Changed [Hackmaggedon_Parser.R], added [ProccessHMRaw] to remove NAs and changed…

… [PArseHMFolder] because of problems with lapply

Changed [Hackmaggedon_Parser.R], added [ProccessHMRaw] to remove NAs and changed…
… [PArseHMFolder] because of problems with lapply
Miguel Tuñón
1 parent a35b1822
Showing 1 changed file with 17 additions and 22 deletions
ISO27001effectiveness/R/Hackmageddon_Parser.R
@@ -4,38 +4,33 @@
  
  
  
-#-----------------------------Library tests / install--------------------------------------------------
-
-
-#' Install and load required libraries
-#'
-#' This function checks if every required library is installed to be loaded, if not they will be installed and then loaded.
-#' Libraries installed:
-#'  xlsx to parse excel files like ISO survey source format
-LoadParserLibraries <- function(){
-  if (!require("xlsx"))
-  {
-    install.packages("xlsx")
-    if (!require("xlsx")) stop("Error while loading package [xlsx]")
-  }
-}
-
-#-----------------------------
-
 ParseHMExcel <- function(file){
   print(file)
   if (!file.exists(file)) {
     stop(paste("Error, file [", file, "] not found"))
   }
-  dataset <- read.xlsx2(file, 1, header = TRUE,colIndex = 2:10)
+  dataset <- read.xlsx2(file, 1, header = TRUE,colIndex = 2:10, colClasses = c("Date", "character", "character", "character", "character", "character", "character", "character"))
+  dataset
+}
+
+ProccesHMRaw <- function(dataset.raw){
+
+  #Remove rows with Date NA
+  dataset <- dataset.raw[!is.na(dataset.raw$Date),]
+
   dataset
 }
  
 ParseHMFolder <- function(folder){
   filelist <- list.files(folder)
-  frames <- lapply(paste(folder,filelist,sep=""),ParseHMExcel)
-  merged_frame =  merge(frames)
-  merged_frame
+  #frames <- lapply(paste(folder,filelist,sep = ""),ParseHMExcel)
+  dataset <- ProccesHMRaw(ParseHMExcel(paste(folder,filelist[1],sep = "/")))
+
+  for (i in 2:length(filelist)) {
+    dataset <- rbind(dataset, ProccesHMRaw(ParseHMExcel(paste(folder,filelist[i],sep = "/"))))
+  }
+
+  dataset
 }