Update app.R

app.R

@@ -1939,6 +1939,75 @@ observe({
       write.csv(scraped_data(), file, row.names = FALSE)
     }
   )
+
+
+observeEvent(input$upload_hf_btn, {
+  req(scraped_data())
+  
+  scrape_status_msg("Downloading existing dataset...")
+  
+  hf_token <- Sys.getenv("HF_WRITE_TOKEN")
+  repo_id <- "CoastalBaseball/_2026MasterData"
+  filename <- paste0(input$scrape_source, "2026_master_pbp.parquet")
+  
+  # Try to download existing data
+  existing <- tryCatch({
+    resp <- httr::GET(
+      paste0("https://huggingface.co/datasets/", repo_id, "/resolve/main/", filename),
+      httr::add_headers(Authorization = paste("Bearer", hf_token))
+    )
+    
+    if (httr::status_code(resp) == 200) {
+      tmp_dl <- tempfile(fileext = ".parquet")
+      writeBin(httr::content(resp, as = "raw"), tmp_dl)
+      arrow::read_parquet(tmp_dl)
+    } else {
+      NULL
+    }
+  }, error = function(e) { NULL })
+  
+  # Combine
+  if (!is.null(existing) && nrow(existing) > 0) {
+    scrape_status_msg("Merging with existing data...")
+    combined <- bind_rows(existing, scraped_data())
+    
+    if ("PitchUID" %in% names(combined)) {
+      combined <- combined %>% distinct(PitchUID, .keep_all = TRUE)
+    } else {
+      combined <- combined %>% distinct()
+    }
+  } else {
+    combined <- scraped_data()
+  }
+  
+  # Upload combined data as parquet
+  scrape_status_msg(paste0("Uploading ", nrow(combined), " total rows..."))
+  
+  tmp <- tempfile(fileext = ".parquet")
+  arrow::write_parquet(combined, tmp)
+  
+  result <- tryCatch({
+    resp <- httr::PUT(
+      paste0("https://huggingface.co/api/datasets/", repo_id, "/upload/main/", filename),
+      httr::add_headers(
+        Authorization = paste("Bearer", hf_token),
+        `Content-Type` = "application/octet-stream"
+      ),
+      body = httr::upload_file(tmp)
+    )
+    
+    if (httr::status_code(resp) %in% c(200, 201)) {
+      paste0("Done! ", nrow(combined), " total rows in dataset.")
+    } else {
+      paste0("Upload failed: ", httr::status_code(resp))
+    }
+  }, error = function(e) {
+    paste("Upload error:", e$message)
+  })
+  
+  file.remove(tmp)
+  scrape_status_msg(result)
+})
     
 }