#this is to extract the state code '##'
state = cityStateText[-2:]
return state
def getBikeName(self, soup):
bikeNameDiv = soup.find_all("div", class_='grid_8 margin-top10')
bikeName = self.getStringFromSoupElement(bikeNameDiv[0].h1)
bikeName = bikeName.replace('Used, ', '')
return bikeName
def unicodeToString(self, unicode):
if unicode is None:
return unicode
else:
return unicode.encode('ascii','ignore')
def getStringFromSoupElement(self, element):
str = self.unicodeToString(element.get_text())
if str is not None:
str = " ".join(str.split())
return str
if __name__ == '__main__':
instance = Extractor()
daoInstance = Dao()
readerHelper = Helper()
for fileName in readerHelper.getAllHtmlFileNames():
dict = instance.extract(fileName)
if bool(dict):
daoInstance.populateAndExecute(dict)
