在R Shiny中无需上传文件生成文件路径列表用于批量处理

I get exactly what you're trying to do—you want to replicate your existing R script's loop logic in Shiny, where you either pick a folder (to process all files inside) or select specific files, without having to upload those large files (which is unnecessary and slow). Here's a straightforward implementation using the shinyFiles package that matches your workflow:

Step 1: Full Working Code #

library(shiny)
library(shinyFiles)

ui <- fluidPage(
  # Buttons for folder and file selection
  shinyDirButton("select_folder", "Choose Folder (Process All Files Inside)", "Pick a folder"),
  shinyFilesButton("select_files", "Choose Specific Files", "Pick one or more files", multiple = TRUE),
  
  # Display selected paths
  h4("Selected Paths:"),
  verbatimTextOutput("display_paths"),
  
  # Optional: Button to trigger processing
  actionButton("run_processing", "Start Processing"),
  verbatimTextOutput("processing_log")
)

server <- function(input, output, session) {
  # Set up accessible volumes (adjust if needed)
  volumes <- getVolumes()
  
  # Reactive variable to store our target file paths
  selected_paths <- reactiveVal(NULL)
  
  # Handle folder selection
  observeEvent(input$select_folder, {
    shinyDirChoose(input, "select_folder", roots = volumes, session = session)
    
    if (!is.null(input$select_folder)) {
      # Parse the selected folder path
      folder_path <- parseDirPath(volumes, input$select_folder)
      
      # Get ALL full file paths in this folder (matches your original list.files call)
      file_list <- list.files(folder_path, full.names = TRUE)
      
      # Update our reactive path list
      selected_paths(file_list)
    }
  })
  
  # Handle file selection
  observeEvent(input$select_files, {
    shinyFileChoose(input, "select_files", roots = volumes, session = session)
    
    if (!is.null(input$select_files)) {
      # Parse selected file paths (datapath is the local full path)
      file_df <- parseFilePaths(volumes, input$select_files)
      
      # Extract the full paths into a list
      file_list <- as.character(file_df$datapath)
      
      # Update our reactive path list
      selected_paths(file_list)
    }
  })
  
  # Display the selected paths
  output$display_paths <- renderPrint({
    if (!is.null(selected_paths())) {
      selected_paths()
    } else {
      "No files/folder selected yet"
    }
  })
  
  # Handle processing when button is clicked
  observeEvent(input$run_processing, {
    req(selected_paths())
    
    # Initialize log
    log_text <- c("Processing started...")
    
    # Loop through each path (exactly like your original script!)
    for (path in selected_paths()) {
      # Replace this with YOUR function call
      # Example: obj <- path; your_custom_function(obj)
      log_text <- c(log_text, paste("Processing:", path))
      
      # Add your actual processing logic here
      # ...
    }
    
    # Update log output
    output$processing_log <- renderPrint({
      cat(paste(log_text, collapse = "\n"))
    })
  })
}

shinyApp(ui = ui, server = server)

Key Details Explained #

• No Uploads Needed: All we're doing is grabbing the local file/folder paths directly—no file upload occurs, which solves your large file problem.
• Matches Your Original Logic:
  • When you select a folder, we use list.files(folder_path, full.names = TRUE)—this is exactly the same as your original listBands <- list.files(myInputDir1, full.names = T) line.
  • The selected_paths() reactive variable holds the list of full paths, which you can loop through just like you did in your script.
• Flexible Selection: Users can choose either a folder (process all files) or specific files, which covers both use cases you mentioned.

How to Integrate Your Custom Function #

Replace the example loop code in observeEvent(input$run_processing, ...) with your actual processing logic. For example:

# Inside the loop
your_custom_function(path)  # path is the full file path, just like your original 'obj'

This approach keeps your existing R script logic intact while wrapping it in a Shiny interface that's intuitive for users.

内容的提问来源于stack exchange，提问作者dtanon