From 841190e3e26f74d76b14d72f028c9a371cdc3340 Mon Sep 17 00:00:00 2001
From: GNajeral <90567992+GNajeral@users.noreply.github.com>
Date: Fri, 3 Mar 2023 13:59:36 +0100
Subject: [PATCH] Updated valid_variables_script2

---
 valid_variables_script2.R | 39 +++++++++++++++++++++++++++++++++------
 1 file changed, 33 insertions(+), 6 deletions(-)

diff --git a/valid_variables_script2.R b/valid_variables_script2.R
index 1b2221d..5ce27d1 100644
--- a/valid_variables_script2.R
+++ b/valid_variables_script2.R
@@ -132,8 +132,8 @@ is_number <- function(x){
 check_values_format <- function(valid_columns, codebook_param){
   res <- ""
   variables_out_of_range = "Variables out of range:"
-  for(i in 1:length(valid_columns[[1]])){
-
+#  for(i in 1:length(valid_columns[[1]])){
+  for(i in 1:9){
     current_column <- valid_columns[[1]][[i]]
     print(current_column)
     variable_type <- codebook_param$Variable.type[codebook$Harmonised.variable.name == current_column]
@@ -141,6 +141,8 @@ check_values_format <- function(valid_columns, codebook_param){
     if (is.na(variable_type)){
       next
     }
+    
+    print(variable_type)
    
     if(variable_type == "Continuous"){
       
@@ -149,8 +151,12 @@ check_values_format <- function(valid_columns, codebook_param){
       ### parse del formato de una variable continua ##
       ## esta sentencia funciona codebook$Possible.values.format[codebook$Harmonised.variable.name == "CMXDE"] pruebala en el interprete.
       value_format <- strsplit(codebook_param$Possible.values.format[codebook_param$Harmonised.variable.name == current_column], " / ")[[1]]
-      high_limit <- gsub(",", ".", (sub("-.*", "", value_format[1])))
-      low_limit <- gsub(",", ".", (sub(".*-", "", value_format[1])))
+      high_limit <- str_trim(gsub(",", ".", (sub("-.*", "", value_format[1]))))
+      low_limit <- str_trim(gsub(",", ".", (sub(".*-", "", value_format[1]))))
+      if(low_limit == ""){
+        high_limit <- str_trim(sub(",.*", "", value_format[1]))
+        low_limit <- str_trim(strtrimsub(".*,", "", value_format[1]))
+      }
       ### parse del formato de una variable continua ##
       
       tryCatch(
@@ -190,8 +196,29 @@ check_values_format <- function(valid_columns, codebook_param){
       ################## FIN ESTO PODRÍA IR EN UNA FUNC DIFERENTE #############
       
     }else if (variable_type == "Categorical" || variable_type == "Binary"){
-      contingency_table <- ds.table(paste("data$",current_column))
-      counts <- contingency_table[[1]][[3]]
+      value_format <- lapply(strsplit(mierda, "/") , str_trim)[[1]]
+      tryCatch(
+        error = function(cnd) {
+          if(grepl("list them with datashield.errors()",cnd))
+            error <- paste("Unable to analyse data" , datashield.errors() , sep = " ")
+          else
+            error <- paste("Unable to analyse data" , cnd, sep = " ")
+          print(error)
+          res <- c(res, error)
+        },
+        {
+          contingency_table <- ds.table(paste("data$",current_column,sep=""))
+          row_names <- rownames(contingency_table[[1]][[3]])
+          for (i in 1:length(row_names)) {
+            if(row_names[i] == "NA")
+              next
+            if(!row_names[i] %in% value_format){
+              variables_out_of_range <- paste(variables_out_of_range, current_column, sep = " ")
+              print(paste(current_column,  "does not follow the established format", sep=" "))
+            }
+          }
+        }
+      )
     }
   }
   return (variables_out_of_range)
-- 
2.24.1