player_team_leaderboard.py

import gradio as gr
import polars as pl
import numpy as np

from datetime import datetime
from functools import partial

from data import data_df
from stats import compute_player_stats, filter_data_by_date_and_game_kind
from convert import team_names_short_to_color, get_text_color_from_team
from plotting import stat_cmap


TEAMS  = [
    'Yomiuri',
    'Yakult',
    'DeNA',
    'Chunichi',
    'Hanshin',
    'Hiroshima',
    'Nipponham',
    'Rakuten',
    'Seibu',
    'Lotte',
    'ORIX',
    'SoftBank'
]


notes = '''**Limitations**
- [Column widths get messed up when filtering](https://github.com/gradio-app/gradio/issues/11564)

**To-do**
- Fix IP calculation
- Add post-season
'''


def get_col_width(col, player_team_type):
  match col:
    case 'Pitcher' | 'Batter':
      return 125
    case 'Team':
      return 75
    case 'Throws' | 'Bats':
      return 60
    case 'IP':
      return 60 if player_team_type == 'team pitching' else 50
    case _:
      return max(50, 10*len(col))


def create_player_team_leaderboard_app(player_team_type):
  pitching = player_team_type in ('pitcher', 'team pitching')
  team = 'team' in player_team_type

  # stats
  if pitching:
    pct_stats = ['K%', 'BB%', 'Swing%', 'Z-Swing%', 'Chase%', 'Contact%', 'Z-Contact%', 'O-Contact%', 'SwStr%', 'Whiff%', 'CSW%', 'GB%', 'FB%', 'LD%', 'Zone%', 'Arm%', 'Glove%', 'High%', 'Low%', 'MM%']
    stats_with_pctls = ['FB Velo', 'K%', 'BB%', 'Swing%', 'Z-Swing%', 'Chase%', 'Contact%', 'Z-Contact%', 'O-Contact%', 'SwStr%', 'Whiff%', 'CSW%', 'GB%', 'FB%', 'LD%', 'Zone%']
    cols = ['Pitcher', 'Team', 'Throws', 'IP', 'TBF', 'FB Velo', 'K%', 'BB%', 'Swing%', 'Z-Swing%', 'Chase%', 'Contact%', 'Z-Contact%', 'O-Contact%', 'SwStr%', 'Whiff%', 'CSW%', 'GB%', 'FB%', 'LD%', 'Zone%', 'Arm%', 'Glove%', 'High%', 'Low%', 'MM%']
    if team:
      cols = [col for col in cols if col not in ('Pitcher', 'Throws')]
  else:
    pct_stats = ['K%', 'BB%', 'Swing%', 'Z-Swing%', 'Chase%', 'Contact%', 'Z-Contact%', 'O-Contact%', 'SwStr%', 'Whiff%', 'CSW%', 'GB%', 'FB%', 'LD%', 'Zone%', 'Arm%', 'Glove%', 'High%', 'Low%', 'MM%']
    stats_with_pctls = ['K%', 'BB%', 'Swing%', 'Z-Swing%', 'Chase%', 'Contact%', 'Z-Contact%', 'O-Contact%', 'SwStr%', 'Whiff%', 'CSW%', 'GB%', 'FB%', 'LD%', 'Zone%']
    cols = ['Batter', 'Team', 'Bats', 'PA', 'K%', 'BB%', 'Swing%', 'Z-Swing%', 'Chase%', 'Contact%', 'Z-Contact%', 'O-Contact%', 'SwStr%', 'Whiff%', 'CSW%', 'GB%', 'FB%', 'LD%']
    if team:
      cols = [col for col in cols if col not in ('Batter', 'Bats')]

  # col names
  player_type = 'pitcher' if pitching else 'batter'
  id_col = f'{player_type[:3].lower()}Id' if not team else f'{player_type}_team_name_short'
  qual_name = 'IP' if pitching else 'PA'
  
  def gr_create_player_team_leaderboard(start_date, end_date, min_qual, qualified, pitcher_lr='Both', batter_lr='Both', include_teams=None):
    assert pitcher_lr in ['Both', 'Left', 'Right']
    assert batter_lr in ['Both', 'Left', 'Right']

    data = data_df.filter(pl.col('ballKind_code') != '-')
    
    data = filter_data_by_date_and_game_kind(data, start_date=start_date, end_date=end_date, game_kind='Regular Season')
    
    rename = {f'{player_type}_team_name_short': 'Team'}
    if not team:
      rename[f'{player_type}_name'] = player_type.title()
    if pitching:
      rename['PA'] = 'TBF'

    # typically "qualified" should be a valid input for min_ip for the current function, 
    # but we separate it from a numerical min_ip argument for API compabtibility
    pitcher_stats = (
          compute_player_stats(
            data, player_type=player_team_type, 
            pitcher_lr='both' if pitcher_lr=='Both' else pitcher_lr[0].lower(), 
            batter_lr='both' if batter_lr == 'Both' else batter_lr[0].lower(), 
            qual='qualified' if qualified else min_qual,
            group_by_team=not team
          )
          .filter(pl.col('qualified'))
          .drop(['qualified'] + ([id_col] if not team else []))
          .rename(rename)
          .with_columns(
            pl.col(stat).mul(100)
            for stat in pct_stats
          )
    )

    # if not team:
    if include_teams is not None:
      pitcher_stats = pitcher_stats.filter(pl.col('Team').is_in(include_teams))

    styling = []
    for i, row in enumerate(pitcher_stats[cols].iter_rows()):
      styling_row = []
      for col, item in zip(pitcher_stats[cols].columns, row):
        _styling = 'font-size: 0.75em; '
        if col in stats_with_pctls:
          r, g, b = (stat_cmap([pitcher_stats[f'{col}_pctl'][i]])[0, :3]*255).astype(np.uint8)
          styling_row.append(f'background-color: rgba({r}, {g}, {b})')
        elif col == 'Team':
          styling_row.append(f'color: {get_text_color_from_team(item)}; background-color: {team_names_short_to_color[item]}')
        else:
          styling_row.append('')
      styling.append(styling_row)

    display_value = []
    for row in pitcher_stats[cols].iter_rows():
      display_value_row = []
      for col, item in zip(cols, row):
        if col in pct_stats:
          display_value_row.append(f'{item:.1f}%')
        elif isinstance(item, float):
          display_value_row.append(f'{item:.1f}')
        else:
          display_value_row.append(item)
      display_value.append(display_value_row)

    value = {
        'data': pitcher_stats[cols].rows(),
        'headers': cols,
        'metadata': {
            'styling': styling,
            'display_value': display_value,
        }
    }
    
    return value


  now = datetime.now()
  start_datetime_init = datetime(now.year, 1, 1)
  end_datetime_init = now
  with gr.Blocks() as app:
    gr.Markdown(f'# {player_team_type.title()} Leaderboard')
    with gr.Row():
      start_date = gr.DateTime(start_datetime_init, include_time=False, type='datetime', label='Start')
      end_date = gr.DateTime(end_datetime_init, include_time=False, type='datetime', label='End')
    with gr.Row():
      if not team:
        with gr.Group():
          min_ip = gr.Number(100, label=f'Min. {qual_name}', precision=0, minimum=0, interactive=False)
          qualified = gr.Checkbox(True, label='Qualified')
      else:
        min_ip = gr.State(0)
        qualified = gr.State(False)
      with gr.Group():
        pitcher_lr = gr.Radio(['Both', 'Left', 'Right'], value='Both', label='Pitcher handedness') 
        batter_lr = gr.Radio(['Both', 'Left', 'Right'], value='Both', label='Batter handedness')
    with gr.Row():
      include_teams = gr.CheckboxGroup(TEAMS, value=TEAMS, label='Teams', scale=3)
      all_teams = gr.Button('Select/Deselect all teams')
        
    search = gr.Button('Search')
    pin_columns = gr.Checkbox(True, label='Pin columns')
    leaderboard = gr.DataFrame(
      pl.DataFrame({'Pitcher': [], 'Pitch': []}),
      column_widths=[get_col_width(col, player_team_type) for col in cols],
      show_copy_button=True,
      show_search='filter',
      pinned_columns=2,
      elem_id='leaderboard'
    )

    gr.Markdown(notes)

    search.click(gr_create_player_team_leaderboard, inputs=[start_date, end_date, min_ip, qualified, pitcher_lr, batter_lr, include_teams], outputs=leaderboard)
    all_teams.click(lambda _teams : [] if _teams == TEAMS else TEAMS, inputs=include_teams, outputs=include_teams)
    qualified.change(lambda qualified: gr.Number(interactive=not qualified), inputs=qualified, outputs=min_ip)
    pin_columns.input(
      lambda pin: gr.DataFrame(pinned_columns=2 if pin else None),
      inputs=pin_columns,
      outputs=leaderboard
    )
    
  return app

create_pitcher_leaderboard = partial(
  create_player_team_leaderboard_app,
  player_team_type='pitcher'
)

create_team_pitching_leaderboard = partial(
  create_player_team_leaderboard_app,
  player_team_type='team pitching'
)

if __name__ == '__main__':
  app = foo()
  app.launch()