DaveM2430 commited on
Commit
ea5e64c
·
verified ·
1 Parent(s): 15854cc

Update download_data.R

Browse files
Files changed (1) hide show
  1. download_data.R +12 -11
download_data.R CHANGED
@@ -4,17 +4,15 @@ token <- Sys.getenv("HF_TOKEN")
4
  repo <- "BU-Baseball/BU-Baseball-Data"
5
  base_url <- paste0("https://huggingface.co/datasets/", repo, "/resolve/main/")
6
 
7
- download_hf <- function(filename, url_filename = NULL) {
8
- dest <- paste0("/app/", filename)
9
- url_filename <- if (is.null(url_filename)) filename else url_filename
10
-
11
  if (file.exists(dest)) {
12
  message(filename, " already exists, skipping.")
13
  return(invisible(NULL))
14
  }
15
  message("Downloading ", filename, "...")
16
  resp <- GET(
17
- paste0(base_url, url_filename),
18
  add_headers(Authorization = paste("Bearer", token)),
19
  write_disk(dest, overwrite = TRUE),
20
  progress()
@@ -24,10 +22,13 @@ download_hf <- function(filename, url_filename = NULL) {
24
  message(filename, " done.")
25
  }
26
 
27
- # Stuff+2.rds is encoded as Stuff%2B2.rds in the URL
28
- download_hf("Stuff+2.rds", "Stuff%2B2.rds")
29
- download_hf("NEW_LeagueStats2.rds")
30
- download_hf("location_plus_model.rds")
31
- download_hf("location_plus_league_stats_pitcher.rds")
 
 
32
 
33
- message("All model files ready.")
 
 
4
  repo <- "BU-Baseball/BU-Baseball-Data"
5
  base_url <- paste0("https://huggingface.co/datasets/", repo, "/resolve/main/")
6
 
7
+ download_hf <- function(filename) {
8
+ dest <- paste0("/app/", filename)
 
 
9
  if (file.exists(dest)) {
10
  message(filename, " already exists, skipping.")
11
  return(invisible(NULL))
12
  }
13
  message("Downloading ", filename, "...")
14
  resp <- GET(
15
+ paste0(base_url, filename),
16
  add_headers(Authorization = paste("Bearer", token)),
17
  write_disk(dest, overwrite = TRUE),
18
  progress()
 
22
  message(filename, " done.")
23
  }
24
 
25
+ files <- c(
26
+ "D1_scores.parquet",
27
+ "D2_scores.parquet",
28
+ "D3_scores.parquet",
29
+ "JUCO_scores.parquet",
30
+ "NAIA_scores.parquet"
31
+ )
32
 
33
+ for (f in files) download_hf(f)
34
+ message("All score files ready.")