cjw123ztt
commented
3 years ago def get_fund_info_ex2(code): url = "http://fundf10.eastmoney.com/ccmx{}.html".format(code) print(url) opt = webdriver.ChromeOptions() opt.set_headless() driver = webdriver.Chrome(options=opt) driver.maximize_window() driver.get(url) driver.implicitly_wait(5) day = datetime.date.today() today = '%s' % day
with open('jijin1.html', 'w', encoding='utf-8') as f:
f.write(driver.page_source)
time.sleep(1)
file = open('jijin1.html', 'r', encoding='utf-8')
soup = BeautifulSoup(file, 'lxml')
# print(soup)
try:
fund = soup.select('#bodydiv > div > div > div.basic-new > div.bs_jz > div.col-left > h4 > a')[0].get_text()
scale = soup.select('#bodydiv > div > div.r_cont > div.basic-new > div.bs_gl > p > label > span')[
2].get_text().strip().split()[0]
tables = soup.select('#cctable > div > div > table')
#trs = tables[0].select('tbody > tr')
index = -1
for table in tables:
#
index = index + 1
#
trs = tables[index].select('tbody > tr') # index
# print(table[1])
# print(trs)
#for tr in trs:
#print(tr)
#code = tr.select('td > a')[0].get_text()
#name = tr.select('td > a')[1].get_text()
# price = tr.select('td > span')[0].get_text()#error
#print(code)
#print(name)
# print(price)
#
QDateH4 = soup.select('#cctable > div > div > h4')[index] # index
QDateLabel = QDateH4.select('label')[1]
QDate = QDateLabel.select('font')[0].get_text()
#print(QDate)
#
for tr in trs:
#print(tr)
code = tr.select('td > a')[0].get_text()
name = tr.select('td > a')[1].get_text()
"""
price = tr.select('td > span')[0].get_text()
try:
round(float(price), 2)
except ValueError:
price = 0
"""
# position
if index == 0:
ratio = tr.select('td.tor')[2].get_text()
position = tr.select('td.tor')[3].get_text()
else:
ratio = tr.select('td.tor')[0].get_text()
position = tr.select('td.tor')[1].get_text()
"""
market = float(num.replace(',', '')) * float(price)
"""
data = {
'crawl_date': QDate,
'code': code,
'fund': fund.split(' (')[0],
'scale': scale,
'name': name,
# 'price': round(float(price), 2),
'position': round(float(position.replace(',', '')), 2),
# 'market_value': round(market, 2),
'ratio': ratio,
'fund_url': url
}
print(data)
# fund_data.insert(data)
except IndexError:
info = {
'url': url
}
# fund_no_data.insert(info)
https://blog.csdn.net/chenlibao0823/article/details/81989002