Update app.R

app.R

@@ -13,19 +13,55 @@ download_private_csv <- function(repo_id, filename) {
   response <- GET(url, add_headers(Authorization = paste("Bearer", Sys.getenv("GETCSV"))))
   
   if (status_code(response) == 200) {
-    content <- content(response, "text")
-    con <- textConnection(content)
+    # Get content as text first to check if it's an LFS pointer
+    content_text <- content(response, "text", encoding = "UTF-8")
     
-    # Try different read options
-    data <- read.csv(con, 
-                     header = TRUE,
-                     check.names = FALSE,  # This prevents R from modifying column names
-                     fileEncoding = "UTF-8",
-                     stringsAsFactors = FALSE)
-    close(con)
-    return(data)
+    # Check if this is an LFS pointer (LFS files start with "version https://git-lfs.github.com/spec/")
+    if (grepl("^version https://git-lfs.github.com/spec/", content_text)) {
+      # This is an LFS file - extract the oid (hash) from the pointer
+      oid_line <- grep("oid sha256:", strsplit(content_text, "\n")[[1]], value = TRUE)
+      oid <- gsub("oid sha256:", "", oid_line)
+      oid <- trimws(oid)
+      
+      # Construct the LFS content URL
+      lfs_url <- paste0("https://huggingface.co/datasets/", repo_id, "/resolve/main/.git/lfs/objects/", 
+                        substr(oid, 1, 2), "/", substr(oid, 3, 4), "/", oid)
+      
+      # Get the actual content from LFS storage
+      lfs_response <- GET(lfs_url, add_headers(Authorization = paste("Bearer", Sys.getenv("GETCSV"))))
+      
+      if (status_code(lfs_response) == 200) {
+        content_text <- content(lfs_response, "text", encoding = "UTF-8")
+      } else {
+        # Alternative LFS URL format
+        lfs_url <- paste0("https://huggingface.co/datasets/", repo_id, "/lfs/resolve/main/", filename, "?download=true")
+        lfs_response <- GET(lfs_url, add_headers(Authorization = paste("Bearer", Sys.getenv("GETCSV"))))
+        
+        if (status_code(lfs_response) == 200) {
+          content_text <- content(lfs_response, "text", encoding = "UTF-8")
+        } else {
+          stop(paste("Failed to download LFS content. Status code:", status_code(lfs_response)))
+        }
+      }
+    }
+    
+    # Process the content (whether it was LFS or regular)
+    con <- textConnection(content_text)
+    tryCatch({
+      data <- read.csv(con, 
+                      header = TRUE,
+                      check.names = FALSE,
+                      fileEncoding = "UTF-8",
+                      stringsAsFactors = FALSE)
+      return(data)
+    }, error = function(e) {
+      close(con)
+      stop(paste("Error parsing CSV:", e$message))
+    }, finally = {
+      close(con)
+    })
   } else {
-    stop("Failed to download dataset")
+    stop(paste("Failed to download dataset. Status code:", status_code(response)))
   }
 }