🚀 go-pugleaf

In article <Bzpxqy.C2E@watserv2.uwaterloo.ca> eric@csg.uwaterloo.ca writes:
 > In article <20180.2b34a24a@ul.ie> murraya@ul.ie writes:
 > >Do anybody out there know anything about an application to `automatically` 
                                                                ^^^^^^^^^^^^^
                                                                Be careful!

 > >generate markuped text, (possibly according to an sgml DTD) from an ordinary 
 > >ASCII file. 
 > > 
 > There is a product called OmniMark from Software Exoterica in Ottawa, Canada
 > which we have been using to translate text to SGML from a number of different
 > formats including ASCII and WordPerfect. We have found this to be an
 > excellent tool and have used it to translate thousands of pages of text.

While OmniMark is an excellent tool for translating text, it is 
far from "automatic". I've used OmniMark and its older brother, 
XTRAN, for several years, and am delighted with its performance 
and reliability. However, it does require a lot of care and feeding. 
The new release of OmniMark (v.2.0) is *very* fast and a joy to 
work with.

Another product, FastTAG from Avalanche Development Co in Boulder 
Colorado, is good at recognizing objects on a page, but takes 
some work to separate objects if its recognition engine 
determines that two objects look like one. FastTAG does not have 
a parser in it, but relies on the programmer to insert start and 
end tags as necessary. FastTAG is very good at recognizing 
tabular material. It is almost human in that respect, and the 
developers should be commended.

FastTAG employs a "Visual Recognition Engine". This allows the 
program to work with "fuzzy" specifications (e.g., heads contain 
60% capitalized words). It is also considerably cheaper than 
OmniMark ($1,500-2,500 vs. $15,000-25,000 -- these could be old 
prices).

In a recent project, I used OmniMark to recognize the basic 
structure of an ASCII input file, then fired off a FastTAG 
session to mark up tables when encountered. It worked quite well, 
and with a minimum of programming. Look for details in an 
upcoming issue of <TAG>.

Then there are the general-purpose and cheap alternatives: AWK, 
Perl, Lex, and the dreaded Snobol. Most of my clients recognize 
that the support provided by vendors selling a specialized tool 
more than outweighs the cost of the product, and select OmniMark 
or FastTAG (or both) as a strategic tool. There are other 
commercial products that do SGML translations, but I have not had 
much experience with them.

As a rule, in any translation system, the quality of the data 
coming out is only as good as the consistency (or regularity) of 
the data going in.

There is no magic.

--
Brian E. Travis                                              brian@sgml.com
SGML Architect,        Managing Editor,               Tele: +1 303 680-0875
InfoDesign Corp.       <TAG> The SGML Newsletter       Fax: +1 303 680-4906