First of all, thanks to all who sent me advice on the matter of normalizing
sgml via sgmls 1.1 output. I received a literal barrage of replies, and
what's more surprising, most of the suggestions were unique.
They fell on the whole into two groups, and the difference was, for me at
least, instructive. Several people sent me pointers to code or actual code
for doing what I had envisaged: reading the output of SGMLS, converting
it back into sgml, with the result that omitted tags are all included.
However, as several other people pointed out to me, relying on the ESIS
format of SGMLS's output has other, undesirable results. Entity references
are all expanded, and marked sections are all either ignored or included.
This makes it much more difficult to produce a portable document instance.
The other strategy suggested was to solve the problem without the helpful
hinderance of sgmls. I was pointed in the direction of several commercial
products which accomplish only the specific goal I had in mind: read a
document instance, supply all omitted tags, and write it. I will try
to investigate as many of these as I can, in the hope of hitting upon a
solution that merits the investment of both money and processing overhead.
Thanks again to all who replied.
- Gregory Murphy
CETH, the Center for Electronic Texts in the Humanities
e-mail: [log in to unmask]
phone: (609) 258-2460