Yes. You can get all the attributes through the python bindings and implement extra filtering in your own plugin. See this PyQGIS Coobook excerpt for the rundown and some examples. You would then just exclude any nonmatching results from the returned dictionary.

As for the visualisation, you'll likely still have to create another layer, as select() does not have fitting arguments. You can use a memory layer to avoid having to create physical files (more on that in the cookbook).

Yes. You can get all the attributes through the python bindings and implement extra filtering in your own plugin. See this PyQGIS Coobook excerpt for the rundown and some examples. You would then just exclude any nonmatching results from the returned dictionary.

As for the visualisation, you'll likely still have to create another layer, as select() does not have fitting arguments. You can use a memory layer to avoid having to create physical files (more on that in the cookbook).

edit:

Actually, you can use selectedFeaturesIds() with setSelectedFeatures(ids) to change the selection to the subset you created. Quoting the implementation directly:

/** Get a copy of the user-selected features */  
QList<QgsFeature> selectedFeatures();

/** Return reference to identifiers of selected features */
const QSet<qint64> &selectedFeaturesIds() const;

/** Change selection to the new set of features */
void setSelectedFeatures(const QSet<qint64> &ids);