| """Find people's names in a text. | |
| pip install onnxruntime tokenizers numpy | |
| python example.py | |
| python example.py letter.txt | |
| """ | |
| import sys | |
| from pathlib import Path | |
| HERE = Path(__file__).resolve().parent | |
| sys.path.insert(0, str(HERE / 'code')) | |
| import runtime | |
| SAMPLE = """Dear Ms Okonjo, | |
| Sam Hill has checked the figures with the Planning team. AR to circulate the draft before the committee meets. | |
| Yours sincerely | |
| Priya Raman | |
| Information Governance Officer | |
| """ | |
| detector = runtime.Detector(HERE) | |
| text = Path(sys.argv[1]).read_text() if len(sys.argv) > 1 else SAMPLE | |
| found = detector.findings(text) | |
| for f in found: | |
| print(f"{f['start']:>6}-{f['end']:<6} {f['score']:.2f} {text[f['start']:f['end']]}") | |
| masked, last = [], 0 | |
| for f in found: | |
| masked += [text[last:f['start']], '[NAME]'] | |
| last = f['end'] | |
| print() | |
| print(''.join(masked) + text[last:]) | |