def insert_race(soup):
# Extract year and month field
year_month = soup.find("h3", {"class": "midashi3rd"}).string
year = formatter("(\d+)年\d+月", year_month, "int")
month = formatter("\d+年(\d+)月", year_month, "int")
for row in soup.select("tbody tr"):
cells = row.findAll('td')
if len(cells) > 2 and cells[0].find("a") is not None:
race_id_without_round = "20" + formatter("\d+", get_from_a(
cells[0]))
day = formatter("(\d+)日（[日月火水木金土]）", str(cells[0]), "int")
place_id = race_id_without_round[4:6]
place_name = to_place_name(place_id)
days = int(race_id_without_round[6:8])
times = int(race_id_without_round[8:10])
for r in range(1, 13):
Race.objects.get_or_create(race_id=race_id_without_round +
str(r).zfill(2),
race_dt=date(year, month, day),
place_id=place_id,
place_name=place_name,
days=days,
times=times,
round=r)
return
def parse_result(cells, race, tof):
result = Result()
result.race = race
result.rank = formatter("\d+", cells[0].string, "int")
result.bracket = formatter("\d+", cells[1].string, "int")
result.horse_num = formatter("\d+", cells[2].string, "int")
result.horse_id = formatter("\d+", get_from_a(cells[3]))
result.horse_name = formatter("[^!-~\xa0]+", get_from_a(cells[3], "name"))
result.key = race.race_id + result.horse_id
result.sex = formatter("[牡牝騸セ]", cells[4].string)
result.age = formatter("\d+", cells[4].string, "int")
result.burden = formatter("(\d+).?\d?", cells[5].string, "float")
result.jockey_id = formatter("\d+", get_from_a(cells[6]))
result.jockey_name = formatter("[^!-~\xa0]+", get_from_a(cells[6], "name"))
if cells[7].string is not None:
tmp_time = re.split(r"[:.]", cells[7].string)
tmp_time = [int(i) for i in tmp_time] # Type conversion
tmp_sec = (tmp_time[0] * 60 + tmp_time[1]) + (tmp_time[2] / 10)
result.finish_time = tmp_sec
result.time_lag = round(
result.finish_time -
tof, 2) if result.finish_time is not None and tof != 0 else 0
result.last3f_time = formatter("\d+.\d+", cells[11].string, "float")
result.odds = formatter("\d+.\d+", cells[12].string, "float")
result.odor = formatter("\d+", cells[13].string, "int")
result.weight = formatter("(\d+)\([+-]?\d*\)", cells[14].string, "int")
result.weight_diff = formatter("\d+\(([+-]?\d+)\)", cells[14].string,
"int")
result.trainer_id = formatter("\d+", get_from_a(cells[15]))
result.trainer_name = formatter("[^!-~\xa0]+",
get_from_a(cells[15], "name"))
result.owner_id = formatter("\d+", get_from_a(cells[19]))
result.owner_name = formatter("[^!-~\xa0]+", get_from_a(cells[19], "name"))
result.prize = formatter("\d*,?\d*.?\d+", cells[20].string, "float")
return result
def parse_entry_8(cells, race):
result = Result()
result.race = race
result.horse_id = formatter("\d+", get_from_a(cells[1]))
result.horse_name = formatter("[^!-~\xa0]+", get_from_a(cells[1], "name"))
result.key = race.race_id + result.horse_id
result.sex = formatter("[牡牝騸セ]", cells[2].string)
result.age = formatter("\d+", cells[2].string, "int")
result.burden = formatter("(\d+).?\d?", cells[3].string, "float")
result.jockey_id = formatter("\d+", get_from_a(cells[4]))
result.jockey_name = formatter("[^!-~\xa0]+", get_from_a(cells[4], "name"))
result.trainer_id = formatter("\d+", get_from_a(cells[5]))
result.trainer_name = formatter("[^!-~\xa0]+",
get_from_a(cells[5], "name"))
result.odds = 0
result.odor = 0
return result
def parse_entry_12(cells, race):
result = Result()
result.race = race
result.bracket = formatter("\d+", cells[0].string, "int")
result.horse_num = formatter("\d+", cells[1].string, "int")
result.horse_id = formatter("\d+", get_from_a(cells[3]))
result.horse_name = formatter("[^!-~\xa0]+", get_from_a(cells[3], "name"))
result.key = race.race_id + result.horse_id
result.sex = formatter("[牡牝騸セ]", cells[4].string)
result.age = formatter("\d+", cells[4].string, "int")
result.burden = formatter("(\d+).?\d?", cells[5].string, "float")
result.jockey_id = formatter("\d+", get_from_a(cells[6]))
result.jockey_name = formatter("[^!-~\xa0]+", get_from_a(cells[6], "name"))
result.trainer_id = formatter("\d+", get_from_a(cells[7]))
result.trainer_name = formatter("[^!-~\xa0]+",
get_from_a(cells[7], "name"))
result.odds = formatter("\d+.\d+", cells[8].string, "float")
result.odor = formatter("\d+", cells[9].string, "int")
return result