def _add_stats_data(teams_list, team_data_dict):
"""
Add a team's stats row to a dictionary.
Pass table contents and a stats dictionary of all teams to accumulate
all stats for each team in a single variable.
Parameters
----------
teams_list : generator
A generator of all row items in a given table.
team_data_dict : {str: {'data': str, 'rank': int}} dictionary
A dictionary where every key is the team's abbreviation and every
value is another dictionary with a 'data' key which contains the
string version of the row data for the matched team, and a 'rank'
key which is the rank of the team.
Returns
-------
dictionary
An updated version of the team_data_dict with the passed table row
information included.
"""
# Teams are listed in terms of rank with the first team being #1
rank = 1
for team_data in teams_list:
if 'class="thead onecell"' in str(team_data):
continue
abbr = utils._parse_field(PARSING_SCHEME, team_data, 'abbreviation')
try:
team_data_dict[abbr]['data'] += team_data
except KeyError:
team_data_dict[abbr] = {'data': team_data, 'rank': rank}
rank += 1
return team_data_dict
def _parse_game_data(self, game_data):
"""
Parses a value for every attribute.
The function looks through every attribute with the exception of those
listed below and retrieves the value according to the parsing scheme
and index of the attribute from the passed HTML data. Once the value
is retrieved, the attribute's value is updated with the returned
result.
Note that this method is called directory once Game is invoked and does
not need to be called manually.
Parameters
----------
game_data : string
A string containing all of the rows of stats for a given game.
"""
for field in self.__dict__:
# Remove the leading '_' from the name
short_name = str(field)[1:]
if short_name == 'datetime' or \
short_name == 'opponent_rank':
continue
elif short_name == 'opponent_abbr':
self._parse_abbreviation(game_data)
continue
elif short_name == 'boxscore':
self._parse_boxscore(game_data)
continue
value = utils._parse_field(SCHEDULE_SCHEME, game_data, short_name)
setattr(self, field, value)
def _add_stats_data(teams_list, team_data_dict):
"""
Add a team's stats row to a dictionary.
Pass table contents and a stats dictionary of all teams to accumulate all
stats for each team in a single variable.
Parameters
----------
teams_list : generator
A generator of all row items in a given table.
team_data_dict : {str: {'data': str}} dictionary
A dictionary where every key is the team's abbreviation and every value
is another dictionary with a 'data' key which contains the string
version of the row data for the matched team.
Returns
-------
dictionary
An updated version of the team_data_dict with the passed table row
information included.
"""
if not teams_list:
return team_data_dict
for team_data in teams_list:
# Skip the sub-header rows
if 'class="over_header thead"' in str(team_data) or \
'class="thead"' in str(team_data):
continue
abbr = utils._parse_field(PARSING_SCHEME, team_data, 'abbreviation')
try:
team_data_dict[abbr]['data'] += team_data
except KeyError:
team_data_dict[abbr] = {'data': team_data}
return team_data_dict
def _parse_game_data(self, uri):
"""
Parses a value for every attribute.
This function looks through every attribute and retrieves the value
according to the parsing scheme and index of the attribute from the
passed HTML data. Once the value is retrieved, the attribute's value is
updated with the returned result.
Note that this method is called directly once Boxscore is invoked and
does not need to be called manually.
Parameters
----------
uri : string
The relative link to the boxscore HTML page, such as
'BOS/BOS201806070'.
"""
boxscore = self._retrieve_html_page(uri)
# If the boxscore is None, the game likely hasn't been played yet and
# no information can be gathered. As there is nothing to grab, the
# class instance should just be empty.
if not boxscore:
return
for field in self.__dict__:
# Remove the '_' from the name
short_field = str(field)[1:]
if short_field == 'winner' or \
short_field == 'winning_name' or \
short_field == 'winning_abbr' or \
short_field == 'losing_name' or \
short_field == 'losing_abbr' or \
short_field == 'uri':
continue
if short_field == 'date' or \
short_field == 'time' or \
short_field == 'venue' or \
short_field == 'attendance' or \
short_field == 'time_of_day' or \
short_field == 'duration':
value = self._parse_game_date_and_location(
short_field, boxscore)
setattr(self, field, value)
continue
if short_field == 'away_name' or \
short_field == 'home_name':
value = self._parse_name(short_field, boxscore)
setattr(self, field, value)
continue
index = 0
if short_field in BOXSCORE_ELEMENT_INDEX.keys():
index = BOXSCORE_ELEMENT_INDEX[short_field]
value = utils._parse_field(BOXSCORE_SCHEME, boxscore, short_field,
index)
setattr(self, field, value)
def test__parse_field_returns_value_for_non_abbreviation(self):
parsing_scheme = {'batters_used': 'td[data-stat="batters_used"]:first'}
html_string = '''<td class="right " data-stat="batters_used">32</td>
<td class="right " data-stat="age_bat">29.1</td>
<td class="right " data-stat="runs_per_game">4.10</td>'''
expected = '32'
result = utils._parse_field(parsing_scheme,
MockHtml(html_string, [expected]),
'batters_used')
assert result == expected
def test_parse_field_returns_none_on_index_error(self):
parsing_scheme = {'batters_used': 'td[data-stat="batters_used"]:first'}
html_string = '''<td class="right " data-stat="batters_used">32</td>
<td class="right " data-stat="age_bat">29.1</td>
<td class="right " data-stat="runs_per_game">4.10</td>'''
expected = None
result = utils._parse_field(parsing_scheme,
MockHtml(html_string, [expected]),
'batters_used',
index=3)
assert result == expected
def test__parse_field_returns_abbreviation(self):
parsing_scheme = {'abbreviation': 'a'}
input_abbreviation = '/teams/ARI/2018.shtml'
expected = 'ARI'
flexmock(utils) \
.should_receive('_parse_abbreviation') \
.and_return('ARI') \
.once()
result = utils._parse_field(parsing_scheme,
MockHtml(input_abbreviation, None),
'abbreviation')
assert result == expected
def test_secondary_index_pulling_values_bad_secondary(self):
parsing_scheme = {'batters_used': 'td[data-stat="batters_used"]'}
html_string = '''<td class="right " data-stat="batters_used">32</td>
<td class="right " data-stat="age_bat">29.1</td>
<td class="right " data-stat="runs_per_game">4.10</td>
<td class="right " data-stat="batters_used">31</td>
<td class="right " data-stat="batters_used">34</td>'''
items = [32, 31, 34]
result = utils._parse_field(parsing_scheme,
MockHtml(html_string, items),
'batters_used',
index=3,
secondary_index=4)
assert not result
def _parse_game_data(self, uri):
"""
Parses a value for every attribute.
This function looks through every attribute and retrieves the value
according to the parsing scheme and index of the attribute from the
passed HTML data. Once the value is retrieved, the attribute's value is
updated with the returned result.
Note that this method is called directly once Boxscore is invoked and
does not need to be called manually.
Parameters
----------
uri : string
The relative link to the boxscore HTML page, such as
'201802040nwe'.
"""
boxscore = self._retrieve_html_page(uri)
# If the boxscore is None, the game likely hasn't been played yet and
# no information can be gathered. As there is nothing to grab, the
# class instance should just be empty.
if not boxscore:
return
fields_to_special_parse = [
'away_even_strength_assists', 'away_power_play_assists',
'away_short_handed_assists', 'away_game_winning_goals',
'away_saves', 'away_save_percentage', 'away_shutout',
'home_even_strength_assists', 'home_power_play_assists',
'home_short_handed_assists', 'home_game_winning_goals',
'home_saves', 'home_save_percentage', 'home_shutout'
]
for field in self.__dict__:
# Remove the '_' from the name
short_field = str(field)[1:]
if short_field == 'winner' or \
short_field == 'winning_name' or \
short_field == 'winning_abbr' or \
short_field == 'losing_name' or \
short_field == 'losing_abbr' or \
short_field == 'uri':
continue
if short_field == 'date' or \
short_field == 'time' or \
short_field == 'arena' or \
short_field == 'attendance' or \
short_field == 'time_of_day' or \
short_field == 'duration':
value = self._parse_game_date_and_location(
short_field, boxscore)
setattr(self, field, value)
continue
if short_field == 'away_name' or \
short_field == 'home_name':
value = self._parse_name(short_field, boxscore)
setattr(self, field, value)
continue
if short_field in fields_to_special_parse:
scheme = BOXSCORE_SCHEME[short_field]
value = [i.text() for i in boxscore(scheme).items()]
setattr(self, field, value)
continue
index = 0
if short_field in BOXSCORE_ELEMENT_INDEX.keys():
index = BOXSCORE_ELEMENT_INDEX[short_field]
value = utils._parse_field(BOXSCORE_SCHEME, boxscore, short_field,
index)
setattr(self, field, value)
self._away_skaters = len(boxscore(BOXSCORE_SCHEME['away_skaters']))
num_away_goalies = boxscore(BOXSCORE_SCHEME['away_goalies']).items()
# Skip the first element as it is dedicated to skaters and not goalies.
next(num_away_goalies)
self._away_goalies = len(next(num_away_goalies)('tbody tr'))