class httpHelper:
logger=None
def __init__(self):
self.logger=AppLog()
def httpExists(self,url):
host, path = urlparse.urlsplit(url)[1:3]
found = None
try:
connection = httplib.HTTPConnection(host) ## Make HTTPConnection Object
connection.request("HEAD", path)
responseOb = connection.getresponse() ## Grab HTTPResponse Object
if responseOb.status in [200,201,204,302]:
found = responseOb.status
except Exception, e:
logvar= e.__class__, e, url
self.logger.logInfo(logvar)
return found
class httpHelper:
logger = None
def __init__(self):
self.logger = AppLog()
def httpExists(self, url):
host, path = urlparse.urlsplit(url)[1:3]
found = None
try:
connection = httplib.HTTPConnection(
host) ## Make HTTPConnection Object
connection.request("HEAD", path)
responseOb = connection.getresponse() ## Grab HTTPResponse Object
if responseOb.status in [200, 201, 204, 302]:
found = responseOb.status
except Exception, e:
logvar = e.__class__, e, url
self.logger.logInfo(logvar)
return found
class UrlHelper:
logger = None
def __init__(self):
self.logger = AppLog()
def isUrl(self, url):
p = re.compile(
'http?://(?:[a-zA-Z]|[0-9]|[$-_@.&+]|[!*\(\),]|(?:%[0-9a-fA-F][0-9a-fA-F]))+'
)
m = p.match(url)
if m:
return True
else:
return False
def getRealURL(self, url):
try:
socket.setdefaulttimeout(30)
realURL = urllib2.urlopen(url).geturl()
return realURL
except urllib2.HTTPError:
self.logger.logInfo(
'Exception in getRealURL method urllib2.HTTPError (helper/urlhelper.py)'
+ str(sys.exc_info()[0]))
return None
except urllib2.URLError:
self.logger.logInfo(
'Exception in getRealURL method urllib2.URLError (helper/urlhelper.py)'
+ str(sys.exc_info()[0]))
return None
except:
return None
def getHeaderInfo(self, soup):
headrDict = {}
title = None
meta = []
description = None
metaTitle = None
try:
title = soup.find('head').title.string
meta = soup.findAll('meta')
except:
self.logger.logInfo(
'getHeaderInfo some meta tags function not tags (helper/urlhelper.py) '
)
for m in meta:
try:
if m['name'].lower() == 'title':
metaTitle = m['content']
if metaTitle:
title = metaTitle
elif m['name'].lower() == 'description':
description = m['content']
except:
self.logger.logInfo(
'getHeaderInfo some meta tags function not tags (helper/urlhelper.py)'
)
headrDict['title'] = title
headrDict['description'] = description
return headrDict
def __init__(self):
self.logger = AppLog()
def __init__(self):
self.logger=AppLog()
#encoding:UTF-8
from dal.urlprovider import Urlprovider
from helper.applog import AppLog
from helper.stringhelper import stringHelper
from helper.urlhelper import UrlHelper
import sys
logger = AppLog()
strObj = stringHelper()
ulrObj = UrlHelper()
urlpObj = Urlprovider()
try:
posts = urlpObj.getPosts()
for post in posts:
postid = post['id']
postText = post['text']
postURLs = strObj.extractURL(postText)
for url in postURLs:
if url:
orgurl = ulrObj.getRealURL(url)
if orgurl:
urlpObj.addURL(url, orgurl, postid)
urlpObj.updatePostUrlExtracted(postid, 1)
except:
class Urlprovider:
logger=None
def __init__(self):
self.logger=AppLog()
def getPosts(self):
try:
dbObj=MySQLHelper()
query="select * from posts where urlextracted=0 order by id;"
result=dbObj.query(query)
return result
except:
self.logger.logInfo('Exception in getPosts method (dal/urlprovider.py)')
def addURL(self,url,orgurl,postid):
try:
dbObj=MySQLHelper()
query="select id,counter from urls where orgurl='"+orgurl+"'"
result=dbObj.query(query)
if len(result) == 0:
tubleData=(url,orgurl,postid)
query="insert into urls (url,orgurl,postid) values (%s,%s,%s)"
result=dbObj.executequery(query,tubleData)
else:
query=''
id=result[0]['id']
counter=result[0]['counter']
if counter is None:
counter=0
counter += 1
query="update urls set counter="+str(counter)+" where id="+str(id)
result=dbObj.query(query)
except:
self.logger.logInfo('Exception in addURL method (dal/urlprovider.py)')
def updatePostUrlExtracted(self,postid,value):
try:
dbObj=MySQLHelper()
query="update posts set urlextracted="+str(value)+" where id="+str(postid)
result=dbObj.query(query)
return result
except:
self.logger.logInfo('Exception in updatePostStatus method (dal/twitterprovider.py)')
def getURLs(self):
try:
dbObj=MySQLHelper()
query="select * from urls where status=0"
result=dbObj.query(query)
return result
except:
self.logger.logInfo('Exception in updatePostStatus method (dal/twitterprovider.py)')
def addUrlHeaderInfo(self,urlid,metatitle,metadescription):
try:
dbObj=MySQLHelper()
tubleData=(metatitle,metadescription)
query="update urls set metatitle=%s,metadescription=%s where id="+str(urlid)
result=dbObj.executequery(query,tubleData)
return result
except:
self.logger.logInfo('Exception in updatePostStatus method (dal/twitterprovider.py)')
def saveUrlIMG(self,urlid,filename):
try:
dbObj=MySQLHelper()
tubleData=(urlid,filename)
query="insert into urlimgs (urlid,filename) values (%s,%s)"
result=dbObj.executequery(query,tubleData)
return result
except:
self.logger.logInfo('Exception in saveUrlIMG method (dal/twitterprovider.py)')
def updateUrlStatus(self,urlid,value):
try:
dbObj=MySQLHelper()
query="update urls set status="+str(value)+" where id="+str(urlid)
result=dbObj.query(query)
return result
except:
self.logger.logInfo('Exception in updateUrlStatus method (dal/twitterprovider.py)')
extension=os.path.splitext(img)[1]
newImgName=str(uuid.uuid1())+str(extension)
im.thumbnail(newSize, Image.ANTIALIAS)
im.save(out_folder+"/"+newImgName,imgformat)
os.remove(imgfile)
saveImgList.append(newImgName)
except:
logger.logInfo("in urldatagrapper.py open images : "+str(sys.exc_info()[0]))
for saveimg in saveImgList:
urlproviderObj.saveUrlIMG(urlid,saveimg)
logger=AppLog()
urlproviderObj=Urlprovider()
urlHelpObj=UrlHelper()
try:
urls=urlproviderObj.getURLs()
for url in urls:
if httpExists(url['orgurl']):
urlid=url['id']
soup=getUrlHTMLsoup(url['orgurl'])
if soup:
headrDict=urlHelpObj.getHeaderInfo(soup)
urlproviderObj.addUrlHeaderInfo(urlid,headrDict['title'], headrDict['description'])
getUrlImgs(url['orgurl'],urlid,soup)
#encoding:UTF-8
from dal.urlprovider import Urlprovider
from helper.applog import AppLog
from helper.stringhelper import stringHelper
from helper.urlhelper import UrlHelper
import sys
logger=AppLog()
strObj=stringHelper()
ulrObj=UrlHelper()
urlpObj=Urlprovider()
try:
posts=urlpObj.getPosts()
for post in posts:
postid=post['id']
postText=post['text']
postURLs=strObj.extractURL(postText)
for url in postURLs:
if url:
orgurl=ulrObj.getRealURL(url)
if orgurl:
urlpObj.addURL(url, orgurl, postid)
urlpObj.updatePostUrlExtracted(postid,1)
except:
class Urlprovider:
logger = None
def __init__(self):
self.logger = AppLog()
def getPosts(self):
try:
dbObj = MySQLHelper()
query = "select * from posts where urlextracted=0 order by id;"
result = dbObj.query(query)
return result
except:
self.logger.logInfo(
'Exception in getPosts method (dal/urlprovider.py)')
def addURL(self, url, orgurl, postid):
try:
dbObj = MySQLHelper()
query = "select id,counter from urls where orgurl='" + orgurl + "'"
result = dbObj.query(query)
if len(result) == 0:
tubleData = (url, orgurl, postid)
query = "insert into urls (url,orgurl,postid) values (%s,%s,%s)"
result = dbObj.executequery(query, tubleData)
else:
query = ''
id = result[0]['id']
counter = result[0]['counter']
if counter is None:
counter = 0
counter += 1
query = "update urls set counter=" + str(
counter) + " where id=" + str(id)
result = dbObj.query(query)
except:
self.logger.logInfo(
'Exception in addURL method (dal/urlprovider.py)')
def updatePostUrlExtracted(self, postid, value):
try:
dbObj = MySQLHelper()
query = "update posts set urlextracted=" + str(
value) + " where id=" + str(postid)
result = dbObj.query(query)
return result
except:
self.logger.logInfo(
'Exception in updatePostStatus method (dal/twitterprovider.py)'
)
def getURLs(self):
try:
dbObj = MySQLHelper()
query = "select * from urls where status=0"
result = dbObj.query(query)
return result
except:
self.logger.logInfo(
'Exception in updatePostStatus method (dal/twitterprovider.py)'
)
def addUrlHeaderInfo(self, urlid, metatitle, metadescription):
try:
dbObj = MySQLHelper()
tubleData = (metatitle, metadescription)
query = "update urls set metatitle=%s,metadescription=%s where id=" + str(
urlid)
result = dbObj.executequery(query, tubleData)
return result
except:
self.logger.logInfo(
'Exception in updatePostStatus method (dal/twitterprovider.py)'
)
def saveUrlIMG(self, urlid, filename):
try:
dbObj = MySQLHelper()
tubleData = (urlid, filename)
query = "insert into urlimgs (urlid,filename) values (%s,%s)"
result = dbObj.executequery(query, tubleData)
return result
except:
self.logger.logInfo(
'Exception in saveUrlIMG method (dal/twitterprovider.py)')
def updateUrlStatus(self, urlid, value):
try:
dbObj = MySQLHelper()
query = "update urls set status=" + str(
value) + " where id=" + str(urlid)
result = dbObj.query(query)
return result
except:
self.logger.logInfo(
'Exception in updateUrlStatus method (dal/twitterprovider.py)')
else:
extension = os.path.splitext(img)[1]
newImgName = str(uuid.uuid1()) + str(extension)
im.thumbnail(newSize, Image.ANTIALIAS)
im.save(out_folder + "/" + newImgName, imgformat)
os.remove(imgfile)
saveImgList.append(newImgName)
except:
logger.logInfo("in urldatagrapper.py open images : " +
str(sys.exc_info()[0]))
for saveimg in saveImgList:
urlproviderObj.saveUrlIMG(urlid, saveimg)
logger = AppLog()
urlproviderObj = Urlprovider()
urlHelpObj = UrlHelper()
try:
urls = urlproviderObj.getURLs()
for url in urls:
if httpExists(url['orgurl']):
urlid = url['id']
soup = getUrlHTMLsoup(url['orgurl'])
if soup:
headrDict = urlHelpObj.getHeaderInfo(soup)
urlproviderObj.addUrlHeaderInfo(urlid, headrDict['title'],
headrDict['description'])