ag.dev.cl: ag.dev.cl

#'
#' @title Checks if all variables do exist and are not empty
#' @description This function check that the variables to analyse are (1) available from all 
#' the studies and (2) that they do not contain only missing values (NAs). It excludes studies 
#' that fail any of these two checks
#' @param opals a list of opal object(s) obtained after login in to opal servers;
#' these objects hold also the data assign to R, as \code{dataframe}, from opal 
#' datasources.
#' @param variables a character vector, the names of the variable(s) to check 
#' @return the opal objects which passed the checks
#' @author Gaye, A.
#' @export
#' @examples {
#' 
#' # load that contains the login details
#' data(logindata)
#' 
#' # login and assign specific variable(s)
#' myvar <- list("LAB_TSC")
#' opals <- ag.ds.login(logins=logindata,assign=TRUE,variables=myvar)
#' 
#' # run checks for the variable LAB_TSC
#' ag.ds.checkvar(opals=opals, variables=list(quote(D$LAB_TSC)))
#' }
#'
ag.ds.checkvar <- function(opals, variables){
  
  # print a message for the user informing of checks
  cat("\nChecks are carried out on the variables used for the analysis\nto ensure they are available from the dataset(s) and not empty.\n\n")
  # get the names of the opal servers/studies
  stdname <- names(opals)
  
  # get the names of the variables to check
  varIDs <- vector("character", length(variables))
  for(i in 1:length(variables)){
    xx <- variables[[i]]
    varIDs[i] <- strsplit(deparse(xx), "\\$", perl=TRUE)[[1]][2]
  }
  
  # a vector that keeps the results of the checks for each study
  toremove <- c()
  
  # loop through the dataset(s) and the variable(s)
  for(i in 1: length(opals)){
    
    # Carry out the first check:  are all the variables to analyse available from dataset
    track <- FALSE
    # get the names of the variables in the assigned dataset
    var.names <- datashield.aggregate(opals[i], quote(colnames(D)))   
    
    # check if any of the variables in the arguments is missing from the assigned dataset
    idx1 <- which(!(varIDs %in% var.names[[1]]))
    missings <- length(idx1)
    if(missings > 0){
      # record that the stduy has failed the first checks and print a message
      track <- TRUE
      cat("The variable(s)", varIDs[idx1], "is/are missing from", stdname[i],"!\n") 
      
    }else{
      # carry out the second check: do any of the variables to anlyse contain only NAs
      # this second check is carried out only if the first check is negative
      # loop through the variables in the argument and if any fails break out the loop
      
      # get the indices, in the assigned dataset, of the variables to check
      idx2 <- which(var.names[[1]] %in% varIDs)
      for(j in idx2){
        # the server side function 'isNA.ds' to check if vector is empty
        cally <- call("ag.isNA.ds", variables[[j]])
        out <- datashield.aggregate(opals[i], cally)
        if(out[[1]]){ 
          track <- TRUE
          cat("The variable", var.names[[1]][j], "in", stdname[i], "is empty (NAs only)!\n")
        }
      }
    }
    
    # if a study fails any of the two checks add it to the list of studies to exclude and print a message
    if(track){ 
      toremove <- append(toremove, i)
      cat(stdname[i], "will not be included in the analysis\n\n") 
    }
    
  }
  
  # remove studies which contain one or more variables that failed the checks
  if(length(toremove) > 0){
    opals <- opals[-toremove] 
  }else{
    cat("The checks went fine: no missing or empty variable(s)!\n\n")
  }
  
  # If none of the datasets passed the checks stop the process
  # ortherwise return the opal objects that passed the checks
  if(length(opals) == 0){
    stop("The variables specified in the arguments are not available or contain only missing values, in all the assigned datasets!")
  }else{
    return(opals)
  }
}