sentences-golden.json

tests/fixtures/sentences-golden.json

1. [
2. {"input": "Hello there. How are you? I'm fine!", "sentences": [
3. {"text": "Hello there.", "offset": 0},
4. {"text": "How are you?", "offset": 13},
5. {"text": "I'm fine!", "offset": 26}
6. ]},
7. {"input": "Dr. Smith met Mr. Jones at 3 p.m. on Oct. 5th. They talked.", "sentences": [
8. {"text": "Dr. Smith met Mr. Jones at 3 p.m. on Oct. 5th.", "offset": 0},
9. {"text": "They talked.", "offset": 47}
10. ]},
11. {"input": "Growth was 3.5 percent. That beat the forecast.", "sentences": [
12. {"text": "Growth was 3.5 percent.", "offset": 0},
13. {"text": "That beat the forecast.", "offset": 24}
14. ]},
15. {"input": "He said \"stop.\" Then he left.", "sentences": [
16. {"text": "He said \"stop.\"", "offset": 0},
17. {"text": "Then he left.", "offset": 16}
18. ]},
19. {"input": "She asked (quietly.) Nobody answered.", "sentences": [
20. {"text": "She asked (quietly.)", "offset": 0},
21. {"text": "Nobody answered.", "offset": 21}
22. ]},
23. {"input": "Um... yeah I think so. Right.", "sentences": [
24. {"text": "Um... yeah I think so.", "offset": 0},
25. {"text": "Right.", "offset": 23}
26. ]},
27. {"input": "Wait... Okay. Go on.", "sentences": [
28. {"text": "Wait...", "offset": 0},
29. {"text": "Okay.", "offset": 8},
30. {"text": "Go on.", "offset": 14}
31. ]},
32. {"input": "The J. R. R. Tolkien box set arrived. It's heavy.", "sentences": [
33. {"text": "The J. R. R. Tolkien box set arrived.", "offset": 0},
34. {"text": "It's heavy.", "offset": 38}
35. ]},
36. {"input": "Acme Inc. and Widgets Ltd. merged, etc. and so on. Done.", "sentences": [
37. {"text": "Acme Inc. and Widgets Ltd. merged, etc. and so on.", "offset": 0},
38. {"text": "Done.", "offset": 51}
39. ]},
40. {"input": "No trailing punctuation here", "sentences": [
41. {"text": "No trailing punctuation here", "offset": 0}
42. ]},
43. {"input": "Really?! Yes. ", "sentences": [
44. {"text": "Really?!", "offset": 0},
45. {"text": "Yes.", "offset": 9}
46. ]},
47. {"input": " leading space. and lower case next.", "sentences": [
48. {"text": "leading space.", "offset": 3},
49. {"text": "and lower case next.", "offset": 18}
50. ]},
51. {"input": "Fig. 2 shows it. Vol. 3 too.", "sentences": [
52. {"text": "Fig. 2 shows it.", "offset": 0},
53. {"text": "Vol. 3 too.", "offset": 17}
54. ]},
55. {"input": "U.S. policy changed. The U.K. followed.", "sentences": [
56. {"text": "U.S. policy changed.", "offset": 0},
57. {"text": "The U.K. followed.", "offset": 21}
58. ]},
59. {"input": "", "sentences": []},
60. {"input": "One.Two. Three", "sentences": [
61. {"text": "One.Two.", "offset": 0},
62. {"text": "Three", "offset": 9}
63. ]},
64. {"input": "It's 10 a.m. Let's go.", "sentences": [
65. {"text": "It's 10 a.m. Let's go.", "offset": 0}
66. ]},
67. {"input": "End with ellipsis...", "sentences": [
68. {"text": "End with ellipsis...", "offset": 0}
69. ]},
70. {"input": "Quote at end.” Next sentence.", "sentences": [
71. {"text": "Quote at end.”", "offset": 0},
72. {"text": "Next sentence.", "offset": 15}
73. ]}
74. ]