The API for the extractors is defined in src/tracker-extract/tracker-extract.h.

The module must define a TrackerExtractorData struct that links mime types with extraction functions. This is returned in a function called tracker_get_extractor_data.

The extraction function has the following signature:

void name_of_the_function (const gchar *filename, 
                           GHashTable  *metadata)
      

ANY extracted meta-data MUST be added to the hash table.