class Getter(metaclass=GetMetaclass):
def __init__(self):
self.fetcher = Fetcher()
def get_info(self, callback, url):
crawl_func = eval('self.{0}'.format(callback))
return crawl_func(url)
def crawl_weixin(self, url):
result = {}
res = self.fetcher.get(url)
soup = BeautifulSoup(res.content, 'html.parser', from_encoding='utf8')
item_imgs = soup.select('#page-content img[data-src!=""]')
imgs = [item_img['data-src'] for item_img in item_imgs]
title = soup.select('h2#activity-name')[0].text.strip()
result['url'] = url
result['imgs'] = imgs
result['title'] = title
result['content'] = res.content
return result
def crawl_youdao(self, url):
report = {}
res = self.fetcher.get(url)
soup = BeautifulSoup(res.content, 'html.parser', from_encoding='utf8')
item_imgs = soup.select('.post-bd img[src!=""]')
imgs = [item_img['src'] for item_img in item_imgs]
report['url'] = url
report['imgs'] = imgs
report['content'] = res.content
return report
print(f' Source file: {src}')
print(f' Source length: {len(insts)}\n')
# src instruction simulation
while True:
'''
if pc > 58 and cycle % 100 == 0:
input(f"cycle={cycle}, full={issuequeue.full()}, block={branch_block}, pc = {pc}")
'''
#debug(f'CYCLE:{cycle}, PC:{pc}')
# if issue queue is not full, i.e. can fetch and decode
if not issuequeue.full() and not branch_block and pc < len(insts):
# fetch
fetcher.send()
pc, branch_block = fetcher.get(pc, insts)
'''
if branch_block:
print(f"BRANCH BLOCKED! at {pc}")
'''
else:
fetcher.send()
# decode
decoder.send()
decoder.decode()
decoder.recv(fetcher.to_decoder())
# issue queue
issuequeue.enqueue(decoder.to_issuequeue())
