def parseSeason(name, seasonUrl):
data = urllib2.urlopen(seasonUrl)
soup = BeautifulSoup(data.read())
teams = []
divisions = []
divisionSet = set()
currentDivision = None
divisionLevels = dict()
seasonId = None
divisionLevel = 1
season = Season()
season.name = name
season.url = seasonUrl
teamId = 1
divisionId = 1
for link in soup.find_all('a'):
hrefVal = link.get('href')
if hrefVal and hrefVal.startswith('display-schedule.php'):
#print 'Team name is: ' + link.string + ' : ' + hrefVal
if seasonId is None:
seasonId = getSeason(hrefVal)
team = Team()
team.url = hrefVal
team.name = link.string
team.teamId = teamId
teamId += 1
team.players = parseTeam(baseUrl + hrefVal)
if (currentDivision) :
currentDivision.teams.append(team)
elif not hrefVal :
test = link.string.split()
division = test[1]
#print 'Division: ' + division
# Only process
if division not in divisionSet:
divisionSet.add(division)
divisions.append(division)
divisionLevels[division] = divisionLevel
divisionLevel += 1
currentDivision = Division()
currentDivision.level = divisionLevel
divisionId += 1
currentDivision.divisionId =divisionId
currentDivision.name = link.string
season.divisions.append(currentDivision )
season.seasonId = seasonId
return season
def get_team(self, team_type, attrib):
try:
team = Team()
team.code = attrib['%s_team_code' % team_type]
team.name = attrib['%s_fname' % team_type]
team.mlb_id = int(attrib['%s_id' % team_type])
return team
except:
return None
