if answerMan:
dic['answerMan'] = answerMan.encode('utf8')
if answerID:
dic['answerID'] = answerID.encode('utf8')
if answerContent:
dic['answerContent'] = answerContent.encode('utf8')
import reverseDay
dic['spiderTime'] = reverseDay.splitTimeStr(startTime)
except Exception, e:
print e
print 'eror at dic[]'
try:
if answerContent != 'N/A':
if answerTime != 'N/A':
dic['answerTime'] = dealTime.dealTime(answerTime, startTime.split('-')[0])
else:
import reverseDay
dic['answerTime'] = reverseDay.splitTimeStr(startTime)
else:
dic['answerTime'] = 'N/A'
if askTime != 'N/A':
dic['askTime'] = dealTime.dealTime(askTime, startTime.split('-')[0])
else:
import reverseDay
dic['askTime'] = reverseDay.splitTimeStr(startTime)
except Exception, e:
print 'error at time'
print e
print '数据录入mongo...'
dic['answerMan'] = answerMan.encode('utf8')
if answerID:
dic['answerID'] = answerID.encode('utf8')
if answerContent:
dic['answerContent'] = answerContent.encode('utf8')
import reverseDay
dic['spiderTime'] = reverseDay.splitTimeStr(startTime)
except Exception, e:
print e
print 'eror at dic[]'
try:
if answerContent != 'N/A':
if answerTime != 'N/A':
dic['answerTime'] = dealTime.dealTime(
answerTime,
startTime.split('-')[0])
else:
import reverseDay
dic['answerTime'] = reverseDay.splitTimeStr(startTime)
else:
dic['answerTime'] = 'N/A'
if askTime != 'N/A':
dic['askTime'] = dealTime.dealTime(askTime,
startTime.split('-')[0])
else:
import reverseDay
dic['askTime'] = reverseDay.splitTimeStr(startTime)
except Exception, e:
print 'error at time'
print e
askMan = askContent = askTime = answerID = answerMan = answerContent = answerTime = 'N/A'
try:
ask = item.find('div', {'class': 'm_feed_detail m_qa_detail'})
if ask == None:
ask = item.find('div', {'class': 'm_feed_detail'})
try:
# 提问人
askMan = ask.find('div', {'class': 'm_feed_face'}).p.text.strip()
# 提问内容
askContent = ask.find('div', {'class': 'm_feed_txt'}).contents[2].strip()
# 上市公司代码
patt = re.compile(r"\((.*?)\)", re.I | re.X)
answerID = patt.findall(ask.find('div', {'class': 'm_feed_txt'}).a.text.strip())[0]
# 提问时间
askTime = ask.find('div', {'class': 'm_feed_from'}).span.text.strip(r'"()"')
askTime = dealTime.dealTime(askTime, '2016')
except Exception, e:
print '提问出错'
print e
try:
answer = item.find('div', {'class':'m_feed_detail m_qa'})
#上市公司
answerMan = answer.find('div', {'class':'m_feed_face'}).p.text.strip()
#上市公司回答内容
answerContent = answer.find('div', {'class':'m_feed_txt'}).text.strip()
#上市公司回答时间
answerTime = answer.find('div',{'class':'m_feed_from'}).span.text.strip(r'"()"')
print askTime
answerTime = dealTime.dealTime(answerTime, '2016')
except Exception, e:
}).p.text.strip()
# 提问内容
askContent = ask.find('div', {
'class': 'm_feed_txt'
}).contents[2].strip()
# 上市公司代码
patt = re.compile(r"\((.*?)\)", re.I | re.X)
answerID = patt.findall(
ask.find('div', {
'class': 'm_feed_txt'
}).a.text.strip())[0]
# 提问时间
askTime = ask.find('div', {
'class': 'm_feed_from'
}).span.text.strip(r'"()"')
askTime = dealTime.dealTime(askTime, '2016')
except Exception, e:
print '提问出错'
print e
try:
answer = item.find('div', {'class': 'm_feed_detail m_qa'})
#上市公司
answerMan = answer.find('div', {
'class': 'm_feed_face'
}).p.text.strip()
#上市公司回答内容
answerContent = answer.find('div', {
'class': 'm_feed_txt'
}).text.strip()
#上市公司回答时间