| Akron | b6efc73 | 2022-03-17 15:03:56 +0100 | [diff] [blame] | 1 | import Cutter |
| 2 | import sys | ||||
| 3 | |||||
| 4 | sys.setrecursionlimit(100000) | ||||
| 5 | |||||
| 6 | cutter = Cutter.Cutter(profile='de') | ||||
| 7 | |||||
| 8 | sent = sys.argv[1] | ||||
| 9 | |||||
| 10 | file = open(sys.argv[2], 'r') | ||||
| 11 | |||||
| 12 | text = file.read() | ||||
| 13 | |||||
| 14 | file.close() | ||||
| 15 | |||||
| 16 | for token in cutter.cut(text): | ||||
| 17 | if token[0]: | ||||
| 18 | print(token[0]) | ||||
| 19 | |||||
| 20 | if sent == "sent": | ||||
| 21 | if token[1].startswith("+EOS"): | ||||
| 22 | print("</eos>") | ||||