Remarkup prevents several classes of text-processing problems by replacing tokens in the text as they are marked up. For example, if you write something like this:

//D12//

It is processed in several stages. First the "D12" matches and is replaced with a token, in the form of "<0x01><ID number><literal "Z">". The first byte, "<0x01>" is a single byte with value 1 that marks a token. If this is token ID "444", the text may now look like this:

//<0x01>444Z//

Now the italics match and are replaced, using the next token ID:

<0x01>445Z

When processing completes, all the tokens are replaced with their final equivalents. For example, token 444 is evaluated to:

<a href="http://...">...</a>

Then token 445 is evaluated:

<em><0x01>444Z</em>

...and all tokens it contains are replaced:

<em><a href="http://...">...</a></em>

If we didn't do this, the italics rule could match the "//" in "http://", or any other number of processing mistakes could occur, some of which create security risks.

This class generates keys, and stores the map of keys to replacement text.