__init__.py 939 B

1234567891011121314151617181920212223242526272829303132333435363738
  1. from .DomAnalysis import DomAnalysis
  2. from .TimeExtractor import TimeExtractor
  3. __all__ = [
  4. 'parse_urls',
  5. 'parser_items',
  6. 'Parser',
  7. 'TimeExtractor'
  8. ]
  9. def parse_urls(dom: str, **kwargs):
  10. dom_handler = DomAnalysis(dom=dom, **kwargs)
  11. return dom_handler.get_urls()
  12. def parser_items(dom: str, **kwargs):
  13. dom_handler = DomAnalysis(dom=dom, **kwargs)
  14. return dom_handler.get_items()
  15. class Parser:
  16. def __init__(self):
  17. self._parse_urls = parse_urls
  18. self._parser_items = parser_items
  19. def _site_items(self, dom, **kwargs):
  20. return self._parser_items(dom, **kwargs)
  21. def urls(self, isogeny, dom, **kwargs):
  22. return self._parse_urls(dom, isogeny=isogeny, **kwargs)
  23. def same_origin(self, dom, host):
  24. return self._site_items(dom, isogeny=True, host=host)
  25. def non_origin(self, dom, url):
  26. return self._site_items(dom, isogeny=False, request_url=url)