29.4 Converting earleme to Earley set ID

In the dense input models, the earleme of every Earley set is equal to the ID of that Earley set. But, because of the sparse input models, this is not true in the general case.

For other applications that want an earleme-to-ID mapping, the most general method for converting earlemes to Earley set IDs is to create a ID-to-earleme array using the marpa_r_earleme() method (see marpa_r_earleme), and invert it. The resulting earleme-to-ID array will be sparse. More precisely, where the function earleme maps IDs to earlemes, the value at index ix of the array should be

• id, if earleme(id) = ix for some id; and
• undefined, otherwise.

The Libmarpa interface does not provide a method for earleme-to-ID conversion. The input model is not known at the Libmarpa level, so that an earleme-to-ID method would have to be fully general. A fully general earleme-to-ID method would be costly, an cost which usually would be unnecessary. For these reasons, it’s up to the application to create its own earleme-to-conversion method, if one is needed.

Here are the alternatives:

• Usually an earleme-to-ID conversion method is not needed.
• If an earleme-to-ID conversion method is needed, and the sparse input model is not in use, the Earley set ID and the earleme are always equal, and the earleme-to-ID conversion is trivial.
• If an earleme-to-ID conversion method is needed, and a sparse input model is in use, the application may be able to implement an efficient earleme-to-ID conversion targeted at that specific input model.
• If an earleme-to-ID conversion method is needed, a sparse input model is in use, and no efficient earleme-to-ID conversion targeted at that specific input model seems to be available, the general earleme-to-ID conversion technique described just above must be used.