Coverage for src / local_deep_research / research_library / downloaders / extraction / base.py: 100%

1""" 

2Base class for HTML content extractors. 

3""" 

4 

5from abc import ABC, abstractmethod 

6from typing import Optional 

7 

8 

9class BaseExtractor(ABC): 

10 """Abstract base for content extraction strategies.""" 

11 

12 @abstractmethod 

13 def extract(self, html: str) -> Optional[str]: 

14 """Extract main content text from HTML. 

15 

16 Args: 

17 html: Raw or partially cleaned HTML string. 

18 

19 Returns: 

20 Extracted plain text, or None if extraction yielded nothing. 

21 """