setwd("C:/Users/victor/Documents/TFG/r-analytics-master")
codebook <- read.csv("new_harmon.csv", sep = ",")

binary <- unlist(codebook[codebook["Variable.type"] == "Binary",]["Harmonised.variable.name"])
categorical <- unlist(codebook[codebook["Variable.type"] == "Categorical",]["Harmonised.variable.name"])
continuous <- unlist(codebook[codebook["Variable.type"] == "Continuous",]["Harmonised.variable.name"])

categoric_vars <- c(binary, categorical)

replace_with_Missing_categoric <- function(column){
  
  for (i in 1:length(column)){
    
    x <- column[i]
    
    if (is.na(x)){
      x <- "Missing"
      
    }else{
      
      if(x == "" | x == "NA" | x == "Unknown")
        x <- "Missing"
    }
    
    column[i] <- x
    
  }
  
  return (column)
}

replace_with_Missing_num_categoric <- function(column){
  
  for (i in 1:length(column)){
    
    x <- column[i]
    
    if (is.na(x)){
      x <- 9999
      
    }else{
      
      if(x == "" | x == "NA" | x == "Unknown")
        x <- 9999
    }
    
    column[i] <- x
    
  }
  
  return (column)
}


replace_with_Missing_continuous <- function(column){
  
  for (i in 1:length(column)){
    
    x <- column[i]
    
    if (is.na(x)){
      x <- ""
      
    }else{
      
      if(x == "NA" | x == "Unknown")
        x <- ""
    }
    
    column[i] <- x
    
  }
  
  return (column)
  
}


add_missing_values <- function(path_to_file){
  
  
  setwd("C:/Users/victor/Documents/TFG/r-analytics-master/ressources/current_db")
  data <- read.csv(path_to_file, sep = ",")
  
  data_colnames <- colnames(data)
  
  num_categoric <- data_colnames[grepl("_numeric", data_colnames)]
  
  aux <- length(data_colnames)
  
  for (i in 1:aux){
    colname <- data_colnames[i]
    
    progress <- round((100*i/aux),digits = 0)
    print(paste(progress,"%", sep = ""))
    
    if(colname %in% categoric_vars){
      column <- unlist(data[colname])
      data[colname] <- replace_with_Missing_categoric(column)
    }
    
    if(colname %in% num_categoric){
      column <- unlist(data[colname])
      data[colname] <- replace_with_Missing_num_categoric(column)
    }
    
    if(colname %in% continuous){
      column <- unlist(data[colname])
      data[colname] <- replace_with_Missing_continuous(column)
    }
    
    
  }
  
  return(data)
  
  
}