From 12a192eb6f739977b6b664de3b1db56b0a7c8a8d Mon Sep 17 00:00:00 2001
From: GNajeral <90567992+GNajeral@users.noreply.github.com>
Date: Fri, 31 Mar 2023 12:35:30 +0200
Subject: [PATCH] valid variables script local, values not duplicated

---
 valid_variables_script_local.R | 212 +++++++++++++++++++++++++--------
 1 file changed, 160 insertions(+), 52 deletions(-)

diff --git a/valid_variables_script_local.R b/valid_variables_script_local.R
index 909c3ef..0c5b2f3 100755
--- a/valid_variables_script_local.R
+++ b/valid_variables_script_local.R
@@ -1,6 +1,7 @@
 rm(list=ls())
 
 dir_name <- readline("Introduce the name of the directory please: ")
+# C:\Users\guill\Documents\harmonize_scripts
 
 setwd(dir_name)
 
@@ -98,94 +99,201 @@ check_valid_name <- function(col_name){
   
 }
 
-check_valid_values_continuous <- function(colname , codebook_param , column){
+# check_valid_values_continuous <- function(colname , codebook_param , column){
+#   
+#   column <- column[column != "."]
+#   possible_values_format <- codebook_param$Possible.values.format[codebook_param$Harmonised.variable.name == colname]
+#   possible_values_list =  str_split(possible_values_format , "/")[[1]]
+#   
+#   range_as_str <- str_trim(possible_values_list[1])
+#   missing_value_format <- str_trim(str_trim(possible_values_list[2]))
+#   
+#   separate_range <- str_split(range_as_str , "-")[[1]]
+#   min_value  <- strtoi(separate_range[1]) 
+#   max_value  <- strtoi(separate_range[2])
+#   
+#   failing_values <- column[column < min_value | column > max_value]
+#   number_of_failing_values <- length(failing_values[!is.na(failing_values)])
+#   
+#   str_res <- ""
+#   if (number_of_failing_values == 0)
+#     str_res <- "No failing values"
+#   else{
+#     failing_values <- failing_values[!is.na(failing_values)]
+#     failing_value_counts <- table(failing_values)
+#     
+#     str_res <- paste(colname, "has failing values:")
+#     
+#     for (i in seq_along(failing_value_counts)) {
+#       value <- names(failing_value_counts)[i]
+#       count <- failing_value_counts[i]
+#       str_res <- paste(str_res, paste(value, "(", count, "times)", collapse = " "), sep = " ")
+#     }
+#     
+#     str_res <- paste(str_res, "should be in range", range_as_str, "(continuous)", sep = " ")
+#   }
+#   
+#   return(str_res)
+# }
+# 
+# 
+# 
+# check_valid_values_categorical <- function(colname, codebook_param, column) {
+#   column <- column[column != "."]
+#   possible_values_format <- codebook_param$Possible.values.format[codebook_param$Harmonised.variable.name == colname]
+#   possible_values_list <- str_split(possible_values_format, "/")[[1]]
+#   
+#   possible_values_list <- lapply(possible_values_list, str_trim)
+#   
+#   str_res <- ""
+#   min_value <- 0
+#   max_value <- 0
+#   if (length(possible_values_list[[1]]) == 2) {
+#     separate_range <- str_split(possible_values_list[[1]][1], "-")[[1]]
+#     min_value <- strtoi(separate_range[1])
+#     max_value <- strtoi(separate_range[2])
+#   } else {
+#     possible_values_list <- lapply(possible_values_list, strtoi)[[1]]
+#     min_value <- possible_values_list[1]
+#     max_value <- possible_values_list[length(possible_values_list) - 1]
+#   }
+#   
+#   failing_values <- column[column < min_value | column > max_value]
+#   number_of_failing_values <- length(failing_values[!is.na(failing_values)])
+#   
+#   if (number_of_failing_values == 0) {
+#     str_res <- "No failing values"
+#   } else {
+#     range_as_str <- paste(min_value, "-", max_value, " (categorical)")
+#     failing_values <- failing_values[!is.na(failing_values)]
+#     failing_value_counts <- table(failing_values)
+#     
+#     str_res <- paste(colname, "has failing values:")
+#     
+#     for (i in seq_along(failing_value_counts)) {
+#       value <- names(failing_value_counts)[i]
+#       count <- failing_value_counts[i]
+#       str_res <- paste(str_res, paste(value, "(", count, "times)", collapse = " "), sep = " ")
+#     }
+#     
+#     str_res <- paste(str_res, "should be in range", range_as_str, sep = " ")
+#   }
+#   
+#   return(str_res)
+# }
+# 
+# check_valid_values_binary <- function(colname, column) {
+#   column <- column[column != "."]
+#   failing_values <- column[column < 0 | column > 1]
+#   number_of_failing_values <- length(failing_values[!is.na(failing_values)])
+#   
+#   str_res <- ""
+#   if (number_of_failing_values == 0)
+#     str_res <- "No failing values"
+#   else {
+#     range_as_str <- "0-1 (binary)"
+#     failing_values <- failing_values[!is.na(failing_values)]
+#     failing_value_counts <- table(failing_values)
+#     
+#     str_res <- paste(colname, "has failing values:")
+#     
+#     for (i in seq_along(failing_value_counts)) {
+#       value <- names(failing_value_counts)[i]
+#       count <- failing_value_counts[i]
+#       str_res <- paste(str_res, paste(value, "(", count, "times)", collapse = " "), sep = " ")
+#     }
+#     
+#     str_res <- paste(str_res, "should be in range", range_as_str, sep = " ")
+#   }
+#   
+#   return(str_res)
+# }
+
+check_valid_values_continuous <- function(colname, codebook_param, column) {
   
   column <- column[column != "."]
   possible_values_format <- codebook_param$Possible.values.format[codebook_param$Harmonised.variable.name == colname]
-  possible_values_list =  str_split(possible_values_format , "/")[[1]]
+  possible_values_list =  str_split(possible_values_format, "/")[[1]]
   
-  # Fallará cuando el codebook no tenga min-max / .
   range_as_str <- str_trim(possible_values_list[1])
   missing_value_format <- str_trim(str_trim(possible_values_list[2]))
   
-  separate_range <- str_split(range_as_str , "-")[[1]]
-  min_value  <- strtoi(separate_range[1]) 
-  max_value  <- strtoi(separate_range[2])
+  separate_range <- str_split(range_as_str, "-")[[1]]
+  min_value <- strtoi(separate_range[1])
+  max_value <- strtoi(separate_range[2])
   
   failing_values <- column[column < min_value | column > max_value]
   number_of_failing_values <- length(failing_values[!is.na(failing_values)])
   
   str_res <- ""
-  if (number_of_failing_values == 0)
+  if (number_of_failing_values == 0) {
     str_res <- "No failing values"
-  else{
-    failing_values <- failing_values[!is.na(failing_values)]
-    str_res <- paste( colname , paste(unlist(failing_values) , collapse =" "))
-    #str_res <- paste(colname , collapse =" ")
-    str_res <- paste(str_res ,  "should be in range" , range_as_str, "(continuous)", sep = " ")
-  }
+  } else {
+    range_as_str <- paste(min_value, "-", max_value, "(continuous)")
     
-  
-  return(str_res)
-}
-
-check_valid_values_binary <- function(colname , column){
-  column <- column[column != "."]
-  failing_values <- column[column < 0 | column > 1]
-  number_of_failing_values <- length(failing_values[!is.na(failing_values)])
-  
-  str_res <- ""
-  if (number_of_failing_values == 0)
-    str_res <- "No failing values"
-  else{
-    range_as_str <- "0-1 (binary)"
-    failing_values <- failing_values[!is.na(failing_values)]
-    str_res <- paste(colname , paste(unlist(failing_values) , collapse =" "))
-    #str_res <- paste(colname , collapse =" ")
-    str_res <- paste(str_res ,  "should be in range" , range_as_str, sep = " ")
+    str_res <- paste(colname, "has", number_of_failing_values, "failing values")
+    str_res <- paste(str_res, "should be in range", range_as_str, sep = " ")
   }
-    
   
   return(str_res)
 }
 
-check_valid_values_categorical <- function(colname , codebook_param , column){
+
+check_valid_values_categorical <- function(colname, codebook_param, column) {
   column <- column[column != "."]
   possible_values_format <- codebook_param$Possible.values.format[codebook_param$Harmonised.variable.name == colname]
-  possible_values_list <-  str_split(possible_values_format , "/")[[1]]
+  possible_values_list <- str_split(possible_values_format, "/")[[1]]
   
-  possible_values_list <- lapply(possible_values_list , str_trim)
+  possible_values_list <- lapply(possible_values_list, str_trim)
   
   str_res <- ""
   min_value <- 0
-  max_value <- 0 
-  if (length(possible_values_list[[1]]) == 2){
-    separate_range <-  str_split(possible_values_list[[1]][1], "-")[[1]]
-    min_value  <- strtoi(separate_range[1]) 
-    max_value  <- strtoi(separate_range[2])
-    
-  }else{
-    
-    possible_values_list <- lapply(possible_values_list , strtoi)[[1]]
+  max_value <- 0
+  if (length(possible_values_list[[1]]) == 2) {
+    separate_range <- str_split(possible_values_list[[1]][1], "-")[[1]]
+    min_value <- strtoi(separate_range[1])
+    max_value <- strtoi(separate_range[2])
+  } else {
+    possible_values_list <- lapply(possible_values_list, strtoi)[[1]]
     min_value <- possible_values_list[1]
     max_value <- possible_values_list[length(possible_values_list) - 1]
   }
   
-  failing_values <- column[column < min_value | column > max_value ]
+  failing_values <- column[column < min_value | column > max_value]
   number_of_failing_values <- length(failing_values[!is.na(failing_values)])
   
-  if(number_of_failing_values == 0){
+  if (number_of_failing_values == 0) {
     str_res <- "No failing values"
-  }else{
-    range_as_str <- paste(min_value , "-" , max_value , " (categorical)")
-    failing_values <- failing_values[!is.na(failing_values)]
-    #str_res <- paste(colname , paste(unlist(failing_values) , collapse =" "))
-    str_res <- paste(colname , collapse =" ")
-    str_res <- paste(str_res ,  "should be in range" , range_as_str, sep = " ")
+  } else {
+    range_as_str <- paste(min_value, "-", max_value, " (categorical)")
+    
+    str_res <- paste(colname, "has", number_of_failing_values, "failing values")
+    str_res <- paste(str_res, "should be in range", range_as_str, sep = " ")
+  }
+  
+  return(str_res)
+}
+
+check_valid_values_binary <- function(colname, column) {
+  column <- column[column != "."]
+  failing_values <- column[column < 0 | column > 1]
+  number_of_failing_values <- length(failing_values[!is.na(failing_values)])
+  
+  str_res <- ""
+  if (number_of_failing_values == 0)
+    str_res <- "No failing values"
+  else {
+    range_as_str <- "0-1 (binary)"
+    
+    str_res <- paste(colname, "has", number_of_failing_values, "failing values")
+    str_res <- paste(str_res, "should be in range", range_as_str, sep = " ")
   }
   
+  return(str_res)
 }
 
+
+
 check_valid_values <- function(valid_colnames, codebook_param){
   
   res <- ""
-- 
2.24.1