def fetch():
options, args = commandline_args("Fetch MLB Gameday data.")
# It seems this should be deprecated.
requests.adapters.DEFAULT_RETRIES = 3
f = FetchMLBAM(**options)
f.fetch()
return f
def run():
options, args = commandline_args('Parse Gameday XML files')
d = DB()
conn, meta = d.connect(options['database'])
for league in options['leagues']:
dir = os.path.join(options['output_dir'], league)
for day in each_day(options['start'], options['end']):
parsed_games = parse_day(dir, day)
d.add_games(parsed_games)
def run():
options, args = commandline_args("Dump BIP into JSON files")
d = DB()
conn, meta = d.connect(options["database"])
park_table = d.model.park
park_dim_table = d.model.park_dimension
bip_table = d.model.bip
game_table = d.model.game
player_table = d.model.mlbam_player
ab_table = d.model.atbat
# Arg, postgres requires every column to be in the ORDER BY clause
park_sql = select(
[park_table.c.id, park_table.c.name, func.count(bip_table.c.id).label("num")],
from_obj=park_table.join(bip_table),
).group_by(park_table.c.id, park_table.c.name)
parks = {}
for row in conn.execute(park_sql):
p = {}
for key in row.keys():
if key in set(["id", "num"]):
p[key] = int(row[key])
else:
p[key] = str(row[key])
p["bip"] = p["num"]
p["years"] = {}
del p["num"]
parks[p["id"]] = p
# Skipping the closing year for simplicity. If a team switches parks in the
# middle of the year, it will be a park with a new id. I believe the only
# issue will be if MLBAM updates the image in the middle of the season.
# This might happen for Citi Field (no bases or foul lines!).
dimension_sql = select(
[
park_dim_table.c.park_id,
park_dim_table.c.image_file,
park_dim_table.c.opening,
park_dim_table.c.hp_x,
park_dim_table.c.hp_y,
park_dim_table.c.image_hp_x,
park_dim_table.c.image_hp_y,
park_dim_table.c.scale,
],
from_obj=park_dim_table,
)
for row in conn.execute(dimension_sql):
id = row["park_id"]
if id not in parks:
continue
opening = row["opening"]
images = {
"file": row["image_file"],
"scale": float(row["scale"]),
"hp_x": float(row["hp_x"]),
"hp_y": float(row["hp_y"]),
"image_hp_x": float(row["image_hp_x"]),
"image_hp_y": float(row["image_hp_y"]),
"opening": opening.year,
}
if "images" not in parks[id]:
parks[id]["images"] = {}
parks[id]["images"][opening.year] = images
years_sql = select([func.distinct(text(get_year(conn, "day")))], from_obj=game_table)
years = [int(row[0]) for row in conn.execute(years_sql)]
p = player_table.alias()
b = player_table.alias()
bip_sql = select(
[
game_table.c.day.label("day"),
bip_table.c.type.label("type"),
bip_table.c.x.label("x"),
bip_table.c.y.label("y"),
ab_table.c.event.label("event"),
(b.c.namelast + ", " + b.c.namefirst).label("batter"),
ab_table.c.batter_stand.label("stand"),
(p.c.namelast + ", " + p.c.namefirst).label("pitcher"),
ab_table.c.pitcher_throw.label("throw"),
],
and_(park_table.c.id == bindparam("park"), text(get_year(conn, "day")) == bindparam("year")),
from_obj=bip_table.join(park_table)
.join(ab_table)
.join(game_table)
.outerjoin(p, onclause=p.c.mlbamid == ab_table.c.pitcher)
.outerjoin(b, onclause=b.c.mlbamid == ab_table.c.batter),
)
for park_id in parks.keys():
for year in years:
bip_list = []
str_y = str(year)
for bip in conn.execute(bip_sql, {"park": park_id, "year": str_y}):
bip_list.append(
{
"x": bip["x"],
"y": bip["y"],
"event": bip["event"],
"type": bip["type"],
"pitcher": bip["pitcher"],
"throw": bip["throw"],
"batter": bip["batter"],
"stand": bip["stand"],
}
)
# No need to write empty files!
if len(bip_list) > 0:
parks[park_id]["years"][year] = True
park_file = os.path.join(options["output_dir"], "park-" + str(park_id) + "-" + str(year) + ".json")
dump_json(park_file, bip_list)
parks_file = os.path.join(options["output_dir"], "parks.json")
dump_json(parks_file, parks)
def run():
args, extra_args = commandline_args('Parse Retrosheet extended game description')
Events(extra_args)
def run():
args, extra_args = commandline_args('Summarize Retrosheet data')
summarize_stats(args, extra_args)
