1234567891011121314151617181920212223242526272829303132333435363738 |
- from .DomAnalysis import DomAnalysis
- from .TimeExtractor import TimeExtractor
- __all__ = [
- 'parse_urls',
- 'parser_items',
- 'Parser',
- 'TimeExtractor'
- ]
- def parse_urls(dom: str, **kwargs):
- dom_handler = DomAnalysis(dom=dom, **kwargs)
- return dom_handler.get_urls()
- def parser_items(dom: str, **kwargs):
- dom_handler = DomAnalysis(dom=dom, **kwargs)
- return dom_handler.get_items()
- class Parser:
- def __init__(self):
- self._parse_urls = parse_urls
- self._parser_items = parser_items
- def _site_items(self, dom, **kwargs):
- return self._parser_items(dom, **kwargs)
- def urls(self, isogeny, dom, **kwargs):
- return self._parse_urls(dom, isogeny=isogeny, **kwargs)
- def same_origin(self, dom, host):
- return self._site_items(dom, isogeny=True, host=host)
- def non_origin(self, dom, url):
- return self._site_items(dom, isogeny=False, request_url=url)
|