def fetchPlayer(line):
def playerError(msg):
print >> Globals.errlog, "Player %s: %s" % (line.encode('utf-8'), msg.encode('utf-8'))
lineWithoutSpaces = ''.join(line.split())
ll = line.lower()
if '{{fs player' in ll or \
'{{football squad player' in ll or \
'{{fs2 player' in ll:
unlinkedline = wikiutils.unlink_wiki(line)
columns = [s.strip() for s in unlinkedline.replace('{', '').replace('}', '').split('|')]
number = None
nationality = None
pos = None
name = None
firstname = None
lastname = None
for column in columns:
if '=' in column:
try:
k, v = wikiutils.getKeyValue(column)
except ValueError:
playerError("Couldn't parse player information column: %s" % column)
continue
if k == 'no':
try:
number = int(v)
except (UnicodeEncodeError, ValueError):
pass # usually dash as a player number
elif k == 'nat':
nationality = v
elif k == 'pos':
pos = v
elif k == 'name':
name = wikiutils.unlinkify(v)[0]
elif k == 'first':
firstname = wikiutils.unlinkify(v)[0]
elif k == 'last':
lastname = wikiutils.unlinkify(v)[0]
if not name and firstname and lastname:
name = firstname + ' ' + lastname
if not number:
number = 0
if not nationality:
nationality = 'NA'
if nationality and pos and name:
return soccer.Player(name, number, pos, nationality)
return None
def parseTeam(team, rvtext, mayGetTemplates):
players = []
teamposition = None
kit = [soccer.Kit(), soccer.Kit()]
finishedReadingPlayers = False
lookForSquadTemplate = False
def teamError(msg):
print >> Globals.errlog, "Team %s: %s" % (team.encode('utf-8'), msg.encode('utf-8'))
for line in rvtext.split('\n'):
lineWithoutSpaces = ''.join(line.split())
if not finishedReadingPlayers:
p = playerparser.fetchPlayer(line)
if p:
players.append(p)
else:
heading = wikiutils.getHeading(line)
if heading:
if mayGetTemplates and 'current squad' in heading.lower() or ('first' in heading.lower() and 'squad' in heading.lower()):
lookForSquadTemplate = True
else:
lookForSquadTemplate = False
elif lookForSquadTemplate:
t = wikiutils.getTemplate(line)
if t:
text = wikiutils.getPage('Template:' + t)
if text:
players = playerparser.fetchPlayers(text)
if len(players) > 15:
finishedReadingPlayers = True
if playerparser.endOfPlayerList(line):
finishedReadingPlayers = True
if lineWithoutSpaces.startswith("|position="):
# this seems to usually be either this or last season's position
if not ('promoted' in lineWithoutSpaces.lower() or 'relegated' in lineWithoutSpaces.lower()):
tp = wikiutils.getNumberKeyValue(line)
if tp:
teamposition = tp
kitresults = kitinfo_re.findall(line)
for kitresult in kitresults:
columns = [x.strip() for x in line.split('|') if 'body' in x or 'shorts' in x or 'socks' in x or 'pattern_b' in x]
# apparently, n may be more than 1 if more than one kit part is on a line
for c in columns:
try:
k, v = wikiutils.getKeyValue(c)
except:
continue
if k.startswith('body'):
k = k[4:]
if not k: continue
n = int(k[0]) - 1
if n == 0 or n == 1:
kit[n].bodycolor = getColorValue(v)
elif k.startswith('shorts'):
k = k[6:]
if not k: continue
n = int(k[0]) - 1
if n == 0 or n == 1:
kit[n].shortscolor = getColorValue(v)
elif k.startswith('socks'):
k = k[5:]
if not k: continue
n = int(k[0]) - 1
if n == 0 or n == 1:
kit[n].sockscolor = getColorValue(v)
elif k.startswith('pattern_b') and k != 'pattern_blue':
k = k[9:]
if not k: continue
n = int(k[0]) - 1
# TODO: body type, second color
if len(players) < 15:
return None
if not teamposition:
teamposition = 0
return soccer.Team(team, kit, teamposition, players)
def getLeagueData(rvtext, leaguedata):
season = ''
relegationleagues = dict()
numteams = 0
levelnum = 0
divisions = 0
class InfoboxState:
Outside = 0
Entered = 1
RelegationLeagues = 2
NumTeams = 3
NumLevel = 4
Season = 5
ibs = InfoboxState.Outside
for line in rvtext.split('\n'):
lineWithoutSpaces = ''.join(line.split())
if not season and lineWithoutSpaces.startswith("|current="):
k, v = wikiutils.getKeyValue(line)
competition, competitionlink = wikiutils.unlinkify(v)
if competitionlink:
season = competitionlink
if not divisions and (lineWithoutSpaces.startswith("|divisions=") or lineWithoutSpaces.startswith("|division=")):
tp = wikiutils.getNumberKeyValue(line)
if tp:
divisions = tp
if not levelnum and (lineWithoutSpaces.startswith("|levels=") or lineWithoutSpaces.startswith("|level=")):
tp = wikiutils.getNumberKeyValue(line)
if tp:
levelnum = tp
if len(relegationleagues) == 0 and lineWithoutSpaces.startswith("|relegation="):
k, v = wikiutils.getKeyValue(line)
candidates = [wikiutils.unlinkify(x.strip()) for x in br_re.split(v)]
for cn, cl in candidates:
if cl:
relegationleagues[cl] = cl
if not numteams and lineWithoutSpaces.startswith('|teams='):
numteams = wikiutils.getNumberKeyValue(line)
if ibs == InfoboxState.Outside and lineWithoutSpaces.startswith('{|class="infoboxfootball"'):
# e.g. Regionalliga_Nord
ibs = InfoboxState.Entered
elif ibs != InfoboxState.Outside:
if lineWithoutSpaces and lineWithoutSpaces[0] == '|':
text = '|'.join(line.split('|')[2:])
if not text and lineWithoutSpaces[0:2] == '|}':
ibs = InfoboxState.Outside
break
elif text:
t, link = wikiutils.unlinkify(text)
tl = t.lower()
if 'background' in line:
if 'relegation' in tl:
ibs = InfoboxState.RelegationLeagues
elif 'number of clubs' in tl:
ibs = InfoboxState.NumTeams
elif 'level' in tl:
ibs = InfoboxState.NumLevel
elif 'current season' in tl:
ibs = InfoboxState.Season
else:
ibs = InfoboxState.Entered
else:
if ibs == InfoboxState.RelegationLeagues:
if not link:
ibs = InfoboxState.Entered
else:
relegationleagues[link] = link
elif ibs == InfoboxState.NumTeams:
pos = re.findall(r'\d+', t)
if len(pos) >= 1:
numteams = int(pos[0])
elif ibs == InfoboxState.NumLevel:
pos = re.findall(r'\d+', t)
if len(pos) >= 1:
levelnum = int(pos[0])
elif ibs == InfoboxState.Season:
if not link:
ibs = InfoboxState.Entered
else:
season = link
if not leaguedata.season:
leaguedata.season = season
if not leaguedata.relegationleagues:
leaguedata.relegationleagues = relegationleagues
if not leaguedata.numteams:
leaguedata.numteams = numteams
if not leaguedata.divisions:
leaguedata.divisions = divisions
if not leaguedata.levelnum:
leaguedata.levelnum = levelnum