Internal: Fix duplicate links handling with improved LP checks and deletion logic - refs BT#22323

main/inc/lib/link.lib.php

@@ -383,22 +383,14 @@ public static function addlinkcategory($type)
 
     /**
      * Used to delete a link or a category.
-     *
-     * @author Patrick Cool <[email protected]>, Ghent University
-     *
-     * @param int    $id
-     * @param string $type The type of item to delete
-     *
-     * @return bool
      */
-    public static function deletelinkcategory($id, $type)
+    public static function deletelinkcategory(int $id, string $type, $courseId = null, bool $removeContentFromDb = false): bool
     {
-        $courseInfo = api_get_course_info();
-        $tbl_link = Database::get_course_table(TABLE_LINK);
-        $tbl_categories = Database::get_course_table(TABLE_LINK_CATEGORY);
-
-        $course_id = $courseInfo['real_id'];
-        $id = intval($id);
+        $courseInfo = api_get_course_info_by_id($courseId);
+        $tblLink = Database::get_course_table(TABLE_LINK);
+        $tblCategories = Database::get_course_table(TABLE_LINK_CATEGORY);
+        $tblItemProperty = Database::get_course_table(TABLE_ITEM_PROPERTY);
+        $courseIdReal = $courseInfo['real_id'];
 
         if (empty($id)) {
             return false;
@@ -407,45 +399,74 @@ public static function deletelinkcategory($id, $type)
         $result = false;
         switch ($type) {
             case 'link':
-                // -> Items are no longer physically deleted,
-                // but the visibility is set to 2 (in item_property).
-                // This will make a restore function possible for the platform administrator.
-                $sql = "UPDATE $tbl_link SET on_homepage='0'
-                        WHERE c_id = $course_id AND id='".$id."'";
-                Database::query($sql);
+                if ($removeContentFromDb) {
+                    // Hard delete: Remove from both c_link and item_property
+                    $sql = "DELETE FROM $tblItemProperty
+                        WHERE c_id = $courseIdReal AND ref = $id AND tool = '".TOOL_LINK."'";
+                    Database::query($sql);
 
-                api_item_property_update(
-                    $courseInfo,
-                    TOOL_LINK,
-                    $id,
-                    'delete',
-                    api_get_user_id()
-                );
-                self::delete_link_from_search_engine(api_get_course_id(), $id);
-                Skill::deleteSkillsFromItem($id, ITEM_TYPE_LINK);
-                Display::addFlash(Display::return_message(get_lang('LinkDeleted')));
-                $result = true;
+                    $sql = "DELETE FROM $tblLink
+                        WHERE c_id = $courseIdReal AND id = $id";
+                    Database::query($sql);
+
+                    self::delete_link_from_search_engine(api_get_course_id(), $id);
+                    Skill::deleteSkillsFromItem($id, ITEM_TYPE_LINK);
+                    $result = true;
+                } else {
+                    // Soft delete: Update visibility in item_property
+                    $sql = "UPDATE $tblLink SET on_homepage='0'
+                        WHERE c_id = $courseIdReal AND id='$id'";
+                    Database::query($sql);
+
+                    api_item_property_update(
+                        $courseInfo,
+                        TOOL_LINK,
+                        $id,
+                        'delete',
+                        api_get_user_id()
+                    );
+                    self::delete_link_from_search_engine(api_get_course_id(), $id);
+                    Skill::deleteSkillsFromItem($id, ITEM_TYPE_LINK);
+                    Display::addFlash(Display::return_message(get_lang('LinkDeleted')));
+                    $result = true;
+                }
                 break;
             case 'category':
-                // First we delete the category itself and afterwards all the links of this category.
-                $sql = "DELETE FROM ".$tbl_categories."
-                        WHERE c_id = $course_id AND id='".$id."'";
-                Database::query($sql);
+                if ($removeContentFromDb) {
+                    // Hard delete: Remove category and its links
+                    $sql = "DELETE FROM $tblCategories
+                        WHERE c_id = $courseIdReal AND id = $id";
+                    Database::query($sql);
 
-                $sql = "DELETE FROM ".$tbl_link."
-                        WHERE c_id = $course_id AND category_id='".$id."'";
-                Database::query($sql);
+                    $sql = "DELETE FROM $tblLink
+                        WHERE c_id = $courseIdReal AND category_id = $id";
+                    Database::query($sql);
 
-                api_item_property_update(
-                    $courseInfo,
-                    TOOL_LINK_CATEGORY,
-                    $id,
-                    'delete',
-                    api_get_user_id()
-                );
+                    $sql = "DELETE FROM $tblItemProperty
+                        WHERE c_id = $courseIdReal AND ref = $id AND tool = '".TOOL_LINK_CATEGORY."'";
+                    Database::query($sql);
+                    $result = true;
+                } else {
+                    // Soft delete: Update visibility in item_property
+                    $sql = "DELETE FROM $tblCategories
+                        WHERE c_id = $courseIdReal AND id = $id";
+                    Database::query($sql);
+
+                    $sql = "DELETE FROM $tblLink
+                        WHERE c_id = $courseIdReal AND category_id = $id";
+                    Database::query($sql);
+
+                    api_item_property_update(
+                        $courseInfo,
+                        TOOL_LINK_CATEGORY,
+                        $id,
+                        'delete',
+                        api_get_user_id()
+                    );
 
-                Display::addFlash(Display::return_message(get_lang('CategoryDeleted')));
-                $result = true;
+                    Display::addFlash(Display::return_message(get_lang('CategoryDeleted')));
+                    $result = true;
+                }
                 break;
         }
 

main/link/link.php

@@ -88,7 +88,7 @@ function check_url(id, url) {
 Event::event_access_tool(TOOL_LINK);
 
 /*	Action Handling */
-$id = isset($_REQUEST['id']) ? $_REQUEST['id'] : null;
+$id = isset($_REQUEST['id']) ? (int) $_REQUEST['id'] : null;
 $scope = isset($_REQUEST['scope']) ? $_REQUEST['scope'] : null;
 $show = isset($_REQUEST['show']) && in_array(trim($_REQUEST['show']), ['all', 'none']) ? $_REQUEST['show'] : 'all';
 $categoryId = isset($_REQUEST['category_id']) ? (int) $_REQUEST['category_id'] : '';

tests/scripts/delete_duplicate_links.php

@@ -0,0 +1,140 @@
+<?php
+/* For licensing terms, see /license.txt */
+/**
+ * This script removes duplicated links.
+ * It identifies duplicate links by URL,
+ * makes sure no usage is associated with the duplicated link, and
+ * that the duplicated link is not used in a learning path.
+ * A duplicated link will generally match the following criteria:
+ * - same URL field as the original
+ * - same title
+ * - same category_id
+ * - same on_homepage field
+ * - same target
+ * - same course, same session (otherwise considered a different link, a voluntary copy)
+ * - each have entries in c_item_property because it was created legitimately
+ * Possible duplicates can be found with a query like:
+ * SELECT iid, c_id, session_id, url, title, description, target, on_homepage FROM c_link WHERE c_id = 470 AND url like '%\__.%' ORDER BY url, title;
+ * This script should be located inside the tests/scripts/ folder to work.
+ * It can be run more than one time as it will only ever affect duplicate
+ * links.
+ * If you have a very large number of links, we recommend you temporarily
+ * comment out the api_item_property_update() calls in
+ * Link::deletelinkcategory() (which deletes a link *or* a category).
+ * Chances are there is not even a registry of those links there in the
+ * first place (they were probably duplicated through a short/broken process) and
+ * this is where most of the time is spent during deletion.
+ * @author Yannick Warnier <[email protected]>
+ */
+exit; //remove this line to execute from the command line
+
+use ChamiloSession as Session;
+
+ini_set('memory_limit', '256M');
+
+if (PHP_SAPI !== 'cli') {
+    die('This script can only be executed from the command line');
+}
+
+require_once __DIR__.'/../../main/inc/global.inc.php';
+
+$debug = true;
+$_user['user_id'] = 1;
+Session::write('_user', $_user);
+
+echo "[" . time() . "] Querying courses\n";
+$sql = "SELECT id, code FROM course ORDER BY id";
+
+$resCourse = Database::query($sql);
+if ($resCourse === false) {
+    exit("Could not find any course\n");
+}
+$countCourses = Database::num_rows($resCourse);
+echo "[" . time() . "] Found $countCourses courses\n";
+
+$duplicatesCount = 0;
+$originalsCount = 0;
+$deletedCount = 0;
+$itemsInLP = 0;
+
+// Iterate through each course
+while ($course = Database::fetch_assoc($resCourse)) {
+    if (empty($course['id'])) {
+        continue; // Skip invalid course IDs
+    }
+
+    $sql2 = "SELECT iid, url, title, description, category_id, on_homepage, target, session_id
+             FROM c_link
+             WHERE c_id = " . $course['id'] . "
+             AND (session_id = 0 OR session_id IS NULL)
+             ORDER BY url, title, iid";
+    $res2 = Database::query($sql2);
+
+    if ($res2 === false) {
+        die("Error querying links in course code " . $course['code'] . "\n");
+    }
+
+    $links = [];
+    while ($item = Database::fetch_assoc($res2)) {
+        $links[] = $item;
+    }
+
+    // Track processed duplicates to avoid redundant operations
+    $processedDuplicates = [];
+
+    foreach ($links as $key => $original) {
+        $originalsCount++;
+        foreach ($links as $key2 => $duplicate) {
+            if (
+                $key !== $key2 &&
+                !in_array($duplicate['iid'], $processedDuplicates) &&
+                $original['url'] === $duplicate['url'] &&
+                $original['title'] === $duplicate['title'] &&
+                $original['description'] === $duplicate['description'] &&
+                $original['category_id'] === $duplicate['category_id'] &&
+                $original['on_homepage'] === $duplicate['on_homepage'] &&
+                $original['target'] === $duplicate['target'] &&
+                $original['session_id'] === $duplicate['session_id'] &&
+                $original['iid'] < $duplicate['iid']
+            ) {
+                $duplicatesCount++;
+                if ($debug) {
+                    echo "\nDuplicate found in Course ID: " . $course['id'] . "\n";
+                    echo "Original IID=" . $original['iid'] . ", Duplicate IID=" . $duplicate['iid'] . "\n";
+                }
+
+                // Check if duplicate exists in c_lp_item
+                $checkSql = "SELECT COUNT(*) as count FROM c_lp_item
+                             WHERE ref = " . $duplicate['iid'] . "
+                             AND c_id = " . $course['id'] . "
+                             AND item_type = 'link'";
+                $checkResult = Database::query($checkSql);
+                $row = Database::fetch_assoc($checkResult);
+
+                if ($row['count'] > 0) {
+                    $itemsInLP++;
+                    if ($debug) {
+                        echo "Duplicate in learning path: IID=" . $duplicate['iid'] . " (Original IID=" . $original['iid'] . ")\n";
+                    }
+                    continue; // Skip duplicates in learning paths
+                }
+
+                // Delete the duplicate
+                Link::deletelinkcategory($duplicate['iid'], 'link', $course['id'], true);
+                $deletedCount++;
+                $processedDuplicates[] = $duplicate['iid']; // Mark as processed
+                if ($debug) {
+                    echo "Deleted Duplicate IID=" . $duplicate['iid'] . "\n";
+                }
+            }
+        }
+    }
+}
+
+// Summary
+echo "\nSummary:\n";
+echo "- Total duplicates detected: $duplicatesCount\n";
+echo "- Duplicates ignored (in learning paths): $itemsInLP\n";
+echo "- Duplicates deleted: $deletedCount\n";
+
+echo "[" . time() . "] Process complete.\n";