def main():
words = ['贸易战']
baidu_url = 'https://www.baidu.com/s?wd=%s'
urls = [baidu_url % (word) for word in words]
# make data flow net
insert = Insert(
"insert into test.baidu (id,name ,url,page_rank,page_no)values('{id}','{name}' ,'{url}',{page_rank},{page_no})",
**dbconf)
p = Pipe(
urls,
HttpLoader(),
Branch(get_all_items, join=True),
Branch(get_all_page_url,
HttpLoader(),
get_all_items,
share=False,
join=True,
route_type=HttpResponse),
insert,
)
Pipe(Timer(delay=2, until=p.finished), show_info)
Bot.render('ex_output/baiduspider')
Bot.run()
def main():
Pipe(
Timer(delay=2, max_time=5),
"http://api.coindesk.com/v1/bpi/currentprice.json",
HttpLoader(),
lambda r: r.json['bpi']['USD']['rate_float'],
print,
)
Bot.render('ex_output/simple_bitcoin_price')
Bot.run()
def main():
words = ['贸易战', '世界杯']
baidu_url = 'https://www.baidu.com/s?wd=%s'
urls = [baidu_url % (word) for word in words]
outputfile = aiofile('ex_output/baidu.txt')
Pipe(
urls,
HttpLoader(),
Branch(get_all_items, outputfile),
Branch(get_all_page_url, HttpLoader(), get_all_items, outputfile),
)
#生成流程图
Bot.render('ex_output/baiduspider')
Bot.run()
