def linkAnalysis(self, link):
temp = link.find("a")
address = temp["href"]
brief = temp["title"]
print 'link address is : ', address
# print 'link content is :', brief
timenow = link.find(attrs={"class": "date"}).string
# print timenow
y = timenow[0:4]
m = timenow[5:7]
d = timenow[8:10]
second = timenow[12:20]
timenow = y + '-' + m + '-' + d + ' ' + second
# print timenow
now = time.strptime(timenow, '%Y-%m-%d %H:%M:%S')
if time.mktime(now) < time.mktime(self.updateTime):
return 0
childSoup = self.getChildSoup(address)
try:
newsContent = childSoup.find(attrs={"class": "art_main"})
except:
newsContent = None
stringTemp = ""
if not newsContent:
return 1
for content in newsContent:
stringTemp += str(content)
# print stringTemp
# print searchdict
tags = self.searchTag(stringTemp)
# print 'tags is ',tags
if tags:
newin = Info()
newin.addLink(address, brief)
for tag in tags:
newin.addTag(tag)
print 'in this news find ', tag
self.addInfo(newin)
print '------------------------- one link analysis end ', '------------------'
return 1
def linkAnalysis(self,link):
temp=link.find("a")
address=temp["href"]
brief=temp["title"]
print 'link address is : ',address
# print 'link content is :', brief
timenow=link.find(attrs={"class":"date"}).string
# print timenow
y=timenow[0:4]
m=timenow[5:7]
d=timenow[8:10]
second=timenow[12:20]
timenow=y+'-'+m+'-'+d+' '+second
# print timenow
now= time.strptime(timenow, '%Y-%m-%d %H:%M:%S')
if time.mktime(now)<time.mktime(self.updateTime):
return 0
childSoup=self.getChildSoup(address)
try:
newsContent=childSoup.find(attrs={"class":"art_main"})
except:
newsContent=None
stringTemp=""
if not newsContent:
return 1
for content in newsContent:
stringTemp+=str(content)
# print stringTemp
# print searchdict
tags=self.searchTag(stringTemp)
# print 'tags is ',tags
if tags:
newin=Info()
newin.addLink(address,brief)
for tag in tags:
newin.addTag(tag)
print 'in this news find ',tag
self.addInfo(newin)
print '------------------------- one link analysis end ','------------------'
return 1
def linkAnalysis(self, link):
temp = link.findAll("a")
dat = link.findAll("span")
i = 0
for da in dat:
strtemp = da.string
if strtemp.find(':') == -1:
return
else:
address = 'http://stock.cfi.cn/' + temp[i]["href"]
brief = temp[i].string
t = time.strftime('%Y-%m-%d %H:%M:%S',
time.localtime(time.time()))
truetime = t[0:11] + strtemp + ':59'
now = time.strptime(truetime, '%Y-%m-%d %H:%M:%S')
if time.mktime(now) < time.mktime(self.updateTime):
return
else:
print 'link address is : ', address
childSoup = self.getChildSoup(address)
try:
newsContent = childSoup.find(attrs={"id": "tdcontent"})
except:
newsContent = None
stringTemp = ""
if not newsContent:
return
for content in newsContent:
stringTemp += str(content)
tags = self.searchTag(stringTemp)
if tags:
newin = Info()
newin.addLink(address, brief)
for tag in tags:
newin.addTag(tag)
print 'in this news find ', tag
self.addInfo(newin)
print '------------------------- one link analysis end ', '------------------'
i += 1
def linkAnalysis(self,link):
temp=link.findAll("a")
dat=link.findAll("span")
i=0
for da in dat:
strtemp=da.string
if strtemp.find(':')==-1:
return
else:
address='http://stock.cfi.cn/'+temp[i]["href"]
brief=temp[i].string
t=time.strftime('%Y-%m-%d %H:%M:%S',time.localtime(time.time()))
truetime=t[0:11]+strtemp+':59'
now= time.strptime(truetime, '%Y-%m-%d %H:%M:%S')
if time.mktime(now)<time.mktime(self.updateTime):
return
else :
print 'link address is : ',address
childSoup=self.getChildSoup(address)
try:
newsContent=childSoup.find(attrs={"id":"tdcontent"})
except:
newsContent=None
stringTemp=""
if not newsContent:
return
for content in newsContent:
stringTemp+=str(content)
tags=self.searchTag(stringTemp)
if tags:
newin=Info()
newin.addLink(address,brief)
for tag in tags:
newin.addTag(tag)
print 'in this news find ',tag
self.addInfo(newin)
print '------------------------- one link analysis end ','------------------'
i+=1
def linkAnalysis(self, link):
temp = link.find("a")
address = temp["href"]
brief = temp["title"]
timenow = link.find("span").string
timenow += ":59"
now = time.strptime(timenow, '%Y-%m-%d %H:%M:%S')
# print 'news time:',timenow,'and last update time :',self.updateTime
if time.mktime(now) < time.mktime(self.updateTime):
return 0
print 'link address is : ', address
childSoup = self.getChildSoup(address)
try:
newsContent = childSoup.find(attrs={
"class": "Body",
"id": "ContentBody"
})
except:
newsContent = None
stringTemp = ""
if newsContent == None:
return 1
for content in newsContent:
stringTemp += str(content)
tags = self.searchTag(stringTemp)
if tags:
newin = Info()
newin.addLink(address, brief)
for tag in tags:
newin.addTag(tag)
print 'in this news find ', tag
self.addInfo(newin)
print '------------------------- one link analysis end -------------------------'
return 1
def linkAnalysis(self,link):
temp=link.find("a")
address=temp["href"]
brief=temp["title"]
timenow=link.find("span").string
timenow+=":59"
now= time.strptime(timenow, '%Y-%m-%d %H:%M:%S')
# print 'news time:',timenow,'and last update time :',self.updateTime
if time.mktime(now)<time.mktime(self.updateTime):
return 0
print 'link address is : ',address
childSoup=self.getChildSoup(address)
try:
newsContent=childSoup.find(attrs={"class":"Body","id":"ContentBody"})
except:
newsContent=None
stringTemp=""
if newsContent==None:
return 1
for content in newsContent:
stringTemp+=str(content)
tags=self.searchTag(stringTemp)
if tags:
newin=Info()
newin.addLink(address,brief)
for tag in tags:
newin.addTag(tag)
print 'in this news find ',tag
self.addInfo(newin)
print '------------------------- one link analysis end -------------------------'
return 1