mirror of
				https://github.com/explosion/spaCy.git
				synced 2025-10-31 07:57:35 +03:00 
			
		
		
		
	
		
			
				
	
	
		
			31 lines
		
	
	
		
			682 B
		
	
	
	
		
			Python
		
	
	
	
	
	
			
		
		
	
	
			31 lines
		
	
	
		
			682 B
		
	
	
	
		
			Python
		
	
	
	
	
	
| from __future__ import unicode_literals
 | |
| 
 | |
| import pytest
 | |
| 
 | |
| def test_hyphen(en_tokenizer):
 | |
|     tokens = en_tokenizer('best-known')
 | |
|     assert len(tokens) == 3
 | |
| 
 | |
| 
 | |
| def test_period(en_tokenizer):
 | |
|     tokens = en_tokenizer('best.Known')
 | |
|     assert len(tokens) == 3
 | |
|     tokens = en_tokenizer('zombo.com')
 | |
|     assert len(tokens) == 1
 | |
| 
 | |
| 
 | |
| def test_ellipsis(en_tokenizer):
 | |
|     tokens = en_tokenizer('best...Known')
 | |
|     assert len(tokens) == 3
 | |
|     tokens = en_tokenizer('best...known')
 | |
|     assert len(tokens) == 3
 | |
| 
 | |
| 
 | |
| def test_email(en_tokenizer):
 | |
|     tokens = en_tokenizer('hello@example.com')
 | |
|     assert len(tokens) == 1
 | |
|     tokens = en_tokenizer('hi+there@gmail.it')
 | |
|     assert len(tokens) == 1
 | |
| 
 | |
| 
 |