while commentFeed is not None:
for comment in commentFeed.entry:
commentText = comment.content.text
commentDate = dateStringToSaturday(comment.updated.text)
commentList.append({'week' : commentDate, 'comment' : commentText})
next_link = commentFeed.GetNextLink()
if next_link is None:
commentFeed = None
else:
commentFeed = self.client.GetYouTubeVideoCommentFeed(next_link.href)
except Exception, e:
print e
return commentList
def extractYoutubeCommentsToDB(self, songList):
db = DBController()
for song in songList:
try:
searchVideoName = song['title'] + ' ' + song['artist']
videoID = self.getVideoID(searchVideoName)
comments = self.getComments(videoID)
db.insertCommentToDB(song['id'], comments)
except Exception as e:
print e
continue
if __name__ == '__main__':
extractor = YoutubeCommentsExtractor()
db = DBController()
songList = db.getSongByWeek(lastSaturday())
extractor.extractYoutubeCommentsToDB(songList)
title = cleanTitle(song.contents[3].contents[0].contents[0].contents[2].contents[0].text)
artist = cleanAtrist(song.contents[3].contents[0].contents[0].contents[2].contents[1].text)
item = (title, artist, rank, lastWeek, peak, weeksOnChart)
chart.append(item)
chart.sort(key= lambda song : song[2])
return chart
def extractSalesRankToDB(self, beginDate=datetime.today(), endDate=datetime.today()):
if beginDate < datetime(2007, 1, 1) or endDate > datetime.today():
raise Exception('Invalid input date!')
beginDate = dateToSaturday(beginDate)
endDate = dateToSaturday(endDate)
endDate = endDate - timedelta(days=7) if endDate > datetime.today() else endDate
iterDate = beginDate
db = DBController()
while iterDate <= endDate:
if db.checkSalesRankExistInDB(iterDate):
iterDate = iterDate + timedelta(days = 7)
continue
URL = self.getURL(iterDate)
chart = self.getSalesChartFromURL(URL)
db.insertSalesChartToDB(iterDate, chart)
iterDate = iterDate + timedelta(days = 7)
def getURL(self, date):
return 'http://top40-charts.com/chart.php?cid=27&date=' + date.strftime('%Y-%m-%d')
if __name__ == '__main__':
extractor = SalesChartExtractor()
extractor.extractSalesRankToDB(lastSaturday())