caused a large amount of downstream performance degradation when fetching documents Therefore the index definition currently used has been tuned over several iterations to the point where we feel that it is quite stable 1 5 Performance Metrics Prior to analysis there was much speculation over which parts of the system were causing performance degradation such as the TEI XML documents XML indexes or the XSLT Processor After performing analyses and profiling on a subset of large TEI XML documents and XSL documents it was determined that the major bottleneck for processing time was within the XSL documents and therefore the XSLT Processor Additionally XSLT processing is a syn chronous blocking function that causes the browser to wait until it is complete This function provides little information to the user about its current action or when it will be complet
18. nt back to the web browser for viewing In the above pipeline XPath is used to grab references and generate foot notes Professor Radcliffe had hand built a number of XML indexes for the XSL TEI XSLT Processor Web Browser Apache HTTP Server Figure 6 Existing architecture built by Professor Radcliffe purpose of looking up someone s full name the works associated with them their birth and death date etc As discussed previously we found the XSLT processing to be far too slow Moreover we felt that maintaining a number of hand built indexes would be far too cumbersome and scales poorly once more data is accumulated Our goal was thus to propose a new system that removes the XSLT processor from the pipeline and attempt to streamline it with XQuery We also wanted to use XQuery along with eXist to reduce the reliance on multiple indexes 3 5 Proposed Architecture In this section we discuss the proposed architecture which was implemented as part of the final project As previously stated we have removed the XSLT processor and moved all templating and styling to the client side Figure 7 shows how styling is done primarily on the client side with CSS Specific CSS classes still need to be attributed to specific XML elements in eXist when a desired effect is required The web pages are displayed using HTML and data is fetched from the eXist db using asynchronous JavaScript HTTP calls using YAHOO s Connection
19. nts Biography Criticism INTRODUCTION CONTENTS SEARCH PERSONS ABOUT LBT LETTERS BY DATE DOCUMENTS BY DATE SEARCH TITLE FILE PERIODICALS ANONYMOUS WORKS BY THE NUMBERS LorD BYRON AND HIS TIMES General Indexes Fetching page 9 of 17 DOCUMENTS IN LORD BYRON AND HIS TIMES Hunt Leigh 1784 1859 Memoir of Mr James Henry Leigh Hunt Written by Himself Monthly Magazine London 1810 04 Vol NS 7 Pp 243 48 Clarke Charles Cowden 1787 1877 An Address to that Quarterly Reviewer London R Jennings1816 Croker John Wilson 1780 1857 Leigh Hunt s Rimini Quarterly Review London 1816 01 Vol 15 Issue 28 Pp 475 81 Hunt Leigh 1784 1859 Distressing Circumstances in High Life Byron s Separation The Examiner London 1816 04 21 Issue 434 Pp 247 50 Hunt Leigh 1784 1859 Distressing Circumstances in High Life Byron s Figure 8 Current LBT System fetching all documents by date using batch loading 5 Installation and Documentation This section aims to provide an overview of installing operating and maintain the eXist db and its associated modules Additionally this sections aims to provide a high level documentation of some of the available functions for XQuery and JavaScript 5 1 eXist db Installation There are two methods for installing eXist db custom or prebuilt Custom Install 1 Follow the QuickStart guide at http exist db org exist quickstart xml for your platform 2 C
20. on Conte Daut Tarata Dahari Madali 1027 Figure 3 Batch loading the Correspondence to and from Sir Walter Scott LORD BYRON so x TIMES Byron SUBMIT s Select from the alphabetical list above or perform a string search whole or partial words forename and sumame forename forename sumame or forename sumame and title etc Searches are cAsE SensitivE Documents Biography Criticism RESULTS 38 INTRODUCTION 1 Bettesworth George Edmund Byron 1780 1808 Captain Bettesworth of the Royal CONTENTS Navy mamied Lady Hannah Althea Whitbread daughter of General Charles Grey SEARCH PERSONS first Earl Grey He was killed at Bergen 25 May 1808 while in command of the Tartar ABOUTLBT He was a descendant of the Travanion family Marchand Sophia Trevanion had married Admiral John Byron LETTERS BY DATE 2 kK N 1852 Y 7 7 DOCUMENTS BY DATE 2 King n e Byron Ada countess of Lovelace 1815 1852 Byron s daughter by Lady Byron in 1835 she married William King eighth Baron King of Ockham 1805 1893 Figure 4 Using Lucene to search and score documents for the term Byron in all indexed documents 3 Developer s Manual 3 1 LBT Overview Professor Radcliffe has been maintaining Lord Byron and his Times LBT in conjunction with the Center for Applied Technology in the Humanities CATH for many years The LBT website contains a large digital archive of various works predominantly written dur
21. onnect to the eXist db WebDav endpoint http lt host gt exist webdav db Create a new database folder called LBT Copy LBT XML Files and Queries from VTechWorks into the LBT folder Create a folder under system config db called LBT oe Oh A oe Copy the collection xconf file into this new folder this sets up the in dexing scheme for eXist db and Lucene An example is located in VTech Works under example_collection xconf 13 7 8 Copy the web site data into the WebApp folder of the eXist db install folder Go to Start Jetty Prebuilt Install l 2 Download the LBT folder from VTech Works Go to Start Jetty Start Jetty 1 2 3 4 5 2 Navigate to the eXist db install location Execute the command export EXIST_HOME lt install location gt Execute the command bin startup sh amp Enjoy eXist db Notes Occasionally the environment variable EXIST_HOME gets reset please be mindful that you may need to set EXIST_HOME 6 VIechWorks Inventory This section outlines the files and inventory loaded on the Byron VTechWorks collection l 2 ProjByron zip the new website files for Project Byron exist tar the full install of eXist db for Project Byron with all files example byron css CSS stylesheet for Byron items divs and elements example byron js JavsScript for accessing XQuery documents on eXist db and other Query functions example collection xconf Col
22. rles Lamb and Mr Coleridge Of Mr Moore there is a very lively pleasant and characteristic description Mr Hunt s anecdotes about the writer of Lalla Rookh Figure 5 An example of what a marked up document looks like system XQuery will be used instead Ultimately the TEI will be processed and displayed in a web browser in a pleasing format which will be seen in the next section 3 3 XSL and XSLT Technically XSL is a language for expressing stylesheets that consists of XPath and XSLT as subcomponents 2 _ XPath is a query language used to ac cess certain parts of XML documents and is used extensively throughout the LBT project to generate references between works and people XSLT or XSL Transformations is a language used to transform XML documents into other XML documents XSLT is used to transform TEI documents which are XML documents as discussed previously into XHTML documents to be displayed in a web browser Figure 5 is an example of an XHTML document generated using TEI XSLT and XPath 3 4 Current Architecture In this section we will describe the architecture designed in implemented by Professor Radcliffe which can be viewed in figure 6 In this architecture a web browser requests a page from an Apache web server A PHP page then fetches the relevant XSL stylesheets XPath queries and TEI documents and the XSLT processor SAXON in this case l 3l then generates an XHTML document This document is se
23. ssible It also allows us to store variables in memory during execution which enables us to perform batch loading on large documents Examples scripts and files can be retrieved from VTechWorks In particular see example_documents xq for XQuery information 4 5 Example Migration This section aims to provide a real world migration example by modifying a section of XSL in the former LBT system for the current LBT system This example focuses on transforming the Documents by Date listing into the new architecture design 1 The first step is to locate the index xsl document and open it in your favorite text editor 2 Locate the tag where we test if the input is AllDocs approximately line number 709 a This is where the index xsl generates the output 3 We need to separate all styling from this so make note the two style attributes for the outermost div tags a There is also another style attribute for each of the items in the list 4 We can add these two styles to our CSS definitions document_title Y margin left 60px text indent 40px document_list 11 10 11 font variant small caps font size 20px letter spacing 1em margin top 250px margin bottom 20px Next we must fetch all the data required by the XSL stylesheet by creating an XQuery script This can include the title all subtitles bibliographic information and publisher informat
24. t between the stakeholder and the developers This section aims to provide high level managerial information on the project s scope challenges solutions and future deliverables 1 1 Overview The project proposed by Professor David Radcliffe was to create a new informa tion server for the current Lord Byron and His Times LBT system with a focus on an architectural design that can accommodate an increasing number of TEL XMLl documents in the repository The new system must be capable of indexing TEI XML documents for quick and effective retrieval of intercon nected documents and persons The system must be scalable and flexible to accommodate new indexes or documents and updates to existing documents The main goals of the project are broken down into two categories perfor mance metrics and development In order to determine a proper architecture recommendation a performance analysis of the former system was necessary to identify any weaknesses The second goal was development specifically the de velopment of a new XML based database system with indexing full text search ing modules and proper templating in the browser Later in this document each goal will be discussed in more detail The overall architectural design envisioned by Radcliffe was one focused on increased performance scalability and flexibility By focusing on the goals mentioned above this project addresses each of these concerns in turn to provide an enhance
Download Pdf Manuals
Related Search