shiny_app_geoprocessing.R

library(rgdal)
library(raster)
library(rgeos)
library(sp)
library(zip)

ui <- fluidPage(
  
  # App title
  
  titlePanel("Rapid weed riskmapr - geoprocessing tools"),
  
  # Sidebar panel for inputs ----
  
  sidebarLayout(
    
    sidebarPanel(width = 6, 
      
      selectInput(
        inputId = "which",
        label = "Select the geoprocessing workflow you need:",
        choices = c(
          "None", 
          "Download case study data",
          "Project detection records", 
          #"Project raster", 
          "Crop raster files", 
          "Propagule supply", 
          "Dispersal by animals (zoochory)", 
          "Dispersal by wind (anemochory)", 
          "Dispersal by water (hydrochory)", 
          "Dispersal by humans (agochory)"#, 
          #"Stream edge to raster", 
          #"Recode stream raster"
          ), 
        selected = "None", 
        multiple = FALSE
      ),
      
      uiOutput("helptext0"),
      
      uiOutput("helptext1"),
      
      uiOutput("helptext2"),
      
      uiOutput("helptext3"),
      
      uiOutput("helptext4a"),
      
      uiOutput("helptext4b"),
      
      uiOutput("helptext4c"),
      
      uiOutput("helptext4d"),

      uiOutput("Stream_raster"),
      
      uiOutput("Stream_value"),
      
      # uiOutput("Other_value"),
      
      uiOutput("Stream_shapefile"),
      
      uiOutput("Generic_shapefile"),
      
      uiOutput("Generic_shapefile_help"),
      
      uiOutput("Generic_raster"),
      
      uiOutput("Crop_generic_raster"),
      
      uiOutput("Detections"),
      
      uiOutput("Detections_help"),
      
      uiOutput("Reference_raster"),
      
      uiOutput("Column"),
      
      uiOutput("Column_help"),
      
      uiOutput("Abundance_thresholds"),
      
      uiOutput("Abundance_thresholds_help"),
      
      uiOutput("Distance_thresholds"),
      
      uiOutput("Distance_thresholds_help"),
      
      uiOutput("Proxy_levels"),
      
      uiOutput("Proxy_levels_help"),
      
      uiOutput("Max_radius"),
      
      uiOutput("Max_radius_help"),
      
      # uiOutput("Proj4string"),
      
      uiOutput("Output_name"),
      
      uiOutput("Crop_output_name"),
      
      uiOutput("submit_button"),
      
      uiOutput("a_readme_note"),
      
      uiOutput("download_button")
      
    ),
    
    mainPanel(
      width = 6,
      plotOutput("plot")
    )
  )
  
)

server <- function(input, output){
  
  # Function to convert stream edges to a raster format
  stream_edges_to_raster <- function(stream_edges, reference_raster){
    
    reference_raster[] <- 1
    mask(reference_raster, stream_edges)
    
  }
  
  # Function to create hydrochory raster
  hydrochory <- function(stream_raster, detections, reference_raster, distance_thresholds, proxy_levels, max_radius = max(distance_thresholds)){
    
    # Check whether the number of distance thresholds and proxy levels agree
    n_p_lvl <- length(proxy_levels)
    n_d_tsd <- length(distance_thresholds)
    if(n_p_lvl + 1 != n_d_tsd){
      stop("The number of proxy levels and distance thresholds must be the same.")
    }
    
    # mask reference raster within maximum distance threshold
    # mask_rs <- mask_within_buffer_streams(detections, stream_raster, max_radius)
    stream_raster <- mask_within_buffer_streams(detections, stream_raster, max_radius)
    
    # Convert stream raster to SpatialPixelsDataFrame and extract coordinates
    # stream_raster <- crop(stream_raster, extent(mask_rs))
    strm_px <- as(stream_raster, "SpatialPixelsDataFrame")
    strm_cd <- coordinates(strm_px)
    
    # Extract coordinates from detections
    dett_cd <- coordinates(detections)
    
    # Calculate distance from stream cells to detections
    n_coord <- nrow(strm_cd)
    n_detec <- nrow(detections@data)
    matrix_ <- matrix(0, nrow = n_coord, ncol = n_detec)
    for(i in 1:n_coord){
      matrix_[i,] <- euc_dist_vec(strm_cd[i,], dett_cd)
    }
    closest <- apply(matrix_, 1, min)
    
    # Compare distances to thresholds
    matrix_ <- matrix(0, nrow = n_coord, ncol = n_p_lvl)
    for(i in 1:n_p_lvl){
      matrix_[,i] <- proxy_levels[i] * 
        (is_between(closest, distance_thresholds[i], distance_thresholds[i+1]) * 1)
    }
    proxies <- apply(matrix_, 1, max)
    
    # Join to SpatialPixelsDataFrame
    strm_px@data$proxy <- proxies
    strm_px@data[] <- strm_px@data[, "proxy"]
    
    # Turn into raster
    hydrochory <- raster(strm_px)
    hydrochory[is.na(hydrochory[])] <- 0
    # hydrochory <- mask_within_buffer_streams(detections, hydrochory, max_radius)
    
    # Return
    return(hydrochory)
    
  }
  
  # Function to create propagule supply raster
  propagule_supply <- function(detections, reference_raster, column, abundance_thresholds, proxy_levels, max_radius){
    
    # Check whether the number of distance thresholds and proxy levels agree
    n_p_lvl <- length(proxy_levels)
    n_a_tsd <- length(abundance_thresholds)
    if(n_p_lvl + 1 != n_a_tsd){
      stop("The number of proxy levels and distance thresholds must be the same.")
    }
    
    # mask reference raster within maximum distance threshold
    mask_rs <- mask_within_buffer(detections, reference_raster, max_radius)
    
    # Turn this into a SpatialPixelsDataFrame
    mask_px <- as(mask_rs, "SpatialPixelsDataFrame")
    mask_cd <- coordinates(mask_px)
    dett_cd <- coordinates(detections)
    n_coord <- nrow(mask_cd)
    n_detec <- nrow(dett_cd)
    
    # Calculate some distances, calculate weed abundance
    matrix_ <- matrix(0, nrow = n_coord, ncol = n_detec)
    for(i in 1:n_coord){
      matrix_[i,] <- euc_dist_vec(mask_cd[i,], dett_cd)
    }
    n_detns <- num_detections_within(matrix_, detections, column, max_radius)
    
    # Match these numbers to the abundance thresholds for the proxies
    matrix_ <- matrix(0, n_coord, n_p_lvl)
    for(i in 1:n_p_lvl){
      matrix_[,i] <- proxy_levels[i] * 
        (is_between(n_detns, abundance_thresholds[i], abundance_thresholds[i+1]) * 1)
    }
    proxies <- apply(matrix_, 1, max)
    
    # Assign the proxies to the SpatialPixelsDataFrame
    mask_px@data$proxy <- proxies
    mask_px@data[] <- mask_px@data$proxy
    
    # Convert this to a raster layer
    propagule_supply <- raster(mask_px)
    
    # Return
    return(propagule_supply)
    
  }
  
  # Function to mask within buffer around detections
  mask_within_buffer <- function(detections, reference_raster, max_radius){
    
    # Check that all arguments are reasonable
    if(max_radius <= 0){
      stop("The argument max_radius must be greater than 0.")
    }
    
    # Buffer and mask
    buffered_detections <- buffer(detections, width = max_radius)
    reference_raster <- crop(reference_raster, extent(buffered_detections))
    reference_raster[] <- 1
    masked_raster <- mask(reference_raster, mask = buffered_detections)
    return(masked_raster)
    
  }
  
  # Function to do the same as above but for stream rasters only
  mask_within_buffer_streams <- function(detections, reference_raster, max_radius){
    
    # Check that all arguments are reasonable
    if(max_radius <= 0){
      stop("The argument max_radius must be greater than 0.")
    }
    
    # Buffer and mask
    buffered_detections <- buffer(detections, width = max_radius)
    reference_raster <- crop(reference_raster, extent(buffered_detections))
    masked_raster <- mask(reference_raster, mask = buffered_detections)
    return(masked_raster)
    
  }
  
  # Function to create a zoochory raster
  zoochory <- function(detections, reference_raster, distance_thresholds, proxy_levels, max_radius = max(distance_thresholds)){
    
    # Check whether the number of distance thresholds and proxy levels agree
    n_p_lvl <- length(proxy_levels)
    n_d_tsd <- length(distance_thresholds)
    if(n_p_lvl + 1 != n_d_tsd){
      stop("The number of proxy levels and distance thresholds must be the same.")
    }
    
    # mask reference raster within maximum distance threshold
    mask_rs <- mask_within_buffer(detections, reference_raster, max_radius)
    
    # Turn this into a SpatialPixelsDataFrame
    mask_px <- as(mask_rs, "SpatialPixelsDataFrame")
    mask_cd <- coordinates(mask_px)
    dett_cd <- coordinates(detections)
    n_coord <- nrow(mask_cd)
    n_detec <- nrow(dett_cd)
    
    # Calculate some distances, apply thresholds, and take the maximum within the rows of the data.
    matrix_ <- matrix(0, nrow = n_coord, ncol = n_detec)
    for(i in 1:n_coord){
      matrix_[i,] <- euc_dist_vec(mask_cd[i,], dett_cd)
    }
    
    # Obtain minimum distance from source
    closest <- apply(matrix_, 1, min)
    
    # Check against distance thresholds
    matrix_ <- matrix(0, n_coord, n_p_lvl)
    for(i in 1:n_p_lvl){
      matrix_[,i] <- proxy_levels[i] * 
        (is_between(closest, distance_thresholds[i], distance_thresholds[i+1]) * 1)
    }
    proxies <- apply(matrix_, 1, max)
    
    # Join back to SpatialPixelsDataFrame
    mask_px@data$proxy <- proxies
    mask_px@data[] <- mask_px$proxy
    
    # Turn into raster
    zoochory <- raster(mask_px)
    
    # Return
    return(zoochory)
    
  }
  
  # Function to count the detections within a fixed radius
  num_detections_within <- function(matrix_, detections, column, distance_threshold){
    
    # Check that inputs are correct
    n_col <- length(column)
    if(n_col != 1){
      stop("Only one column can be specified for the argument column.")
    }
    
    # Calculate the number of detections within the region
    ((matrix_ <= distance_threshold) * 1) %*% detections@data[, column]
    
  }
  
  # Same as above but requiring no column argument
  num_detections_within_no_column <- function(matrix_, detections, distance_threshold){
    
    # Calculate the number of detections within the region
    sum((matrix_ <= distance_threshold) * 1)
    
  }
  
  # Convert stream rasters with two values (one for streams, another for others) to a raster with ones for stream cells and NAs elsewhere
  binary_stream_to_unary_stream <- function(stream_raster, stream_value){
    
    # Turn stream cells to 1
    if(stream_value != 1){
      stream_raster[stream_raster[] == stream_value] <- 1
    }
    
    # Turn non-stream cells to NA
    stream_raster[stream_raster[] != stream_value] <- NA
    
    # Return streams
    stream_raster
    
  }
  
  # Function to map thresholds to proxy levels
  is_between <- function(v, b1, b2){
    
    # Sort the bounds
    lower <- min(c(b1, b2))
    upper <- max(c(b1, b2))
    
    # Return logical for comparison
    lower <= v & v < upper # [LWR, UPR)
    
  }
  
  # Function to measure Euclidean distance
  euc_dist <- function(x, y){
    
    d <- x - y
    sqrt(sum(d^2))
    
  }
  
  # Function to vectorise the above process
  euc_dist_vec <- function(x, y){
    
    apply(y, 1, euc_dist, x)
    
  }
  
  # This code creates the readme note for the Download case study data tool
  output$a_readme_note <- renderUI(
    {
      if(input$which == "Download case study data"){
        helpText("The downloaded .zip archive contains a README file that describes the contents of the data package and how they should be used in riskmapr.")
      }
    }
  )
  
  # This code updates the UI
  output$helptext0 <- renderUI(
    {
      if(input$which == "None"){
        helpText("0. Download case study data: Use this tool to download the preprocessed data that are needed to reproduce the case studies in Froese, Pearse & Hamilton (2019).")
      }
    }
  )
  
  output$helptext1 <- renderUI(
    {
      if(input$which == "None"){
        helpText("1. Project detection records: Use this tool to ensure a consistent spatial reference between weed detection records and the study area reference raster. Output is a zipped .SHP file.")
       }
    }
  )
  
  output$helptext2 <- renderUI(
    {
      if(input$which == "None"){
        helpText("2. Crop raster files: Use this tool to crop the study area reference raster and all spatial proxies for risk factors affecting suitability (establishment and persistence) to the specified dispersal risk area around weed detection records. REQUIRED to limit computational demands on the riskmapr - susceptibility model app. Outputs are zipped .TIF files.")
      }
    }
  )
  
  output$helptext3 <- renderUI(
    {
      if(input$which == "None"){
        helpText("3. Propagule supply: Use this tool to generate a spatial proxy for propagule supply within the specified dispersal risk area around weed detection records. Abundance thresholds are used to classify the output into discrete states (the numerical values assigned to the defined 'risk levels' of 'Propagule supply'). Output is a .TIF file.")
      }
    }
  )
  
  output$helptext4d <- renderUI(
    {
      if(input$which == "None"){
        helpText("4d. Dispersal by humans (agochory): Use this tool to generate a spatial proxy for propagule dispersal via human transportation along linear features (e.g. roads) within the specified dispersal risk area. Distance thresholds are used to classify the output into discrete states (the numerical values assigned to the defined 'risk levels' of 'Agochory'). Output is a .TIF file.")
      }
    }
  )
  
  output$helptext4b <- renderUI(
    {
      if(input$which == "None"){
        helpText("4b. Dispersal by wind (anemochory): Use this tool to generate a spatial proxy for propagule dispersal via wind within the specified dispersal risk area. Distance thresholds are used to classify the output into discrete states (the numerical values assigned to the defined 'risk levels' of 'Anemochory'). Output is a .TIF file.")
      }
    }
  )
  
  output$helptext4c <- renderUI(
    {
      if(input$which == "None"){
        helpText("4c. Dispersal by water (hydrochory): Use this tool to generate a spatial proxy for propagule dispersal via water along linear features (e.g. streams) within the specified dispersal risk area. Distance thresholds are used to classify the output into discrete states (the numerical values assigned to the defined 'risk levels' of 'Hydrochory'). Output is a .TIF file.")
      }
    }
  )
  
  output$helptext4a <- renderUI(
    {
      if(input$which == "None"){
        helpText("4a. Dispersal by animals (zoochory): Use this tool to generate a spatial proxy for propagule dispersal via animal ingestion (endoozoochory) or attachment (epizoochory) within the specified dispersal risk area. Distance thresholds are used to classify the output into discrete states (the numerical values assigned to the defined 'risk levels' of 'Zoochory'). Output is a .TIF file.")
      }
    }
  )
  
  output$Stream_raster <- renderUI(
    {
      if(input$which %in% c("Dispersal by water (hydrochory)", "Dispersal by humans (agochory)", "Recode stream raster")){
        fileInput("stream_raster", "Upload raster of a linear feature (e.g. streams, roads) (.tif extension)", FALSE, ".tif")
      }
    }
  )
  
  output$Stream_value <- renderUI(
    {
      if(input$which == "Recode stream raster"){
        numericInput("stream_value", "Enter the value for the stream cells", 1, min = -1e9, max = 1e9)
      }
    }
  )
  
  # output$Other_value <- renderUI(
  #   {
  #     if(input$which == "Recode stream raster"){
  #       numericInput("other_value", "Enter the value for the non-stream cells", 0, min = -1e9, max = 1e9)
  #     }
  #   }
  # )
  
  output$Stream_shapefile <- renderUI(
    {
      if(input$which == "Stream edge to raster"){
        fileInput("stream_shapefile", "Upload a shapefile of streams", TRUE)
      }
    }
  )
  
  output$Detections <- renderUI(
    {
      if(input$which %in% c("Propagule supply", "Dispersal by animals (zoochory)", "Dispersal by wind (anemochory)", "Dispersal by water (hydrochory)", "Dispersal by humans (agochory)", "Crop raster files")){
        fileInput("detections", "Upload shapefile of projected weed detection records", TRUE)
      }
    }
  )
  
  output$Detections_help <- renderUI(
    {
      if(input$which %in% c("Propagule supply", "Dispersal by animals (zoochory)", "Dispersal by wind (anemochory)", "Dispersal by water (hydrochory)", "Dispersal by humans (agochory)", "Crop raster files")){
        helpText("Select the entire collection of files which make up the shapefile (same filename before extension).")
      }
    }
  )
  
  output$Generic_raster <- renderUI(
    {
      if(input$which %in% c("Project raster", "Project detection records")){
        fileInput("generic_raster", "Upload study area reference raster (.tif extension)", FALSE, ".tif")
      }
    }
  )
  
  output$Crop_generic_raster <- renderUI(
    {
      if(input$which == "Crop raster files"){
        fileInput("crop_generic_raster", "Upload rasters (.tif extension, allows multiple)", TRUE, ".tif")
      }
    }
  )
  
  output$Generic_shapefile <- renderUI(
    {
      if(input$which %in% c("Project detection records")){
        fileInput("generic_shapefile", "Upload shapefile of weed detection records", TRUE)
      }
    }
  )
  
  output$Generic_shapefile_help <- renderUI(
    {
      if(input$which %in% c("Project detection records")){
        helpText("Select the entire collection of files which make up the shapefile (same filename before extension).")
      }
    }
  )
  
  # output$Proj4string <- renderUI(
  #   {
  #     if(input$which %in% c("Project detection records", "Project raster")){
  #       textInput("proj4string", "Enter a proj4string (see spatialreference.org for proj4strings):", "+init=epsg:4326")
  #     }
  #   }
  # )
  
  output$Reference_raster <- renderUI(
    {
      if(input$which %in% c("Propagule supply", "Dispersal by animals (zoochory)", "Dispersal by wind (anemochory)", "Dispersal by water (hydrochory)", "Dispersal by animals (zoochory)", "Dispersal by humans (agochory)", "Stream edge to raster")){
        fileInput("reference_raster", "Upload cropped study area reference raster (.tif extension)", FALSE, ".tif")
      }
    }
  )
  
  output$Column <- renderUI(
    {
      if(input$which == "Propagule supply"){
        textInput("column", "Field name for abundance per weed detection record", "Abundance")
      }
    }
  )
  
  output$Column_help <- renderUI(
    {
      if(input$which %in% c("Propagule supply")){
        helpText("Specify the name of the field/column in the shapefile's attribute table, which contains a (observed or estimated) measure of abundance for each detection record (must be numerical).The default is 'Abundance'")
      }
    }
  )
  
  output$Abundance_thresholds <- renderUI(
    {
      if(input$which %in% c("Propagule supply")){
        textInput("abundance_thresholds", "Abundance thresholds for each risk level", "")
      }
    }
  )
  
  output$Abundance_thresholds_help <- renderUI(
    {
      if(input$which %in% c("Propagule supply")){
        helpText("Specify abundance thresholds to classify the output into discrete states (corresponding to the 'risk levels' defined for propagule supply). Each interval is inclusive of the lower threshold and exclusive of the higher threshold. The number of thresholds must equal the number of risk levels n + 1. Set the highest threshold arbitrarily high to ensure valid computations.")
      }
    }
  )
  
  output$Distance_thresholds <- renderUI(
    {
      if(input$which %in% c("Dispersal by animals (zoochory)", "Dispersal by water (hydrochory)", "Dispersal by humans (agochory)", "Dispersal by wind (anemochory)")){
        textInput("distance_thresholds", "Distance thresholds for each risk level", "")
      }
    }
  )
  
  output$Distance_thresholds_help <- renderUI(
    {
      if(input$which %in% c("Dispersal by animals (zoochory)", "Dispersal by water (hydrochory)", "Dispersal by humans (agochory)", "Dispersal by wind (anemochory)")){
        helpText("Specify distance thresholds to classify the output into discrete states (corresponding to the 'risk levels' defined for the dispersal mode). Each interval is inclusive of the lower threshold and exclusive of the higher threshold. The number of thresholds must equal the number of risk levels n + 1. Set the lowest threshold to '0' and the highest threshold equal to the 'dispersal risk area' to ensure valid computations. Where the upper limit for the dispersal mode [u] is less than the dispersal risk area [d], set the penultimate threshold to equal [u], and the ultimate threshold to equal [d].")
      }
    }
  )
  
  output$Proxy_levels <- renderUI(
    {
      if(input$which %in% c("Propagule supply", "Dispersal by animals (zoochory)", "Dispersal by water (hydrochory)", "Dispersal by wind (anemochory)", "Dispersal by humans (agochory)")){
        textInput("proxy_levels", "Risk levels corresponding to distance or abundance thresholds", "")
      }
    }
  )
  
  output$Proxy_levels_help <- renderUI(
    {
      if(input$which %in% c("Propagule supply", "Dispersal by animals (zoochory)", "Dispersal by water (hydrochory)", "Dispersal by wind (anemochory)", "Dispersal by humans (agochory)")){
        helpText("Enter the numerical value assigned to each discrete state ('risk level') of propagule supply or the dispersal mode. The number of values must equal the number of abundance/distance thresholds n - 1. Where the upper limit for the dispersal mode [u] is less than the dispersal risk area [d], assign the value '0' to the distance band [u,d].")
      }
    }
  )
  
  output$Max_radius <- renderUI(
    {
      if(input$which %in% c("Propagule supply", "Dispersal by animals (zoochory)", "Dispersal by water (hydrochory)", "Dispersal by wind (anemochory)", "Dispersal by humans (agochory)", "Crop raster files")){
        numericInput("max_radius", "Dispersal risk area (m)", 1000, min = 0, max = 1e6)
      }
    }
  )
  
  output$Max_radius_help <- renderUI(
    if(input$which %in% c("Propagule supply", "Dispersal by animals (zoochory)", "Dispersal by water (hydrochory)", "Dispersal by wind (anemochory)", "Dispersal by humans (agochory)", "Crop raster files")){
       helpText("Specify the area at risk of propagule introduction from any of the source infestations included in the shapefile. Use the upper limit dispersal distance of the furthest-reaching disperal mode identified in your susceptibility model.")
    }
  )
  
  output$Output_name <- renderUI(
    if(!(input$which %in% c("None", "Crop raster files", "Download case study data"))){
      textInput("output_name", "Enter descriptive name of output file (no extension)", "Output_File")
    }
  )
  
  output$Crop_output_name <- renderUI(
    if(input$which == "Crop raster files"){
      textInput("crop_output_name", "Enter descriptive suffix to append to each output file (no extension)", "Crop")
    }
  )
  
  output$submit_button <- renderUI(
    {
      if(!input$which %in% c("None", "Download case study data")){
        actionButton("submit", "RUN GEOPROCESSING TOOL")
      }
    }
  )
  
  output$download_button <- renderUI(
    {
      if(input$which != "None"){
        if(input$which != "Download case study data"){
          downloadButton("Download", "DOWNLOAD OUTPUT(S)")
        } else {
          downloadButton("Download", "DOWNLOAD CASE STUDY DATA")
        }
      }
    }
  )
  
  the_data <- eventReactive(
    input$submit,
    {
      if(input$which == "Crop raster files"){
        
        # Ingest detection records
        files <- input$detections$datapath
        files_renamed <- gsub("/[0-9]{1}\\.", "/REPLACE\\.", files)
        for(i in 1:length(files)){
          file.rename(files[i], files_renamed[i])
        }
        detections <- shapefile(files_renamed[grep(".shp$", files_renamed)])
        
        # Ingest generic raster
        generic_raster <- stack(input$crop_generic_raster$datapath)
        
        # Max buffer distance
        max_radius <- input$max_radius
        
        # Buffer and crop
        buffered <- buffer(detections, width = max_radius)
        result <- crop(generic_raster, extent(buffered))
        
      }
      if(input$which == "Propagule supply"){
        
        # Ingest detection records
        files <- input$detections$datapath
        files_renamed <- gsub("/[0-9]{1}\\.", "/REPLACE\\.", files)
        for(i in 1:length(files)){
          file.rename(files[i], files_renamed[i])
        }
        detections <- shapefile(files_renamed[grep(".shp$", files_renamed)])
        
        # Ingest reference raster
        reference_raster <- input$reference_raster
        reference_raster <- reference_raster$datapath
        reference_raster <- raster(reference_raster)
        
        # Ingest column
        column <- input$column
        
        # Ingest distance thresholds
        abundance_thresholds <- input$abundance_thresholds
        abundance_thresholds <- as.numeric(strsplit(abundance_thresholds, ",")[[1]])
        
        # Ingest proxy levels
        proxy_levels <- input$proxy_levels
        proxy_levels <- as.numeric(strsplit(proxy_levels, ",")[[1]])
        
        # Check these
        if(any(is.na(abundance_thresholds)) | any(is.na(proxy_levels))){
          stop("Please re-enter the distance thresholds and proxy levels. Please make sure that these are entered as numbers separated by commas.")
        }
        
        # Ingest number for maximum buffer distance
        max_radius <- input$max_radius
        
        # Compute hydrochory raster
        result <- propagule_supply(detections, reference_raster, column, abundance_thresholds, proxy_levels, max_radius)
        
      }
      
      if(input$which %in% c("Dispersal by animals (zoochory)", "Dispersal by wind (anemochory)")){
        
        # Ingest detection records
        files <- input$detections$datapath
        files_renamed <- gsub("/[0-9]{1}\\.", "/REPLACE\\.", files)
        for(i in 1:length(files)){
          file.rename(files[i], files_renamed[i])
        }
        detections <- shapefile(files_renamed[grep(".shp$", files_renamed)])
        
        # Ingest reference raster
        reference_raster <- input$reference_raster
        reference_raster <- reference_raster$datapath
        reference_raster <- raster(reference_raster)
        
        # Ingest distance thresholds
        distance_thresholds <- input$distance_thresholds
        distance_thresholds <- as.numeric(strsplit(distance_thresholds, ",")[[1]])
        
        # Ingest proxy levels
        proxy_levels <- input$proxy_levels
        proxy_levels <- as.numeric(strsplit(proxy_levels, ",")[[1]])
        
        # Check these
        if(any(is.na(distance_thresholds)) | any(is.na(proxy_levels))){
          stop("Please re-enter the distance thresholds and proxy levels. Please make sure that these are entered as numbers separated by commas.")
        }
        
        # Ingest number for maximum buffer distance
        max_radius <- input$max_radius
        
        # Compute zoochory raster
        result <- zoochory(detections, reference_raster, distance_thresholds, proxy_levels, max_radius)
      }
      
      if(input$which %in% c("Dispersal by water (hydrochory)", "Dispersal by humans (agochory)")){
        
        # Ingest stream raster
        stream_raster <- input$stream_raster
        stream_raster <- stream_raster$datapath
        stream_raster <- raster(stream_raster)
        
        # Ingest detection records
        files <- input$detections$datapath
        files_renamed <- gsub("/[0-9]{1}\\.", "/REPLACE\\.", files)
        for(i in 1:length(files)){
          file.rename(files[i], files_renamed[i])
        }
        detections <- shapefile(files_renamed[grep(".shp$", files_renamed)])
        
        # Ingest reference raster
        reference_raster <- input$reference_raster
        reference_raster <- reference_raster$datapath
        reference_raster <- raster(reference_raster)
        
        # Ingest distance thresholds
        distance_thresholds <- input$distance_thresholds
        distance_thresholds <- as.numeric(strsplit(distance_thresholds, ",")[[1]])
        
        # Ingest proxy levels
        proxy_levels <- input$proxy_levels
        proxy_levels <- as.numeric(strsplit(proxy_levels, ",")[[1]])
        
        # Check these
        if(any(is.na(distance_thresholds)) | any(is.na(proxy_levels))){
          stop("Please re-enter the distance thresholds and proxy levels. Please make sure that these are entered as numbers separated by commas.")
        }
        
        # Ingest number for maximum buffer distance
        max_radius <- input$max_radius
        
        # Compute hydrochory raster
        result <- hydrochory(stream_raster, detections, reference_raster, distance_thresholds, proxy_levels, max_radius)
      
      }
      
      # if(input$which == "Stream edge to raster"){
      #   
      #   # Ingest shapefile
      #   files <- input$stream_shapefile$datapath
      #   files_renamed <- gsub("/[0-9]{1}\\.", "/REPLACE\\.", files)
      #   for(i in 1:length(files)){
      #     file.rename(files[i], files_renamed[i])
      #   }
      #   stream_shapefile <- shapefile(files_renamed[grep(".shp$", files_renamed)])
      #   
      #   # Ingest reference raster
      #   reference_raster <- raster(input$reference_raster$datapath)
      #   
      #   # Mask over reference raster
      #   reference_raster[] <- 1
      #   result <- mask(reference_raster, stream_shapefile)
      #   
      # }
      
      # if(input$which == "Recode stream raster"){
      #   
      #   # Ingest inputs
      #   stream_raster <- input$stream_raster
      #   stream_raster <- stream_raster$datapath
      #   stream_raster <- raster(stream_raster)
      #   stream_value <- input$stream_value
      #   # Return recoded stream raster
      #   result <- binary_stream_to_unary_stream(stream_raster, stream_value)
      #   
      # }
      # 
      # if(input$which == "Project raster"){
      #   
      #   # Load data
      #   project_raster <- raster(input$generic_raster$datapath)
      #   
      #   # Project the raster
      #   result <- projectRaster(project_raster, crs = CRS(input$proj4string))
      #   
      # }
      
      if(input$which == "Project detection records"){

        # Ingest shapefile
        files <- input$generic_shapefile$datapath
        files_renamed <- gsub("/[0-9]{1}\\.", "/REPLACE\\.", files)
        for(i in 1:length(files)){
          file.rename(files[i], files_renamed[i])
        }
        project_shapefile <- shapefile(files_renamed[grep(".shp$", files_renamed)])
        
        # Ingest raster with target coordinate reference system
        project_raster <- raster(input$generic_raster$datapath)

        # Project it
        result <- spTransform(project_shapefile, crs(project_raster))

      }
      
      result

    }
  )
  
  output$plot <- renderPlot(
    {
      plot(the_data())
    }
  )
  
  output$Download <- downloadHandler(
    
    filename = function(){
      if(!(input$which %in% c("Project detection records", "Crop raster files"))){
        # Single raster outputs will be available as .tif
        outname <- paste0(input$output_name, ".tif")
      }
      if(input$which == "Project detection records"){
        # Bundles of multiple files will be downloadable as a .zip archive
        outname <- paste0(input$output_name, ".zip")
      }
      if(input$which == "Crop raster files"){
        # Bundles of (potentially) multiple files will be downloadable as a .zip archive
        outname <- "Downloads.zip"
      }
      if(input$which == "Download case study data"){
        # Download case study data from hard-coded file uploaded with the app
        outname <- "Case_Study_Data.zip"
      }
      outname
      
    },
    content = function(file){
      efficiently_write_raster <- function(r, fn, ...){
        # MODIFIED FROM SOURCE CODE IN THE PACKAGE "RASTER", FROM FUNCTION WRITE RASTER
        # MODIFIED 3 FEB, 2019
        
        # Find good chunk characteristics for writing to disk
        tr <- blockSize(r)
        # Function to write out the raster WITHOUT copying it several times in memory
        f <- writeStart(r, fn, ...)
        for(i in 1:tr$n){
          vals <- getValuesBlock(r, row=tr$row[i], nrows=tr$nrows[i])
          f <- writeValues(f, vals, tr$row[i])
        }
        f <- writeStop(f)
        return(f)
      }
      if(length(Sys.glob("*.tif")) > 0){
        file.remove(Sys.glob("*.tif"))
      }
      if(!(input$which %in% c("Project detection records", "Crop raster files", "Download case study data"))){
        efficiently_write_raster(the_data(), file)
      } 
      if(input$which == "Crop raster files") {
        num_rasters <- nlayers(the_data())
        nam_rasters <- input$crop_generic_raster$name
        nam_rasters <- gsub(".tif$", "", nam_rasters)
        nam_rasters <- paste0(nam_rasters, "_", input$crop_output_name)
        writeRaster(the_data(), nam_rasters, bylayer = TRUE, format = "GTiff")
        zip(zipfile = file, files = Sys.glob("*.tif"))
      }
      if(input$which == "Project detection records") {
        shapefile(the_data(), paste0(input$output_name, ".shp"), overwrite = TRUE)
        zip(zipfile = file, files = Sys.glob(paste0("*", input$output_name, ".*")))
      }
      if(input$which == "Download case study data"){
        file.copy("CaseStudyData.zip", file)
      }
      
    }
    
  )
  
}

shinyApp(ui, server)