A screenscraper for the australian open scores page

ausopen.py

#! /usr/bin/env python
# A Screenscraper for the Australian Open scores page.
# P.C. Shyamshankar
# January 2009
#
# Can handle ongoing as well as completed matches, but you'll have to change the
# urls yourself.
#
# Note : Match numbers do not stay fixed! As matches finish and are pulled out
# of the list, they affect all below them (-1) and their own, of course.
#
# Important matches are _usually_ at the top, and stay there for sometime, so
# they should work out in most cases.

from BeautifulSoup import BeautifulSoup
from urllib import urlopen
from sys import argv

try :
    match_num = int(argv[1])
except :
    match_num = -1

# Matches in progress
url = urlopen('http://www.australianopen.com/en_AU/scores/index2.html')

# Completed matches
# url = urlopen('http://www.australianopen.com/en_AU/scores/cmatch/12ms.html')

soup = BeautifulSoup(url)

tables = soup.findAll('table', attrs = {'summary' : True})

for index, table in enumerate(tables) :
    if match_num == -1 :
        pass
    elif match_num == index :
        pass
    else :
        continue

    cells = [i.contents for i in table.findAll('td', attrs = {'valign' : True})]
    players = [i.contents[0] for i in table.findAll('a')]

    title = cells[0][0].find('td', attrs = {'class' : 'medBold'}).contents[0]
    title = title.replace(u' ', ' ')

    img = table.findAll('img')[0]
    if img.get('alt') == u'Winner' :
        status = 'completed'
        print "Match (%d) completed : %s" % (index, title)
        template = "%40s %1s %4s %4s %4s %4s %4s"
        print template % ("Player", "W", "S1", "S2", "S3", "S4", "S5")
    else :
        status = 'progress'
        print "Match (%d) in progress : %s" % (index, title)
        template = "%40s %1s %4s %4s %4s %4s %4s %4s"
        print template % ("Player", "S", "G", "S1", "S2", "S3", "S4", "S5")

    if cells[2] :
        serving = 0
    else :
        serving = 1

    if u'Doubles' in title :
        if status == 'completed' :
            players[serving * 2] = players[serving * 2].contents[0]
            players[serving * 2 + 1] = players[serving * 2 + 1].contents[0]
        players[0] = players[0] + '/' + players[1]
        players[1] = players[2] + '/' + players[3]
    else :

        players = []
        players.append(
            [cells[1][1].contents[0], cells[1][2].replace(u' ', u' ')]
        )
        players.append(
            [cells[10][1].contents[0], cells[10][2].replace(u' ', u' ')]
        )


        if status == 'completed' :
            players[serving] =  players[serving][1] + ' ' + players[serving][0].contents[0]
            players[1 - serving] = players[1 - serving][1] + ' ' +  players[1 - serving][0]
        else :
            for i in 0, 1 :
                players[i] = players[i][1] + ' ' + players[i][0]

    player0 = []
    player1 = []

    if status == 'progress' :
        if serving == 0 :
            player0.append(cells[3][1].contents[0])
            player1.append(cells[9 + 3][0])
        elif serving == 1 :
            player0.append(cells[3][0])
            player1.append(cells[9 + 3][1].contents[0])

    for i in range(4, 9) :
        try :
            player0.append(
                cells[i][0] +
                ('(' + cells[i][1].contents[0] + ')'
                 if cells[i][1].contents else ''))
            player1.append(
                cells[9 + i][0] +
                ('(' + cells[9 + i][1].contents[0] + ')'
                 if cells[9 + i][1].contents else ''))
        except :
            break

    player_0_string = template % (
        (players[0],) +
        (("*" if serving == 0 else " "),) +
        tuple(player0) +
        tuple(["-"] * ((5 if status =='completed' else 6) - len(player0)))
    )
    player_1_string = template % (
        (players[1],) +
        (("*" if serving == 1 else " "),) +
        tuple(player1) +
        tuple(["-"] * ((5 if status == 'completed' else 6) - len(player1)))
    )

    player_0_string = player_0_string.replace(u' ', u'%4s' % u'-')
    player_1_string = player_1_string.replace(u' ', u'%4s' % u'-')

    print player_0_string
    print player_1_string
    print