Update github_data to get stats and player rosters

src/queries/nflverse/github_data.py

@@ -8,6 +8,7 @@ duckdb.default_connection.execute("SET GLOBAL pandas_analyze_sample=100000")
 
 BASE_URL = "https://github.com/nflverse/nflverse-data/releases/download/"
 
+SEASON = "2023"
 
 FANTASY_POSITIONS = [
     "QB",
@@ -19,73 +20,40 @@ FANTASY_POSITIONS = [
 ]
 
 
-def get_snap_counts(season_int: int) -> pd.DataFrame:
-    df = duckdb.sql(f"SELECT * from snap_counts_snap_counts_{season_int}").df()
-    df["fantasy_position"] = df["position"].isin(FANTASY_POSITIONS)
-    return df
-
-
-def get_play_by_play(season_int: int) -> pd.DataFrame:
-    df = duckdb.sql(f"SELECT * from pbp_play_by_play_{season_int}").df()
-    return df
-
-
-def get_player_stats(season_int: int) -> pd.DataFrame:
-    df = duckdb.sql("SELECT * from player_stats_player_stats").df()
-    return df
-
-
-def get_ftn_charting(season_int: int) -> pd.DataFrame:
-    df = duckdb.sql(f"SELECT * from ftn_charting_ftn_charting_{season_int}").df()
-    return df
-
-
-def get_pbp_participation(season_int: int) -> pd.DataFrame:
+def get_weekly_rosters(season_int: int = SEASON) -> pd.DataFrame:
     df = duckdb.sql(
         f"""
+                    with grouped_stats as (
+                    select player_id, sum(fantasy_points) as fantasy_points
+                    from player_stats_player_stats_{season_int}
+                    group by player_id
+                    )
                     SELECT 
-                        a.*
-                        , b.*
-                        , 1 as count_col
-                    from pbp_participation_pbp_participation_{season_int} a
-                    left join pbp_play_by_play_{season_int} b
-                    on a.play_id = b.play_id
-                    and a.nflverse_game_id = b.game_id
-                    where b.week is not null
+                    team
+                    , wr.position
+                    , wr.jersey_number
+                    , wr.status
+                    , wr.full_name
+                    , wr.headshot_url
+                    , wr.week
+                    , wr.gsis_id
+                    , gs.fantasy_points
+                    from weekly_rosters_roster_weekly_{season_int} wr
+                    join grouped_stats gs
+                    on wr.gsis_id = gs.player_id
 """
     ).df()
     return df
 
 
-def get_nextgen_stats(season_int: int, stat_category: str) -> pd.DataFrame:
-    df = duckdb.sql(f"SELECT * from nextgen_stats_ngs_{stat_category} where season = {season_int}").df()
+def get_player_stats(season_int: int = SEASON) -> pd.DataFrame:
+    df = duckdb.sql(f"SELECT * from player_stats_player_stats_{season_int}").df()
     return df
 
 
-SEASON = "2023"
-
 NFLVERSE_ASSETS = [
-    ("ftn_charting", f"ftn_charting_{SEASON}.parquet"),
-    ("espn_data", "qbr_season_level.parquet"),
-    ("espn_data", "qbr_week_level.parquet"),
-    ("players", "players.parquet"),
-    ("pbp_participation", f"pbp_participation_{SEASON}.parquet"),
-    ("snap_counts", f"snap_counts_{SEASON}.parquet"),
+    ("weekly_rosters", f"roster_weekly_{SEASON}.parquet"),
     ("player_stats", f"player_stats_{SEASON}.parquet"),
-    ("player_stats", f"player_stats_def_{SEASON}.parquet"),
-    ("player_stats", f"player_stats_kicking_{SEASON}.parquet"),
-    ("pfr_advstats", "advstats_season_def.parquet"),
-    ("pfr_advstats", "advstats_season_pass.parquet"),
-    ("pfr_advstats", "advstats_season_rec.parquet"),
-    ("pfr_advstats", "advstats_season_rush.parquet"),
-    ("pfr_advstats", f"advstats_week_def_{SEASON}.parquet"),
-    ("pfr_advstats", f"advstats_week_pass_{SEASON}.parquet"),
-    ("pfr_advstats", f"advstats_week_rec_{SEASON}.parquet"),
-    ("pfr_advstats", f"advstats_week_rush_{SEASON}.parquet"),
-    ("pbp", f"play_by_play_{SEASON}.parquet"),
-    ("nextgen_stats", "ngs_passing.parquet"),
-    ("nextgen_stats", "ngs_receiving.parquet"),
-    ("nextgen_stats", "ngs_rushing.parquet"),
 ]