fffiloni commited on
Commit
43bfecb
·
verified ·
1 Parent(s): 6d1acb0

Deploy ZONOS3 Mini ZeroGPU integration v0.1.0 (part 2)

Browse files
This view is limited to 50 files because it contains too many changes.   See raw diff
Files changed (50) hide show
  1. vendor/zonos3/vendor/nemo_text_processing/text_normalization/en/data/whitelist/tts.tsv +3890 -0
  2. vendor/zonos3/vendor/nemo_text_processing/text_normalization/en/graph_utils.py +337 -0
  3. vendor/zonos3/vendor/nemo_text_processing/text_normalization/en/taggers/__init__.py +13 -0
  4. vendor/zonos3/vendor/nemo_text_processing/text_normalization/en/taggers/abbreviation.py +51 -0
  5. vendor/zonos3/vendor/nemo_text_processing/text_normalization/en/taggers/cardinal.py +151 -0
  6. vendor/zonos3/vendor/nemo_text_processing/text_normalization/en/taggers/date.py +386 -0
  7. vendor/zonos3/vendor/nemo_text_processing/text_normalization/en/taggers/decimal.py +130 -0
  8. vendor/zonos3/vendor/nemo_text_processing/text_normalization/en/taggers/electronic.py +179 -0
  9. vendor/zonos3/vendor/nemo_text_processing/text_normalization/en/taggers/fraction.py +56 -0
  10. vendor/zonos3/vendor/nemo_text_processing/text_normalization/en/taggers/measure.py +332 -0
  11. vendor/zonos3/vendor/nemo_text_processing/text_normalization/en/taggers/money.py +199 -0
  12. vendor/zonos3/vendor/nemo_text_processing/text_normalization/en/taggers/ordinal.py +62 -0
  13. vendor/zonos3/vendor/nemo_text_processing/text_normalization/en/taggers/punctuation.py +66 -0
  14. vendor/zonos3/vendor/nemo_text_processing/text_normalization/en/taggers/range.py +128 -0
  15. vendor/zonos3/vendor/nemo_text_processing/text_normalization/en/taggers/roman.py +115 -0
  16. vendor/zonos3/vendor/nemo_text_processing/text_normalization/en/taggers/serial.py +143 -0
  17. vendor/zonos3/vendor/nemo_text_processing/text_normalization/en/taggers/telephone.py +134 -0
  18. vendor/zonos3/vendor/nemo_text_processing/text_normalization/en/taggers/time.py +133 -0
  19. vendor/zonos3/vendor/nemo_text_processing/text_normalization/en/taggers/tokenize_and_classify.py +241 -0
  20. vendor/zonos3/vendor/nemo_text_processing/text_normalization/en/taggers/tokenize_and_classify_lm.py +227 -0
  21. vendor/zonos3/vendor/nemo_text_processing/text_normalization/en/taggers/tokenize_and_classify_with_audio.py +229 -0
  22. vendor/zonos3/vendor/nemo_text_processing/text_normalization/en/taggers/whitelist.py +153 -0
  23. vendor/zonos3/vendor/nemo_text_processing/text_normalization/en/taggers/word.py +91 -0
  24. vendor/zonos3/vendor/nemo_text_processing/text_normalization/en/utils.py +60 -0
  25. vendor/zonos3/vendor/nemo_text_processing/text_normalization/en/verbalizers/__init__.py +13 -0
  26. vendor/zonos3/vendor/nemo_text_processing/text_normalization/en/verbalizers/abbreviation.py +36 -0
  27. vendor/zonos3/vendor/nemo_text_processing/text_normalization/en/verbalizers/cardinal.py +48 -0
  28. vendor/zonos3/vendor/nemo_text_processing/text_normalization/en/verbalizers/date.py +107 -0
  29. vendor/zonos3/vendor/nemo_text_processing/text_normalization/en/verbalizers/decimal.py +85 -0
  30. vendor/zonos3/vendor/nemo_text_processing/text_normalization/en/verbalizers/electronic.py +122 -0
  31. vendor/zonos3/vendor/nemo_text_processing/text_normalization/en/verbalizers/fraction.py +89 -0
  32. vendor/zonos3/vendor/nemo_text_processing/text_normalization/en/verbalizers/measure.py +122 -0
  33. vendor/zonos3/vendor/nemo_text_processing/text_normalization/en/verbalizers/money.py +85 -0
  34. vendor/zonos3/vendor/nemo_text_processing/text_normalization/en/verbalizers/ordinal.py +57 -0
  35. vendor/zonos3/vendor/nemo_text_processing/text_normalization/en/verbalizers/post_processing.py +182 -0
  36. vendor/zonos3/vendor/nemo_text_processing/text_normalization/en/verbalizers/roman.py +69 -0
  37. vendor/zonos3/vendor/nemo_text_processing/text_normalization/en/verbalizers/telephone.py +64 -0
  38. vendor/zonos3/vendor/nemo_text_processing/text_normalization/en/verbalizers/time.py +103 -0
  39. vendor/zonos3/vendor/nemo_text_processing/text_normalization/en/verbalizers/verbalize.py +82 -0
  40. vendor/zonos3/vendor/nemo_text_processing/text_normalization/en/verbalizers/verbalize_final.py +75 -0
  41. vendor/zonos3/vendor/nemo_text_processing/text_normalization/en/verbalizers/whitelist.py +40 -0
  42. vendor/zonos3/vendor/nemo_text_processing/text_normalization/en/verbalizers/word.py +36 -0
  43. vendor/zonos3/vendor/nemo_text_processing/text_normalization/es/README.md +22 -0
  44. vendor/zonos3/vendor/nemo_text_processing/text_normalization/es/__init__.py +15 -0
  45. vendor/zonos3/vendor/nemo_text_processing/text_normalization/es/data/__init__.py +13 -0
  46. vendor/zonos3/vendor/nemo_text_processing/text_normalization/es/data/dates/__init__.py +13 -0
  47. vendor/zonos3/vendor/nemo_text_processing/text_normalization/es/data/dates/days_abbr.tsv +27 -0
  48. vendor/zonos3/vendor/nemo_text_processing/text_normalization/es/data/dates/months.tsv +12 -0
  49. vendor/zonos3/vendor/nemo_text_processing/text_normalization/es/data/dates/months_abbr.tsv +44 -0
  50. vendor/zonos3/vendor/nemo_text_processing/text_normalization/es/data/dates/year_suffix.tsv +51 -0
vendor/zonos3/vendor/nemo_text_processing/text_normalization/en/data/whitelist/tts.tsv ADDED
@@ -0,0 +1,3890 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ Ph.D. PHD
2
+ Hon. honorable
3
+ Mt. Mount
4
+ Maj. Major
5
+ Rev. Reverend
6
+ w/o without
7
+ a/c number account number
8
+ c/o care of
9
+ Gov. governor
10
+ vs. versus
11
+ vs versus
12
+ dept. department
13
+ prof. professor
14
+ est. estimated
15
+ vol volume
16
+ vol. volume
17
+ bldg. building
18
+ Bldg. Building
19
+ apt. apartment
20
+ Apt. Apartment
21
+ World War I World War one
22
+ World War II World War two
23
+ etc. etcetera.
24
+ SnO2 tin four oxide
25
+ dept department
26
+ HVAC H-vac
27
+ SPDR spider
28
+ ZIP zip
29
+ ~ approximately
30
+ κ kappa
31
+ ω omega
32
+ α alpha
33
+ ν nu
34
+ δ delta
35
+ Δ delta
36
+ Α alpha
37
+ β beta
38
+ Β beta
39
+ χ chi
40
+ Χ chi
41
+ ε epsilon
42
+ Ε epsilon
43
+ φ phi
44
+ Φ phi
45
+ γ gamma
46
+ Γ gamma
47
+ η eta
48
+ Η eta
49
+ ι iota
50
+ Ι iota
51
+ κ kappa
52
+ Κ kappa
53
+ λ lambda
54
+ Λ lambda
55
+ μ mu
56
+ Μ mu
57
+ ν nu
58
+ Ν nu
59
+ ο omicron
60
+ Ο omicron
61
+ π pi
62
+ Π pi
63
+ θ theta
64
+ Θ theta
65
+ ρ rho
66
+ Ρ rho
67
+ σ sigma
68
+ Σ sigma
69
+ τ tau
70
+ Τ tau
71
+ υ upsilon
72
+ Υ upsilon
73
+ ω omega
74
+ Ω omega
75
+ ξ xi
76
+ Ξ xi
77
+ ψ psi
78
+ Ψ psi
79
+ ζ zeta
80
+ Ζ zeta
81
+ ltd limited
82
+ int'l international
83
+ A. D AD
84
+ A.D AD
85
+ a. d AD
86
+ a.d AD
87
+ a. d. AD
88
+ a.d. AD
89
+ B. C BC
90
+ B.C BC
91
+ b. c BC
92
+ b.c BC
93
+ A. D. AD
94
+ A.D. AD
95
+ B. C. BC
96
+ B.C. BC
97
+ b. c. BC
98
+ b.c. BC
99
+ A. A. a a
100
+ A.A. AA
101
+ A&A AANDA
102
+ A100 AONEHUNDRED
103
+ AAAs AAA's
104
+ AaB AAB
105
+ Aabo AABO
106
+ aac AAC
107
+ AACNo AACNO
108
+ AADTs AADT's
109
+ Aadu AADU
110
+ AAFld AAFLD
111
+ Aap AAP
112
+ A. B. AB
113
+ A.B. AB
114
+ Abhi ABHI
115
+ abl ABL
116
+ Abl ABL
117
+ Abp ABP
118
+ Abr ABR
119
+ abv ABV
120
+ Abz ABZ
121
+ A&C AANDC
122
+ A. C. AC
123
+ A.C. AC
124
+ A/C AC
125
+ acac ACAC
126
+ Acad ACAD
127
+ ACC&S ACC and S
128
+ Achs ACH's
129
+ Acis ACI's
130
+ acls ACLS
131
+ ACMs ACM's
132
+ AcpA ACPA
133
+ AcpB ACPB
134
+ acq ACQ
135
+ A&CR AANDCR
136
+ AcSOC ACSOC
137
+ acu ACU
138
+ AC&W AC and W
139
+ Acy ACY
140
+ A&D AANDD
141
+ Adab ADAB
142
+ A. D. AD
143
+ A.D. AD
144
+ A.D.A.M. adam
145
+ AdaSL ADASL
146
+ AdC ADC
147
+ ADCs ADC's
148
+ AD&D AD and D
149
+ AdK ADK
150
+ Adl ADL
151
+ ADLs ADL's
152
+ adm ADM
153
+ Adm ADM
154
+ admn ADMN
155
+ ADNs ADN's
156
+ AdP ADP
157
+ ADRs ADR's
158
+ ADSRs ADSR's
159
+ ADSs ADS's
160
+ Adwa ADWA
161
+ A&E AANDE
162
+ A. E. AE
163
+ A.E. AE
164
+ Aed AED
165
+ aegte AEGTE
166
+ Aeka AEKA
167
+ ael AEL
168
+ Aema AEMA
169
+ aen AEN
170
+ aere AERE
171
+ AEre AERE
172
+ AErn AERN
173
+ Aert AERT
174
+ Aes AE's
175
+ aet AET
176
+ Afa AFA
177
+ af AF
178
+ Af AF
179
+ A. F. AF
180
+ A.F. AF
181
+ Afaf AFAF
182
+ afc AFC
183
+ AFCSThe AFCSTHE
184
+ Afd AFD
185
+ AfD AFD
186
+ AfDB AFDB
187
+ Afe AFE
188
+ Aff AFF
189
+ Afi AFI
190
+ Afiq AFIQ
191
+ AFMs AFM's
192
+ AfNS AFNS
193
+ Afo AFO
194
+ Afr AFR
195
+ A. G. AG
196
+ A.G. AG
197
+ Agco AGCO
198
+ Agda AGDA
199
+ AG&G AG and G
200
+ agg AGG
201
+ Agi AGI
202
+ Agis AGI's
203
+ agli AGLI
204
+ Agn AGN
205
+ AgN AGN
206
+ AGNs AGN's
207
+ AGRs AGR's
208
+ Agsu AGSU
209
+ agt AGT
210
+ Agte AGTE
211
+ Agu AGU
212
+ A. H. AH
213
+ A.H. AH
214
+ Ahdhra AHDHRA
215
+ ahe AHE
216
+ Ahe AHE
217
+ Ahirs AHIR's
218
+ Ahk AHK
219
+ Ahsa'i AHSAI
220
+ Ahta AHTA
221
+ aht AHT
222
+ Ahu AHU
223
+ Ahva AHVA
224
+ A&I AANDI
225
+ A. I. AI
226
+ A.I. AI
227
+ Aias AIA's
228
+ aici AICI
229
+ Aicme AICME
230
+ AICs AIC's
231
+ Aiel AIEL
232
+ Aife AIFE
233
+ AiG AIG
234
+ Aigis AIGI's
235
+ Aija AIJA
236
+ Aik AIK
237
+ AIs AI's
238
+ Aist AIST
239
+ Aitu AITU
240
+ Aiud AIUD
241
+ aius AIUS
242
+ A. J. AJ
243
+ A.J. AJ
244
+ Ajba AJBA
245
+ Ajmi AJMI
246
+ Ajok AJOK
247
+ Ajsa AJSA
248
+ Aju AJU
249
+ Akaa AKAA
250
+ a.k.a. AKA
251
+ aka. AKA
252
+ aka AKA
253
+ Aka. AKA
254
+ Aka AKA
255
+ A&K AANDK
256
+ A. K. AK
257
+ A.K. AK
258
+ AKAPs AKAP's
259
+ akas AKA's
260
+ AKAs AKA's
261
+ akh AKH
262
+ Akpa AKPA
263
+ Aku AKU
264
+ A&L AANDL
265
+ A. L. AL
266
+ A.L. AL
267
+ Alh ALH
268
+ ALUs ALU's
269
+ A'ma AMA
270
+ A&M AANDM
271
+ a.m. AM
272
+ a.m AM
273
+ A. M. AM
274
+ A.M. AM
275
+ AMAs AMA's
276
+ amb AMB
277
+ Amb AMB
278
+ AmBX AMBX
279
+ AMCs AMC's
280
+ Amda AMDA
281
+ Amd AMD
282
+ AM&FM AM and FM
283
+ Amha AMHA
284
+ aml AML
285
+ Amli AMLI
286
+ Amlwch AMLWCH
287
+ Amm AMM
288
+ AM&O AM and O
289
+ Ampt AMPT
290
+ amr AMR
291
+ A&M's AANDM's
292
+ Ams' AM's
293
+ amsl AMSL
294
+ Amta AMTA
295
+ amt AMT
296
+ Amt AMT
297
+ Amts AMT's
298
+ Amu AMU
299
+ A. N. AN
300
+ A.N. AN
301
+ andFHL ANDFHL
302
+ andPPP ANDPPP
303
+ Anej ANEJ
304
+ Angti ANGTI
305
+ Angu ANGU
306
+ Anhe ANHE
307
+ Anr ANR
308
+ Anrep ANREP
309
+ A&O AANDO
310
+ A. O. AO
311
+ A.O. AO
312
+ AOCCs AOCC's
313
+ Aodh AODH
314
+ Aodla AODLA
315
+ Aogo AOGO
316
+ Aoke AOKE
317
+ Aoko AOKO
318
+ Aola AOLA
319
+ Aol AOL
320
+ Aone AONE
321
+ Aoni AONI
322
+ Aoos AOO's
323
+ aor AOR
324
+ AORs AOR's
325
+ aos AOS
326
+ AoS AOS
327
+ Aotus AOTU's
328
+ aov AOV
329
+ aovf AOVF
330
+ A&P AANDP
331
+ A. P. AP
332
+ A.P. AP
333
+ Apc APC
334
+ APCh APCH
335
+ APCs APC's
336
+ apg APG
337
+ Aph APH
338
+ Apiao APIAO
339
+ Api API
340
+ APi API
341
+ Apic APIC
342
+ APICv APICV
343
+ Apl APL
344
+ Aplu APLU
345
+ Apphttp APPHTTP
346
+ Appl APPL
347
+ appr APPR
348
+ Appts APPT's
349
+ appu APPU
350
+ Appu APPU
351
+ A&P's AANDP's
352
+ Aps AP's
353
+ APs AP's
354
+ AP&T AP and T
355
+ Aqa AQA
356
+ Aql AQL
357
+ Aqr AQR
358
+ A&R AANDR
359
+ A. R. AR
360
+ A.R. AR
361
+ ArgR ARGR
362
+ ARGs ARG's
363
+ ArH ARH
364
+ ARMv ARMV
365
+ Arnd ARND
366
+ arv ARV
367
+ Arv ARV
368
+ Arwi ARWI
369
+ As'ad ASAD
370
+ Asai ASAI
371
+ Asao ASAO
372
+ A.s A's
373
+ A. S. AS
374
+ A.S. AS
375
+ Asasp ASASP
376
+ asbl ASBL
377
+ asci ASCI
378
+ Asci ASCI
379
+ Asdis ASDI's
380
+ Asfi ASFI
381
+ Asn ASN
382
+ Aso ASO
383
+ Aspe ASPE
384
+ Asr ASR
385
+ Assn ASSN
386
+ assoc ASSOC
387
+ Assoc ASSOC
388
+ ASSPs ASSP's
389
+ Asst ASST
390
+ Aste ASTE
391
+ Astt ASTT
392
+ Aswa ASWA
393
+ A&T AANDT
394
+ atac ATAC
395
+ Atac ATAC
396
+ Atad ATAD
397
+ Ata's ATA's
398
+ Atas ATA's
399
+ A. T. AT
400
+ A.T. AT
401
+ ATBs ATB's
402
+ atc ATC
403
+ atcc ATCC
404
+ Atg ATG
405
+ ath ATH
406
+ Ath ATH
407
+ Atia's ATIA's
408
+ Ati ATI
409
+ ATi ATI
410
+ Atid ATID
411
+ Atiiq ATIIQ
412
+ Atil ATIL
413
+ Atiq ATIQ
414
+ Ativ ATIV
415
+ Atka ATKA
416
+ Atl ATL
417
+ Atli ATLI
418
+ atm ATM
419
+ Atm ATM
420
+ ATMs ATM's
421
+ Atos ATO's
422
+ ATOs ATO's
423
+ atpB ATPB
424
+ atri ATRI
425
+ Atri ATRI
426
+ A&T's AANDT's
427
+ A.T.s AT's
428
+ AT&SF AT and SF
429
+ AT&T ATANDT
430
+ attd ATTD
431
+ AT&T's ATANDTS
432
+ atv ATV
433
+ ATVs ATV's
434
+ AtxA ATXA
435
+ A. U. AU
436
+ A.U. AU
437
+ Aub AUB
438
+ Aubl AUBL
439
+ AUVs AUV's
440
+ Auw AUW
441
+ avab AVAB
442
+ A. V. AV
443
+ A.V. AV
444
+ AVCs AVC's
445
+ avg AVG
446
+ Avn AVN
447
+ Avo AVO
448
+ AvP AVP
449
+ AVRs AVR's
450
+ Avs' AV's
451
+ Avs AV's
452
+ avvo AVVO
453
+ Awa AWA
454
+ AWAs AWA's
455
+ A. W. AW
456
+ A.W. AW
457
+ Awo AWO
458
+ Awwa AWWA
459
+ A. X. AX
460
+ A. Y. AY
461
+ A.Y. AY
462
+ A. Z. AZ
463
+ A.Z. AZ
464
+ Azg AZG
465
+ Azi AZI
466
+ azm AZM
467
+ AzTV AZTV
468
+ Baad BAAD
469
+ Ba'al BAAL
470
+ Baal BAAL
471
+ Baam BAAM
472
+ Baap BAAP
473
+ ba'as BAA's
474
+ Baat BAAT
475
+ Ba'ath BAATH
476
+ B. A. BA
477
+ B.A. BA
478
+ B&A B and A
479
+ Bac BAC
480
+ bae BAE
481
+ Bae BAE
482
+ BAe BAE
483
+ Bauw BAUW
484
+ B'Av BAV
485
+ B&B B and B
486
+ bb BB
487
+ B. B. BB
488
+ B.B. BB
489
+ bbc BBC
490
+ BB&CI BB and CI
491
+ BBC&PJR BBC and PJR
492
+ BBQ'er BBQER
493
+ B&Bs B and B's
494
+ BB&S BB and S
495
+ BBSes BBSE's
496
+ BB&T BB and T
497
+ BbvCI BBVCI
498
+ bc BC
499
+ B. C. BC
500
+ B.C. BC
501
+ Bcci BCCI
502
+ B.C.'s BC's
503
+ BCs BC's
504
+ BCSic BCSIC
505
+ BdA BDA
506
+ bd BD
507
+ B. D. BD
508
+ B.D. BD
509
+ Bde BDE
510
+ bds BDS
511
+ BdU BDU
512
+ B. E. BE
513
+ B.E. BE
514
+ Bedw BEDW
515
+ bef BEF
516
+ bei BEI
517
+ Bei BEI
518
+ Beih BEIH
519
+ Beis BEI's
520
+ bej BEJ
521
+ BEMs BEM's
522
+ Ber BER
523
+ Bes BE's
524
+ bf BF
525
+ B. F. BF
526
+ B.F. BF
527
+ BFFs BFF's
528
+ bfy BFY
529
+ B. G. BG
530
+ B.G. BG
531
+ BglII BGLII
532
+ bgs BGS
533
+ Bgy BGY
534
+ Bha BHA
535
+ B&H B and H
536
+ B. H. BH
537
+ B.H. BH
538
+ BH&E BH and E
539
+ Bhe BHE
540
+ BHMs BHM's
541
+ bhttp BHTTP
542
+ B. I. BI
543
+ B.I. BI
544
+ Bie BIE
545
+ biedt BIEDT
546
+ Biem BIEM
547
+ B.I.G.'s BIG's
548
+ BiH BIH
549
+ B. J. BJ
550
+ B.J. BJ
551
+ BJPs BJP's
552
+ B.J.'s BJ's
553
+ B&K B and K
554
+ B. K. BK
555
+ B.K. BK
556
+ BKCa BKCA
557
+ B. L. BL
558
+ B.L. BL
559
+ ble BLE
560
+ bleg BLEG
561
+ blev BLEV
562
+ bli BLI
563
+ Blla BLLA
564
+ BLR&D BLR and D
565
+ BLTs BLT's
566
+ B&M B and M
567
+ B. M. BM
568
+ B.M. BM
569
+ bmd BMD
570
+ bmi BMI
571
+ BMPs BMP's
572
+ BMRs BMR's
573
+ bmt BMT
574
+ B&N B and N
575
+ bnb BNB
576
+ B. N. BN
577
+ B.N. BN
578
+ Boac BOAC
579
+ B&O B and O
580
+ B. O. BO
581
+ B.O. BO
582
+ Boc BOC
583
+ BoC BOC
584
+ BOCs BOC's
585
+ boj BOJ
586
+ Boj BOJ
587
+ B&O's B and O's
588
+ bp BP
589
+ B. P. BP
590
+ B.P. BP
591
+ bpl BPL
592
+ B&Q B and Q
593
+ B. Q. BQ
594
+ Brbic BRBIC
595
+ B. R. BR
596
+ B.R. BR
597
+ BRBs BRB's
598
+ Brne BRNE
599
+ Brno BRNO
600
+ BRTs BRT's
601
+ Brza BRZA
602
+ brzu BRZU
603
+ brzy BRZY
604
+ Brzyk BRZYK
605
+ BsaL BSAL
606
+ B. s B's
607
+ B.'s B's
608
+ Bs B's
609
+ B. S. BS
610
+ B.S. BS
611
+ bsd BSD
612
+ bt BT
613
+ B. T. BT
614
+ B.T. BT
615
+ Btry BTRY
616
+ btsan BTSAN
617
+ btus BTUS
618
+ BTUs BTU's
619
+ B.U. BU
620
+ B. V. BV
621
+ B.V. BV
622
+ Bwa BWA
623
+ B&W B and W
624
+ B. W. BW
625
+ B.W. BW
626
+ BWTs BWT's
627
+ bwwtv BWWTV
628
+ Bxa BXA
629
+ Bxe BXE
630
+ B. Y. BY
631
+ byc BYC
632
+ bygd BYGD
633
+ Byk BYK
634
+ byn BYN
635
+ Byo BYO
636
+ Byou BYOU
637
+ Byrl BYRL
638
+ bySLC BYSLC
639
+ byt BYT
640
+ B.Z. BZ
641
+ bzhed BZHED
642
+ Bzik BZIK
643
+ Bzyb BZYB
644
+ Caat CAAT
645
+ C. A. CA
646
+ C.A. CA
647
+ C&A CANDA
648
+ CA&CC CA and CC
649
+ Cadw CADW
650
+ caeca CAECA
651
+ Cae CAE
652
+ C.A.M.'s CAM's
653
+ Capt. captain
654
+ cas CAS
655
+ Cas CA's
656
+ Casc CASC
657
+ cb CB
658
+ C. B. CB
659
+ C.B. CB
660
+ cbc CBC
661
+ CB&CNS CB and CNS
662
+ CBDs CBD's
663
+ CBGBs CBGB's
664
+ CbiXS CBIXS
665
+ CB&Q CB and Q
666
+ CBs CB's
667
+ 'c C
668
+ CcaA CCAA
669
+ cca CCA
670
+ Cca CCA
671
+ C&C C and C
672
+ C. C. CC
673
+ C.C. CC
674
+ ccd CCD
675
+ CCDev CCDEV
676
+ CCDs CCD's
677
+ CCEd CCED
678
+ CCi CCI
679
+ ccm CCM
680
+ CCRCs CCRC's
681
+ cct CCT
682
+ ccus CCUS
683
+ C&D C and D
684
+ cdc CDC
685
+ cd CD
686
+ C. D. CD
687
+ C.D. CD
688
+ CD&DR CD and DR
689
+ CDi CDI
690
+ cDNA CDNA
691
+ cdnas CDNAS
692
+ cDNAs CDNA's
693
+ CDOs CDO's
694
+ CDPs CDP's
695
+ CD&R CD and R
696
+ cds CDS
697
+ CDs CD's
698
+ CDSPCo CDSPCO
699
+ CdtA CDTA
700
+ CdTe CDTE
701
+ CDTi CDTI
702
+ CD&V CD and V
703
+ Cec CEC
704
+ c'e CE
705
+ ce CE
706
+ C'e CE
707
+ Ce CE
708
+ C. E. CE
709
+ C.E. CE
710
+ Ceel CEEL
711
+ Cefn CEFN
712
+ C&EI's C and EI's
713
+ C&EN C and EN
714
+ cen CEN
715
+ Cen CEN
716
+ ceo CEO
717
+ Ceol CEOL
718
+ Ceorl CEORL
719
+ Ceos CEO's
720
+ CEOs' CEO's
721
+ CEOs CEO's
722
+ cer CER
723
+ Cer CER
724
+ ces CES
725
+ Ces CE's
726
+ Cex CEX
727
+ CeX CEX
728
+ Cfa CFA
729
+ C&F C and F
730
+ cf CF
731
+ C. F. CF
732
+ C.F. CF
733
+ CFOs CFO's
734
+ CFTs CFT's
735
+ C&G C and G
736
+ C. G. CG
737
+ C.G. CG
738
+ cgl CGL
739
+ C&H C and H
740
+ C. H. CH
741
+ C.H. CH
742
+ Chhu CHHU
743
+ Chirs CHIR's
744
+ CHP'den CHPDEN
745
+ CHPs CHP's
746
+ Chy CHY
747
+ Cia CIA
748
+ ci CI
749
+ Ci CI
750
+ C. I. CI
751
+ C.I. CI
752
+ Cicic CICIC
753
+ cii CII
754
+ cim CIM
755
+ Cim CIM
756
+ Cio CIO
757
+ CIOs CIO's
758
+ Cip CIP
759
+ cis CIS
760
+ CIs CI's
761
+ cit CIT
762
+ CiTD CITD
763
+ citS CITS
764
+ CitX CITX
765
+ Ciuc CIUC
766
+ Ciu CIU
767
+ CiU CIU
768
+ Cixi CIXI
769
+ C. J. CJ
770
+ C.J. CJ
771
+ C. K. CK
772
+ C.K. CK
773
+ ckx CKX
774
+ CLBs CLB's
775
+ clc CLC
776
+ cl CL
777
+ C. L. CL
778
+ C.L. CL
779
+ Cle CLE
780
+ cllr CLLR
781
+ CL&N CL and N
782
+ Clo CLO
783
+ CL&P CL and P
784
+ CLs CL's
785
+ CLTs CLT's
786
+ Clu CLU
787
+ Clwyd CLWYD
788
+ C&MA C and MA
789
+ cm CM
790
+ C. M. CM
791
+ C.M. CM
792
+ CMCs CMC's
793
+ Cmde CMDE
794
+ cmdr CMDR
795
+ cml CML
796
+ CMMs CMM's
797
+ cmn CMN
798
+ Cmte CMTE
799
+ cn CN
800
+ C. N. CN
801
+ C.N. CN
802
+ CNMs CNM's
803
+ cnr CNR
804
+ CNs CN's
805
+ CNTs CNT's
806
+ Cnut's CNUT's
807
+ CNVs CNV's
808
+ C&O C and O
809
+ C. O. CO
810
+ C.O. CO
811
+ Co. company
812
+ Col. colonel
813
+ C&P C and P
814
+ cp CP
815
+ C. P. CP
816
+ C.P. CP
817
+ CPCs CPC's
818
+ cpo CPO
819
+ CPUs CPU's
820
+ cq CQ
821
+ CQDs CQD's
822
+ cr CR
823
+ C. R. CR
824
+ C.R. CR
825
+ Cre CRE
826
+ crk CRK
827
+ Crkva CRKVA
828
+ crkve CRKVE
829
+ CRLs CRL's
830
+ crm CRM
831
+ Crne CRNE
832
+ Crni CRNI
833
+ Crno CRNO
834
+ CRs CR's
835
+ csa CSA
836
+ Csa CSA
837
+ Csak CSAK
838
+ Csaky CSAKY
839
+ C&S C and S
840
+ cs CS
841
+ C.s C's
842
+ Cs C's
843
+ C. S. CS
844
+ C.S. CS
845
+ C&SF C and SF
846
+ csg CSG
847
+ CsI CSI
848
+ CSis CSI's
849
+ CSIs CSI's
850
+ CSOs CSO's
851
+ cspA CSPA
852
+ csp CSP
853
+ CSPs CSP's
854
+ CSSNCs CSSNC's
855
+ Csuz CSUZ
856
+ CTAs CTA's
857
+ C&T C and T
858
+ cte CTE
859
+ CTLs CTL's
860
+ ctr CTR
861
+ Ctrip CTRIP
862
+ ctv CTV
863
+ cu CU
864
+ C. U. CU
865
+ Cuyp CUYP
866
+ C&V C and V
867
+ cvcp CVCP
868
+ 'cv CV
869
+ cv CV
870
+ C. V. CV
871
+ C.V. CV
872
+ CVEs CVE's
873
+ CVOs CVO's
874
+ CVs CV's
875
+ CVTs CVT's
876
+ C. W. CW
877
+ C.W. CW
878
+ Cwej CWEJ
879
+ Cwele CWELE
880
+ cwm CWM
881
+ cwrt CWRT
882
+ C.X. CX
883
+ C. Y. CY
884
+ C.Y. CY
885
+ cyl CYL
886
+ Cyn CYN
887
+ CyP CYP
888
+ Cyrl CYRL
889
+ Cys CY's
890
+ czci CZCI
891
+ cz CZ
892
+ C.Z. CZ
893
+ czy CZY
894
+ Daai DAAI
895
+ D. A. DA
896
+ D.A. DA
897
+ D&AD DANDAD
898
+ D.A.N.C.E. dance
899
+ Dav DAV
900
+ dBa DBA
901
+ db DB
902
+ D. B. DB
903
+ D.B. DB
904
+ DCCs DCC's
905
+ D&C D and C
906
+ dc DC
907
+ D. C. DC
908
+ D.C. DC
909
+ DCom DCOM
910
+ D.C.'s DC's
911
+ DCs DC's
912
+ DCUs DCU's
913
+ D&D D and D
914
+ D. D. DD
915
+ D.D. DD
916
+ DDi DDI
917
+ Ddoc DDOC
918
+ Ddu DDU
919
+ Deah DEAH
920
+ D. E. DE
921
+ D.E. DE
922
+ deg DEG
923
+ dei DEI
924
+ Dei DEI
925
+ DeI DEI
926
+ Deijm DEIJM
927
+ Deip DEIP
928
+ dez DEZ
929
+ Dez DEZ
930
+ Dfa DFA
931
+ DFCs DFC's
932
+ D. F. DF
933
+ D.F. DF
934
+ DfE DFE
935
+ DfES DFES
936
+ DFPs DFP's
937
+ DFs DF's
938
+ dfx DFX
939
+ D&G D and G
940
+ D. G. DG
941
+ D.G. DG
942
+ Dha DHA
943
+ DHBs DHB's
944
+ D&H D and H
945
+ dh DH
946
+ D. H. DH
947
+ D.H. DH
948
+ d'HEC DHEC
949
+ dhe DHE
950
+ Dheku DHEKU
951
+ Dhod DHOD
952
+ dhol DHOL
953
+ Dhol DHOL
954
+ Dhoo DHOO
955
+ Dhor DHOR
956
+ Dhou DHOU
957
+ dhr DHR
958
+ DHSs DHS's
959
+ D.I. DI
960
+ diy DIY
961
+ DiY DIY
962
+ Diyn DIYN
963
+ DIYs DIY's
964
+ Djam DJAM
965
+ dj DJ
966
+ D. J. DJ
967
+ D.J. DJ
968
+ DJ'ed DJED
969
+ Djem DJEM
970
+ Dji DJI
971
+ Djo DJO
972
+ D'Jok DJOK
973
+ djr DJR
974
+ D.J.s DJ's
975
+ DJs DJ's
976
+ DjVu DJVU
977
+ DjVus DJVU's
978
+ D. K. DK
979
+ D.K. DK
980
+ dla DLA
981
+ dl DL
982
+ D. L. DL
983
+ D.L. DL
984
+ dlia DLIA
985
+ DL&LR DL and LR
986
+ DLLs DLL's
987
+ dlo DLO
988
+ DL&W DL and W
989
+ D.M.C.'s DMC's
990
+ D&M D and M
991
+ D. M. DM
992
+ D.M. DM
993
+ DMed DMED
994
+ DM&E DM and E
995
+ DMs DM's
996
+ DMUs DMU's
997
+ Dna DNA
998
+ DnA DNA
999
+ DnaJ DNAJ
1000
+ D. N. DN
1001
+ D.N. DN
1002
+ Dnepr DNEPR
1003
+ dnes DNES
1004
+ Dnes DNE's
1005
+ DNFs DNF's
1006
+ DNSBLs DNSBL's
1007
+ dns DNS
1008
+ D.O.A.'s DOA's
1009
+ dod DOD
1010
+ Dod DOD
1011
+ DoD DOD
1012
+ D. O. DO
1013
+ D.O. DO
1014
+ D'oh DOH
1015
+ dok DOK
1016
+ Dok DOK
1017
+ domt DOMT
1018
+ Domt DOMT
1019
+ Dop DOP
1020
+ dotCMS DOTCMS
1021
+ DotSVN DOTSVN
1022
+ Douw DOUW
1023
+ dozd DOZD
1024
+ d'OZ DOZ
1025
+ Dozhd DOZHD
1026
+ dp DP
1027
+ D. P. DP
1028
+ D.P. DP
1029
+ dpon DPON
1030
+ DProf DPROF
1031
+ DPs DP's
1032
+ D.Q. DQ
1033
+ drc DRC
1034
+ Dr. doctor
1035
+ D. R. DR
1036
+ D.R. DR
1037
+ dri DRI
1038
+ Dri DRI
1039
+ DRIs DRI's
1040
+ Driu DRIU
1041
+ Driv DRIV
1042
+ Drnis DRNI's
1043
+ Dro DRO
1044
+ Drs. doctors
1045
+ Drska DRSKA
1046
+ druj DRUJ
1047
+ DsbA DSBA
1048
+ DS&BB DS and BB
1049
+ dsDNA DSDNA
1050
+ ds DS
1051
+ Ds D's
1052
+ D. S. DS
1053
+ D.S. DS
1054
+ DSGi DSGI
1055
+ dsi DSI
1056
+ DSi DSI
1057
+ dsl DSL
1058
+ DSMs DSM's
1059
+ dsn DSN
1060
+ D&SNGRR D and SNGRR
1061
+ dsu' DSU
1062
+ DTCs DTC's
1063
+ DTDs DTD's
1064
+ D. T. DT
1065
+ D.T. DT
1066
+ DT&E DT and E
1067
+ DT&I DT and I
1068
+ dtl DTL
1069
+ DTs DT's
1070
+ dty DTY
1071
+ D.U. DU
1072
+ dva DVA
1073
+ Dva DVA
1074
+ DVCs DVC's
1075
+ dvd DVD
1076
+ DVDs DVD's
1077
+ dv DV
1078
+ D. V. DV
1079
+ D.V. DV
1080
+ dve DVE
1081
+ dvfb DVFB
1082
+ dvi DVI
1083
+ dvije DVIJE
1084
+ Dvin DVIN
1085
+ Dvir DVIR
1086
+ Dwa DWA
1087
+ D&W D and W
1088
+ D. W. DW
1089
+ D.W. DW
1090
+ Dwedw DWEDW
1091
+ Dwi DWI
1092
+ dwr DWR
1093
+ dwur DWUR
1094
+ dxa DXA
1095
+ dx DX
1096
+ D.X. DX
1097
+ DXers DXER's
1098
+ Dydd DYDD
1099
+ dy DY
1100
+ Dy DY
1101
+ D. Y. DY
1102
+ dypl DYPL
1103
+ dyr DYR
1104
+ dz DZ
1105
+ dzis DZIS
1106
+ Dzog DZOG
1107
+ Dzor DZOR
1108
+ Dzus DZU's
1109
+ E. A. EA
1110
+ E.A. EA
1111
+ EAMs EAM's
1112
+ eas EAS
1113
+ Eas EA's
1114
+ Eav EAV
1115
+ E. B. EB
1116
+ E.B. EB
1117
+ Ebn EBN
1118
+ E&BR E and BR
1119
+ Ebru EBRU
1120
+ EBWs EBW's
1121
+ ecc ECC
1122
+ Eccl ECCL
1123
+ ec EC
1124
+ Ec EC
1125
+ E. C. EC
1126
+ E.C. EC
1127
+ ece ECE
1128
+ Ece ECE
1129
+ Ecem ECEM
1130
+ ecg ECG
1131
+ ecma ECMA
1132
+ ECNs ECN's
1133
+ ECP&DA's ECP and DA's
1134
+ eCRM ECRM
1135
+ ect ECT
1136
+ ECVs ECV's
1137
+ edb EDB
1138
+ E. D. ED
1139
+ E.D. ED
1140
+ Edh EDH
1141
+ edhe EDHE
1142
+ EDMs EDM's
1143
+ Edmx EDMX
1144
+ edn EDN
1145
+ Edn EDN
1146
+ eds EDS
1147
+ Eds ED's
1148
+ Edw EDW
1149
+ Eeb EEB
1150
+ E. E. EE
1151
+ E.E. EE
1152
+ Eef EEF
1153
+ Eega EEGA
1154
+ eene EENE
1155
+ eep EEP
1156
+ Ees EE's
1157
+ eeuw EEUW
1158
+ Eeuw EEUW
1159
+ EEZs EEZ's
1160
+ efc EFC
1161
+ ef EF
1162
+ Ef EF
1163
+ E. F. EF
1164
+ E.F. EF
1165
+ eFS EFS
1166
+ Efs EF's
1167
+ Efu EFU
1168
+ Egba EGBA
1169
+ egfl EGFL
1170
+ e.g. for example
1171
+ e. g. for example
1172
+ EG&G EG and G
1173
+ Egi EGI
1174
+ E&GR E and GR
1175
+ egy EGY
1176
+ Egyl EGYL
1177
+ EHAs EHA's
1178
+ E. H. EH
1179
+ E.H. EH
1180
+ ehf EHF
1181
+ Ehttp EHTTP
1182
+ EiCs EIC's
1183
+ E. I. EI
1184
+ E.I. EI
1185
+ eIF EIF
1186
+ Eitr EITR
1187
+ EITs EIT's
1188
+ EJBs EJB's
1189
+ ej EJ
1190
+ E. J. EJ
1191
+ E.J. EJ
1192
+ E. K. EK
1193
+ E.K. EK
1194
+ eki EKI
1195
+ Ekma EKMA
1196
+ Ekow EKOW
1197
+ eks EKS
1198
+ Eks EK's
1199
+ E. L. EL
1200
+ E.L. EL
1201
+ ELTs ELT's
1202
+ ELUs ELU's
1203
+ Embd EMBD
1204
+ eMC EMC
1205
+ EMDs EMD's
1206
+ Emea EMEA
1207
+ E&M E and M
1208
+ E. M. EM
1209
+ E.M. EM
1210
+ Emge EMGE
1211
+ emra EMRA
1212
+ E&NA E and NA
1213
+ Enas ENA's
1214
+ EnBW ENBW
1215
+ EncFS ENCFS
1216
+ Encyc ENCYC
1217
+ E. N. EN
1218
+ E.N. EN
1219
+ Enes ENE's
1220
+ engl ENGL
1221
+ Engl ENGL
1222
+ Engr ENGR
1223
+ Eni ENI
1224
+ enn ENN
1225
+ Enn ENN
1226
+ ens ENS
1227
+ Ens EN's
1228
+ Enso ENSO
1229
+ ENTJs ENTJ's
1230
+ EoD EOD
1231
+ E. O. EO
1232
+ E.O. EO
1233
+ Eois EOI's
1234
+ eok EOK
1235
+ Eole EOLE
1236
+ Eol EOL
1237
+ Eolss EOLS's
1238
+ Eom EOM
1239
+ eop EOP
1240
+ eoptta EOPTTA
1241
+ Eora EORA
1242
+ Eorl EORL
1243
+ Eors EOR's
1244
+ E&P E and P
1245
+ Epe EPE
1246
+ ep EP
1247
+ Ep EP
1248
+ E. P. EP
1249
+ E.P. EP
1250
+ Eph EPH
1251
+ Ephs EPH's
1252
+ epi EPI
1253
+ Epi EPI
1254
+ ePO EPO
1255
+ epos' EPO's
1256
+ epos EPOS
1257
+ Epos EPO's
1258
+ E.P.s EP's
1259
+ eq EQ
1260
+ Eq EQ
1261
+ E. R. ER
1262
+ E.R. ER
1263
+ ERJs ERJ's
1264
+ Eru ERU
1265
+ Eruv ERUV
1266
+ Erv ERV
1267
+ ES3 ES three
1268
+ Esa ESA
1269
+ Esam ESAM
1270
+ Esa's ESA's
1271
+ Esat ESAT
1272
+ ESCs ESC's
1273
+ Esd ESD
1274
+ E. S. ES
1275
+ E.S. ES
1276
+ esi ESI
1277
+ espn ESPN
1278
+ ESPNhttp ESPNHTTP
1279
+ esq ESQ
1280
+ Esq ESQ
1281
+ Esq. esquire
1282
+ Esraa ESRAA
1283
+ Esra ESRA
1284
+ Esref ESREF
1285
+ Esri ESRI
1286
+ estd ESTD
1287
+ Estd ESTD
1288
+ esu ESU
1289
+ Esva ESVA
1290
+ Etad ETAD
1291
+ ETBs ETB's
1292
+ E. T. ET
1293
+ E.T. ET
1294
+ ETFs ETF's
1295
+ Et'hem ETHEM
1296
+ Eti ETI
1297
+ etj ETJ
1298
+ Eto ETO
1299
+ ets ETS
1300
+ Ets ET's
1301
+ ETs ET's
1302
+ ett ETT
1303
+ Ett ETT
1304
+ Ettre ETTRE
1305
+ Etts ETT's
1306
+ Etz ETZ
1307
+ EUBs EUB's
1308
+ E. U. EU
1309
+ E.U. EU
1310
+ eup EUP
1311
+ Eurwg EURWG
1312
+ Euse EUSE
1313
+ Eusi EUSI
1314
+ EvaGT EVAGT
1315
+ E. V. EV
1316
+ E.V. EV
1317
+ evl EVL
1318
+ Evna EVNA
1319
+ evnt EVNT
1320
+ EVs EV's
1321
+ evv EVV
1322
+ Ewa's EWA's
1323
+ E. W. EW
1324
+ E.W. EW
1325
+ Ewha EWHA
1326
+ EWOs EWO's
1327
+ Ewu EWU
1328
+ excl EXCL
1329
+ exd EXD
1330
+ E.X. EX
1331
+ exhb EXHB
1332
+ exh EXH
1333
+ Exptl EXPTL
1334
+ exsul EXSUL
1335
+ E&Y E and Y
1336
+ E. Y. EY
1337
+ Eyk EYK
1338
+ eyu EYU
1339
+ Ezaa EZAA
1340
+ ez EZ
1341
+ Ez EZ
1342
+ Ezh EZH
1343
+ Ezu EZU
1344
+ Ezy EZY
1345
+ Faaa FAAA
1346
+ faa FAA
1347
+ Faa FAA
1348
+ faama FAAMA
1349
+ FabH FABH
1350
+ FabR FABR
1351
+ F.A.B.'s FAB's
1352
+ Fadl FADL
1353
+ fa FA
1354
+ Fa FA
1355
+ F. A. FA
1356
+ F.A. FA
1357
+ Faf FAF
1358
+ Fafhrd FAFHRD
1359
+ fai FAI
1360
+ Fai FAI
1361
+ FAIPs FAIP's
1362
+ Faiq FAIQ
1363
+ fajn FAJN
1364
+ Fal FAL
1365
+ farw FARW
1366
+ fasc FASC
1367
+ Fasc FASC
1368
+ fas FAS
1369
+ Fas FA's
1370
+ FAs FA's
1371
+ FasL FASL
1372
+ F. B. FB
1373
+ F.B. FB
1374
+ F.B.G.s FBG's
1375
+ FBOs FBO's
1376
+ fcb FCB
1377
+ FCBs FCB's
1378
+ F&C F and C
1379
+ fc FC
1380
+ F. C. FC
1381
+ F.C. FC
1382
+ F.C.'s FC's
1383
+ FD&C FD and C
1384
+ FDCPAs FDCPA's
1385
+ F. D. FD
1386
+ F.D. FD
1387
+ FDICs FDIC's
1388
+ FdI FDI
1389
+ F. E. FE
1390
+ F.E. FE
1391
+ FEGs FEG's
1392
+ Fes FE's
1393
+ ffc FFC
1394
+ ff FF
1395
+ F. F. FF
1396
+ F.F. FF
1397
+ ffm FFM
1398
+ FFs FF's
1399
+ F. G. FG
1400
+ F.G. FG
1401
+ fgk FGK
1402
+ FGs FG's
1403
+ F. H. FH
1404
+ F.H. FH
1405
+ fhm FHM
1406
+ fiadh FIADH
1407
+ fia FIA
1408
+ FiBL FIBL
1409
+ Fi'd FID
1410
+ F.I F
1411
+ F&I F and I
1412
+ F. I. FI
1413
+ F.I. FI
1414
+ Figl FIGL
1415
+ fija FIJA
1416
+ Fio FIO
1417
+ FiO FIO
1418
+ F. J. FJ
1419
+ F.J. FJ
1420
+ F. K. FK
1421
+ F.K. FK
1422
+ Fla FLA
1423
+ Fles FLE's
1424
+ fl FL
1425
+ F. L. FL
1426
+ F.L. FL
1427
+ fli FLI
1428
+ flyr FLYR
1429
+ fm FM
1430
+ F. M. FM
1431
+ F.M. FM
1432
+ fMRI FMRI
1433
+ FMs FM's
1434
+ F. N. FN
1435
+ F.N. FN
1436
+ F. O. FO
1437
+ FPDs FPD's
1438
+ fp FP
1439
+ F. P. FP
1440
+ F.P. FP
1441
+ FPGAs FPGA's
1442
+ FPM&SA FPM and SA
1443
+ fps FPS
1444
+ FPSs FPS's
1445
+ F.Q. FQ
1446
+ fr FR
1447
+ F. R. FR
1448
+ F.R. FR
1449
+ frs FRS
1450
+ Fru FRU
1451
+ Frwydr FRWYDR
1452
+ fsf FSF
1453
+ fs FS
1454
+ F. s F's
1455
+ Fs F's
1456
+ F. S. FS
1457
+ F.S. FS
1458
+ fsn FSN
1459
+ FSSs FSS's
1460
+ FTAs FTA's
1461
+ Ft. Fort
1462
+ F. T. FT
1463
+ F.T. FT
1464
+ ftp FTP
1465
+ FtsA FTSA
1466
+ Ftuh FTUH
1467
+ F. V. FV
1468
+ F.V. FV
1469
+ F&W F and W
1470
+ F. W. FW
1471
+ F.W. FW
1472
+ fwr FWR
1473
+ fx FX
1474
+ Fyb FYB
1475
+ F.Y. FY
1476
+ Fyn FYN
1477
+ fyi FYI
1478
+ fyr FYR
1479
+ Fyw FYW
1480
+ F.Z. FZ
1481
+ Gaac GAAC
1482
+ G. A. GA
1483
+ G.A. GA
1484
+ Gbe GBE
1485
+ GbE GBE
1486
+ G. B. GB
1487
+ G.B. GB
1488
+ GB&NDR GB and NDR
1489
+ G&C G and C
1490
+ G. C. GC
1491
+ G.C. GC
1492
+ GCSEs GCSE's
1493
+ GC&SF GC and SF
1494
+ Gdal GDAL
1495
+ Gde GDE
1496
+ gd GD
1497
+ G. D. GD
1498
+ G.D. GD
1499
+ Gdow GDOW
1500
+ GDPs GDP's
1501
+ gdr GDR
1502
+ Gebr GEBR
1503
+ Gebt GEBT
1504
+ Ged GED
1505
+ ge GE
1506
+ G. E. GE
1507
+ G.E. GE
1508
+ Gek GEK
1509
+ Gen. general
1510
+ geq GEQ
1511
+ Geu GEU
1512
+ Gev GEV
1513
+ GeV GEV
1514
+ Gfa GFA
1515
+ GF&A GFANDA
1516
+ G. F. GF
1517
+ G.F. GF
1518
+ G. G. GG
1519
+ G.G. GG
1520
+ Ghe GHE
1521
+ G. H. GH
1522
+ G.H. GH
1523
+ GHGs GHG's
1524
+ Ghir GHIR
1525
+ ghra GHRA
1526
+ GH&SA GH and SA
1527
+ Giei GIEI
1528
+ G. I. GI
1529
+ G.I. GI
1530
+ G. J. GJ
1531
+ G.J. GJ
1532
+ G. K. GK
1533
+ G.K. GK
1534
+ Gla GLA
1535
+ G&L G and L
1536
+ G. L. GL
1537
+ G.L. GL
1538
+ gli GLI
1539
+ Gli GLI
1540
+ Glis GLI's
1541
+ glnA GLNA
1542
+ Glos GLO's
1543
+ gma GMA
1544
+ GmbH GMBH
1545
+ Gmel GMEL
1546
+ gm GM
1547
+ G. M. GM
1548
+ G.M. GM
1549
+ gmin GMIN
1550
+ GMOs GMO's
1551
+ GMs GM's
1552
+ gmt GMT
1553
+ Gmul GMUL
1554
+ gnb GNB
1555
+ G. N. GN
1556
+ G.N. GN
1557
+ GNP&BR GNP and BR
1558
+ gnp GNP
1559
+ GNPs GNP's
1560
+ G.O. GO
1561
+ GPCRs GPCR's
1562
+ GPdI GPDI
1563
+ G&P G and P
1564
+ gp GP
1565
+ G. P. GP
1566
+ G.P. GP
1567
+ GpIIb GPIIB
1568
+ GPMGs GPMG's
1569
+ GPRs GPR's
1570
+ GPUs GPU's
1571
+ G. Q. GQ
1572
+ grac GRAC
1573
+ gra GRA
1574
+ Gra GRA
1575
+ Grbac GRBAC
1576
+ GRBs GRB's
1577
+ grc GRC
1578
+ GRCs GRC's
1579
+ gre GRE
1580
+ Gre GRE
1581
+ gr GR
1582
+ G. R. GR
1583
+ G.R. GR
1584
+ grrl GRRL
1585
+ Gryf GRYF
1586
+ gry GRY
1587
+ Gry GRY
1588
+ grz GRZ
1589
+ G&S G and S
1590
+ Gs G's
1591
+ G. S. GS
1592
+ G.S. GS
1593
+ GSIs GSI's
1594
+ GSOp GSOP
1595
+ GTAs GTA's
1596
+ gt GT
1597
+ G. T. GT
1598
+ G.T. GT
1599
+ GTi GTI
1600
+ GTs GT's
1601
+ G. U. GU
1602
+ G.U. GU
1603
+ GUIs GUI's
1604
+ G. V. GV
1605
+ G.V. GV
1606
+ GvpA GVPA
1607
+ GWe GWE
1608
+ Gwet GWET
1609
+ gw GW
1610
+ G. W. GW
1611
+ G.W. GW
1612
+ Gwi GWI
1613
+ gwr GWR
1614
+ Gy GY
1615
+ G. Y. GY
1616
+ Gyn GYN
1617
+ Gyps GYP's
1618
+ gyu GYU
1619
+ Gyu GYU
1620
+ H. A. HA
1621
+ H.A. HA
1622
+ H&A H and A
1623
+ HBCo HBCO
1624
+ HBGAs HBGA's
1625
+ H. B. HB
1626
+ H.B. HB
1627
+ H&BR H and BR
1628
+ HCEs HCE's
1629
+ H. C. HC
1630
+ H.C. HC
1631
+ hcl HCL
1632
+ HClO HCLO
1633
+ HCoV HCOV
1634
+ HCPs HCP's
1635
+ Hcy HCY
1636
+ Hdad HDAD
1637
+ hDAF HDAF
1638
+ HDDs HDD's
1639
+ hd HD
1640
+ H. D. HD
1641
+ H.D. HD
1642
+ hdh HDH
1643
+ HDi HDI
1644
+ H. E. HE
1645
+ H.E. HE
1646
+ hezb HEZB
1647
+ Hezb HEZB
1648
+ H&F H and F
1649
+ H. F. HF
1650
+ H.F. HF
1651
+ hgcA HGCA
1652
+ hg HG
1653
+ H. G. HG
1654
+ H.G. HG
1655
+ HgO HGO
1656
+ HGPs HGP's
1657
+ HgU HGU
1658
+ HGVs HGV's
1659
+ H&H H and H
1660
+ H. H. HH
1661
+ H.H. HH
1662
+ H. I. HI
1663
+ H.I. HI
1664
+ Hizb HIZB
1665
+ hja HJA
1666
+ H. J. HJ
1667
+ H.J. HJ
1668
+ Hkam HKAM
1669
+ H&K H and K
1670
+ H. K. HK
1671
+ H.K. HK
1672
+ Hla HLA
1673
+ H. L. HL
1674
+ H.L. HL
1675
+ Hluk HLUK
1676
+ hlutr HLUTR
1677
+ Hly HLY
1678
+ Hman HMAN
1679
+ H&M H and M
1680
+ hm HM
1681
+ H. M. HM
1682
+ H.M. HM
1683
+ HMOs HMO's
1684
+ hMRE HMRE
1685
+ H&M's H and M's
1686
+ H. N. HN
1687
+ H.N. HN
1688
+ H. O. HO
1689
+ H.O. HO
1690
+ hpc HPC
1691
+ hp HP
1692
+ H. P. HP
1693
+ H.P. HP
1694
+ H.Q. HQ
1695
+ H&R H and R
1696
+ H. R. HR
1697
+ H.R. HR
1698
+ hroa HROA
1699
+ HRo HRO
1700
+ hrs HRS
1701
+ HRs HR's
1702
+ hscy HSCY
1703
+ Hsee HSEE
1704
+ hs HS
1705
+ Hs H's
1706
+ H. S. HS
1707
+ H.S. HS
1708
+ hsi HSI
1709
+ hSlo HSLO
1710
+ HSPs HSP's
1711
+ HSTs HST's
1712
+ hsv HSV
1713
+ hTAS HTAS
1714
+ HTAs HTA's
1715
+ htc HTC
1716
+ ht HT
1717
+ H. T. HT
1718
+ H.T. HT
1719
+ hti HTI
1720
+ Htin HTIN
1721
+ Htoo HTOO
1722
+ HtrA HTRA
1723
+ HTRs HTR's
1724
+ http HTTP
1725
+ https HTTPS
1726
+ Htwa HTWA
1727
+ H. U. HU
1728
+ HVCs HVC's
1729
+ hvcv HVCV
1730
+ hvem HVEM
1731
+ Hvem HVEM
1732
+ H. V. HV
1733
+ H.V. HV
1734
+ H. W. HW
1735
+ H.W. HW
1736
+ hwy HWY
1737
+ Hwy HWY
1738
+ hxt HXT
1739
+ H. Y. HY
1740
+ H.Y. HY
1741
+ H. Z. HZ
1742
+ HZ&PC HZ and PC
1743
+ iaaf IAAF
1744
+ IaaS IAAS
1745
+ Iaca IACA
1746
+ Iacob IACOB
1747
+ IACTs IACT's
1748
+ Iacub IACUB
1749
+ iagt IAGT
1750
+ I. A. IA
1751
+ I.A. IA
1752
+ Iapa IAPA
1753
+ iar IAR
1754
+ Iar IAR
1755
+ IAUCs IAUC's
1756
+ Iax IAX
1757
+ Ibac IBAC
1758
+ Ibaes IBAE's
1759
+ IBCs IBC's
1760
+ IBDA'A IBDAA
1761
+ Ibda IBDA
1762
+ ibdal IBDAL
1763
+ I. B. IB
1764
+ I.B. IB
1765
+ Ibm IBM
1766
+ Ibne IBNE
1767
+ ibn IBN
1768
+ Ibra IBRA
1769
+ Ibsa IBSA
1770
+ Ibs IB's
1771
+ iBT IBT
1772
+ Ibu IBU
1773
+ iby IBY
1774
+ ICBMs ICBM's
1775
+ Icche ICCHE
1776
+ icf ICF
1777
+ I&C I and C
1778
+ ic IC
1779
+ Ic IC
1780
+ I. C. IC
1781
+ I.C. IC
1782
+ Ici ICI
1783
+ iCN ICN
1784
+ Ico ICO
1785
+ IcRn ICRN
1786
+ ics ICS
1787
+ ICs IC's
1788
+ ICTs ICT's
1789
+ ICTVdB ICTVDB
1790
+ ICv ICV
1791
+ iCyt ICYT
1792
+ Iddaa IDDAA
1793
+ Idd IDD
1794
+ ID'd IDD
1795
+ id ID
1796
+ I. D. ID
1797
+ I.D. ID
1798
+ Idi IDI
1799
+ Idir IDIR
1800
+ IDLHs IDLH's
1801
+ Idm IDM
1802
+ IDPs IDP's
1803
+ ids IDS
1804
+ IDs ID's
1805
+ IDx IDX
1806
+ i.e. that is
1807
+ Ifa IFA
1808
+ ifc IFC
1809
+ IfF IFF
1810
+ Ifft IFFT
1811
+ ifi IFI
1812
+ Ifi IFI
1813
+ IFIs IFI's
1814
+ IfM IFM
1815
+ Ifni IFNI
1816
+ IFNs IFN's
1817
+ Ifop IFOP
1818
+ IFRSs IFRS's
1819
+ Iga IGA
1820
+ IgA IGA
1821
+ IGFs IGF's
1822
+ 'ig IG
1823
+ Ig IG
1824
+ I.G. IG
1825
+ igi IGI
1826
+ IgM IGM
1827
+ ign IGN
1828
+ Ign IGN
1829
+ Ih IH
1830
+ I. H. IH
1831
+ ihi IHI
1832
+ ihm IHM
1833
+ ihn IHN
1834
+ Ihn IHN
1835
+ Ihor IHOR
1836
+ Iht IHT
1837
+ IIb IIB
1838
+ IIc IIC
1839
+ IIfx IIFX
1840
+ Iiga IIGA
1841
+ IIga IIGA
1842
+ IIgs IIG's
1843
+ IIGs IIG's
1844
+ IIIb IIIB
1845
+ IIIBy IIIBY
1846
+ IIIc IIIC
1847
+ IIId IIID
1848
+ I. I. II
1849
+ I.I. II
1850
+ ija IJA
1851
+ Ija IJA
1852
+ ij IJ
1853
+ I. J. IJ
1854
+ Ijok IJOK
1855
+ Ijui IJUI
1856
+ Ikh IKH
1857
+ I.K. IK
1858
+ IKr IKR
1859
+ Iksa IKSA
1860
+ Iku IKU
1861
+ Ilbe ILBE
1862
+ IL&FS IL and FS
1863
+ ili ILI
1864
+ Ili ILI
1865
+ I. L. IL
1866
+ I.L. IL
1867
+ Ilm ILM
1868
+ ilu ILU
1869
+ Ilu ILU
1870
+ IMbd IMBD
1871
+ imc IMC
1872
+ imdb IMDB
1873
+ IMDb. IMDB
1874
+ IMDb IMDB
1875
+ ime IME
1876
+ I&M I and M
1877
+ I. M. IM
1878
+ I.M. IM
1879
+ Iml IML
1880
+ Imm IMM
1881
+ Imms' IMM's
1882
+ imoa IMOA
1883
+ impr IMPR
1884
+ Impr IMPR
1885
+ iMSNs IMSN's
1886
+ Imst IMST
1887
+ Inba INBA
1888
+ inb INB
1889
+ incl INCL
1890
+ ind IND
1891
+ Ind IND
1892
+ Infs INF's
1893
+ inHg INHG
1894
+ I. N. IN
1895
+ I.N. IN
1896
+ InlB INLB
1897
+ InP INP
1898
+ Inre INRE
1899
+ Inspx INSPX
1900
+ Ints INT's
1901
+ INTs INT's
1902
+ Intu INTU
1903
+ inv INV
1904
+ Ioba IOBA
1905
+ IODs IOD's
1906
+ io. IO
1907
+ io IO
1908
+ .I.o. IO
1909
+ Io. IO
1910
+ Io IO
1911
+ .IO IO
1912
+ I/O IO
1913
+ Ioka IOKA
1914
+ Iok IOK
1915
+ IoM IOM
1916
+ IOPs IOP's
1917
+ iOS IOS
1918
+ Io's IO's
1919
+ I/Os IO's
1920
+ I.O.'s IO's
1921
+ Iosu IOSU
1922
+ IoT IOT
1923
+ IoW IOW
1924
+ IPCs IPC's
1925
+ ip IP
1926
+ Ip IP
1927
+ I.P. IP
1928
+ Ipo IPO
1929
+ IPPs IPP's
1930
+ Ippu IPPU
1931
+ iPPV IPPV
1932
+ iPSC IPSC
1933
+ iPSCs IPSC's
1934
+ ipse IPSE
1935
+ iPS IPS
1936
+ IPs IP's
1937
+ IPv IPV
1938
+ iQ IQ
1939
+ I.Q. IQ
1940
+ Irgm IRGM
1941
+ Irig IRIG
1942
+ Iril IRIL
1943
+ i'r IR
1944
+ ir IR
1945
+ Ir IR
1946
+ I. R. IR
1947
+ I.R. IR
1948
+ Irla IRLA
1949
+ Irmis IRMI's
1950
+ iro IRO
1951
+ Iro IRO
1952
+ Irra IRRA
1953
+ IRs IR's
1954
+ iru IRU
1955
+ Iru IRU
1956
+ ISAv ISAV
1957
+ Isba ISBA
1958
+ isbn ISBN
1959
+ ISBNs ISBN's
1960
+ isCf ISCF
1961
+ iSC ISC
1962
+ Isc ISC
1963
+ Isd ISD
1964
+ isdn ISDN
1965
+ Isgec ISGEC
1966
+ ishq ISHQ
1967
+ Ishq ISHQ
1968
+ I. S. IS
1969
+ I.S. IS
1970
+ Isl ISL
1971
+ ISMNs ISMN's
1972
+ ISPs ISP's
1973
+ Isra'il ISRAIL
1974
+ Isra ISRA
1975
+ iss ISS
1976
+ Iss IS's
1977
+ ISSNs ISSN's
1978
+ Isu ISU
1979
+ Ite ITE
1980
+ ITHs ITH's
1981
+ iti ITI
1982
+ Iti ITI
1983
+ I.T. IT
1984
+ itk ITK
1985
+ Itk ITK
1986
+ ITNNs ITNN's
1987
+ Itoi ITOI
1988
+ iTP ITP
1989
+ Itse ITSE
1990
+ itt ITT
1991
+ Itu ITU
1992
+ Itz ITZ
1993
+ Iucn IUCN
1994
+ IUDs IUD's
1995
+ Iuz IUZ
1996
+ I. V. IV
1997
+ I.V. IV
1998
+ IVs IV's
1999
+ Ivu IVU
2000
+ Iwas IWA's
2001
+ Iwaz IWAZ
2002
+ IWBs IWB's
2003
+ iwi IWI
2004
+ iw IW
2005
+ I. W. IW
2006
+ iwrg IWRG
2007
+ iwspy IWSPY
2008
+ Iwuh IWUH
2009
+ IXb IXB
2010
+ IXBs IXB's
2011
+ IXCs IXC's
2012
+ IXe IXE
2013
+ iXL IXL
2014
+ ixtle IXTLE
2015
+ Iya IYA
2016
+ Iyar IYAR
2017
+ iyem IYEM
2018
+ Iyi IYI
2019
+ iy IY
2020
+ I. Y. IY
2021
+ I.Y. IY
2022
+ Jaf JAF
2023
+ J. A. JA
2024
+ J&B J and B
2025
+ jb JB
2026
+ J. B. JB
2027
+ J.B. JB
2028
+ jcis JCIS
2029
+ jc JC
2030
+ J. C. JC
2031
+ J.C. JC
2032
+ JCRs JCR's
2033
+ JdeBP JDEBP
2034
+ jdi JDI
2035
+ jd JD
2036
+ J. D. JD
2037
+ J.D. JD
2038
+ jdk JDK
2039
+ J&D's J and D's
2040
+ J.D.s JD's
2041
+ J. E. JE
2042
+ J.E. JE
2043
+ jf JF
2044
+ J. F. JF
2045
+ J.F. JF
2046
+ J. G. JG
2047
+ J.G. JG
2048
+ jh JH
2049
+ J. H. JH
2050
+ J.H. JH
2051
+ JHs JH's
2052
+ JHSVs JHSV's
2053
+ JHud JHUD
2054
+ J. I. JI
2055
+ J.I. JI
2056
+ J&J JANDJ
2057
+ J. J. JJ
2058
+ J.J. JJ
2059
+ Jka JKA
2060
+ jkd JKD
2061
+ J&K J and K
2062
+ J. K. JK
2063
+ J.K. JK
2064
+ jkx JKX
2065
+ J. L. JL
2066
+ J.L. JL
2067
+ J. M. JM
2068
+ J.M. JM
2069
+ jnb JNB
2070
+ J. N. JN
2071
+ J.N. JN
2072
+ Jno JNO
2073
+ jnr JNR
2074
+ J. O. JO
2075
+ J.O. JO
2076
+ jpg JPG
2077
+ J&P J and P
2078
+ jp JP
2079
+ J.P. JP
2080
+ J.Q. JQ
2081
+ J&R J and R
2082
+ J. R. JR
2083
+ J.R. JR
2084
+ Jr. junior
2085
+ jr. junior
2086
+ Jr junior
2087
+ jr junior
2088
+ J.R.'s JR's
2089
+ jsb JSB
2090
+ Js J's
2091
+ J. S. JS
2092
+ J.S. JS
2093
+ J&T J and T
2094
+ J. T. JT
2095
+ J.T. JT
2096
+ JT&KW JT and KW
2097
+ J. U. JU
2098
+ J&V J and V
2099
+ J. V. JV
2100
+ J.V. JV
2101
+ Jwa JWA
2102
+ J. W. JW
2103
+ J.W. JW
2104
+ jx JX
2105
+ J. Y. JY
2106
+ J.Y. JY
2107
+ Jym JYM
2108
+ Jymn JYMN
2109
+ J. Z. JZ
2110
+ J.Z. JZ
2111
+ K8S KUBERNETES
2112
+ K. A. KA
2113
+ K.A. KA
2114
+ kbi KBI
2115
+ K&B K and B
2116
+ K. B. KB
2117
+ K.B. KB
2118
+ K. C. KC
2119
+ K.C. KC
2120
+ K.C.'s KC's
2121
+ kDa KDA
2122
+ kdal KDAL
2123
+ K&D K and D
2124
+ kd KD
2125
+ K. D. KD
2126
+ K.D. KD
2127
+ K. D.'s KD's
2128
+ K.D.'s KD's
2129
+ ke KE
2130
+ Ke KE
2131
+ K. E. KE
2132
+ K.E. KE
2133
+ kfc KFC
2134
+ K.F. KF
2135
+ kgr KGR
2136
+ kgt KGT
2137
+ KgU KGU
2138
+ kgv KGV
2139
+ Khizr KHIZR
2140
+ K. H. KH
2141
+ K.H. KH
2142
+ Khlav KHLAV
2143
+ Khmu KHMU
2144
+ Khri KHRI
2145
+ Khru KHRU
2146
+ K. I. KI
2147
+ K.I. KI
2148
+ Kjer KJER
2149
+ kj KJ
2150
+ K. J. KJ
2151
+ K.J. KJ
2152
+ Kjop KJOP
2153
+ KJo's KJO's
2154
+ K. K. KK
2155
+ K.K. KK
2156
+ Kle KLE
2157
+ kl KL
2158
+ K. L. KL
2159
+ K.L. KL
2160
+ K&M K and M
2161
+ km KM
2162
+ K. M. KM
2163
+ K.M. KM
2164
+ K&N K and N
2165
+ kn KN
2166
+ K. N. KN
2167
+ K.N. KN
2168
+ k'o KO
2169
+ K. O. KO
2170
+ K.O. KO
2171
+ KPIs KPI's
2172
+ K&P K and P
2173
+ K. P. KP
2174
+ K.P. KP
2175
+ Kppen KPPEN
2176
+ kptm KPTM
2177
+ krc KRC
2178
+ KRCs KRC's
2179
+ K.R.I.T.'s KRIT's
2180
+ Krka KRKA
2181
+ kr KR
2182
+ K. R. KR
2183
+ K.R. KR
2184
+ Krne KRNE
2185
+ kroz KROZ
2186
+ Kroz KROZ
2187
+ Kru KRU
2188
+ ksa KSA
2189
+ kset KSET
2190
+ Kseur KSEUR
2191
+ Ksevt KSEVT
2192
+ ks KS
2193
+ K. S. KS
2194
+ K.S. KS
2195
+ Ktav KTAV
2196
+ KT&K KT and K
2197
+ kt KT
2198
+ K. T. KT
2199
+ K.T. KT
2200
+ Kuaa KUAA
2201
+ Kud KUD
2202
+ kuih KUIH
2203
+ K. U. KU
2204
+ K.U. KU
2205
+ Kutb KUTB
2206
+ Kvik KVIK
2207
+ Kvit KVIT
2208
+ K. V. KV
2209
+ K.V. KV
2210
+ Kvyat KVYAT
2211
+ kwa KWA
2212
+ kwe KWE
2213
+ Kyse KYSE
2214
+ kyt KYT
2215
+ Kyt KYT
2216
+ kz KZ
2217
+ K. Z. KZ
2218
+ L. A. LA
2219
+ L.A.'s LA's
2220
+ Lay's LAY's
2221
+ LBi LBI
2222
+ L. B. LB
2223
+ L.B. LB
2224
+ LBPs LBP's
2225
+ LCAs LCA's
2226
+ lcc LCC
2227
+ LCCs LCC's
2228
+ LC&DR LC and DR
2229
+ LCDs LCD's
2230
+ LCIs LCI's
2231
+ L. C. LC
2232
+ L.C. LC
2233
+ lcn LCN
2234
+ LCSs LCS's
2235
+ LCTs LCT's
2236
+ LCVPs LCVP's
2237
+ Lda LDA
2238
+ LDCs LDC's
2239
+ L. D. LD
2240
+ L.D. LD
2241
+ L. E. LE
2242
+ L.E. LE
2243
+ LFFCs LFFC's
2244
+ L. F. LF
2245
+ L.F. LF
2246
+ LFs LF's
2247
+ LFSRs LFSR's
2248
+ L. G. LG
2249
+ L.G. LG
2250
+ LGMs LGM's
2251
+ LG&RDD LG and RDD
2252
+ LGs LG's
2253
+ LGUs LGU's
2254
+ LGVs LGV's
2255
+ Lha LHA
2256
+ L. H. LH
2257
+ L.H. LH
2258
+ L. I. LI
2259
+ L.I. LI
2260
+ Lje LJE
2261
+ L. J. LJ
2262
+ L.J. LJ
2263
+ L. K. LK
2264
+ lks LKS
2265
+ Llapi LLAPI
2266
+ lle LLE
2267
+ lli LLI
2268
+ L&L L and L
2269
+ L. L. LL
2270
+ L.L. LL
2271
+ llp LLP
2272
+ llu LLU
2273
+ LMGs LMG's
2274
+ lm LM
2275
+ L. M. LM
2276
+ L.M. LM
2277
+ LMPs LMP's
2278
+ LMQs LMQ's
2279
+ LnAIB LNAIB
2280
+ L&N L and N
2281
+ L. N. LN
2282
+ L.N. LN
2283
+ L.O.C.'s LOC's
2284
+ Lokk LOKK
2285
+ l'OL LOL
2286
+ L. O. LO
2287
+ L.O. LO
2288
+ LPARs LPAR's
2289
+ L&P L and P
2290
+ lp LP
2291
+ L. P. LP
2292
+ L.P. LP
2293
+ L&PM L and PM
2294
+ LPMud LPMUD
2295
+ LPs LP's
2296
+ LPThe LPTHE
2297
+ lr LR
2298
+ L. R. LR
2299
+ L.R. LR
2300
+ LRTs LRT's
2301
+ LRVs LRV's
2302
+ LSDs LSD's
2303
+ LSi LSI
2304
+ ls LS
2305
+ Ls L's
2306
+ L. S. LS
2307
+ L.S. LS
2308
+ LSTs LST's
2309
+ LSVCCs LSVCC's
2310
+ Ltda LTDA
2311
+ Ltd. limited
2312
+ L&T L and T
2313
+ Lt. lieutenant
2314
+ lt LT
2315
+ L. T. LT
2316
+ L.T. LT
2317
+ ltoh LTOH
2318
+ L. U. LU
2319
+ lv LV
2320
+ L. V. LV
2321
+ L.V. LV
2322
+ L. W. LW
2323
+ L.W. LW
2324
+ L.Y. LY
2325
+ L&YR L and YR
2326
+ M. A. MA
2327
+ M.A. MA
2328
+ MAX-Q MAXQ
2329
+ M&A M and A
2330
+ Mbewu MBEWU
2331
+ mbi MBI
2332
+ M. B. MB
2333
+ M.B. MB
2334
+ mBo MBO
2335
+ Mbre MBRE
2336
+ mbr MBR
2337
+ MCCs MCC's
2338
+ M. C. MC
2339
+ M.C. MC
2340
+ MCs MC's
2341
+ mcyG MCYG
2342
+ mdb MDB
2343
+ MDCs MDC's
2344
+ mdDA MDDA
2345
+ MD&DI MD and DI
2346
+ Mde MDE
2347
+ MDGs MDG's
2348
+ MDHUs MDHU's
2349
+ mDia MDIA
2350
+ MDic MDIC
2351
+ mdla MDLA
2352
+ md MD
2353
+ M. D. MD
2354
+ M.D. MD
2355
+ mdr MDR
2356
+ M. E. ME
2357
+ M.E. ME
2358
+ MFDs MFD's
2359
+ MFe MFE
2360
+ M. F. MF
2361
+ M.F. MF
2362
+ MFTs MFT's
2363
+ Mgadla MGADLA
2364
+ Mgal MGAL
2365
+ mga MGA
2366
+ Mga MGA
2367
+ Mgbo MGBO
2368
+ MGen MGEN
2369
+ M&G M and G
2370
+ M. G. MG
2371
+ M.G. MG
2372
+ M&GN M and GN
2373
+ M&GR's M and GR's
2374
+ Mha MHA
2375
+ MHCs MHC's
2376
+ MH&L MH and L
2377
+ M&H M and H
2378
+ M. H. MH
2379
+ M.H. MH
2380
+ Mhor MHOR
2381
+ Mhow MHOW
2382
+ M.I.A.'s MIA's
2383
+ M&I M and I
2384
+ M. I. MI
2385
+ M.I. MI
2386
+ M&J M and J
2387
+ M. J. MJ
2388
+ M.J. MJ
2389
+ M. K. MK
2390
+ M.K. MK
2391
+ MKs MK's
2392
+ MLAs MLA's
2393
+ mlc MLC
2394
+ MLCs MLC's
2395
+ mlg MLG
2396
+ mli MLI
2397
+ Mlle MLLE
2398
+ M&L M and L
2399
+ M. L. ML
2400
+ M.L. ML
2401
+ MLPs MLP's
2402
+ MM&A MMANDA
2403
+ M&M M and M
2404
+ M. M. MM
2405
+ M.M. MM
2406
+ M&M's M and M's
2407
+ MNCs MNC's
2408
+ M&NF M and NF
2409
+ M. N. MN
2410
+ M.N. MN
2411
+ M.O.G.U.E.R.A.'s MOGUERA's
2412
+ M. O. MO
2413
+ M.O. MO
2414
+ MpA MPA
2415
+ MPAs MPA's
2416
+ MPBu MPBU
2417
+ MP&I MP and I
2418
+ MPi MPI
2419
+ MPLMs MPLM's
2420
+ M&P M and P
2421
+ M. P. MP
2422
+ M.P. MP
2423
+ MPPs MPP's
2424
+ M.P.'s MP's
2425
+ MPs MP's
2426
+ MPThe MPTHE
2427
+ Mpu MPU
2428
+ M. Q. MQ
2429
+ Mra MRA
2430
+ MRBs MRB's
2431
+ mre MRE
2432
+ MR&LE MR and LE
2433
+ Mr. mister
2434
+ M. R. MR
2435
+ M.R. MR
2436
+ mRNA MRNA
2437
+ mRNAs MRNA's
2438
+ Mrs. misses
2439
+ MRTs MRT's
2440
+ msd MSD
2441
+ Mse MSE
2442
+ M&S M and S
2443
+ Ms. miss
2444
+ M. S. MS
2445
+ M.S. MS
2446
+ MSPs MSP's
2447
+ mst MST
2448
+ Mta MTA
2449
+ MTAs MTA's
2450
+ M&T M and T
2451
+ M. T. MT
2452
+ M.T. MT
2453
+ Mtor MTOR
2454
+ mtvU MTVU
2455
+ MTVu MTVU
2456
+ M. U. MU
2457
+ M.U.s MU's
2458
+ MVMs MVM's
2459
+ M. V. MV
2460
+ M.V. MV
2461
+ MVPs MVP's
2462
+ MVs MV's
2463
+ MWe MWE
2464
+ M. W. MW
2465
+ M.W. MW
2466
+ M. X. MX
2467
+ myb MYB
2468
+ myc MYC
2469
+ Myc MYC
2470
+ MyDD MYDD
2471
+ M. Y. MY
2472
+ M.Y. MY
2473
+ MySQL MYSQL
2474
+ M. Z. MZ
2475
+ N. A. NA
2476
+ N.A. NA
2477
+ N.B.A.'s NBA's
2478
+ NBAs NBA's
2479
+ N. B. NB
2480
+ N.B. NB
2481
+ nbs NBS
2482
+ NBTwo NBTWO
2483
+ NCAAs NCAA's
2484
+ NCCs NCC's
2485
+ N. C. NC
2486
+ N.C. NC
2487
+ NCOs NCO's
2488
+ N. D. ND
2489
+ N.D. ND
2490
+ N. E. NE
2491
+ N.E. NE
2492
+ ner NER
2493
+ NFATc NFATC
2494
+ nfed NFED
2495
+ N. F. NF
2496
+ N.F. NF
2497
+ Nge NGE
2498
+ N. G. NG
2499
+ N.G. NG
2500
+ NGOs NGO's
2501
+ Nha NHA
2502
+ NHCEs NHCE's
2503
+ NHCs NHC's
2504
+ nhi NHI
2505
+ NHM&W NHM and W
2506
+ N. H. NH
2507
+ N.H. NH
2508
+ NHPs NHP's
2509
+ N. I. NI
2510
+ N.I. NI
2511
+ N'I NI
2512
+ N. J. NJ
2513
+ N.J. NJ
2514
+ Njoo NJOO
2515
+ N.J.'s NJ's
2516
+ N. K. NK
2517
+ N.K. NK
2518
+ nkvd NKVD
2519
+ nkv NKV
2520
+ N. L. NL
2521
+ N.L. NL
2522
+ NLRs NLR's
2523
+ N. M. NM
2524
+ N.M. NM
2525
+ N. N. NN
2526
+ N.N. NN
2527
+ NPCs NPC's
2528
+ NPMs NPM's
2529
+ N. P. NP
2530
+ N.P. NP
2531
+ NPs NP's
2532
+ N. Q. NQ
2533
+ nri NRI
2534
+ Nri NRI
2535
+ NRIs NRI's
2536
+ NRJs NRJ's
2537
+ nr NR
2538
+ N. R. NR
2539
+ N.R. NR
2540
+ N. S. NS
2541
+ N.S. NS
2542
+ NSOs NSO's
2543
+ Nta NTA
2544
+ ntb NTB
2545
+ N. T. NT
2546
+ N.T. NT
2547
+ ntw NTW
2548
+ N.U. NU
2549
+ NvDA's NVDA's
2550
+ N. V. NV
2551
+ N.V. NV
2552
+ nwa NWA
2553
+ N&W N and W
2554
+ N. W. NW
2555
+ N.W. NW
2556
+ nyc NYC
2557
+ N. Y. NY
2558
+ N.Y. NY
2559
+ NX-SOC NXSOC
2560
+ N'Zif NZIF
2561
+ NZiK NZIK
2562
+ N'Zi NZI
2563
+ N.Z. NZ
2564
+ O. A. OA
2565
+ O.A. OA
2566
+ obl OBL
2567
+ Oblt OBLT
2568
+ O.B. OB
2569
+ O. C. OC
2570
+ O.C. OC
2571
+ Octl OCTL
2572
+ O. D. OD
2573
+ O.E. OE
2574
+ O.F. OF
2575
+ ofr OFR
2576
+ O&G O and G
2577
+ O. G. OG
2578
+ O.G. OG
2579
+ O. H. OH
2580
+ O.H. OH
2581
+ O. I. OI
2582
+ O.I. OI
2583
+ O. J. OJ
2584
+ O.J. OJ
2585
+ OK'd OKD
2586
+ OKd OKD
2587
+ oke OKE
2588
+ Oke OKE
2589
+ O&K O and K
2590
+ O. K. OK
2591
+ O.K. OK
2592
+ O. L. OL
2593
+ O.L. OL
2594
+ Olo OLO
2595
+ Olov OLOV
2596
+ ols OLS
2597
+ OLs OL's
2598
+ olvwm OLVWM
2599
+ olwm OLWM
2600
+ O&MFL O and MFL
2601
+ omg OMG
2602
+ OMGs OMG's
2603
+ O. M. OM
2604
+ O.M. OM
2605
+ OmOm OMOM
2606
+ Om's OM's
2607
+ oncu ONCU
2608
+ ond OND
2609
+ onf ONF
2610
+ O. N. ON
2611
+ O.N. ON
2612
+ O&O O and O
2613
+ O. O. OO
2614
+ O.O. OO
2615
+ Oop OOP
2616
+ oor OOR
2617
+ oose OOSE
2618
+ Oo's OO's
2619
+ Ootw OOTW
2620
+ OPMs OPM's
2621
+ O. P. OP
2622
+ O.P. OP
2623
+ opr OPR
2624
+ Opr OPR
2625
+ ORFs ORF's
2626
+ OR&N OR and N
2627
+ O. R. OR
2628
+ osaa OSAA
2629
+ osa OSA
2630
+ Osa OSA
2631
+ OSBs OSB's
2632
+ O. S. OS
2633
+ O.S. OS
2634
+ OSTs OST's
2635
+ osv OSV
2636
+ O.T. OT
2637
+ OTs OT's
2638
+ Otu OTU
2639
+ otv OTV
2640
+ O.U. OU
2641
+ ovca OVCA
2642
+ Ovca OVCA
2643
+ Ovda OVDA
2644
+ Ovo OVO
2645
+ ov OV
2646
+ Ov OV
2647
+ O. V. OV
2648
+ O.V. OV
2649
+ OWCs OWC's
2650
+ O. W. OW
2651
+ O.W. OW
2652
+ pa PA
2653
+ P. A. PA
2654
+ P.A. PA
2655
+ pBCE PBCE
2656
+ pbc PBC
2657
+ pb PB
2658
+ P. B. PB
2659
+ P.B. PB
2660
+ PCBs PCB's
2661
+ PCDDs PCDD's
2662
+ PCeU PCEU
2663
+ PCIe PCIE
2664
+ pci PCI
2665
+ pcl PCL
2666
+ PcoA PCOA
2667
+ pc PC
2668
+ P. C. PC
2669
+ P.C. PC
2670
+ PCRev PCREV
2671
+ pcs PCS
2672
+ P.C.s PC's
2673
+ pDAB PDAB
2674
+ PDAs PDA's
2675
+ pdbp PDBP
2676
+ PDBsum PDBSUM
2677
+ PDCs PDC's
2678
+ PD&D PD and D
2679
+ PDEs PDE's
2680
+ pdf PDF
2681
+ PDFs PDF's
2682
+ P&D P and D
2683
+ P. D. PD
2684
+ P.D. PD
2685
+ PDPs PDP's
2686
+ PDs PD's
2687
+ P.E.I.'s PEI's
2688
+ P. E. PE
2689
+ P.E. PE
2690
+ PFs PF's
2691
+ PFW&C PFW and C
2692
+ PGMs PGM's
2693
+ PG&N PG and N
2694
+ P&G P and G
2695
+ P. G. PG
2696
+ P.G. PG
2697
+ Phlo PHLO
2698
+ P&H P and H
2699
+ P. H. PH
2700
+ P.H. PH
2701
+ php PHP
2702
+ PHPs PHP's
2703
+ P&I P and I
2704
+ P. I. PI
2705
+ P.I. PI
2706
+ Piz PIZ
2707
+ pj PJ
2708
+ P. J. PJ
2709
+ P.J. PJ
2710
+ PJs PJ's
2711
+ P. K. PK
2712
+ P.K. PK
2713
+ plc PLC
2714
+ PLCs PLC's
2715
+ P. L. PL
2716
+ P.L. PL
2717
+ PMMoV PMMOV
2718
+ p.m. PM
2719
+ p.m PM
2720
+ P. M. PM
2721
+ P.M. PM
2722
+ PMs PM's
2723
+ PNaCl PNACL
2724
+ pna PNA
2725
+ Pnau PNAU
2726
+ pnb PNB
2727
+ PNEs PNE's
2728
+ PNETs PNET's
2729
+ pneus PNEUS
2730
+ png PNG
2731
+ PNNs PNN's
2732
+ PNoy PNOY
2733
+ pn PN
2734
+ P. N. PN
2735
+ P.N. PN
2736
+ P.O.D.'s POD's
2737
+ P&O P and O
2738
+ P. O. PO
2739
+ P.O. PO
2740
+ P.O.W.'s POW's
2741
+ P&PH P and PH
2742
+ PPi PPI
2743
+ P. P. PP
2744
+ P.P. PP
2745
+ PPVs PPV's
2746
+ PPy PPY
2747
+ pr PR
2748
+ P. R. PR
2749
+ P.R. PR
2750
+ P. s P's
2751
+ P.'s P's
2752
+ Ps P's
2753
+ P. S. PS
2754
+ P.S. PS
2755
+ PSSAs PSSA's
2756
+ pTA PTA
2757
+ PTAs PTA's
2758
+ Pte PTE
2759
+ PTEs PTE's
2760
+ ptf PTF
2761
+ Ptie PTIE
2762
+ PTLs PTL's
2763
+ Ptol PTOL
2764
+ pt PT
2765
+ P. T. PT
2766
+ P.T. PT
2767
+ PTTs PTT's
2768
+ P'Twa PTWA
2769
+ Puiu PUIU
2770
+ Pul PUL
2771
+ PVAs PVA's
2772
+ PvdA PVDA
2773
+ pve PVE
2774
+ P. V. PV
2775
+ P.V. PV
2776
+ pvr PVR
2777
+ PVs PV's
2778
+ pwll PWLL
2779
+ P&W P and W
2780
+ P. W. PW
2781
+ P.W. PW
2782
+ Pyi PYI
2783
+ Pyk PYK
2784
+ Pyl PYL
2785
+ PyL PYL
2786
+ P. Y. PY
2787
+ Pyu PYU
2788
+ Pyw PYW
2789
+ Pyx PYX
2790
+ P. Z. PZ
2791
+ Q. A. QA
2792
+ Q&A QANDA
2793
+ Q&A's QANDA's
2794
+ Q&As QANDAS
2795
+ QbA QBA
2796
+ Q.B. QB
2797
+ Q.C. QC
2798
+ Q. E. QE
2799
+ Q.H. QH
2800
+ Q.I. QI
2801
+ Q. J. QJ
2802
+ Q. N. QN
2803
+ QPOs QPO's
2804
+ Q.V. QV
2805
+ R. A. RA
2806
+ R.A. RA
2807
+ R&A R and A
2808
+ R&AW R and AW
2809
+ RbAg RBAG
2810
+ RBCs RBC's
2811
+ RBIs RBI's
2812
+ R&B R and B
2813
+ R. B. RB
2814
+ R.B. RB
2815
+ rca RCA
2816
+ R&C R and C
2817
+ rc RC
2818
+ R. C. RC
2819
+ R.C. RC
2820
+ RCTs RCT's
2821
+ RdE RDE
2822
+ R&D R and D
2823
+ R. D. RD
2824
+ R.D. RD
2825
+ R.E.M.'s REM's
2826
+ REPLs REPL's
2827
+ R. E. RE
2828
+ R.E. RE
2829
+ R. F. RF
2830
+ R.F. RF
2831
+ R&G R and G
2832
+ R. G. RG
2833
+ R.G. RG
2834
+ RHIBs RHIB's
2835
+ Rhiw RHIW
2836
+ rhl RHL
2837
+ RhoG RHOG
2838
+ Rho's RHO's
2839
+ R. H. RH
2840
+ R.H. RH
2841
+ R&I R and I
2842
+ R. I. RI
2843
+ R.I. RI
2844
+ riu RIU
2845
+ Riu RIU
2846
+ rivs RIVS
2847
+ Rivu RIVU
2848
+ Rixt RIXT
2849
+ rjf RJF
2850
+ R. J. RJ
2851
+ R.J. RJ
2852
+ R. K. RK
2853
+ R.K. RK
2854
+ R. L. RL
2855
+ R.L. RL
2856
+ rly RLY
2857
+ RMDs RMD's
2858
+ RMLs RML's
2859
+ rm RM
2860
+ R. M. RM
2861
+ R.M. RM
2862
+ RMSDs RMSD's
2863
+ RMs RM's
2864
+ rna RNA
2865
+ RNAs RNA's
2866
+ Rnet RNET
2867
+ Rnic RNIC
2868
+ RNNs RNN's
2869
+ RNPs RNP's
2870
+ R. N. RN
2871
+ R.N. RN
2872
+ rOmpB ROMPB
2873
+ Rooi ROOI
2874
+ R. O. RO
2875
+ R.O. RO
2876
+ ROVs ROV's
2877
+ Roxb ROXB
2878
+ Roxx ROXX
2879
+ RP&C RP and C
2880
+ rpc RPC
2881
+ Rpe RPE
2882
+ rpg RPG
2883
+ RPGs RPG's
2884
+ rpm RPM
2885
+ R. P. RP
2886
+ R.P. RP
2887
+ RPs RP's
2888
+ R. Q. RQ
2889
+ R.Q. RQ
2890
+ rra RRA
2891
+ rrd RRD
2892
+ Rreli RRELI
2893
+ rre RRE
2894
+ rRNA RRNA
2895
+ rRNAs RRNA's
2896
+ R&R R and R
2897
+ rr RR
2898
+ R. R. RR
2899
+ R.R. RR
2900
+ rsh RSH
2901
+ Rsis RSI's
2902
+ rsly RSLY
2903
+ RsmA RSMA
2904
+ RSpec RSPEC
2905
+ rs RS
2906
+ Rs R's
2907
+ R. S. RS
2908
+ R.S. RS
2909
+ RSu RSU
2910
+ rtb RTB
2911
+ RTCs RTC's
2912
+ RTEjr RTEJR
2913
+ RTeOR RTEOR
2914
+ RTOs RTO's
2915
+ rtPA RTPA
2916
+ rt RT
2917
+ R. T. RT
2918
+ R.T. RT
2919
+ rts RTS
2920
+ RTVFBiH RTVFBIH
2921
+ Ruao RUAO
2922
+ Rukn RUKN
2923
+ Ruk RUK
2924
+ Rupf RUPF
2925
+ Rupr RUPR
2926
+ rup RUP
2927
+ Rup RUP
2928
+ R. U. RU
2929
+ R.U. RU
2930
+ RUN:AI RUN AI
2931
+ RutB RUTB
2932
+ Ruu RUU
2933
+ Ruwa RUWA
2934
+ Ruy RUY
2935
+ Ruyt RUYT
2936
+ Rvat RVAT
2937
+ Rvo RVO
2938
+ R. V. RV
2939
+ R.V. RV
2940
+ rvs RVS
2941
+ rwa RWA
2942
+ Rwa RWA
2943
+ rwb RWB
2944
+ RWEs RWE's
2945
+ rwjf RWJF
2946
+ R. W. RW
2947
+ R.W. RW
2948
+ Rxa RXA
2949
+ rz RZ
2950
+ R. Z. RZ
2951
+ Sa'id SAID
2952
+ S. A. SA
2953
+ S.A. SA
2954
+ Saxl SAXL
2955
+ Sa'yo SAYO
2956
+ Sbai SBAI
2957
+ sbc SBC
2958
+ SBCs SBC's
2959
+ sbk SBK
2960
+ Sborz SBORZ
2961
+ sb SB
2962
+ S. B. SB
2963
+ S.B. SB
2964
+ Sbu SBU
2965
+ scr SCR
2966
+ SCRs SCR's
2967
+ scry SCRY
2968
+ sc SC
2969
+ S. C. SC
2970
+ S.C. SC
2971
+ Scuf SCUF
2972
+ Scymn SCYMN
2973
+ SD&AE SD and AE
2974
+ Sdei SDEI
2975
+ sde SDE
2976
+ Sde SDE
2977
+ SDG&E SDG and E
2978
+ sdk SDK
2979
+ S. D. SD
2980
+ S.D. SD
2981
+ SDSM&T's SDSM and T's
2982
+ Sejms SEJM's
2983
+ sejr SEJR
2984
+ Sekl SEKL
2985
+ Sek SEK
2986
+ S. E. SE
2987
+ S.E. SE
2988
+ ses SES
2989
+ Ses SE's
2990
+ SEs SE's
2991
+ S.E.S.'s SES's
2992
+ Seyh SEYH
2993
+ sfadb SFADB
2994
+ Sfax SFAX
2995
+ SFFCo SFFCO
2996
+ sfn SFN
2997
+ 'sf SF
2998
+ sf SF
2999
+ S. F. SF
3000
+ S.F. SF
3001
+ sfs SFS
3002
+ S.F.'s SF's
3003
+ SG&A SGANDA
3004
+ sgb SGB
3005
+ Sgip SGIP
3006
+ sgml SGML
3007
+ sgra SGRA
3008
+ S. G. SG
3009
+ S.G. SG
3010
+ SGSNs SGSN's
3011
+ Sgt. sergeant
3012
+ sgt SGT
3013
+ S.H.I.E.L.D.'s SHIELD's
3014
+ Shma SHMA
3015
+ S. H. SH
3016
+ S.H. SH
3017
+ S. I. SI
3018
+ S.I. SI
3019
+ sjef SJEF
3020
+ Sjon SJON
3021
+ S. J. SJ
3022
+ S.J. SJ
3023
+ S. K. SK
3024
+ S.K. SK
3025
+ S.L.A.A.'s SLAA's
3026
+ sla SLA
3027
+ SLAs SLA's
3028
+ S&L S and L
3029
+ SLS&E SLS and E
3030
+ S. L. SL
3031
+ S.L. SL
3032
+ SMe SME
3033
+ SMEs SME's
3034
+ SmI SMI
3035
+ SMPSs SMPS's
3036
+ smr SMR
3037
+ S&M S and M
3038
+ sm SM
3039
+ S. M. SM
3040
+ S.M. SM
3041
+ S.M.'s SM's
3042
+ SMs SMS
3043
+ Smyl SMYL
3044
+ SNESjr SNESJR
3045
+ sngle SNGLE
3046
+ SNPs SNP's
3047
+ SnSe SNSE
3048
+ S. N. SN
3049
+ S.N. SN
3050
+ S. O. SO
3051
+ S.O. SO
3052
+ Sos SO's
3053
+ Sovn SOVN
3054
+ Sov SOV
3055
+ S&P 500 S and P five hundred
3056
+ SPCAs SPCA's
3057
+ Spe SPE
3058
+ Spoa SPOA
3059
+ SpPIn SPPIN
3060
+ S&P S and P
3061
+ S. P. SP
3062
+ S.P. SP
3063
+ SPs SP's
3064
+ SPUs SPU's
3065
+ sql SQL
3066
+ sq SQ
3067
+ S. Q. SQ
3068
+ Sra SRA
3069
+ SRAs SRA's
3070
+ Srba SRBA
3071
+ Srbi SRBI
3072
+ SRBs SRB's
3073
+ SRGs SRG's
3074
+ Srhir SRHIR
3075
+ SRLGs SRLG's
3076
+ SRMs SRM's
3077
+ Srni SRNI
3078
+ srp SRP
3079
+ sr SR
3080
+ S. R. SR
3081
+ S.R. SR
3082
+ Srul SRUL
3083
+ srx SRX
3084
+ SSAs SSA's
3085
+ SSDs SSD's
3086
+ ssl SSL
3087
+ SSoSV SSOSV
3088
+ SSRIs SSRI's
3089
+ S&S S and S
3090
+ ss SS
3091
+ S. s S's
3092
+ Ss S's
3093
+ S. S. SS
3094
+ S.S. SS
3095
+ Ssu SSU
3096
+ ssw SSW
3097
+ Ssy SSY
3098
+ ST&AJ ST and AJ
3099
+ STDs STD's
3100
+ stfv STFV
3101
+ STG-TWO STGTWO
3102
+ STGs STG's
3103
+ STIs STI's
3104
+ STi STI
3105
+ stl STL
3106
+ Stryj STRYJ
3107
+ S&T S and T
3108
+ STScI STSCI
3109
+ S. T. ST
3110
+ S.T. ST
3111
+ suo SUO
3112
+ Suo SUO
3113
+ S. U. SU
3114
+ Susz SUSZ
3115
+ SUTs SUT's
3116
+ Suu SUU
3117
+ SUVs SUV's
3118
+ Suy SUY
3119
+ Svac SVAC
3120
+ svar SVAR
3121
+ sve SVE
3122
+ Sve SVE
3123
+ SVMs SVM's
3124
+ svn SVN
3125
+ svom SVOM
3126
+ sv SV
3127
+ S. V. SV
3128
+ S.V. SV
3129
+ SVTs SVT's
3130
+ Swe SWE
3131
+ SwRI SWRI
3132
+ S&W S and W
3133
+ sw SW
3134
+ S. W. SW
3135
+ S.W. SW
3136
+ sx SX
3137
+ S.X. SX
3138
+ SysML SYSML
3139
+ SysRq SYSRQ
3140
+ Sys SY's
3141
+ SysV SYSV
3142
+ S. Y. SY
3143
+ S.Y. SY
3144
+ SyT SYT
3145
+ syv SYV
3146
+ Syxx SYXX
3147
+ Szasz SZASZ
3148
+ Szer SZER
3149
+ szkic SZKIC
3150
+ Szklo SZKLO
3151
+ Szlak SZLAK
3152
+ SzMME SZMME
3153
+ Szpir SZPIR
3154
+ sz SZ
3155
+ S. Z. SZ
3156
+ Szu SZU
3157
+ Szyk SZYK
3158
+ taf TAF
3159
+ Taf TAF
3160
+ T. A. TA
3161
+ T.A. TA
3162
+ T&A T and A
3163
+ TBCs TBC's
3164
+ TBMs TBM's
3165
+ tbh TBH
3166
+ T. B. TB
3167
+ T.B. TB
3168
+ TCiAP TCIAP
3169
+ TCKs TCK's
3170
+ tc TC
3171
+ T. C. TC
3172
+ T.C. TC
3173
+ TCUs TCU's
3174
+ tcu TCU
3175
+ tdb TDB
3176
+ TDCi TDCI
3177
+ TdIF TDIF
3178
+ TDs TD's
3179
+ T. D. TD
3180
+ T.D. TD
3181
+ Teatr TEATR
3182
+ T. E. TE
3183
+ T.E. TE
3184
+ T&F T and F
3185
+ tf TF
3186
+ T. F. TF
3187
+ T.F. TF
3188
+ tge TGE
3189
+ tgf TGF
3190
+ tg TG
3191
+ T. G. TG
3192
+ T.G. TG
3193
+ TGVs TGV's
3194
+ tgv TGV
3195
+ Thok THOK
3196
+ ThSe THSE
3197
+ T. H. TH
3198
+ T.H. TH
3199
+ T.I.'s TI's
3200
+ T.I. TI
3201
+ Tiu TIU
3202
+ tiv TIV
3203
+ Tiv TIV
3204
+ Tiy's TIY's
3205
+ T. J. TJ
3206
+ T.J. TJ
3207
+ t'ju TJU
3208
+ tjz TJZ
3209
+ tko TKO
3210
+ Tko TKO
3211
+ tk TK
3212
+ T. K. TK
3213
+ T.K. TK
3214
+ Tlas TLA's
3215
+ TlCu TLCU
3216
+ TLDs TLD's
3217
+ tli TLI
3218
+ tlp TLP
3219
+ TLRs TLR's
3220
+ TLs TL's
3221
+ tl TL
3222
+ T. L. TL
3223
+ T.L. TL
3224
+ Tluk TLUK
3225
+ Tluszcz TLUSZCZ
3226
+ tmc TMC
3227
+ TMGs TMG's
3228
+ tmos TMOS
3229
+ tmRNA TMRNA
3230
+ TMSs TMS's
3231
+ T.M.s TM's
3232
+ tm TM
3233
+ T. M. TM
3234
+ T.M. TM
3235
+ tna TNA
3236
+ tnbc TNBC
3237
+ TNTAs TNTA's
3238
+ tn TN
3239
+ T. N. TN
3240
+ T.N. TN
3241
+ T.O.'s TO's
3242
+ T. O. TO
3243
+ T.O. TO
3244
+ tou TOU
3245
+ Tou TOU
3246
+ ToU TOU
3247
+ tPA TPA
3248
+ tpr TPR
3249
+ tp TP
3250
+ T. P. TP
3251
+ T.P. TP
3252
+ TPVs TPV's
3253
+ T.Q. TQ
3254
+ tra TRA
3255
+ Tra TRA
3256
+ Trbic TRBIC
3257
+ Trcek TRCEK
3258
+ TrkA TRKA
3259
+ Trmcic TRMCIC
3260
+ tRNA TRNA
3261
+ TRPs TRP's
3262
+ trs TRS
3263
+ tr TR
3264
+ T. R. TR
3265
+ T.R. TR
3266
+ truTV TRUTV
3267
+ TruTV TRUTV
3268
+ Trve TRVE
3269
+ Tsa TSA
3270
+ tsit TSIT
3271
+ Tsiv TSIV
3272
+ tso TSO
3273
+ Tso TSO
3274
+ tsr TSR
3275
+ Ts T's
3276
+ T. S. TS
3277
+ T.S. TS
3278
+ tsus TSUS
3279
+ Tsvi TSVI
3280
+ Tta TTA
3281
+ tteok TTEOK
3282
+ tte TTE
3283
+ ttp TTP
3284
+ T&T T and T
3285
+ T. T. TT
3286
+ T.T. TT
3287
+ T'uqu TUQU
3288
+ T.U. TU
3289
+ Tuzk TUZK
3290
+ tvaan TVAAN
3291
+ Tvam TVAM
3292
+ tva TVA
3293
+ TVETs TVET's
3294
+ TVii TVII
3295
+ TVIn TVIN
3296
+ TVi TVI
3297
+ tvo TVO
3298
+ Tvo TVO
3299
+ TVRi TVRI
3300
+ tvr TVR
3301
+ TVSpy TVSPY
3302
+ tvs TVS
3303
+ TVs TV's
3304
+ tVTA TVTA
3305
+ tv TV
3306
+ T. V. TV
3307
+ T.V. TV
3308
+ Twa TWA
3309
+ Twi TWI
3310
+ TWTs TWT's
3311
+ T. W. TW
3312
+ T.W. TW
3313
+ TxDOT TXDOT
3314
+ T.X. TX
3315
+ tya TYA
3316
+ Tyk TYK
3317
+ Tza'ar TZAAR
3318
+ Tze TZE
3319
+ Tzrif TZRIF
3320
+ Tzuh TZUH
3321
+ Tzvi TZVI
3322
+ Ua's UA's
3323
+ UAs UA's
3324
+ uat UAT
3325
+ U.A. UA
3326
+ UAVs UAV's
3327
+ Uba UBA
3328
+ UbcM UBCM
3329
+ Ube UBE
3330
+ ubi UBI
3331
+ Ubi UBI
3332
+ Ucar UCAR
3333
+ Uca UCA
3334
+ Ucmak UCMAK
3335
+ Ucn UCN
3336
+ uCs UC's
3337
+ Uc UC
3338
+ U.C. UC
3339
+ Ucuncu UCUNCU
3340
+ Uczta UCZTA
3341
+ Uda UDA
3342
+ Udit UDIT
3343
+ Udny UDNY
3344
+ UDTs UDT's
3345
+ ud UD
3346
+ Ud UD
3347
+ U. D. UD
3348
+ udu UDU
3349
+ Udu UDU
3350
+ Ueda's UEDA's
3351
+ Uéda UE acute DA
3352
+ U. E. UE
3353
+ UFOs UFO's
3354
+ Ufot UFOT
3355
+ ufo UFO
3356
+ Ufo UFO
3357
+ U. F. UF
3358
+ Uga UGA
3359
+ Uge UGE
3360
+ Ugni UGNI
3361
+ Ugra UGRA
3362
+ Ugrszke UGRSZKE
3363
+ Ug UG
3364
+ Uhha UHHA
3365
+ uhn UHN
3366
+ UHTCs UHTC's
3367
+ Uhud UHUD
3368
+ U.H. UH
3369
+ uhur UHUR
3370
+ Uiam UIAM
3371
+ Uibh UIBH
3372
+ Uible UIBLE
3373
+ Uig UIG
3374
+ uisae UISAE
3375
+ UiS UIS
3376
+ UiTM UITM
3377
+ uit UIT
3378
+ Uit UIT
3379
+ Uiy UIY
3380
+ Ujed UJED
3381
+ Ujsag UJSAG
3382
+ uj UJ
3383
+ U.J. UJ
3384
+ Uka UKA
3385
+ Ukhra UKHRA
3386
+ Ukic UKIC
3387
+ Uki UKI
3388
+ Ukoh UKOH
3389
+ Uko UKO
3390
+ ukr UKR
3391
+ Ukr UKR
3392
+ Ukui UKUI
3393
+ u'k UK
3394
+ uk UK
3395
+ Uk UK
3396
+ U.K. UK
3397
+ Ukwu UKWU
3398
+ Ulic ULIC
3399
+ Ull ULL
3400
+ Ulms ULM's
3401
+ UlSU ULSU
3402
+ Uluj ULUJ
3403
+ U.L. UL
3404
+ ulus ULUS
3405
+ Ulus ULU's
3406
+ Ulwa ULWA
3407
+ Ulwe ULWE
3408
+ Umbr UMBR
3409
+ umelcu UMELCU
3410
+ UMe UME
3411
+ Umla UMLA
3412
+ Umme UMME
3413
+ Umno UMNO
3414
+ Umri UMRI
3415
+ U. M. UM
3416
+ U.M. UM
3417
+ Unli UNLI
3418
+ Unlu UNLU
3419
+ unm UNM
3420
+ unnd UNND
3421
+ UNSh UNSH
3422
+ Unst UNST
3423
+ Uns UN's
3424
+ U.N.'s UN's
3425
+ Unt UNT
3426
+ U.N. UN
3427
+ Unz UNZ
3428
+ UOCl UOCL
3429
+ UofM UOFM
3430
+ Uoho UOHO
3431
+ UoMs UOM's
3432
+ uORF UORF
3433
+ UoW UOW
3434
+ UPBs UPB's
3435
+ Upd UPD
3436
+ Upf UPF
3437
+ Upir UPIR
3438
+ UPnP UPNP
3439
+ Uppu UPPU
3440
+ U.P. UP
3441
+ UPyD UPYD
3442
+ uq UQ
3443
+ Urbz URBZ
3444
+ URCs URC's
3445
+ Urdd URDD
3446
+ Urei UREI
3447
+ urf URF
3448
+ Urla URLA
3449
+ URLhttp URLHTTP
3450
+ URLs URL's
3451
+ Usal USAL
3452
+ usan USAN
3453
+ Usan USAN
3454
+ usao USAO
3455
+ usata USATA
3456
+ usa USA
3457
+ .USA USA
3458
+ USAya USAYA
3459
+ USBs USB's
3460
+ Uscie USCIE
3461
+ U.S.C.'s USC's
3462
+ USDoE USDOE
3463
+ usd USD
3464
+ USFbA USFBA
3465
+ USF&WS USF and WS
3466
+ UShs USH's
3467
+ Usia USIA
3468
+ usih USIH
3469
+ Uslu USLU
3470
+ Usmar USMAR
3471
+ Usna USNA
3472
+ Usnic USNIC
3473
+ Usoi USOI
3474
+ Usos USO's
3475
+ Uso USO
3476
+ USPs USP's
3477
+ U.S.'s US's
3478
+ Usti USTI
3479
+ Ustka USTKA
3480
+ Usui USUI
3481
+ usum USUM
3482
+ U. S. US
3483
+ U.S. US
3484
+ Usut USUT
3485
+ Usvit USVIT
3486
+ Uta's UTA's
3487
+ uta UTA
3488
+ Uta UTA
3489
+ Utca UTCA
3490
+ Utd's UTD's
3491
+ Utd UTD
3492
+ utea UTEA
3493
+ Utes UTE's
3494
+ Uth UTH
3495
+ uti UTI
3496
+ Uti UTI
3497
+ Utne UTNE
3498
+ Utnur UTNUR
3499
+ Uto UTO
3500
+ utrci UTRCI
3501
+ utsav UTSAV
3502
+ Utsav UTSAV
3503
+ Utsu UTSU
3504
+ U.T. UT
3505
+ Utu UTU
3506
+ Utva UTVA
3507
+ Uuh UUH
3508
+ Uul UUL
3509
+ Uulu UULU
3510
+ Uusi UUSI
3511
+ UUs UU's
3512
+ Uuto UUTO
3513
+ Uvac UVAC
3514
+ Uvea UVEA
3515
+ uvnitr UVNITR
3516
+ Uvo UVO
3517
+ U.V. UV
3518
+ Uwasa UWASA
3519
+ UWFi UWFI
3520
+ V. A. VA
3521
+ V.A. VA
3522
+ V&A V and A
3523
+ Vav VAV
3524
+ Vay VAY
3525
+ vb VB
3526
+ V. B. VB
3527
+ V.B. VB
3528
+ VCDs VCD's
3529
+ VCRs VCR's
3530
+ VCSELs VCSEL's
3531
+ VCs VC's
3532
+ VCTs VCT's
3533
+ vc VC
3534
+ V. C. VC
3535
+ V.C. VC
3536
+ vda VDA
3537
+ Vda VDA
3538
+ VDCs VDC's
3539
+ vdiq VDIQ
3540
+ vdm VDM
3541
+ V. D. VD
3542
+ V.D. VD
3543
+ vez VEZ
3544
+ Vez VEZ
3545
+ vfp VFP
3546
+ vfr VFR
3547
+ V. F. VF
3548
+ V.F. VF
3549
+ V'Ger VGER
3550
+ vgmdb VGMDB
3551
+ VGo VGO
3552
+ VGSCs VGSC's
3553
+ VGSoM VGSOM
3554
+ V. G. VG
3555
+ V.G. VG
3556
+ V. H. VH
3557
+ V.I.C.'s VIC's
3558
+ Vict VICT
3559
+ viita VIITA
3560
+ vijf VIJF
3561
+ vij VIJ
3562
+ Vij VIJ
3563
+ V.I.P.s VIP's
3564
+ VIPs VIP's
3565
+ V. I. VI
3566
+ V.I. VI
3567
+ VJs VJ's
3568
+ V. J. VJ
3569
+ V.J. VJ
3570
+ V. K. VK
3571
+ V.K. VK
3572
+ Vlah VLAH
3573
+ VLCCs VLCC's
3574
+ vlei VLEI
3575
+ Vlijt VLIJT
3576
+ V. L. VL
3577
+ V.L. VL
3578
+ VMAs VMA's
3579
+ vm VM
3580
+ V. M. VM
3581
+ V.M. VM
3582
+ vner VNER
3583
+ V. N. VN
3584
+ V.N. VN
3585
+ V.O. VO
3586
+ Vov VOV
3587
+ Voz VOZ
3588
+ vpis VPIS
3589
+ VPNs VPN's
3590
+ vpu VPU
3591
+ V. P. VP
3592
+ V.P. VP
3593
+ vq VQ
3594
+ vrak VRAK
3595
+ Vrba VRBA
3596
+ Vrbuv VRBUV
3597
+ Vrej VREJ
3598
+ vrem VREM
3599
+ Vrin VRIN
3600
+ vrj VRJ
3601
+ vrn VRN
3602
+ vroee VROEE
3603
+ vrou VROU
3604
+ vrouw VROUW
3605
+ Vrouw VROUW
3606
+ Vrsac VRSAC
3607
+ Vrtis VRTI's
3608
+ V&R V and R
3609
+ vr VR
3610
+ V. R. VR
3611
+ V.R. VR
3612
+ VSANs VSAN's
3613
+ VSATs VSAT's
3614
+ Vsekh VSEKH
3615
+ vse VSE
3616
+ vso VSO
3617
+ VSPs VSP's
3618
+ vs. versus
3619
+ _vs._ versus
3620
+ vsyo VSYO
3621
+ VTE VT eL
3622
+ Vtic VTIC
3623
+ VTi VTI
3624
+ V&T's V and T's
3625
+ V&T V and T
3626
+ V.T. VT
3627
+ Vuur VUUR
3628
+ VVIPs VVIP's
3629
+ V. V. VV
3630
+ V.V. VV
3631
+ VWs VW's
3632
+ Vyg VYG
3633
+ vyr VYR
3634
+ vy VY
3635
+ vz VZ
3636
+ V. Z. VZ
3637
+ Waay WAAY
3638
+ Wa'il WAIL
3639
+ Wakf WAKF
3640
+ wa'l WAL
3641
+ waqf WAQF
3642
+ waqt WAQT
3643
+ Waqt WAQT
3644
+ Wasl WASL
3645
+ Watfa WATFA
3646
+ wau WAU
3647
+ Wau WAU
3648
+ W. A. WA
3649
+ W.A. WA
3650
+ waw WAW
3651
+ waza WAZA
3652
+ Waza WAZA
3653
+ WBs WB's
3654
+ W. B. WB
3655
+ W.B. WB
3656
+ W&CBR W and CBR
3657
+ wc WC
3658
+ W. C. WC
3659
+ W.C. WC
3660
+ W. D. WD
3661
+ W.D. WD
3662
+ W. E. WE
3663
+ W.E. WE
3664
+ WF&NW WF and NW
3665
+ W&F W and F
3666
+ W. F. WF
3667
+ W.F. WF
3668
+ wga WGA
3669
+ wgn WGN
3670
+ wg WG
3671
+ W. G. WG
3672
+ W.G. WG
3673
+ W&H W and H
3674
+ W. H. WH
3675
+ W.H. WH
3676
+ Whyld WHYLD
3677
+ Wica WICA
3678
+ wici WICI
3679
+ Wif WIF
3680
+ Wijk WIJK
3681
+ Wiwa WIWA
3682
+ W. I. WI
3683
+ wjaz WJAZ
3684
+ W&J's W and J's
3685
+ W&J W and J
3686
+ W. J. WJ
3687
+ W.J. WJ
3688
+ W. K. WK
3689
+ W.K. WK
3690
+ W&LE W and LE
3691
+ Wley WLEY
3692
+ wlrs WLRS
3693
+ W. L. WL
3694
+ W.L. WL
3695
+ wml WML
3696
+ W&M W and M
3697
+ W. M. WM
3698
+ W.M. WM
3699
+ W. N. WN
3700
+ W.N. WN
3701
+ wnzaa WNZAA
3702
+ W&OD W and OD
3703
+ Worh WORH
3704
+ W. O. WO
3705
+ W.O. WO
3706
+ wpt WPT
3707
+ wp WP
3708
+ W. P. WP
3709
+ W.P. WP
3710
+ Wrec WREC
3711
+ W. R. WR
3712
+ W.R. WR
3713
+ ws WS
3714
+ W. S. WS
3715
+ W.S. WS
3716
+ WTs WT's
3717
+ WTTs WTT's
3718
+ W. T. WT
3719
+ W.T. WT
3720
+ WUAs WUA's
3721
+ Wudl WUDL
3722
+ Wuhr WUHR
3723
+ Wuhu WUHU
3724
+ Wu's WU's
3725
+ wuv WUV
3726
+ W. U. WU
3727
+ Wuz WUZ
3728
+ WVa WVA
3729
+ W. V. WV
3730
+ W.V. WV
3731
+ Wwe WWE
3732
+ W. W. WW
3733
+ W.W. WW
3734
+ wx WX
3735
+ W.Y. WY
3736
+ wyzc WYZC
3737
+ W. Z. WZ
3738
+ xbg XBG
3739
+ Xbra XBRA
3740
+ xCo XCO
3741
+ X. C. XC
3742
+ xda XDA
3743
+ XDRs XDR's
3744
+ X. D. XD
3745
+ Xfce XFCE
3746
+ xf XF
3747
+ X.F. XF
3748
+ xh XH
3749
+ X. H. XH
3750
+ xk XK
3751
+ xmc XMC
3752
+ xml XML
3753
+ X.M. XM
3754
+ xO XO
3755
+ X.O. XO
3756
+ XPe XPE
3757
+ X.Q. XQ
3758
+ XSi XSI
3759
+ xsr XSR
3760
+ XTwas XTWA's
3761
+ xt XT
3762
+ X. W. XW
3763
+ xy XY
3764
+ xyz XYZ
3765
+ xyZ XYZ
3766
+ Y. A. YA
3767
+ Y.A. YA
3768
+ Y. B. YB
3769
+ Y.B. YB
3770
+ ycia YCIA
3771
+ ycie YCIE
3772
+ yc YC
3773
+ Y. C. YC
3774
+ Y.C. YC
3775
+ Ydby YDBY
3776
+ Yde YDE
3777
+ YdiB YDIB
3778
+ yd YD
3779
+ Y. D. YD
3780
+ Y.D. YD
3781
+ Y.E. YE
3782
+ YFCs YFC's
3783
+ yfle YFLE
3784
+ YF&R YF and R
3785
+ yg YG
3786
+ Y. G. YG
3787
+ Y.G. YG
3788
+ Y. H. YH
3789
+ Y. I. YI
3790
+ Y.I. YI
3791
+ yj YJ
3792
+ Y. J. YJ
3793
+ Ykt YKT
3794
+ Y. K. YK
3795
+ Y.K. YK
3796
+ Yle YLE
3797
+ yl YL
3798
+ Y. L. YL
3799
+ Ymke YMKE
3800
+ ym YM
3801
+ Y. M. YM
3802
+ yndi YNDI
3803
+ Yndi YNDI
3804
+ yne YNE
3805
+ Yngve YNGVE
3806
+ Ynis YNI's
3807
+ yn YN
3808
+ Yn YN
3809
+ Y. N. YN
3810
+ Y.N. YN
3811
+ Y.O. YO
3812
+ ypa YPA
3813
+ Y.P. YP
3814
+ Y. Q. YQ
3815
+ yra YRA
3816
+ Yra YRA
3817
+ Y&R Y and R
3818
+ YSK'da YSKDA
3819
+ YSK'ya YSKYA
3820
+ Ys Y's
3821
+ Y.S. YS
3822
+ yta YTA
3823
+ Ytre YTRE
3824
+ Y&T Y and T
3825
+ Y. T. YT
3826
+ Y.T. YT
3827
+ Y. V. YV
3828
+ Y. W. YW
3829
+ Y.Y. YY
3830
+ Z. A. ZA
3831
+ Z.A. ZA
3832
+ Zdar ZDAR
3833
+ zda ZDA
3834
+ zg ZG
3835
+ Z.G. ZG
3836
+ zh ZH
3837
+ Z. H. ZH
3838
+ Z.H. ZH
3839
+ zijn ZIJN
3840
+ zij ZIJ
3841
+ Zij ZIJ
3842
+ Z. I. ZI
3843
+ Z. J. ZJ
3844
+ Z.K. ZK
3845
+ Z.L. ZL
3846
+ Z.M. ZM
3847
+ ZnO ZNO
3848
+ Zpav ZPAV
3849
+ Z.P. ZP
3850
+ ZrI ZRI
3851
+ Zsuzsa ZSUZSA
3852
+ Z.W. ZW
3853
+ Z.X. ZX
3854
+ Z. Y. ZY
3855
+ Z.Y. ZY
3856
+ Z. Z. ZZ
3857
+ Z.Z. ZZ
3858
+ 2.5G two point five g
3859
+ 7-eleven seven eleven
3860
+ AAA triple a
3861
+ C# c sharp
3862
+ DeepMind deep mind
3863
+ DeepStream deep stream
3864
+ DevOps dev ops
3865
+ DisplayPort display port
3866
+ HuggingFace hugging face
3867
+ ServiceNow service now
3868
+ SuperCloud super cloud
3869
+ Wi-Fi wifi
3870
+ 7-Eleven seven eleven
3871
+ Deepmind deep mind
3872
+ Deepstream deep stream
3873
+ Devops dev ops
3874
+ Displayport display port
3875
+ Huggingface hugging face
3876
+ Servicenow service now
3877
+ Supercloud super cloud
3878
+ wi-fi wifi
3879
+ deepmind deep mind
3880
+ deepstream deep stream
3881
+ devops dev ops
3882
+ displayport display port
3883
+ huggingface hugging face
3884
+ servicenow service now
3885
+ supercloud super cloud
3886
+ wi-fi wifi
3887
+ 3-D three d
3888
+ 401k four oh one k
3889
+ 401(k) four oh one k
3890
+ 401 (k) four oh one k
vendor/zonos3/vendor/nemo_text_processing/text_normalization/en/graph_utils.py ADDED
@@ -0,0 +1,337 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # Copyright (c) 2021, NVIDIA CORPORATION. All rights reserved.
2
+ # Copyright 2015 and onwards Google, Inc.
3
+ #
4
+ # Licensed under the Apache License, Version 2.0 (the "License");
5
+ # you may not use this file except in compliance with the License.
6
+ # You may obtain a copy of the License at
7
+ #
8
+ # http://www.apache.org/licenses/LICENSE-2.0
9
+ #
10
+ # Unless required by applicable law or agreed to in writing, software
11
+ # distributed under the License is distributed on an "AS IS" BASIS,
12
+ # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
13
+ # See the License for the specific language governing permissions and
14
+ # limitations under the License.
15
+
16
+ import os
17
+ import string
18
+ from pathlib import Path
19
+ from typing import Dict
20
+
21
+ import pynini
22
+ from pynini import Far
23
+ from pynini.examples import plurals
24
+ from pynini.export import export
25
+ from pynini.lib import byte, pynutil, utf8
26
+
27
+ from zonos3.vendor.nemo_text_processing.text_normalization.en.utils import get_abs_path, load_labels
28
+ from zonos3.vendor.nemo_text_processing.utils.logging import logger
29
+
30
+ NEMO_CHAR = utf8.VALID_UTF8_CHAR
31
+
32
+ NEMO_DIGIT = byte.DIGIT
33
+ NEMO_LOWER = pynini.union(*string.ascii_lowercase).optimize()
34
+ NEMO_UPPER = pynini.union(*string.ascii_uppercase).optimize()
35
+ NEMO_ALPHA = pynini.union(NEMO_LOWER, NEMO_UPPER).optimize()
36
+ NEMO_ALNUM = pynini.union(NEMO_DIGIT, NEMO_ALPHA).optimize()
37
+ NEMO_HEX = pynini.union(*string.hexdigits).optimize()
38
+ NEMO_NON_BREAKING_SPACE = "\u00a0"
39
+ NEMO_SPACE = " "
40
+ NEMO_WHITE_SPACE = pynini.union(" ", "\t", "\n", "\r", "\u00a0").optimize()
41
+ NEMO_NOT_SPACE = pynini.difference(NEMO_CHAR, NEMO_WHITE_SPACE).optimize()
42
+ NEMO_NOT_QUOTE = pynini.difference(NEMO_CHAR, r'"').optimize()
43
+
44
+ NEMO_PUNCT = pynini.union(*map(pynini.escape, string.punctuation)).optimize()
45
+ NEMO_GRAPH = pynini.union(NEMO_ALNUM, NEMO_PUNCT).optimize()
46
+
47
+ NEMO_SIGMA = pynini.closure(NEMO_CHAR)
48
+ NEMO_LOWER_NOT_A = pynini.union(
49
+ "b",
50
+ "c",
51
+ "d",
52
+ "e",
53
+ "f",
54
+ "g",
55
+ "h",
56
+ "i",
57
+ "j",
58
+ "k",
59
+ "l",
60
+ "m",
61
+ "n",
62
+ "o",
63
+ "p",
64
+ "q",
65
+ "r",
66
+ "s",
67
+ "t",
68
+ "u",
69
+ "v",
70
+ "w",
71
+ "x",
72
+ "y",
73
+ "z",
74
+ ).optimize()
75
+
76
+ delete_space = pynutil.delete(pynini.closure(NEMO_WHITE_SPACE))
77
+ delete_zero_or_one_space = pynutil.delete(pynini.closure(NEMO_WHITE_SPACE, 0, 1))
78
+ insert_space = pynutil.insert(" ")
79
+ delete_extra_space = pynini.cross(pynini.closure(NEMO_WHITE_SPACE, 1), " ")
80
+ delete_preserve_order = pynini.closure(
81
+ pynutil.delete(" preserve_order: true")
82
+ | (pynutil.delete(' field_order: "') + NEMO_NOT_QUOTE + pynutil.delete('"'))
83
+ )
84
+
85
+
86
+ # Common string literals; expand as you see fit.
87
+ username_string = "username"
88
+ double_quotes = '"'
89
+ domain_string = "domain"
90
+ protocol_string = "protocol"
91
+ slash = "/"
92
+ double_slash = "//"
93
+ triple_slash = "///"
94
+ file = "file"
95
+ period = "."
96
+ at = "@"
97
+ colon = ":"
98
+ https = "https"
99
+ http = "http"
100
+ www = "www"
101
+
102
+
103
+ suppletive = pynini.string_file(get_abs_path("data/suppletive.tsv"))
104
+ # _v = pynini.union("a", "e", "i", "o", "u")
105
+ _c = pynini.union(
106
+ "b",
107
+ "c",
108
+ "d",
109
+ "f",
110
+ "g",
111
+ "h",
112
+ "j",
113
+ "k",
114
+ "l",
115
+ "m",
116
+ "n",
117
+ "p",
118
+ "q",
119
+ "r",
120
+ "s",
121
+ "t",
122
+ "v",
123
+ "w",
124
+ "x",
125
+ "y",
126
+ "z",
127
+ )
128
+ _ies = NEMO_SIGMA + _c + pynini.cross("y", "ies")
129
+ _es = NEMO_SIGMA + pynini.union("s", "sh", "ch", "x", "z") + pynutil.insert("es")
130
+ _s = NEMO_SIGMA + pynutil.insert("s")
131
+
132
+ graph_plural = plurals._priority_union(
133
+ suppletive,
134
+ plurals._priority_union(_ies, plurals._priority_union(_es, _s, NEMO_SIGMA), NEMO_SIGMA),
135
+ NEMO_SIGMA,
136
+ ).optimize()
137
+
138
+ SINGULAR_TO_PLURAL = graph_plural
139
+ PLURAL_TO_SINGULAR = pynini.invert(graph_plural)
140
+ TO_LOWER = pynini.union(*[pynini.cross(x, y) for x, y in zip(string.ascii_uppercase, string.ascii_lowercase)])
141
+ TO_UPPER = pynini.invert(TO_LOWER)
142
+ MIN_NEG_WEIGHT = -0.0001
143
+ MIN_POS_WEIGHT = 0.0001
144
+ INPUT_CASED = "cased"
145
+ INPUT_LOWER_CASED = "lower_cased"
146
+ MINUS = pynini.union("minus", "Minus").optimize()
147
+
148
+
149
+ def capitalized_input_graph(
150
+ graph: "pynini.FstLike",
151
+ original_graph_weight: float = None,
152
+ capitalized_graph_weight: float = None,
153
+ ) -> "pynini.FstLike":
154
+ """
155
+ Allow graph input to be capitalized, e.g. for ITN)
156
+
157
+ Args:
158
+ graph: FstGraph
159
+ original_graph_weight: weight to add to the original `graph`
160
+ capitalized_graph_weight: weight to add to the capitalized graph
161
+ """
162
+ capitalized_graph = pynini.compose(TO_LOWER + NEMO_SIGMA, graph).optimize()
163
+
164
+ if original_graph_weight is not None:
165
+ graph = pynutil.add_weight(graph, weight=original_graph_weight)
166
+
167
+ if capitalized_graph_weight is not None:
168
+ capitalized_graph = pynutil.add_weight(capitalized_graph, weight=capitalized_graph_weight)
169
+
170
+ graph |= capitalized_graph
171
+ return graph
172
+
173
+
174
+ def generator_main(file_name: str, graphs: Dict[str, "pynini.FstLike"]):
175
+ """
176
+ Exports graph as OpenFst finite state archive (FAR) file with given file name and rule name.
177
+
178
+ Args:
179
+ file_name: exported file name
180
+ graphs: Mapping of a rule name and Pynini WFST graph to be exported
181
+ """
182
+ exporter = export.Exporter(file_name)
183
+ for rule, graph in graphs.items():
184
+ exporter[rule] = graph.optimize()
185
+ exporter.close()
186
+ logger.info(f"Created {file_name}")
187
+
188
+
189
+ def get_plurals(fst):
190
+ """
191
+ Given singular returns plurals
192
+
193
+ Args:
194
+ fst: Fst
195
+
196
+ Returns plurals to given singular forms
197
+ """
198
+ return SINGULAR_TO_PLURAL @ fst
199
+
200
+
201
+ def get_singulars(fst):
202
+ """
203
+ Given plural returns singulars
204
+
205
+ Args:
206
+ fst: Fst
207
+
208
+ Returns singulars to given plural forms
209
+ """
210
+ return PLURAL_TO_SINGULAR @ fst
211
+
212
+
213
+ def convert_space(fst) -> "pynini.FstLike":
214
+ """
215
+ Converts space to nonbreaking space.
216
+ Used only in tagger grammars for transducing token values within quotes, e.g. name: "hello kitty"
217
+ This is making transducer significantly slower, so only use when there could be potential spaces within quotes, otherwise leave it.
218
+
219
+ Args:
220
+ fst: input fst
221
+
222
+ Returns output fst where breaking spaces are converted to non breaking spaces
223
+ """
224
+ return fst @ pynini.cdrewrite(pynini.cross(NEMO_SPACE, NEMO_NON_BREAKING_SPACE), "", "", NEMO_SIGMA)
225
+
226
+
227
+ def string_map_cased(input_file: str, input_case: str = INPUT_LOWER_CASED):
228
+ labels = load_labels(input_file)
229
+
230
+ if input_case == INPUT_CASED:
231
+ additional_labels = []
232
+ for written, spoken, *weight in labels:
233
+ written_capitalized = written[0].upper() + written[1:]
234
+ additional_labels.extend(
235
+ [
236
+ [
237
+ written_capitalized,
238
+ spoken.capitalize(),
239
+ ], # first letter capitalized
240
+ [
241
+ written_capitalized,
242
+ spoken.upper().replace(" AND ", " and "),
243
+ ], # # add pairs with the all letters capitalized
244
+ ]
245
+ )
246
+
247
+ spoken_no_space = spoken.replace(" ", "")
248
+ # add abbreviations without spaces (both lower and upper case), i.e. "BMW" not "B M W"
249
+ if len(spoken) == (2 * len(spoken_no_space) - 1):
250
+ logger.debug(f"This is weight {weight}")
251
+ if len(weight) == 0:
252
+ additional_labels.extend(
253
+ [
254
+ [written, spoken_no_space],
255
+ [written_capitalized, spoken_no_space.upper()],
256
+ ]
257
+ )
258
+ else:
259
+ additional_labels.extend(
260
+ [
261
+ [written, spoken_no_space, weight[0]],
262
+ [written_capitalized, spoken_no_space.upper(), weight[0]],
263
+ ]
264
+ )
265
+ labels += additional_labels
266
+
267
+ whitelist = pynini.string_map(labels).invert().optimize()
268
+ return whitelist
269
+
270
+
271
+ class GraphFst:
272
+ """
273
+ Base class for all grammar fsts.
274
+
275
+ Args:
276
+ name: name of grammar class
277
+ kind: either 'classify' or 'verbalize'
278
+ deterministic: if True will provide a single transduction option,
279
+ for False multiple transduction are generated (used for audio-based normalization)
280
+ """
281
+
282
+ def __init__(self, name: str, kind: str, deterministic: bool = True):
283
+ self.name = name
284
+ self.kind = kind
285
+ self._fst = None
286
+ self.deterministic = deterministic
287
+
288
+ self.far_path = Path(os.path.dirname(__file__) + "/grammars/" + kind + "/" + name + ".far")
289
+ if self.far_exist():
290
+ self._fst = Far(self.far_path, mode="r", arc_type="standard", far_type="default").get_fst()
291
+
292
+ def far_exist(self) -> bool:
293
+ """
294
+ Returns true if FAR can be loaded
295
+ """
296
+ return self.far_path.exists()
297
+
298
+ @property
299
+ def fst(self) -> "pynini.FstLike":
300
+ return self._fst
301
+
302
+ @fst.setter
303
+ def fst(self, fst):
304
+ self._fst = fst
305
+
306
+ def add_tokens(self, fst) -> "pynini.FstLike":
307
+ """
308
+ Wraps class name around to given fst
309
+
310
+ Args:
311
+ fst: input fst
312
+
313
+ Returns:
314
+ Fst: fst
315
+ """
316
+ return pynutil.insert(f"{self.name} {{ ") + fst + pynutil.insert(" }")
317
+
318
+ def delete_tokens(self, fst) -> "pynini.FstLike":
319
+ """
320
+ Deletes class name wrap around output of given fst
321
+
322
+ Args:
323
+ fst: input fst
324
+
325
+ Returns:
326
+ Fst: fst
327
+ """
328
+ res = (
329
+ pynutil.delete(f"{self.name}")
330
+ + delete_space
331
+ + pynutil.delete("{")
332
+ + delete_space
333
+ + fst
334
+ + delete_space
335
+ + pynutil.delete("}")
336
+ )
337
+ return res @ pynini.cdrewrite(pynini.cross("\u00a0", " "), "", "", NEMO_SIGMA)
vendor/zonos3/vendor/nemo_text_processing/text_normalization/en/taggers/__init__.py ADDED
@@ -0,0 +1,13 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # Copyright (c) 2021, NVIDIA CORPORATION. All rights reserved.
2
+ #
3
+ # Licensed under the Apache License, Version 2.0 (the "License");
4
+ # you may not use this file except in compliance with the License.
5
+ # You may obtain a copy of the License at
6
+ #
7
+ # http://www.apache.org/licenses/LICENSE-2.0
8
+ #
9
+ # Unless required by applicable law or agreed to in writing, software
10
+ # distributed under the License is distributed on an "AS IS" BASIS,
11
+ # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
12
+ # See the License for the specific language governing permissions and
13
+ # limitations under the License.
vendor/zonos3/vendor/nemo_text_processing/text_normalization/en/taggers/abbreviation.py ADDED
@@ -0,0 +1,51 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # Copyright (c) 2021, NVIDIA CORPORATION. All rights reserved.
2
+ #
3
+ # Licensed under the Apache License, Version 2.0 (the "License");
4
+ # you may not use this file except in compliance with the License.
5
+ # You may obtain a copy of the License at
6
+ #
7
+ # http://www.apache.org/licenses/LICENSE-2.0
8
+ #
9
+ # Unless required by applicable law or agreed to in writing, software
10
+ # distributed under the License is distributed on an "AS IS" BASIS,
11
+ # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
12
+ # See the License for the specific language governing permissions and
13
+ # limitations under the License.
14
+
15
+
16
+ import pynini
17
+ from pynini.lib import pynutil
18
+
19
+ from zonos3.vendor.nemo_text_processing.text_normalization.en.graph_utils import NEMO_UPPER, GraphFst, insert_space
20
+
21
+
22
+ class AbbreviationFst(GraphFst):
23
+ """
24
+ Finite state transducer for classifying electronic: as URLs, email addresses, etc.
25
+ e.g. "ABC" -> tokens { abbreviation { value: "A B C" } }
26
+
27
+ Args:
28
+ whitelist: whitelist FST
29
+ deterministic: if True will provide a single transduction option,
30
+ for False multiple transduction are generated (used for audio-based normalization)
31
+ """
32
+
33
+ def __init__(self, whitelist: 'pynini.FstLike', deterministic: bool = True):
34
+ super().__init__(name="abbreviation", kind="classify", deterministic=deterministic)
35
+
36
+ dot = pynini.accep(".")
37
+ # A.B.C. -> A. B. C.
38
+ graph = NEMO_UPPER + dot + pynini.closure(insert_space + NEMO_UPPER + dot, 1)
39
+ # A.B.C. -> A.B.C.
40
+ graph |= NEMO_UPPER + dot + pynini.closure(NEMO_UPPER + dot, 1)
41
+ # ABC -> A B C
42
+ graph |= NEMO_UPPER + pynini.closure(insert_space + NEMO_UPPER, 1)
43
+
44
+ # exclude words that are included in the whitelist
45
+ graph = pynini.compose(
46
+ pynini.difference(pynini.project(graph, "input"), pynini.project(whitelist.graph, "input")), graph
47
+ )
48
+
49
+ graph = pynutil.insert("value: \"") + graph.optimize() + pynutil.insert("\"")
50
+ graph = self.add_tokens(graph)
51
+ self.fst = graph.optimize()
vendor/zonos3/vendor/nemo_text_processing/text_normalization/en/taggers/cardinal.py ADDED
@@ -0,0 +1,151 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # Copyright (c) 2021, NVIDIA CORPORATION. All rights reserved.
2
+ #
3
+ # Licensed under the Apache License, Version 2.0 (the "License");
4
+ # you may not use this file except in compliance with the License.
5
+ # You may obtain a copy of the License at
6
+ #
7
+ # http://www.apache.org/licenses/LICENSE-2.0
8
+ #
9
+ # Unless required by applicable law or agreed to in writing, software
10
+ # distributed under the License is distributed on an "AS IS" BASIS,
11
+ # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
12
+ # See the License for the specific language governing permissions and
13
+ # limitations under the License.
14
+
15
+
16
+ import pynini
17
+ from pynini.examples import plurals
18
+ from pynini.lib import pynutil
19
+
20
+ from zonos3.vendor.nemo_text_processing.text_normalization.en.graph_utils import (
21
+ NEMO_DIGIT,
22
+ NEMO_NOT_QUOTE,
23
+ NEMO_SIGMA,
24
+ GraphFst,
25
+ insert_space,
26
+ )
27
+ from zonos3.vendor.nemo_text_processing.text_normalization.en.taggers.date import get_four_digit_year_graph
28
+ from zonos3.vendor.nemo_text_processing.text_normalization.en.utils import get_abs_path
29
+
30
+
31
+ class CardinalFst(GraphFst):
32
+ """
33
+ Finite state transducer for classifying cardinals, e.g.
34
+ -23 -> cardinal { negative: "true" integer: "twenty three" } }
35
+
36
+ Args:
37
+ deterministic: if True will provide a single transduction option,
38
+ for False multiple transduction are generated (used for audio-based normalization)
39
+ """
40
+
41
+ def __init__(self, deterministic: bool = True, lm: bool = False):
42
+ super().__init__(name="cardinal", kind="classify", deterministic=deterministic)
43
+
44
+ self.lm = lm
45
+ self.deterministic = deterministic
46
+ # TODO replace to have "oh" as a default for "0"
47
+ graph = pynini.Far(get_abs_path("data/number/cardinal_number_name.far")).get_fst()
48
+ graph_au = pynini.Far(get_abs_path("data/number/cardinal_number_name_au.far")).get_fst()
49
+ self.graph_hundred_component_at_least_one_none_zero_digit = (
50
+ pynini.closure(NEMO_DIGIT, 2, 3) | pynini.difference(NEMO_DIGIT, pynini.accep("0"))
51
+ ) @ graph
52
+
53
+ graph_digit = pynini.string_file(get_abs_path("data/number/digit.tsv"))
54
+ graph_zero = pynini.string_file(get_abs_path("data/number/zero.tsv"))
55
+
56
+ single_digits_graph = pynini.invert(graph_digit | graph_zero)
57
+ self.single_digits_graph = single_digits_graph + pynini.closure(insert_space + single_digits_graph)
58
+
59
+ if not deterministic:
60
+ # for a single token allow only the same normalization
61
+ # "007" -> {"oh oh seven", "zero zero seven"} not {"oh zero seven"}
62
+ single_digits_graph_zero = pynini.invert(graph_digit | graph_zero)
63
+ single_digits_graph_oh = pynini.invert(graph_digit) | pynini.cross("0", "oh")
64
+
65
+ self.single_digits_graph = single_digits_graph_zero + pynini.closure(
66
+ insert_space + single_digits_graph_zero
67
+ )
68
+ self.single_digits_graph |= single_digits_graph_oh + pynini.closure(insert_space + single_digits_graph_oh)
69
+
70
+ single_digits_graph_with_commas = pynini.closure(
71
+ self.single_digits_graph + insert_space, 1, 3
72
+ ) + pynini.closure(
73
+ pynutil.delete(",")
74
+ + single_digits_graph
75
+ + insert_space
76
+ + single_digits_graph
77
+ + insert_space
78
+ + single_digits_graph,
79
+ 1,
80
+ )
81
+
82
+ optional_minus_graph = pynini.closure(pynutil.insert("negative: ") + pynini.cross("-", "\"true\" "), 0, 1)
83
+
84
+ graph = (
85
+ pynini.closure(NEMO_DIGIT, 1, 3)
86
+ + (pynini.closure(pynutil.delete(",") + NEMO_DIGIT**3) | pynini.closure(NEMO_DIGIT**3))
87
+ ) @ graph
88
+
89
+ self.graph = graph
90
+ self.graph_with_and = self.add_optional_and(graph)
91
+
92
+ if deterministic:
93
+ long_numbers = pynini.compose(NEMO_DIGIT ** (5, ...), self.single_digits_graph).optimize()
94
+ self.long_numbers = plurals._priority_union(long_numbers, self.graph_with_and, NEMO_SIGMA).optimize()
95
+ cardinal_with_leading_zeros = pynini.compose(
96
+ pynini.accep("0") + pynini.closure(NEMO_DIGIT), self.single_digits_graph
97
+ )
98
+ final_graph = self.long_numbers | cardinal_with_leading_zeros
99
+ final_graph |= self.add_optional_and(graph_au)
100
+ else:
101
+ leading_zeros = pynini.compose(pynini.closure(pynini.accep("0"), 1), self.single_digits_graph)
102
+ cardinal_with_leading_zeros = (
103
+ leading_zeros + pynutil.insert(" ") + pynini.compose(pynini.closure(NEMO_DIGIT), self.graph_with_and)
104
+ )
105
+ self.long_numbers = self.graph_with_and | pynutil.add_weight(self.single_digits_graph, 0.0001)
106
+ # add small weight to non-default graphs to make sure the deterministic option is listed first
107
+ final_graph = (
108
+ self.long_numbers
109
+ | get_four_digit_year_graph() # allows e.g. 4567 be pronounced as forty five sixty seven
110
+ | pynutil.add_weight(single_digits_graph_with_commas, 0.0001)
111
+ | cardinal_with_leading_zeros
112
+ ).optimize()
113
+
114
+ one_to_a_replacement_graph = (
115
+ pynini.cross("one hundred", "a hundred")
116
+ | pynini.cross("one thousand", "thousand")
117
+ | pynini.cross("one million", "a million")
118
+ )
119
+ final_graph |= pynini.compose(final_graph, one_to_a_replacement_graph.optimize() + NEMO_SIGMA).optimize()
120
+ # remove commas for 4 digits numbers
121
+ four_digit_comma_graph = (NEMO_DIGIT - "0") + pynutil.delete(",") + NEMO_DIGIT**3
122
+ final_graph |= pynini.compose(four_digit_comma_graph.optimize(), final_graph).optimize()
123
+
124
+ self.final_graph = final_graph
125
+ final_graph = optional_minus_graph + pynutil.insert("integer: \"") + final_graph + pynutil.insert("\"")
126
+ final_graph = self.add_tokens(final_graph)
127
+ self.fst = final_graph.optimize()
128
+
129
+ def add_optional_and(self, graph):
130
+ graph_with_and = graph
131
+
132
+ if not self.lm:
133
+ graph_with_and = pynutil.add_weight(graph, 0.00001)
134
+ not_quote = pynini.closure(NEMO_NOT_QUOTE)
135
+ no_thousand_million = pynini.difference(
136
+ not_quote, not_quote + pynini.union("thousand", "million") + not_quote
137
+ ).optimize()
138
+ integer = (
139
+ not_quote + pynutil.add_weight(pynini.cross("hundred ", "hundred and ") + no_thousand_million, -0.0001)
140
+ ).optimize()
141
+
142
+ no_hundred = pynini.difference(NEMO_SIGMA, not_quote + pynini.accep("hundred") + not_quote).optimize()
143
+ integer |= (
144
+ not_quote + pynutil.add_weight(pynini.cross("thousand ", "thousand and ") + no_hundred, -0.0001)
145
+ ).optimize()
146
+
147
+ optional_hundred = pynini.compose((NEMO_DIGIT - "0") ** 3, graph).optimize()
148
+ optional_hundred = pynini.compose(optional_hundred, NEMO_SIGMA + pynini.cross(" hundred", "") + NEMO_SIGMA)
149
+ graph_with_and |= pynini.compose(graph, integer).optimize()
150
+ graph_with_and |= optional_hundred
151
+ return graph_with_and
vendor/zonos3/vendor/nemo_text_processing/text_normalization/en/taggers/date.py ADDED
@@ -0,0 +1,386 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # Copyright (c) 2021, NVIDIA CORPORATION. All rights reserved.
2
+ #
3
+ # Licensed under the Apache License, Version 2.0 (the "License");
4
+ # you may not use this file except in compliance with the License.
5
+ # You may obtain a copy of the License at
6
+ #
7
+ # http://www.apache.org/licenses/LICENSE-2.0
8
+ #
9
+ # Unless required by applicable law or agreed to in writing, software
10
+ # distributed under the License is distributed on an "AS IS" BASIS,
11
+ # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
12
+ # See the License for the specific language governing permissions and
13
+ # limitations under the License.
14
+
15
+ import pynini
16
+ from pynini.examples import plurals
17
+ from pynini.lib import pynutil
18
+
19
+ from zonos3.vendor.nemo_text_processing.text_normalization.en.graph_utils import (
20
+ NEMO_CHAR,
21
+ NEMO_DIGIT,
22
+ NEMO_LOWER,
23
+ NEMO_SIGMA,
24
+ TO_LOWER,
25
+ GraphFst,
26
+ delete_extra_space,
27
+ delete_space,
28
+ insert_space,
29
+ )
30
+ from zonos3.vendor.nemo_text_processing.text_normalization.en.utils import (
31
+ augment_labels_with_punct_at_end,
32
+ get_abs_path,
33
+ load_labels,
34
+ )
35
+
36
+ graph_teen = pynini.invert(pynini.string_file(get_abs_path("data/number/teen.tsv"))).optimize()
37
+ graph_digit = pynini.invert(pynini.string_file(get_abs_path("data/number/digit.tsv"))).optimize()
38
+ ties_graph = pynini.invert(pynini.string_file(get_abs_path("data/number/ty.tsv"))).optimize()
39
+ year_suffix = load_labels(get_abs_path("data/date/year_suffix.tsv"))
40
+ year_suffix.extend(augment_labels_with_punct_at_end(year_suffix))
41
+ year_suffix = pynini.string_map(year_suffix).optimize()
42
+
43
+
44
+ def get_ties_graph(deterministic: bool = True):
45
+ """
46
+ Returns two digit transducer, e.g.
47
+ 03 -> o three
48
+ 12 -> thirteen
49
+ 20 -> twenty
50
+ """
51
+ graph = graph_teen | ties_graph + pynutil.delete("0") | ties_graph + insert_space + graph_digit
52
+
53
+ if deterministic:
54
+ graph = graph | pynini.cross("0", "oh") + insert_space + graph_digit
55
+ else:
56
+ graph = graph | (pynini.cross("0", "oh") | pynini.cross("0", "zero")) + insert_space + graph_digit
57
+
58
+ return graph.optimize()
59
+
60
+
61
+ def get_four_digit_year_graph(deterministic: bool = True):
62
+ """
63
+ Returns a four digit transducer which is combination of ties/teen or digits
64
+ (using hundred instead of thousand format), e.g.
65
+ 1219 -> twelve nineteen
66
+ 3900 -> thirty nine hundred
67
+ """
68
+ graph_ties = get_ties_graph(deterministic)
69
+
70
+ graph_with_s = (
71
+ (graph_ties + insert_space + graph_ties)
72
+ | (graph_teen + insert_space + (ties_graph | pynini.cross("1", "ten")))
73
+ ) + pynutil.delete("0s")
74
+
75
+ graph_with_s |= (graph_teen | graph_ties) + insert_space + pynini.cross("00", "hundred") + pynutil.delete("s")
76
+ graph_with_s = graph_with_s @ pynini.cdrewrite(
77
+ pynini.cross("y", "ies") | pynutil.insert("s"), "", "[EOS]", NEMO_SIGMA
78
+ )
79
+
80
+ graph = graph_ties + insert_space + graph_ties
81
+ graph |= (graph_teen | graph_ties) + insert_space + pynini.cross("00", "hundred")
82
+
83
+ thousand_graph = (
84
+ graph_digit
85
+ + insert_space
86
+ + pynini.cross("00", "thousand")
87
+ + (pynutil.delete("0") | insert_space + graph_digit)
88
+ )
89
+ thousand_graph |= (
90
+ graph_digit
91
+ + insert_space
92
+ + pynini.cross("000", "thousand")
93
+ + pynini.closure(pynutil.delete(" "), 0, 1)
94
+ + pynini.accep("s")
95
+ )
96
+
97
+ graph |= graph_with_s
98
+ if deterministic:
99
+ graph = plurals._priority_union(thousand_graph, graph, NEMO_SIGMA)
100
+ else:
101
+ graph |= thousand_graph
102
+
103
+ return graph.optimize()
104
+
105
+
106
+ def _get_two_digit_year_with_s_graph():
107
+ # to handle '70s -> seventies
108
+ graph = (
109
+ pynini.closure(pynutil.delete("'"), 0, 1)
110
+ + pynini.compose(
111
+ ties_graph + pynutil.delete("0s"), pynini.cdrewrite(pynini.cross("y", "ies"), "", "[EOS]", NEMO_SIGMA)
112
+ )
113
+ ).optimize()
114
+ return graph
115
+
116
+
117
+ def _get_year_graph(cardinal_graph, deterministic: bool = True):
118
+ """
119
+ Transducer for year, only from 1000 - 2999 e.g.
120
+ 1290 -> twelve nineteen
121
+ 2000 - 2009 will be verbalized as two thousand.
122
+
123
+ Transducer for 3 digit year, e.g. 123-> one twenty three
124
+
125
+ Transducer for year with suffix
126
+ 123 A.D., 4200 B.C
127
+ """
128
+ graph = get_four_digit_year_graph(deterministic)
129
+ graph = (pynini.union("1", "2") + (NEMO_DIGIT**3) + pynini.closure(pynini.cross(" s", "s") | "s", 0, 1)) @ graph
130
+
131
+ graph |= _get_two_digit_year_with_s_graph()
132
+
133
+ three_digit_year = (NEMO_DIGIT @ cardinal_graph) + insert_space + (NEMO_DIGIT**2) @ cardinal_graph
134
+ year_with_suffix = (
135
+ (get_four_digit_year_graph(deterministic=True) | three_digit_year) + delete_space + insert_space + year_suffix
136
+ )
137
+ graph |= year_with_suffix
138
+ return graph.optimize()
139
+
140
+
141
+ def _get_two_digit_year(cardinal_graph, single_digits_graph):
142
+ two_digit_year = NEMO_DIGIT ** (2) @ plurals._priority_union(cardinal_graph, single_digits_graph, NEMO_SIGMA)
143
+ return two_digit_year
144
+
145
+
146
+ def _get_financial_period_graph():
147
+ # 1H23 -> first half of twenty three
148
+ # 3Q22 -> third quarter of twenty two
149
+
150
+ h_ordinals = pynini.cross('1', 'first') | pynini.cross('2', 'second')
151
+ q_ordinals = h_ordinals | pynini.cross('3', 'third') | pynini.cross('4', 'fourth')
152
+
153
+ h_graph = h_ordinals + pynini.cross('H', ' half')
154
+ q_graph = q_ordinals + pynini.cross('Q', ' quarter')
155
+ period_graph = h_graph | q_graph
156
+
157
+ return period_graph
158
+
159
+
160
+ class DateFst(GraphFst):
161
+ """
162
+ Finite state transducer for classifying date, e.g.
163
+ jan. 5, 2012 -> date { month: "january" day: "five" year: "twenty twelve" preserve_order: true }
164
+ jan. 5 -> date { month: "january" day: "five" preserve_order: true }
165
+ 5 january 2012 -> date { day: "five" month: "january" year: "twenty twelve" preserve_order: true }
166
+ 2012-01-05 -> date { year: "twenty twelve" month: "january" day: "five" }
167
+ 2012.01.05 -> date { year: "twenty twelve" month: "january" day: "five" }
168
+ 2012/01/05 -> date { year: "twenty twelve" month: "january" day: "five" }
169
+ 2012 -> date { year: "twenty twelve" }
170
+
171
+ Args:
172
+ cardinal: CardinalFst
173
+ deterministic: if True will provide a single transduction option,
174
+ for False multiple transduction are generated (used for audio-based normalization)
175
+ """
176
+
177
+ def __init__(self, cardinal: GraphFst, deterministic: bool, lm: bool = False):
178
+ super().__init__(name="date", kind="classify", deterministic=deterministic)
179
+
180
+ # january
181
+ month_graph = pynini.string_file(get_abs_path("data/date/month_name.tsv")).optimize()
182
+ # January, JANUARY
183
+ month_graph |= pynini.compose(TO_LOWER + pynini.closure(NEMO_CHAR), month_graph) | pynini.compose(
184
+ TO_LOWER ** (2, ...), month_graph
185
+ )
186
+
187
+ # jan
188
+ month_abbr_graph = pynini.string_file(get_abs_path("data/date/month_abbr.tsv")).optimize()
189
+ # jan, Jan, JAN
190
+ month_abbr_graph = (
191
+ month_abbr_graph
192
+ | pynini.compose(TO_LOWER + pynini.closure(NEMO_LOWER, 1), month_abbr_graph).optimize()
193
+ | pynini.compose(TO_LOWER ** (2, ...), month_abbr_graph).optimize()
194
+ ) + pynini.closure(pynutil.delete("."), 0, 1)
195
+ month_graph |= month_abbr_graph.optimize()
196
+
197
+ month_numbers_labels = pynini.string_file(get_abs_path("data/date/month_number.tsv")).optimize()
198
+ cardinal_graph = cardinal.graph_hundred_component_at_least_one_none_zero_digit
199
+
200
+ year_graph = _get_year_graph(cardinal_graph=cardinal_graph, deterministic=deterministic)
201
+
202
+ # three_digit_year = (NEMO_DIGIT @ cardinal_graph) + insert_space + (NEMO_DIGIT ** 2) @ cardinal_graph
203
+ # year_graph |= three_digit_year
204
+
205
+ month_graph = pynutil.insert("month: \"") + month_graph + pynutil.insert("\"")
206
+ month_numbers_graph = pynutil.insert("month: \"") + month_numbers_labels + pynutil.insert("\"")
207
+
208
+ endings = ["rd", "th", "st", "nd"]
209
+ endings += [x.upper() for x in endings]
210
+ endings = pynini.union(*endings)
211
+
212
+ day_graph = (
213
+ pynutil.insert("day: \"")
214
+ + pynini.closure(pynutil.delete("the "), 0, 1)
215
+ + (
216
+ ((pynini.union("1", "2") + NEMO_DIGIT) | NEMO_DIGIT | (pynini.accep("3") + pynini.union("0", "1")))
217
+ + pynini.closure(pynutil.delete(endings), 0, 1)
218
+ )
219
+ @ cardinal_graph
220
+ + pynutil.insert("\"")
221
+ )
222
+
223
+ two_digit_year = _get_two_digit_year(
224
+ cardinal_graph=cardinal_graph, single_digits_graph=cardinal.single_digits_graph
225
+ )
226
+ two_digit_year = pynutil.insert("year: \"") + two_digit_year + pynutil.insert("\"")
227
+
228
+ # if lm:
229
+ # two_digit_year = pynini.compose(pynini.difference(NEMO_DIGIT, "0") + NEMO_DIGIT ** (3), two_digit_year)
230
+ # year_graph = pynini.compose(pynini.difference(NEMO_DIGIT, "0") + NEMO_DIGIT ** (2), year_graph)
231
+ # year_graph |= pynini.compose(pynini.difference(NEMO_DIGIT, "0") + NEMO_DIGIT ** (4, ...), year_graph)
232
+
233
+ graph_year = pynutil.insert(" year: \"") + pynutil.delete(" ") + year_graph + pynutil.insert("\"")
234
+ graph_year |= (
235
+ pynutil.insert(" year: \"")
236
+ + pynini.accep(",")
237
+ + pynini.closure(pynini.accep(" "), 0, 1)
238
+ + year_graph
239
+ + pynutil.insert("\"")
240
+ )
241
+ optional_graph_year = pynini.closure(graph_year, 0, 1)
242
+
243
+ year_graph = pynutil.insert("year: \"") + year_graph + pynutil.insert("\"")
244
+
245
+ graph_mdy = month_graph + (
246
+ (delete_extra_space + day_graph)
247
+ | (pynini.accep(" ") + day_graph)
248
+ | graph_year
249
+ | (delete_extra_space + day_graph + graph_year)
250
+ )
251
+
252
+ graph_mdy |= (
253
+ month_graph
254
+ + pynini.cross("-", " ")
255
+ + day_graph
256
+ + pynini.closure(((pynini.cross("-", " ") + NEMO_SIGMA) @ graph_year), 0, 1)
257
+ )
258
+
259
+ for x in ["-", "/", "."]:
260
+ delete_sep = pynutil.delete(x)
261
+ graph_mdy |= (
262
+ month_numbers_graph
263
+ + delete_sep
264
+ + insert_space
265
+ + pynini.closure(pynutil.delete("0"), 0, 1)
266
+ + day_graph
267
+ + delete_sep
268
+ + insert_space
269
+ + (year_graph | two_digit_year)
270
+ )
271
+
272
+ graph_dmy = day_graph + delete_extra_space + month_graph + optional_graph_year
273
+ day_ex_month = (NEMO_DIGIT**2 - pynini.project(month_numbers_graph, "input")) @ day_graph
274
+ for x in ["-", "/", "."]:
275
+ delete_sep = pynutil.delete(x)
276
+ graph_dmy |= (
277
+ day_ex_month
278
+ + delete_sep
279
+ + insert_space
280
+ + month_numbers_graph
281
+ + delete_sep
282
+ + insert_space
283
+ + (year_graph | two_digit_year)
284
+ )
285
+
286
+ graph_ymd = pynini.accep("")
287
+ for x in ["-", "/", "."]:
288
+ delete_sep = pynutil.delete(x)
289
+ graph_ymd |= (
290
+ (year_graph | two_digit_year)
291
+ + delete_sep
292
+ + insert_space
293
+ + month_numbers_graph
294
+ + delete_sep
295
+ + insert_space
296
+ + pynini.closure(pynutil.delete("0"), 0, 1)
297
+ + day_graph
298
+ )
299
+
300
+ final_graph = graph_mdy | graph_dmy
301
+
302
+ if not deterministic or lm:
303
+ final_graph += pynini.closure(pynutil.insert(" preserve_order: true"), 0, 1)
304
+ m_sep_d = (
305
+ month_numbers_graph
306
+ + pynutil.delete(pynini.union("-", "/"))
307
+ + insert_space
308
+ + pynini.closure(pynutil.delete("0"), 0, 1)
309
+ + day_graph
310
+ )
311
+ final_graph |= m_sep_d
312
+ else:
313
+ final_graph += pynutil.insert(" preserve_order: true")
314
+
315
+ period_fy = pynutil.insert("text: \"") + _get_financial_period_graph() + pynutil.insert("\"")
316
+ graph_fy = period_fy + insert_space + two_digit_year
317
+
318
+ final_graph |= graph_ymd | year_graph | graph_fy
319
+
320
+ if not deterministic or lm:
321
+ ymd_to_mdy_graph = None
322
+ ymd_to_dmy_graph = None
323
+ mdy_to_dmy_graph = None
324
+ md_to_dm_graph = None
325
+
326
+ for month in [x[0] for x in load_labels(get_abs_path("data/date/month_name.tsv"))]:
327
+ for day in [x[0] for x in load_labels(get_abs_path("data/date/day.tsv"))]:
328
+ ymd_to_mdy_curr = (
329
+ pynutil.insert("month: \"" + month + "\" day: \"" + day + "\" ")
330
+ + pynini.accep('year:')
331
+ + NEMO_SIGMA
332
+ + pynutil.delete(" month: \"" + month + "\" day: \"" + day + "\"")
333
+ )
334
+
335
+ # YY-MM-DD -> MM-DD-YY
336
+ ymd_to_mdy_curr = pynini.compose(graph_ymd, ymd_to_mdy_curr)
337
+ ymd_to_mdy_graph = (
338
+ ymd_to_mdy_curr
339
+ if ymd_to_mdy_graph is None
340
+ else pynini.union(ymd_to_mdy_curr, ymd_to_mdy_graph)
341
+ )
342
+
343
+ ymd_to_dmy_curr = (
344
+ pynutil.insert("day: \"" + day + "\" month: \"" + month + "\" ")
345
+ + pynini.accep('year:')
346
+ + NEMO_SIGMA
347
+ + pynutil.delete(" month: \"" + month + "\" day: \"" + day + "\"")
348
+ )
349
+
350
+ # YY-MM-DD -> MM-DD-YY
351
+ ymd_to_dmy_curr = pynini.compose(graph_ymd, ymd_to_dmy_curr).optimize()
352
+ ymd_to_dmy_graph = (
353
+ ymd_to_dmy_curr
354
+ if ymd_to_dmy_graph is None
355
+ else pynini.union(ymd_to_dmy_curr, ymd_to_dmy_graph)
356
+ )
357
+
358
+ mdy_to_dmy_curr = (
359
+ pynutil.insert("day: \"" + day + "\" month: \"" + month + "\" ")
360
+ + pynutil.delete("month: \"" + month + "\" day: \"" + day + "\" ")
361
+ + pynini.accep('year:')
362
+ + NEMO_SIGMA
363
+ ).optimize()
364
+ # MM-DD-YY -> verbalize as MM-DD-YY (February fourth 1991) or DD-MM-YY (the fourth of February 1991)
365
+ mdy_to_dmy_curr = pynini.compose(graph_mdy, mdy_to_dmy_curr).optimize()
366
+ mdy_to_dmy_graph = (
367
+ mdy_to_dmy_curr
368
+ if mdy_to_dmy_graph is None
369
+ else pynini.union(mdy_to_dmy_curr, mdy_to_dmy_graph).optimize()
370
+ ).optimize()
371
+
372
+ md_to_dm_curr = pynutil.insert("day: \"" + day + "\" month: \"" + month + "\"") + pynutil.delete(
373
+ "month: \"" + month + "\" day: \"" + day + "\""
374
+ )
375
+ md_to_dm_curr = pynini.compose(m_sep_d, md_to_dm_curr).optimize()
376
+
377
+ md_to_dm_graph = (
378
+ md_to_dm_curr
379
+ if md_to_dm_graph is None
380
+ else pynini.union(md_to_dm_curr, md_to_dm_graph).optimize()
381
+ ).optimize()
382
+
383
+ final_graph |= mdy_to_dmy_graph | md_to_dm_graph | ymd_to_mdy_graph | ymd_to_dmy_graph
384
+
385
+ final_graph = self.add_tokens(final_graph)
386
+ self.fst = final_graph.optimize()
vendor/zonos3/vendor/nemo_text_processing/text_normalization/en/taggers/decimal.py ADDED
@@ -0,0 +1,130 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # Copyright (c) 2021, NVIDIA CORPORATION. All rights reserved.
2
+ #
3
+ # Licensed under the Apache License, Version 2.0 (the "License");
4
+ # you may not use this file except in compliance with the License.
5
+ # You may obtain a copy of the License at
6
+ #
7
+ # http://www.apache.org/licenses/LICENSE-2.0
8
+ #
9
+ # Unless required by applicable law or agreed to in writing, software
10
+ # distributed under the License is distributed on an "AS IS" BASIS,
11
+ # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
12
+ # See the License for the specific language governing permissions and
13
+ # limitations under the License.
14
+
15
+ import pynini
16
+ from pynini.lib import pynutil
17
+
18
+ from zonos3.vendor.nemo_text_processing.text_normalization.en.graph_utils import NEMO_SIGMA, TO_UPPER, GraphFst, get_abs_path
19
+
20
+ delete_space = pynutil.delete(" ")
21
+ quantities = pynini.string_file(get_abs_path("data/number/thousand.tsv"))
22
+ quantities_abbr = pynini.string_file(get_abs_path("data/number/quantity_abbr.tsv"))
23
+ quantities_abbr |= TO_UPPER @ quantities_abbr
24
+
25
+
26
+ def get_quantity(
27
+ decimal: 'pynini.FstLike', cardinal_up_to_hundred: 'pynini.FstLike', include_abbr: bool
28
+ ) -> 'pynini.FstLike':
29
+ """
30
+ Returns FST that transforms either a cardinal or decimal followed by a quantity into a numeral,
31
+ e.g. 1 million -> integer_part: "one" quantity: "million"
32
+ e.g. 1.5 million -> integer_part: "one" fractional_part: "five" quantity: "million"
33
+
34
+ Args:
35
+ decimal: decimal FST
36
+ cardinal_up_to_hundred: cardinal FST
37
+ """
38
+ quantity_wo_thousand = pynini.project(quantities, "input") - pynini.union("k", "K", "thousand")
39
+ if include_abbr:
40
+ quantity_wo_thousand |= pynini.project(quantities_abbr, "input") - pynini.union("k", "K", "thousand")
41
+ res = (
42
+ pynutil.insert("integer_part: \"")
43
+ + cardinal_up_to_hundred
44
+ + pynutil.insert("\"")
45
+ + pynini.closure(pynutil.delete(" "), 0, 1)
46
+ + pynutil.insert(" quantity: \"")
47
+ + (quantity_wo_thousand @ (quantities | quantities_abbr))
48
+ + pynutil.insert("\"")
49
+ )
50
+ if include_abbr:
51
+ quantity = quantities | quantities_abbr
52
+ else:
53
+ quantity = quantities
54
+ res |= (
55
+ decimal
56
+ + pynini.closure(pynutil.delete(" "), 0, 1)
57
+ + pynutil.insert("quantity: \"")
58
+ + quantity
59
+ + pynutil.insert("\"")
60
+ )
61
+ return res
62
+
63
+
64
+ class DecimalFst(GraphFst):
65
+ """
66
+ Finite state transducer for classifying decimal, e.g.
67
+ -12.5006 billion -> decimal { negative: "true" integer_part: "12" fractional_part: "five o o six" quantity: "billion" }
68
+ 1 billion -> decimal { integer_part: "one" quantity: "billion" }
69
+
70
+ cardinal: CardinalFst
71
+ """
72
+
73
+ def __init__(self, cardinal: GraphFst, deterministic: bool):
74
+ super().__init__(name="decimal", kind="classify", deterministic=deterministic)
75
+
76
+ cardinal_graph = cardinal.graph_with_and
77
+ cardinal_graph_hundred_component_at_least_one_none_zero_digit = (
78
+ cardinal.graph_hundred_component_at_least_one_none_zero_digit
79
+ )
80
+
81
+ self.graph = cardinal.single_digits_graph.optimize()
82
+
83
+ if not deterministic:
84
+ self.graph = self.graph | cardinal_graph
85
+
86
+ point = pynutil.delete(".")
87
+ optional_graph_negative = pynini.closure(pynutil.insert("negative: ") + pynini.cross("-", "\"true\" "), 0, 1)
88
+
89
+ self.graph_fractional = pynutil.insert("fractional_part: \"") + self.graph + pynutil.insert("\"")
90
+ self.graph_integer = pynutil.insert("integer_part: \"") + cardinal_graph + pynutil.insert("\"")
91
+ final_graph_wo_sign = (
92
+ pynini.closure(self.graph_integer + pynutil.insert(" "), 0, 1)
93
+ + point
94
+ + pynutil.insert(" ")
95
+ + self.graph_fractional
96
+ )
97
+
98
+ quantity_w_abbr = get_quantity(
99
+ final_graph_wo_sign, cardinal_graph_hundred_component_at_least_one_none_zero_digit, include_abbr=True
100
+ )
101
+ quantity_wo_abbr = get_quantity(
102
+ final_graph_wo_sign, cardinal_graph_hundred_component_at_least_one_none_zero_digit, include_abbr=False
103
+ )
104
+ self.final_graph_wo_negative_w_abbr = final_graph_wo_sign | quantity_w_abbr
105
+ self.final_graph_wo_negative = final_graph_wo_sign | quantity_wo_abbr
106
+
107
+ # reduce options for non_deterministic and allow either "oh" or "zero", but not combination
108
+ if not deterministic:
109
+ no_oh_zero = pynini.difference(
110
+ NEMO_SIGMA,
111
+ (NEMO_SIGMA + "oh" + NEMO_SIGMA + "zero" + NEMO_SIGMA)
112
+ | (NEMO_SIGMA + "zero" + NEMO_SIGMA + "oh" + NEMO_SIGMA),
113
+ ).optimize()
114
+ no_zero_oh = pynini.difference(
115
+ NEMO_SIGMA, NEMO_SIGMA + pynini.accep("zero") + NEMO_SIGMA + pynini.accep("oh") + NEMO_SIGMA
116
+ ).optimize()
117
+
118
+ self.final_graph_wo_negative |= pynini.compose(
119
+ self.final_graph_wo_negative,
120
+ pynini.cdrewrite(
121
+ pynini.cross("integer_part: \"zero\"", "integer_part: \"oh\""), NEMO_SIGMA, NEMO_SIGMA, NEMO_SIGMA
122
+ ),
123
+ )
124
+ self.final_graph_wo_negative = pynini.compose(self.final_graph_wo_negative, no_oh_zero).optimize()
125
+ self.final_graph_wo_negative = pynini.compose(self.final_graph_wo_negative, no_zero_oh).optimize()
126
+
127
+ final_graph = optional_graph_negative + self.final_graph_wo_negative
128
+
129
+ final_graph = self.add_tokens(final_graph)
130
+ self.fst = final_graph.optimize()
vendor/zonos3/vendor/nemo_text_processing/text_normalization/en/taggers/electronic.py ADDED
@@ -0,0 +1,179 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # Copyright (c) 2024, NVIDIA CORPORATION. All rights reserved.
2
+ #
3
+ # Licensed under the Apache License, Version 2.0 (the "License");
4
+ # you may not use this file except in compliance with the License.
5
+ # You may obtain a copy of the License at
6
+ #
7
+ # http://www.apache.org/licenses/LICENSE-2.0
8
+ #
9
+ # Unless required by applicable law or agreed to in writing, software
10
+ # distributed under the License is distributed on an "AS IS" BASIS,
11
+ # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
12
+ # See the License for the specific language governing permissions and
13
+ # limitations under the License.
14
+
15
+
16
+ import pynini
17
+ from pynini.lib import pynutil
18
+
19
+ from zonos3.vendor.nemo_text_processing.text_normalization.en.graph_utils import ( # common string literals
20
+ MIN_NEG_WEIGHT,
21
+ NEMO_ALPHA,
22
+ NEMO_DIGIT,
23
+ NEMO_NOT_SPACE,
24
+ NEMO_SIGMA,
25
+ NEMO_UPPER,
26
+ TO_UPPER,
27
+ GraphFst,
28
+ get_abs_path,
29
+ insert_space,
30
+ )
31
+
32
+
33
+ class ElectronicFst(GraphFst):
34
+ """
35
+ Finite state transducer for classifying electronic: as URLs, email addresses, etc.
36
+ e.g. cdf1@abc.edu -> tokens { electronic { username: "cdf1" domain: "abc.edu" } }
37
+
38
+ Args:
39
+ cardinal: CardinalFst
40
+ deterministic: if True will provide a single transduction option,
41
+ for False multiple transduction are generated (used for audio-based normalization)
42
+ """
43
+
44
+ def __init__(self, cardinal: GraphFst, deterministic: bool = True):
45
+ super().__init__(name="electronic", kind="classify", deterministic=deterministic)
46
+
47
+ if deterministic:
48
+ numbers = NEMO_DIGIT
49
+ else:
50
+ numbers = pynutil.insert(" ") + cardinal.long_numbers + pynutil.insert(" ")
51
+
52
+ cc_cues = pynutil.add_weight(
53
+ pynini.string_file(get_abs_path("data/electronic/cc_cues.tsv")),
54
+ MIN_NEG_WEIGHT,
55
+ )
56
+
57
+ cc_cues = pynutil.add_weight(
58
+ pynini.string_file(get_abs_path("data/electronic/cc_cues.tsv")),
59
+ MIN_NEG_WEIGHT,
60
+ )
61
+
62
+ accepted_symbols = pynini.project(pynini.string_file(get_abs_path("data/electronic/symbol.tsv")), "input")
63
+
64
+ accepted_common_domains = pynini.project(
65
+ pynini.string_file(get_abs_path("data/electronic/domain.tsv")), "input"
66
+ )
67
+
68
+ dict_words = pynutil.add_weight(
69
+ pynini.string_file(get_abs_path("data/electronic/words.tsv")),
70
+ MIN_NEG_WEIGHT,
71
+ )
72
+
73
+ dict_words_without_delimiter = dict_words + pynini.closure(
74
+ pynutil.add_weight(pynutil.insert(" ") + dict_words, MIN_NEG_WEIGHT),
75
+ 1,
76
+ )
77
+ dict_words_graph = dict_words_without_delimiter | dict_words
78
+
79
+ all_accepted_symbols_start = (
80
+ dict_words_graph | pynini.closure(TO_UPPER) | pynini.closure(NEMO_UPPER) | accepted_symbols
81
+ ).optimize()
82
+
83
+ all_accepted_symbols_end = (
84
+ dict_words_graph | numbers | pynini.closure(TO_UPPER) | pynini.closure(NEMO_UPPER) | accepted_symbols
85
+ ).optimize()
86
+
87
+ graph_symbols = pynini.string_file(get_abs_path("data/electronic/symbol.tsv")).optimize()
88
+
89
+ username = (NEMO_ALPHA | dict_words_graph) + pynini.closure(
90
+ NEMO_ALPHA | numbers | accepted_symbols | dict_words_graph
91
+ )
92
+
93
+ username = pynutil.insert('username: "') + username + pynutil.insert('"') + pynini.cross("@", " ")
94
+
95
+ domain_graph = all_accepted_symbols_start + pynini.closure(
96
+ all_accepted_symbols_end | pynutil.add_weight(accepted_common_domains, MIN_NEG_WEIGHT)
97
+ )
98
+
99
+ protocol_symbols = pynini.closure((graph_symbols | pynini.cross(":", "colon")) + pynutil.insert(" "))
100
+
101
+ protocol_start = (pynini.cross("https", "HTTPS ") | pynini.cross("http", "HTTP ")) + (
102
+ pynini.accep("://") @ protocol_symbols
103
+ )
104
+
105
+ protocol_file_start = pynini.accep("file") + insert_space + (pynini.accep(":///") @ protocol_symbols)
106
+
107
+ protocol_end = pynutil.add_weight(pynini.cross("www", "WWW ") + pynini.accep(".") @ protocol_symbols, -1000)
108
+
109
+ protocol = protocol_file_start | protocol_start | protocol_end | (protocol_start + protocol_end)
110
+
111
+ domain_graph_with_class_tags = (
112
+ pynutil.insert('domain: "')
113
+ + pynini.compose(
114
+ NEMO_ALPHA + pynini.closure(NEMO_NOT_SPACE) + (NEMO_ALPHA | NEMO_DIGIT | pynini.accep("/")),
115
+ domain_graph,
116
+ ).optimize()
117
+ + pynutil.insert('"')
118
+ )
119
+
120
+ protocol = pynutil.insert('protocol: "') + pynutil.add_weight(protocol, MIN_NEG_WEIGHT) + pynutil.insert('"')
121
+
122
+ # email
123
+ graph = pynini.compose(
124
+ NEMO_SIGMA + pynini.accep("@") + NEMO_SIGMA + pynini.accep(".") + NEMO_SIGMA,
125
+ username + domain_graph_with_class_tags,
126
+ )
127
+
128
+ full_stop_accep = pynini.accep(".")
129
+ dollar_accep = pynini.accep("$") # Include for the correct transduction of the money graph
130
+ excluded_symbols = full_stop_accep | dollar_accep | pynini.accep(",")
131
+ filtered_symbols = pynini.difference(accepted_symbols, excluded_symbols)
132
+ accepted_characters = NEMO_ALPHA | NEMO_DIGIT | filtered_symbols
133
+ domain_component = full_stop_accep + pynini.closure(accepted_characters, 2)
134
+ graph_domain = pynutil.add_weight(
135
+ pynutil.insert('domain: "')
136
+ + (pynini.closure(accepted_characters, 1) + pynini.closure(domain_component, 1))
137
+ + pynutil.insert('"'),
138
+ 0.1,
139
+ ).optimize()
140
+
141
+ graph |= pynutil.add_weight(graph_domain, MIN_NEG_WEIGHT)
142
+
143
+ # www.abc.com/sdafsdf, or https://www.abc.com/asdfad or www.abc.abc/asdfad
144
+ graph |= protocol + pynutil.insert(" ") + domain_graph_with_class_tags
145
+
146
+ # recursively handles the "/" in strings like:
147
+ # update/upgrade -> update slash upgrade
148
+ # update/upgrade/downgrade -> update slash upgrade slash downgrade
149
+
150
+ slash_string = (
151
+ pynini.accep(" ").ques + pynini.accep("/") + pynini.accep(" ").ques + pynini.closure(NEMO_ALPHA, 1)
152
+ )
153
+
154
+ graph |= (
155
+ pynutil.insert('domain: "')
156
+ + (pynini.closure(NEMO_ALPHA, 1) - "and")
157
+ + pynini.closure(slash_string, 1)
158
+ + pynutil.insert('"')
159
+ ).optimize()
160
+
161
+ if deterministic:
162
+ # credit card cues
163
+ numbers = pynini.closure(NEMO_DIGIT, 4, 16)
164
+ cc_phrases = (
165
+ pynutil.insert('protocol: "') + cc_cues + pynutil.insert('" domain: "') + numbers + pynutil.insert('"')
166
+ )
167
+ graph |= cc_phrases
168
+
169
+ if deterministic:
170
+ # credit card cues
171
+ numbers = pynini.closure(NEMO_DIGIT, 4, 16)
172
+ cc_phrases = (
173
+ pynutil.insert('protocol: "') + cc_cues + pynutil.insert('" domain: "') + numbers + pynutil.insert('"')
174
+ )
175
+ graph |= cc_phrases
176
+
177
+ final_graph = self.add_tokens(graph)
178
+
179
+ self.fst = final_graph.optimize()
vendor/zonos3/vendor/nemo_text_processing/text_normalization/en/taggers/fraction.py ADDED
@@ -0,0 +1,56 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # Copyright (c) 2021, NVIDIA CORPORATION. All rights reserved.
2
+ #
3
+ # Licensed under the Apache License, Version 2.0 (the "License");
4
+ # you may not use this file except in compliance with the License.
5
+ # You may obtain a copy of the License at
6
+ #
7
+ # http://www.apache.org/licenses/LICENSE-2.0
8
+ #
9
+ # Unless required by applicable law or agreed to in writing, software
10
+ # distributed under the License is distributed on an "AS IS" BASIS,
11
+ # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
12
+ # See the License for the specific language governing permissions and
13
+ # limitations under the License.
14
+
15
+ import pynini
16
+ from pynini.lib import pynutil
17
+
18
+ from zonos3.vendor.nemo_text_processing.text_normalization.en.graph_utils import GraphFst, get_abs_path
19
+
20
+
21
+ class FractionFst(GraphFst):
22
+ """
23
+ Finite state transducer for classifying fraction
24
+ "23 4/5" ->
25
+ tokens { fraction { integer: "twenty three" numerator: "four" denominator: "five" } }
26
+ "23 4/5th" ->
27
+ tokens { fraction { integer: "twenty three" numerator: "four" denominator: "five" } }
28
+
29
+ Args:
30
+ deterministic: if True will provide a single transduction option,
31
+ for False multiple transduction are generated (used for audio-based normalization)
32
+ """
33
+
34
+ def __init__(self, cardinal, deterministic: bool = True):
35
+ super().__init__(name="fraction", kind="classify", deterministic=deterministic)
36
+ cardinal_graph = cardinal.graph
37
+
38
+ integer = pynutil.insert("integer_part: \"") + cardinal_graph + pynutil.insert("\"")
39
+ numerator = (
40
+ pynutil.insert("numerator: \"") + cardinal_graph + (pynini.cross("/", "\" ") | pynini.cross(" / ", "\" "))
41
+ )
42
+
43
+ endings = ["rd", "th", "st", "nd"]
44
+ endings += [x.upper() for x in endings]
45
+ optional_end = pynini.closure(pynini.cross(pynini.union(*endings), ""), 0, 1)
46
+
47
+ denominator = pynutil.insert("denominator: \"") + cardinal_graph + optional_end + pynutil.insert("\"")
48
+
49
+ graph = pynini.closure(integer + pynini.accep(" "), 0, 1) + (numerator + denominator)
50
+ graph |= pynini.closure(integer + (pynini.accep(" ") | pynutil.insert(" ")), 0, 1) + pynini.compose(
51
+ pynini.string_file(get_abs_path("data/number/fraction.tsv")), (numerator + denominator)
52
+ )
53
+
54
+ self.graph = graph
55
+ final_graph = self.add_tokens(self.graph)
56
+ self.fst = final_graph.optimize()
vendor/zonos3/vendor/nemo_text_processing/text_normalization/en/taggers/measure.py ADDED
@@ -0,0 +1,332 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # Copyright (c) 2021, NVIDIA CORPORATION. All rights reserved.
2
+ #
3
+ # Licensed under the Apache License, Version 2.0 (the "License");
4
+ # you may not use this file except in compliance with the License.
5
+ # You may obtain a copy of the License at
6
+ #
7
+ # http://www.apache.org/licenses/LICENSE-2.0
8
+ #
9
+ # Unless required by applicable law or agreed to in writing, software
10
+ # distributed under the License is distributed on an "AS IS" BASIS,
11
+ # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
12
+ # See the License for the specific language governing permissions and
13
+ # limitations under the License.
14
+
15
+ import pynini
16
+ from pynini.examples import plurals
17
+ from pynini.lib import pynutil
18
+
19
+ from zonos3.vendor.nemo_text_processing.text_normalization.en.graph_utils import (
20
+ NEMO_ALPHA,
21
+ NEMO_DIGIT,
22
+ NEMO_NON_BREAKING_SPACE,
23
+ NEMO_SIGMA,
24
+ NEMO_SPACE,
25
+ NEMO_UPPER,
26
+ SINGULAR_TO_PLURAL,
27
+ TO_LOWER,
28
+ GraphFst,
29
+ convert_space,
30
+ delete_space,
31
+ delete_zero_or_one_space,
32
+ insert_space,
33
+ )
34
+ from zonos3.vendor.nemo_text_processing.text_normalization.en.taggers.ordinal import OrdinalFst as OrdinalTagger
35
+ from zonos3.vendor.nemo_text_processing.text_normalization.en.taggers.whitelist import get_formats
36
+ from zonos3.vendor.nemo_text_processing.text_normalization.en.utils import get_abs_path, load_labels
37
+ from zonos3.vendor.nemo_text_processing.text_normalization.en.verbalizers.ordinal import OrdinalFst as OrdinalVerbalizer
38
+
39
+
40
+ class MeasureFst(GraphFst):
41
+ """
42
+ Finite state transducer for classifying measure, suppletive aware, e.g.
43
+ -12kg -> measure { negative: "true" cardinal { integer: "twelve" } units: "kilograms" }
44
+ 1kg -> measure { cardinal { integer: "one" } units: "kilogram" }
45
+ .5kg -> measure { decimal { fractional_part: "five" } units: "kilograms" }
46
+
47
+ Args:
48
+ cardinal: CardinalFst
49
+ decimal: DecimalFst
50
+ fraction: FractionFst
51
+ deterministic: if True will provide a single transduction option,
52
+ for False multiple transduction are generated (used for audio-based normalization)
53
+ """
54
+
55
+ def __init__(
56
+ self,
57
+ cardinal: GraphFst,
58
+ decimal: GraphFst,
59
+ fraction: GraphFst,
60
+ deterministic: bool = True,
61
+ ):
62
+ super().__init__(name="measure", kind="classify", deterministic=deterministic)
63
+ cardinal_graph = cardinal.graph_with_and | self.get_range(cardinal.graph_with_and)
64
+
65
+ graph_unit = pynini.string_file(get_abs_path("data/measure/unit.tsv"))
66
+ if not deterministic:
67
+ graph_unit |= pynini.string_file(get_abs_path("data/measure/unit_alternatives.tsv"))
68
+
69
+ graph_unit |= pynini.compose(
70
+ pynini.closure(TO_LOWER, 1) + (NEMO_ALPHA | TO_LOWER) + pynini.closure(NEMO_ALPHA | TO_LOWER),
71
+ graph_unit,
72
+ ).optimize()
73
+
74
+ graph_unit_plural = convert_space(graph_unit @ SINGULAR_TO_PLURAL)
75
+ graph_unit = convert_space(graph_unit)
76
+
77
+ optional_graph_negative = pynini.closure(pynutil.insert("negative: ") + pynini.cross("-", '"true" '), 0, 1)
78
+
79
+ graph_unit2 = (
80
+ pynini.cross("/", "per") + delete_zero_or_one_space + pynutil.insert(NEMO_NON_BREAKING_SPACE) + graph_unit
81
+ )
82
+
83
+ optional_graph_unit2 = pynini.closure(
84
+ delete_zero_or_one_space + pynutil.insert(NEMO_NON_BREAKING_SPACE) + graph_unit2,
85
+ 0,
86
+ 1,
87
+ )
88
+
89
+ unit_plural = (
90
+ pynutil.insert('units: "') + (graph_unit_plural + optional_graph_unit2 | graph_unit2) + pynutil.insert('"')
91
+ )
92
+
93
+ unit_singular = (
94
+ pynutil.insert('units: "') + (graph_unit + optional_graph_unit2 | graph_unit2) + pynutil.insert('"')
95
+ )
96
+
97
+ subgraph_decimal = (
98
+ pynutil.insert("decimal { ")
99
+ + optional_graph_negative
100
+ + decimal.final_graph_wo_negative
101
+ + delete_space
102
+ + pynutil.insert(" } ")
103
+ + unit_plural
104
+ )
105
+
106
+ # support radio FM/AM
107
+ subgraph_decimal |= (
108
+ pynutil.insert("decimal { ")
109
+ + decimal.final_graph_wo_negative
110
+ + delete_space
111
+ + pynutil.insert(" } ")
112
+ + pynutil.insert('units: "')
113
+ + pynini.union("AM", "FM")
114
+ + pynutil.insert('"')
115
+ )
116
+
117
+ subgraph_cardinal = (
118
+ pynutil.insert("cardinal { ")
119
+ + optional_graph_negative
120
+ + pynutil.insert('integer: "')
121
+ + ((NEMO_SIGMA - "1") @ cardinal_graph)
122
+ + delete_space
123
+ + pynutil.insert('"')
124
+ + pynutil.insert(" } ")
125
+ + unit_plural
126
+ )
127
+
128
+ subgraph_cardinal |= (
129
+ pynutil.insert("cardinal { ")
130
+ + optional_graph_negative
131
+ + pynutil.insert('integer: "')
132
+ + pynini.cross("1", "one")
133
+ + delete_space
134
+ + pynutil.insert('"')
135
+ + pynutil.insert(" } ")
136
+ + unit_singular
137
+ )
138
+
139
+ unit_graph = (
140
+ pynutil.insert('cardinal { integer: "-" } units: "')
141
+ + ((pynini.cross("/", "per") + delete_zero_or_one_space) | (pynini.accep("per") + pynutil.delete(" ")))
142
+ + pynutil.insert(NEMO_NON_BREAKING_SPACE)
143
+ + graph_unit
144
+ + pynutil.insert('" preserve_order: true')
145
+ )
146
+
147
+ decimal_dash_alpha = (
148
+ pynutil.insert("decimal { ")
149
+ + decimal.final_graph_wo_negative
150
+ + pynini.cross("-", "")
151
+ + pynutil.insert(' } units: "')
152
+ + pynini.closure(NEMO_ALPHA, 1)
153
+ + pynutil.insert('"')
154
+ )
155
+
156
+ decimal_times = (
157
+ pynutil.insert("decimal { ")
158
+ + decimal.final_graph_wo_negative
159
+ + pynutil.insert(' } units: "')
160
+ + (pynini.cross(pynini.union("x", "X"), "x") | pynini.cross(pynini.union("x", "X"), " times"))
161
+ + pynutil.insert('"')
162
+ )
163
+
164
+ alpha_dash_decimal = (
165
+ pynutil.insert('units: "')
166
+ + pynini.closure(NEMO_ALPHA, 1)
167
+ + pynini.accep("-")
168
+ + pynutil.insert('"')
169
+ + pynutil.insert(" decimal { ")
170
+ + decimal.final_graph_wo_negative
171
+ + pynutil.insert(" } preserve_order: true")
172
+ )
173
+
174
+ subgraph_fraction = (
175
+ pynutil.insert("fraction { ") + fraction.graph + delete_space + pynutil.insert(" } ") + unit_plural
176
+ )
177
+
178
+ address = self.get_address_graph(cardinal)
179
+ address = (
180
+ pynutil.insert('units: "address" cardinal { integer: "')
181
+ + address
182
+ + pynutil.insert('" } preserve_order: true')
183
+ )
184
+
185
+ math_operations = pynini.string_file(get_abs_path("data/measure/math_operation.tsv"))
186
+ delimiter = pynini.accep(" ") | pynutil.insert(" ")
187
+
188
+ math = (
189
+ (cardinal_graph | NEMO_ALPHA)
190
+ + delimiter
191
+ + math_operations
192
+ + (delimiter | NEMO_ALPHA)
193
+ + cardinal_graph
194
+ + delimiter
195
+ + pynini.cross("=", "equals")
196
+ + delimiter
197
+ + (cardinal_graph | NEMO_ALPHA)
198
+ )
199
+
200
+ math |= (
201
+ (cardinal_graph | NEMO_ALPHA)
202
+ + delimiter
203
+ + pynini.cross("=", "equals")
204
+ + delimiter
205
+ + (cardinal_graph | NEMO_ALPHA)
206
+ + delimiter
207
+ + math_operations
208
+ + delimiter
209
+ + cardinal_graph
210
+ )
211
+
212
+ math = (
213
+ pynutil.insert('units: "math" cardinal { integer: "') + math + pynutil.insert('" } preserve_order: true')
214
+ )
215
+ final_graph = (
216
+ subgraph_decimal
217
+ | subgraph_cardinal
218
+ | unit_graph
219
+ | decimal_dash_alpha
220
+ | decimal_times
221
+ | alpha_dash_decimal
222
+ | subgraph_fraction
223
+ | address
224
+ | math
225
+ )
226
+
227
+ final_graph = self.add_tokens(final_graph)
228
+ self.fst = final_graph.optimize()
229
+
230
+ def get_range(self, cardinal: GraphFst):
231
+ """
232
+ Returns range forms for measure tagger, e.g. 2-3, 2x3, 2*2
233
+
234
+ Args:
235
+ cardinal: cardinal GraphFst
236
+ """
237
+ range_graph = cardinal + pynini.cross(pynini.union("-", " - "), " to ") + cardinal
238
+
239
+ for x in [" x ", "x"]:
240
+ range_graph |= cardinal + pynini.cross(x, " by ") + cardinal
241
+ if not self.deterministic:
242
+ range_graph |= cardinal + pynini.cross(x, " times ") + pynini.closure(cardinal, 0, 1)
243
+
244
+ for x in ["*", " * "]:
245
+ range_graph |= cardinal + pynini.cross(x, " times ") + cardinal
246
+ return range_graph.optimize()
247
+
248
+ def get_address_graph(self, cardinal):
249
+ """
250
+ Finite state transducer for classifying serial.
251
+ The serial is a combination of digits, letters and dashes, e.g.:
252
+ 2788 San Tomas Expy, Santa Clara, CA 95051 ->
253
+ units: "address" cardinal
254
+ { integer: "two seven eight eight San Tomas Expressway Santa Clara California nine five zero five one" }
255
+ preserve_order: true
256
+ """
257
+ ordinal_verbalizer = OrdinalVerbalizer().graph
258
+ ordinal_tagger = OrdinalTagger(cardinal=cardinal).graph
259
+ ordinal_num = pynini.compose(
260
+ pynutil.insert('integer: "') + ordinal_tagger + pynutil.insert('"'),
261
+ ordinal_verbalizer,
262
+ )
263
+
264
+ address_num = NEMO_DIGIT ** (1, 2) @ cardinal.graph_hundred_component_at_least_one_none_zero_digit
265
+ address_num += insert_space + NEMO_DIGIT**2 @ (
266
+ pynini.closure(pynini.cross("0", "zero "), 0, 1)
267
+ + cardinal.graph_hundred_component_at_least_one_none_zero_digit
268
+ )
269
+ # to handle the rest of the numbers
270
+ address_num = pynini.compose(NEMO_DIGIT ** (3, 4), address_num)
271
+ address_num = plurals._priority_union(address_num, cardinal.graph, NEMO_SIGMA)
272
+
273
+ direction = (
274
+ pynini.cross("E", "East")
275
+ | pynini.cross("S", "South")
276
+ | pynini.cross("W", "West")
277
+ | pynini.cross("N", "North")
278
+ ) + pynini.closure(pynutil.delete("."), 0, 1)
279
+
280
+ direction = pynini.closure(pynini.accep(NEMO_SPACE) + direction, 0, 1)
281
+ address_words = get_formats(get_abs_path("data/address/address_word.tsv"))
282
+ address_words = (
283
+ pynini.accep(NEMO_SPACE)
284
+ + (pynini.closure(ordinal_num, 0, 1) | NEMO_UPPER + pynini.closure(NEMO_ALPHA, 1))
285
+ + NEMO_SPACE
286
+ + pynini.closure(NEMO_UPPER + pynini.closure(NEMO_ALPHA) + NEMO_SPACE)
287
+ + address_words
288
+ )
289
+
290
+ city = pynini.closure(NEMO_ALPHA | pynini.accep(NEMO_SPACE), 1)
291
+ city = pynini.closure(pynini.accep(",") + pynini.accep(NEMO_SPACE) + city, 0, 1)
292
+
293
+ states = load_labels(get_abs_path("data/address/state.tsv"))
294
+
295
+ additional_options = []
296
+ for x, y in states:
297
+ additional_options.append((x, f"{y[0]}.{y[1:]}"))
298
+ states.extend(additional_options)
299
+ state_graph = pynini.string_map(states)
300
+ state = pynini.invert(state_graph)
301
+ state = pynini.closure(pynini.accep(",") + pynini.accep(NEMO_SPACE) + state, 0, 1)
302
+
303
+ zip_code = pynini.compose(NEMO_DIGIT**5, cardinal.single_digits_graph)
304
+ zip_code = pynini.closure(
305
+ pynini.closure(pynini.accep(","), 0, 1) + pynini.accep(NEMO_SPACE) + zip_code,
306
+ 0,
307
+ 1,
308
+ )
309
+
310
+ address = address_num + direction + address_words + pynini.closure(city + state + zip_code, 0, 1)
311
+
312
+ # Handles sentences ending in terminal period abbreviations, when followed by another sentence
313
+ # (e.g. Nancy lived on 1428 Elm St. It was a strange place-> Nancy lived on fourteen twenty eight Elm Street. It was a strange place.)
314
+
315
+ hyphen = pynini.accep("-")
316
+ alpha_chars = NEMO_ALPHA | hyphen
317
+
318
+ address_eos = (
319
+ address_num
320
+ + direction
321
+ + address_words
322
+ + pynini.accep(".")
323
+ + pynini.closure(NEMO_SPACE, 1, 2)
324
+ + NEMO_UPPER
325
+ + pynini.closure(alpha_chars)
326
+ )
327
+
328
+ address |= pynutil.add_weight(address_eos, -0.001)
329
+
330
+ address |= address_num + direction + address_words + pynini.closure(pynini.cross(".", ""), 0, 1)
331
+
332
+ return address
vendor/zonos3/vendor/nemo_text_processing/text_normalization/en/taggers/money.py ADDED
@@ -0,0 +1,199 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # Copyright (c) 2021, NVIDIA CORPORATION. All rights reserved.
2
+ #
3
+ # Licensed under the Apache License, Version 2.0 (the "License");
4
+ # you may not use this file except in compliance with the License.
5
+ # You may obtain a copy of the License at
6
+ #
7
+ # http://www.apache.org/licenses/LICENSE-2.0
8
+ #
9
+ # Unless required by applicable law or agreed to in writing, software
10
+ # distributed under the License is distributed on an "AS IS" BASIS,
11
+ # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
12
+ # See the License for the specific language governing permissions and
13
+ # limitations under the License.
14
+
15
+ import pynini
16
+ from pynini.lib import pynutil
17
+
18
+ from zonos3.vendor.nemo_text_processing.text_normalization.en.graph_utils import (
19
+ NEMO_ALPHA,
20
+ NEMO_DIGIT,
21
+ NEMO_SIGMA,
22
+ SINGULAR_TO_PLURAL,
23
+ GraphFst,
24
+ convert_space,
25
+ insert_space,
26
+ )
27
+ from zonos3.vendor.nemo_text_processing.text_normalization.en.utils import get_abs_path, load_labels
28
+
29
+ min_singular = pynini.string_file(get_abs_path("data/money/currency_minor_singular.tsv"))
30
+ min_plural = pynini.string_file(get_abs_path("data/money/currency_minor_plural.tsv"))
31
+ maj_singular = pynini.string_file((get_abs_path("data/money/currency_major.tsv")))
32
+ per_units = pynini.string_file(get_abs_path("data/money/per_unit.tsv"))
33
+
34
+
35
+ class MoneyFst(GraphFst):
36
+ """
37
+ Finite state transducer for classifying money, suppletive aware, e.g.
38
+ $12.05 -> money { integer_part: "twelve" currency_maj: "dollars" fractional_part: "five" currency_min: "cents" preserve_order: true }
39
+ $12.0500 -> money { integer_part: "twelve" currency_maj: "dollars" fractional_part: "five" currency_min: "cents" preserve_order: true }
40
+ $1 -> money { currency_maj: "dollar" integer_part: "one" }
41
+ $1.00 -> money { currency_maj: "dollar" integer_part: "one" }
42
+ $0.05 -> money { fractional_part: "five" currency_min: "cents" preserve_order: true }
43
+ $1 million -> money { currency_maj: "dollars" integer_part: "one" quantity: "million" }
44
+ $1.2 million -> money { currency_maj: "dollars" integer_part: "one" fractional_part: "two" quantity: "million" }
45
+ $1.2320 -> money { currency_maj: "dollars" integer_part: "one" fractional_part: "two three two" }
46
+
47
+ Args:
48
+ cardinal: CardinalFst
49
+ decimal: DecimalFst
50
+ deterministic: if True will provide a single transduction option,
51
+ for False multiple transduction are generated (used for audio-based normalization)
52
+ """
53
+
54
+ def __init__(self, cardinal: GraphFst, decimal: GraphFst, deterministic: bool = True):
55
+ super().__init__(name="money", kind="classify", deterministic=deterministic)
56
+ cardinal_graph = cardinal.graph_with_and
57
+ graph_decimal_final = decimal.final_graph_wo_negative_w_abbr
58
+
59
+ maj_singular_labels = load_labels(get_abs_path("data/money/currency_major.tsv"))
60
+ maj_unit_plural = convert_space(maj_singular @ SINGULAR_TO_PLURAL)
61
+ maj_unit_singular = convert_space(maj_singular)
62
+
63
+ graph_maj_singular = pynutil.insert('currency_maj: "') + maj_unit_singular + pynutil.insert('"')
64
+ graph_maj_plural = pynutil.insert('currency_maj: "') + maj_unit_plural + pynutil.insert('"')
65
+
66
+ optional_delete_fractional_zeros = pynini.closure(
67
+ pynutil.delete(".") + pynini.closure(pynutil.delete("0"), 1), 0, 1
68
+ )
69
+
70
+ graph_integer_one = pynutil.insert('integer_part: "') + pynini.cross("1", "one") + pynutil.insert('"')
71
+ # only for decimals where third decimal after comma is non-zero or with quantity
72
+ decimal_delete_last_zeros = (
73
+ pynini.closure(NEMO_DIGIT | pynutil.delete(","))
74
+ + pynini.accep(".")
75
+ + pynini.closure(NEMO_DIGIT, 2)
76
+ + (NEMO_DIGIT - "0")
77
+ + pynini.closure(pynutil.delete("0"))
78
+ )
79
+ decimal_with_quantity = NEMO_SIGMA + NEMO_ALPHA
80
+
81
+ graph_decimal = (
82
+ graph_maj_plural + insert_space + (decimal_delete_last_zeros | decimal_with_quantity) @ graph_decimal_final
83
+ )
84
+
85
+ graph_integer = pynutil.insert('integer_part: "') + ((NEMO_SIGMA - "1") @ cardinal_graph) + pynutil.insert('"')
86
+
87
+ graph_integer_only = graph_maj_singular + insert_space + graph_integer_one
88
+ graph_integer_only |= graph_maj_plural + insert_space + graph_integer
89
+
90
+ final_graph = (graph_integer_only + optional_delete_fractional_zeros) | graph_decimal
91
+
92
+ # remove trailing zeros of non zero number in the first 2 digits and fill up to 2 digits
93
+ # e.g. 2000 -> 20, 0200->02, 01 -> 01, 10 -> 10
94
+ # not accepted: 002, 00, 0,
95
+ two_digits_fractional_part = (
96
+ pynini.closure(NEMO_DIGIT) + (NEMO_DIGIT - "0") + pynini.closure(pynutil.delete("0"))
97
+ ) @ (
98
+ (pynutil.delete("0") + (NEMO_DIGIT - "0"))
99
+ | ((NEMO_DIGIT - "0") + pynutil.insert("0"))
100
+ | ((NEMO_DIGIT - "0") + NEMO_DIGIT)
101
+ )
102
+
103
+ graph_min_singular = pynutil.insert(' currency_min: "') + min_singular + pynutil.insert('"')
104
+ graph_min_plural = pynutil.insert(' currency_min: "') + min_plural + pynutil.insert('"')
105
+ # format ** dollars ** cent
106
+ decimal_graph_with_minor = None
107
+ integer_graph_reordered = None
108
+ decimal_default_reordered = None
109
+ for curr_symbol, _ in maj_singular_labels:
110
+ preserve_order = pynutil.insert(" preserve_order: true")
111
+ integer_plus_maj = graph_integer + insert_space + pynutil.insert(curr_symbol) @ graph_maj_plural
112
+ integer_plus_maj |= graph_integer_one + insert_space + pynutil.insert(curr_symbol) @ graph_maj_singular
113
+
114
+ integer_plus_maj_with_comma = pynini.compose(
115
+ NEMO_DIGIT - "0" + pynini.closure(NEMO_DIGIT | pynutil.delete(",")),
116
+ integer_plus_maj,
117
+ )
118
+ integer_plus_maj = pynini.compose(pynini.closure(NEMO_DIGIT) - "0", integer_plus_maj)
119
+ integer_plus_maj |= integer_plus_maj_with_comma
120
+
121
+ graph_fractional_one = two_digits_fractional_part @ pynini.cross("1", "one")
122
+ graph_fractional_one = pynutil.insert('fractional_part: "') + graph_fractional_one + pynutil.insert('"')
123
+ graph_fractional = (
124
+ two_digits_fractional_part
125
+ @ (pynini.closure(NEMO_DIGIT, 1, 2) - "1")
126
+ @ cardinal.graph_hundred_component_at_least_one_none_zero_digit
127
+ )
128
+ graph_fractional = pynutil.insert('fractional_part: "') + graph_fractional + pynutil.insert('"')
129
+
130
+ fractional_plus_min = graph_fractional + insert_space + pynutil.insert(curr_symbol) @ graph_min_plural
131
+ fractional_plus_min |= (
132
+ graph_fractional_one + insert_space + pynutil.insert(curr_symbol) @ graph_min_singular
133
+ )
134
+
135
+ decimal_graph_with_minor_curr = integer_plus_maj + pynini.cross(".", " ") + fractional_plus_min
136
+
137
+ if not deterministic:
138
+ decimal_graph_with_minor_curr |= pynutil.add_weight(
139
+ integer_plus_maj
140
+ + pynini.cross(".", " ")
141
+ + pynutil.insert('fractional_part: "')
142
+ + two_digits_fractional_part @ cardinal.graph_hundred_component_at_least_one_none_zero_digit
143
+ + pynutil.insert('"'),
144
+ weight=0.0001,
145
+ )
146
+ default_fraction_graph = (decimal_delete_last_zeros | decimal_with_quantity) @ graph_decimal_final
147
+ decimal_graph_with_minor_curr |= (
148
+ pynini.closure(pynutil.delete("0"), 0, 1) + pynutil.delete(".") + fractional_plus_min
149
+ )
150
+ decimal_graph_with_minor_curr = (
151
+ pynutil.delete(curr_symbol) + decimal_graph_with_minor_curr + preserve_order
152
+ )
153
+
154
+ decimal_graph_with_minor = (
155
+ decimal_graph_with_minor_curr
156
+ if decimal_graph_with_minor is None
157
+ else pynini.union(decimal_graph_with_minor, decimal_graph_with_minor_curr).optimize()
158
+ )
159
+
160
+ if not deterministic:
161
+ integer_graph_reordered_curr = (
162
+ pynutil.delete(curr_symbol) + integer_plus_maj + preserve_order
163
+ ).optimize()
164
+
165
+ integer_graph_reordered = (
166
+ integer_graph_reordered_curr
167
+ if integer_graph_reordered is None
168
+ else pynini.union(integer_graph_reordered, integer_graph_reordered_curr).optimize()
169
+ )
170
+ decimal_default_reordered_curr = (
171
+ pynutil.delete(curr_symbol)
172
+ + default_fraction_graph
173
+ + insert_space
174
+ + pynutil.insert(curr_symbol) @ graph_maj_plural
175
+ )
176
+
177
+ decimal_default_reordered = (
178
+ decimal_default_reordered_curr
179
+ if decimal_default_reordered is None
180
+ else pynini.union(decimal_default_reordered, decimal_default_reordered_curr)
181
+ ).optimize()
182
+
183
+ # weight for SH
184
+ final_graph |= pynutil.add_weight(decimal_graph_with_minor, -0.0001)
185
+
186
+ # utilizes morphosyntactic features to append "per units"
187
+ graph_per_units = pynutil.insert(' morphosyntactic_features: "') + per_units + pynutil.insert('"')
188
+
189
+ if not deterministic:
190
+ final_graph |= integer_graph_reordered | decimal_default_reordered
191
+ # to handle "$2.00" cases
192
+ final_graph |= pynini.compose(
193
+ NEMO_SIGMA + pynutil.delete(".") + pynini.closure(pynutil.delete("0"), 1),
194
+ integer_graph_reordered,
195
+ )
196
+
197
+ final_graph += graph_per_units.ques
198
+ final_graph = self.add_tokens(final_graph.optimize())
199
+ self.fst = final_graph.optimize()
vendor/zonos3/vendor/nemo_text_processing/text_normalization/en/taggers/ordinal.py ADDED
@@ -0,0 +1,62 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # Copyright (c) 2021, NVIDIA CORPORATION. All rights reserved.
2
+ #
3
+ # Licensed under the Apache License, Version 2.0 (the "License");
4
+ # you may not use this file except in compliance with the License.
5
+ # You may obtain a copy of the License at
6
+ #
7
+ # http://www.apache.org/licenses/LICENSE-2.0
8
+ #
9
+ # Unless required by applicable law or agreed to in writing, software
10
+ # distributed under the License is distributed on an "AS IS" BASIS,
11
+ # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
12
+ # See the License for the specific language governing permissions and
13
+ # limitations under the License.
14
+
15
+
16
+ import pynini
17
+ from pynini.lib import pynutil
18
+
19
+ from zonos3.vendor.nemo_text_processing.text_normalization.en.graph_utils import NEMO_DIGIT, GraphFst
20
+
21
+
22
+ class OrdinalFst(GraphFst):
23
+ """
24
+ Finite state transducer for classifying ordinal, e.g.
25
+ 13th -> ordinal { integer: "thirteen" }
26
+
27
+ Args:
28
+ cardinal: CardinalFst
29
+ deterministic: if True will provide a single transduction option,
30
+ for False multiple transduction are generated (used for audio-based normalization)
31
+ """
32
+
33
+ def __init__(self, cardinal: GraphFst, deterministic: bool = True):
34
+ super().__init__(name="ordinal", kind="classify", deterministic=deterministic)
35
+
36
+ cardinal_graph = cardinal.graph
37
+ cardinal_format = pynini.closure(NEMO_DIGIT | pynini.accep(","))
38
+ st_format = (
39
+ pynini.closure(cardinal_format + (NEMO_DIGIT - "1"), 0, 1)
40
+ + pynini.accep("1")
41
+ + pynutil.delete(pynini.union("st", "ST", "ˢᵗ"))
42
+ )
43
+ nd_format = (
44
+ pynini.closure(cardinal_format + (NEMO_DIGIT - "1"), 0, 1)
45
+ + pynini.accep("2")
46
+ + pynutil.delete(pynini.union("nd", "ND", "ⁿᵈ"))
47
+ )
48
+ rd_format = (
49
+ pynini.closure(cardinal_format + (NEMO_DIGIT - "1"), 0, 1)
50
+ + pynini.accep("3")
51
+ + pynutil.delete(pynini.union("rd", "RD", "ʳᵈ"))
52
+ )
53
+ th_format = pynini.closure(
54
+ (NEMO_DIGIT - "1" - "2" - "3")
55
+ | (cardinal_format + "1" + NEMO_DIGIT)
56
+ | (cardinal_format + (NEMO_DIGIT - "1") + (NEMO_DIGIT - "1" - "2" - "3")),
57
+ 1,
58
+ ) + pynutil.delete(pynini.union("th", "TH", "ᵗʰ"))
59
+ self.graph = (st_format | nd_format | rd_format | th_format) @ cardinal_graph
60
+ final_graph = pynutil.insert("integer: \"") + self.graph + pynutil.insert("\"")
61
+ final_graph = self.add_tokens(final_graph)
62
+ self.fst = final_graph.optimize()
vendor/zonos3/vendor/nemo_text_processing/text_normalization/en/taggers/punctuation.py ADDED
@@ -0,0 +1,66 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # Copyright (c) 2021, NVIDIA CORPORATION. All rights reserved.
2
+ #
3
+ # Licensed under the Apache License, Version 2.0 (the "License");
4
+ # you may not use this file except in compliance with the License.
5
+ # You may obtain a copy of the License at
6
+ #
7
+ # http://www.apache.org/licenses/LICENSE-2.0
8
+ #
9
+ # Unless required by applicable law or agreed to in writing, software
10
+ # distributed under the License is distributed on an "AS IS" BASIS,
11
+ # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
12
+ # See the License for the specific language governing permissions and
13
+ # limitations under the License.
14
+
15
+ import sys
16
+ from unicodedata import category
17
+
18
+ import pynini
19
+ from pynini.examples import plurals
20
+ from pynini.lib import pynutil
21
+
22
+ from zonos3.vendor.nemo_text_processing.text_normalization.en.graph_utils import NEMO_NOT_SPACE, NEMO_SIGMA, GraphFst
23
+ from zonos3.vendor.nemo_text_processing.text_normalization.en.utils import get_abs_path, load_labels
24
+
25
+
26
+ class PunctuationFst(GraphFst):
27
+ """
28
+ Finite state transducer for classifying punctuation
29
+ e.g. a, -> tokens { name: "a" } tokens { name: "," }
30
+
31
+ Args:
32
+ deterministic: if True will provide a single transduction option,
33
+ for False multiple transduction are generated (used for audio-based normalization)
34
+
35
+ """
36
+
37
+ def __init__(self, deterministic: bool = True):
38
+ super().__init__(name="punctuation", kind="classify", deterministic=deterministic)
39
+ s = "!#%&\'()*+,-./:;<=>?@^_`{|}~\""
40
+
41
+ punct_symbols_to_exclude = ["[", "]"]
42
+ punct_unicode = [
43
+ chr(i)
44
+ for i in range(sys.maxunicode)
45
+ if category(chr(i)).startswith("P") and chr(i) not in punct_symbols_to_exclude
46
+ ] + [r"\[", r"\]"]
47
+
48
+ whitelist_symbols = load_labels(get_abs_path("data/whitelist/symbol.tsv"))
49
+ whitelist_symbols = [x[0] for x in whitelist_symbols]
50
+ self.punct_marks = [p for p in punct_unicode + list(s) if p not in whitelist_symbols]
51
+
52
+ punct = pynini.union(*self.punct_marks)
53
+ punct = pynini.closure(punct, 1)
54
+
55
+ emphasis = (
56
+ pynini.accep("<")
57
+ + (
58
+ (pynini.closure(NEMO_NOT_SPACE - pynini.union("<", ">"), 1) + pynini.closure(pynini.accep("/"), 0, 1))
59
+ | (pynini.accep("/") + pynini.closure(NEMO_NOT_SPACE - pynini.union("<", ">"), 1))
60
+ )
61
+ + pynini.accep(">")
62
+ )
63
+ punct = plurals._priority_union(emphasis, punct, NEMO_SIGMA)
64
+
65
+ self.graph = punct
66
+ self.fst = (pynutil.insert("name: \"") + self.graph + pynutil.insert("\"")).optimize()
vendor/zonos3/vendor/nemo_text_processing/text_normalization/en/taggers/range.py ADDED
@@ -0,0 +1,128 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # Copyright (c) 2022, NVIDIA CORPORATION. All rights reserved.
2
+ #
3
+ # Licensed under the Apache License, Version 2.0 (the "License");
4
+ # you may not use this file except in compliance with the License.
5
+ # You may obtain a copy of the License at
6
+ #
7
+ # http://www.apache.org/licenses/LICENSE-2.0
8
+ #
9
+ # Unless required by applicable law or agreed to in writing, software
10
+ # distributed under the License is distributed on an "AS IS" BASIS,
11
+ # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
12
+ # See the License for the specific language governing permissions and
13
+ # limitations under the License.
14
+
15
+
16
+ import pynini
17
+ from pynini.lib import pynutil
18
+
19
+ from zonos3.vendor.nemo_text_processing.text_normalization.en.graph_utils import NEMO_DIGIT, GraphFst, convert_space
20
+
21
+
22
+ class RangeFst(GraphFst):
23
+ """
24
+ This class is a composite class of two other class instances
25
+
26
+ Args:
27
+ time: composed tagger and verbalizer
28
+ date: composed tagger and verbalizer
29
+ cardinal: tagger
30
+ deterministic: if True will provide a single transduction option,
31
+ for False multiple transduction are generated (used for audio-based normalization)
32
+ lm: whether to use for hybrid LM
33
+ """
34
+
35
+ def __init__(
36
+ self,
37
+ time: GraphFst,
38
+ date: GraphFst,
39
+ cardinal: GraphFst,
40
+ deterministic: bool = True,
41
+ lm: bool = False,
42
+ ):
43
+ super().__init__(name="range", kind="classify", deterministic=deterministic)
44
+
45
+ delete_space = pynini.closure(pynutil.delete(" "), 0, 1)
46
+
47
+ approx = pynini.cross("~", "approximately")
48
+
49
+ # TIME
50
+ time_graph = time + delete_space + pynini.cross("-", " to ") + delete_space + time
51
+ self.graph = time_graph | (approx + time)
52
+
53
+ cardinal = cardinal.graph_with_and
54
+ # YEAR
55
+ date_year_four_digit = (NEMO_DIGIT**4 + pynini.closure(pynini.accep("s"), 0, 1)) @ date
56
+ date_year_two_digit = (NEMO_DIGIT**2 + pynini.closure(pynini.accep("s"), 0, 1)) @ date
57
+ year_to_year_graph = (
58
+ date_year_four_digit
59
+ + delete_space
60
+ + pynini.cross("-", " to ")
61
+ + delete_space
62
+ + (date_year_four_digit | date_year_two_digit | (NEMO_DIGIT**2 @ cardinal))
63
+ )
64
+ mid_year_graph = pynini.accep("mid") + pynini.cross("-", " ") + (date_year_four_digit | date_year_two_digit)
65
+
66
+ self.graph |= year_to_year_graph
67
+ self.graph |= mid_year_graph
68
+
69
+ # ADDITION
70
+ range_graph = cardinal + pynini.closure(pynini.cross("+", " plus ") + cardinal, 1)
71
+ range_graph |= cardinal + pynini.closure(pynini.cross(" + ", " plus ") + cardinal, 1)
72
+ range_graph |= approx + cardinal
73
+ range_graph |= cardinal + (pynini.cross("...", " ... ") | pynini.accep(" ... ")) + cardinal
74
+
75
+ if not deterministic or lm:
76
+ # cardinal ----
77
+ cardinal_to_cardinal_graph = (
78
+ cardinal + delete_space + pynini.cross("-", pynini.union(" to ", " minus ")) + delete_space + cardinal
79
+ )
80
+
81
+ range_graph |= cardinal_to_cardinal_graph | (
82
+ cardinal + delete_space + pynini.cross(":", " to ") + delete_space + cardinal
83
+ )
84
+
85
+ # MULTIPLY
86
+ for x in [" x ", "x"]:
87
+ range_graph |= cardinal + pynini.cross(x, pynini.union(" by ", " times ")) + cardinal
88
+
89
+ # 40x -> "40 times" ("40 x" cases is covered in serial)
90
+ for x in [" x", "x"]:
91
+ range_graph |= cardinal + pynini.cross(x, " times")
92
+
93
+ # 5x to 7x-> five to seven x/times
94
+ range_graph |= (
95
+ cardinal
96
+ + pynutil.delete(x)
97
+ + pynini.union(" to ", "-", " - ")
98
+ + cardinal
99
+ + pynini.cross(x, pynini.union(" x", " times"))
100
+ )
101
+
102
+ for x in ["*", " * "]:
103
+ range_graph |= cardinal + pynini.closure(pynini.cross(x, " times ") + cardinal, 1)
104
+
105
+ # supports "No. 12" -> "Number 12"
106
+ range_graph |= (
107
+ (pynini.cross(pynini.union("NO", "No"), "Number") | pynini.cross("no", "number"))
108
+ + pynini.closure(pynini.union(". ", " "), 0, 1)
109
+ + cardinal
110
+ )
111
+
112
+ for x in ["/", " / "]:
113
+ range_graph |= cardinal + pynini.closure(pynini.cross(x, " divided by ") + cardinal, 1)
114
+
115
+ # 10% to 20% -> ten to twenty percent
116
+ range_graph |= (
117
+ cardinal
118
+ + pynini.closure(pynini.cross("%", " percent") | pynutil.delete("%"), 0, 1)
119
+ + pynini.union(" to ", "-", " - ")
120
+ + cardinal
121
+ + pynini.cross("%", " percent")
122
+ )
123
+
124
+ self.graph |= range_graph
125
+
126
+ self.graph = self.graph.optimize()
127
+ graph = pynutil.insert("name: \"") + convert_space(self.graph).optimize() + pynutil.insert("\"")
128
+ self.fst = graph.optimize()
vendor/zonos3/vendor/nemo_text_processing/text_normalization/en/taggers/roman.py ADDED
@@ -0,0 +1,115 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # Copyright (c) 2021, NVIDIA CORPORATION. All rights reserved.
2
+ #
3
+ # Licensed under the Apache License, Version 2.0 (the "License");
4
+ # you may not use this file except in compliance with the License.
5
+ # You may obtain a copy of the License at
6
+ #
7
+ # http://www.apache.org/licenses/LICENSE-2.0
8
+ #
9
+ # Unless required by applicable law or agreed to in writing, software
10
+ # distributed under the License is distributed on an "AS IS" BASIS,
11
+ # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
12
+ # See the License for the specific language governing permissions and
13
+ # limitations under the License.
14
+
15
+
16
+ import pynini
17
+ from pynini.lib import pynutil
18
+
19
+ from zonos3.vendor.nemo_text_processing.text_normalization.en.graph_utils import NEMO_ALPHA, NEMO_SIGMA, GraphFst
20
+ from zonos3.vendor.nemo_text_processing.text_normalization.en.utils import get_abs_path, load_labels
21
+
22
+
23
+ class RomanFst(GraphFst):
24
+ """
25
+ Finite state transducer for classifying roman numbers:
26
+ e.g. "IV" -> tokens { roman { integer: "four" } }
27
+
28
+ Args:
29
+ deterministic: if True will provide a single transduction option,
30
+ for False multiple transduction are generated (used for audio-based normalization)
31
+ """
32
+
33
+ def __init__(self, deterministic: bool = True, lm: bool = False):
34
+ super().__init__(name="roman", kind="classify", deterministic=deterministic)
35
+
36
+ roman_dict = load_labels(get_abs_path("data/roman/roman_to_spoken.tsv"))
37
+ default_graph = pynini.string_map(roman_dict).optimize()
38
+ default_graph = pynutil.insert("integer: \"") + default_graph + pynutil.insert("\"")
39
+ ordinal_limit = 19
40
+
41
+ if deterministic:
42
+ # exclude "I"
43
+ start_idx = 1
44
+ else:
45
+ start_idx = 0
46
+
47
+ graph_teens = pynini.string_map([x[0] for x in roman_dict[start_idx:ordinal_limit]]).optimize()
48
+
49
+ # roman numerals up to ordinal_limit with a preceding name are converted to ordinal form
50
+ names = get_names()
51
+ graph = (
52
+ pynutil.insert("key_the_ordinal: \"")
53
+ + names
54
+ + pynutil.insert("\"")
55
+ + pynini.accep(" ")
56
+ + graph_teens @ default_graph
57
+ ).optimize()
58
+
59
+ # single symbol roman numerals with preceding key words (multiple formats) are converted to cardinal form
60
+ key_words = []
61
+ for k_word in load_labels(get_abs_path("data/roman/key_word.tsv")):
62
+ key_words.append(k_word)
63
+ key_words.append([k_word[0][0].upper() + k_word[0][1:]])
64
+ key_words.append([k_word[0].upper()])
65
+
66
+ key_words = pynini.string_map(key_words).optimize()
67
+ graph |= (
68
+ pynutil.insert("key_cardinal: \"") + key_words + pynutil.insert("\"") + pynini.accep(" ") + default_graph
69
+ ).optimize()
70
+
71
+ if deterministic or lm:
72
+ # two digit roman numerals up to 49
73
+ roman_to_cardinal = pynini.compose(
74
+ pynini.closure(NEMO_ALPHA, 2),
75
+ (
76
+ pynutil.insert("default_cardinal: \"default\" ")
77
+ + (pynini.string_map([x[0] for x in roman_dict[:50]]).optimize()) @ default_graph
78
+ ),
79
+ )
80
+ graph |= roman_to_cardinal
81
+ elif not lm:
82
+ # two or more digit roman numerals
83
+ roman_to_cardinal = pynini.compose(
84
+ pynini.difference(NEMO_SIGMA, "I"),
85
+ (
86
+ pynutil.insert("default_cardinal: \"default\" integer: \"")
87
+ + pynini.string_map(roman_dict).optimize()
88
+ + pynutil.insert("\"")
89
+ ),
90
+ ).optimize()
91
+ graph |= roman_to_cardinal
92
+
93
+ # convert three digit roman or up with suffix to ordinal
94
+ roman_to_ordinal = pynini.compose(
95
+ pynini.closure(NEMO_ALPHA, 3),
96
+ (pynutil.insert("default_ordinal: \"default\" ") + graph_teens @ default_graph + pynutil.delete("th")),
97
+ )
98
+
99
+ graph |= roman_to_ordinal
100
+ graph = self.add_tokens(graph.optimize())
101
+
102
+ self.fst = graph.optimize()
103
+
104
+
105
+ def get_names():
106
+ """
107
+ Returns the graph that matched common male and female names.
108
+ """
109
+ male_labels = load_labels(get_abs_path("data/roman/male.tsv"))
110
+ female_labels = load_labels(get_abs_path("data/roman/female.tsv"))
111
+ male_labels.extend([[x[0].upper()] for x in male_labels])
112
+ female_labels.extend([[x[0].upper()] for x in female_labels])
113
+ names = pynini.string_map(male_labels).optimize()
114
+ names |= pynini.string_map(female_labels).optimize()
115
+ return names
vendor/zonos3/vendor/nemo_text_processing/text_normalization/en/taggers/serial.py ADDED
@@ -0,0 +1,143 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # Copyright (c) 2022, NVIDIA CORPORATION. All rights reserved.
2
+ #
3
+ # Licensed under the Apache License, Version 2.0 (the "License");
4
+ # you may not use this file except in compliance with the License.
5
+ # You may obtain a copy of the License at
6
+ #
7
+ # http://www.apache.org/licenses/LICENSE-2.0
8
+ #
9
+ # Unless required by applicable law or agreed to in writing, software
10
+ # distributed under the License is distributed on an "AS IS" BASIS,
11
+ # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
12
+ # See the License for the specific language governing permissions and
13
+ # limitations under the License.
14
+
15
+
16
+ import pynini
17
+ from pynini.examples import plurals
18
+ from pynini.lib import pynutil
19
+
20
+ from zonos3.vendor.nemo_text_processing.text_normalization.en.graph_utils import (
21
+ NEMO_ALPHA,
22
+ NEMO_DIGIT,
23
+ NEMO_NOT_SPACE,
24
+ NEMO_SIGMA,
25
+ GraphFst,
26
+ convert_space,
27
+ )
28
+ from zonos3.vendor.nemo_text_processing.text_normalization.en.utils import get_abs_path, load_labels
29
+
30
+
31
+ class SerialFst(GraphFst):
32
+ """
33
+ This class is a composite class of two other class instances
34
+
35
+ Args:
36
+ time: composed tagger and verbalizer
37
+ date: composed tagger and verbalizer
38
+ cardinal: tagger
39
+ deterministic: if True will provide a single transduction option,
40
+ for False multiple transduction are generated (used for audio-based normalization)
41
+ lm: whether to use for hybrid LM
42
+ """
43
+
44
+ def __init__(self, cardinal: GraphFst, ordinal: GraphFst, deterministic: bool = True, lm: bool = False):
45
+ super().__init__(name="integer", kind="classify", deterministic=deterministic)
46
+
47
+ """
48
+ Finite state transducer for classifying serial (handles only cases without delimiters,
49
+ values with delimiters are handled by default).
50
+ The serial is a combination of digits, letters and dashes, e.g.:
51
+ c325b -> tokens { cardinal { integer: "c three two five b" } }
52
+ """
53
+ if deterministic:
54
+ num_graph = pynini.compose(NEMO_DIGIT ** (6, ...), cardinal.single_digits_graph).optimize()
55
+ num_graph |= pynini.compose(NEMO_DIGIT ** (1, 5), cardinal.graph).optimize()
56
+ # to handle numbers starting with zero
57
+ num_graph |= pynini.compose(
58
+ pynini.accep("0") + pynini.closure(NEMO_DIGIT), cardinal.single_digits_graph
59
+ ).optimize()
60
+ else:
61
+ num_graph = cardinal.final_graph
62
+
63
+ # TODO: "#" doesn't work from the file
64
+ symbols_graph = pynini.string_file(get_abs_path("data/whitelist/symbol.tsv")).optimize() | pynini.cross(
65
+ "#", "hash"
66
+ )
67
+ num_graph |= symbols_graph
68
+
69
+ if not self.deterministic and not lm:
70
+ num_graph |= cardinal.single_digits_graph
71
+ num_graph |= pynini.compose(num_graph, NEMO_SIGMA + pynutil.delete("hundred ") + NEMO_SIGMA)
72
+ # also allow double digits to be pronounced as integer in serial number
73
+ num_graph |= pynutil.add_weight(
74
+ NEMO_DIGIT**2 @ cardinal.graph_hundred_component_at_least_one_none_zero_digit, weight=0.0001
75
+ )
76
+
77
+ # add space between letter and digit/symbol
78
+ symbols = [x[0] for x in load_labels(get_abs_path("data/whitelist/symbol.tsv"))]
79
+ symbols = pynini.union(*symbols)
80
+ digit_symbol = NEMO_DIGIT | symbols
81
+
82
+ graph_with_space = pynini.compose(
83
+ pynini.cdrewrite(pynutil.insert(" "), NEMO_ALPHA | symbols, digit_symbol, NEMO_SIGMA),
84
+ pynini.cdrewrite(pynutil.insert(" "), digit_symbol, NEMO_ALPHA | symbols, NEMO_SIGMA),
85
+ )
86
+
87
+ # serial graph with delimiter
88
+ delimiter = pynini.accep("-") | pynini.accep("/") | pynini.accep(" ")
89
+ if not deterministic:
90
+ delimiter |= pynini.cross("-", " dash ") | pynini.cross("/", " slash ")
91
+
92
+ alphas = pynini.closure(NEMO_ALPHA, 1)
93
+ letter_num = alphas + delimiter + num_graph
94
+ num_letter = pynini.closure(num_graph + delimiter, 1) + alphas
95
+ next_alpha_or_num = pynini.closure(delimiter + (alphas | num_graph))
96
+ next_alpha_or_num |= pynini.closure(
97
+ delimiter
98
+ + num_graph
99
+ + plurals._priority_union(pynini.accep(" "), pynutil.insert(" "), NEMO_SIGMA).optimize()
100
+ + alphas
101
+ )
102
+
103
+ serial_graph = letter_num + next_alpha_or_num
104
+ serial_graph |= num_letter + next_alpha_or_num
105
+ # numbers only with 2+ delimiters
106
+ serial_graph |= (
107
+ num_graph + delimiter + num_graph + delimiter + num_graph + pynini.closure(delimiter + num_graph)
108
+ )
109
+ # 2+ symbols
110
+ serial_graph |= pynini.compose(NEMO_SIGMA + symbols + NEMO_SIGMA, num_graph + delimiter + num_graph)
111
+
112
+ # exclude ordinal numbers from serial options
113
+ serial_graph = pynini.compose(
114
+ pynini.difference(NEMO_SIGMA, pynini.project(ordinal.graph, "input")), serial_graph
115
+ ).optimize()
116
+
117
+ serial_graph = pynutil.add_weight(serial_graph, 0.0001)
118
+ serial_graph |= (
119
+ pynini.closure(NEMO_NOT_SPACE, 1)
120
+ + (pynini.cross("^2", " squared") | pynini.cross("^3", " cubed")).optimize()
121
+ )
122
+
123
+ # at least one serial graph with alpha numeric value and optional additional serial/num/alpha values
124
+ serial_graph = (
125
+ pynini.closure((serial_graph | num_graph | alphas) + delimiter)
126
+ + serial_graph
127
+ + pynini.closure(delimiter + (serial_graph | num_graph | alphas))
128
+ )
129
+
130
+ serial_graph |= pynini.compose(graph_with_space, serial_graph.optimize()).optimize()
131
+ serial_graph = pynini.compose(pynini.closure(NEMO_NOT_SPACE, 2), serial_graph).optimize()
132
+
133
+ # this is not to verbolize "/" as "slash" in cases like "import/export"
134
+ serial_graph = pynini.compose(
135
+ pynini.difference(
136
+ NEMO_SIGMA, pynini.closure(NEMO_ALPHA, 1) + pynini.accep("/") + pynini.closure(NEMO_ALPHA, 1)
137
+ ),
138
+ serial_graph,
139
+ )
140
+
141
+ self.graph = serial_graph.optimize()
142
+ graph = pynutil.insert("name: \"") + convert_space(self.graph).optimize() + pynutil.insert("\"")
143
+ self.fst = graph.optimize()
vendor/zonos3/vendor/nemo_text_processing/text_normalization/en/taggers/telephone.py ADDED
@@ -0,0 +1,134 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # Copyright (c) 2021, NVIDIA CORPORATION. All rights reserved.
2
+ #
3
+ # Licensed under the Apache License, Version 2.0 (the "License");
4
+ # you may not use this file except in compliance with the License.
5
+ # You may obtain a copy of the License at
6
+ #
7
+ # http://www.apache.org/licenses/LICENSE-2.0
8
+ #
9
+ # Unless required by applicable law or agreed to in writing, software
10
+ # distributed under the License is distributed on an "AS IS" BASIS,
11
+ # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
12
+ # See the License for the specific language governing permissions and
13
+ # limitations under the License.
14
+
15
+ import pynini
16
+ from pynini.lib import pynutil
17
+
18
+ from zonos3.vendor.nemo_text_processing.text_normalization.en.graph_utils import (
19
+ NEMO_ALPHA,
20
+ NEMO_DIGIT,
21
+ NEMO_SIGMA,
22
+ GraphFst,
23
+ delete_extra_space,
24
+ delete_space,
25
+ insert_space,
26
+ plurals,
27
+ )
28
+ from zonos3.vendor.nemo_text_processing.text_normalization.en.utils import get_abs_path
29
+
30
+
31
+ class TelephoneFst(GraphFst):
32
+ """
33
+ Finite state transducer for classifying telephone, and IP, and SSN which includes country code, number part and extension
34
+ country code optional: +***
35
+ number part: ***-***-****, or (***) ***-****
36
+ extension optional: 1-9999
37
+ E.g
38
+ +1 123-123-5678-1 -> telephone { country_code: "one" number_part: "one two three, one two three, five six seven eight" extension: "one" }
39
+ 1-800-GO-U-HAUL -> telephone { country_code: "one" number_part: "one, eight hundred GO U HAUL" }
40
+ Args:
41
+ deterministic: if True will provide a single transduction option,
42
+ for False multiple transduction are generated (used for audio-based normalization)
43
+ """
44
+
45
+ def __init__(self, deterministic: bool = True):
46
+ super().__init__(name="telephone", kind="classify", deterministic=deterministic)
47
+
48
+ add_separator = pynutil.insert(", ") # between components
49
+ zero = pynini.cross("0", "zero")
50
+ if not deterministic:
51
+ zero |= pynini.cross("0", pynini.union("o", "oh"))
52
+ digit = pynini.invert(pynini.string_file(get_abs_path("data/number/digit.tsv"))).optimize() | zero
53
+
54
+ telephone_prompts = pynini.string_file(get_abs_path("data/telephone/telephone_prompt.tsv"))
55
+ country_code = (
56
+ pynini.closure(telephone_prompts + delete_extra_space, 0, 1)
57
+ + pynini.closure(pynini.cross("+", "plus "), 0, 1)
58
+ + pynini.closure(digit + insert_space, 0, 2)
59
+ + digit
60
+ + pynutil.insert(",")
61
+ )
62
+ country_code |= telephone_prompts
63
+ country_code = pynutil.insert("country_code: \"") + country_code + pynutil.insert("\"")
64
+ country_code = country_code + pynini.closure(pynutil.delete("-"), 0, 1) + delete_space + insert_space
65
+
66
+ area_part_default = pynini.closure(digit + insert_space, 2, 2) + digit
67
+ area_part = pynini.cross("800", "eight hundred") | pynini.compose(
68
+ pynini.difference(NEMO_SIGMA, "800"), area_part_default
69
+ )
70
+
71
+ area_part = (
72
+ (area_part + (pynutil.delete("-") | pynutil.delete(".")))
73
+ | (
74
+ pynutil.delete("(")
75
+ + area_part
76
+ + ((pynutil.delete(")") + pynini.closure(pynutil.delete(" "), 0, 1)) | pynutil.delete(")-"))
77
+ )
78
+ ) + add_separator
79
+
80
+ del_separator = pynini.closure(pynini.union("-", " ", "."), 0, 1)
81
+ number_length = ((NEMO_DIGIT + del_separator) | (NEMO_ALPHA + del_separator)) ** 7
82
+ number_words = pynini.closure(
83
+ (NEMO_DIGIT @ digit) + (insert_space | (pynini.cross("-", ', ')))
84
+ | NEMO_ALPHA
85
+ | (NEMO_ALPHA + pynini.cross("-", ' '))
86
+ )
87
+ number_words |= pynini.closure(
88
+ (NEMO_DIGIT @ digit) + (insert_space | (pynini.cross(".", ', ')))
89
+ | NEMO_ALPHA
90
+ | (NEMO_ALPHA + pynini.cross(".", ' '))
91
+ )
92
+ number_words = pynini.compose(number_length, number_words)
93
+ number_part = area_part + number_words
94
+ number_part = pynutil.insert("number_part: \"") + number_part + pynutil.insert("\"")
95
+ extension = (
96
+ pynutil.insert("extension: \"") + pynini.closure(digit + insert_space, 0, 3) + digit + pynutil.insert("\"")
97
+ )
98
+ extension = pynini.closure(insert_space + extension, 0, 1)
99
+
100
+ graph = plurals._priority_union(country_code + number_part, number_part, NEMO_SIGMA).optimize()
101
+ graph = plurals._priority_union(country_code + number_part + extension, graph, NEMO_SIGMA).optimize()
102
+ graph = plurals._priority_union(number_part + extension, graph, NEMO_SIGMA).optimize()
103
+
104
+ # ip
105
+ ip_prompts = pynini.string_file(get_abs_path("data/telephone/ip_prompt.tsv"))
106
+ digit_to_str_graph = digit + pynini.closure(pynutil.insert(" ") + digit, 0, 2)
107
+ ip_graph = digit_to_str_graph + (pynini.cross(".", " dot ") + digit_to_str_graph) ** 3
108
+ graph |= (
109
+ pynini.closure(
110
+ pynutil.insert("country_code: \"") + ip_prompts + pynutil.insert("\"") + delete_extra_space, 0, 1
111
+ )
112
+ + pynutil.insert("number_part: \"")
113
+ + ip_graph.optimize()
114
+ + pynutil.insert("\"")
115
+ )
116
+ # ssn
117
+ ssn_prompts = pynini.string_file(get_abs_path("data/telephone/ssn_prompt.tsv"))
118
+ three_digit_part = digit + (pynutil.insert(" ") + digit) ** 2
119
+ two_digit_part = digit + pynutil.insert(" ") + digit
120
+ four_digit_part = digit + (pynutil.insert(" ") + digit) ** 3
121
+ ssn_separator = pynini.cross("-", ", ")
122
+ ssn_graph = three_digit_part + ssn_separator + two_digit_part + ssn_separator + four_digit_part
123
+
124
+ graph |= (
125
+ pynini.closure(
126
+ pynutil.insert("country_code: \"") + ssn_prompts + pynutil.insert("\"") + delete_extra_space, 0, 1
127
+ )
128
+ + pynutil.insert("number_part: \"")
129
+ + ssn_graph.optimize()
130
+ + pynutil.insert("\"")
131
+ )
132
+
133
+ final_graph = self.add_tokens(graph)
134
+ self.fst = final_graph.optimize()
vendor/zonos3/vendor/nemo_text_processing/text_normalization/en/taggers/time.py ADDED
@@ -0,0 +1,133 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # Copyright (c) 2021, NVIDIA CORPORATION. All rights reserved.
2
+ #
3
+ # Licensed under the Apache License, Version 2.0 (the "License");
4
+ # you may not use this file except in compliance with the License.
5
+ # You may obtain a copy of the License at
6
+ #
7
+ # http://www.apache.org/licenses/LICENSE-2.0
8
+ #
9
+ # Unless required by applicable law or agreed to in writing, software
10
+ # distributed under the License is distributed on an "AS IS" BASIS,
11
+ # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
12
+ # See the License for the specific language governing permissions and
13
+ # limitations under the License.
14
+
15
+
16
+ import pynini
17
+ from pynini.lib import pynutil
18
+
19
+ from zonos3.vendor.nemo_text_processing.text_normalization.en.graph_utils import (
20
+ NEMO_DIGIT,
21
+ GraphFst,
22
+ convert_space,
23
+ delete_space,
24
+ insert_space,
25
+ )
26
+ from zonos3.vendor.nemo_text_processing.text_normalization.en.utils import (
27
+ augment_labels_with_punct_at_end,
28
+ get_abs_path,
29
+ load_labels,
30
+ )
31
+
32
+
33
+ class TimeFst(GraphFst):
34
+ """
35
+ Finite state transducer for classifying time, e.g.
36
+ 12:30 a.m. est -> time { hours: "twelve" minutes: "thirty" suffix: "a m" zone: "e s t" }
37
+ 2.30 a.m. -> time { hours: "two" minutes: "thirty" suffix: "a m" }
38
+ 02.30 a.m. -> time { hours: "two" minutes: "thirty" suffix: "a m" }
39
+ 2.00 a.m. -> time { hours: "two" suffix: "a m" }
40
+ 2 a.m. -> time { hours: "two" suffix: "a m" }
41
+ 02:00 -> time { hours: "two" }
42
+ 2:00 -> time { hours: "two" }
43
+ 10:00:05 a.m. -> time { hours: "ten" minutes: "zero" seconds: "five" suffix: "a m" }
44
+
45
+ Args:
46
+ cardinal: CardinalFst
47
+ deterministic: if True will provide a single transduction option,
48
+ for False multiple transduction are generated (used for audio-based normalization)
49
+ """
50
+
51
+ def __init__(self, cardinal: GraphFst, deterministic: bool = True):
52
+ super().__init__(name="time", kind="classify", deterministic=deterministic)
53
+ suffix_labels = load_labels(get_abs_path("data/time/suffix.tsv"))
54
+ suffix_labels.extend(augment_labels_with_punct_at_end(suffix_labels))
55
+ suffix_graph = pynini.string_map(suffix_labels)
56
+
57
+ time_zone_graph = pynini.string_file(get_abs_path("data/time/zone.tsv"))
58
+
59
+ # only used for < 1000 thousand -> 0 weight
60
+ cardinal = cardinal.graph
61
+
62
+ labels_hour = [str(x) for x in range(0, 24)]
63
+ labels_minute_single = [str(x) for x in range(1, 10)]
64
+ labels_minute_double = [str(x) for x in range(10, 60)]
65
+
66
+ delete_leading_zero_to_double_digit = (NEMO_DIGIT + NEMO_DIGIT) | (
67
+ pynini.closure(pynutil.delete("0"), 0, 1) + NEMO_DIGIT
68
+ )
69
+
70
+ graph_hour = delete_leading_zero_to_double_digit @ pynini.union(*labels_hour) @ cardinal
71
+
72
+ graph_minute_single = pynini.union(*labels_minute_single) @ cardinal
73
+ graph_minute_double = pynini.union(*labels_minute_double) @ cardinal
74
+
75
+ final_graph_hour = pynutil.insert("hours: \"") + graph_hour + pynutil.insert("\"")
76
+ final_graph_minute = (
77
+ pynutil.insert("minutes: \"")
78
+ + (pynini.cross("0", "o") + insert_space + graph_minute_single | graph_minute_double)
79
+ + pynutil.insert("\"")
80
+ )
81
+ final_graph_second = (
82
+ pynutil.insert("seconds: \"")
83
+ + (pynini.cross("0", "o") + insert_space + graph_minute_single | graph_minute_double)
84
+ + pynutil.insert("\"")
85
+ )
86
+ final_suffix = pynutil.insert("suffix: \"") + convert_space(suffix_graph) + pynutil.insert("\"")
87
+ final_suffix_optional = pynini.closure(delete_space + insert_space + final_suffix, 0, 1)
88
+ final_time_zone_optional = pynini.closure(
89
+ delete_space
90
+ + insert_space
91
+ + pynutil.insert("zone: \"")
92
+ + convert_space(time_zone_graph)
93
+ + pynutil.insert("\""),
94
+ 0,
95
+ 1,
96
+ )
97
+
98
+ # 2:30 pm, 02:30, 2:00
99
+ graph_hm = (
100
+ final_graph_hour
101
+ + pynutil.delete(":")
102
+ + (pynutil.delete("00") | insert_space + final_graph_minute)
103
+ + final_suffix_optional
104
+ + final_time_zone_optional
105
+ )
106
+
107
+ # 10:30:05 pm,
108
+ graph_hms = (
109
+ final_graph_hour
110
+ + pynutil.delete(":")
111
+ + (pynini.cross("00", " minutes: \"zero\"") | insert_space + final_graph_minute)
112
+ + pynutil.delete(":")
113
+ + (pynini.cross("00", " seconds: \"zero\"") | insert_space + final_graph_second)
114
+ + final_suffix_optional
115
+ + final_time_zone_optional
116
+ )
117
+
118
+ # 2.xx pm/am
119
+ graph_hm2 = (
120
+ final_graph_hour
121
+ + pynutil.delete(".")
122
+ + (pynutil.delete("00") | insert_space + final_graph_minute)
123
+ + delete_space
124
+ + insert_space
125
+ + final_suffix
126
+ + final_time_zone_optional
127
+ )
128
+ # 2 pm est
129
+ graph_h = final_graph_hour + delete_space + insert_space + final_suffix + final_time_zone_optional
130
+ final_graph = (graph_hm | graph_h | graph_hm2 | graph_hms).optimize()
131
+
132
+ final_graph = self.add_tokens(final_graph)
133
+ self.fst = final_graph.optimize()
vendor/zonos3/vendor/nemo_text_processing/text_normalization/en/taggers/tokenize_and_classify.py ADDED
@@ -0,0 +1,241 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # Copyright (c) 2021, NVIDIA CORPORATION. All rights reserved.
2
+ #
3
+ # Licensed under the Apache License, Version 2.0 (the "License");
4
+ # you may not use this file except in compliance with the License.
5
+ # You may obtain a copy of the License at
6
+ #
7
+ # http://www.apache.org/licenses/LICENSE-2.0
8
+ #
9
+ # Unless required by applicable law or agreed to in writing, software
10
+ # distributed under the License is distributed on an "AS IS" BASIS,
11
+ # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
12
+ # See the License for the specific language governing permissions and
13
+ # limitations under the License.
14
+
15
+ import os
16
+ import time
17
+
18
+ import pynini
19
+ from pynini.lib import pynutil
20
+
21
+ from zonos3.vendor.nemo_text_processing.text_normalization.en.graph_utils import (
22
+ NEMO_WHITE_SPACE,
23
+ GraphFst,
24
+ delete_extra_space,
25
+ delete_space,
26
+ generator_main,
27
+ )
28
+ from zonos3.vendor.nemo_text_processing.text_normalization.en.taggers.abbreviation import AbbreviationFst
29
+ from zonos3.vendor.nemo_text_processing.text_normalization.en.taggers.cardinal import CardinalFst
30
+ from zonos3.vendor.nemo_text_processing.text_normalization.en.taggers.date import DateFst
31
+ from zonos3.vendor.nemo_text_processing.text_normalization.en.taggers.decimal import DecimalFst
32
+ from zonos3.vendor.nemo_text_processing.text_normalization.en.taggers.electronic import ElectronicFst
33
+ from zonos3.vendor.nemo_text_processing.text_normalization.en.taggers.fraction import FractionFst
34
+ from zonos3.vendor.nemo_text_processing.text_normalization.en.taggers.measure import MeasureFst
35
+ from zonos3.vendor.nemo_text_processing.text_normalization.en.taggers.money import MoneyFst
36
+ from zonos3.vendor.nemo_text_processing.text_normalization.en.taggers.ordinal import OrdinalFst
37
+ from zonos3.vendor.nemo_text_processing.text_normalization.en.taggers.punctuation import PunctuationFst
38
+ from zonos3.vendor.nemo_text_processing.text_normalization.en.taggers.range import RangeFst as RangeFst
39
+ from zonos3.vendor.nemo_text_processing.text_normalization.en.taggers.roman import RomanFst
40
+ from zonos3.vendor.nemo_text_processing.text_normalization.en.taggers.serial import SerialFst
41
+ from zonos3.vendor.nemo_text_processing.text_normalization.en.taggers.telephone import TelephoneFst
42
+ from zonos3.vendor.nemo_text_processing.text_normalization.en.taggers.time import TimeFst
43
+ from zonos3.vendor.nemo_text_processing.text_normalization.en.taggers.whitelist import WhiteListFst
44
+ from zonos3.vendor.nemo_text_processing.text_normalization.en.taggers.word import WordFst
45
+ from zonos3.vendor.nemo_text_processing.text_normalization.en.verbalizers.date import DateFst as vDateFst
46
+ from zonos3.vendor.nemo_text_processing.text_normalization.en.verbalizers.ordinal import OrdinalFst as vOrdinalFst
47
+ from zonos3.vendor.nemo_text_processing.text_normalization.en.verbalizers.time import TimeFst as vTimeFst
48
+ from zonos3.vendor.nemo_text_processing.utils.logging import logger
49
+
50
+
51
+ class ClassifyFst(GraphFst):
52
+ """
53
+ Final class that composes all other classification grammars. This class can process an entire sentence including punctuation.
54
+ For deployment, this grammar will be compiled and exported to OpenFst Finite State Archive (FAR) File.
55
+ More details to deployment at NeMo/tools/text_processing_deployment.
56
+
57
+ Args:
58
+ input_case: accepting either "lower_cased" or "cased" input.
59
+ deterministic: if True will provide a single transduction option,
60
+ for False multiple options (used for audio-based normalization)
61
+ cache_dir: path to a dir with .far grammar file. Set to None to avoid using cache.
62
+ overwrite_cache: set to True to overwrite .far files
63
+ whitelist: path to a file with whitelist replacements
64
+ """
65
+
66
+ def __init__(
67
+ self,
68
+ input_case: str,
69
+ deterministic: bool = True,
70
+ cache_dir: str = None,
71
+ overwrite_cache: bool = False,
72
+ whitelist: str = None,
73
+ ):
74
+ super().__init__(name="tokenize_and_classify", kind="classify", deterministic=deterministic)
75
+
76
+ far_file = None
77
+ if cache_dir is not None and cache_dir != "None":
78
+ os.makedirs(cache_dir, exist_ok=True)
79
+ whitelist_file = os.path.basename(whitelist) if whitelist else ""
80
+ far_file = os.path.join(
81
+ cache_dir,
82
+ f"en_tn_{deterministic}_deterministic_{input_case}_{whitelist_file}_tokenize.far",
83
+ )
84
+ if not overwrite_cache and far_file and os.path.exists(far_file):
85
+ self.fst = pynini.Far(far_file, mode="r")["tokenize_and_classify"]
86
+ logger.info(f"ClassifyFst.fst was restored from {far_file}.")
87
+ else:
88
+ logger.info(f"Creating ClassifyFst grammars.")
89
+
90
+ start_time = time.time()
91
+ cardinal = CardinalFst(deterministic=deterministic)
92
+ cardinal_graph = cardinal.fst
93
+ logger.debug(f"cardinal: {time.time() - start_time: .2f}s -- {cardinal_graph.num_states()} nodes")
94
+
95
+ start_time = time.time()
96
+ ordinal = OrdinalFst(cardinal=cardinal, deterministic=deterministic)
97
+ ordinal_graph = ordinal.fst
98
+ logger.debug(f"ordinal: {time.time() - start_time: .2f}s -- {ordinal_graph.num_states()} nodes")
99
+
100
+ start_time = time.time()
101
+ decimal = DecimalFst(cardinal=cardinal, deterministic=deterministic)
102
+ decimal_graph = decimal.fst
103
+ logger.debug(f"decimal: {time.time() - start_time: .2f}s -- {decimal_graph.num_states()} nodes")
104
+
105
+ start_time = time.time()
106
+ fraction = FractionFst(deterministic=deterministic, cardinal=cardinal)
107
+ fraction_graph = fraction.fst
108
+ logger.debug(f"fraction: {time.time() - start_time: .2f}s -- {fraction_graph.num_states()} nodes")
109
+
110
+ start_time = time.time()
111
+ measure = MeasureFst(
112
+ cardinal=cardinal,
113
+ decimal=decimal,
114
+ fraction=fraction,
115
+ deterministic=deterministic,
116
+ )
117
+ measure_graph = measure.fst
118
+ logger.debug(f"measure: {time.time() - start_time: .2f}s -- {measure_graph.num_states()} nodes")
119
+
120
+ start_time = time.time()
121
+ date_graph = DateFst(cardinal=cardinal, deterministic=deterministic).fst
122
+ logger.debug(f"date: {time.time() - start_time: .2f}s -- {date_graph.num_states()} nodes")
123
+
124
+ start_time = time.time()
125
+ time_graph = TimeFst(cardinal=cardinal, deterministic=deterministic).fst
126
+ logger.debug(f"time: {time.time() - start_time: .2f}s -- {time_graph.num_states()} nodes")
127
+
128
+ start_time = time.time()
129
+ telephone_graph = TelephoneFst(deterministic=deterministic).fst
130
+ logger.debug(f"telephone: {time.time() - start_time: .2f}s -- {telephone_graph.num_states()} nodes")
131
+
132
+ start_time = time.time()
133
+ electonic_graph = ElectronicFst(cardinal=cardinal, deterministic=deterministic).fst
134
+ logger.debug(f"electronic: {time.time() - start_time: .2f}s -- {electonic_graph.num_states()} nodes")
135
+
136
+ start_time = time.time()
137
+ money_graph = MoneyFst(cardinal=cardinal, decimal=decimal, deterministic=deterministic).fst
138
+ logger.debug(f"money: {time.time() - start_time: .2f}s -- {money_graph.num_states()} nodes")
139
+
140
+ start_time = time.time()
141
+ whitelist_graph = WhiteListFst(
142
+ input_case=input_case, deterministic=deterministic, input_file=whitelist
143
+ ).fst
144
+ logger.debug(f"whitelist: {time.time() - start_time: .2f}s -- {whitelist_graph.num_states()} nodes")
145
+
146
+ start_time = time.time()
147
+ punctuation = PunctuationFst(deterministic=deterministic)
148
+ punct_graph = punctuation.fst
149
+ logger.debug(f"punct: {time.time() - start_time: .2f}s -- {punct_graph.num_states()} nodes")
150
+
151
+ start_time = time.time()
152
+ word_graph = WordFst(punctuation=punctuation, deterministic=deterministic).fst
153
+ logger.debug(f"word: {time.time() - start_time: .2f}s -- {word_graph.num_states()} nodes")
154
+
155
+ start_time = time.time()
156
+ serial_graph = SerialFst(cardinal=cardinal, ordinal=ordinal, deterministic=deterministic).fst
157
+ logger.debug(f"serial: {time.time() - start_time: .2f}s -- {serial_graph.num_states()} nodes")
158
+
159
+ start_time = time.time()
160
+ v_time_graph = vTimeFst(deterministic=deterministic).fst
161
+ v_ordinal_graph = vOrdinalFst(deterministic=deterministic)
162
+ v_date_graph = vDateFst(ordinal=v_ordinal_graph, deterministic=deterministic).fst
163
+ time_final = pynini.compose(time_graph, v_time_graph)
164
+ date_final = pynini.compose(date_graph, v_date_graph)
165
+ range_graph = RangeFst(
166
+ time=time_final,
167
+ date=date_final,
168
+ cardinal=cardinal,
169
+ deterministic=deterministic,
170
+ ).fst
171
+ logger.debug(f"range: {time.time() - start_time: .2f}s -- {range_graph.num_states()} nodes")
172
+
173
+ # A quick fix to address money ranges:
174
+ # $150-$200 -> one hundred and fifty dollars to two hundred dollars
175
+
176
+ dash = (pynutil.insert('name: "') + pynini.cross("-", "to") + pynutil.insert('"')).optimize()
177
+
178
+ graph_range_money = pynini.closure(
179
+ money_graph
180
+ + pynutil.insert(" }")
181
+ + pynutil.insert(" tokens { ")
182
+ + dash
183
+ + pynutil.insert(" } ")
184
+ + pynutil.insert("tokens { ")
185
+ + money_graph,
186
+ 1,
187
+ )
188
+
189
+ classify = (
190
+ pynutil.add_weight(whitelist_graph, 1.01)
191
+ | pynutil.add_weight(time_graph, 1.1)
192
+ | pynutil.add_weight(date_graph, 1.09)
193
+ | pynutil.add_weight(decimal_graph, 1.1)
194
+ | pynutil.add_weight(measure_graph, 1.1)
195
+ | pynutil.add_weight(cardinal_graph, 1.1)
196
+ | pynutil.add_weight(ordinal_graph, 1.1)
197
+ | pynutil.add_weight(money_graph, 1.1)
198
+ | pynutil.add_weight(telephone_graph, 1.1)
199
+ | pynutil.add_weight(electonic_graph, 1.11)
200
+ | pynutil.add_weight(fraction_graph, 1.1)
201
+ | pynutil.add_weight(range_graph, 1.1)
202
+ | pynutil.add_weight(serial_graph, 1.12) # should be higher than the rest of the classes
203
+ | pynutil.add_weight(graph_range_money, 1.1)
204
+ )
205
+
206
+ # roman_graph = RomanFst(deterministic=deterministic).fst
207
+ # classify |= pynutil.add_weight(roman_graph, 1.1)
208
+
209
+ if not deterministic:
210
+ abbreviation_graph = AbbreviationFst(deterministic=deterministic).fst
211
+ classify |= pynutil.add_weight(abbreviation_graph, 100)
212
+
213
+ punct = pynutil.insert("tokens { ") + pynutil.add_weight(punct_graph, weight=2.1) + pynutil.insert(" }")
214
+ punct = pynini.closure(
215
+ pynini.compose(pynini.closure(NEMO_WHITE_SPACE, 1), delete_extra_space)
216
+ | (pynutil.insert(" ") + punct)
217
+ | punct,
218
+ 1,
219
+ )
220
+
221
+ classify |= pynutil.add_weight(word_graph, 100)
222
+ token = pynutil.insert("tokens { ") + classify + pynutil.insert(" }")
223
+ token_plus_punct = (
224
+ pynini.closure(punct + pynutil.insert(" ")) + token + pynini.closure(pynutil.insert(" ") + punct)
225
+ )
226
+
227
+ graph = token_plus_punct + pynini.closure(
228
+ (
229
+ pynini.compose(pynini.closure(NEMO_WHITE_SPACE, 1), delete_extra_space)
230
+ | (pynutil.insert(" ") + punct + pynutil.insert(" "))
231
+ )
232
+ + token_plus_punct
233
+ )
234
+
235
+ graph = delete_space + graph + delete_space
236
+ graph |= punct
237
+
238
+ self.fst = graph.optimize()
239
+
240
+ if far_file:
241
+ generator_main(far_file, {"tokenize_and_classify": self.fst})
vendor/zonos3/vendor/nemo_text_processing/text_normalization/en/taggers/tokenize_and_classify_lm.py ADDED
@@ -0,0 +1,227 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # Copyright (c) 2021, NVIDIA CORPORATION. All rights reserved.
2
+ #
3
+ # Licensed under the Apache License, Version 2.0 (the "License");
4
+ # you may not use this file except in compliance with the License.
5
+ # You may obtain a copy of the License at
6
+ #
7
+ # http://www.apache.org/licenses/LICENSE-2.0
8
+ #
9
+ # Unless required by applicable law or agreed to in writing, software
10
+ # distributed under the License is distributed on an "AS IS" BASIS,
11
+ # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
12
+ # See the License for the specific language governing permissions and
13
+ # limitations under the License.
14
+
15
+ import os
16
+
17
+ import pynini
18
+ from pynini.examples import plurals
19
+ from pynini.lib import pynutil
20
+
21
+ from zonos3.vendor.nemo_text_processing.text_normalization.en.graph_utils import (
22
+ NEMO_CHAR,
23
+ NEMO_DIGIT,
24
+ NEMO_NOT_SPACE,
25
+ NEMO_SIGMA,
26
+ NEMO_WHITE_SPACE,
27
+ GraphFst,
28
+ delete_extra_space,
29
+ delete_space,
30
+ generator_main,
31
+ )
32
+ from zonos3.vendor.nemo_text_processing.text_normalization.en.taggers.cardinal import CardinalFst
33
+ from zonos3.vendor.nemo_text_processing.text_normalization.en.taggers.date import DateFst
34
+ from zonos3.vendor.nemo_text_processing.text_normalization.en.taggers.decimal import DecimalFst
35
+ from zonos3.vendor.nemo_text_processing.text_normalization.en.taggers.electronic import ElectronicFst
36
+ from zonos3.vendor.nemo_text_processing.text_normalization.en.taggers.fraction import FractionFst
37
+ from zonos3.vendor.nemo_text_processing.text_normalization.en.taggers.measure import MeasureFst
38
+ from zonos3.vendor.nemo_text_processing.text_normalization.en.taggers.money import MoneyFst
39
+ from zonos3.vendor.nemo_text_processing.text_normalization.en.taggers.ordinal import OrdinalFst
40
+ from zonos3.vendor.nemo_text_processing.text_normalization.en.taggers.punctuation import PunctuationFst
41
+ from zonos3.vendor.nemo_text_processing.text_normalization.en.taggers.range import RangeFst as RangeFst
42
+ from zonos3.vendor.nemo_text_processing.text_normalization.en.taggers.roman import RomanFst
43
+ from zonos3.vendor.nemo_text_processing.text_normalization.en.taggers.serial import SerialFst
44
+ from zonos3.vendor.nemo_text_processing.text_normalization.en.taggers.telephone import TelephoneFst
45
+ from zonos3.vendor.nemo_text_processing.text_normalization.en.taggers.time import TimeFst
46
+ from zonos3.vendor.nemo_text_processing.text_normalization.en.taggers.whitelist import WhiteListFst
47
+ from zonos3.vendor.nemo_text_processing.text_normalization.en.taggers.word import WordFst
48
+ from zonos3.vendor.nemo_text_processing.text_normalization.en.verbalizers.cardinal import CardinalFst as vCardinal
49
+ from zonos3.vendor.nemo_text_processing.text_normalization.en.verbalizers.date import DateFst as vDate
50
+ from zonos3.vendor.nemo_text_processing.text_normalization.en.verbalizers.decimal import DecimalFst as vDecimal
51
+ from zonos3.vendor.nemo_text_processing.text_normalization.en.verbalizers.electronic import ElectronicFst as vElectronic
52
+ from zonos3.vendor.nemo_text_processing.text_normalization.en.verbalizers.fraction import FractionFst as vFraction
53
+ from zonos3.vendor.nemo_text_processing.text_normalization.en.verbalizers.measure import MeasureFst as vMeasure
54
+ from zonos3.vendor.nemo_text_processing.text_normalization.en.verbalizers.money import MoneyFst as vMoney
55
+ from zonos3.vendor.nemo_text_processing.text_normalization.en.verbalizers.ordinal import OrdinalFst as vOrdinal
56
+ from zonos3.vendor.nemo_text_processing.text_normalization.en.verbalizers.roman import RomanFst as vRoman
57
+ from zonos3.vendor.nemo_text_processing.text_normalization.en.verbalizers.telephone import TelephoneFst as vTelephone
58
+ from zonos3.vendor.nemo_text_processing.text_normalization.en.verbalizers.time import TimeFst as vTime
59
+ from zonos3.vendor.nemo_text_processing.text_normalization.en.verbalizers.word import WordFst as vWord
60
+ from zonos3.vendor.nemo_text_processing.utils.logging import logger
61
+
62
+
63
+ class ClassifyFst(GraphFst):
64
+ """
65
+ Final class that composes all other classification grammars. This class can process an entire sentence including punctuation.
66
+ For deployment, this grammar will be compiled and exported to OpenFst Finite State Archive (FAR) File.
67
+ More details to deployment at NeMo/tools/text_processing_deployment.
68
+
69
+ Args:
70
+ input_case: accepting either "lower_cased" or "cased" input.
71
+ deterministic: if True will provide a single transduction option,
72
+ for False multiple options (used for audio-based normalization)
73
+ cache_dir: path to a dir with .far grammar file. Set to None to avoid using cache.
74
+ overwrite_cache: set to True to overwrite .far files
75
+ whitelist: path to a file with whitelist replacements
76
+ """
77
+
78
+ def __init__(
79
+ self,
80
+ input_case: str,
81
+ deterministic: bool = True,
82
+ cache_dir: str = None,
83
+ overwrite_cache: bool = True,
84
+ whitelist: str = None,
85
+ ):
86
+ super().__init__(name="tokenize_and_classify", kind="classify", deterministic=deterministic)
87
+
88
+ far_file = None
89
+ if cache_dir is not None and cache_dir != 'None':
90
+ os.makedirs(cache_dir, exist_ok=True)
91
+ whitelist_file = os.path.basename(whitelist) if whitelist else ""
92
+ far_file = os.path.join(
93
+ cache_dir, f"_{input_case}_en_tn_{deterministic}_deterministic{whitelist_file}_lm.far"
94
+ )
95
+ if not overwrite_cache and far_file and os.path.exists(far_file):
96
+ self.fst = pynini.Far(far_file, mode='r')['tokenize_and_classify']
97
+ no_digits = pynini.closure(pynini.difference(NEMO_CHAR, NEMO_DIGIT))
98
+ self.fst_no_digits = pynini.compose(self.fst, no_digits).optimize()
99
+ logger.info(f'ClassifyFst.fst was restored from {far_file}.')
100
+ else:
101
+ logger.info(f'Creating ClassifyFst grammars. This might take some time...')
102
+ # TAGGERS
103
+ cardinal = CardinalFst(deterministic=True, lm=True)
104
+ cardinal_tagger = cardinal
105
+ cardinal_graph = cardinal.fst
106
+
107
+ ordinal = OrdinalFst(cardinal=cardinal, deterministic=True)
108
+ ordinal_graph = ordinal.fst
109
+
110
+ decimal = DecimalFst(cardinal=cardinal, deterministic=True)
111
+ decimal_graph = decimal.fst
112
+ fraction = FractionFst(deterministic=True, cardinal=cardinal)
113
+ fraction_graph = fraction.fst
114
+
115
+ measure = MeasureFst(cardinal=cardinal, decimal=decimal, fraction=fraction, deterministic=True)
116
+ measure_graph = measure.fst
117
+ date = DateFst(cardinal=cardinal, deterministic=True, lm=True)
118
+ date_graph = date.fst
119
+ punctuation = PunctuationFst(deterministic=True)
120
+ punct_graph = punctuation.graph
121
+ word_graph = WordFst(punctuation=punctuation, deterministic=deterministic).graph
122
+ time_graph = TimeFst(cardinal=cardinal, deterministic=True).fst
123
+ telephone_graph = TelephoneFst(deterministic=True).fst
124
+ electronic_graph = ElectronicFst(cardinal=cardinal, deterministic=True).fst
125
+ money_graph = MoneyFst(cardinal=cardinal, decimal=decimal, deterministic=False).fst
126
+ whitelist = WhiteListFst(input_case=input_case, deterministic=False, input_file=whitelist)
127
+ whitelist_graph = whitelist.graph
128
+ serial_graph = SerialFst(cardinal=cardinal, ordinal=ordinal, deterministic=deterministic, lm=True).fst
129
+
130
+ # VERBALIZERS
131
+ cardinal = vCardinal(deterministic=True)
132
+ v_cardinal_graph = cardinal.fst
133
+ decimal = vDecimal(cardinal=cardinal, deterministic=True)
134
+ v_decimal_graph = decimal.fst
135
+ ordinal = vOrdinal(deterministic=True)
136
+ v_ordinal_graph = ordinal.fst
137
+ fraction = vFraction(deterministic=True, lm=True)
138
+ v_fraction_graph = fraction.fst
139
+ v_telephone_graph = vTelephone(deterministic=True).fst
140
+ v_electronic_graph = vElectronic(deterministic=True).fst
141
+ measure = vMeasure(decimal=decimal, cardinal=cardinal, fraction=fraction, deterministic=False)
142
+ v_measure_graph = measure.fst
143
+ v_time_graph = vTime(deterministic=True).fst
144
+ v_date_graph = vDate(ordinal=ordinal, deterministic=deterministic, lm=True).fst
145
+ v_money_graph = vMoney(decimal=decimal, deterministic=deterministic).fst
146
+ v_roman_graph = vRoman(deterministic=deterministic).fst
147
+ v_word_graph = vWord(deterministic=deterministic).fst
148
+
149
+ cardinal_or_date_final = plurals._priority_union(date_graph, cardinal_graph, NEMO_SIGMA)
150
+ cardinal_or_date_final = pynini.compose(cardinal_or_date_final, (v_cardinal_graph | v_date_graph))
151
+
152
+ time_final = pynini.compose(time_graph, v_time_graph)
153
+ ordinal_final = pynini.compose(ordinal_graph, v_ordinal_graph)
154
+ sem_w = 1
155
+ word_w = 100
156
+ punct_w = 2
157
+ classify_and_verbalize = (
158
+ pynutil.add_weight(time_final, sem_w)
159
+ | pynutil.add_weight(pynini.compose(decimal_graph, v_decimal_graph), sem_w)
160
+ | pynutil.add_weight(pynini.compose(measure_graph, v_measure_graph), sem_w)
161
+ | pynutil.add_weight(ordinal_final, sem_w)
162
+ | pynutil.add_weight(pynini.compose(telephone_graph, v_telephone_graph), sem_w)
163
+ | pynutil.add_weight(pynini.compose(electronic_graph, v_electronic_graph), sem_w)
164
+ | pynutil.add_weight(pynini.compose(fraction_graph, v_fraction_graph), sem_w)
165
+ | pynutil.add_weight(pynini.compose(money_graph, v_money_graph), sem_w)
166
+ | pynutil.add_weight(cardinal_or_date_final, sem_w)
167
+ | pynutil.add_weight(whitelist_graph, sem_w)
168
+ | pynutil.add_weight(
169
+ pynini.compose(serial_graph, v_word_graph), 1.1001
170
+ ) # should be higher than the rest of the classes
171
+ ).optimize()
172
+
173
+ roman_graph = RomanFst(deterministic=deterministic, lm=True).fst
174
+ # the weight matches the word_graph weight for "I" cases in long sentences with multiple semiotic tokens
175
+ classify_and_verbalize |= pynutil.add_weight(pynini.compose(roman_graph, v_roman_graph), sem_w)
176
+
177
+ date_final = pynini.compose(date_graph, v_date_graph)
178
+ range_graph = RangeFst(
179
+ time=time_final, cardinal=cardinal_tagger, date=date_final, deterministic=deterministic
180
+ ).fst
181
+ classify_and_verbalize |= pynutil.add_weight(pynini.compose(range_graph, v_word_graph), sem_w)
182
+ classify_and_verbalize = pynutil.insert("< ") + classify_and_verbalize + pynutil.insert(" >")
183
+ classify_and_verbalize |= pynutil.add_weight(word_graph, word_w)
184
+
185
+ punct_only = pynutil.add_weight(punct_graph, weight=punct_w)
186
+ punct = pynini.closure(
187
+ pynini.compose(pynini.closure(NEMO_WHITE_SPACE, 1), delete_extra_space)
188
+ | (pynutil.insert(" ") + punct_only),
189
+ 1,
190
+ )
191
+
192
+ def get_token_sem_graph(classify_and_verbalize):
193
+ token_plus_punct = (
194
+ pynini.closure(punct + pynutil.insert(" "))
195
+ + classify_and_verbalize
196
+ + pynini.closure(pynutil.insert(" ") + punct)
197
+ )
198
+
199
+ graph = token_plus_punct + pynini.closure(
200
+ (
201
+ pynini.compose(pynini.closure(NEMO_WHITE_SPACE, 1), delete_extra_space)
202
+ | (pynutil.insert(" ") + punct + pynutil.insert(" "))
203
+ )
204
+ + token_plus_punct
205
+ )
206
+
207
+ graph |= punct_only + pynini.closure(punct)
208
+ graph = delete_space + graph + delete_space
209
+
210
+ remove_extra_spaces = pynini.closure(NEMO_NOT_SPACE, 1) + pynini.closure(
211
+ delete_extra_space + pynini.closure(NEMO_NOT_SPACE, 1)
212
+ )
213
+ remove_extra_spaces |= (
214
+ pynini.closure(pynutil.delete(" "), 1)
215
+ + pynini.closure(NEMO_NOT_SPACE, 1)
216
+ + pynini.closure(delete_extra_space + pynini.closure(NEMO_NOT_SPACE, 1))
217
+ )
218
+
219
+ graph = pynini.compose(graph.optimize(), remove_extra_spaces).optimize()
220
+ return graph
221
+
222
+ self.fst = get_token_sem_graph(classify_and_verbalize)
223
+ no_digits = pynini.closure(pynini.difference(NEMO_CHAR, NEMO_DIGIT))
224
+ self.fst_no_digits = pynini.compose(self.fst, no_digits).optimize()
225
+
226
+ if far_file:
227
+ generator_main(far_file, {"tokenize_and_classify": self.fst})
vendor/zonos3/vendor/nemo_text_processing/text_normalization/en/taggers/tokenize_and_classify_with_audio.py ADDED
@@ -0,0 +1,229 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # Copyright (c) 2021, NVIDIA CORPORATION. All rights reserved.
2
+ #
3
+ # Licensed under the Apache License, Version 2.0 (the "License");
4
+ # you may not use this file except in compliance with the License.
5
+ # You may obtain a copy of the License at
6
+ #
7
+ # http://www.apache.org/licenses/LICENSE-2.0
8
+ #
9
+ # Unless required by applicable law or agreed to in writing, software
10
+ # distributed under the License is distributed on an "AS IS" BASIS,
11
+ # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
12
+ # See the License for the specific language governing permissions and
13
+ # limitations under the License.
14
+
15
+ import os
16
+
17
+ import pynini
18
+ from pynini.lib import pynutil
19
+
20
+ from zonos3.vendor.nemo_text_processing.text_normalization.en.graph_utils import (
21
+ NEMO_CHAR,
22
+ NEMO_DIGIT,
23
+ NEMO_NOT_SPACE,
24
+ NEMO_WHITE_SPACE,
25
+ GraphFst,
26
+ delete_extra_space,
27
+ delete_space,
28
+ generator_main,
29
+ )
30
+ from zonos3.vendor.nemo_text_processing.text_normalization.en.taggers.abbreviation import AbbreviationFst
31
+ from zonos3.vendor.nemo_text_processing.text_normalization.en.taggers.cardinal import CardinalFst
32
+ from zonos3.vendor.nemo_text_processing.text_normalization.en.taggers.date import DateFst
33
+ from zonos3.vendor.nemo_text_processing.text_normalization.en.taggers.decimal import DecimalFst
34
+ from zonos3.vendor.nemo_text_processing.text_normalization.en.taggers.electronic import ElectronicFst
35
+ from zonos3.vendor.nemo_text_processing.text_normalization.en.taggers.fraction import FractionFst
36
+ from zonos3.vendor.nemo_text_processing.text_normalization.en.taggers.measure import MeasureFst
37
+ from zonos3.vendor.nemo_text_processing.text_normalization.en.taggers.money import MoneyFst
38
+ from zonos3.vendor.nemo_text_processing.text_normalization.en.taggers.ordinal import OrdinalFst
39
+ from zonos3.vendor.nemo_text_processing.text_normalization.en.taggers.punctuation import PunctuationFst
40
+ from zonos3.vendor.nemo_text_processing.text_normalization.en.taggers.range import RangeFst as RangeFst
41
+ from zonos3.vendor.nemo_text_processing.text_normalization.en.taggers.roman import RomanFst
42
+ from zonos3.vendor.nemo_text_processing.text_normalization.en.taggers.serial import SerialFst
43
+ from zonos3.vendor.nemo_text_processing.text_normalization.en.taggers.telephone import TelephoneFst
44
+ from zonos3.vendor.nemo_text_processing.text_normalization.en.taggers.time import TimeFst
45
+ from zonos3.vendor.nemo_text_processing.text_normalization.en.taggers.whitelist import WhiteListFst
46
+ from zonos3.vendor.nemo_text_processing.text_normalization.en.taggers.word import WordFst
47
+ from zonos3.vendor.nemo_text_processing.text_normalization.en.verbalizers.abbreviation import AbbreviationFst as vAbbreviation
48
+ from zonos3.vendor.nemo_text_processing.text_normalization.en.verbalizers.cardinal import CardinalFst as vCardinal
49
+ from zonos3.vendor.nemo_text_processing.text_normalization.en.verbalizers.date import DateFst as vDate
50
+ from zonos3.vendor.nemo_text_processing.text_normalization.en.verbalizers.decimal import DecimalFst as vDecimal
51
+ from zonos3.vendor.nemo_text_processing.text_normalization.en.verbalizers.electronic import ElectronicFst as vElectronic
52
+ from zonos3.vendor.nemo_text_processing.text_normalization.en.verbalizers.fraction import FractionFst as vFraction
53
+ from zonos3.vendor.nemo_text_processing.text_normalization.en.verbalizers.measure import MeasureFst as vMeasure
54
+ from zonos3.vendor.nemo_text_processing.text_normalization.en.verbalizers.money import MoneyFst as vMoney
55
+ from zonos3.vendor.nemo_text_processing.text_normalization.en.verbalizers.ordinal import OrdinalFst as vOrdinal
56
+ from zonos3.vendor.nemo_text_processing.text_normalization.en.verbalizers.roman import RomanFst as vRoman
57
+ from zonos3.vendor.nemo_text_processing.text_normalization.en.verbalizers.telephone import TelephoneFst as vTelephone
58
+ from zonos3.vendor.nemo_text_processing.text_normalization.en.verbalizers.time import TimeFst as vTime
59
+ from zonos3.vendor.nemo_text_processing.text_normalization.en.verbalizers.word import WordFst as vWord
60
+ from zonos3.vendor.nemo_text_processing.utils.logging import logger
61
+
62
+
63
+ class ClassifyFst(GraphFst):
64
+ """
65
+ Final class that composes all other classification grammars. This class can process an entire sentence including punctuation.
66
+ For deployment, this grammar will be compiled and exported to OpenFst Finite State Archive (FAR) File.
67
+ More details to deployment at NeMo/tools/text_processing_deployment.
68
+
69
+ Args:
70
+ input_case: accepting either "lower_cased" or "cased" input.
71
+ deterministic: if True will provide a single transduction option,
72
+ for False multiple options (used for audio-based normalization)
73
+ cache_dir: path to a dir with .far grammar file. Set to None to avoid using cache.
74
+ overwrite_cache: set to True to overwrite .far files
75
+ whitelist: path to a file with whitelist replacements
76
+ """
77
+
78
+ def __init__(
79
+ self,
80
+ input_case: str,
81
+ deterministic: bool = True,
82
+ cache_dir: str = None,
83
+ overwrite_cache: bool = True,
84
+ whitelist: str = None,
85
+ ):
86
+ super().__init__(name="tokenize_and_classify", kind="classify", deterministic=deterministic)
87
+
88
+ far_file = None
89
+ if cache_dir is not None and cache_dir != 'None':
90
+ os.makedirs(cache_dir, exist_ok=True)
91
+ whitelist_file = os.path.basename(whitelist) if whitelist else ""
92
+ far_file = os.path.join(
93
+ cache_dir, f"_{input_case}_en_tn_{deterministic}_deterministic{whitelist_file}.far"
94
+ )
95
+ if not overwrite_cache and far_file and os.path.exists(far_file):
96
+ self.fst = pynini.Far(far_file, mode='r')['tokenize_and_classify']
97
+ logger.info(f'ClassifyFst.fst was restored from {far_file}.')
98
+ else:
99
+ logger.info(f'Creating ClassifyFst grammars. This might take some time...')
100
+ # TAGGERS
101
+ cardinal = CardinalFst(deterministic=deterministic)
102
+ cardinal_graph = cardinal.fst
103
+
104
+ ordinal = OrdinalFst(cardinal=cardinal, deterministic=deterministic)
105
+ deterministic_ordinal = OrdinalFst(cardinal=cardinal, deterministic=True)
106
+ ordinal_graph = ordinal.fst
107
+
108
+ decimal = DecimalFst(cardinal=cardinal, deterministic=deterministic)
109
+ decimal_graph = decimal.fst
110
+ fraction = FractionFst(deterministic=deterministic, cardinal=cardinal)
111
+ fraction_graph = fraction.fst
112
+
113
+ measure = MeasureFst(cardinal=cardinal, decimal=decimal, fraction=fraction, deterministic=deterministic)
114
+ measure_graph = measure.fst
115
+ date_graph = DateFst(cardinal=cardinal, deterministic=deterministic).fst
116
+ punctuation = PunctuationFst(deterministic=True)
117
+ punct_graph = punctuation.graph
118
+ word_graph = WordFst(punctuation=punctuation, deterministic=deterministic).graph
119
+ time_graph = TimeFst(cardinal=cardinal, deterministic=deterministic).fst
120
+ telephone_graph = TelephoneFst(deterministic=deterministic).fst
121
+ electronic_graph = ElectronicFst(cardinal=cardinal, deterministic=deterministic).fst
122
+ money_graph = MoneyFst(cardinal=cardinal, decimal=decimal, deterministic=deterministic).fst
123
+ whitelist = WhiteListFst(input_case=input_case, deterministic=deterministic, input_file=whitelist)
124
+ whitelist_graph = whitelist.graph
125
+ serial_graph = SerialFst(cardinal=cardinal, ordinal=deterministic_ordinal, deterministic=deterministic).fst
126
+
127
+ # VERBALIZERS
128
+ cardinal = vCardinal(deterministic=deterministic)
129
+ v_cardinal_graph = cardinal.fst
130
+ decimal = vDecimal(cardinal=cardinal, deterministic=deterministic)
131
+ v_decimal_graph = decimal.fst
132
+ ordinal = vOrdinal(deterministic=deterministic)
133
+ v_ordinal_graph = ordinal.fst
134
+ fraction = vFraction(deterministic=deterministic)
135
+ v_fraction_graph = fraction.fst
136
+ v_telephone_graph = vTelephone(deterministic=deterministic).fst
137
+ v_electronic_graph = vElectronic(deterministic=deterministic).fst
138
+ measure = vMeasure(decimal=decimal, cardinal=cardinal, fraction=fraction, deterministic=deterministic)
139
+ v_measure_graph = measure.fst
140
+ v_time_graph = vTime(deterministic=deterministic).fst
141
+ v_date_graph = vDate(ordinal=ordinal, deterministic=deterministic).fst
142
+ v_money_graph = vMoney(decimal=decimal, deterministic=deterministic).fst
143
+ v_roman_graph = vRoman(deterministic=deterministic).fst
144
+ v_abbreviation = vAbbreviation(deterministic=deterministic).fst
145
+
146
+ det_v_time_graph = vTime(deterministic=True).fst
147
+ det_v_date_graph = vDate(ordinal=vOrdinal(deterministic=True), deterministic=True).fst
148
+ time_final = pynini.compose(time_graph, det_v_time_graph)
149
+ date_final = pynini.compose(date_graph, det_v_date_graph)
150
+ range_graph = RangeFst(
151
+ time=time_final, date=date_final, cardinal=CardinalFst(deterministic=True), deterministic=deterministic
152
+ ).fst
153
+ v_word_graph = vWord(deterministic=deterministic).fst
154
+
155
+ sem_w = 1
156
+ word_w = 100
157
+ punct_w = 2
158
+ classify_and_verbalize = (
159
+ pynutil.add_weight(whitelist_graph, sem_w)
160
+ | pynutil.add_weight(pynini.compose(time_graph, v_time_graph), sem_w)
161
+ | pynutil.add_weight(pynini.compose(decimal_graph, v_decimal_graph), sem_w)
162
+ | pynutil.add_weight(pynini.compose(measure_graph, v_measure_graph), sem_w)
163
+ | pynutil.add_weight(pynini.compose(cardinal_graph, v_cardinal_graph), sem_w)
164
+ | pynutil.add_weight(pynini.compose(ordinal_graph, v_ordinal_graph), sem_w)
165
+ | pynutil.add_weight(pynini.compose(telephone_graph, v_telephone_graph), sem_w)
166
+ | pynutil.add_weight(pynini.compose(electronic_graph, v_electronic_graph), sem_w)
167
+ | pynutil.add_weight(pynini.compose(fraction_graph, v_fraction_graph), sem_w)
168
+ | pynutil.add_weight(pynini.compose(money_graph, v_money_graph), sem_w)
169
+ | pynutil.add_weight(word_graph, word_w)
170
+ | pynutil.add_weight(pynini.compose(date_graph, v_date_graph), sem_w - 0.01)
171
+ | pynutil.add_weight(pynini.compose(range_graph, v_word_graph), sem_w)
172
+ | pynutil.add_weight(
173
+ pynini.compose(serial_graph, v_word_graph), 1.1001
174
+ ) # should be higher than the rest of the classes
175
+ ).optimize()
176
+
177
+ if not deterministic:
178
+ roman_graph = RomanFst(deterministic=deterministic).fst
179
+ # the weight matches the word_graph weight for "I" cases in long sentences with multiple semiotic tokens
180
+ classify_and_verbalize |= pynutil.add_weight(pynini.compose(roman_graph, v_roman_graph), word_w)
181
+
182
+ abbreviation_graph = AbbreviationFst(whitelist=whitelist, deterministic=deterministic).fst
183
+ classify_and_verbalize |= pynutil.add_weight(
184
+ pynini.compose(abbreviation_graph, v_abbreviation), word_w
185
+ )
186
+
187
+ punct_only = pynutil.add_weight(punct_graph, weight=punct_w)
188
+ punct = pynini.closure(
189
+ pynini.compose(pynini.closure(NEMO_WHITE_SPACE, 1), delete_extra_space)
190
+ | (pynutil.insert(" ") + punct_only),
191
+ 1,
192
+ )
193
+
194
+ token_plus_punct = (
195
+ pynini.closure(punct + pynutil.insert(" "))
196
+ + classify_and_verbalize
197
+ + pynini.closure(pynutil.insert(" ") + punct)
198
+ )
199
+
200
+ graph = token_plus_punct + pynini.closure(
201
+ (
202
+ pynini.compose(pynini.closure(NEMO_WHITE_SPACE, 1), delete_extra_space)
203
+ | (pynutil.insert(" ") + punct + pynutil.insert(" "))
204
+ )
205
+ + token_plus_punct
206
+ )
207
+
208
+ graph |= punct_only + pynini.closure(punct)
209
+ graph = delete_space + graph + delete_space
210
+
211
+ remove_extra_spaces = pynini.closure(NEMO_NOT_SPACE, 1) + pynini.closure(
212
+ delete_extra_space + pynini.closure(NEMO_NOT_SPACE, 1)
213
+ )
214
+ remove_extra_spaces |= (
215
+ pynini.closure(pynutil.delete(" "), 1)
216
+ + pynini.closure(NEMO_NOT_SPACE, 1)
217
+ + pynini.closure(delete_extra_space + pynini.closure(NEMO_NOT_SPACE, 1))
218
+ )
219
+
220
+ graph = pynini.compose(graph.optimize(), remove_extra_spaces).optimize()
221
+ self.fst = graph
222
+
223
+ if far_file:
224
+ generator_main(far_file, {"tokenize_and_classify": self.fst})
225
+
226
+ # to remove normalization options that still contain digits and some special symbols
227
+ # e.g., "P&E" -> {P and E, P&E}, "P & E" will be removed from the list of normalization options
228
+ no_digits = pynini.closure(pynini.difference(NEMO_CHAR, pynini.union(NEMO_DIGIT, "&")))
229
+ self.fst_no_digits = pynini.compose(self.fst, no_digits).optimize()
vendor/zonos3/vendor/nemo_text_processing/text_normalization/en/taggers/whitelist.py ADDED
@@ -0,0 +1,153 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # Copyright (c) 2021, NVIDIA CORPORATION. All rights reserved.
2
+ #
3
+ # Licensed under the Apache License, Version 2.0 (the "License");
4
+ # you may not use this file except in compliance with the License.
5
+ # You may obtain a copy of the License at
6
+ #
7
+ # http://www.apache.org/licenses/LICENSE-2.0
8
+ #
9
+ # Unless required by applicable law or agreed to in writing, software
10
+ # distributed under the License is distributed on an "AS IS" BASIS,
11
+ # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
12
+ # See the License for the specific language governing permissions and
13
+ # limitations under the License.
14
+
15
+ import pynini
16
+ from pynini.lib import pynutil
17
+
18
+ from zonos3.vendor.nemo_text_processing.text_normalization.en.graph_utils import (
19
+ INPUT_CASED,
20
+ INPUT_LOWER_CASED,
21
+ NEMO_CHAR,
22
+ NEMO_NOT_SPACE,
23
+ NEMO_SIGMA,
24
+ NEMO_UPPER,
25
+ SINGULAR_TO_PLURAL,
26
+ GraphFst,
27
+ convert_space,
28
+ )
29
+ from zonos3.vendor.nemo_text_processing.text_normalization.en.taggers.roman import get_names
30
+ from zonos3.vendor.nemo_text_processing.text_normalization.en.utils import (
31
+ augment_labels_with_punct_at_end,
32
+ get_abs_path,
33
+ load_labels,
34
+ )
35
+
36
+
37
+ class WhiteListFst(GraphFst):
38
+ """
39
+ Finite state transducer for classifying whitelist, e.g.
40
+ misses -> tokens { name: "mrs" }
41
+ for non-deterministic case: "Dr. Abc" ->
42
+ tokens { name: "drive" } tokens { name: "Abc" }
43
+ tokens { name: "doctor" } tokens { name: "Abc" }
44
+ tokens { name: "Dr." } tokens { name: "Abc" }
45
+ This class has highest priority among all classifier grammars. Whitelisted tokens are defined and loaded from "data/whitelist.tsv".
46
+
47
+ Args:
48
+ input_case: accepting either "lower_cased" or "cased" input.
49
+ deterministic: if True will provide a single transduction option,
50
+ for False multiple options (used for audio-based normalization)
51
+ input_file: path to a file with whitelist replacements
52
+ """
53
+
54
+ def __init__(self, input_case: str, deterministic: bool = True, input_file: str = None):
55
+ super().__init__(name="whitelist", kind="classify", deterministic=deterministic)
56
+
57
+ def _get_whitelist_graph(input_case, file, keep_punct_add_end: bool = False):
58
+ whitelist = load_labels(file)
59
+ if input_case == INPUT_LOWER_CASED:
60
+ whitelist = [[x.lower(), y] for x, y in whitelist]
61
+ else:
62
+ whitelist = [[x, y] for x, y in whitelist]
63
+
64
+ if keep_punct_add_end:
65
+ whitelist.extend(augment_labels_with_punct_at_end(whitelist))
66
+
67
+ graph = pynini.string_map(whitelist)
68
+ return graph
69
+
70
+ graph = _get_whitelist_graph(input_case, get_abs_path("data/whitelist/tts.tsv"))
71
+ graph |= pynini.compose(
72
+ pynini.difference(NEMO_SIGMA, pynini.accep("/")).optimize(),
73
+ _get_whitelist_graph(input_case, get_abs_path("data/whitelist/symbol.tsv")),
74
+ ).optimize()
75
+
76
+ if deterministic:
77
+ names = get_names()
78
+ graph |= (
79
+ pynini.cross(pynini.union("st", "St", "ST"), "Saint")
80
+ + pynini.closure(pynutil.delete("."))
81
+ + pynini.accep(" ")
82
+ + names
83
+ )
84
+ else:
85
+ graph |= _get_whitelist_graph(
86
+ input_case, get_abs_path("data/whitelist/alternatives.tsv"), keep_punct_add_end=True
87
+ )
88
+
89
+ for x in [".", ". "]:
90
+ graph |= (
91
+ NEMO_UPPER
92
+ + pynini.closure(pynutil.delete(x) + NEMO_UPPER, 2)
93
+ + pynini.closure(pynutil.delete("."), 0, 1)
94
+ )
95
+
96
+ if not deterministic:
97
+ multiple_forms_whitelist_graph = get_formats(get_abs_path("data/whitelist/alternatives_all_format.tsv"))
98
+ graph |= multiple_forms_whitelist_graph
99
+
100
+ graph_unit = pynini.string_file(get_abs_path("data/measure/unit.tsv")) | pynini.string_file(
101
+ get_abs_path("data/measure/unit_alternatives.tsv")
102
+ )
103
+ graph_unit_plural = graph_unit @ SINGULAR_TO_PLURAL
104
+ units_graph = pynini.compose(NEMO_CHAR ** (3, ...), convert_space(graph_unit | graph_unit_plural))
105
+ graph |= units_graph
106
+
107
+ # convert to states only if comma is present before the abbreviation to avoid converting all caps words,
108
+ # e.g. "IN", "OH", "OK"
109
+ # TODO or only exclude above?
110
+ states = load_labels(get_abs_path("data/address/state.tsv"))
111
+ additional_options = []
112
+ for x, y in states:
113
+ if input_case == INPUT_LOWER_CASED:
114
+ x = x.lower()
115
+ additional_options.append((x, f"{y[0]}.{y[1:]}"))
116
+ if not deterministic:
117
+ additional_options.append((x, f"{y[0]}.{y[1:]}."))
118
+
119
+ states.extend(additional_options)
120
+ state_graph = pynini.string_map(states)
121
+ graph |= pynini.closure(NEMO_NOT_SPACE, 1) + pynini.union(", ", ",") + pynini.invert(state_graph).optimize()
122
+
123
+ if input_file:
124
+ whitelist_provided = _get_whitelist_graph(input_case, input_file)
125
+ if not deterministic:
126
+ graph |= whitelist_provided
127
+ else:
128
+ graph = whitelist_provided
129
+
130
+ self.graph = (convert_space(graph)).optimize()
131
+
132
+ self.fst = (pynutil.insert("name: \"") + self.graph + pynutil.insert("\"")).optimize()
133
+
134
+
135
+ def get_formats(input_f, input_case=INPUT_CASED, is_default=True):
136
+ """
137
+ Adds various abbreviation format options to the list of acceptable input forms
138
+ """
139
+ multiple_formats = load_labels(input_f)
140
+ additional_options = []
141
+ for x, y in multiple_formats:
142
+ if input_case == INPUT_LOWER_CASED:
143
+ x = x.lower()
144
+ additional_options.append((f"{x}.", y)) # default "dr" -> doctor, this includes period "dr." -> doctor
145
+ additional_options.append((f"{x[0].upper() + x[1:]}", f"{y[0].upper() + y[1:]}")) # "Dr" -> Doctor
146
+ additional_options.append((f"{x[0].upper() + x[1:]}.", f"{y[0].upper() + y[1:]}")) # "Dr." -> Doctor
147
+ multiple_formats.extend(additional_options)
148
+
149
+ if not is_default:
150
+ multiple_formats = [(x, f"|raw_start|{x}|raw_end||norm_start|{y}|norm_end|") for (x, y) in multiple_formats]
151
+
152
+ multiple_formats = pynini.string_map(multiple_formats)
153
+ return multiple_formats
vendor/zonos3/vendor/nemo_text_processing/text_normalization/en/taggers/word.py ADDED
@@ -0,0 +1,91 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # Copyright (c) 2021, NVIDIA CORPORATION. All rights reserved.
2
+ #
3
+ # Licensed under the Apache License, Version 2.0 (the "License");
4
+ # you may not use this file except in compliance with the License.
5
+ # You may obtain a copy of the License at
6
+ #
7
+ # http://www.apache.org/licenses/LICENSE-2.0
8
+ #
9
+ # Unless required by applicable law or agreed to in writing, software
10
+ # distributed under the License is distributed on an "AS IS" BASIS,
11
+ # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
12
+ # See the License for the specific language governing permissions and
13
+ # limitations under the License.
14
+
15
+ import pynini
16
+ from pynini.examples import plurals
17
+ from pynini.lib import pynutil
18
+
19
+ from zonos3.vendor.nemo_text_processing.text_normalization.en.graph_utils import (
20
+ MIN_NEG_WEIGHT,
21
+ NEMO_ALPHA,
22
+ NEMO_DIGIT,
23
+ NEMO_NOT_SPACE,
24
+ NEMO_SIGMA,
25
+ GraphFst,
26
+ convert_space,
27
+ get_abs_path,
28
+ )
29
+ from zonos3.vendor.nemo_text_processing.text_normalization.en.taggers.punctuation import PunctuationFst
30
+
31
+
32
+ class WordFst(GraphFst):
33
+ """
34
+ Finite state transducer for classifying word. Considers sentence boundary exceptions.
35
+ e.g. sleep -> tokens { name: "sleep" }
36
+
37
+ Args:
38
+ punctuation: PunctuationFst
39
+ deterministic: if True will provide a single transduction option,
40
+ for False multiple transduction are generated (used for audio-based normalization)
41
+ """
42
+
43
+ def __init__(self, punctuation: GraphFst, deterministic: bool = True):
44
+ super().__init__(name="word", kind="classify", deterministic=deterministic)
45
+
46
+ punct = PunctuationFst().graph
47
+ default_graph = pynini.closure(pynini.difference(NEMO_NOT_SPACE, punct.project("input")), 1)
48
+ symbols_to_exclude = (pynini.union("$", "€", "₩", "£", "¥", "#", "%") | NEMO_DIGIT).optimize()
49
+ graph = pynini.closure(pynini.difference(NEMO_NOT_SPACE, symbols_to_exclude), 1)
50
+ graph = pynutil.add_weight(graph, MIN_NEG_WEIGHT) | default_graph
51
+
52
+ # leave phones of format [HH AH0 L OW1] untouched
53
+ phoneme_unit = pynini.closure(NEMO_ALPHA, 1) + pynini.closure(NEMO_DIGIT)
54
+ phoneme = (
55
+ pynini.accep(pynini.escape("["))
56
+ + pynini.closure(phoneme_unit + pynini.accep(" "))
57
+ + phoneme_unit
58
+ + pynini.accep(pynini.escape("]"))
59
+ )
60
+
61
+ # leave IPA phones of format [ˈdoʊv] untouched, single words and sentences with punctuation marks allowed
62
+ punct_marks = pynini.union(*punctuation.punct_marks).optimize()
63
+ stress = pynini.union("ˈ", "'", "ˌ")
64
+ ipa_phoneme_unit = pynini.string_file(get_abs_path("data/whitelist/ipa_symbols.tsv"))
65
+ # word in ipa form
66
+ ipa_phonemes = (
67
+ pynini.closure(stress, 0, 1)
68
+ + pynini.closure(ipa_phoneme_unit, 1)
69
+ + pynini.closure(stress | ipa_phoneme_unit)
70
+ )
71
+ # allow sentences of words in IPA format separated with spaces or punct marks
72
+ delim = (punct_marks | pynini.accep(" ")) ** (1, ...)
73
+ ipa_phonemes = ipa_phonemes + pynini.closure(delim + ipa_phonemes) + pynini.closure(delim, 0, 1)
74
+ ipa_phonemes = (pynini.accep(pynini.escape("[")) + ipa_phonemes + pynini.accep(pynini.escape("]"))).optimize()
75
+
76
+ if not deterministic:
77
+ phoneme = (
78
+ pynini.accep(pynini.escape("["))
79
+ + pynini.closure(pynini.accep(" "), 0, 1)
80
+ + pynini.closure(phoneme_unit + pynini.accep(" "))
81
+ + phoneme_unit
82
+ + pynini.closure(pynini.accep(" "), 0, 1)
83
+ + pynini.accep(pynini.escape("]"))
84
+ ).optimize()
85
+ ipa_phonemes = (
86
+ pynini.accep(pynini.escape("[")) + ipa_phonemes + pynini.accep(pynini.escape("]"))
87
+ ).optimize()
88
+
89
+ phoneme |= ipa_phonemes
90
+ self.graph = plurals._priority_union(convert_space(phoneme.optimize()), graph, NEMO_SIGMA)
91
+ self.fst = (pynutil.insert("name: \"") + self.graph + pynutil.insert("\"")).optimize()
vendor/zonos3/vendor/nemo_text_processing/text_normalization/en/utils.py ADDED
@@ -0,0 +1,60 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # Copyright (c) 2021, NVIDIA CORPORATION. All rights reserved.
2
+ #
3
+ # Licensed under the Apache License, Version 2.0 (the "License");
4
+ # you may not use this file except in compliance with the License.
5
+ # You may obtain a copy of the License at
6
+ #
7
+ # http://www.apache.org/licenses/LICENSE-2.0
8
+ #
9
+ # Unless required by applicable law or agreed to in writing, software
10
+ # distributed under the License is distributed on an "AS IS" BASIS,
11
+ # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
12
+ # See the License for the specific language governing permissions and
13
+ # limitations under the License.
14
+
15
+ import csv
16
+ import os
17
+
18
+
19
+ def get_abs_path(rel_path):
20
+ """
21
+ Get absolute path
22
+
23
+ Args:
24
+ rel_path: relative path to this file
25
+
26
+ Returns absolute path
27
+ """
28
+ return os.path.dirname(os.path.abspath(__file__)) + '/' + rel_path
29
+
30
+
31
+ def load_labels(abs_path):
32
+ """
33
+ loads relative path file as dictionary
34
+
35
+ Args:
36
+ abs_path: absolute path
37
+
38
+ Returns dictionary of mappings
39
+ """
40
+ with open(abs_path, encoding="utf-8") as label_tsv:
41
+ labels = list(csv.reader(label_tsv, delimiter="\t"))
42
+ return labels
43
+
44
+
45
+ def augment_labels_with_punct_at_end(labels):
46
+ """
47
+ augments labels: if key ends on a punctuation that value does not have, add a new label
48
+ where the value maintains the punctuation
49
+
50
+ Args:
51
+ labels : input labels
52
+ Returns:
53
+ additional labels
54
+ """
55
+ res = []
56
+ for label in labels:
57
+ if len(label) > 1:
58
+ if label[0][-1] == "." and label[1][-1] != ".":
59
+ res.append([label[0], label[1] + "."] + label[2:])
60
+ return res
vendor/zonos3/vendor/nemo_text_processing/text_normalization/en/verbalizers/__init__.py ADDED
@@ -0,0 +1,13 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # Copyright (c) 2021, NVIDIA CORPORATION. All rights reserved.
2
+ #
3
+ # Licensed under the Apache License, Version 2.0 (the "License");
4
+ # you may not use this file except in compliance with the License.
5
+ # You may obtain a copy of the License at
6
+ #
7
+ # http://www.apache.org/licenses/LICENSE-2.0
8
+ #
9
+ # Unless required by applicable law or agreed to in writing, software
10
+ # distributed under the License is distributed on an "AS IS" BASIS,
11
+ # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
12
+ # See the License for the specific language governing permissions and
13
+ # limitations under the License.
vendor/zonos3/vendor/nemo_text_processing/text_normalization/en/verbalizers/abbreviation.py ADDED
@@ -0,0 +1,36 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # Copyright (c) 2021, NVIDIA CORPORATION. All rights reserved.
2
+ #
3
+ # Licensed under the Apache License, Version 2.0 (the "License");
4
+ # you may not use this file except in compliance with the License.
5
+ # You may obtain a copy of the License at
6
+ #
7
+ # http://www.apache.org/licenses/LICENSE-2.0
8
+ #
9
+ # Unless required by applicable law or agreed to in writing, software
10
+ # distributed under the License is distributed on an "AS IS" BASIS,
11
+ # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
12
+ # See the License for the specific language governing permissions and
13
+ # limitations under the License.
14
+
15
+ import pynini
16
+ from pynini.lib import pynutil
17
+
18
+ from zonos3.vendor.nemo_text_processing.text_normalization.en.graph_utils import NEMO_NOT_QUOTE, GraphFst
19
+
20
+
21
+ class AbbreviationFst(GraphFst):
22
+ """
23
+ Finite state transducer for verbalizing abbreviations
24
+ e.g. tokens { abbreviation { value: "A B C" } } -> "ABC"
25
+
26
+ Args:
27
+ deterministic: if True will provide a single transduction option,
28
+ for False multiple transduction are generated (used for audio-based normalization)
29
+ """
30
+
31
+ def __init__(self, deterministic: bool = True):
32
+ super().__init__(name="abbreviation", kind="verbalize", deterministic=deterministic)
33
+
34
+ graph = pynutil.delete("value: \"") + pynini.closure(NEMO_NOT_QUOTE, 1) + pynutil.delete("\"")
35
+ delete_tokens = self.delete_tokens(graph)
36
+ self.fst = delete_tokens.optimize()
vendor/zonos3/vendor/nemo_text_processing/text_normalization/en/verbalizers/cardinal.py ADDED
@@ -0,0 +1,48 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # Copyright (c) 2021, NVIDIA CORPORATION. All rights reserved.
2
+ #
3
+ # Licensed under the Apache License, Version 2.0 (the "License");
4
+ # you may not use this file except in compliance with the License.
5
+ # You may obtain a copy of the License at
6
+ #
7
+ # http://www.apache.org/licenses/LICENSE-2.0
8
+ #
9
+ # Unless required by applicable law or agreed to in writing, software
10
+ # distributed under the License is distributed on an "AS IS" BASIS,
11
+ # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
12
+ # See the License for the specific language governing permissions and
13
+ # limitations under the License.
14
+
15
+ import pynini
16
+ from pynini.lib import pynutil
17
+
18
+ from zonos3.vendor.nemo_text_processing.text_normalization.en.graph_utils import NEMO_NOT_QUOTE, GraphFst, delete_space
19
+
20
+
21
+ class CardinalFst(GraphFst):
22
+ """
23
+ Finite state transducer for verbalizing cardinal, e.g.
24
+ cardinal { negative: "true" integer: "23" } -> minus twenty three
25
+
26
+ Args:
27
+ deterministic: if True will provide a single transduction option,
28
+ for False multiple options (used for audio-based normalization)
29
+ """
30
+
31
+ def __init__(self, deterministic: bool = True):
32
+ super().__init__(name="cardinal", kind="verbalize", deterministic=deterministic)
33
+
34
+ self.optional_sign = pynini.cross("negative: \"true\"", "minus ")
35
+ if not deterministic:
36
+ self.optional_sign |= pynini.cross("negative: \"true\"", "negative ")
37
+ self.optional_sign |= pynini.cross("negative: \"true\"", "dash ")
38
+
39
+ self.optional_sign = pynini.closure(self.optional_sign + delete_space, 0, 1)
40
+
41
+ integer = pynini.closure(NEMO_NOT_QUOTE)
42
+
43
+ self.integer = delete_space + pynutil.delete("\"") + integer + pynutil.delete("\"")
44
+ integer = pynutil.delete("integer:") + self.integer
45
+
46
+ self.numbers = self.optional_sign + integer
47
+ delete_tokens = self.delete_tokens(self.numbers)
48
+ self.fst = delete_tokens.optimize()
vendor/zonos3/vendor/nemo_text_processing/text_normalization/en/verbalizers/date.py ADDED
@@ -0,0 +1,107 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # Copyright (c) 2021, NVIDIA CORPORATION. All rights reserved.
2
+ #
3
+ # Licensed under the Apache License, Version 2.0 (the "License");
4
+ # you may not use this file except in compliance with the License.
5
+ # You may obtain a copy of the License at
6
+ #
7
+ # http://www.apache.org/licenses/LICENSE-2.0
8
+ #
9
+ # Unless required by applicable law or agreed to in writing, software
10
+ # distributed under the License is distributed on an "AS IS" BASIS,
11
+ # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
12
+ # See the License for the specific language governing permissions and
13
+ # limitations under the License.
14
+
15
+ import pynini
16
+ from pynini.examples import plurals
17
+ from pynini.lib import pynutil
18
+
19
+ from zonos3.vendor.nemo_text_processing.text_normalization.en.graph_utils import (
20
+ NEMO_NOT_QUOTE,
21
+ NEMO_SIGMA,
22
+ GraphFst,
23
+ delete_extra_space,
24
+ delete_space,
25
+ )
26
+
27
+
28
+ class DateFst(GraphFst):
29
+ """
30
+ Finite state transducer for verbalizing date, e.g.
31
+ date { month: "february" day: "five" year: "twenty twelve" preserve_order: true } -> february fifth twenty twelve
32
+ date { day: "five" month: "february" year: "twenty twelve" preserve_order: true } -> the fifth of february twenty twelve
33
+
34
+ Args:
35
+ ordinal: OrdinalFst
36
+ deterministic: if True will provide a single transduction option,
37
+ for False multiple transduction are generated (used for audio-based normalization)
38
+ """
39
+
40
+ def __init__(self, ordinal: GraphFst, deterministic: bool = True, lm: bool = False):
41
+ super().__init__(name="date", kind="verbalize", deterministic=deterministic)
42
+
43
+ phrase = pynini.closure(NEMO_NOT_QUOTE, 1)
44
+ day_cardinal = (
45
+ pynutil.delete("day:")
46
+ + delete_space
47
+ + pynutil.delete("\"")
48
+ + pynini.closure(NEMO_NOT_QUOTE, 1)
49
+ + pynutil.delete("\"")
50
+ )
51
+ day = day_cardinal @ ordinal.suffix
52
+ period = pynutil.delete("text:") + delete_space + pynutil.delete("\"") + phrase + pynutil.delete("\"")
53
+ month = pynutil.delete("month:") + delete_space + pynutil.delete("\"") + phrase + pynutil.delete("\"")
54
+
55
+ year = (
56
+ pynutil.delete("year:")
57
+ + delete_space
58
+ + pynutil.delete("\"")
59
+ + pynini.closure(NEMO_NOT_QUOTE, 1)
60
+ + delete_space
61
+ + pynutil.delete("\"")
62
+ )
63
+
64
+ # financial period
65
+ graph_fy = (
66
+ pynutil.insert("the ") + period + pynutil.insert(" of ") + pynini.closure(delete_extra_space + year, 0, 1)
67
+ )
68
+
69
+ # month (day) year
70
+ graph_mdy = (
71
+ month + pynini.closure(delete_extra_space + day, 0, 1) + pynini.closure(delete_extra_space + year, 0, 1)
72
+ )
73
+ # may 5 -> may five
74
+ if not deterministic and not lm:
75
+ graph_mdy |= (
76
+ month
77
+ + pynini.closure(delete_extra_space + day_cardinal, 0, 1)
78
+ + pynini.closure(delete_extra_space + year, 0, 1)
79
+ )
80
+
81
+ # day month year
82
+ graph_dmy = (
83
+ pynutil.insert("the ")
84
+ + day
85
+ + delete_extra_space
86
+ + pynutil.insert("of ")
87
+ + month
88
+ + pynini.closure(delete_extra_space + year, 0, 1)
89
+ )
90
+
91
+ optional_preserve_order = pynini.closure(
92
+ pynutil.delete("preserve_order:") + delete_space + pynutil.delete("true") + delete_space
93
+ | pynutil.delete("field_order:")
94
+ + delete_space
95
+ + pynutil.delete("\"")
96
+ + NEMO_NOT_QUOTE
97
+ + pynutil.delete("\"")
98
+ + delete_space
99
+ )
100
+
101
+ final_graph = (
102
+ (plurals._priority_union(graph_mdy, pynutil.add_weight(graph_dmy, 0.0001), NEMO_SIGMA) | year | graph_fy)
103
+ + delete_space
104
+ + optional_preserve_order
105
+ )
106
+ delete_tokens = self.delete_tokens(final_graph)
107
+ self.fst = delete_tokens.optimize()
vendor/zonos3/vendor/nemo_text_processing/text_normalization/en/verbalizers/decimal.py ADDED
@@ -0,0 +1,85 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # Copyright (c) 2021, NVIDIA CORPORATION. All rights reserved.
2
+ #
3
+ # Licensed under the Apache License, Version 2.0 (the "License");
4
+ # you may not use this file except in compliance with the License.
5
+ # You may obtain a copy of the License at
6
+ #
7
+ # http://www.apache.org/licenses/LICENSE-2.0
8
+ #
9
+ # Unless required by applicable law or agreed to in writing, software
10
+ # distributed under the License is distributed on an "AS IS" BASIS,
11
+ # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
12
+ # See the License for the specific language governing permissions and
13
+ # limitations under the License.
14
+
15
+ import pynini
16
+ from pynini.lib import pynutil
17
+
18
+ from zonos3.vendor.nemo_text_processing.text_normalization.en.graph_utils import (
19
+ NEMO_NOT_QUOTE,
20
+ NEMO_SIGMA,
21
+ GraphFst,
22
+ delete_space,
23
+ insert_space,
24
+ )
25
+
26
+
27
+ class DecimalFst(GraphFst):
28
+ """
29
+ Finite state transducer for verbalizing decimal, e.g.
30
+ decimal { negative: "true" integer_part: "twelve" fractional_part: "five o o six" quantity: "billion" } -> minus twelve point five o o six billion
31
+
32
+ Args:
33
+ deterministic: if True will provide a single transduction option,
34
+ for False multiple transduction are generated (used for audio-based normalization)
35
+ """
36
+
37
+ def __init__(self, cardinal, deterministic: bool = True):
38
+ super().__init__(name="decimal", kind="verbalize", deterministic=deterministic)
39
+ self.optional_sign = pynini.cross("negative: \"true\"", "minus ")
40
+ if not deterministic:
41
+ self.optional_sign |= pynini.cross("negative: \"true\"", "negative ")
42
+ self.optional_sign = pynini.closure(self.optional_sign + delete_space, 0, 1)
43
+ self.integer = pynutil.delete("integer_part:") + cardinal.integer
44
+ self.optional_integer = pynini.closure(self.integer + delete_space + insert_space, 0, 1)
45
+ self.fractional_default = (
46
+ pynutil.delete("fractional_part:")
47
+ + delete_space
48
+ + pynutil.delete("\"")
49
+ + pynini.closure(NEMO_NOT_QUOTE, 1)
50
+ + pynutil.delete("\"")
51
+ )
52
+
53
+ self.fractional = pynutil.insert("point ") + self.fractional_default
54
+
55
+ self.quantity = (
56
+ delete_space
57
+ + insert_space
58
+ + pynutil.delete("quantity:")
59
+ + delete_space
60
+ + pynutil.delete("\"")
61
+ + pynini.closure(NEMO_NOT_QUOTE, 1)
62
+ + pynutil.delete("\"")
63
+ )
64
+ self.optional_quantity = pynini.closure(self.quantity, 0, 1)
65
+
66
+ graph = self.optional_sign + (
67
+ self.integer
68
+ | (self.integer + self.quantity)
69
+ | (self.optional_integer + self.fractional + self.optional_quantity)
70
+ )
71
+
72
+ self.numbers = graph
73
+ delete_tokens = self.delete_tokens(graph)
74
+ if not deterministic:
75
+ delete_tokens |= pynini.compose(
76
+ delete_tokens,
77
+ NEMO_SIGMA
78
+ + (
79
+ pynini.cross(" point five", " and a half")
80
+ | pynini.cross("zero point five", "half")
81
+ | pynini.cross(" point two five", " and a quarter")
82
+ | pynini.cross("zero point two five", "quarter")
83
+ ),
84
+ ).optimize()
85
+ self.fst = delete_tokens.optimize()
vendor/zonos3/vendor/nemo_text_processing/text_normalization/en/verbalizers/electronic.py ADDED
@@ -0,0 +1,122 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # Copyright (c) 2021, NVIDIA CORPORATION. All rights reserved.
2
+ #
3
+ # Licensed under the Apache License, Version 2.0 (the "License");
4
+ # you may not use this file except in compliance with the License.
5
+ # You may obtain a copy of the License at
6
+ #
7
+ # http://www.apache.org/licenses/LICENSE-2.0
8
+ #
9
+ # Unless required by applicable law or agreed to in writing, software
10
+ # distributed under the License is distributed on an "AS IS" BASIS,
11
+ # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
12
+ # See the License for the specific language governing permissions and
13
+ # limitations under the License.
14
+
15
+ import pynini
16
+ from pynini.examples import plurals
17
+ from pynini.lib import pynutil
18
+
19
+ from zonos3.vendor.nemo_text_processing.text_normalization.en.graph_utils import (
20
+ MIN_NEG_WEIGHT,
21
+ NEMO_ALPHA,
22
+ NEMO_CHAR,
23
+ NEMO_LOWER,
24
+ NEMO_NOT_QUOTE,
25
+ NEMO_SIGMA,
26
+ NEMO_SPACE,
27
+ TO_LOWER,
28
+ GraphFst,
29
+ delete_extra_space,
30
+ delete_space,
31
+ insert_space,
32
+ )
33
+ from zonos3.vendor.nemo_text_processing.text_normalization.en.utils import get_abs_path
34
+
35
+
36
+ class ElectronicFst(GraphFst):
37
+ """
38
+ Finite state transducer for verbalizing electronic
39
+ e.g. tokens { electronic { username: "cdf1" domain: "abc.edu" } } -> c d f one at a b c dot e d u
40
+
41
+ Args:
42
+ deterministic: if True will provide a single transduction option,
43
+ for False multiple transduction are generated (used for audio-based normalization)
44
+ """
45
+
46
+ def __init__(self, deterministic: bool = True):
47
+ super().__init__(name="electronic", kind="verbalize", deterministic=deterministic)
48
+ graph_digit_no_zero = pynini.invert(pynini.string_file(get_abs_path("data/number/digit.tsv"))).optimize()
49
+ graph_zero = pynini.cross("0", "zero")
50
+ long_numbers = pynutil.add_weight(graph_digit_no_zero + pynini.cross("000", " thousand"), MIN_NEG_WEIGHT)
51
+
52
+ if not deterministic:
53
+ graph_zero |= pynini.cross("0", "o") | pynini.cross("0", "oh")
54
+
55
+ graph_digit = graph_digit_no_zero | graph_zero
56
+ graph_symbols = pynini.string_file(get_abs_path("data/electronic/symbol.tsv")).optimize()
57
+
58
+ NEMO_NOT_BRACKET = pynini.difference(NEMO_CHAR, pynini.union("{", "}")).optimize()
59
+ dict_words = pynini.project(pynini.string_file(get_abs_path("data/electronic/words.tsv")), "output")
60
+ default_chars_symbols = pynini.cdrewrite(
61
+ pynutil.insert(" ") + (graph_symbols | graph_digit | long_numbers) + pynutil.insert(" "),
62
+ "",
63
+ "",
64
+ NEMO_SIGMA,
65
+ )
66
+ default_chars_symbols = pynini.compose(
67
+ pynini.closure(NEMO_NOT_BRACKET), default_chars_symbols.optimize()
68
+ ).optimize()
69
+
70
+ # this is far cases when user name was split by dictionary words, i.e. "sevicepart@ab.com" -> "service part"
71
+ space_separated_dict_words = pynutil.add_weight(
72
+ NEMO_ALPHA
73
+ + pynini.closure(NEMO_ALPHA | NEMO_SPACE)
74
+ + NEMO_SPACE
75
+ + pynini.closure(NEMO_ALPHA | NEMO_SPACE),
76
+ MIN_NEG_WEIGHT,
77
+ )
78
+
79
+ user_name = (
80
+ pynutil.delete("username:")
81
+ + delete_space
82
+ + pynutil.delete("\"")
83
+ + (default_chars_symbols | space_separated_dict_words).optimize()
84
+ + pynutil.delete("\"")
85
+ )
86
+
87
+ domain_common = pynini.string_file(get_abs_path("data/electronic/domain.tsv"))
88
+
89
+ # this will be used for a safe fallback
90
+ domain_all = pynini.compose(
91
+ default_chars_symbols,
92
+ pynini.closure(TO_LOWER | NEMO_LOWER | NEMO_SPACE | pynutil.add_weight(dict_words, MIN_NEG_WEIGHT)),
93
+ )
94
+
95
+ domain = (
96
+ domain_all
97
+ + insert_space
98
+ + plurals._priority_union(
99
+ domain_common, pynutil.add_weight(pynini.cross(".", "dot"), weight=0.0001), NEMO_SIGMA
100
+ )
101
+ + pynini.closure(insert_space + default_chars_symbols, 0, 1)
102
+ )
103
+
104
+ domain = (
105
+ pynutil.delete("domain:")
106
+ + delete_space
107
+ + pynutil.delete("\"")
108
+ + (domain | pynutil.add_weight(domain_all, weight=100)).optimize()
109
+ + delete_space
110
+ + pynutil.delete("\"")
111
+ ).optimize()
112
+
113
+ protocol = pynutil.delete("protocol: \"") + pynini.closure(NEMO_NOT_QUOTE, 1) + pynutil.delete("\"")
114
+ graph = (
115
+ pynini.closure(protocol + delete_space, 0, 1)
116
+ + pynini.closure(user_name + delete_space + pynutil.insert(" at ") + delete_space, 0, 1)
117
+ + domain
118
+ + delete_space
119
+ ).optimize() @ pynini.cdrewrite(delete_extra_space, "", "", NEMO_SIGMA)
120
+
121
+ delete_tokens = self.delete_tokens(graph)
122
+ self.fst = delete_tokens.optimize()
vendor/zonos3/vendor/nemo_text_processing/text_normalization/en/verbalizers/fraction.py ADDED
@@ -0,0 +1,89 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # Copyright (c) 2021, NVIDIA CORPORATION. All rights reserved.
2
+ #
3
+ # Licensed under the Apache License, Version 2.0 (the "License");
4
+ # you may not use this file except in compliance with the License.
5
+ # You may obtain a copy of the License at
6
+ #
7
+ # http://www.apache.org/licenses/LICENSE-2.0
8
+ #
9
+ # Unless required by applicable law or agreed to in writing, software
10
+ # distributed under the License is distributed on an "AS IS" BASIS,
11
+ # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
12
+ # See the License for the specific language governing permissions and
13
+ # limitations under the License.
14
+
15
+ import pynini
16
+ from pynini.examples import plurals
17
+ from pynini.lib import pynutil
18
+
19
+ from zonos3.vendor.nemo_text_processing.text_normalization.en.graph_utils import NEMO_NOT_QUOTE, NEMO_SIGMA, GraphFst, insert_space
20
+ from zonos3.vendor.nemo_text_processing.text_normalization.en.verbalizers.ordinal import OrdinalFst
21
+
22
+
23
+ class FractionFst(GraphFst):
24
+ """
25
+ Finite state transducer for verbalizing fraction
26
+ e.g. tokens { fraction { integer: "twenty three" numerator: "four" denominator: "five" } } ->
27
+ twenty three and four fifth
28
+
29
+ Args:
30
+ deterministic: if True will provide a single transduction option,
31
+ for False multiple transduction are generated (used for audio-based normalization)
32
+ """
33
+
34
+ def __init__(self, deterministic: bool = True, lm: bool = False):
35
+ super().__init__(name="fraction", kind="verbalize", deterministic=deterministic)
36
+ suffix = OrdinalFst().suffix
37
+
38
+ integer = pynutil.delete("integer_part: \"") + pynini.closure(NEMO_NOT_QUOTE) + pynutil.delete("\" ")
39
+ denominator_one = pynini.cross("denominator: \"one\"", "over one")
40
+ denominator_half = pynini.cross("denominator: \"two\"", "half")
41
+ denominator_quarter = pynini.cross("denominator: \"four\"", "quarter")
42
+
43
+ denominator_rest = (
44
+ pynutil.delete("denominator: \"") + pynini.closure(NEMO_NOT_QUOTE) @ suffix + pynutil.delete("\"")
45
+ )
46
+
47
+ denominators = plurals._priority_union(
48
+ denominator_one,
49
+ plurals._priority_union(
50
+ denominator_half,
51
+ plurals._priority_union(denominator_quarter, denominator_rest, NEMO_SIGMA),
52
+ NEMO_SIGMA,
53
+ ),
54
+ NEMO_SIGMA,
55
+ ).optimize()
56
+ if not deterministic:
57
+ denominators |= pynutil.delete("denominator: \"") + (pynini.accep("four") @ suffix) + pynutil.delete("\"")
58
+
59
+ numerator_one = pynutil.delete("numerator: \"") + pynini.accep("one") + pynutil.delete("\" ")
60
+ numerator_one = numerator_one + insert_space + denominators
61
+ numerator_rest = (
62
+ pynutil.delete("numerator: \"")
63
+ + (pynini.closure(NEMO_NOT_QUOTE) - pynini.accep("one"))
64
+ + pynutil.delete("\" ")
65
+ )
66
+ numerator_rest = numerator_rest + insert_space + denominators
67
+ numerator_rest @= pynini.cdrewrite(
68
+ plurals._priority_union(pynini.cross("half", "halves"), pynutil.insert("s"), NEMO_SIGMA),
69
+ "",
70
+ "[EOS]",
71
+ NEMO_SIGMA,
72
+ )
73
+
74
+ graph = numerator_one | numerator_rest
75
+
76
+ conjunction = pynutil.insert("and ")
77
+ if not deterministic and not lm:
78
+ conjunction = pynini.closure(conjunction, 0, 1)
79
+
80
+ integer = pynini.closure(integer + insert_space + conjunction, 0, 1)
81
+
82
+ graph = integer + graph
83
+ graph @= pynini.cdrewrite(
84
+ pynini.cross("and one half", "and a half") | pynini.cross("over ones", "over one"), "", "[EOS]", NEMO_SIGMA
85
+ )
86
+
87
+ self.graph = graph
88
+ delete_tokens = self.delete_tokens(self.graph)
89
+ self.fst = delete_tokens.optimize()
vendor/zonos3/vendor/nemo_text_processing/text_normalization/en/verbalizers/measure.py ADDED
@@ -0,0 +1,122 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # Copyright (c) 2021, NVIDIA CORPORATION. All rights reserved.
2
+ #
3
+ # Licensed under the Apache License, Version 2.0 (the "License");
4
+ # you may not use this file except in compliance with the License.
5
+ # You may obtain a copy of the License at
6
+ #
7
+ # http://www.apache.org/licenses/LICENSE-2.0
8
+ #
9
+ # Unless required by applicable law or agreed to in writing, software
10
+ # distributed under the License is distributed on an "AS IS" BASIS,
11
+ # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
12
+ # See the License for the specific language governing permissions and
13
+ # limitations under the License.
14
+
15
+ import pynini
16
+ from pynini.lib import pynutil
17
+
18
+ from zonos3.vendor.nemo_text_processing.text_normalization.en.graph_utils import (
19
+ NEMO_NOT_QUOTE,
20
+ NEMO_SIGMA,
21
+ GraphFst,
22
+ delete_space,
23
+ insert_space,
24
+ )
25
+
26
+
27
+ class MeasureFst(GraphFst):
28
+ """
29
+ Finite state transducer for verbalizing measure, e.g.
30
+ measure { negative: "true" cardinal { integer: "twelve" } units: "kilograms" } -> minus twelve kilograms
31
+ measure { decimal { integer_part: "twelve" fractional_part: "five" } units: "kilograms" } -> twelve point five kilograms
32
+ tokens { measure { units: "covid" decimal { integer_part: "nineteen" fractional_part: "five" } } } -> covid nineteen point five
33
+
34
+ Args:
35
+ decimal: DecimalFst
36
+ cardinal: CardinalFst
37
+ fraction: FractionFst
38
+ deterministic: if True will provide a single transduction option,
39
+ for False multiple transduction are generated (used for audio-based normalization)
40
+ """
41
+
42
+ def __init__(self, decimal: GraphFst, cardinal: GraphFst, fraction: GraphFst, deterministic: bool = True):
43
+ super().__init__(name="measure", kind="verbalize", deterministic=deterministic)
44
+ optional_sign = cardinal.optional_sign
45
+ unit = (
46
+ pynutil.delete("units: \"")
47
+ + pynini.difference(pynini.closure(NEMO_NOT_QUOTE, 1), pynini.union("address", "math"))
48
+ + pynutil.delete("\"")
49
+ + delete_space
50
+ )
51
+
52
+ if not deterministic:
53
+ unit |= pynini.compose(unit, pynini.cross(pynini.union("inch", "inches"), "\""))
54
+
55
+ graph_decimal = (
56
+ pynutil.delete("decimal {")
57
+ + delete_space
58
+ + optional_sign
59
+ + delete_space
60
+ + decimal.numbers
61
+ + delete_space
62
+ + pynutil.delete("}")
63
+ )
64
+
65
+ if not deterministic:
66
+ graph_decimal |= pynini.compose(
67
+ graph_decimal,
68
+ NEMO_SIGMA
69
+ + (
70
+ pynini.cross(" point five", " and a half")
71
+ | pynini.cross("zero point five", "half")
72
+ | pynini.cross(" point two five", " and a quarter")
73
+ | pynini.cross("zero point two five", "quarter")
74
+ ),
75
+ ).optimize()
76
+
77
+ graph_cardinal = (
78
+ pynutil.delete("cardinal {")
79
+ + delete_space
80
+ + optional_sign
81
+ + delete_space
82
+ + cardinal.numbers
83
+ + delete_space
84
+ + pynutil.delete("}")
85
+ )
86
+
87
+ graph_fraction = (
88
+ pynutil.delete("fraction {") + delete_space + fraction.graph + delete_space + pynutil.delete("}")
89
+ )
90
+
91
+ graph = (graph_cardinal | graph_decimal | graph_fraction) + delete_space + insert_space + unit
92
+
93
+ # SH adds "preserve_order: true" by default
94
+ preserve_order = pynutil.delete("preserve_order:") + delete_space + pynutil.delete("true") + delete_space
95
+ graph |= unit + insert_space + (graph_cardinal | graph_decimal) + delete_space + pynini.closure(preserve_order)
96
+ # for only unit
97
+ graph |= (
98
+ pynutil.delete("cardinal { integer: \"-\"")
99
+ + delete_space
100
+ + pynutil.delete("}")
101
+ + delete_space
102
+ + unit
103
+ + pynini.closure(preserve_order)
104
+ )
105
+ address = (
106
+ pynutil.delete("units: \"address\" ")
107
+ + delete_space
108
+ + graph_cardinal
109
+ + delete_space
110
+ + pynini.closure(preserve_order)
111
+ )
112
+ math = (
113
+ pynutil.delete("units: \"math\" ")
114
+ + delete_space
115
+ + graph_cardinal
116
+ + delete_space
117
+ + pynini.closure(preserve_order)
118
+ )
119
+ graph |= address | math
120
+
121
+ delete_tokens = self.delete_tokens(graph)
122
+ self.fst = delete_tokens.optimize()
vendor/zonos3/vendor/nemo_text_processing/text_normalization/en/verbalizers/money.py ADDED
@@ -0,0 +1,85 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # Copyright (c) 2021, NVIDIA CORPORATION. All rights reserved.
2
+ #
3
+ # Licensed under the Apache License, Version 2.0 (the "License");
4
+ # you may not use this file except in compliance with the License.
5
+ # You may obtain a copy of the License at
6
+ #
7
+ # http://www.apache.org/licenses/LICENSE-2.0
8
+ #
9
+ # Unless required by applicable law or agreed to in writing, software
10
+ # distributed under the License is distributed on an "AS IS" BASIS,
11
+ # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
12
+ # See the License for the specific language governing permissions and
13
+ # limitations under the License.
14
+
15
+ import pynini
16
+ from pynini.lib import pynutil
17
+
18
+ from zonos3.vendor.nemo_text_processing.text_normalization.en.graph_utils import (
19
+ NEMO_NOT_QUOTE,
20
+ GraphFst,
21
+ delete_extra_space,
22
+ delete_preserve_order,
23
+ )
24
+
25
+ from zonos3.vendor.nemo_text_processing.text_normalization.en.utils import get_abs_path
26
+
27
+ per_units = pynini.string_file(get_abs_path("data/money/per_unit.tsv"))
28
+
29
+
30
+ class MoneyFst(GraphFst):
31
+ """
32
+ Finite state transducer for verbalizing money, e.g.
33
+ money { integer_part: "twelve" fractional_part: "o five" currency: "dollars" } -> twelve o five dollars
34
+
35
+ Args:
36
+ decimal: DecimalFst
37
+ deterministic: if True will provide a single transduction option,
38
+ for False multiple transduction are generated (used for audio-based normalization)
39
+ """
40
+
41
+ def __init__(self, decimal: GraphFst, deterministic: bool = True):
42
+ super().__init__(name="money", kind="verbalize", deterministic=deterministic)
43
+ keep_space = pynini.accep(" ")
44
+ maj = pynutil.delete('currency_maj: "') + pynini.closure(NEMO_NOT_QUOTE, 1) + pynutil.delete('"')
45
+ min = pynutil.delete('currency_min: "') + pynini.closure(NEMO_NOT_QUOTE, 1) + pynutil.delete('"')
46
+
47
+ fractional_part = (
48
+ pynutil.delete('fractional_part: "') + pynini.closure(NEMO_NOT_QUOTE, 1) + pynutil.delete('"')
49
+ )
50
+
51
+ integer_part = decimal.integer
52
+
53
+ # *** currency_maj
54
+ graph_integer = integer_part + keep_space + maj
55
+
56
+ # *** currency_maj + (***) | ((and) *** current_min)
57
+ fractional = fractional_part + delete_extra_space + min
58
+
59
+ if not deterministic:
60
+ fractional |= pynutil.insert("and ") + fractional
61
+
62
+ graph_integer_with_minor = integer_part + keep_space + maj + keep_space + fractional + delete_preserve_order
63
+
64
+ # *** point *** currency_maj
65
+ graph_decimal = decimal.numbers + keep_space + maj
66
+
67
+ # *** current_min
68
+ graph_minor = fractional_part + delete_extra_space + min + delete_preserve_order
69
+
70
+ graph = graph_integer | graph_integer_with_minor | graph_decimal | graph_minor
71
+
72
+ if not deterministic:
73
+ graph |= graph_integer + delete_preserve_order
74
+
75
+ per_units_normalized = pynini.project(per_units, "output")
76
+ remove_per_units_normalized = (
77
+ pynutil.delete(' morphosyntactic_features: "')
78
+ + pynutil.insert(" ")
79
+ + per_units_normalized
80
+ + pynutil.delete('" ')
81
+ )
82
+ graph += remove_per_units_normalized.ques
83
+
84
+ delete_tokens = self.delete_tokens(graph)
85
+ self.fst = delete_tokens.optimize()
vendor/zonos3/vendor/nemo_text_processing/text_normalization/en/verbalizers/ordinal.py ADDED
@@ -0,0 +1,57 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # Copyright (c) 2021, NVIDIA CORPORATION. All rights reserved.
2
+ #
3
+ # Licensed under the Apache License, Version 2.0 (the "License");
4
+ # you may not use this file except in compliance with the License.
5
+ # You may obtain a copy of the License at
6
+ #
7
+ # http://www.apache.org/licenses/LICENSE-2.0
8
+ #
9
+ # Unless required by applicable law or agreed to in writing, software
10
+ # distributed under the License is distributed on an "AS IS" BASIS,
11
+ # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
12
+ # See the License for the specific language governing permissions and
13
+ # limitations under the License.
14
+
15
+
16
+ import pynini
17
+ from pynini.lib import pynutil
18
+
19
+ from zonos3.vendor.nemo_text_processing.text_normalization.en.graph_utils import NEMO_NOT_QUOTE, NEMO_SIGMA, GraphFst, delete_space
20
+ from zonos3.vendor.nemo_text_processing.text_normalization.en.utils import get_abs_path
21
+
22
+
23
+ class OrdinalFst(GraphFst):
24
+ """
25
+ Finite state transducer for verbalizing ordinal, e.g.
26
+ ordinal { integer: "thirteen" } } -> thirteenth
27
+
28
+ Args:
29
+ deterministic: if True will provide a single transduction option,
30
+ for False multiple transduction are generated (used for audio-based normalization)
31
+ """
32
+
33
+ def __init__(self, deterministic: bool = True):
34
+ super().__init__(name="ordinal", kind="verbalize", deterministic=deterministic)
35
+
36
+ graph_digit = pynini.string_file(get_abs_path("data/ordinal/digit.tsv")).invert()
37
+ graph_teens = pynini.string_file(get_abs_path("data/ordinal/teen.tsv")).invert()
38
+
39
+ graph = (
40
+ pynutil.delete("integer:")
41
+ + delete_space
42
+ + pynutil.delete("\"")
43
+ + pynini.closure(NEMO_NOT_QUOTE, 1)
44
+ + pynutil.delete("\"")
45
+ )
46
+ convert_rest = pynutil.insert("th")
47
+
48
+ suffix = pynini.cdrewrite(
49
+ graph_digit | graph_teens | pynini.cross("ty", "tieth") | convert_rest,
50
+ "",
51
+ "[EOS]",
52
+ NEMO_SIGMA,
53
+ ).optimize()
54
+ self.graph = pynini.compose(graph, suffix)
55
+ self.suffix = suffix
56
+ delete_tokens = self.delete_tokens(self.graph)
57
+ self.fst = delete_tokens.optimize()
vendor/zonos3/vendor/nemo_text_processing/text_normalization/en/verbalizers/post_processing.py ADDED
@@ -0,0 +1,182 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # Copyright (c) 2022, NVIDIA CORPORATION & AFFILIATES. All rights reserved.
2
+ #
3
+ # Licensed under the Apache License, Version 2.0 (the "License");
4
+ # you may not use this file except in compliance with the License.
5
+ # You may obtain a copy of the License at
6
+ #
7
+ # http://www.apache.org/licenses/LICENSE-2.0
8
+ #
9
+ # Unless required by applicable law or agreed to in writing, software
10
+ # distributed under the License is distributed on an "AS IS" BASIS,
11
+ # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
12
+ # See the License for the specific language governing permissions and
13
+ # limitations under the License.
14
+
15
+ import os
16
+
17
+ import pynini
18
+ from pynini.lib import pynutil
19
+
20
+ from zonos3.vendor.nemo_text_processing.text_normalization.en.graph_utils import (
21
+ MIN_NEG_WEIGHT,
22
+ NEMO_ALPHA,
23
+ NEMO_CHAR,
24
+ NEMO_SIGMA,
25
+ NEMO_SPACE,
26
+ generator_main,
27
+ )
28
+ from zonos3.vendor.nemo_text_processing.text_normalization.en.taggers.punctuation import PunctuationFst
29
+ from zonos3.vendor.nemo_text_processing.utils.logging import logger
30
+
31
+
32
+ class PostProcessingFst:
33
+ """
34
+ Finite state transducer that post-processing an entire sentence after verbalization is complete, e.g.
35
+ removes extra spaces around punctuation marks " ( one hundred and twenty three ) " -> "(one hundred and twenty three)"
36
+
37
+ Args:
38
+ cache_dir: path to a dir with .far grammar file. Set to None to avoid using cache.
39
+ overwrite_cache: set to True to overwrite .far files
40
+ """
41
+
42
+ def __init__(self, cache_dir: str = None, overwrite_cache: bool = False):
43
+
44
+ far_file = None
45
+ if cache_dir is not None and cache_dir != "None":
46
+ os.makedirs(cache_dir, exist_ok=True)
47
+ far_file = os.path.join(cache_dir, "en_tn_post_processing.far")
48
+ if not overwrite_cache and far_file and os.path.exists(far_file):
49
+ self.fst = pynini.Far(far_file, mode="r")["post_process_graph"]
50
+ logger.info(f'Post processing graph was restored from {far_file}.')
51
+ else:
52
+ self.set_punct_dict()
53
+ self.fst = self.get_punct_postprocess_graph()
54
+
55
+ if far_file:
56
+ generator_main(far_file, {"post_process_graph": self.fst})
57
+
58
+ def set_punct_dict(self):
59
+ self.punct_marks = {
60
+ "'": [
61
+ "'",
62
+ '´',
63
+ 'ʹ',
64
+ 'ʻ',
65
+ 'ʼ',
66
+ 'ʽ',
67
+ 'ʾ',
68
+ 'ˈ',
69
+ 'ˊ',
70
+ 'ˋ',
71
+ '˴',
72
+ 'ʹ',
73
+ '΄',
74
+ '՚',
75
+ '՝',
76
+ 'י',
77
+ '׳',
78
+ 'ߴ',
79
+ 'ߵ',
80
+ 'ᑊ',
81
+ 'ᛌ',
82
+ '᾽',
83
+ '᾿',
84
+ '`',
85
+ '´',
86
+ '῾',
87
+ '‘',
88
+ '’',
89
+ '‛',
90
+ '′',
91
+ '‵',
92
+ 'ꞌ',
93
+ ''',
94
+ '`',
95
+ '𖽑',
96
+ '𖽒',
97
+ ],
98
+ }
99
+
100
+ def get_punct_postprocess_graph(self):
101
+ """
102
+ Returns graph to post process punctuation marks.
103
+
104
+ {``} quotes are converted to {"}. Note, if there are spaces around single quote {'}, they will be kept.
105
+ By default, a space is added after a punctuation mark, and spaces are removed before punctuation marks.
106
+ """
107
+ punct_marks_all = PunctuationFst().punct_marks
108
+
109
+ # no_space_before_punct assume no space before them
110
+ quotes = ["'", "\"", "``", "«"]
111
+ dashes = ["-", "—"]
112
+ brackets = ["<", "{", "(", r"\["]
113
+ open_close_single_quotes = [
114
+ ("`", "`"),
115
+ ]
116
+
117
+ open_close_double_quotes = [('"', '"'), ("``", "``"), ("“", "”")]
118
+ open_close_symbols = open_close_single_quotes + open_close_double_quotes
119
+ allow_space_before_punct = ["&"] + quotes + dashes + brackets + [k[0] for k in open_close_symbols]
120
+
121
+ no_space_before_punct = [m for m in punct_marks_all if m not in allow_space_before_punct]
122
+ no_space_before_punct = pynini.union(*no_space_before_punct)
123
+ no_space_after_punct = pynini.union(*brackets)
124
+ delete_space = pynutil.delete(" ")
125
+ delete_space_optional = pynini.closure(delete_space, 0, 1)
126
+
127
+ # non_punct allows space
128
+ # delete space before no_space_before_punct marks, if present
129
+ non_punct = pynini.difference(NEMO_CHAR, no_space_before_punct).optimize()
130
+ graph = (
131
+ pynini.closure(non_punct)
132
+ + pynini.closure(
133
+ no_space_before_punct | pynutil.add_weight(delete_space + no_space_before_punct, MIN_NEG_WEIGHT)
134
+ )
135
+ + pynini.closure(non_punct)
136
+ )
137
+ graph = pynini.closure(graph).optimize()
138
+ graph = pynini.compose(
139
+ graph, pynini.cdrewrite(pynini.cross("``", '"'), "", "", NEMO_SIGMA).optimize()
140
+ ).optimize()
141
+
142
+ # remove space after no_space_after_punct (even if there are no matching closing brackets)
143
+ no_space_after_punct = pynini.cdrewrite(delete_space, no_space_after_punct, NEMO_SIGMA, NEMO_SIGMA).optimize()
144
+ graph = pynini.compose(graph, no_space_after_punct).optimize()
145
+
146
+ # remove space around text in quotes
147
+ single_quote = pynutil.add_weight(pynini.accep("`"), MIN_NEG_WEIGHT)
148
+ double_quotes = pynutil.add_weight(pynini.accep('"'), MIN_NEG_WEIGHT)
149
+ quotes_graph = (
150
+ single_quote + delete_space_optional + NEMO_ALPHA + NEMO_SIGMA + delete_space_optional + single_quote
151
+ ).optimize()
152
+
153
+ # this is to make sure multiple quotes are tagged from right to left without skipping any quotes in the left
154
+ not_alpha = pynini.difference(NEMO_CHAR, NEMO_ALPHA).optimize() | pynutil.add_weight(
155
+ NEMO_SPACE, MIN_NEG_WEIGHT
156
+ )
157
+ end = pynini.closure(pynutil.add_weight(not_alpha, MIN_NEG_WEIGHT))
158
+ quotes_graph |= (
159
+ double_quotes
160
+ + delete_space_optional
161
+ + NEMO_ALPHA
162
+ + NEMO_SIGMA
163
+ + delete_space_optional
164
+ + double_quotes
165
+ + end
166
+ )
167
+
168
+ quotes_graph = pynutil.add_weight(quotes_graph, MIN_NEG_WEIGHT)
169
+ quotes_graph = NEMO_SIGMA + pynini.closure(NEMO_SIGMA + quotes_graph + NEMO_SIGMA)
170
+
171
+ graph = pynini.compose(graph, quotes_graph).optimize()
172
+
173
+ # remove space between a word and a single quote followed by s
174
+ remove_space_around_single_quote = pynini.cdrewrite(
175
+ delete_space_optional + pynini.union(*self.punct_marks["'"]) + delete_space,
176
+ NEMO_ALPHA,
177
+ pynini.union("s ", "s[EOS]"),
178
+ NEMO_SIGMA,
179
+ )
180
+
181
+ graph = pynini.compose(graph, remove_space_around_single_quote).optimize()
182
+ return graph
vendor/zonos3/vendor/nemo_text_processing/text_normalization/en/verbalizers/roman.py ADDED
@@ -0,0 +1,69 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # Copyright (c) 2021, NVIDIA CORPORATION. All rights reserved.
2
+ #
3
+ # Licensed under the Apache License, Version 2.0 (the "License");
4
+ # you may not use this file except in compliance with the License.
5
+ # You may obtain a copy of the License at
6
+ #
7
+ # http://www.apache.org/licenses/LICENSE-2.0
8
+ #
9
+ # Unless required by applicable law or agreed to in writing, software
10
+ # distributed under the License is distributed on an "AS IS" BASIS,
11
+ # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
12
+ # See the License for the specific language governing permissions and
13
+ # limitations under the License.
14
+
15
+ import pynini
16
+ from pynini.lib import pynutil
17
+
18
+ from zonos3.vendor.nemo_text_processing.text_normalization.en.graph_utils import NEMO_NOT_QUOTE, GraphFst
19
+ from zonos3.vendor.nemo_text_processing.text_normalization.en.verbalizers.ordinal import OrdinalFst
20
+
21
+
22
+ class RomanFst(GraphFst):
23
+ """
24
+ Finite state transducer for verbalizing roman numerals
25
+ e.g. tokens { roman { integer: "one" } } -> one
26
+
27
+ Args:
28
+ deterministic: if True will provide a single transduction option,
29
+ for False multiple transduction are generated (used for audio-based normalization)
30
+ """
31
+
32
+ def __init__(self, deterministic: bool = True):
33
+ super().__init__(name="roman", kind="verbalize", deterministic=deterministic)
34
+ suffix = OrdinalFst().suffix
35
+
36
+ cardinal = pynini.closure(NEMO_NOT_QUOTE)
37
+ ordinal = pynini.compose(cardinal, suffix)
38
+
39
+ graph = (
40
+ pynutil.delete("key_cardinal: \"")
41
+ + pynini.closure(NEMO_NOT_QUOTE, 1)
42
+ + pynutil.delete("\"")
43
+ + pynini.accep(" ")
44
+ + pynutil.delete("integer: \"")
45
+ + cardinal
46
+ + pynutil.delete("\"")
47
+ ).optimize()
48
+
49
+ graph |= (
50
+ pynutil.delete("default_cardinal: \"default\" integer: \"") + cardinal + pynutil.delete("\"")
51
+ ).optimize()
52
+
53
+ graph |= (
54
+ pynutil.delete("default_ordinal: \"default\" integer: \"") + ordinal + pynutil.delete("\"")
55
+ ).optimize()
56
+
57
+ graph |= (
58
+ pynutil.delete("key_the_ordinal: \"")
59
+ + pynini.closure(NEMO_NOT_QUOTE, 1)
60
+ + pynutil.delete("\"")
61
+ + pynini.accep(" ")
62
+ + pynutil.delete("integer: \"")
63
+ + pynini.closure(pynutil.insert("the "), 0, 1)
64
+ + ordinal
65
+ + pynutil.delete("\"")
66
+ ).optimize()
67
+
68
+ delete_tokens = self.delete_tokens(graph)
69
+ self.fst = delete_tokens.optimize()
vendor/zonos3/vendor/nemo_text_processing/text_normalization/en/verbalizers/telephone.py ADDED
@@ -0,0 +1,64 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # Copyright (c) 2021, NVIDIA CORPORATION. All rights reserved.
2
+ #
3
+ # Licensed under the Apache License, Version 2.0 (the "License");
4
+ # you may not use this file except in compliance with the License.
5
+ # You may obtain a copy of the License at
6
+ #
7
+ # http://www.apache.org/licenses/LICENSE-2.0
8
+ #
9
+ # Unless required by applicable law or agreed to in writing, software
10
+ # distributed under the License is distributed on an "AS IS" BASIS,
11
+ # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
12
+ # See the License for the specific language governing permissions and
13
+ # limitations under the License.
14
+
15
+ import pynini
16
+ from pynini.lib import pynutil
17
+
18
+ from zonos3.vendor.nemo_text_processing.text_normalization.en.graph_utils import NEMO_NOT_QUOTE, GraphFst, delete_space, insert_space
19
+
20
+
21
+ class TelephoneFst(GraphFst):
22
+ """
23
+ Finite state transducer for verbalizing telephone numbers, e.g.
24
+ telephone { country_code: "one" number_part: "one two three, one two three, five six seven eight" extension: "one" }
25
+ -> one, one two three, one two three, five six seven eight, one
26
+
27
+ Args:
28
+ deterministic: if True will provide a single transduction option,
29
+ for False multiple transduction are generated (used for audio-based normalization)
30
+ """
31
+
32
+ def __init__(self, deterministic: bool = True):
33
+ super().__init__(name="telephone", kind="verbalize", deterministic=deterministic)
34
+
35
+ optional_country_code = pynini.closure(
36
+ pynutil.delete("country_code: \"")
37
+ + pynini.closure(NEMO_NOT_QUOTE, 1)
38
+ + pynutil.delete("\"")
39
+ + delete_space
40
+ + insert_space,
41
+ 0,
42
+ 1,
43
+ )
44
+
45
+ number_part = (
46
+ pynutil.delete("number_part: \"")
47
+ + pynini.closure(NEMO_NOT_QUOTE, 1)
48
+ + pynini.closure(pynutil.add_weight(pynutil.delete(" "), -0.0001), 0, 1)
49
+ + pynutil.delete("\"")
50
+ )
51
+
52
+ optional_extension = pynini.closure(
53
+ delete_space
54
+ + insert_space
55
+ + pynutil.delete("extension: \"")
56
+ + pynini.closure(NEMO_NOT_QUOTE, 1)
57
+ + pynutil.delete("\""),
58
+ 0,
59
+ 1,
60
+ )
61
+
62
+ graph = optional_country_code + number_part + optional_extension
63
+ delete_tokens = self.delete_tokens(graph)
64
+ self.fst = delete_tokens.optimize()
vendor/zonos3/vendor/nemo_text_processing/text_normalization/en/verbalizers/time.py ADDED
@@ -0,0 +1,103 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # Copyright (c) 2021, NVIDIA CORPORATION. All rights reserved.
2
+ #
3
+ # Licensed under the Apache License, Version 2.0 (the "License");
4
+ # you may not use this file except in compliance with the License.
5
+ # You may obtain a copy of the License at
6
+ #
7
+ # http://www.apache.org/licenses/LICENSE-2.0
8
+ #
9
+ # Unless required by applicable law or agreed to in writing, software
10
+ # distributed under the License is distributed on an "AS IS" BASIS,
11
+ # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
12
+ # See the License for the specific language governing permissions and
13
+ # limitations under the License.
14
+
15
+ import pynini
16
+ from pynini.lib import pynutil
17
+
18
+ from zonos3.vendor.nemo_text_processing.text_normalization.en.graph_utils import (
19
+ NEMO_NOT_QUOTE,
20
+ NEMO_SIGMA,
21
+ GraphFst,
22
+ delete_space,
23
+ insert_space,
24
+ )
25
+
26
+
27
+ class TimeFst(GraphFst):
28
+ """
29
+ Finite state transducer for verbalizing time, e.g.
30
+ time { hours: "twelve" minutes: "thirty" suffix: "a m" zone: "e s t" } -> twelve thirty a m e s t
31
+ time { hours: "twelve" } -> twelve o'clock
32
+
33
+ Args:
34
+ deterministic: if True will provide a single transduction option,
35
+ for False multiple transduction are generated (used for audio-based normalization)
36
+ """
37
+
38
+ def __init__(self, deterministic: bool = True):
39
+ super().__init__(name="time", kind="verbalize", deterministic=deterministic)
40
+ hour = (
41
+ pynutil.delete("hours:")
42
+ + delete_space
43
+ + pynutil.delete("\"")
44
+ + pynini.closure(NEMO_NOT_QUOTE, 1)
45
+ + pynutil.delete("\"")
46
+ )
47
+ minute = (
48
+ pynutil.delete("minutes:")
49
+ + delete_space
50
+ + pynutil.delete("\"")
51
+ + pynini.closure(NEMO_NOT_QUOTE, 1)
52
+ + pynutil.delete("\"")
53
+ )
54
+ suffix = (
55
+ pynutil.delete("suffix:")
56
+ + delete_space
57
+ + pynutil.delete("\"")
58
+ + pynini.closure(NEMO_NOT_QUOTE, 1)
59
+ + pynutil.delete("\"")
60
+ )
61
+ optional_suffix = pynini.closure(delete_space + insert_space + suffix, 0, 1)
62
+ zone = (
63
+ pynutil.delete("zone:")
64
+ + delete_space
65
+ + pynutil.delete("\"")
66
+ + pynini.closure(NEMO_NOT_QUOTE, 1)
67
+ + pynutil.delete("\"")
68
+ )
69
+ optional_zone = pynini.closure(delete_space + insert_space + zone, 0, 1)
70
+ second = (
71
+ pynutil.delete("seconds:")
72
+ + delete_space
73
+ + pynutil.delete("\"")
74
+ + pynini.closure(NEMO_NOT_QUOTE, 1)
75
+ + pynutil.delete("\"")
76
+ )
77
+ graph_hms = (
78
+ hour
79
+ + pynutil.insert(" hours ")
80
+ + delete_space
81
+ + minute
82
+ + pynutil.insert(" minutes and ")
83
+ + delete_space
84
+ + second
85
+ + pynutil.insert(" seconds")
86
+ + optional_suffix
87
+ + optional_zone
88
+ )
89
+ graph_hms @= pynini.cdrewrite(pynini.cross("one hours", "one hour"), "[BOS]", "", NEMO_SIGMA)
90
+ graph_hms @= pynini.cdrewrite(
91
+ pynutil.delete("o ")
92
+ | pynutil.add_weight(pynini.cross("o one minutes", "one minute"), -0.01)
93
+ | pynutil.add_weight(pynini.cross("o one seconds", "one second"), -0.01),
94
+ " ",
95
+ "",
96
+ NEMO_SIGMA,
97
+ )
98
+ graph = hour + delete_space + insert_space + minute + optional_suffix + optional_zone
99
+ graph |= hour + insert_space + pynutil.insert("o'clock") + optional_zone
100
+ graph |= hour + delete_space + insert_space + suffix + optional_zone
101
+ graph |= graph_hms
102
+ delete_tokens = self.delete_tokens(graph)
103
+ self.fst = delete_tokens.optimize()
vendor/zonos3/vendor/nemo_text_processing/text_normalization/en/verbalizers/verbalize.py ADDED
@@ -0,0 +1,82 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # Copyright (c) 2021, NVIDIA CORPORATION. All rights reserved.
2
+ #
3
+ # Licensed under the Apache License, Version 2.0 (the "License");
4
+ # you may not use this file except in compliance with the License.
5
+ # You may obtain a copy of the License at
6
+ #
7
+ # http://www.apache.org/licenses/LICENSE-2.0
8
+ #
9
+ # Unless required by applicable law or agreed to in writing, software
10
+ # distributed under the License is distributed on an "AS IS" BASIS,
11
+ # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
12
+ # See the License for the specific language governing permissions and
13
+ # limitations under the License.
14
+
15
+ from zonos3.vendor.nemo_text_processing.text_normalization.en.graph_utils import GraphFst
16
+ from zonos3.vendor.nemo_text_processing.text_normalization.en.verbalizers.abbreviation import AbbreviationFst
17
+ from zonos3.vendor.nemo_text_processing.text_normalization.en.verbalizers.cardinal import CardinalFst
18
+ from zonos3.vendor.nemo_text_processing.text_normalization.en.verbalizers.date import DateFst
19
+ from zonos3.vendor.nemo_text_processing.text_normalization.en.verbalizers.decimal import DecimalFst
20
+ from zonos3.vendor.nemo_text_processing.text_normalization.en.verbalizers.electronic import ElectronicFst
21
+ from zonos3.vendor.nemo_text_processing.text_normalization.en.verbalizers.fraction import FractionFst
22
+ from zonos3.vendor.nemo_text_processing.text_normalization.en.verbalizers.measure import MeasureFst
23
+ from zonos3.vendor.nemo_text_processing.text_normalization.en.verbalizers.money import MoneyFst
24
+ from zonos3.vendor.nemo_text_processing.text_normalization.en.verbalizers.ordinal import OrdinalFst
25
+ from zonos3.vendor.nemo_text_processing.text_normalization.en.verbalizers.roman import RomanFst
26
+ from zonos3.vendor.nemo_text_processing.text_normalization.en.verbalizers.telephone import TelephoneFst
27
+ from zonos3.vendor.nemo_text_processing.text_normalization.en.verbalizers.time import TimeFst
28
+ from zonos3.vendor.nemo_text_processing.text_normalization.en.verbalizers.whitelist import WhiteListFst
29
+
30
+
31
+ class VerbalizeFst(GraphFst):
32
+ """
33
+ Composes other verbalizer grammars.
34
+ For deployment, this grammar will be compiled and exported to OpenFst Finite State Archive (FAR) File.
35
+ More details to deployment at NeMo/tools/text_processing_deployment.
36
+
37
+ Args:
38
+ deterministic: if True will provide a single transduction option,
39
+ for False multiple options (used for audio-based normalization)
40
+ """
41
+
42
+ def __init__(self, deterministic: bool = True):
43
+ super().__init__(name="verbalize", kind="verbalize", deterministic=deterministic)
44
+ cardinal = CardinalFst(deterministic=deterministic)
45
+ cardinal_graph = cardinal.fst
46
+ decimal = DecimalFst(cardinal=cardinal, deterministic=deterministic)
47
+ decimal_graph = decimal.fst
48
+ ordinal = OrdinalFst(deterministic=deterministic)
49
+ ordinal_graph = ordinal.fst
50
+ fraction = FractionFst(deterministic=deterministic)
51
+ fraction_graph = fraction.fst
52
+ telephone_graph = TelephoneFst(deterministic=deterministic).fst
53
+ electronic_graph = ElectronicFst(deterministic=deterministic).fst
54
+ measure = MeasureFst(decimal=decimal, cardinal=cardinal, fraction=fraction, deterministic=deterministic)
55
+ measure_graph = measure.fst
56
+ time_graph = TimeFst(deterministic=deterministic).fst
57
+ date_graph = DateFst(ordinal=ordinal, deterministic=deterministic).fst
58
+ money_graph = MoneyFst(decimal=decimal, deterministic=deterministic).fst
59
+ whitelist_graph = WhiteListFst(deterministic=deterministic).fst
60
+
61
+ graph = (
62
+ time_graph
63
+ | date_graph
64
+ | money_graph
65
+ | measure_graph
66
+ | ordinal_graph
67
+ | decimal_graph
68
+ | cardinal_graph
69
+ | telephone_graph
70
+ | electronic_graph
71
+ | fraction_graph
72
+ | whitelist_graph
73
+ )
74
+
75
+ roman_graph = RomanFst(deterministic=deterministic).fst
76
+ graph |= roman_graph
77
+
78
+ if not deterministic:
79
+ abbreviation_graph = AbbreviationFst(deterministic=deterministic).fst
80
+ graph |= abbreviation_graph
81
+
82
+ self.fst = graph
vendor/zonos3/vendor/nemo_text_processing/text_normalization/en/verbalizers/verbalize_final.py ADDED
@@ -0,0 +1,75 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # Copyright (c) 2021, NVIDIA CORPORATION. All rights reserved.
2
+ #
3
+ # Licensed under the Apache License, Version 2.0 (the "License");
4
+ # you may not use this file except in compliance with the License.
5
+ # You may obtain a copy of the License at
6
+ #
7
+ # http://www.apache.org/licenses/LICENSE-2.0
8
+ #
9
+ # Unless required by applicable law or agreed to in writing, software
10
+ # distributed under the License is distributed on an "AS IS" BASIS,
11
+ # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
12
+ # See the License for the specific language governing permissions and
13
+ # limitations under the License.
14
+
15
+ import os
16
+
17
+ import pynini
18
+ from pynini.lib import pynutil
19
+
20
+ from zonos3.vendor.nemo_text_processing.text_normalization.en.graph_utils import (
21
+ GraphFst,
22
+ delete_extra_space,
23
+ delete_space,
24
+ generator_main,
25
+ )
26
+ from zonos3.vendor.nemo_text_processing.text_normalization.en.verbalizers.verbalize import VerbalizeFst
27
+ from zonos3.vendor.nemo_text_processing.text_normalization.en.verbalizers.word import WordFst
28
+ from zonos3.vendor.nemo_text_processing.utils.logging import logger
29
+
30
+
31
+ class VerbalizeFinalFst(GraphFst):
32
+ """
33
+ Finite state transducer that verbalizes an entire sentence, e.g.
34
+ tokens { name: "its" } tokens { time { hours: "twelve" minutes: "thirty" } } tokens { name: "now" } tokens { name: "." } -> its twelve thirty now .
35
+
36
+ Args:
37
+ deterministic: if True will provide a single transduction option,
38
+ for False multiple options (used for audio-based normalization)
39
+ cache_dir: path to a dir with .far grammar file. Set to None to avoid using cache.
40
+ overwrite_cache: set to True to overwrite .far files
41
+ """
42
+
43
+ def __init__(self, deterministic: bool = True, cache_dir: str = None, overwrite_cache: bool = False):
44
+ super().__init__(name="verbalize_final", kind="verbalize", deterministic=deterministic)
45
+
46
+ far_file = None
47
+ if cache_dir is not None and cache_dir != "None":
48
+ os.makedirs(cache_dir, exist_ok=True)
49
+ far_file = os.path.join(cache_dir, f"en_tn_{deterministic}_deterministic_verbalizer.far")
50
+ if not overwrite_cache and far_file and os.path.exists(far_file):
51
+ self.fst = pynini.Far(far_file, mode="r")["verbalize"]
52
+ logger.info(f'VerbalizeFinalFst graph was restored from {far_file}.')
53
+ else:
54
+ verbalize = VerbalizeFst(deterministic=deterministic).fst
55
+ word = WordFst(deterministic=deterministic).fst
56
+ types = verbalize | word
57
+
58
+ if deterministic:
59
+ graph = (
60
+ pynutil.delete("tokens")
61
+ + delete_space
62
+ + pynutil.delete("{")
63
+ + delete_space
64
+ + types
65
+ + delete_space
66
+ + pynutil.delete("}")
67
+ )
68
+ else:
69
+ graph = delete_space + types + delete_space
70
+
71
+ graph = delete_space + pynini.closure(graph + delete_extra_space) + graph + delete_space
72
+
73
+ self.fst = graph.optimize()
74
+ if far_file:
75
+ generator_main(far_file, {"verbalize": self.fst})
vendor/zonos3/vendor/nemo_text_processing/text_normalization/en/verbalizers/whitelist.py ADDED
@@ -0,0 +1,40 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # Copyright (c) 2021, NVIDIA CORPORATION. All rights reserved.
2
+ #
3
+ # Licensed under the Apache License, Version 2.0 (the "License");
4
+ # you may not use this file except in compliance with the License.
5
+ # You may obtain a copy of the License at
6
+ #
7
+ # http://www.apache.org/licenses/LICENSE-2.0
8
+ #
9
+ # Unless required by applicable law or agreed to in writing, software
10
+ # distributed under the License is distributed on an "AS IS" BASIS,
11
+ # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
12
+ # See the License for the specific language governing permissions and
13
+ # limitations under the License.
14
+ import pynini
15
+ from pynini.lib import pynutil
16
+
17
+ from zonos3.vendor.nemo_text_processing.text_normalization.en.graph_utils import NEMO_CHAR, NEMO_SIGMA, GraphFst, delete_space
18
+
19
+
20
+ class WhiteListFst(GraphFst):
21
+ """
22
+ Finite state transducer for verbalizing whitelist
23
+ e.g. tokens { name: "misses" } } -> misses
24
+
25
+ Args:
26
+ deterministic: if True will provide a single transduction option,
27
+ for False multiple transduction are generated (used for audio-based normalization)
28
+ """
29
+
30
+ def __init__(self, deterministic: bool = True):
31
+ super().__init__(name="whitelist", kind="verbalize", deterministic=deterministic)
32
+ graph = (
33
+ pynutil.delete("name:")
34
+ + delete_space
35
+ + pynutil.delete("\"")
36
+ + pynini.closure(NEMO_CHAR - " ", 1)
37
+ + pynutil.delete("\"")
38
+ )
39
+ graph = graph @ pynini.cdrewrite(pynini.cross(u"\u00a0", " "), "", "", NEMO_SIGMA)
40
+ self.fst = graph.optimize()
vendor/zonos3/vendor/nemo_text_processing/text_normalization/en/verbalizers/word.py ADDED
@@ -0,0 +1,36 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # Copyright (c) 2021, NVIDIA CORPORATION. All rights reserved.
2
+ #
3
+ # Licensed under the Apache License, Version 2.0 (the "License");
4
+ # you may not use this file except in compliance with the License.
5
+ # You may obtain a copy of the License at
6
+ #
7
+ # http://www.apache.org/licenses/LICENSE-2.0
8
+ #
9
+ # Unless required by applicable law or agreed to in writing, software
10
+ # distributed under the License is distributed on an "AS IS" BASIS,
11
+ # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
12
+ # See the License for the specific language governing permissions and
13
+ # limitations under the License.
14
+ import pynini
15
+ from pynini.lib import pynutil
16
+
17
+ from zonos3.vendor.nemo_text_processing.text_normalization.en.graph_utils import NEMO_CHAR, NEMO_SIGMA, GraphFst, delete_space
18
+
19
+
20
+ class WordFst(GraphFst):
21
+ """
22
+ Finite state transducer for verbalizing word
23
+ e.g. tokens { name: "sleep" } -> sleep
24
+
25
+ Args:
26
+ deterministic: if True will provide a single transduction option,
27
+ for False multiple transduction are generated (used for audio-based normalization)
28
+ """
29
+
30
+ def __init__(self, deterministic: bool = True):
31
+ super().__init__(name="word", kind="verbalize", deterministic=deterministic)
32
+ chars = pynini.closure(NEMO_CHAR - " ", 1)
33
+ char = pynutil.delete("name:") + delete_space + pynutil.delete("\"") + chars + pynutil.delete("\"")
34
+ graph = char @ pynini.cdrewrite(pynini.cross(u"\u00a0", " "), "", "", NEMO_SIGMA)
35
+
36
+ self.fst = graph.optimize()
vendor/zonos3/vendor/nemo_text_processing/text_normalization/es/README.md ADDED
@@ -0,0 +1,22 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # Localization Note
2
+
3
+ Depending on locale, Spanish number strings will vary in formatting. In the EU and South American countries, it is common to use a period (".") or space to delineate groupings of three
4
+ digits. e.g.
5
+ `1.000.000` -> "un millón"
6
+ `1 000 000` -> "un millón"
7
+
8
+ and commas (",") to seperate cardinal and decimal strings. e.g.
9
+
10
+ `1,00` -> "uno coma cero cero"
11
+
12
+ While Central and Northern America will use commas (",") to delineate groupings of three digits, e.g.
13
+ `1,000,000` -> "un millón"
14
+
15
+ and periods (".") to seperate cardinal and decimal strings. e.g.
16
+
17
+ `1.00` -> "uno coma cero cero"
18
+
19
+ As inclusion of both forms will create inherrent ambiguity for verbalization, this module defaults to the former formatting (periods for cardinal delineation and commas for decimals).
20
+
21
+ To toggle the alternate formatting, you may edit the `LOCALIZATION` variable in `nemo_text_processing.text_normalization.es.__init__` with the value of `'am'`. This will perform necessary
22
+ adjustments to all affected classes.
vendor/zonos3/vendor/nemo_text_processing/text_normalization/es/__init__.py ADDED
@@ -0,0 +1,15 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # Copyright (c) 2022, NVIDIA CORPORATION & AFFILIATES. All rights reserved.
2
+ #
3
+ # Licensed under the Apache License, Version 2.0 (the "License");
4
+ # you may not use this file except in compliance with the License.
5
+ # You may obtain a copy of the License at
6
+ #
7
+ # http://www.apache.org/licenses/LICENSE-2.0
8
+ #
9
+ # Unless required by applicable law or agreed to in writing, software
10
+ # distributed under the License is distributed on an "AS IS" BASIS,
11
+ # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
12
+ # See the License for the specific language governing permissions and
13
+ # limitations under the License.
14
+
15
+ LOCALIZATION = "eu" # Set to am for alternate formatting
vendor/zonos3/vendor/nemo_text_processing/text_normalization/es/data/__init__.py ADDED
@@ -0,0 +1,13 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # Copyright (c) 2022, NVIDIA CORPORATION & AFFILIATES. All rights reserved.
2
+ #
3
+ # Licensed under the Apache License, Version 2.0 (the "License");
4
+ # you may not use this file except in compliance with the License.
5
+ # You may obtain a copy of the License at
6
+ #
7
+ # http://www.apache.org/licenses/LICENSE-2.0
8
+ #
9
+ # Unless required by applicable law or agreed to in writing, software
10
+ # distributed under the License is distributed on an "AS IS" BASIS,
11
+ # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
12
+ # See the License for the specific language governing permissions and
13
+ # limitations under the License.
vendor/zonos3/vendor/nemo_text_processing/text_normalization/es/data/dates/__init__.py ADDED
@@ -0,0 +1,13 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ # Copyright (c) 2022, NVIDIA CORPORATION & AFFILIATES. All rights reserved.
2
+ #
3
+ # Licensed under the Apache License, Version 2.0 (the "License");
4
+ # you may not use this file except in compliance with the License.
5
+ # You may obtain a copy of the License at
6
+ #
7
+ # http://www.apache.org/licenses/LICENSE-2.0
8
+ #
9
+ # Unless required by applicable law or agreed to in writing, software
10
+ # distributed under the License is distributed on an "AS IS" BASIS,
11
+ # WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
12
+ # See the License for the specific language governing permissions and
13
+ # limitations under the License.
vendor/zonos3/vendor/nemo_text_processing/text_normalization/es/data/dates/days_abbr.tsv ADDED
@@ -0,0 +1,27 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ LU lunes
2
+ lun. lunes
3
+ LUN lunes
4
+ ma. martes
5
+ MA martes
6
+ mar. martes
7
+ mi. miércoles
8
+ mié. miércoles
9
+ miér. miércoles
10
+ MIÉ miércoles
11
+ mié miércoles
12
+ miér miércoles
13
+ ju. jueves
14
+ JU jueves
15
+ jue jueves
16
+ JUE jueves
17
+ vie. viernes
18
+ vier. viernes
19
+ vie viernes
20
+ vier. viernes
21
+ sá. sábado
22
+ sáb sábado
23
+ sáb. sábado
24
+ SÁB sábado
25
+ do. domingo
26
+ dom. domingo
27
+ DOM domingo
vendor/zonos3/vendor/nemo_text_processing/text_normalization/es/data/dates/months.tsv ADDED
@@ -0,0 +1,12 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ 1 enero
2
+ 2 febrero
3
+ 3 marzo
4
+ 4 abril
5
+ 5 mayo
6
+ 6 junio
7
+ 7 julio
8
+ 8 agosto
9
+ 9 septiembre
10
+ 10 octubre
11
+ 11 noviembre
12
+ 12 diciembre
vendor/zonos3/vendor/nemo_text_processing/text_normalization/es/data/dates/months_abbr.tsv ADDED
@@ -0,0 +1,44 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ene. enero
2
+ en. enero
3
+ ENE enero
4
+ febr. febrero
5
+ FEB febrero
6
+ feb. febrero
7
+ mzo. marzo
8
+ mzo marzo
9
+ MAR marzo
10
+ mar. marzo
11
+ abr. abril
12
+ ABR abril
13
+ abr abril
14
+ my. mayo
15
+ MAY mayo
16
+ may. mayo
17
+ may mayo
18
+ jun. junio
19
+ JUN junio
20
+ jun junio
21
+ jul. julio
22
+ JUL julio
23
+ jul julio
24
+ AGO agosto
25
+ ag. agosto
26
+ agt. agosto
27
+ ago. agosto
28
+ set. septiembre
29
+ sept. septiembre
30
+ setbre. septiembre
31
+ sept septiembre
32
+ setbre septiembre
33
+ SEP septiembre
34
+ oct. octubre
35
+ OCT octubre
36
+ oct octubre
37
+ nov. noviembre
38
+ novbre. noviembre
39
+ NOV noviembre
40
+ nov noviembre
41
+ novbre noviembre
42
+ dic. diciembre
43
+ dicbre. diciembre
44
+ DIC diciembre
vendor/zonos3/vendor/nemo_text_processing/text_normalization/es/data/dates/year_suffix.tsv ADDED
@@ -0,0 +1,51 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ane antes de nuestra era
2
+ a.n.e. antes de nuestra era
3
+ a. n. e. antes de nuestra era
4
+ aec antes de la era común
5
+ a.e.c antes de la era común
6
+ a. e. c. antes de la era común
7
+ ac antes de cristo
8
+ a.c. antes de cristo
9
+ a. c. antes de cristo
10
+ adc antes de cristo
11
+ a.d.c. antes de cristo
12
+ a. d. c. antes de cristo
13
+ a. de c. antes de cristo
14
+ ajc antes de jesucristo
15
+ a.j.c. antes de jesucristo
16
+ a. j. c. antes de jesucristo
17
+ adjc antes de jesucristo
18
+ a.d.j.c. antes de jesucristo
19
+ a. d. j. c. antes de jesucristo
20
+ a. de j.c. antes de jesucristo
21
+ a. de j. c. antes de jesucristo
22
+ ap antes del presente
23
+ a.p. antes del presente
24
+ a. p. antes del presente
25
+ bp antes del presente
26
+ b.p. antes del presente
27
+ b. p. antes del presente
28
+ ne nuestra era
29
+ n.e. nuestra era
30
+ n. e. nuestra era
31
+ ec era común
32
+ e.c. era común
33
+ e. c. era común
34
+ dc después de cristo
35
+ d.c. después de cristo
36
+ d. c. después de cristo
37
+ ddc después de cristo
38
+ d.d.c después de cristo
39
+ d. d. c. después de cristo
40
+ d. de c. después de cristo
41
+ ddjc después de jesucristo
42
+ d.d.j.c después de jesucristo
43
+ d. d. j. c. después de jesucristo
44
+ d. de j.c. después de jesucristo
45
+ d. de j. c. después de jesucristo
46
+ bc b c
47
+ b.c. b c
48
+ b. c. b c
49
+ ad a d
50
+ a.d. a d
51
+ a. d. a d