From 9282ab3164eace28c7a4df9ad97b10dfe017cd84 Mon Sep 17 00:00:00 2001
From: Imanol-Mikel Barba Sabariego <imanol@imanolbarba.net>
Date: Tue, 20 Dec 2016 20:32:10 +0100
Subject: [PATCH] Filtered invalid HM entries (N/A countries, >1 countries, etc..)

---
 ISO27001effectiveness/DESCRIPTION             |  3 ++-
 ISO27001effectiveness/R/Hackmageddon_Parser.R | 23 +++++++++++++++++------
 2 files changed, 19 insertions(+), 7 deletions(-)

diff --git a/ISO27001effectiveness/DESCRIPTION b/ISO27001effectiveness/DESCRIPTION
index 28ba219..d07d399 100644
--- a/ISO27001effectiveness/DESCRIPTION
+++ b/ISO27001effectiveness/DESCRIPTION
@@ -10,4 +10,5 @@ License: What license is it under?
 Encoding: UTF-8
 LazyData: true
 RoxygenNote: 5.0.1
-Imports: xlsx
+Imports: xlsx,
+    ggplot2
diff --git a/ISO27001effectiveness/R/Hackmageddon_Parser.R b/ISO27001effectiveness/R/Hackmageddon_Parser.R
index 3d9add8..c407e7b 100644
--- a/ISO27001effectiveness/R/Hackmageddon_Parser.R
+++ b/ISO27001effectiveness/R/Hackmageddon_Parser.R
@@ -43,14 +43,22 @@ ParseHMExcel <- function(file, cols){
 #' @export
 #'
 #' @examples
-#' data.pro <- ProccesHMRaw(data.raw, "1899-12-30")
-ProccesHMRaw <- function(dataset.raw, dateOffset){
+#' data.pro <- ProcessHMRaw(data.raw, "1899-12-30")
+ProcessHMRaw <- function(dataset.raw, dateOffset){
 
   #Standar names to the columns
   dataset <- setNames(dataset.raw, c("Date", "Attack", "Target", "Country"))
 
   #Remove rows with Date NA
   dataset <- dataset[!is.na(dataset$Date),]
+  dataset <- dataset[!is.na(dataset$Country),]
+  dataset <- dataset[!dataset$Country == ">1",]
+  dataset <- dataset[!dataset$Country == ">A",]
+  dataset <- dataset[!dataset$Country == "INT",]
+  dataset <- dataset[!grepl(">",dataset$Country),]
+  dataset$Country <- gsub("\n"," ",dataset$Country)
+  dataset <- FilterMultiCountry(dataset)
+  dataset <- dataset[!dataset$Country == "",]
 
   #Format properly the date
   dataset$Date <- as.POSIXct(dataset$Date*86400, tz = "GMT", origin = dateOffset)
@@ -58,7 +66,10 @@ ProccesHMRaw <- function(dataset.raw, dateOffset){
   dataset
 }
 
-
+FilterMultiCountry <- function(dataset) {
+  multi <- dataset[grepl(" ",Attacks$Country),]
+  #TODO
+}
 
 #' Parse every excel file into a folder
 #'
@@ -70,16 +81,16 @@ ProccesHMRaw <- function(dataset.raw, dateOffset){
 #' @export
 #'
 #' @examples
-#' data.pro <- ProccesHMRaw("./data/hackmaggedon/", c(1, 5, 3) "1899-12-30")
+#' data.pro <- ProcessHMRaw("./data/hackmaggedon/", c(1, 5, 3) "1899-12-30")
 ParseHMFolder <- function(folder, cols, dateOffset){
 
   #List excel files into the folder
   filelist <- list.files(folder, pattern = "*.xls*")
 
   #Iterate for each file appending the returned data.frame
-  dataset <- ProccesHMRaw(ParseHMExcel(paste(folder,filelist[1],sep = ""), cols), dateOffset)
+  dataset <- ProcessHMRaw(ParseHMExcel(paste(folder,filelist[1],sep = ""), cols), dateOffset)
   for (i in 2:length(filelist)) {
-    dataset <- rbind(dataset, ProccesHMRaw(ParseHMExcel(paste(folder,filelist[i],sep = ""), cols), dateOffset))
+    dataset <- rbind(dataset, ProcessHMRaw(ParseHMExcel(paste(folder,filelist[i],sep = ""), cols), dateOffset))
   }
 
   dataset
--
libgit2 0.22.2