Dear Weiming,

You could use the following:

import nazca as nd

cells = nd.load_gds('input_abc.gds', asdict=True)

# Assume this file has top cells 'a', 'b' and 'c'.

# Only export 'a' and 'b'.
export_cells = [cells['a'], cells['b']]

nd.export_gds(export_cells, filename='export_ab.gds')

Please note that the name of the exported cell can be different when it was already in use when importing the gds file. The actual cell name is in e.g. cells['a'].cell_name

Xaveer