NetNews Usenet Archive 1993 #1/NN_1993_1.iso/spool/comp/text/sgml/1263

Internet Message Format | 1993-01-06 | 5.3 KB

Path: sparky!uunet!ralvm13.VNET.IBM.COM
From: [e-mail]
Message-ID: <[e-mail]>
Date: Wed, 6 Jan 93 18:51:16 EST
Newsgroups: comp.text.sgml
Subject: Re: SGML for data querying
Disclaimer: This posting represents the poster's views, not those of IBM
News-Software: UReply 3.0
References: <[e-mail]>
Lines: 96

In <[e-mail]> "David Herron" <[e-mail]> writes:

>Taking the `forms' example.  A tag might be `<NAME>' and the value at
>that tag is the persons name.  Or a purchase requisition form might
>have <AMOUNT-REQUIRED> <ITEM-DESCRIPTION> <CATALOG-NUMBER> <UNIT-PRICE>
>and <PRICE> tags, with associated values.  Is there a concept wherein
>some `tags' have `values'?  Or does a tag only create a context, and the
>data within that context is interpreted in a particular way?

The SGML markup only defines a structural context, e.g., Ammount-Required
must precede Item-Description, or Name contains <lastname> followed
by <firstname>.  It is up to a particular processing application to
define what those elements mean, which they do by associating some
processing with the elements and their data.  This processing can be
anything you can program.  SGML neither defines nor constrains the
purpose to which you put SGML-encoded data.

Elements (tags are the syntactic constructs that delimit element
boundaries) can have values in a sense.
All elements (except one special case we can ignore for simplicity)
can be plain text (PCDATA or "parsed character data"), other
elements only, or elements and PCDATA ("mixed content").  When elements
only take PCDATA as their content, the PCDATA is generally intended
as that element's "value" in the way you mean.  Note that in
SGML, elements always contain things, they do not simply represent
start and end flags, and that containment is tracked and reported
by the SGML parser, and knowledge of that containment structure can
be used by applications to make decisions based on the containment
hierarchy for any element or data.

>In any case it appears that for `querying' type applications the
>SGML parser must build up some data structure, and that the application
>program must then be able to probe the data structure at its leisure.
>For a normal text processing application (SGML -> TeX -> PS conversion)
>you can get away with creating and destroying the data structures on
>the fly.  But to be consulted at the applications leisure the data
>structures must stay around for a long time.
>
>So... Are there any SGML programming toolkits available which allow
>for this sort of thing?  Is the ARCSGML/JClark suitable in this way?
>(I'm on a small budget)

Someone has provided a Rexx interface to ARCSGML for use with
Personal Rexx under MS-DOS.  With this, you could fairly easily
write "query" applications.  The basic idea is
to first define the data structures you need to hold the data
you want, just as you would for any other data processing
application.  You then process the document, grabbing data
off of the relevant elements as they go by and putting it into your data
structures.  The rest you just let flow through and ignore.

When you're done processing the document, you do whatever needs
to be done with the data you've collected, such as create
reports, load databases, etc.  This should be fairly straight-forward
processing.

<digression>

The examples above, and most SGML applications in existence today,
process SGML documents sequentially from start to finish.  However,
SGML does not require sequential processing, and it can make just
as much sense to define applications that work with SGML documents
wholistically as a tree, rather than sequentially, at
least in the abstract.  In this processing model, access is
via queries, rather than sequential access by waiting for the
element you want to flow by.  Some SGML-based online presentation
systems, such as EBT's Dynatext, present this sort of view of
documents, providing query-based direct access to elements and
and data, which makes sense in a dynamic, direct access application
like online presentation.

You can also define processing models that use both sequential
and query-based processing.  For example, you might use a query
to access the root of some subtree within a document, and then
process that subtree sequentially to build some output, or in
the course of sequential processing, you might use a query to
do lookahead to resolve a cross reference (as opposed to
doing some sort of two-pass process).  This sort of system might
make the most sense when your documents are already stored in
an SGML-knowledgeable database system that provides the retrieval
function you need.

The point is that even though bytehead hackers like myself who
knock together quick programs to do pragmatic stuff tend to
think about processing SGML sequentially, especially because
that's what our tools give us, SGML processing is not limited
to that one processing model, and that by thinking of SGML documents
as databases against which you can do queries, you might start
to think of very interesting and useful ways to use your
SGML documents you might not have thought of before.

Eliot Kimber                      Internet:  [e-mail]
Dept E14/B500                     IBMMAIL:   USIB2DK9@IBMMAIL
Network Programs Information Development     Phone: [phone]
IBM Corporation
Research Triangle Park, NC 27709