def getSongListFromXML(XMLFileDir):
root = ET.parse(XMLFileDir).getroot()
songDict = {}
songList = []
for song in root.iter('song'):
title = cleanTitle(song.attrib['title'])
artist = cleanAtrist(song.attrib['artist'])
key = title + '_' + artist
if key not in songDict:
songDict[key] = True
songList.append((title, artist))
return songList
def getTop50ChartFromURL(self, url):
page = urllib2.urlopen(url)
soup = BeautifulSoup(page.read())
songs = soup.findAll('tr')
chart = []
for song in songs:
rank = int(song.contents[0].text.strip())
title = cleanTitle(song.contents[1].text)
artist = cleanAtrist(song.contents[2].text)
item = (rank, title, artist)
chart.append(item)
chart.sort(key=lambda song : song[0])
return chart
def getSalesChartFromURL(self, URL):
page = urllib2.urlopen(URL)
soup = BeautifulSoup(page.read())
songs = soup.findAll(attrs = {'class' : 'latc_song'})
chart = []
for song in songs:
rank = stringToInt(song.contents[0].text)
lastWeek = song.contents[2].text
if lastWeek == 'New':
lastWeek = None
elif lastWeek == 'RE':
lastWeek = rank
else:
lastWeek = stringToInt(lastWeek)
peak = stringToInt(song.contents[6].text)
weeksOnChart = stringToInt(song.contents[7].text)
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
