def __init__(self, playerid, name, home): """ Create a new Player instance with the player id (from NFL.com's GameCenter), the player's name (e.g., "T.Brady") and whether the player is playing in a home game or not. """ self.playerid = playerid self.name = name self.home = home self._stats = OrderedDict()
def _json_play_players(play, data): """ Takes a single JSON play entry (data) and converts it to an OrderedDict of player statistics. play is the instance of Play that this data is part of. It is used to determine whether the player belong to the home team or not. """ players = OrderedDict() for playerid, statcats in data.iteritems(): if playerid == '0': continue for info in statcats: if info['statId'] not in nflgame.statmap.idmap: continue if playerid not in players: home = play.drive.game.is_home(info['clubcode']) if home: team_name = play.drive.game.home else: team_name = play.drive.game.away stats = nflgame.player.PlayPlayerStats(playerid, info['playerName'], home, team_name) players[playerid] = stats statvals = nflgame.statmap.values(info['statId'], info['yards']) players[playerid]._add_stats(statvals) return players
def _json_game_player_stats(game, data): """ Parses the 'home' and 'away' team stats and returns an OrderedDict mapping player id to their total game statistics as instances of nflgame.player.GamePlayerStats. """ players = OrderedDict() for team in ('home', 'away'): for category in nflgame.statmap.categories: if category not in data[team]['stats']: continue for pid, raw in data[team]['stats'][category].iteritems(): stats = {} for k, v in raw.iteritems(): if k == 'name': continue stats['%s_%s' % (category, k)] = v if pid not in players: home = team == 'home' if home: team_name = game.home else: team_name = game.away players[pid] = nflgame.player.GamePlayerStats( pid, raw['name'], home, team_name) players[pid]._add_stats(stats) return players
def max_player_stats(self): """ Returns a GenPlayers sequence of player statistics that combines game statistics and play statistics by taking the max value of each corresponding statistic. This is useful when accuracy is desirable. Namely, using only play-by-play data or using only game statistics can be unreliable. That is, both are inconsistently correct. Taking the max values of each statistic reduces the chance of being wrong (particularly for stats that are in both play-by-play data and game statistics), but does not eliminate them. """ game_players = list(self.players) play_players = list(self.drives.plays().players()) max_players = OrderedDict() # So this is a little tricky. It's possible for a player to have # only statistics at the play level, and therefore not be represented # in the game level statistics. Therefore, we initialize our # max_players with play-by-play stats first. Then go back through # and combine them with available game statistics. for pplay in play_players: newp = nflgame.player.GamePlayerStats(pplay.playerid, pplay.name, pplay.home, pplay.team) maxstats = {} for stat, val in pplay._stats.iteritems(): maxstats[stat] = val newp._overwrite_stats(maxstats) max_players[pplay.playerid] = newp for newp in max_players.itervalues(): for pgame in game_players: if pgame.playerid != newp.playerid: continue maxstats = {} for stat, val in pgame._stats.iteritems(): maxstats[stat] = max([val, newp._stats.get(stat, -_MAX_INT)]) newp._overwrite_stats(maxstats) break return nflgame.seq.GenPlayerStats(max_players)
def new_schedule(): """ Builds an entire schedule from scratch. """ sched = OrderedDict() for year, stype, week in year_phase_week(): update_week(sched, year, stype, week) return sched
def __add__(self, other): """ Adds two sequences of players by combining repeat players and summing their statistics. """ players = OrderedDict() for p in itertools.chain(self, other): if p.playerid not in players: players[p.playerid] = p else: players[p.playerid] += p return GenPlayerStats(players)
def players(self): """ Returns the combined player stats for every play in the sequence. """ players = OrderedDict() for play in self: for player in play.players: if player.playerid not in players: players[player.playerid] = player else: players[player.playerid] += player return GenPlayerStats(players)
def diff(before, after): """ Returns the difference between two points of time in a game in terms of plays and player statistics. The return value is a GameDiff namedtuple with two attributes: plays and players. Each contains *only* the data that is in the after game but not in the before game. This is useful for sending alerts where you're guaranteed to see each play statistic only once (assuming NFL.com behaves itself). XXX: There is an assertion that requires after's game clock be the same or later than before's game clock. This may need to be removed if NFL.com allows its game clock to be rolled back due to corrections from refs. """ assert after.time >= before.time, \ 'When diffing two games, "after" (%s) must be later or the ' \ 'same time as "before" (%s).' % (after.time, before.time) assert after.eid == before.eid plays = [] after_plays = list(after.drives.plays()) before_plays = list(before.drives.plays()) for play in after_plays: if play not in before_plays: plays.append(play) # You might think that updated play data is enough. You could scan # it for statistics you're looking for (like touchdowns). # But sometimes a play can sneak in twice if its description gets # updated (late call? play review? etc.) # Thus, we do a diff on the play statistics for player data too. _players = OrderedDict() after_players = list(after.drives.players()) before_players = list(before.drives.players()) for aplayer in after_players: has_before = False for bplayer in before_players: if aplayer.playerid == bplayer.playerid: has_before = True pdiff = aplayer - bplayer if pdiff is not None: _players[aplayer.playerid] = pdiff if not has_before: _players[aplayer.playerid] = aplayer players = nflgame.seq.GenPlayerStats(_players) return GameDiff(plays=plays, players=players)
def build_old(nfl_schedules_path): sched = OrderedDict() xml_filenames = get_filenames(nfl_schedules_path, "", ".xml") sort_nicely(xml_filenames) xml_filenames.reverse() cur_year = DETAILED_STATS_START_YEAR for xml_file in xml_filenames: year, week, stype = xml_file.split(".xml")[0].split("-") year = int(year) week = int(week) if year < cur_year: print(str(year)) cur_year = year if year < DETAILED_STATS_START_YEAR: print('Building (%d, %s, %d)...' % (year, stype, week)) update_week(sched, year, stype, week, nfl_schedules_path) return sched
def diff(before, after): """ Returns the difference between two points of time in a game in terms of plays and player statistics. The return value is a GameDiff namedtuple with two attributes: plays and players. Each contains *only* the data that is in the after game but not in the before game. This is useful for sending alerts where you're guaranteed to see each play statistic only once (assuming NFL.com behaves itself). """ assert after.eid == before.eid plays = [] after_plays = list(after.drives.plays()) before_plays = list(before.drives.plays()) for play in after_plays: if play not in before_plays: plays.append(play) # You might think that updated play data is enough. You could scan # it for statistics you're looking for (like touchdowns). # But sometimes a play can sneak in twice if its description gets # updated (late call? play review? etc.) # Thus, we do a diff on the play statistics for player data too. _players = OrderedDict() after_players = list(after.max_player_stats()) before_players = list(before.max_player_stats()) for aplayer in after_players: has_before = False for bplayer in before_players: if aplayer.playerid == bplayer.playerid: has_before = True pdiff = aplayer - bplayer if pdiff is not None: _players[aplayer.playerid] = pdiff if not has_before: _players[aplayer.playerid] = aplayer players = nflgame.seq.GenPlayerStats(_players) return GameDiff(before=before, after=after, plays=plays, players=players)
class PlayerStats(object): """ Player represents a single player and all of his statistical categories. Every player has 'playerid', 'name' and 'home' fields. Additionally, depending upon which statistical categories that player was involved in for the game, he'll have properties such as 'passing_tds', 'rushing_yds', 'defense_int' and 'kicking_fgm'. In order to know whether a paricular player belongs to a statical category, you may use the filtering methods of a player sequence or alternatively, use the has_cat method with arguments like 'passing', 'rushing', 'kicking', etc. (A player sequence in this case would be an instance of GenPlayerStats.) You may also inspect whether a player has a certain property by using the special __dict__ attribute. For example:: if 'passing_yds' in player.__dict__: # Do something with player.passing_yds """ def __init__(self, playerid, name, home, team): """ Create a new Player instance with the player id (from NFL.com's GameCenter), the player's name (e.g., "T.Brady") and whether the player is playing in a home game or not. """ self.playerid = playerid self.name = name self.home = home self.team = team self._stats = OrderedDict() self.player = None if self.playerid in nflgame.players: self.player = nflgame.players[self.playerid] def has_cat(self, cat): for f in self._stats: if f.startswith(cat): return True return False @property def guess_position(self): """ Guesses the position of this player based on the statistical categories present in this object when player meta is not present. Note that if this resorts to a guess, then it will be more effective on aggregate data rather than data from just a single play. (e.g., if a QB runs the ball, and that's the only data available, the position returned will be RB.) When a position is guessed, only the following positions will be returned: QB, RB, WR, DEF, K and P. """ # Look for the player meta first. Duh. if self.player is not None: return self.player.position stats = [ (self.passing_att, "QB"), (self.rushing_att, "RB"), (self.receiving_tar, "WR"), (self.defense_tkl, "DEF"), (self.defense_ast, "DEF"), (self.kicking_tot, "K"), (self.kicking_fga, "K"), (self.punting_tot, "P"), ] return sorted(stats, reverse=True)[0][1] @property def tds(self): """ Returns the total number of touchdowns credited to this player across all statistical categories. """ n = 0 for f, v in self.__dict__.iteritems(): if f.endswith("tds"): n += v return n @property def twopta(self): """ Returns the total number of two point conversion attempts for the passing, rushing and receiving categories. """ return self.passing_twopta + self.rushing_twopta + self.receiving_twopta @property def twoptm(self): """ Returns the total number of two point conversions for the passing, rushing and receiving categories. """ return self.passing_twoptm + self.rushing_twoptm + self.receiving_twoptm @property def twoptmissed(self): """ Returns the total number of two point conversion failures for the passing, rushing and receiving categories. """ return self.passing_twoptmissed + self.rushing_twoptmissed + self.receiving_twoptmissed @property def stats(self): """ Returns a dict of all stats for the player. """ return self._stats def formatted_stats(self): """ Returns a roughly-formatted string of all statistics for this player. """ s = [] for stat, val in self._stats.iteritems(): s.append("%s: %s" % (stat, val)) return ", ".join(s) def _add_stats(self, stats): for k, v in stats.iteritems(): self.__dict__[k] = self.__dict__.get(k, 0) + v self._stats[k] = self.__dict__[k] def _overwrite_stats(self, stats): for k, v in stats.iteritems(): self.__dict__[k] = v self._stats[k] = self.__dict__[k] def __str__(self): """ Simply returns the player's name, e.g., "T.Brady". """ return self.name def __add__(self, other): """ Adds two players together. Only two player objects that correspond to the same human (i.e., GameCenter identifier) can be added together. If two different players are added together, an assertion will be raised. The effect of adding two player objects simply corresponds to the sums of all statistical values. Note that as soon as two players have been added, the 'home' property becomes undefined if the two operands have different values of 'home'. """ assert self.playerid == other.playerid assert type(self) == type(other) if self.home != other.home: home = None else: home = self.home new_player = self.__class__(self.playerid, self.name, home, self.team) new_player._add_stats(self._stats) new_player._add_stats(other._stats) return new_player def __sub__(self, other): assert self.playerid == other.playerid assert type(self) == type(other) new_player = GamePlayerStats(self.playerid, self.name, self.home, self.team) new_player._add_stats(self._stats) for bk, bv in other._stats.iteritems(): if bk not in new_player._stats: # stat was taken away? ignore. continue new_player._stats[bk] -= bv if new_player._stats[bk] == 0: del new_player._stats[bk] else: new_player.__dict__[bk] = new_player._stats[bk] anydiffs = False for k, v in new_player._stats.iteritems(): if v > 0: anydiffs = True break if not anydiffs: return None return new_player def __getattr__(self, name): # If name has one of the categories as a prefix, then return # a default value of zero for cat in nflgame.statmap.categories: if name.startswith(cat): return 0 raise AttributeError def passer_rating(self): """ Calculate and return the passer rating using the NFL formula. Passer rating is calculated using a player's passing attempts, completions, yards, touchdowns, and interceptions. Passer rating in the NFL is on a scale from 0 to 158.3. """ l = [((self.passing_cmp / self.passing_att) - 0.3) * 5] l.append(((self.passing_yds / self.passing_att) - 3) * 0.25) l.append((self.tds / self.passing_att) * 20) l.append(2.375 - (self.passing_ints / self.passing_att * 25)) m = [] for a in l: if a < 0: a = 0 m.append(a) elif a > 2.375: a = 2.375 m.append(a) else: m.append(a) rating = round((sum(m) / 6) * 100, 1) return rating
class PlayerStats (object): """ Player represents a single player and all of his statistical categories. Every player has 'playerid', 'name' and 'home' fields. Additionally, depending upon which statistical categories that player was involved in for the game, he'll have properties such as 'passing_tds', 'rushing_yds', 'defense_int' and 'kicking_fgm'. In order to know whether a paricular player belongs to a statical category, you may use the filtering methods of a player sequence or alternatively, use the has_cat method with arguments like 'passing', 'rushing', 'kicking', etc. (A player sequence in this case would be an instance of GenPlayerStats.) You may also inspect whether a player has a certain property by using the special __dict__ attribute. For example:: if 'passing_yds' in player.__dict__: # Do something with player.passing_yds """ def __init__(self, playerid, name, home): """ Create a new Player instance with the player id (from NFL.com's GameCenter), the player's name (e.g., "T.Brady") and whether the player is playing in a home game or not. """ self.playerid = playerid self.name = name self.home = home self._stats = OrderedDict() def has_cat(self, cat): return self.__dict__.get(cat, False) def __refresh_categories(self): for cat in nflgame.statmap.categories: for f in self.__dict__: if f.startswith(cat): self.__dict__[cat] = True break @property def tds(self): """ Returns the total number of touchdowns credited to this player across all statistical categories. """ n = 0 for f, v in self.__dict__.iteritems(): if f.endswith('tds'): n += v return n @property def stats(self): """ Returns a dict of all stats for the player. """ return self._stats def formatted_stats(self): """ Returns a roughly-formatted string of all statistics for this player. """ s = [] for stat, val in self._stats.iteritems(): s.append('%s: %s' % (stat, val)) return ', '.join(s) def _add_stats(self, stats): for k, v in stats.iteritems(): self.__dict__[k] = self.__dict__.get(k, 0) + v self._stats[k] = self.__dict__[k] self.__refresh_categories() def __str__(self): """ Simply returns the player's name, e.g., "T.Brady". """ return self.name def __add__(self, other): """ Adds two players together. Only two player objects that correspond to the same human (i.e., GameCenter identifier) can be added together. If two different players are added together, an assertion will be raised. The effect of adding two player objects simply corresponds to the sums of all statistical values. Note that as soon as two players have been added, the 'home' property becomes undefined. """ assert self.playerid == other.playerid assert type(self) == type(other) new_player = self.__class__(self.playerid, self.name, None) new_player._add_stats(self._stats) new_player._add_stats(other._stats) return new_player def __sub__(self, other): assert self.playerid == other.playerid assert type(self) == type(other) new_player = GamePlayerStats(self.playerid, self.name, self.home) new_player._add_stats(self._stats) for bk, bv in other._stats.iteritems(): new_player._stats[bk] -= bv if new_player._stats[bk] == 0: del new_player._stats[bk] else: new_player.__dict__[bk] = new_player._stats[bk] anydiffs = False for k, v in new_player._stats.iteritems(): if v > 0: anydiffs = True break if not anydiffs: return None return new_player def __getattr__(self, name): # If name has one of the categories as a prefix, then return # a default value of zero for cat in nflgame.statmap.categories: if name.startswith(cat): return 0 print name raise AttributeError
def _xml_plays(data, coach=True): """ Parses the XML raw string `data` given into an ordered dictionary of `nflvid.Play` objects corresponding to coach play timings. If `coach` is set to `False`, then play timings for the broadcast are retrieved. The dictionary is keyed by play id. A second return value, the ending time of the broadcast footage, is also returned. (This is used to compute an offset between the ArchiveTCIN time and when the play really starts.) """ if data is None: return None soup = bs4.BeautifulSoup(data) game_end_time = soup.find('dataset').get('endtime', None) if game_end_time is not None: game_end_time = PlayTime(game_end_time.strip()) # Load everything into a list first, since we need to look ahead to see # the next play's start time to compute the current play's duration. rows = [] for row in soup.find_all('row'): playid = row.find('id') if not playid: playid = row.get('playid', None) if not playid: continue playid = playid.strip() else: playid = playid.get_text().strip() if coach: start = row.find('catin') else: start = row.find('archivetcin') if not start: continue start = PlayTime(start.get_text().strip()) rows.append((playid, start, row)) # A predicate for determining whether to ignore a row or not in our final # result set. For example, timeouts take a lot of time but aren't needed # for play-by-play footage. def ignore(row): if 'playdescription' in row.attrs: if row['playdescription'].lower().startswith('timeout'): return True if row['playdescription'].lower().startswith('two-minute'): return True # Did we miss anything? if 'preplaybyplay' in row.attrs: if row['preplaybyplay'].lower().startswith('timeout'): return True return False d = OrderedDict() for i, (playid, start, row) in enumerate(rows): if ignore(row): continue end = None if i < len(rows) - 1: end = rows[i+1][1] d[playid] = Play(start, end, playid, game_end_time) return d
class PlayerStats(object): """ Player represents a single player and all of his statistical categories. Every player has 'playerid', 'name' and 'home' fields. Additionally, depending upon which statistical categories that player was involved in for the game, he'll have properties such as 'passing_tds', 'rushing_yds', 'defense_int' and 'kicking_fgm'. In order to know whether a paricular player belongs to a statical category, you may use the filtering methods of a player sequence or alternatively, use the has_cat method with arguments like 'passing', 'rushing', 'kicking', etc. (A player sequence in this case would be an instance of GenPlayerStats.) You may also inspect whether a player has a certain property by using the special __dict__ attribute. For example:: if 'passing_yds' in player.__dict__: # Do something with player.passing_yds """ def __init__(self, playerid, name, home, team): """ Create a new Player instance with the player id (from NFL.com's GameCenter), the player's name (e.g., "T.Brady") and whether the player is playing in a home game or not. """ self.playerid = playerid self.name = name self.home = home self.team = team self._stats = OrderedDict() self.player = None if self.playerid in nflgame.players: self.player = nflgame.players[self.playerid] def has_cat(self, cat): for f in self._stats: if f.startswith(cat): return True return False @property def guess_position(self): """ Guesses the position of this player based on the statistical categories present in this object when player meta is not present. Note that if this resorts to a guess, then it will be more effective on aggregate data rather than data from just a single play. (e.g., if a QB runs the ball, and that's the only data available, the position returned will be RB.) When a position is guessed, only the following positions will be returned: QB, RB, WR, DEF, K and P. """ # Look for the player meta first. Duh. if self.player is not None: return self.player.position stats = [ (self.passing_att, 'QB'), (self.rushing_att, 'RB'), (self.receiving_tar, 'WR'), (self.defense_tkl, 'DEF'), (self.defense_ast, 'DEF'), (self.kicking_tot, 'K'), (self.kicking_fga, 'K'), (self.punting_tot, 'P'), ] return sorted(stats, reverse=True)[0][1] @property def tds(self): """ Returns the total number of touchdowns credited to this player across all statistical categories. """ n = 0 for f, v in self.__dict__.iteritems(): if f.endswith('tds'): n += v return n @property def twopta(self): """ Returns the total number of two point conversion attempts for the passing, rushing and receiving categories. """ return (self.passing_twopta + self.rushing_twopta + self.receiving_twopta) @property def twoptm(self): """ Returns the total number of two point conversions for the passing, rushing and receiving categories. """ return (self.passing_twoptm + self.rushing_twoptm + self.receiving_twoptm) @property def twoptmissed(self): """ Returns the total number of two point conversion failures for the passing, rushing and receiving categories. """ return (self.passing_twoptmissed + self.rushing_twoptmissed + self.receiving_twoptmissed) @property def stats(self): """ Returns a dict of all stats for the player. """ return self._stats def formatted_stats(self): """ Returns a roughly-formatted string of all statistics for this player. """ s = [] for stat, val in self._stats.iteritems(): s.append('%s: %s' % (stat, val)) return ', '.join(s) def _add_stats(self, stats): for k, v in stats.iteritems(): self.__dict__[k] = self.__dict__.get(k, 0) + v self._stats[k] = self.__dict__[k] def _overwrite_stats(self, stats): for k, v in stats.iteritems(): self.__dict__[k] = v self._stats[k] = self.__dict__[k] def __str__(self): """ Simply returns the player's name, e.g., "T.Brady". """ return self.name def __add__(self, other): """ Adds two players together. Only two player objects that correspond to the same human (i.e., GameCenter identifier) can be added together. If two different players are added together, an assertion will be raised. The effect of adding two player objects simply corresponds to the sums of all statistical values. Note that as soon as two players have been added, the 'home' property becomes undefined if the two operands have different values of 'home'. """ assert self.playerid == other.playerid assert type(self) == type(other) if self.home != other.home: home = None else: home = self.home new_player = self.__class__(self.playerid, self.name, home, self.team) new_player._add_stats(self._stats) new_player._add_stats(other._stats) return new_player def __sub__(self, other): assert self.playerid == other.playerid assert type(self) == type(other) new_player = GamePlayerStats(self.playerid, self.name, self.home, self.team) new_player._add_stats(self._stats) for bk, bv in other._stats.iteritems(): if bk not in new_player._stats: # stat was taken away? ignore. continue new_player._stats[bk] -= bv if new_player._stats[bk] == 0: del new_player._stats[bk] else: new_player.__dict__[bk] = new_player._stats[bk] anydiffs = False for k, v in new_player._stats.iteritems(): if v > 0: anydiffs = True break if not anydiffs: return None return new_player def __getattr__(self, name): # If name has one of the categories as a prefix, then return # a default value of zero for cat in nflgame.statmap.categories: if name.startswith(cat): return 0 raise AttributeError
class PlayerStats(object): """ Player represents a single player and all of his statistical categories. Every player has 'playerid', 'name' and 'home' fields. Additionally, depending upon which statistical categories that player was involved in for the game, he'll have properties such as 'passing_tds', 'rushing_yds', 'defense_int' and 'kicking_fgm'. In order to know whether a paricular player belongs to a statical category, you may use the filtering methods of a player sequence or alternatively, use the has_cat method with arguments like 'passing', 'rushing', 'kicking', etc. (A player sequence in this case would be an instance of GenPlayerStats.) You may also inspect whether a player has a certain property by using the special __dict__ attribute. For example:: if 'passing_yds' in player.__dict__: # Do something with player.passing_yds """ def __init__(self, playerid, name, home): """ Create a new Player instance with the player id (from NFL.com's GameCenter), the player's name (e.g., "T.Brady") and whether the player is playing in a home game or not. """ self.playerid = playerid self.name = name self.home = home self._stats = OrderedDict() def has_cat(self, cat): return self.__dict__.get(cat, False) def __refresh_categories(self): for cat in nflgame.statmap.categories: for f in self.__dict__: if f.startswith(cat): self.__dict__[cat] = True break @property def tds(self): """ Returns the total number of touchdowns credited to this player across all statistical categories. """ n = 0 for f, v in self.__dict__.iteritems(): if f.endswith("tds"): n += v return n @property def stats(self): """ Returns a dict of all stats for the player. """ return self._stats def formatted_stats(self): """ Returns a roughly-formatted string of all statistics for this player. """ s = [] for stat, val in self._stats.iteritems(): s.append("%s: %s" % (stat, val)) return ", ".join(s) def _add_stats(self, stats): for k, v in stats.iteritems(): self.__dict__[k] = self.__dict__.get(k, 0) + v self._stats[k] = self.__dict__[k] self.__refresh_categories() def __str__(self): """ Simply returns the player's name, e.g., "T.Brady". """ return self.name def __add__(self, other): """ Adds two players together. Only two player objects that correspond to the same human (i.e., GameCenter identifier) can be added together. If two different players are added together, an assertion will be raised. The effect of adding two player objects simply corresponds to the sums of all statistical values. Note that as soon as two players have been added, the 'home' property becomes undefined. """ assert self.playerid == other.playerid assert type(self) == type(other) new_player = self.__class__(self.playerid, self.name, None) new_player._add_stats(self._stats) new_player._add_stats(other._stats) return new_player def __sub__(self, other): assert self.playerid == other.playerid assert type(self) == type(other) new_player = GamePlayerStats(self.playerid, self.name, self.home) new_player._add_stats(self._stats) for bk, bv in other._stats.iteritems(): new_player._stats[bk] -= bv if new_player._stats[bk] == 0: del new_player._stats[bk] else: new_player.__dict__[bk] = new_player._stats[bk] anydiffs = False for k, v in new_player._stats.iteritems(): if v > 0: anydiffs = True break if not anydiffs: return None return new_player def __getattr__(self, name): # If name has one of the categories as a prefix, then return # a default value of zero for cat in nflgame.statmap.categories: if name.startswith(cat): return 0 print name raise AttributeError