Add features to pitch leaderboard

app.py

@@ -11,4 +11,5 @@ if __name__ == '__main__':
     with gr.Tab('Pitch Leaderboard'):
       create_pitch_leaderboard()
 
+    gr.Markdown('Last updated: 2025-07-19')
   app.launch()

pitch_leaderboard.py

@@ -12,30 +12,51 @@ STATS = ['Count', 'Usage', 'SwStr%', 'Whiff%', 'CSW%', 'GB%', 'FB%', 'LD%']
 PCT_STATS = ['Usage', 'SwStr%', 'Whiff%', 'CSW%', 'GB%', 'FB%', 'LD%']
 STATS_WITH_PCTLS = ['SwStr%', 'Whiff%', 'CSW%', 'GB%', 'FB%', 'LD%']
 
-def gr_create_pitch_leaderboard(start_date, end_date, min_pitches, include_pitches):
+todo = '''
+**To-do**
+- Color cells according to percentiles
+'''
+
+def gr_create_pitch_leaderboard(start_date, end_date, min_pitches, pitcher_lr, include_pitches):
+  assert pitcher_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')
-  both, left, right = [
-    (
-      compute_pitch_stats(df, player_type='pitcher', min_pitches=min_pitches, pitch_class_type='specific')
-      .filter(pl.col('qualified') & (pl.col('ballKind').is_in(include_pitches)))
-      .drop('qualified')
-      .rename({'pitcher_name': 'Pitcher', 'count': 'Count', 'usage': 'Usage', 'ballKind': 'Pitch', 'general_ballKind': 'Pitch (General)'} | {f'{stat}_pctl': f'{stat} (Pctl)' for stat in STATS_WITH_PCTLS})
-      .with_columns(
-        pl.col(stat).mul(100).round(1)
-        for stat in PCT_STATS + [f'{stat} (Pctl)' for stat in STATS_WITH_PCTLS]
-      )
-      [['pitId', 'ballKind_code', 'Pitcher', 'Pitch', 'Pitch (General)', 'Count', 'Usage'] + STATS_WITH_PCTLS]
-    )
-    for df
-    in [data, data.filter(pl.col('batLR') == 'l'), data.filter(pl.col('batLR') == 'r')]
-  ]
+  if pitcher_lr != 'Both':
+    data = data.filter(pl.col('batLR') == pitcher_lr[0].lower())
+    
+  # both, left, right = [
+    # (
+      # compute_pitch_stats(df, player_type='pitcher', min_pitches=min_pitches, pitch_class_type='specific')
+      # .filter(pl.col('qualified') & (pl.col('ballKind').is_in(include_pitches)))
+      # .drop('qualified')
+      # .rename({'pitcher_name': 'Pitcher', 'count': 'Count', 'usage': 'Usage', 'ballKind': 'Pitch', 'general_ballKind': 'Pitch (General)'} | {f'{stat}_pctl': f'{stat} (Pctl)' for stat in STATS_WITH_PCTLS})
+      # .with_columns(
+        # pl.col(stat).mul(100).round(1)
+        # for stat in PCT_STATS + [f'{stat} (Pctl)' for stat in STATS_WITH_PCTLS]
+      # )
+      # [['pitId', 'ballKind_code', 'Pitcher', 'Pitch', 'Pitch (General)', 'Count', 'Usage'] + STATS_WITH_PCTLS]
+    # )
+    # for df
+    # in [data, data.filter(pl.col('batLR') == 'l'), data.filter(pl.col('batLR') == 'r')]
+  # ]
+  # pitch_stats = (
+    # both
+    # .join(left, on=['pitId', 'ballKind_code'], suffix=' (LHH)', how='full')
+    # .join(right, on=['pitId', 'ballKind_code'], suffix=' (RHH)', how='full')
+    # .drop('pitId', 'ballKind_code', *list(chain.from_iterable([[f'{col} ({handedness}HH)' for col in ['pitId', 'ballKind_code', 'Pitcher', 'Pitch', 'Pitch (General)']] for handedness in ('L', 'R')])))
+  # )
   pitch_stats = (
-    both
-    .join(left, on=['pitId', 'ballKind_code'], suffix=' (LHH)', how='full')
-    .join(right, on=['pitId', 'ballKind_code'], suffix=' (RHH)', how='full')
-    .drop('pitId', 'ballKind_code', *list(chain.from_iterable([[f'{col} ({handedness}HH)' for col in ['pitId', 'ballKind_code', 'Pitcher', 'Pitch', 'Pitch (General)']] for handedness in ('L', 'R')])))
+        compute_pitch_stats(data, player_type='pitcher', min_pitches=min_pitches, pitch_class_type='specific')
+        .filter(pl.col('qualified') & (pl.col('ballKind').is_in(include_pitches)))
+        .drop('pitId', 'ballKind_code', 'qualified')
+        .rename({'pitcher_name': 'Pitcher', 'count': 'Count', 'usage': 'Usage', 'ballKind': 'Pitch', 'general_ballKind': 'Pitch (General)'} | {f'{stat}_pctl': f'{stat} (Pctl)' for stat in STATS_WITH_PCTLS})
+        .with_columns(
+          pl.col(stat).mul(100).round(1)
+          for stat in PCT_STATS + [f'{stat} (Pctl)' for stat in STATS_WITH_PCTLS]
+        )
+        [['Pitcher', 'Pitch', 'Pitch (General)', 'Count', 'Usage'] + STATS_WITH_PCTLS]
   )
   return pitch_stats
 
@@ -50,14 +71,16 @@ def create_pitch_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')
-      min_pitches = gr.Number(100, label='Min. Pitches', precision=0, minimum=0)
     with gr.Row():
-      include_pitches = gr.CheckboxGroup(pitch_types, value=pitch_types, label='Pitches', scale=2)
-      # with gr.Column(scale=1):
-        # all_pitches = gr.Checkbox(label='Select all pitches')
+      include_pitches = gr.CheckboxGroup(pitch_types, value=pitch_types, label='Pitches', scale=3)
+      with gr.Column(scale=1):
+        all_pitches = gr.Button('Select/Deselect all pitches', scale=1)
+        min_pitches = gr.Number(100, label='Min. Pitches', precision=0, minimum=0)
+        pitcher_lr = gr.Radio(['Both', 'Left', 'Right'], value='Both', label='Batter handedness')
         
     search = gr.Button('Search')
     leaderboard = gr.DataFrame(
+      pl.DataFrame({'Pitcher': [], 'Pitch': []}),
       # gr_create_pitch_leaderboard(start_date=start_date_init, end_date=end_date_init, min_pitches=100),
       column_widths=[200]*3 + [100]*(3*len(STATS)),
       show_copy_button=True,
@@ -66,10 +89,9 @@ def create_pitch_leaderboard():
     )
 
 
-    search.click(gr_create_pitch_leaderboard, inputs=[start_date, end_date, min_pitches, include_pitches], outputs=leaderboard)
-    # include_pitches.input(lambda _pitch_types: (_pitch_types == pitch_types), inputs=include_pitches, outputs=all_pitches)
-    # all_pitches.input(lambda _all_pitches : pitch_types, inputs=all_outputs=all_pitches)
-  
+    search.click(gr_create_pitch_leaderboard, inputs=[start_date, end_date, min_pitches, pitcher_lr, include_pitches], outputs=leaderboard)
+    all_pitches.click(lambda _pitch_types : [] if _pitch_types == pitch_types else pitch_types, inputs=include_pitches, outputs=include_pitches)
+    gr.Markdown(todo)
   return app
 
 if __name__ == '__main__':

pitcher_overview.py

@@ -6,6 +6,17 @@ from data import SEASONS, data_df
 
 from plotting import create_pitcher_overview_card
 
+notes = '''**Limitations**
+- Foreign players names are in Hebpurn romanization. Contact me if you need a card for a foreign player.
+
+**To-do**
+- Fix names of foreign playeres
+- Add teams insignias
+- Measure percentiles per pitcher handedness
+- Allow for arbitrary date ranges
+- Improve readability of pitch velocities
+'''
+
 def dummy(*inputs):
   return inputs
 
@@ -31,7 +42,7 @@ def gr_create_pitcher_overview_card(name, season):
   
 def create_pitcher_overview(data_df):
   with gr.Blocks() as app:
-    gr.Markdown('# Pitcher overview')
+    gr.Markdown('# Pitcher Overview')
 
     with gr.Row():
       with gr.Column():
@@ -43,21 +54,7 @@ def create_pitcher_overview(data_df):
         # season_end = gr.Dropdown(SEASONS, label='Season end')
         # game_type = gr.Dropdown(['Spring Training', 'Regular Season', 'Postseason'], label='Game Type'])
         view = gr.Button('View')
-        gr.Markdown(
-          '''
-**Limitations**
-- Foreign players names are in Hebpurn romanization. Contact me if you need a card for a foreign player.
-
-**To-do**
-- Fix names of foreign playeres
-- Add teams insignias
-- Measure percentiles per pitcher handedness
-- Allow for arbitrary date ranges
-- Improve readability of pitch velocities
-
-Last updated: 2025-07-19
-'''
-        )
+        gr.Markdown(notes)
 
       with gr.Column():
         overview_card = gr.Image(label='Overview')