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_scoreboard(self, boxscore, away_team_name, home_team_name):
"""
scrape scoreboard
:param boxscore: boxscore contents
:param away_team_name: away team name
:param home_team_name: home team name
"""
score_all = Utils.find_tag(boxscore, 'linescore')
away_total_runs = Utils.find_attr(score_all, 'away_team_runs')
home_total_runs = Utils.find_attr(score_all, 'home_team_runs')
away_total_hits = Utils.find_attr(score_all, 'away_team_hits')
home_total_hits = Utils.find_attr(score_all, 'home_team_hits')
away_total_errs = Utils.find_attr(score_all, 'away_team_errors')
home_total_errs = Utils.find_attr(score_all, 'home_team_errors')
innings = Utils.find_all_tags(score_all, 'inning_line_score')
table_contents = []
header = []
away = []
home = []
aligns = []
header.append('TEAM')
away.append(away_team_name)
home.append(home_team_name)
for inning in innings:
aligns.append('c')
header.append(Utils.find_attr(inning, 'inning'))
away.append(Utils.find_attr(inning, 'away'))
home.append(Utils.find_attr(inning, 'home'))
header.extend(['R', 'H', 'E'])
away.extend([away_total_runs, away_total_hits, away_total_errs])
home.extend([home_total_runs, home_total_hits, home_total_errs])
aligns.extend(['c', 'c', 'c', 'c'])
table_contents.append(header)
table_contents.append(away)
table_contents.append(home)
Utils.draw_table(table_contents, aligns, True)
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)