Jon Solow commited on
Commit
c998657
·
1 Parent(s): 962e128

Replace get_parquet_github with duckdb queries

Browse files
Files changed (1) hide show
  1. src/queries/nflverse/github_data.py +8 -10
src/queries/nflverse/github_data.py CHANGED
@@ -7,12 +7,6 @@ from typing import Callable
7
  BASE_URL = "https://github.com/nflverse/nflverse-data/releases/download/"
8
 
9
 
10
- def get_parquet_github(season_int: int, parquet_prefix: str):
11
- location = f"https://github.com/nflverse/nflverse-data/releases/download/{parquet_prefix}_{season_int}.parquet"
12
- df = pd.read_parquet(location)
13
- return df
14
-
15
-
16
  FANTASY_POSITIONS = [
17
  "QB",
18
  "RB",
@@ -30,19 +24,23 @@ def get_snap_counts(season_int: int) -> pd.DataFrame:
30
 
31
 
32
  def get_play_by_play(season_int: int) -> pd.DataFrame:
33
- return get_parquet_github(season_int, "pbp/play_by_play")
 
34
 
35
 
36
  def get_player_stats(season_int: int) -> pd.DataFrame:
37
- return get_parquet_github(season_int, "player_stats/player_stats")
 
38
 
39
 
40
  def get_ftn_charting(season_int: int) -> pd.DataFrame:
41
- return get_parquet_github(season_int, "ftn_charting/ftn_charting")
 
42
 
43
 
44
  def get_pbp_participation(season_int: int) -> pd.DataFrame:
45
- return get_parquet_github(season_int, "pbp_participation/pbp_participation")
 
46
 
47
 
48
  SEASON = "2023"
 
7
  BASE_URL = "https://github.com/nflverse/nflverse-data/releases/download/"
8
 
9
 
 
 
 
 
 
 
10
  FANTASY_POSITIONS = [
11
  "QB",
12
  "RB",
 
24
 
25
 
26
  def get_play_by_play(season_int: int) -> pd.DataFrame:
27
+ df = duckdb.sql(f"SELECT * from pbp_play_by_play_{season_int}").df()
28
+ return df
29
 
30
 
31
  def get_player_stats(season_int: int) -> pd.DataFrame:
32
+ df = duckdb.sql("SELECT * from player_stats_player_stats").df()
33
+ return df
34
 
35
 
36
  def get_ftn_charting(season_int: int) -> pd.DataFrame:
37
+ df = duckdb.sql(f"SELECT * from ftn_charting_ftn_charting_{season_int}").df()
38
+ return df
39
 
40
 
41
  def get_pbp_participation(season_int: int) -> pd.DataFrame:
42
+ df = duckdb.sql(f"SELECT * from pbp_participation_pbp_participation_{season_int}").df()
43
+ return df
44
 
45
 
46
  SEASON = "2023"