def fix_dates():
for match in matches.find({'startDate': {'$type': 2}}).sort('matchId', -1).batch_size(100):
print(match['matchId'])
match['startDate'] = datetime.strptime(match['startDate'], '%m/%d/%Y %H:%M:%S')
match['startTime'] = datetime.strptime(match['startTime'], '%m/%d/%Y %H:%M:%S')
if 'timeStamp' in match:
try:
match['timeStamp'] = datetime.strptime(match['timeStamp'], '%d/%m/%Y %H:%M:%S')
except ValueError:
match['timeStamp'] = datetime.strptime(match['timeStamp'], '%Y-%m-%d %H:%M:%S')
matches.save(match)
def load_events():
print(datetime.now(), 'Starting')
for match in matches.find({'events': {'$exists': True}}).sort('matchId', -1):
count = events.count({'matchId': match['matchId']})
if count and len(match['events']) == count:
# print('Events already written')
continue
else:
print(datetime.now(), match['matchId'], '{0} events'.format(len(match['events'])))
events.remove({'matchId': match['matchId']})
for event in match['events']:
for k in ['matchId', 'stageId', 'seasonId', 'tournamentId', 'regionId']:
if k in match:
event[k] = match[k]
events.insert_one(event)
print(datetime.now(), 'Complete')
def update_matches(status_code=None):
# 0: Error
# 1: Pending
# 2: Postponed
# 3: In-Play
# 4: (Not seen)
# 5: Abandoned
# 6: Complete
# 7: Cancelled
# 8: Ignore
if status_code is None or type(status_code) is not int:
status_code = {'$in': [1, 3]}
for match in matches.find({'statusCode': status_code,
'error': {'$exists': False},
'startDate': {'$lte': datetime.today()},
'startTime': {'$lte': datetime.today()},
}).sort([('statusCode', 1), ('startDate', -1)]):
print(match['matchId'], match['statusCode'], match['startTime'])
get_match(match['matchId'], overwrite=True)
if geocode_result:
elevation_result = gmaps.elevation(
convert.normalize_lat_lng(
geocode_result[0]['geometry']['location']))
stadium_data_to_insert = {
'venueName': stadium,
'location_data': geocode_result[0],
'altitude_data': elevation_result[0]
}
stadiums.replace_one({'venueName': stadium},
stadium_data_to_insert,
upsert=True)
if wiki_name:
wiki_page = wptools.page(wiki_name[0], silent=True)
wiki_page.get_parse()
wiki_data = wiki_page.data['infobox']
wiki_data_to_insert = {
'venueName': stadium,
'wiki_name': wiki_name[0],
'wiki_data': wiki_data
}
wiki.replace_one({'venueName': stadium},
wiki_data_to_insert,
upsert=True)
if __name__ == "__main__":
stads = matches.find().distinct('venueName')
stads = [x for x in stads if x != None]
for stadium in stads:
scrape(stadium)
def missing_stadiums():
for stadium in matches.find().distinct('venueName'):
if not stadiums.find_one({'venueName': stadium}):
print(stadium)
def load_players():
for match in matches.find({'playerIdNameDictionary': {'$exists': True}}).sort('matchId', 1):
print(match['matchId'])
for k, v in match['playerIdNameDictionary'].items():
players.update_one({'playerId': int(k)}, {'$setOnInsert': {'name': v}}, upsert=True)
for attendance in content.xpath(
'//div[@class="sb-spieldaten"]/p[3]/span/strong/text()'):
match['attendance'] = int(
attendance.replace(' Spectators', '').replace('.', ''))
for referee in content.xpath('//div[@class="sb-spieldaten"]/p[3]/a'):
referees.update_one(
{'referee': int(referee.xpath('@href')[0].split('/')[-1])},
{'$setOnInsert': {
'name': referee.xpath('@title')[0]
}},
upsert=True)
match['referee'] = int(referee.xpath('@href')[0].split('/')[-1])
matches.save(match)
wait()
if __name__ == '__main__':
get_regions()
for region in regions.find().sort('name'):
get_tournaments(region['region'])
for tournament in tournaments.find().sort('tournament'):
get_seasons(tournament['tournament'])
for season in seasons.find().sort([('season', -1),
('tournament', 1)]).batch_size(1):
get_fixtures(season['tournament'], season['season'])
for match in matches.find().sort('date', -1):
get_lineups(match['match'])
