Add files

app.py

@@ -0,0 +1,12 @@
+import gradio as gr
+
+from data import data_df
+from pitcher_overview import create_pitcher_overview
+
+
+with gr.Blocks() as demo:
+
+  create_pitcher_overview(data_df)
+
+if __name__ == '__main__':
+  demo.launch()

convert.py

@@ -0,0 +1,259 @@
+aux_global_id_to_code = {
+    7911: 'G',
+    7912: 'S',
+    7913: 'DB',
+    7914: 'D',
+    7915: 'T',
+    7916: 'C',
+    7917: 'F',
+    7918: 'E',
+    7919: 'L',
+    7920: 'M',
+    7921: 'B',
+    7922: 'H',
+    7925: 'PL',
+    7926: 'CL'
+}
+
+ball_kind = {
+    -1: '-',
+    31: 'Fastball (4-seam)',
+    32: 'Slider',
+    33: 'Vertical Slider',
+    34: 'Slurve',
+    35: 'Curve',
+    36: 'Slow Curve',
+    37: 'Power Curve',
+    38: 'Knuckle Curve',
+    39: 'Forkball',
+    40: 'Splitter',
+    41: 'Changeup',
+    42: 'Sinker',
+    43: 'Screwball',
+    44: 'Palmball',
+    45: 'Knuckleball',
+    46: 'Shootball',
+    47: 'Fastball (2-seam)',
+    48: 'Fastball (1-seam)',
+    49: 'Cutter',
+    50: 'Eephus', # technically "super" eephus but I haven't encountered a normal one yet
+    51: 'Hard Sinker',
+    52: 'Hard Slider',
+}
+
+ball_kind_code = {
+    -1: '-',
+    31: 'FF',
+    32: 'SL',
+    33: 'VS',
+    34: 'SV',
+    35: 'CU',
+    36: 'SC',
+    37: 'PC',
+    38: 'KC',
+    39: 'FO',
+    40: 'FS',
+    41: 'CH',
+    42: 'SI',
+    43: 'SB',
+    44: 'PB',
+    45: 'KN',
+    46: 'SH',
+    47: 'FT',
+    48: 'FW',
+    49: 'FC',
+    50: 'EP', # technically "super" eephus but I haven't encountered a normal one yet
+    51: 'HS',
+    52: 'HL'
+}
+
+lr = {1: 'l', 2: 'r'}
+
+presult = {
+    0: 'None',
+    101: 'Foul',
+    102: 'Single',
+    103: 'Hit by pitch',
+    104: 'Double',
+    105: 'Triple',
+    106: 'Home run',
+    107: 'Error',
+    108: 'Groundout',
+    109: 'Flyout',
+    110: 'Lineout',
+    111: 'Groundout (Double play)',
+    112: 'Foul fly',
+    113: 'Foul line (?)',
+    114: 'Sacrifice bunt',
+    115: 'Sacrifice fly',
+    116: 'Swinging strike',
+    117: 'Looking strike',
+    118: 'Ball',
+    119: 'Walk',
+    120: 'Balk',
+    121: 'Batter interference',
+    122: 'Catcher interference',
+    123: 'Uncaught third strike',
+    124: 'Sacrifice hit error',
+    125: 'Sacrifice fly, error',
+    126: "Fielder's choice",
+    128: "Sacrifice fielder's choice",
+    129: 'Bunt strikeout',
+    130: 'Swinging strikeout',
+    131: 'Looking strikeout',
+    133: 'Inside-the-park home run',
+    134: 'Pitcher delay',
+    135: 'Balk',
+    139: 'Intentional walk',
+    140: 'Groundout (Double play)',
+    141: 'Unknown'
+}
+
+bresult = {
+    0: '空振り三振',
+    1: '単打', # 1b gb p
+    2: '単打', # 1b gb c
+    3: '単打', # 1b gb 1b
+    4: '単打', # 1b gb 2b
+    5: '単打', # 1b gb 3b? ld
+    6: '単打', # 1b gb ss
+    7: '単打', # 1b gb lf? ld
+    8: '単打', # 1b gb cf
+    9: '単打', # 1b gb rf
+    10: '２塁打', # 2b gb p
+
+    12: '２塁打', # 2b gb 1b
+    13: '２塁打', # 2b gb 2b
+    14: '２塁打', # 2b gb 3b? ld
+    15: '２塁打', # 2b gb ss
+    16: '２塁打', # 2b gb lf? ld
+    17: '２塁打', # 2b gb cf
+    18: '２塁打', # 2b gb rf
+
+    22: '３塁打', # 3b gb 2b
+
+    25: '３塁打', # 3b gb lf? ld
+    26: '３塁打', # 3b gb cf
+    27: '３塁打', # 3b gb rf
+    28: '本塁打(ランニング)', # ihr lf?
+    29: '本塁打(ランニング)', # ihr cf?
+    30: '本塁打(ランニング)', # ihr rf?
+
+    31: '空振り三振',
+    32: '振逃げ',
+    33: '振逃げ',
+    34: '四球',
+    35: 'フライ', # fb cf
+    36: '死球',
+    37: '打撃妨害',
+    38: '守備妨害',
+
+    39: '犠打野選',
+    40: 'ゴロ', #gb p
+    41: 'ゴロ', #gb c
+    42: 'ゴロ', #gb 1b
+    43: 'ゴロ', #gb 2b
+    44: 'ゴロ', #gb 3b
+    45: 'ゴロ', #gb ss
+    46: 'ゴロ', #gb lf
+    47: 'ゴロ', #gb cf
+    48: 'ゴロ', #gb rf
+
+    49: 'フライ', # fb p
+    50: 'フライ', # fb c
+    51: 'フライ', # fb 1b
+    52: 'フライ', # fb 2b
+    53: 'フライ', # fb 3b
+    54: 'フライ', # fb ss
+    55: 'フライ', # fb lf
+    56: 'フライ', # fb cf
+    57: 'フライ', # fb rf
+    58: 'ライナー', # ld p
+    60: 'ライナー', # ld 1b
+    61: 'ライナー', # ld 2b
+    62: 'ライナー', # ld 3b
+    63: 'ライナー', # ld ss
+    64: 'ライナー', # ld lf
+
+    66: 'ライナー', # ld rf
+    67: '犠打',
+    68: '犠打',
+    69: '犠打',
+    70: '犠打',
+    71: '犠打',
+    79: '犠飛', # sac fly lf
+    80: '犠飛', # sac fly cf
+    81:'犠飛', # sac fly rf
+    82: '邪飛',
+    83: '邪飛',
+    84: '邪飛',
+    85: '邪飛',
+    86: '邪飛',
+    87: '邪飛',
+    88: '邪飛',
+
+    90: '邪飛',
+    91: 'ゴロ(併殺打)',
+    92: 'ゴロ(併殺打)',
+    93: 'ゴロ(併殺打)',
+    94: 'ゴロ(併殺打)',
+    95: 'ゴロ(併殺打)',
+    96: 'ゴロ(併殺打)',
+
+    100: '失策出塁',
+    101: '失策出塁',
+    102: '失策出塁',
+    103: '失策出塁',
+    104: '失策出塁',
+    105: '失策出塁',
+    106: '失策出塁',
+    107: '失策出塁',
+    108: '失策出塁',
+    109: '野選',
+    110: '野選',
+    111: '野選',
+    112: '野選',
+    113: '野選',
+    114: '野選',
+    115: '犠打失策',
+    116: '犠飛失策',
+
+    208: '本塁打', # hr lf?
+
+    210: '本塁打', # hr cf?
+    212: '本塁打', # hr rf?
+
+    214: '見送り三振',
+    215: '犠打野選',
+    216: '犠打野選',
+
+    218: '犠打野選',
+
+    229: '振逃げ',
+
+    234: '邪直',
+
+    236: '邪直',
+
+    241: '３バント失敗',
+
+    258: '規則違反',
+
+    267: '打撃妨害',
+
+    338: '?'
+}
+
+game_kind = {
+    1: 'CL Regular Season',
+    2: 'PL Regular Season',
+    3: 'Nippon Series',
+    4: 'All-Star Game',
+    5: 'Spring Training',
+    11: 'Farm Championship',
+    26: 'Interleague',
+    35: 'CL Climax Series First Stage',
+    36: 'CL Climax Series Final Stage',
+    37: 'PL Climax Series First Stage',
+    38: 'PL Climax Series Final Stage'
+}

data.py

@@ -0,0 +1,168 @@
+import polars as pl
+from glob import glob
+import os
+from tqdm.auto import tqdm
+
+from convert import aux_global_id_to_code, presult, ball_kind, ball_kind_code, lr, game_kind
+
+DATA_PATH = os.path.expanduser('~/Documents/npb_data_collector/npb')
+# SEASONS = list(range(2021, 2025+1))
+SEASONS = [2021, 2022, 2023, 2024, 2025]
+
+data_df = pl.DataFrame()
+text_df = pl.DataFrame()
+aux_df = pl.DataFrame()
+sched_df = pl.DataFrame()
+aux_sched_df = pl.DataFrame()
+
+for season in tqdm(SEASONS):
+  _data_df = pl.read_parquet(os.path.join(DATA_PATH, str(season), 'pbp_data.parquet'))
+  data_df = pl.concat((data_df, _data_df))
+
+  _text_df = pl.read_parquet(os.path.join(DATA_PATH, str(season), 'pbp_text.parquet'))
+  text_df = pl.concat((text_df, _text_df))
+
+  _aux_df = pl.read_parquet(os.path.join(DATA_PATH, str(season), 'pbp_aux.parquet'))
+  aux_df = pl.concat((aux_df, _aux_df), how='diagonal_relaxed')
+
+  _sched_df = pl.read_parquet(os.path.join(DATA_PATH, str(season), 'schedule.parquet'))
+  sched_df = pl.concat((sched_df, _sched_df))
+
+  _aux_sched_df = pl.read_parquet(os.path.join(DATA_PATH, str(season), 'aux_schedule.parquet'))
+  aux_sched_df = pl.concat((aux_sched_df, _aux_sched_df))
+
+
+# sched_df = sched_df.
+
+aux_df = (
+    aux_df
+    .filter(pl.col('type') != 'RUNNER')
+    .join(aux_sched_df[['gameGlobalId', 'gameDate']], on='gameGlobalId')
+    .with_columns(
+        pl.col('gameDate').str.to_date().dt.strftime('%Y%m%d'),
+        pl.col('home').struct.field('globalId').replace_strict(aux_global_id_to_code).alias('home'),
+        pl.col('visitor').struct.field('globalId').replace_strict(aux_global_id_to_code).alias('visitor'),
+        pl.when(pl.col('tob') == 'Top').then(pl.lit('1')).otherwise(pl.lit('2')).alias('tob_code'),
+    )
+    .filter(
+        # pl.col('pitch').struct.field('count') > 0
+
+        # either one alone should be enough but let's use them together to be safe
+        ~((pl.col('code') == 98) & (pl.col('id') == 1))
+    )
+    .with_columns(
+        (pl.col('pitch').struct.field('count') == 1).cum_sum().over(['gameGlobalId', 'inning', 'tob']).alias('pa_count')
+    )
+    .with_columns(
+        pl.col('code').is_in([6402, 6404, 6406, 6405]).any().over(['gameGlobalId', 'inning', 'tob', 'pa_count']).alias('ibb')
+    )
+    .with_columns(
+        pl.when(~pl.col('ibb')).then(pl.col('pitch').struct.field('count') == 1).cum_sum().over(['gameGlobalId', 'inning', 'tob']).alias('new_pa_count')
+    )
+    .with_columns(
+        pl.len().over(['gameGlobalId', 'inning', 'tob', 'new_pa_count']).alias('pa_pitches'),
+        pl.max('new_pa_count').over(['gameGlobalId', 'inning', 'tob']).alias('inning_pas')
+    )
+    .with_columns(
+        (
+            pl.col('gameDate') + '_' + \
+            pl.col('visitor') + '_' + \
+            pl.col('home') + '_' + \
+            pl.col('inning').str.zfill(2) + pl.when(pl.col('tob') == 'Top').then(pl.lit('1')).otherwise(pl.lit('2')) + pl.col('new_pa_count').cast(pl.String).str.zfill(2) + '_' +\
+            pl.col('pitch').struct.field('count').cast(pl.String)
+        ).alias('universal_code'),
+        (
+            pl.col('gameDate') + '_' + \
+            pl.col('visitor') + '_' + \
+            pl.col('home') + '_' + \
+            pl.col('inning').str.zfill(2) + pl.when(pl.col('tob') == 'Top').then(pl.lit('1')).otherwise(pl.lit('2'))
+        ).alias('inning_code'),
+        (
+            pl.col('gameDate') + '_' + \
+            pl.col('visitor') + '_' + \
+            pl.col('home') + '_' + \
+            pl.col('inning').str.zfill(2) + pl.when(pl.col('tob') == 'Top').then(pl.lit('1')).otherwise(pl.lit('2')) + pl.col('new_pa_count').cast(pl.String).str.zfill(2)
+        ).alias('pa_code')
+    )
+)
+
+data_df = data_df
+
+data_df = (
+    data_df
+    .with_columns(
+        *[
+            pl.col(col).cast(pl.Int32)
+            for col
+            in ['gameId', 'ballKind', 'ballSpeed', 'x', 'y', 'presult', 'bresult', 'battedX', 'battedY']
+        ],
+        pl.col('UpdatedAt').str.to_datetime(),
+        pl.col('fiveDigitSerialNumber').str.slice(offset=0, length=3).alias('half_inning'),
+        pl.col('fiveDigitSerialNumber').str.slice(offset=3, length=2).alias('batter'),
+    )
+    .with_columns(
+        # pl.count('ID').over(['gameId', 'fiveDigitSerialNumber']).alias('pa_pitches')
+        (~pl.col('presult').is_in([0])).sum().over(['gameId', 'fiveDigitSerialNumber']).alias('pa_pitches'),
+        pl.col('presult').is_in([139]).any().over(['gameId', 'fiveDigitSerialNumber']).alias('ibb')
+    )
+    .filter(
+        (pl.col('pa_pitches') > 0)
+    )
+    .with_columns(
+        pl.when(~pl.col('ibb')).then(pl.col('batter'))
+    )
+    .with_columns(
+        pl.when(~pl.col('ibb')).then(pl.col('batter').rank('dense')).over(['gameId', 'half_inning']).cast(pl.String).str.zfill(2).alias('new_batter')
+    )
+    .with_columns(
+        (pl.col('half_inning') + pl.col('new_batter')).alias('newFiveDigitSerialNumber')
+    )
+    .with_columns(pl.max('new_batter').cast(pl.Int32).over(['gameId', pl.col('newFiveDigitSerialNumber').str.slice(offset=0, length=3)]).alias('inning_pas'))
+    .join(sched_df[['GameID', 'HomeTeamNameES', 'VisitorTeamNameES']].rename({'GameID': 'gameId'}), on='gameId')
+    .with_columns(pl.col('UpdatedAt').dt.strftime('%Y%m%d').alias('date'))
+    .with_columns(
+        (pl.col('date') + '_' + pl.col('VisitorTeamNameES') + '_' + pl.col('HomeTeamNameES') + '_' + pl.col('newFiveDigitSerialNumber')).alias('universal_code') + '_' + pl.col('atBatBallCount'),
+        (pl.col('date') + '_' + pl.col('VisitorTeamNameES') + '_' + pl.col('HomeTeamNameES') + '_' + pl.col('newFiveDigitSerialNumber').str.slice(offset=0, length=3)).alias('inning_code'),
+        (pl.col('date') + '_' + pl.col('VisitorTeamNameES') + '_' + pl.col('HomeTeamNameES') + '_' + pl.col('newFiveDigitSerialNumber')).alias('pa_code')
+    )
+    .join(
+        (
+            aux_df.filter(~pl.col('ibb'))[['universal_code', 'battingResult', 'inning_pas', 'pa_pitches']]
+            .rename({'battingResult': 'aux_bresult', 'inning_pas': 'aux_inning_pas', 'pa_pitches': 'aux_pa_pitches'})
+        ),
+        on='universal_code',
+        how='left'
+    )
+    .with_columns(
+
+    )
+    .join(
+        text_df[['GameID', 'GameKindID']].with_columns(
+            pl.col('GameID').cast(pl.Int32),
+            pl.col('GameKindID').cast(pl.Int32),
+        ).unique(),
+        how='left',
+        left_on='gameId',
+        right_on='GameID'
+    )
+    .with_columns(pl.col('GameKindID').replace_strict(game_kind).alias('GameKindName'))
+    .with_columns(
+        pl.when((pl.col('inning_pas') == pl.col('aux_inning_pas')) & (pl.col('pa_pitches') == pl.col('aux_pa_pitches')))
+        .then('aux_bresult')
+        .alias('aux_bresult'),
+
+        pl.col('x').add(-100).mul(-1),
+        pl.col('y').neg().add(250),
+        pl.col('presult').replace_strict(presult),
+        pl.col('ballKind').replace_strict(ball_kind),
+        pl.col('ballKind').replace_strict(ball_kind_code).alias('ballKind_code'),
+        pl.col('batLR').replace_strict(lr),
+        
+        pl.when(pl.col('GameKindName').str.contains('Regular Season') | (pl.col('GameKindName') == 'Interleague'))
+        .then(pl.lit('Regular Season'))
+        .when(~pl.col('GameKindName').is_in(['Spring Training', 'All-Star Game']))
+        .then(pl.lit('Postseason'))
+        .otherwise('GameKindName')
+        .alias('coarse_game_kind')
+    )
+)

pitcher_overview.py

@@ -0,0 +1,36 @@
+import gradio as gr
+
+from data import SEASONS
+
+def dummy(*inputs):
+  return inputs
+
+def adjust_season_end_based_on_season_start(season_start, season_end):
+  return max(season_start, season_end)
+
+def adjust_season_start_based_on_season_end(season_end, season_start):
+  return min(season_start, season_end)
+    
+  
+def create_pitcher_overview(data_df):
+  with gr.Blocks() as app:
+    gr.Markdown('Test')
+
+    name = gr.Dropdown(sorted(data_df['pitId'].unique().to_list()), label='Name')
+    season_start = gr.Dropdown(SEASONS, label='Season start')
+    season_end = gr.Dropdown(SEASONS, label='Season end')
+
+    season_start.input(adjust_season_end_based_on_season_start, inputs=[season_start, season_end], outputs=season_end)
+    season_end.input(adjust_season_start_based_on_season_end, inputs=[season_end, season_start], outputs=season_start)
+
+    game_type = gr.Dropdown(['Spring Training', 'Regular Season', 'Postseason'], label='Game Type'])
+    
+    generate = gr.Button('Generate')
+
+    dummy_io = [name, season_start, season_end, game_type
+    generate.click(dummy, inputs=dummy_io, outputs=dummy_io)
+
+  return app
+
+if __name__ == '__main__':
+  create_pitcher_overview().launch()