Authoring SGML DTD's with GELLMU

William F. Hammond

Email: hammond@math.albany.edu

September 28, 1998

It is certainly not a new idea that SGML markup can be used to manage many things including, in particular, SGML DTD's. (For the moment, I am at a loss to find a publicly visible example.)

The extensible nature of GELLMU (Generalized Extensible LaTeX-Like MarkUp) is rooted in the fact that GELLMU may be used to author SGML documents.

It follows, inasmuch as GELLMU may be used to prepare an SGML document, that it may be used to prepare an SGML document that is used to generate an SGML DTD.

The following included verbatim segment is the text of a simple GELLMU document "prexdt.glm" that flows in one step, i.e., GELLMU stage 1, to "prexdt.sgml". It is followed by a copy of the DTD for a simple SGML document type that I have named ``dtd''. "prexdt.sgml" is a document under the ``dtd'' document type. With a small collection of stage 3 codelets for sgmlspl, one may then generate "prexdt.dtd", which is the third included verbatim segment below.

Of course, if this is to make any sense, it must follow that "boot.dtd" can itself be authored with the document type that it defines, and that is, indeed, the case.


prexdt.glm:


\documenttype[boot.dtd]{dtd}
\doctype{test}
\dtdref{prexdt.dtd}
\cmt{A Test DTD}
\cmt{Does the recommended usage match this?
  <!DOCTYPE test SYSTEM "prexdt.dtd">
}
\begin{document}
\elt{test}{oo}{(preamble,body)}
\elt{preamble}{rr}{(a)}
\elt{body}{rr}{((p|t)*)}{    BETTER}
\enp{ts}{a|p}
\elt{(\%ts)}{rr}{(#PCDATA)*}
\elt{t}{rr}{((r)+)}
\elt{r}{oo}{((c)+)}
\elt{s}{oo}{(i,(c)*)}
\elt{c}{oo}{(#PCDATA)*}
\elt{i}{oo}{(#PCDATA)+}
\end{document}

boot.dtd:


<!-- Use <!DOCTYPE dtd SYSTEM "boot.dtd"> -->
<!-- Use
     <!DOCTYPE dtd SYSTEM "boot.dtd">
 -->
<!ELEMENT doctype - - (#PCDATA) -- the name of the document type,
    i.e., the name of the root element (which needs to be defined) -->
<!ELEMENT dtdref - - (#PCDATA) -- the (unquoted) system resource name
   of the DTD to be used with documents of the type "doctype"; this
   name is only used to generate a comment.
   Example: for this DTD:  doctype = dtd   dtdref = boot.dtd -->
<!ENTITY % dtd "elt|enp|ent|empt|cmt">
<!ENTITY % tags "rr|ro|or|oo">
<!ENTITY % charr "spc|cma|per|ltc|gtc|qum|sol|col|scl|quo|lsq|rsq">
<!ENTITY % charn "lbr|rbr|exc|atc|hsh|dol|pct|crt|amp|ast|lpr|rpr|und">
<!ENTITY % chare "lsb|rsb|hyp|eqc|plu|vbr|bsl|tld">
<!ENTITY % chara "%charr|%charn|%chare">
<!ENTITY % charh "rdash">
<!ENTITY % chars "%chara|%charh|brk">
<!ENTITY % eos "eos|eoq|eoe">
<!ELEMENT (%eos) - O EMPTY -- ends of sentence made by GELLMU -->
<!ELEMENT (%chars) - O EMPTY -- 33 printable, non-alphanumeric -->
<!ELEMENT (%tags) - O EMPTY -- the tag requirement combinations -->
<!ENTITY % string "#PCDATA|%chars">
<!ELEMENT ag0 - - (%string|%eos)* -- command argument -->
<!ELEMENT dtd O O (preamble, body)>
<!ELEMENT preamble - - (doctype,dtdref,(cmt)*) -- zero or more comments -->
<!ELEMENT body - - (%dtd)*>
<!ELEMENT elt - O (ag0,ag0,ag0,(ag0)*) -- element:
             name, tags, defn, comments? -->
<!ELEMENT enp - O (ag0,ag0,(ag0)*) -- parameter entity:
             name, defn, comments? -->
<!ELEMENT ent - O (ag0,ag0,(ag0)*) -- general entity:
             name, defn, comments? -->
<!ELEMENT empt - O (ag0,(ag0)*) -- empty element:
             name, comments? -->
<!ELEMENT cmt - - ((%string|%eos|verb)*) -- comment -->
<!ELEMENT verb - - ((%string|%eos)*) -- inline verbatim, OK in cmt -->


prexdt.dtd:


<!-- Use <!DOCTYPE test SYSTEM "prexdt.dtd"> -->
<!ELEMENT test - - (#PCDATA) -- name of root tag>
<!-- A Test DTD -->
<!-- Does the recommended usage match this?
  <!DOCTYPE test SYSTEM "prexdt.dtd"> -->
<!ELEMENT test O O (preamble,body)>
<!ELEMENT preamble - - (a)>
<!ELEMENT body - - ((p|t)*) --     BETTER -->
<!ENTITY % ts "a|p">
<!ELEMENT (%ts) - - (#PCDATA)*>
<!ELEMENT t - - ((r)+)>
<!ELEMENT r O O ((c)+)>
<!ELEMENT s O O (i,(c)*)>
<!ELEMENT c O O (#PCDATA)*>
<!ELEMENT i O O (#PCDATA)+>


About the GELLMU verbatim command

Since GELLMU documents are amenable for processing to many ultimate target formats, including possible future formats, it is essential to make verbatim sections safe. This means that verbatim sections cannot be prepared in the way that a LaTeX ``verbatim environment'' is prepared. Even the HTML "pre" command -- for pre-formatted text -- requires the preparer to escape the characters `<', `>', and `&' where they occur in the pre-formatted material.

There are two reasonable ways to handle verbatim material in a GELLMU system:

  1. Prepare the material, as for HTML, for direct inclusion in the SGML image of a GELLMU document.

  2. Filter the material to use symbolic character names for any character that is not strictly alpha-numeric before inserting the material inside a GELLMU verbatim command.


About GELLMU

GELLMU is an acronym for ``Generalized Extensible LaTeX-like MarkUp''. As of October, 1998 all GELLMU materials are didactic, and there has been no ``release'' of related code. The GELLMU materials, under frequent revision, are available for examination. This document on the web is
http://www.albany.edu/%7Ehammond/gellmu/boot/dboot.html, and the durable GELLMU web entrance is
http://math.albany.edu:8000/math/pers/hammond/igl.html.


AUTHOR  |  COMMENT

[Processed from GELLMU to HTML: Sat 12 Aug 2000 at 18:10:39 EDT]