def _get_games(self):
"""
get all games from base url
"""
base_url = self.DELIMITER.join(
[self.BASE_URL,
self.YMD_FORMAT.format(**self.params), 'epg.xml'])
html = Utils.get_content(base_url, self.PARSER)
return Utils.find_all_tags(html, 'game')
def scrape(self):
games = self._get_games()
selected_game = games[int(self.box)]
boxscore_path = Utils.find_attr(selected_game, 'game_data_directory')
boxscore_url = 'http://gd2.mlb.com' + boxscore_path + '/' + 'boxscore.xml'
html = Utils.get_content(boxscore_url, self.PARSER)
boxscore = Utils.find_tag(html, 'boxscore')
# basic info
away_team_name = Utils.find_attr(boxscore, 'away_fname')
home_team_name = Utils.find_attr(boxscore, 'home_fname')
self._scrape_scoreboard(boxscore, away_team_name, home_team_name)
self._scrape_boxscore(boxscore, away_team_name, home_team_name)
def scrape(self):
games = self._get_games()
selected_game = games[int(self.roster)]
game_info_path = Utils.find_attr(selected_game,'game_data_directory')
roster_url = 'http://gd2.mlb.com' + game_info_path + '/' + 'players.xml'
away_roster = []
home_roster = []
header = ['TEAM','NAME','POS']
aligns = ['c','c','c']
away_roster.append(header)
home_roster.append(header)
html = Utils.get_content(roster_url,self.PARSER)
teams = Utils.find_all_tags(html,'team')
for team in teams:
roster = []
type = Utils.find_attr(team,'type')
team_nm = Utils.find_attr(team,'name')
members = Utils.find_all_tags(team,'player')
for member in members:
members_info = []
first_name = Utils.find_attr(member,'first')
last_name = Utils.find_attr(member,'last')
full_name = ' '.join([first_name,last_name])
members_info.append(team_nm)
members_info.append(full_name)
members_info.append(Utils.find_attr(member,'position'))
roster.append(members_info)
if type == 'away':
away_roster.extend(roster)
else:
home_roster.extend(roster)
print('\t'.join(['***','away','***']))
Utils.draw_table(away_roster,aligns,False)
print('')
print('\t'.join(['***','home','***']))
Utils.draw_table(home_roster,aligns,False)
def scrape(self):
games = self._get_games()
selected_game = games[int(self.play)]
game_info_path = Utils.find_attr(selected_game, 'game_data_directory')
play_by_play_url = 'http://gd2.mlb.com' + game_info_path + '/' + 'game_events.xml'
table_contents = []
header = ['INNING', 'BATTER_NO', 'DESCRIPTION', 'SCORE']
aligns = ['c', 'c', 'l', 'c']
table_contents.append(header)
html = Utils.get_content(play_by_play_url, self.PARSER)
all_innings = Utils.find_all_tags(html, 'inning')
for single_inning in all_innings:
inning = []
inning_no = Utils.find_attr(single_inning, 'num')
top = Utils.find_tag(single_inning, 'top')
bottom = Utils.find_tag(single_inning, 'bottom')
top_atbats = Utils.find_all_tags(top, 'atbat')
bottom_atbats = Utils.find_all_tags(bottom, 'atbat')
self._get_results(top_atbats, inning, inning_no)
self._get_results(bottom_atbats, inning, inning_no)
table_contents.extend(inning)
if self.score:
score_plays_table = []
score_plays_table_aligns = ['c', 'l', 'c']
score_plays_table.append(['INNING', 'DESCRIPTION', 'SCORE'])
for table_content in table_contents:
score_play = []
if 'scores' in table_content[2] or 'homers' in table_content[2]:
score_play.append(table_content[0])
score_play.append(table_content[2])
score_play.append(table_content[3])
score_plays_table.append(score_play)
Utils.draw_table(score_plays_table, score_plays_table_aligns,
False)
elif self.grep != 'no_grep':
grep_result_table = []
grep_result_table_aligns = ['c', 'l', 'c']
grep_result_table.append(['INNING', 'DESCRIPTION', 'SCORE'])
for table_content in table_contents:
grep_play = []
if self.grep in table_content[2]:
grep_play.append(table_content[0])
grep_play.append(table_content[2])
grep_play.append(table_content[3])
grep_result_table.append(grep_play)
Utils.draw_table(grep_result_table, grep_result_table_aligns,
False)
else:
Utils.draw_table(table_contents, aligns, False)