TPD / download_data.R
DaveM2430's picture
Update download_data.R
ad74b7d verified
Raw
History Blame Contribute Delete
1.4 kB
library(httr)
token <- Sys.getenv("HF_TOKEN")
repo <- "BU-Baseball/BU-Baseball-Data"
base_url <- paste0("https://huggingface.co/datasets/", repo, "/resolve/main/")
download_hf <- function(filename) {
dest <- paste0("/app/", filename)
if (file.exists(dest)) {
message(filename, " already exists, skipping.")
return(invisible(NULL))
}
message("Downloading ", filename, "...")
resp <- GET(
paste0(base_url, filename),
add_headers(Authorization = paste("Bearer", token)),
write_disk(dest, overwrite = TRUE),
progress()
)
if (status_code(resp) != 200)
stop("Failed to download ", filename, " — status ", status_code(resp))
message(filename, " done.")
}
files <- c(
"D1.parquet",
"D2.parquet",
"D3.parquet",
"JUCO.parquet",
"NAIA.parquet",
"D1_scores.parquet",
"D2_scores.parquet",
"D3_scores.parquet",
"JUCO_scores.parquet",
"NAIA_scores.parquet",
"transfer_portal.parquet",
"hitter_portal.parquet",
"D1_pitcher_leaderboard.parquet",
"D2_pitcher_leaderboard.parquet",
"D3_pitcher_leaderboard.parquet",
"JUCO_pitcher_leaderboard.parquet",
"NAIA_pitcher_leaderboard.parquet",
"D1_hitter_leaderboard.parquet",
"D2_hitter_leaderboard.parquet",
"D3_hitter_leaderboard.parquet",
"JUCO_hitter_leaderboard.parquet",
"NAIA_hitter_leaderboard.parquet"
)
for (f in files) download_hf(f)
message("All files ready.")