def parse(self, reg, range_postfix):
bodies = []
start, end = tuple(range_postfix)
for i in tqdm(range(start, end), desc=f'{self.name}'):
url = self.url.link(i)
text = Text.create_by_url(url)
bodies_sub = text.findall(reg)
bodies.extend(bodies_sub)
urls = self.url.concat_site_with_bodies(bodies)
return urls
def __init__(self, site, url):
self.url = url
self.site = site
self.text = Text.create_by_url(url)
def urls(self):
reg = r'<div class="content-list_item" data-reactid=".*?"><div class="content-list_item-info" data-reactid=".*?"><div class="media-preview" data-reactid=".*?"><a class="media-preview_img-wrap" href="(.*?)"'
text = Text.create_by_url(self.url.url)
urls = text.findall(reg)
urls = self.url.concat_site_with_bodies(urls)
return urls
