Building Blocks
Turning the Web Into a Data Source
By Leslie
Marable
What if the Web could
expose its information to any application in any way? This is the concept
behind webMethods Inc. of Fairfax,
Va., which has defined an eXtensible Markup Language (XML) that can directly
address any HTML document or set of documents as a data source.
WebMethods' Web Interface
Definition Language (WIDL) is at the heart of the webMethods Web Automation
Toolkit, a Java class creator that "reads" a document's structure,
transforms it into its constituent objects, and stores these objects in
a repository for use in any Java application.
These classes can
be called from existing business applications to get data from a Web site,
which lets the developer bypass the browser and treat the Web as a data
source. As a Java Bean, these classes can be incorporated into COM- or
CORBA-based applications through the Java-ActiveX bridge or other interobject
communication schema. Beans support is coming in a future release, according
to Charles Allen, vice president of business development at webMethods.
The Web Interface
Definition Language (WIDL) consists of six XML-compliant HTML extenders
that define a universal schema for HTML documents based on the Document
Object Model (DOM) as it is being defined by the World
Wide Web Consortium.
Currently, the Automation
Toolkit parses the HTML level supported by Microsoft
and Netscape 3.x on top of the W3C's
HTML 3.2 specification (what will be known as DOM level 0).
By dealing with HTML
documents as a set of abstract objects, the Automation Toolkit can extract
data from any place in any document at an extremely detailed level. For
instance, defining a news story as a headline, a byline, and body copy
organized as paragraphs, an application designed to monitor the Associated
Press wire could enumerate, manipulate, and extract the specific paragraphs
containing the word Zaire in any story by a specific reporter. It could
then pull that story, paragraph, or whatever into a local application.
The company added
native support for the Secure Sockets Layer 3.0 de facto standard earlier
this month, enabling secure transactions with any Web-based source from
nonbrowser applications.
Web developers working
for webMethods clients, including DHL Worldwide
Express, Texas Instruments, and the U.S. Postal Service, are building
applications used to automate the exchange of data from suppliers, business
partners, and customers.
Mark Lussier, the
lead software engineer for DHL's customer automation division, said he
used the technology to build a single-source package tracking facility.
"In the case of [package] tracking, the Web results that a user sees
are different depending on the current situation of the package. If it's
been delivered, there's one response, if the package is in transit, there's
a different response," he said. "The webMethods Toolkit allows
us to handle all those conditions with one piece of code, so we didn't
have to write code for each variation of the page." Ultimately, Lussier
said, the Toolkit saved him time and saved DHL money, because Netscape
priced its proposed solution for the same task at $27,000. "I did
the same thing in two hours for one-tenth the cost." A 30-day evaluation
version of the complete Web Automation Toolkit can be downloaded from the
company's Web site. A beta version of Web Automation Toolkit 2.0 is also
available. Individual developer licenses for the Toolkit begin at $995.
Server run-time licenses begin at $2,495. Server run-time licenses with
the SSL plug-in begin at $4,995.
The SSL plug-in requires
the Phaos Technology Corp. SSL libraries
and an encryption license from RSA Data Security
Inc.
Reprinted from Web Week, Volume 3, Issue 11, April
21, 1997 © Mecklermedia
Corp. All rights reserved. Keywords: development_tools
java Date: 19970421
