The GLOSS-xml system, part 2

An additional feature of text processing in the GLOSS-xml system is the ability to push back in to element-mode from text-mode with a further [.

The idea is that, when typing a section of mixed data it is often convenient to think of simple markup such as <dfn>...</dfn> or <q>...</q> to be part of the text than as stopping the text, starting something else, then restarting the text again. So [ pushes back into element mode for a while, a subsequent ] pops back to text mode, and so on. The [ ] must balance, and elements cannot straddle [ ], so [ ] behaves a bit like { } as well.

XML doesn't have such a feature of nested modes, so GLOSS has to simulate it by ending text and then re-starting more text, but conceptually and structurally this feature of GLOSS-xml is very useful, easy to type, and remarkably legible.

For example,

which yields

The inner text can also have pushed element sections too, and so on. (Of course, care is needed that brackets match.) The form [i[kumquats]] is common when writing HTML for italicising text. It seems a little lengthy because of the double brackets (though a touch better than <i>kumquats</i>), but it turns out to be very quick to type in practice as the square bracket symbols are usually very conveniently located on the keyboard and much more easily accessed than < >.

To insert a comment in the input text, just use the ; sign. The comment lasts to the end of the line. Thus

yields

If you want your output XML file to contain comments, not just your input file, then clearly a different mechanism is required. Instead use the pseudo tag name ! just like you would an xml element name and make its content a text block. Thus:

yields

The sequence of characters -- is not allowed in comment text. If you try to do this your text is modified to something appropriate. Thus

yields

The encoding used currently is to replace any -- by &#x2D;&#x2D;.

CDATA sections also textual input as described above. To define a CDATA section use the pseudo tag name !CDATA in the same way as you used ! for comments. Thus:

yields

All the same rules apply as for comments. In particular just because the output will be CDATA doesn't mean the rule of escaping [, ], {, } and \ is relaxed. (If you think about it you will realise that they can't be.) But note that CDATA sections have different forbidden text. In the case of CDATA it is ]]> that may not appear in a CDATA section. This is handled as follows:

yields

LITERAL sections also use text as described above, and are provided as a last-ditch resort to place literal text in the output. Do not use these unless there really is no alternative! It works just like CDATA with the pseudo tag name !LITERAL.

yields the non-well-formed

We have seen that !CDATA, !LITERAL and even ! are useful as special elements. Any other elements of the form !name may be used too. In fact, the normal behavious is that these are converted to normal elements. This

yields

For other applications, xtending GLOSS-xml, it is useful to have two separate sets of elements. Thus in GLOSS-xsl the !name forms generate xsl commands, and normal forms generate normal elements. Similarly in GLOSS-mv the !name forms generate gloss mv commands.