Fix error in some cases Union find algorithm

README.md

@@ -14,17 +14,21 @@ Example of input data (based on the csv file):
 ```
 ID,PARENT_ID,EMAIL,CARD,PHONE,TMP
 1,NULL,email1,card1,phone1,
-2,NULL,email2,card2,phone2,
+2,NULL,email2,card1,phone2,
 3,NULL,email3,card3,phone3,
 4,NULL,email1,card2,phone4,
 5,NULL,email5,card5,phone2,
 6,NULL,email6,card6,phone6,
 7,NULL,email3,card9,phone7,
 8,NULL,email8,card10,phone8,
-9,NULL,email9,card9,phone3,     
-10,NULL,email10,card10,phone10,
+9,NULL,email9,card9,phone3,
+10,NULL,email2,card10,phone10,
 ```
 
+In the example of the element with **ID 10** it was associated with 2,8,4,1. Original duplicate 1. Brief visualization of dependencies:
+- **ID1 => ID2 => ID10 => ID8**
+
+
 Require
 --
 - php

README_RU.md

@@ -14,17 +14,20 @@ Problem/Motivation
 ```
 ID,PARENT_ID,EMAIL,CARD,PHONE,TMP
 1,NULL,email1,card1,phone1,
-2,NULL,email2,card2,phone2,
+2,NULL,email2,card1,phone2,
 3,NULL,email3,card3,phone3,
 4,NULL,email1,card2,phone4,
 5,NULL,email5,card5,phone2,
 6,NULL,email6,card6,phone6,
 7,NULL,email3,card9,phone7,
 8,NULL,email8,card10,phone8,
-9,NULL,email9,card9,phone3,     
-10,NULL,email10,card10,phone10,
+9,NULL,email9,card9,phone3,
+10,NULL,email2,card10,phone10,
 ```
 
+На примере элемента с **ID 10** его связали с 2,8,4,1. Оригинальный дубликат 1. Краткая визуализация зависимостей:
+- **ID1 => ID2 => ID10 => ID8**
+
 Require
 --
 - php

README_UA.md

@@ -14,17 +14,23 @@ Problem/Motivation
 ```
 ID,PARENT_ID,EMAIL,CARD,PHONE,TMP
 1,NULL,email1,card1,phone1,
-2,NULL,email2,card2,phone2,
+2,NULL,email2,card1,phone2,
 3,NULL,email3,card3,phone3,
 4,NULL,email1,card2,phone4,
 5,NULL,email5,card5,phone2,
 6,NULL,email6,card6,phone6,
 7,NULL,email3,card9,phone7,
 8,NULL,email8,card10,phone8,
-9,NULL,email9,card9,phone3,     
-10,NULL,email10,card10,phone10,
+9,NULL,email9,card9,phone3,
+10,NULL,email2,card10,phone10,
 ```
 
+На прикладі елементу з **ID 10** його зв'язали з 2,8,4,1. Оригінальний дублікат 1. Коротка візуалізація залежностей:
+ID1 > ID2> ID10 > ID8
+- **ID1 => ID2 => ID10 => ID8**
+
+
+
 Require
 --
 - php

index.php

@@ -7,22 +7,23 @@
  */
 
 // Define constants.
-define('FIELDS', ['EMAIL', 'CARD', 'PHONE']);
+define('DUPLICATES_FIELDS', ['EMAIL', 'CARD', 'PHONE']);
 
 // Default example data.
 $csv = 'ID,PARENT_ID,EMAIL,CARD,PHONE,TMP
 1,NULL,email1,card1,phone1,
 2,NULL,email2,card1,phone2,
 3,NULL,email3,card3,phone3,
-4,NULL,email1,card2,phone4,                                                                                                                                                          
+4,NULL,email1,card2,phone4,
 5,NULL,email5,card5,phone2,
 6,NULL,email6,card6,phone6,
 7,NULL,email3,card9,phone7,
 8,NULL,email8,card10,phone8,
-9,NULL,email9,card9,phone3,                                                                                                                                                          
-10,NULL,email10,card10,phone10,';
+9,NULL,email9,card9,phone3,
+10,NULL,email2,card10,phone10,';
 
 $rows = explode(PHP_EOL, $csv);
+$fields_array = [];
 
 // Prepare array data.
 foreach ($rows as $key => $row) {
@@ -41,6 +42,7 @@
 $csv_string = 'ID,PARENT_ID' . PHP_EOL;
 
 $mapping_fields = [];
+$grouping_key = [];
 
 // Find duplicates and save to mapping.
 foreach ($fields_array as $key => $array) {
@@ -51,32 +53,40 @@
   }
 
   // Set default value for each iteration.
-  $group = NULL;
+  $group = $group_key = NULL;
   $group_to_merge = [];
 
   // Grouping by fields.
-  foreach (FIELDS as $field) {
+  foreach (DUPLICATES_FIELDS as $field) {
     $field_value = $array[$field];
     if (array_key_exists($array[$field], $mapping_fields)) {
-      $group = $mapping_fields[$field_value];
-      $group_to_merge[] = $group;
+      $group_key = $mapping_fields[$field_value];
+      $group_to_merge[] = $group_key;
     }
   }
 
-  // Setting minimal group if have more one group ID.
-  if (count($group_to_merge) > 1) {
-    $group = min($group_to_merge);
+  // Setting group if do not have any duplicates.
+  if ($group_key === NULL) {
+    $grouping_key[] = $array['ID'];
+    $group_key = array_search($array['ID'], $grouping_key);
   }
+  $group = $grouping_key[$group_key];
 
-  // Setting group if do not have any duplicates.
-  if ($group === NULL) {
-    $group = $array['ID'];
+  // Setting minimal group if have more one group ID.
+  if (count($group_to_merge) > 1) {
+    for ($i = 0; $i < count($group_to_merge); $i++) {
+      $merging_array[] = $grouping_key[$group_to_merge[$i]];
+    }
+    if (!empty($merging_array)) {
+      $group = min($merging_array);
+      $group_key = array_search($group, $grouping_key);
+    }
   }
 
   // Save fields to mapping.
-  $mapping_fields[$array['EMAIL']] = $group;
-  $mapping_fields[$array['CARD']] = $group;
-  $mapping_fields[$array['PHONE']] = $group;
+  $mapping_fields[$array['EMAIL']] = $group_key;
+  $mapping_fields[$array['CARD']] = $group_key;
+  $mapping_fields[$array['PHONE']] = $group_key;
 
 }
 
@@ -85,8 +95,13 @@
   if ($key === 0) {
     continue;
   }
-  // Searching PARENT_ID by email field. May be any field (like: CARD, PHONE).
-  $fields_array[$key]['PARENT_ID'] = $mapping_fields[$array['EMAIL']];
+
+  $parent_ids = NULL;
+  // Searching PARENT_ID by fields.
+  foreach (DUPLICATES_FIELDS as $field) {
+    $parent_ids[] = $grouping_key[$mapping_fields[$array[$field]]];
+  }
+  $fields_array[$key]['PARENT_ID'] = min($parent_ids);
 
   // Prepare data from csv.
   if ($key !== 0) {