def parse_fighter_page(ref):
# parse main info from Sherdog
logger.info("Parsing Sherdog info for {}".format(ref))
sherdog_page = _download_page(_sherdog_ref(ref))
fighter_name = sherdog_page.body.find('span', attrs={'class': 'fn'}).text
ufc_url = "http://www.ufc.com/fighter/" + fighter_name.strip().replace(
" ", "-")
wiki_url = "https://en.wikipedia.org/wiki/" + fighter_name.strip().replace(
" ", "_")
fighter_data = FighterData(sherdog_page, _download_page(wiki_url),
_download_page(ufc_url))
# extract each field
fighter = Fighter(ref=ref, name=fighter_name)
fighter.country = fighter_data.extract(country_extractor)
fighter.city = fighter_data.extract(city_extractor)
fighter.birthday = fighter_data.extract(sherdog_birthday_extractor,
wiki_birthday_extractor)
fighter.height = fighter_data.extract(height_extractor)
fighter.weight = fighter_data.extract(weight_extractor)
fighter.reach = fighter_data.extract(wiki_reach_extractor,
ufc_reach_extractor)
fighter.specialization = fighter_data.extract(spec_extractor)
fight_infos = _parse_fight_infos(ref, sherdog_page)
return fighter, fight_infos
def parse_fighter_page(ref):
# parse main info from Sherdog
logger.info("Parsing Sherdog info for {}".format(ref))
sherdog_page = _download_page(_sherdog_ref(ref))
fighter_name = sherdog_page.body.find('span', attrs={'class': 'fn'}).text
ufc_url = "http://www.ufc.com/fighter/" + fighter_name.strip().replace(" ", "-")
wiki_url = "https://en.wikipedia.org/wiki/" + fighter_name.strip().replace(" ", "_")
fighter_data = FighterData(
sherdog_page,
_download_page(wiki_url),
_download_page(ufc_url)
)
# extract each field
fighter = Fighter(ref=ref, name=fighter_name)
fighter.country = fighter_data.extract(country_extractor)
fighter.city = fighter_data.extract(city_extractor)
fighter.birthday = fighter_data.extract(sherdog_birthday_extractor, wiki_birthday_extractor)
fighter.height = fighter_data.extract(height_extractor)
fighter.weight = fighter_data.extract(weight_extractor)
fighter.reach = fighter_data.extract(wiki_reach_extractor, ufc_reach_extractor)
fighter.specialization = fighter_data.extract(spec_extractor)
fight_infos = _parse_fight_infos(ref, sherdog_page)
return fighter, fight_infos
from storage.models.fight import Fight
from storage.models.fighter import Fighter
from storage import Session, init_db
from storage.models.event import Event
if __name__ == "__main__":
init_db()
session = Session()
tagir = Fighter(name='Tagir', ref='/fighter/tagir-1001')
session.add(tagir)
sergey = Fighter(name='Sergey', ref='/fighter/sergey-1002')
session.add(sergey)
ufc01 = Event(name='UFC 01', ref='/event/ufc-01')
session.add(ufc01)
f1 = Fight(fighter1=tagir, fighter2=sergey, event=ufc01, outcome='Tagir crushed him')
session.add(f1)
# e1 = session.query(Event).filter_by(name='UFC 01').first()
# print(e1.fights[0].first_fighter.name)
#t = session.query(Fighter).filter_by(name='Tagir').first()
session.commit()
t = session.query(Fight).first()
tagir.name = "TTT"