ctc_token_fst.py 707 B

123456789101112131415161718192021222324
  1. #!/usr/bin/env python
  2. # Apache 2.0
  3. import sys
  4. with open(sys.argv[1], 'r') as fread:
  5. print('0 0 <blank> <eps>')
  6. nodeX = 1
  7. for entry in fread.readlines():
  8. entry = entry.replace('\n', '').strip()
  9. fields = entry.split(' ')
  10. phone = fields[0]
  11. if phone == '<eps>' or phone == '<blank>':
  12. continue
  13. if '#' in phone:
  14. print(str(0) + ' ' + str(0) + ' ' + '<eps>' + ' ' + phone)
  15. else:
  16. print(str(0) + ' ' + str(nodeX) + ' ' + phone + ' ' + phone)
  17. print(str(nodeX) + ' ' + str(nodeX) + ' ' + phone + ' <eps>')
  18. print(str(nodeX) + ' ' + str(0) + ' ' + '<eps> <eps>')
  19. nodeX += 1
  20. print('0')