def testFilledExporter(self):
"""Export two FARs and check that they contain the right FSTs."""
filename_a = os.path.join(FLAGS.test_tmpdir, 'test_a.far')
filename_b = os.path.join(FLAGS.test_tmpdir, 'test_b.far')
FLAGS.outputs = 'a=' + filename_a + ',b=' + filename_b
with self.assertRaises(SystemExit):
multi_grm.run(generator_method)
stored_fsts_a = _read_fst_map(filename_a)
self.assertLen(stored_fsts_a, 1)
self.assertTrue(stored_fsts_a['FST1'])
stored_fsts_b = _read_fst_map(filename_b)
self.assertLen(stored_fsts_b, 2)
self.assertTrue(stored_fsts_b['FST2'])
self.assertTrue(stored_fsts_b['FST3'])
cluster_or_vowel_with_coda = pynini.union(
independent_vowel,
cluster_with_vowel) + (uf.QuesSafe(vowel_length_sign) +
uf.QuesSafe(coda) + uf.QuesSafe(dead_consonant))
akshara = pynini.union(cluster_or_vowel_with_coda, cluster_and_virama)
return uf.StarSafe(pynini.union(akshara, standalone, accept)).optimize()
def generator_main(exporter_map: multi_grm.ExporterMapping):
"""Generate unweighted FSAs accepting the language of each Brahmic script."""
for token_type in ('byte', 'utf8'):
with pynini.default_token_type(token_type):
exporter = exporter_map[token_type]
for script in u.SCRIPTS:
exporter[script.upper()] = accept_well_formed(
u.SCRIPT_DIR / script / 'script_config.textproto',
u.SCRIPT_DIR / script / 'consonant.tsv',
u.SCRIPT_DIR / script / 'dead_consonant.tsv',
u.SCRIPT_DIR / script / 'subjoined_consonant.tsv',
u.SCRIPT_DIR / script / 'vowel_sign.tsv',
u.SCRIPT_DIR / script / 'vowel.tsv',
u.SCRIPT_DIR / script / 'vowel_length_sign.tsv',
u.SCRIPT_DIR / script / 'coda.tsv', u.SCRIPT_DIR / script /
'standalone.tsv', u.SCRIPT_DIR / script / 'virama.tsv',
u.SCRIPT_DIR / script / 'accept.tsv',
u.SCRIPT_DIR / script / 'preserve.tsv')
if __name__ == '__main__':
multi_grm.run(generator_main)