def extract(gender, year, div, org, game, path):
if not exists(join(path, 'Box Score - All (Parsed).csv')):
print("Missing: {}".format(path), file=sys.stderr)
return
try:
box = loadTable(join(path, 'Box Score - All (Parsed).csv'))
row = [r for r in box[1:] if r[3] == org][-1]
isHome = row[3] == box[-1][3]
home = "Home" if isHome else "Away"
opponent = box[1 if isHome else -1][3]
info = loadTable(join(path, 'Info.csv'))
location, attendance = "", ""
for (c, v) in zip(info[0], info[1]):
if c == "Location:":
location = v
elif c == "Attendance:":
attendance = v
row = [gender, year, div] + row[:2] + [org, opponent, home] + row[6:] + [location, attendance]
output.append(row)
except:
print("Exception: {}".format(path), file=sys.stderr)
def loadDateTime():
reDateTime = re.compile(r"^([^ ]+)(?: +(.+))?$")
infoTable = loadTable(infofile)
m = reDateTime.match(infoTable[1][0])
return [withDefault(m.group(1), default=""), withDefault(m.group(2), default="")]
def loadPlayers(team):
filename = os.path.join(
os.path.split(datafile)[0],
period + " - {}.csv".format(team)
)
if not os.path.exists(filename):
filename = os.path.join(
os.path.split(datafile)[0],
"Box Score - {}.csv".format(team)
)
boxScoreTable = loadTable(filename)
return [(r[0], r[2] != "") for r in boxScoreTable[1:-2]] + [("Team", None)]
def parse(datafile, period, infofile):
def computeSeconds(t):
m, s = tuple(int(v) for v in t.split(":"))
return m * 60 + s
def computeTime(t):
(x, y) = divmod(t, 60)
return "{}:{:02d}".format(x, y)
pbpTable = loadTable(datafile)
rePBP = re.compile(
r"^([0-9]+|[A-Z.`'-?\xbf]*(?:(?:, *| +)[A-Z.`'-?\xbf]*)*)" +
r"(?:(?:^| +)(made|missed))?" +
r"(?:(?:^| +)((?:[A-Z][a-z]*|[0-9]+)(?: +[A-Z][a-z]*|[0-9]+)*))$",
re.U
)
teams = [pbpTable[0][1], pbpTable[0][3]]
def loadDateTime():
reDateTime = re.compile(r"^([^ ]+)(?: +(.+))?$")
infoTable = loadTable(infofile)
m = reDateTime.match(infoTable[1][0])
return [withDefault(m.group(1), default=""), withDefault(m.group(2), default="")]
def loadPlayers(team):
filename = os.path.join(
os.path.split(datafile)[0],
period + " - {}.csv".format(team)
)
if not os.path.exists(filename):
filename = os.path.join(
os.path.split(datafile)[0],
"Box Score - {}.csv".format(team)
)
boxScoreTable = loadTable(filename)
return [(r[0], r[2] != "") for r in boxScoreTable[1:-2]] + [("Team", None)]
dateTime = loadDateTime()
players = [loadPlayers(team) for team in teams]
def buildTable():
table = [["Date", "Time", "Period", "Time Left", "Score", "Team", "Player", "Status", "Action"]]
for r in pbpTable[1:]:
if len(r) < 4:
break
flag = r[1] != ""
m = rePBP.match(r[1 if flag else 3])
if m is None:
continue
table.append(
dateTime + \
[period] + \
[
r[0],
r[2],
teams[0 if flag else 1],
best(
[p[0] for p in players[0 if flag else 1]],
withDefault(m.group(1), "").replace(",", ", ")
),
withDefault(m.group(2), ""),
withDefault(m.group(3), "")
]
)
return table
table = buildTable()
def getPeriodLength(table):
maxTime = table[1][3]
n = int(maxTime.split(':')[0])
if 10 < n <= 20:
return "20:00"
if 5 < n <= 10:
return "10:00"
elif 0 < n <= 5:
return "5:00"
periodLength = getPeriodLength(table)
def addTimer(table):
table[0].append("Shot Clock")
lastTime = periodLength
for r in table[1:]:
time = r[3]
if r[-1] == "Turnover":
lastTime = time
if r[-1] in attacks or r[-1] in defends:
diff = min(30, computeSeconds(lastTime) - computeSeconds(time))
r.append(str(30 - diff) if diff > 0 else "")
lastTime = time
else:
r.append("")
addTimer(table)
def addLineup(table):
initPlayers = (
set(p[0] for p in players[0] if p[1]),
set(p[0] for p in players[1] if p[1])
)
for r in reversed(table[1:]):
player = r[6]
i = teams.index(r[5])
if player == "Team":
continue
if r[-2] == "Enters Game":
if player in initPlayers[i]:
initPlayers[i].remove(player)
elif r[-2] == "Leaves Game":
initPlayers[i].add(player)
table[0].extend(["Lineup", "Lineup Time"])
lineupTimes = [{}, {}]
lastLineup, lastLineupTime = [None, None], [periodLength, periodLength]
for r in table[1:]:
player = r[6]
i = teams.index(r[5])
if player != "Team":
if r[-2] == "Leaves Game":
if player in initPlayers[i]:
initPlayers[i].remove(player)
else:
initPlayers[i].add(player)
lineup = "; ".join(sorted(initPlayers[i]))
r.append(lineup)
lineupTime = r[3]
if lastLineup[i] == None:
lastLineup[i] = lineup
lineupTimes[i][lastLineup[i]] = lineupTimes[i].setdefault(lastLineup[i], 0) + computeSeconds(lastLineupTime[i]) - computeSeconds(lineupTime)
r.append(computeTime(lineupTimes[i].setdefault(lineup, 0)))
lastLineup[i], lastLineupTime[i] = lineup, lineupTime
addLineup(table)
return table
def crawl(sport, year, division, org, game, url, neutral=False):
global data
data = data.format(sport, year, division)
gamename = game.replace('/', '.')
def readFlag(flag):
if not os.path.exists(os.path.join(data, org, gamename)):
os.mkdir(os.path.join(data, org, gamename))
return os.path.exists(os.path.join(data, org, gamename, flag))
def setFlag(flag):
with open(os.path.join(data, org, gamename, flag), 'w') as f:
pass
if neutral and not readFlag(".neutral"):
setFlag(".neutral")
filename = os.path.join(data, org, gamename, "{}.csv")
if not readFlag(".done"):
try:
gamelink = urljoin(domain, url)
log("{} {} {} {} {} {}".format(sport, year, division, org, game, dumpURL(gamelink)))
gs = parseURL(gamelink)
sleep(2)
gamescore = None
gameinfo = None
periods = []
teams = []
nextPeriod = 0
for table in gs.select("div.header_menu a"):
if (
table["href"] == "#" or
not (
table["href"].startswith("/game/box_score") or
table["href"].startswith("/game/play_by_play")
)
):
continue
tablelink = urljoin(domain, table["href"])
print2("{} \033[4m{}\033[0m".format(table.text.strip(), tablelink))
ts = parseURL(tablelink)
if gamescore is None:
gamescore = parseTable(ts.select("table:nth-of-type(1)")[0])
dumpTable(
gamescore,
filename.format("Score")
)
if gameinfo is None:
gameinfo = transposeTable(
parseTable(ts.select("table:nth-of-type(3)")[0]) +
parseTable(ts.select("table:nth-of-type(4)")[0])
)
dumpTable(
gameinfo,
filename.format("Info")
)
teams = [gamescore[1][0].text.strip(), gamescore[2][0].text.strip()]
periods = [v.text.strip() for v in gamescore[0][1:]]
if table["href"].startswith("/game/box_score"):
if table.text.strip() == "Box Score":
sfilename = filename.format("Box Score - {}")
else:
sfilename = filename.format(periods[nextPeriod] + " - {}")
nextPeriod += 1
dumpTable(
parseTable(ts.select("table:nth-of-type(5)")[0], header=1),
sfilename.format(teams[0])
)
dumpTable(
parseTable(ts.select("table:nth-of-type(6)")[0], header=1),
sfilename.format(teams[1])
)
elif table["href"].startswith("/game/play_by_play"):
sfilename = filename.format("Play by Play - {}")
for (i, period) in enumerate(periods[:-1]):
dumpTable(
parseTable(ts.select("table:nth-of-type({})".format(6 + 2 * i))[0], header=0),
sfilename.format(period)
)
sleep(2)
if gamescore == gameinfo == None:
raise Exception("Not a game.")
setFlag(".done")
sleep(2)
except Exception as e:
print2(colored("Error: ", "red"), e)
finally:
print2()
if not readFlag(".parsed"):
try:
gamelink = urljoin(domain, url)
log("{} {} {} {} {} {}".format(sport, year, division, org, game, dumpURL(gamelink)))
print2("Parsing...")
gamescore = loadTable(filename.format("Score"))
sfilename = filename.format("Box Score - {}")
teams = [gamescore[1][0], gamescore[2][0]]
with open(filename.format("Box Score - All (Parsed)"), "w") as af:
for team in teams:
boxScore = parseBoxScore(
sfilename.format(team),
filename.format("Info"),
team,
"All"
)
rawDumpTable(boxScore[(0 if team == teams[0] else 1):], af)
sfilename = filename.format("Play by Play - {}")
periods = gamescore[0][1:]
with open(filename.format("Play by Play - All (Parsed)"), "w") as af:
for period in periods[:-1]:
playByPlay = parsePlayByPlay(
sfilename.format(period),
period,
filename.format("Info")
)
rawDumpTable(playByPlay[(0 if period == periods[0] else 1):], af)
setFlag(".parsed")
except Exception as e:
print2(colored("Error: ", "red"), e)
finally:
print2()