BA372 - XML & Web Services

• The Web for humans: HyperText Markup Language (HTML):
  • is a language for formatting documents.
  • sections of text are marked-up according to how they should be displayed:

    this should be in italics and bolded

    <i>this should be in italics <b>and bolded</b></i>

• But what about programs?
  • I am an on-line bookstore and I want to sell other stores' books (and get a commission).
  • I am an on-line newspaper and I want to show the NYSE and NASDAQ stock quotes on my Web site.
  • I am an on-line weather site and I want to simultaneously display the current temperatures in 10 large cities.
  • I am a mortgage broker and I need to know the current rates of my top 50 lenders.
  • I am the OSU library and I would like to show my patrons which books are available in the UofO's library.
  
  
  • I am a financial spreadsheet and I would like to know the value of my portfolio.
    
  • I am www.teachengineering.com and I would like to show my users how many NSDL documents have something to do with 'water'.
  • etc.
    

• HTML is no good for programmatic data retrieval:
  • Each site that I would need to visit provides its own, specially formatted and laid-out HTML pages.
  
  
  • Every time their HTML front-end changes, I (might) have to change my program parsing it.
    

  • I'd like to first collect all the information from multiple sites and then organize it on a single web page.
  • So I need some HTML, but only at the very end -- facing my customer.
    
  • How do I get the first bit (collecting all the information in the first place) done?

• eXtensible Markup Language (XML) to the rescue:
  • Like HTML, XML is a markup language.
  • Adopted as W3C standard in 1998.
    
  • However, instead of formatting instructions, it communicates contents.

  example: from Harold, E.R. & Means, W.S. (2001) XML in a Nutshell; O'Reilly & Associates

  <collection>
  
    <painting>
      <title>Memory of the Garden at Etten</title>
      <artist>Vincent van Gogh</artist>
      <date>Nov. 1888</date>
      <description>Two women look to the left. A third works in her garden</description>
    </painting>
  
    <painting>
      <title>The Swing</title>
      <artist>Pierre Auguste Renoir</artist>
      <date>1886</date>
      <description>Girl on a swing. Two men and a toddler watch</description>
    </painting>
  
  </collection>
  

• Notice how the markup tags convey meaning, yet say nothing about how this information must be displayed.
• Also notice how the concepts are hierarchical; i.e., 'nested:'
  • a collection can contain more than one painting.
  • titles and artists appear only within paintings.

• This strict hierarchy is easy for programs to 'parse.'
• An XML document or fragment can be represented as a tree (similar to a directory/folder tree).
• Problem: Can you draw the <collection> tree?
  

• Trees are 'nice' data structures to represent and parse.
• e.g., look at how your browser shows an arbitrary XML document. (vs. the document rendered as HTML).

• Document Object Model (DOM) is a W3C-governed standard for representing XML documents as 'trees.'
• An entire XML document can be read into a tree and stored in memory; e.g., as a DOM object.
  

• When the tree is properly constructed; i.e., as a tree,  the document is said to be well formed.
• Only well-formed documents can be reliably parsed ==> XML makes things very rigid & predictable ==>  easy on programs to parse and 'take apart.'
• Problem: what are some examples of a not well-formed XML tree?
• Problem: make an XML document, introduce some well formedness errors and try pick up the 'broken' document with your browser; what do you observe?

• Since XML (like HTML) is text-based, it is compatible across computing platforms ==> important implications for 'interoperability.'
  

• Problem: Who determines which 'tags' or 'elements' are available? What if I would like to create some XML to represent course student enrollment? Which XML elements are there for me to use?
• Answer: you make them up yourself!

  For example:

  <course name="Business Systems Design & Development">
  
    <abbreviation>BA372</abbreviation>
  
    <section number="001">
      <instructor>
         <first_name>Rene</first_name>
         <last_name>Reitsma<last_name>
      </instructor> 
  
      <student>
         <first_name>John</first_name>
         <last_name>Smith<last_name>
      </student> 
  
      <student>
         <first_name>Mary</first_name>
         <last_name>Jones<last_name>
      </student> 
    </section >
  
    <section number="002">
       ...
    </section >
  
  </course>
  

• Note how an element or 'node' such as <section> can have attributes, similar to attributes in HTML tags.
• Problem: When to make something an attribute rather than an element?

• But how does the receiving computer know which XML elements I made up? How, therefore, can it 'parse' my XML?
• Answer 1: If the document is well formed; a document tree can always be formed.
• However, what about more complex constraints?
• <student> must have both a <first_name> and a <last_name>.
• <first_name>s and <last_name>s are atomic; i.e., they must contain nothing but characters.
• A <section> has an attribute number.
  

• Answer 2: you have to provide a declaration of your syntax (grammar + vocabulary): Document Type Definition (DTD) or XML Schema (XSD).

• Notice that DTDs and XSDs themselves are XML documents. However!!! their syntax is fixed and predetermined!
• Look how Altova's XMLSpy visualizes an XML Schema (XSD) as a tree (not a public site).

• Problem: So how do the XML document and the DTD/XSD get associated with each other? How does the receiving computer know which DTD/XSD goes with which XML document?

• Internal DTD: DTD and XML stored in the same document:

• External DTD: XML document contains a reference to a DTD elsewhere on the Web.

• Examples of XML-based formats:
• MathML
  
• XBRL: Extensible Business Reporting Language
• Music(X)ML
• And a zillion others.
  

• So XML makes it 'easy' to represent contents... So what do we do with that?
• Store internal information; e.g.,
• TeachEngineering documents.
• Microsoft Office 2007 default document storage is XML.
• System and application configuration files (recall environment variables).
• Anything else we want to store in a well-formed format.

• Make systems interoperate with each other.

• Interoperability with XML & Web services.
• Traditional: Application X running on platform P cannot communicate with application Y running on platform Q.

• But we want...

• So how to make these apps talk to each other?
• We're a Microsoft shop; We're a UNIX/ORACLE shop.
• Middleware, or...

• Proprietary data exchange standard CORBA, COM, ActiveX, etc.)
• Web services: HTTP/XML-based data provider. Two types:

• REpresentational State Transfer (REST): HTTP response contains unadorned, unencapsulated, native XML. Request sending mixes HTTP and XML.
  
• Digital Library for Earth System Education (DLESE) search for the term "ocean" and return a maximum of 10 search results, starting at position 0: http://www.dlese.org/dds/services/ddsws1-1?verb=Search&q=ocean&s=0&n=10
• Yahoo REST search service: http://search.yahooapis.com/WebSearchService/V1/webSearch?appid=foo&query=eggplant
• Example application:
• TeachEngineering, CAT, ASN & NSDL.
• Notice how NSDL 'knows' about TeachEngineering curriculum.
• Reason: TeachEngineering exposes its metadata as XML to NSDL for harvesting.

• Notice how TeachEngineering 'knows' that some of its curriculum aligns with K-12 educational standards in CT.
• But the curriculum authors never aligned their curriculum with CT!

• Solution: All TeachEngineering curriculum is submitted to CAT for alignment.

• Notice how both CAT & TeachEngineering identify educational standards with S... identifiers.
• Reason, both CAT & TeachEngineering get their standards from the Achievement Standard Network (ASN) webservice.

• Notice how in both the DLESE and yahoo request the search strings are passed through the URL ==> not so nice:
• Message reply is pure XML, but request is a mixture of HTTP & XML.
• Whereas the full functionality of XML is available for sending the message reply, it is not for the message request.

• Service-Oriented (previously: Simple Object) Access Protocol (SOAP):
• XML specification for messaging (Web services) on the Web; governed by W3C.
• Provides a structure for Web-based messaging; both request & response:
• Can easily be wrapped inside HTTP.
  

• Often used in conjunction with Web Services Description Language (WSDL).
• XML specification for describing the programmatic interface to a (SOAP) Web service.
• Example from W3C's pages.
  
• e.g., Amazon's E-Commerce Service's WSDL.

• UDDI? (Universal Description and Discovery Integration):
• Web services registry/directory service (sponsored by OASIS).
• Applications can
  1. Search the UDDI for applicable services,
  2. then use their WSDLs to find out how to use them,
  3. then use SOAP to use them.

• Tragedy of metadata (Havenstein article).

• Overall: Web services are being considered as a general mechanism for interoperability:
• Ws-*: evolving set of best practices and standards related to Web services.
• WS-BPEL.

• Problem: We're moving to so-called Service-Oriented Architectures (SOA):

• Problem: Isn't this what OOP was supposed to give us?

• Problem: So, what's the difference? Why would XML-based Web services be able to accomplish what OOP did not?