From bd683608665115206300ff18da44e8ad13c09989 Mon Sep 17 00:00:00 2001
From: GbArc <gabriel.arczynski@devtactics.net>
Date: Fri, 19 May 2023 13:41:26 +0200
Subject: [PATCH 1/4] params to remove, udpated dirs

---
 cassandra_consistency_script.php | 69 +++++++++++++++++++++-----------
 1 file changed, 45 insertions(+), 24 deletions(-)

diff --git a/cassandra_consistency_script.php b/cassandra_consistency_script.php
index 68e0ef9..d7f253a 100644
--- a/cassandra_consistency_script.php
+++ b/cassandra_consistency_script.php
@@ -12,6 +12,8 @@ class DataConsistencyChecker
     const DEFAULT_PAGE_SIZE = 3;
     const CASSANDRA_USERNAME = 'cassandra';
     const CASSANDRA_PASSWORD = 'cassandra';
+    const CASSANDRA_KEYSPACE = 'tr_key';
+
     private $_cluster;
     private $session;
     private $cassandra;
@@ -19,7 +21,7 @@ class DataConsistencyChecker
     private $directory;
     private $structured_directory = '/usr/share/nginx/html/testrail/db/cassandra';
 
-    private $retrived_csv = '/usr/share/nginx/html/testrail/db/cassandra/result_from_physical_files.csv';
+    private $retrived_csv;
     static $clientId;
     private static $schemaVersion = 1;
     static $fileBucketCounter = 1;
@@ -39,12 +41,16 @@ class DataConsistencyChecker
     public function init(): void
     {
         $this->_cluster = Cassandra::cluster()
-            ->withContactPoints('cassandra')
-            ->withPort(9042)
+            ->withContactPoints(static::$cassandraHost)
+	    ->withPort(9042)
+            ->withCredentials(
+                static::CASSANDRA_USERNAME,
+                static::CASSANDRA_PASSWORD
+            )
             ->build();
         if ($this->_cluster) {
             try {
-                $this->cassandra = $this->_cluster->connect("tr_key");
+                $this->cassandra = $this->_cluster->connect(static::CASSANDRA_KEYSPACE);
             } catch (Exception $e) {
                 echo "err\n";
             }
@@ -55,8 +61,8 @@ class DataConsistencyChecker
 
     public function runFromCommandLine($arguments)
     {
-        $shortOptions = "hd:v";
-        $longOptions = ["help", "directory:", "version:", "v"];
+        $shortOptions = "hd:v:o:r:s:";
+        $longOptions = ["help", "directory:", "version:", "v", "output:", "o", "remove:", "r", "source:", "s"];
 
         $options = getopt($shortOptions, $longOptions);
 
@@ -68,18 +74,32 @@ class DataConsistencyChecker
         $directory = isset($options['directory']) ? $options['directory'] : (isset($options['d']) ? $options['d'] : null);
         $schemaVersion = isset($options['version']) ? $options['version'] : (isset($options['v']) ? $options['v'] : null);
 
-        if ($directory === null || $schemaVersion === null) {
+        $source = isset($options['source']) ? $options['source'] : (isset($options['s']) ? $options['s'] : null);
+        $remove = isset($options['remove']) ? $options['remove'] : (isset($options['r']) ? $options['r'] : null);
+
+        $this->structured_directory = isset($options['output']) ? $options['output'] : (isset($options['o']) ? $options['o'] : null);
+
+        if (($directory === null || $schemaVersion === null) && $remove === null && $source === null) {
             echo "Missing Attachment directory or schema version.\n";
             exit;
         }
-        if (!in_array($schemaVersion, [1, 2])) {
+	    if ($schemaVersion && !in_array($schemaVersion, [1, 2])) {
             echo "Invalid schema version. Only versions 1 and 2 are supported.\n";
             exit;
-        }
+	    }
+	    if ($this->structured_directory == null) {
+		    $this->structured_directory = './';
+	    }
 
-        static::$schemaVersion = (int)$schemaVersion;
-    
-        $this->checkConsistency('attachment_file_info');
+	    static::$schemaVersion = (int)$schemaVersion;
+	    $this->directory = $directory;
+	    $this->retrived_csv =  './result_from_physical_files.csv';
+        if ($remove && $source) {
+            $this->processAttachmentDeletionCSV($remove, $source);
+        }
+        else {
+            $this->checkConsistency('attachment_file_info');
+        }
         exit;
     }   
 
@@ -109,7 +129,7 @@ class DataConsistencyChecker
         $dbEntries = $this->getDbEntries($tableName);
         $fileEntries = $this->getFileEntries($this->directory);
         $this->process_files_in_directory($this->structured_directory);
-        $this->createCSVById($this->retrived_csv);
+        //$this->createCSVById($this->retrived_csv);
     }
 
     private function getFileEntries($directory)
@@ -177,7 +197,7 @@ class DataConsistencyChecker
 
     private function createPhysicalFileCSV($clientId, $entries)
     {
-        $fileName = "physical_" . $clientId . ".csv";
+        $fileName = $this->structured_directory . "physical_" . $clientId . ".csv";
         $csvFile = fopen($fileName, 'w');
         fputcsv($csvFile, ['id', 'size', 'creation_time']);
         foreach ($entries as $entry) {
@@ -194,7 +214,7 @@ class DataConsistencyChecker
 
     private function createDBFileCSV($clientId, $entries)
     {
-        $fileName = "cassandra_" . (string) $clientId . ".csv";
+        $fileName = $this->structured_directory . "cassandra_" . (string) $clientId . ".csv";
         $csvFile = fopen($fileName, 'w');
 
         $headers = ['id', 'size', 'creation_time', 'filename', 'bucket', 'client_id', 'attachment_id'];
@@ -305,7 +325,10 @@ class DataConsistencyChecker
         $old_migrated_data = array_search('migrated_files_id', $headers);
 
 
-        $query = "SELECT * FROM attachment_file_info WHERE id = ? ALLOW FILTERING";
+	$query = "SELECT * FROM attachment_file_info WHERE id = ? and client_id = ?";
+	if ($this->schema_version() == 2) {
+		$query = $query . ' and bucket = ?';
+	}
         $statement = $this->cassandra->prepare($query);
 
         foreach ($fileData as $row) {
@@ -492,7 +515,7 @@ class DataConsistencyChecker
                     if (!isset($physical_files_assoc[$id])) {
                         $missing_physical_files[] = [
                             'id' => $id,
-                            'file1' => [$id, $data[0], $data[1], $data[2], $data[3], $data[4], $data[5], $data[6]],
+                            'file1' => [$id, $data[0], $data[1], $data[2], $data[3], $data[4], $data[5]],
                         ];
                     }
                 }
@@ -527,7 +550,7 @@ class DataConsistencyChecker
             }
             if (count($values) > 3) {
 
-                $file_assoc[$values[0]] = [$values[1], $values[2], $values[3], $values[4], $values[5], $values[0]];
+                $file_assoc[$values[0]] = [$values[1], $values[2], $values[3], $values[4], $values[5], $values[6]];
             }
         }
 
@@ -893,7 +916,7 @@ class DataConsistencyChecker
         ];
 
         if ($this->schema_version() == 1) {
-            $query = $this->cassandra->prepare('SELECT * FROM attachment_file_info WHERE id = ? AND client_id = ? ALLOW FILTERING');
+            $query = $this->cassandra->prepare('SELECT * FROM attachment_file_info WHERE id = ? AND client_id = ?');
 	} else {
 		$q = 'SELECT * FROM attachment_file_info WHERE id = ? AND client_id = ? AND bucket = \'' . $bucketId . '\'';
 		echo "Q: [" . $q . "]\n";
@@ -991,7 +1014,7 @@ class DataConsistencyChecker
 	    $key = $this->_get_attachment_key($attachmentId);
             if ($key) {
                 if ($this->schema_version() === 1) {
-                    $query = $this->cassandra->prepare('SELECT ' . $properties . ' FROM attachments WHERE id = ? AND client_id = ? AND project_id = ? AND entity_type = ? ALLOW FILTERING');
+                    $query = $this->cassandra->prepare('SELECT ' . $properties . ' FROM attachments WHERE id = ? AND client_id = ? AND project_id = ? AND entity_type = ?');
                     $arguments = [
                         'client_id' => static::$clientId,
                         'id' => $key->id,
@@ -1261,11 +1284,10 @@ class DataConsistencyChecker
                     $data->bucket = $values[8];
                     $data->id = $values[9];
                 }
-                if ($data->source == $src) {
-                    if ($src === 'cassandra') {
+                    if ($data->source === 'Attachment' && $src === 'cassandra') {
                         echo "will delete " . $data->clientId . " : " . $data->bucket . " : "  . $data->id . PHP_EOL;
                         fwrite($logHandle, "Deleted attachment: $data->id" . PHP_EOL);
-                    } else {
+                    } else if ($src === 'File' && $source === 'file'){
                         $filePath = $directory . $values[1];
                         $thumb1Path = $directory . $values[1];
                         $thumb2Path = $directory . $values[1];
@@ -1280,7 +1302,6 @@ class DataConsistencyChecker
                             echo "File not found: $filePath" . PHP_EOL;
                         }
                     }
-                }
             }
         }
 

From a35654a07152b64f5d316f99ed8562177f10f22f Mon Sep 17 00:00:00 2001
From: Spase <spase.papcev@gmail.com>
Date: Fri, 19 May 2023 19:01:57 +0200
Subject: [PATCH 2/4] Added additional changes and documentation

---
 cassandra_consistency_script.php | 647 ++++++++++++++++---------------
 config.ini                       |  15 +
 2 files changed, 357 insertions(+), 305 deletions(-)
 create mode 100644 config.ini

diff --git a/cassandra_consistency_script.php b/cassandra_consistency_script.php
index 68e0ef9..5a6522d 100644
--- a/cassandra_consistency_script.php
+++ b/cassandra_consistency_script.php
@@ -1,28 +1,17 @@
 <?php
-ini_set('display_errors', 1);
-ini_set('display_startup_errors', 1);
-error_reporting(E_ALL);
-use Cassandra\Cluster;
-use Cassandra\ExecutionOptions;
-use Cassandra\Session;
 
 class DataConsistencyChecker
 {
-    const OLD_ATTACHMENTS_PREFIX = 'E_';
-    const DEFAULT_PAGE_SIZE = 3;
-    const CASSANDRA_USERNAME = 'cassandra';
-    const CASSANDRA_PASSWORD = 'cassandra';
+    const DEFAULT_PAGE_SIZE = 30;
     private $_cluster;
     private $session;
     private $cassandra;
-    //private $directory = '/usr/share/nginx/html/testrail/attachments';
     private $directory;
-    private $structured_directory = '/usr/share/nginx/html/testrail/db/cassandra';
-
-    private $retrived_csv = '/usr/share/nginx/html/testrail/db/cassandra/result_from_physical_files.csv';
+    private $structured_directory;
+    private $retrived_csv;
     static $clientId;
     private static $schemaVersion = 1;
-    static $fileBucketCounter = 1;
+
     static $bucketMagic = 4;
     static $cassandraHost = 'cassandra';
 
@@ -35,28 +24,43 @@ class DataConsistencyChecker
         $this->runFromCommandLine($_SERVER['argv']);
     }
 
-
+    /**
+     * Initializes the Cassandra connection based on the configuration settings.
+     *
+     * @return void
+     */
     public function init(): void
     {
+        $config = parse_ini_file("config.ini", true);
         $this->_cluster = Cassandra::cluster()
-            ->withContactPoints('cassandra')
+            ->withContactPoints($config['CASSANDRA']['host'])
             ->withPort(9042)
+            ->withCredentials(
+                $config['CASSANDRA']['user'],
+                $config['CASSANDRA']['password']
+            )
             ->build();
         if ($this->_cluster) {
             try {
-                $this->cassandra = $this->_cluster->connect("tr_key");
+                $this->cassandra = $this->_cluster->connect($config['CASSANDRA']['keyspace']);
             } catch (Exception $e) {
                 echo "err\n";
             }
         }
-        // static::$bucketMagic = defined('CASSANDRA_BUCKET_MAGIC') ? (int) CASSANDRA_BUCKET_MAGIC : 4;
+
 
     }
 
+    /**
+     * Runs the script from the command line with the provided arguments.
+     *
+     * @param array $arguments The command line arguments.
+     * @return void
+     */
     public function runFromCommandLine($arguments)
     {
-        $shortOptions = "hd:v";
-        $longOptions = ["help", "directory:", "version:", "v"];
+        $shortOptions = "hd:v:o:r:s:";
+        $longOptions = ["help", "directory:", "version:", "v", "output:", "o", "remove:", "r", "source:", "s"];
 
         $options = getopt($shortOptions, $longOptions);
 
@@ -68,22 +72,47 @@ class DataConsistencyChecker
         $directory = isset($options['directory']) ? $options['directory'] : (isset($options['d']) ? $options['d'] : null);
         $schemaVersion = isset($options['version']) ? $options['version'] : (isset($options['v']) ? $options['v'] : null);
 
-        if ($directory === null || $schemaVersion === null) {
+        $source = isset($options['source']) ? $options['source'] : (isset($options['s']) ? $options['s'] : null);
+        $remove = isset($options['remove']) ? $options['remove'] : (isset($options['r']) ? $options['r'] : null);
+
+        $structured_directory = isset($options['output']) ? $options['output'] : (isset($options['o']) ? $options['o'] : null);
+        if (!file_exists($structured_directory)) {
+            mkdir($structured_directory, 0777, true);
+        }
+        $this->structured_directory = $structured_directory;
+
+        if (($directory === null || $schemaVersion === null) && $remove === null && $source === null) {
             echo "Missing Attachment directory or schema version.\n";
             exit;
         }
-        if (!in_array($schemaVersion, [1, 2])) {
+        if ($schemaVersion && !in_array($schemaVersion, [1, 2])) {
             echo "Invalid schema version. Only versions 1 and 2 are supported.\n";
             exit;
         }
+        if ($this->structured_directory == null) {
+            $this->structured_directory = './';
+        }
 
-        static::$schemaVersion = (int)$schemaVersion;
-    
-        $this->checkConsistency('attachment_file_info');
+        static::$schemaVersion = (int) $schemaVersion;
+        $this->directory = $directory;
+        $this->retrived_csv = './result_from_physical_files.csv';
+        if ($remove && $source) {
+            $this->processAttachmentDeletionCSV($remove, $source);
+        } else {
+            $this->checkConsistency('attachment_file_info');
+            if (is_dir($this->structured_directory)) {
+                $this->removeDirectory($this->structured_directory);
+            }
+        }
         exit;
-    }   
+    }
 
-    private function displayHelpMessage()
+    /**
+     * Displays the help message with instructions on how to use the script.
+     *
+     * @return void
+     */
+    private function displayHelpMessage(): void
     {
         $helpMessage = <<<EOT
         Usage:
@@ -93,26 +122,67 @@ class DataConsistencyChecker
         -h, --help     Display this help screen.
         --version      Set the schema version.
         --directory    Set the directory path for attachments.
+        --output       Set the folder for temp files
+        --remove       We need to choose between file and cassandra what we want to remove
 
         Example:
-        php script.php --directory /path/to/directory --version schema_version
-        
+        php script_name --version schema_version --directory=/path/to/directory --output ./out/
+        For Delete:
+        php script_name --remove result_from_cassandra_entries.csv --source file - to remove missing physical files
+        php script_name --remove result_from_physical_files.csv --source cassandra - to remove missing cassandra entries
         EOT;
         echo $helpMessage;
     }
 
+    /**
+     * Recursively removes a directory and its contents.
+     *
+     * @param string $directory The directory path to be removed.
+     * @return void
+     */
+    private function removeDirectory($directory): void
+    {
+        if (!is_dir($directory)) {
+            return;
+        }
+
+        $files = array_diff(scandir($directory), ['.', '..']);
+
+        foreach ($files as $file) {
+            $path = $directory . '/' . $file;
+
+            if (is_dir($path)) {
+                $this->removeDirectory($path);
+            } else {
+                unlink($path);
+            }
+        }
+
+        rmdir($directory);
+    }
+
+    /**
+     * Checks the consistency between database entries and file entries.
+     *
+     * @param string $tableName The name of the table in the database to check consistency for.
+     * @return void
+     */
     public function checkConsistency($tableName)
     {
-
-        //initialize the cassandra connection
         $this->init();
         $dbEntries = $this->getDbEntries($tableName);
         $fileEntries = $this->getFileEntries($this->directory);
         $this->process_files_in_directory($this->structured_directory);
-        $this->createCSVById($this->retrived_csv);
     }
 
-    private function getFileEntries($directory)
+    /**
+     * Retrieves file entries from a directory and organizes them based on dynamic values.
+     *
+     * @param string $directory The directory path to retrieve file entries from.
+     * @return array An array containing the file entries organized by dynamic values.
+     */
+
+    private function getFileEntries($directory): array
     {
         $files = glob($directory . '/*');
         $entries = [];
@@ -120,15 +190,11 @@ class DataConsistencyChecker
         foreach ($files as $file) {
             if (is_file($file)) {
                 $fileName = basename($file);
-
-                // Skip files with specific suffixes
                 if (strpos($fileName, '-thumb1') !== false || strpos($fileName, '-thumb2') !== false) {
                     continue;
                 }
-
                 $dashParts = explode("-", $fileName, 2);
                 $dotParts = explode(".", $fileName);
-
                 if (count($dashParts) === 2) {
                     $clientId = $dashParts[0];
                     $id = $dashParts[1];
@@ -140,16 +206,12 @@ class DataConsistencyChecker
                     $fileParts = $dotParts;
                     $delimiter = '.';
                 } else {
-                    // Handle cases where the file name does not contain either a dash or a dot
                     continue;
                 }
 
                 $filePath = $file;
                 $size = filesize($filePath);
-
-                // Use file modification time (filemtime()) instead of file creation time (filectime())
                 $creationTime = date('Y-m-d H:i:s', filemtime($filePath));
-
                 $dynamicValue = substr($id, 0, 2);
                 $creationTime = str_replace('"', '', $creationTime);
 
@@ -162,7 +224,6 @@ class DataConsistencyChecker
                 ];
             }
         }
-
         // create CSV file for each client ID's physical file entries
         foreach ($entries as $clientId => $clientEntries) {
             $this->createPhysicalFileCSV($clientId, $clientEntries);
@@ -172,29 +233,41 @@ class DataConsistencyChecker
     }
 
 
-
-
+    /**
+     * Creates a CSV file containing the physical file entries for a specific client ID.
+     *
+     * @param string $clientId The client ID.
+     * @param array $entries An array containing the physical file entries for the client.
+     * @return void
+     */
 
     private function createPhysicalFileCSV($clientId, $entries)
     {
-        $fileName = "physical_" . $clientId . ".csv";
+        $fileName = $this->structured_directory . "physical_" . $clientId . ".csv";
         $csvFile = fopen($fileName, 'w');
         fputcsv($csvFile, ['id', 'size', 'creation_time']);
         foreach ($entries as $entry) {
             fputcsv($csvFile, [
                 $entry['file_name'],
                 $entry['size'],
-		$entry['creation_time']
+                $entry['creation_time']
             ]);
         }
         fclose($csvFile);
     }
 
 
+    /**
+     * Creates a CSV file containing the entries for a specific client ID.
+     *
+     * @param string $clientId The client ID.
+     * @param array $entries An array containing the entries for the client.
+     * @return void
+     */
 
-    private function createDBFileCSV($clientId, $entries)
+    private function createDBFileCSV($clientId, $entries): void
     {
-        $fileName = "cassandra_" . (string) $clientId . ".csv";
+        $fileName = $this->structured_directory . "cassandra_" . (string) $clientId . ".csv";
         $csvFile = fopen($fileName, 'w');
 
         $headers = ['id', 'size', 'creation_time', 'filename', 'bucket', 'client_id', 'attachment_id'];
@@ -207,7 +280,7 @@ class DataConsistencyChecker
                 $entry['creation_time'],
                 $entry['filename'],
                 $entry['bucket'],
-		        $entry['client_id'],
+                $entry['client_id'],
                 $entry['attachment_id'],
             ];
 
@@ -217,12 +290,18 @@ class DataConsistencyChecker
         fclose($csvFile);
     }
 
+    /**
+     * Retrieves entries from a database table.
+     *
+     * @param string $tableName The name of the database table.
+     * @return array An array containing the retrieved entries.
+     */
 
 
     private function getDbEntries($tableName)
-    {   
-       
-        if($this->schema_version() === 1) {
+    {
+
+        if ($this->schema_version() === 1) {
             $query = "SELECT client_id, id, size, filename, created_on FROM $tableName";
         } else {
             $query = "SELECT client_id, id, size, filename, created_on, bucket FROM $tableName";
@@ -238,6 +317,7 @@ class DataConsistencyChecker
         $entries = [];
         while ($result) {
             foreach ($result as $row) {
+
                 if (preg_match('/^[0-9]+$/', $row['id'])) {
                     $dotParts = explode(".", $row['filename'], 2);
                     $dynamicValue = substr($dotParts[1], 0, 2);
@@ -245,7 +325,6 @@ class DataConsistencyChecker
                     $date = date('Y-m-d H:i:s', $timestamp);
                     $creationTime = str_replace('"', '', $date);
                     $entry = [
-
                         "id" => $row['id'],
                         "size" => (string) $row['size'],
                         "creation_time" => $creationTime,
@@ -291,82 +370,17 @@ class DataConsistencyChecker
 
         return $entries;
     }
- 
-    private function createCSVById($csvFile)
-    {
-        $data = [];
 
 
-        $fileData = array_map('str_getcsv', file($csvFile));
-        $headers = array_shift($fileData);
+    /**
+     * Compares a file entry with its corresponding entry in the Cassandra file association.
+     *
+     * @param string $id The ID of the file entry.
+     * @param array $data An array containing data of the file entry.
+     * @param array $cassandra_file_assoc The Cassandra file association.
+     * @return array|null An array containing mismatched entries, or null if the entries match.
+     */
 
-
-        $idIndex = array_search('File path', $headers);
-        $old_migrated_data = array_search('migrated_files_id', $headers);
-
-
-        $query = "SELECT * FROM attachment_file_info WHERE id = ? ALLOW FILTERING";
-        $statement = $this->cassandra->prepare($query);
-
-        foreach ($fileData as $row) {
-            if ($old_migrated_data !== false && isset($row[$old_migrated_data])) {
-                $id = $row[$old_migrated_data];
-            } else {
-                $id = $row[2];
-            }
-
-            $options = ['arguments' => [$id]];
-
-
-            $result = $this->cassandra->execute($statement, $options);
-
-
-            foreach ($result as $row) {
-                $data[] = $row;
-            }
-        }
-
-
-        $fileName = "data_by_id.csv";
-        $csvFile = fopen($fileName, 'w');
-
-        fputcsv($csvFile, $headers);
-
-        foreach ($data as $entry) {
-            $rowData = [];
-
-            foreach ($headers as $field) {
-                $value = isset($entry[$field]) ? $entry[$field] : '';
-                $rowData[] = $value;
-            }
-
-            fputcsv($csvFile, $rowData);
-        }
-
-        fclose($csvFile);
-    }
-
-    private function parseCSVFile($file)
-    {
-        $file_contents = file_get_contents($file);
-        $file_lines = explode("\n", $file_contents);
-        $file_assoc = array();
-
-        foreach ($file_lines as $line) {
-            if ($line === reset($file_lines)) {
-                continue;
-            }
-            $values = explode(",", $line);
-            if (count($values) == 3) {
-                $values[2] = str_replace('"', '', $values[2]);
-                $file_assoc[$values[0]] = [$values[1], $values[2]];
-            }
-        }
-
-        return $file_assoc;
-    }
-
-    // Helper function to read file contents and parse CSV lines
     private function compareFileEntries($id, $data, $cassandra_file_assoc)
     {
         if (!isset($cassandra_file_assoc[$id])) {
@@ -391,13 +405,27 @@ class DataConsistencyChecker
         return null;
     }
 
-    private function getFileLines($file)
+    /**
+     * Retrieves the lines of a file and returns them as an array.
+     *
+     * @param string $file The path to the file.
+     * @return array An array containing the lines of the file.
+     */
+
+    private function getFileLines($file): array
     {
         $file_contents = file_get_contents($file);
         return explode("\n", $file_contents);
     }
 
-    private function filterAndMapEntries($entries)
+    /**
+     * Filters out null entries and maps the remaining entries to their first element.
+     *
+     * @param array $entries An array containing entries to be filtered and mapped.
+     * @return array An array of filtered and mapped entries.
+     */
+
+    private function filterAndMapEntries($entries): array
     {
         $filtered_entries = array_filter($entries);
         $mapped_entries = array_map(function ($entry) {
@@ -405,7 +433,14 @@ class DataConsistencyChecker
         }, $filtered_entries);
         return array_values($mapped_entries);
     }
-    private function process_files_in_directory($dir)
+
+    /**
+     * Processes files in a directory, performs comparisons, and generates CSV and HTML reports.
+     *
+     * @param string $dir The directory path containing the files to be processed.
+     * @return void
+     */
+    private function process_files_in_directory($dir): void
     {
         $files = glob($dir . '/*.csv');
         $physical_files = array();
@@ -436,15 +471,13 @@ class DataConsistencyChecker
                 $compared_physical[] = $this->compare_csv_files($physical_file, $cassandra_file);
                 $compared_cassandra[] = $this->compare_csv_files($cassandra_file, $physical_file);
                 $physical_entries = $this->filterAndMapEntries($compared_cassandra);
-            
-		        //echo "phe: " . var_dump($physical_entries) . PHP_EOL;
                 $cassandra_entries = $this->filterAndMapEntries($compared_physical);
 
             } else {
-		    if (!file_exists($physical_file)) {
-	    		$missing_physical_files[] = $physical_file;
+                if (!file_exists($physical_file)) {
+                    $missing_physical_files[] = $physical_file;
 
-		        echo "cfl: " . var_dump($missing_physical_files) . PHP_EOL;
+                    echo "cfl: " . var_dump($missing_physical_files) . PHP_EOL;
 
                 }
                 if (!file_exists($cassandra_file)) {
@@ -460,13 +493,13 @@ class DataConsistencyChecker
                     if ($line === reset($physical_file_lines)) {
                         continue;
                     }
-		    $values = explode(",", $line);
+                    $values = explode(",", $line);
                     if (count($values) == 3) {
                         $values[2] = str_replace('"', '', $values[2]);
                         $file_assoc[$values[0]] = [$values[1], $values[2]];
                     }
                 }
-		foreach ($file_assoc as $id => $data) {
+                foreach ($file_assoc as $id => $data) {
                     if (!isset($cassandra_files_assoc[$id])) {
                         $missing_cassandra_entries[] = [
                             'id' => $id,
@@ -475,7 +508,7 @@ class DataConsistencyChecker
                     }
                 }
             } else {
-            
+
             }
         }
 
@@ -483,34 +516,36 @@ class DataConsistencyChecker
             $physical_file = $dir . '/physical_' . $file_num . '.csv';
 
             if (!file_exists($physical_file)) {
-		    $cassandra_file_lines = $this->getFileLines($cassandra_file);
-		// echo "cfl: " . var_dump($cassandra_file_lines) . PHP_EOL;
+                $cassandra_file_lines = $this->getFileLines($cassandra_file);
                 $file_assoc = $this->buildFileAssociation($cassandra_file_lines);
 
-		foreach ($file_assoc as $id => $data) {
-
+                foreach ($file_assoc as $id => $data) {
                     if (!isset($physical_files_assoc[$id])) {
                         $missing_physical_files[] = [
                             'id' => $id,
-                            'file1' => [$id, $data[0], $data[1], $data[2], $data[3], $data[4], $data[5], $data[6]],
+                            'file1' => [$id, $data[0], $data[1], $data[2], $data[3], $data[4], $data[5]],
                         ];
                     }
                 }
             }
         }
+
         $result_from_cassandra_entries = array_unique(array_merge($cassandra_entries, $missing_cassandra_entries), SORT_REGULAR);
-
-       // var_dump($missing_physical_files);
-	$result_from_physical_files = array_unique(array_merge($physical_entries, $missing_physical_files), SORT_REGULAR);
-
+        $result_from_physical_files = array_unique(array_merge($physical_entries, $missing_physical_files), SORT_REGULAR);
         $this->generateCsvReportForDbEntries($result_from_cassandra_entries, 'result_from_cassandra_entries.csv');
         $this->generateHtmlReport($result_from_cassandra_entries, 'cassandra.html');
         $this->generateCsvReportForPhysicalFiles($result_from_physical_files, 'result_from_physical_files.csv');
         $this->generateHtmlReport($result_from_physical_files, 'physical.html');
-        // $this->compareCSVFilesTransform($this->structured_directory . 'cassandra_all_entries.csv', $this->structured_directory . 'result_from_cassandra_entries.csv', $this->structured_directory . 'final_file.csv');
-        // $this->deletePhysicalFilesFromCsv('result_from_cassandra_entries.csv');
+
     }
-    private function buildFileAssociation($file_lines)
+    /**
+     * Builds an associative array from file lines.
+     *
+     * @param array $file_lines An array containing lines of a file.
+     * @return array An associative array representing the file association.
+     */
+
+    private function buildFileAssociation($file_lines): array
     {
         $file_assoc = [];
 
@@ -520,22 +555,28 @@ class DataConsistencyChecker
             }
 
             $values = explode(",", $line);
-            $values[2] = str_replace('"', '', $values[2]);
             if (count($values) == 3) {
 
                 $file_assoc[$values[0]] = [$values[1], $values[2]];
             }
             if (count($values) > 3) {
 
-                $file_assoc[$values[0]] = [$values[1], $values[2], $values[3], $values[4], $values[5], $values[0]];
+                $file_assoc[$values[0]] = [$values[1], $values[2], $values[3], $values[4], $values[5], $values[6]];
             }
         }
 
         return $file_assoc;
     }
 
+    /**
+     * Compares two CSV files and returns missing entries or entries with mismatched data.
+     *
+     * @param string $file1_path The file path of the first CSV file.
+     * @param string $file2_path The file path of the second CSV file.
+     * @return array An array containing missing entries or entries with mismatched data.
+     */
 
-    private function compare_csv_files($file1_path, $file2_path)
+    private function compare_csv_files($file1_path, $file2_path): array
     {
         $file1_data = array_map('str_getcsv', file($file1_path));
         $file2_data = array_map('str_getcsv', file($file2_path));
@@ -566,7 +607,7 @@ class DataConsistencyChecker
                     $row[$time_index_1],
                     isset($row[$filename_index_1]) ? $row[$filename_index_1] : null,
                     isset($row[$clientid_index_1]) ? $row[$clientid_index_1] : null,
-		    isset($row[$bucket_index_1]) ? $row[$bucket_index_1] : null,
+                    isset($row[$bucket_index_1]) ? $row[$bucket_index_1] : null,
                     isset($row[$attachment_id_index_1]) ? $row[$attachment_id_index_1] : null
                 ],
             ];
@@ -594,7 +635,7 @@ class DataConsistencyChecker
                 ];
             } else {
                 $file2_data = $file2_assoc[$id]['file2'];
-                if ($data['file1'][1] !== $file2_data[1] || $data['file1'][2] !== $file2_data[2]) {
+                if ($data['file1'][0] !== $file2_data[0] || $data['file1'][1] !== $file2_data[1]) {
                     $missing_entries[] = [
                         'id' => $id,
                         'file1' => $data['file1'],
@@ -619,19 +660,16 @@ class DataConsistencyChecker
     private function generateCsvReportForPhysicalFiles(array $inconsistentFiles, string $name): void
     {
         $fp = fopen($name, 'w');
-        fputcsv($fp, ['File/Attachment', 'File path', 'File name', 'Thumb 1', 'Thumb 2', 'Size', 'Creation Time', 'ClientId', 'Bucket' , 'Id']);
+        fputcsv($fp, ['File/Attachment', 'File path', 'File name', 'Thumb 1', 'Thumb 2', 'Size', 'Creation Time', 'ClientId', 'Bucket', 'Id']);
 
         foreach ($inconsistentFiles as $row) {
-
-            $file1Value = $row['file1'][0];
             $check_value = $row['id'];
-            //$check_value = is_numeric($row['id']) ? $row['file1'][3] : $file1Value; 
-            $filePath = $this->directory . '/' . $check_value ? $this->directory . '/' . $check_value : $this->directory . '/' . $row['file2'][0];
+            $filePath = $check_value ? $check_value : $row['file2'][0];
             if (is_numeric($row['id'])) {
-                $filePath = $this->directory . '/' . $row['file1'][3];
+                $filePath = $row['file1'][3];
                 $check_value = $row['file1'][3];
-	    }
-	
+            }
+
             $size = isset($row['file1'][1]) ? (string) $row['file1'][1] : filesize($filePath);
             $creationTime = isset($row['file1'][2]) ? (string) $row['file1'][2] : date('Y-m-d H:i:s', filectime($filePath));
             $thumb1 = isset($row['thumb1']) ? $row['thumb1'] : '';
@@ -658,7 +696,13 @@ class DataConsistencyChecker
         chmod($name, 0666);
     }
 
-
+    /**
+     * Generates a CSV report for inconsistent database entries.
+     *
+     * @param array $inconsistentFiles An array containing inconsistent file data.
+     * @param string $name The name of the CSV report file to be generated.
+     * @return void
+     */
     private function generateCsvReportForDbEntries(array $inconsistentFiles, string $name): void
     {
 
@@ -666,20 +710,24 @@ class DataConsistencyChecker
         fputcsv($fp, ['File/Attachment', 'Entry Path', 'Entry Name', 'Thumb 1', 'Thumb 2', 'Size', 'Creation Time']);
 
         foreach ($inconsistentFiles as $row) {
-
-            $filePath = $row['file1'][0] ? $row['file1'][0] : $row['file2'][0];
+            $filePath = $this->directory . '/' . $row['file1'][0] ? $this->directory . '/' . $row['file1'][0] : $this->directory . '/' . $row['file2'][0];
             $size = isset($row['file1'][1]) ? (string) $row['file1'][1] : filesize($filePath);
             $creationTime = isset($row['file1'][2]) ? (string) $row['file1'][2] : date('Y-m-d H:i:s', filectime($filePath));
-            $thumb1 = isset($row['thumb1']) ? $row['thumb1'] : '';
-	        $thumb2 = isset($row['thumb2']) ? $row['thumb2'] : '';
+            $thumb1 = $row['file1'][0] . '-thumb1';
+            $thumb2 = $row['file1'][0] . '-thumb2';
+            if (is_string($row['id']) && strpos($row['id'], '.') !== false) {
+                $old_attachment = explode('.', $row['id'])[0];
+                $thumb1 = '';
+                $thumb2 = '';
+            }
             fputcsv($fp, [
-                'Attachment',
+                'File',
                 $filePath,
                 $row['file1'][0],
                 $thumb1,
                 $thumb2,
                 $size,
-		        $creationTime
+                $creationTime
             ]);
         }
 
@@ -696,6 +744,7 @@ class DataConsistencyChecker
      */
     private function generateHtmlReport(array $inconsistentFiles, string $name): void
     {
+
         $file = fopen($name, 'w');
         if (!$file) {
             throw new Exception('Failed to open the file for writing.');
@@ -734,7 +783,16 @@ class DataConsistencyChecker
         fclose($file);
     }
 
-    function compareCSVFilesTransform($firstFile, $secondFile, $finalFile)
+    /**
+     * Compares two CSV files and creates a new CSV file containing the matching entries.
+     *
+     * @param string $firstFile The path to the first CSV file.
+     * @param string $secondFile The path to the second CSV file.
+     * @param string $finalFile The path to the final CSV file to be created.
+     * @return void
+     */
+
+    function compareCSVFilesTransform($firstFile, $secondFile, $finalFile): void
     {
         // Read the first CSV file
         $firstData = array_map('str_getcsv', file($firstFile));
@@ -790,6 +848,12 @@ class DataConsistencyChecker
         fclose($finalCsvFile);
     }
 
+    /**
+     * Deletes physical files based on the entries listed in a CSV file.
+     *
+     * @param string $csvFile The path to the CSV file containing the list of files to delete.
+     * @return void
+     */
 
     private function deletePhysicalFilesFromCsv(string $csvFile): void
     {
@@ -842,35 +906,13 @@ class DataConsistencyChecker
     }
 
 
-    private static function _is_uuid(string $id): bool
-    {
-        $regex = '/^[a-z0-9]{8}-[a-z0-9]{4}-[a-z0-9]{4}-[a-z0-9]{4}-[a-z0-9]{12}$/';
-        return preg_match($regex, $id);
-    }
-
-    /**
-     * Check if the attachment string is a valid migrated attachment
-     *
-     * @param string $attachmentId
-     *
-     * @return bool
-     */
-    private static function _isValidMigratedAttachment(string $attachmentId = ''): bool
-    {
-        return str::sub($attachmentId, 0, 2) === OLD_ATTACHMENTS_PREFIX;
-    }
-
-
-
-    /* ATTACHMENT DELETION  ===================================================================================================================== */
-
     /**
      * Returns cassandra schema version
      *
      *
      * @return int
      */
-    
+
     private function schema_version()
     {
         return static::$schemaVersion;
@@ -893,10 +935,10 @@ class DataConsistencyChecker
         ];
 
         if ($this->schema_version() == 1) {
-            $query = $this->cassandra->prepare('SELECT * FROM attachment_file_info WHERE id = ? AND client_id = ? ALLOW FILTERING');
-	} else {
-		$q = 'SELECT * FROM attachment_file_info WHERE id = ? AND client_id = ? AND bucket = \'' . $bucketId . '\'';
-		echo "Q: [" . $q . "]\n";
+            $query = $this->cassandra->prepare('SELECT * FROM attachment_file_info WHERE id = ? AND client_id = ?');
+        } else {
+            $q = 'SELECT * FROM attachment_file_info WHERE id = ? AND client_id = ? AND bucket = \'' . $bucketId . '\'';
+            echo "Q: [" . $q . "]\n";
             $query = $this->cassandra->prepare($q);
         }
         $res = $this->cassandra->execute(
@@ -906,14 +948,14 @@ class DataConsistencyChecker
             ]
         );
         if ($res && $res->valid()) {
-	  $tmp = $res->current();
+            $tmp = $res->current();
 
-          $attachment = (object) $tmp;
-          $attachment->id = (string) $tmp['id'];
-          if (array_key_exists('size', $tmp)) {
-            $attachment->size = (int) $tmp['size'];
-	  }
-	}
+            $attachment = (object) $tmp;
+            $attachment->id = (string) $tmp['id'];
+            if (array_key_exists('size', $tmp)) {
+                $attachment->size = (int) $tmp['size'];
+            }
+        }
         return $attachment;
     }
 
@@ -927,7 +969,7 @@ class DataConsistencyChecker
     {
         return substr($id, 0, static::$bucketMagic);
     }
-    
+
     /**
      * updates attachment count and size
      *
@@ -956,22 +998,18 @@ class DataConsistencyChecker
      */
     private function _get_attachment_key(string $id): ?object
     {
-	//echo "GET KEY!\n";
         $result = null;
         $query = $this->cassandra->prepare('SELECT * from attachment_ids where id = ? AND bucket = ? AND client_id = ?');
         $arguments = [
             'client_id' => static::$clientId,
             'bucket' => $this->_set_bucket($id),
             'id' => $id,
-	];
-	//echo "args: " . var_dump($arguments, true) . PHP_EOL;
-	$data = $this->cassandra->execute($query, ['arguments' => $arguments]);
-	echo "GK2!!!!";
-	//var_dump($data);
+        ];
+        $data = $this->cassandra->execute($query, ['arguments' => $arguments]);
         if ($data && $data->valid()) {
             $result = (object) $data->current();
-	}
-	//var_dump($result);
+        }
+        ;
         return $result;
     }
 
@@ -986,12 +1024,12 @@ class DataConsistencyChecker
     {
         $attachment = null;
         try {
-		$properties = '*';
-		echo "attid: ". $attachmentId . PHP_EOL;
-	    $key = $this->_get_attachment_key($attachmentId);
+            $properties = '*';
+            echo "attid: " . $attachmentId . PHP_EOL;
+            $key = $this->_get_attachment_key($attachmentId);
             if ($key) {
                 if ($this->schema_version() === 1) {
-                    $query = $this->cassandra->prepare('SELECT ' . $properties . ' FROM attachments WHERE id = ? AND client_id = ? AND project_id = ? AND entity_type = ? ALLOW FILTERING');
+                    $query = $this->cassandra->prepare('SELECT ' . $properties . ' FROM attachments WHERE id = ? AND client_id = ? AND project_id = ? AND entity_type = ?');
                     $arguments = [
                         'client_id' => static::$clientId,
                         'id' => $key->id,
@@ -1011,9 +1049,9 @@ class DataConsistencyChecker
                 $attachment = $this->cassandra->execute($query, ['arguments' => $arguments]);
             }
         } catch (Cassandra\Exception\InvalidArgumentException $e) {
-	}
-	//echo "ATTA: \n";
-	//var_dump($attachment);
+        }
+        //echo "ATTA: \n";
+        //var_dump($attachment);
 
         return $attachment != null && $attachment->valid() ? $this->_convert_to_object($attachment->current()) : null;
     }
@@ -1079,28 +1117,27 @@ class DataConsistencyChecker
      */
     public function deleteAttachment(int $clientId, ?string $bucketId, string $id): bool
     {
-	static::$clientId = $clientId;
+        static::$clientId = $clientId;
         $refData = [
             'client_id' => $clientId,
             'id' => $id,
         ];
-	$fileInfo = $this->get_info($clientId, $bucketId, $id);
-	//var_dump($fileInfo);
+        $fileInfo = $this->get_info($clientId, $bucketId, $id);
         if ($fileInfo) {
             $this->_update_attachment_stats(false, $fileInfo->size);
-	} else {
-	  return false;
-	}
-	
-	$q = 'DELETE FROM attachment_file_info WHERE id = ? AND client_id = ? ';
-	if ($this->schema_version() === 2) {
-		$q = $q . ' AND bucket = \'' . $bucketId . '\'';
-	}
-        $query = $this->cassandra->prepare($q);
-	$this->cassandra->execute($query, ['arguments' => $refData]);
-	
+        } else {
+            return false;
+        }
 
-	echo "DEL REFS\n";
+        $q = 'DELETE FROM attachment_file_info WHERE id = ? AND client_id = ? ';
+        if ($this->schema_version() === 2) {
+            $q = $q . ' AND bucket = \'' . $bucketId . '\'';
+        }
+        $query = $this->cassandra->prepare($q);
+        $this->cassandra->execute($query, ['arguments' => $refData]);
+
+
+        echo "DEL REFS\n";
         $refData['bucket'] = $this->_set_bucket($id);
         $query = $this->cassandra->prepare(
             'DELETE FROM attachment_file_refs WHERE bucket = ? AND id = ? AND client_id = ?'
@@ -1113,7 +1150,7 @@ class DataConsistencyChecker
         );
         $result = $this->cassandra->execute($query, ['arguments' => $refData]);
 
-	echo "DEL ATTACHMENTS\n";
+        echo "DEL ATTACHMENTS\n";
         if ($this->schema_version() === 1) {
             $delQuery = $this->cassandra->prepare(
                 'DELETE FROM attachments WHERE id = ? AND client_id = ? AND project_id = ? AND entity_type = ?'
@@ -1122,15 +1159,15 @@ class DataConsistencyChecker
             $delQuery = $this->cassandra->prepare(
                 'DELETE FROM attachments WHERE id = ? AND client_id = ? AND project_id = ? AND entity_type = ? AND entity_id = ?'
             );
-	}
+        }
         while ($result && $result->valid()) {
             $attachmentId = $result->current()['attachment_id'];
-	    $key = $this->_get_attachment_key((string) $attachmentId);
-	    if ($key == null)  {
-            	$result->next();
-		continue;
-	    }
-            if ($this->schema_version()  === 1) {
+            $key = $this->_get_attachment_key((string) $attachmentId);
+            if ($key == null) {
+                $result->next();
+                continue;
+            }
+            if ($this->schema_version() === 1) {
                 $selectQuery = $this->cassandra->prepare(
                     'SELECT entity_id FROM attachments WHERE id = ? AND client_id = ? AND project_id = ? AND entity_type = ?'
                 );
@@ -1166,40 +1203,40 @@ class DataConsistencyChecker
         $query = $this->cassandra->prepare(
             'DELETE FROM attachment_file_ids WHERE bucket = ? AND id = ? AND client_id = ?'
         );
-	$this->cassandra->execute($query, ['arguments' => $refData]);
-	$result = false;
-	$attachment = $this->_get_attachment_by_id($id);
-	if ($attachment) {
-		if ($this->schema_version() === 1) {
-		    $query = $this->cassandra->prepare("DELETE FROM attachments WHERE id = ? AND project_id = ? AND entity_type = ? AND client_id = ?");
-		    $arguments = [
-			'arguments' => [
-			    'client_id' => static::$clientId,
-			    'id' => $id,
-			    'project_id' => $attachment->project_id,
-			    'entity_type' => $attachment->entity_type
-			],
-		    ];
-		} else {
-		    $query = $this->cassandra->prepare("DELETE FROM attachments WHERE id = ? AND project_id = ? AND entity_id = ? AND entity_type = ? AND client_id = ?");
-		    $arguments = [
-			'arguments' => [
-			    'client_id' => static::$clientId,
-			    'id' => $id,
-			    'project_id' => $attachment->project_id,
-			    'entity_type' => $attachment->entity_type,
-			    'entity_id' => $attachment->entity_id
-			],
-		    ];
-		}
-		$queryResult = $this->cassandra->execute($query, $arguments) != null;
+        $this->cassandra->execute($query, ['arguments' => $refData]);
+        $result = false;
+        $attachment = $this->_get_attachment_by_id($id);
+        if ($attachment) {
+            if ($this->schema_version() === 1) {
+                $query = $this->cassandra->prepare("DELETE FROM attachments WHERE id = ? AND project_id = ? AND entity_type = ? AND client_id = ?");
+                $arguments = [
+                    'arguments' => [
+                        'client_id' => static::$clientId,
+                        'id' => $id,
+                        'project_id' => $attachment->project_id,
+                        'entity_type' => $attachment->entity_type
+                    ],
+                ];
+            } else {
+                $query = $this->cassandra->prepare("DELETE FROM attachments WHERE id = ? AND project_id = ? AND entity_id = ? AND entity_type = ? AND client_id = ?");
+                $arguments = [
+                    'arguments' => [
+                        'client_id' => static::$clientId,
+                        'id' => $id,
+                        'project_id' => $attachment->project_id,
+                        'entity_type' => $attachment->entity_type,
+                        'entity_id' => $attachment->entity_id
+                    ],
+                ];
+            }
+            $queryResult = $this->cassandra->execute($query, $arguments) != null;
 
-		if ($queryResult) {
-		    $result = true;
-		    $this->_update_file_refs($attachment->data_id, false, $attachment->id);
-		    $this->_delete_attachment_key($id);
-		}
-	}
+            if ($queryResult) {
+                $result = true;
+                $this->_update_file_refs($attachment->data_id, false, $attachment->id);
+                $this->_delete_attachment_key($id);
+            }
+        }
 
         return $result;
     }
@@ -1221,7 +1258,7 @@ class DataConsistencyChecker
      * file,/tmp/testx3f,testx3f,testx3f-thumb1,testx3f-thumb2,4343,20-02-22 13:30,,,
      * cassandra,/test1/testx3,testx3,testx3-thumb1,testx3-thumb2,4343,20-02-22 13:30,1,3,3abc-def
      */
-    public function processAttachmentDeletionCSV(string $file, string $src) : void
+    public function processAttachmentDeletionCSV(string $file, string $src): void
     {
 
         echo "Before proceeding with the deletion, make sure you have a backup of your data." . PHP_EOL;
@@ -1236,7 +1273,6 @@ class DataConsistencyChecker
         $file_lines = explode("\n", $file_contents);
         $logFile = 'deleted_files.log';
         $logHandle = fopen($logFile, 'a');
-        $directory = $this->directory . '/';
 
         foreach ($file_lines as $line) {
             if ($line === reset($file_lines)) {
@@ -1245,40 +1281,46 @@ class DataConsistencyChecker
             $values = explode(",", $line);
 
             if ($values[0] === 'File/Attachment') {
-                    continue;
+                continue;
             }
             if (count($values) >= 7) {
                 $data = (object) array();
-                $data->source  = $values[0];
+                $data->source = $values[0];
                 $data->path = $values[1];
                 $data->name = $values[2];
                 $data->thumb1 = $values[3];
                 $data->thumb2 = $values[4];
                 $data->size = $values[5];
                 $data->created = $values[6];
-                if(count($values) >= 10) {
+                $path = dirname($data->path);
+                if (count($values) >= 10) {
                     $data->clientId = (int) $values[7];
                     $data->bucket = $values[8];
                     $data->id = $values[9];
                 }
-                if ($data->source == $src) {
-                    if ($src === 'cassandra') {
-                        echo "will delete " . $data->clientId . " : " . $data->bucket . " : "  . $data->id . PHP_EOL;
-                        fwrite($logHandle, "Deleted attachment: $data->id" . PHP_EOL);
-                    } else {
-                        $filePath = $directory . $values[1];
-                        $thumb1Path = $directory . $values[1];
-                        $thumb2Path = $directory . $values[1];
-                        if (file_exists($filePath)) {
-                            //   unlink($filePath);
-                            //   unlink($thumb1Path);
-                            //   unlink($thumb2Path);
-                            echo "File deleted: $filePath" . PHP_EOL;
+                if ($data->source === 'Attachment' && $src === 'cassandra') {
+                    echo "will delete " . $data->clientId . " : " . $data->bucket . " : " . $data->id . PHP_EOL;
+                    $this->deleteAttachment($data->clientId, $data->bucket, $data->id);
+                    fwrite($logHandle, "Deleted attachment: $data->id" . PHP_EOL);
+                } else if ($data->source === 'File' && $src === 'file') {
+                    $filePath = $values[1];
+                    $thumb1Path = $path . '/' . $values[3];
+                    $thumb2Path = $path . '/' . $values[4];
+                    if (file_exists($thumb1Path) && file_exists($thumb2Path)) {
+                        files::delete($thumb1Path);
+                        files::delete($thumb2Path);
+                        fwrite($logHandle, "Deleted thumbnail: " . $thumb1Path  . PHP_EOL);
+                        fwrite($logHandle, "Deleted thumbnail: " . $thumb2Path  . PHP_EOL);
+                    }
+                    if (file_exists($filePath)) {
+                        unlink($filePath);
+                        unlink($thumb1Path);
+                        unlink($thumb2Path);
+                        echo "File deleted: $filePath" . PHP_EOL;
                         // Write the deleted file path to the log file
-                            fwrite($logHandle, "Deleted file: $filePath" . PHP_EOL);
-                        } else {
-                            echo "File not found: $filePath" . PHP_EOL;
-                        }
+                        fwrite($logHandle, "Deleted file: $filePath" . PHP_EOL);
+                    } else {
+                        echo "File not found: $filePath" . PHP_EOL;
                     }
                 }
             }
@@ -1290,10 +1332,5 @@ class DataConsistencyChecker
     }
 }
 
-#$options = getopt('', ['directory:']);
 $checker = new DataConsistencyChecker();
-$checker->checkConsistency("attachment_file_info", true);
-#$checker::$cassandraHost = 'localhost';
-#$checker->init($options);
-#$checker->deleteAttachment("1", "f", "ff29ead0-8696-4ef1-8120-538d6dd7efd1");
-#$checker->processAttachmentDeletionCSV("todelete.csv", "file");
+$checker->checkConsistency("attachment_file_info", true);
\ No newline at end of file
diff --git a/config.ini b/config.ini
new file mode 100644
index 0000000..c3f6a72
--- /dev/null
+++ b/config.ini
@@ -0,0 +1,15 @@
+[SQL]
+driver = mysql
+host = 127.0.0.1
+databaseName = testrail
+user = testrail
+password = 123456789
+port = 6666
+
+[CASSANDRA]
+host = cassandra
+databaseName = testrail
+user = casandra
+password = cassandra
+port = 9042
+keyspace = testrail

From 4f7d180d2c20a704cdaabc68c87c6a3cadff368d Mon Sep 17 00:00:00 2001
From: Spase <spase.papcev@gmail.com>
Date: Mon, 29 May 2023 08:23:39 +0200
Subject: [PATCH 3/4] Changes from feedback

---
 cassandra_consistency_script.php | 579 ++++++++++++++-----------------
 report_template.html             |  37 ++
 2 files changed, 303 insertions(+), 313 deletions(-)
 create mode 100644 report_template.html

diff --git a/cassandra_consistency_script.php b/cassandra_consistency_script.php
index 5a6522d..24fa4e0 100644
--- a/cassandra_consistency_script.php
+++ b/cassandra_consistency_script.php
@@ -1,19 +1,41 @@
 <?php
 
+/**
+ * Class DataConsistencyChecker
+ *
+ * The DataConsistencyChecker class is responsible for checking the consistency of data.
+ */
+
 class DataConsistencyChecker
 {
-    const DEFAULT_PAGE_SIZE = 30;
     private $_cluster;
     private $session;
     private $cassandra;
     private $directory;
     private $structured_directory;
     private $retrived_csv;
-    static $clientId;
+    private static $clientId;
     private static $schemaVersion = 1;
+    private static $bucketMagic = 4;
 
-    static $bucketMagic = 4;
-    static $cassandraHost = 'cassandra';
+    private const DEFAULT_PAGE_SIZE = 30;
+    private const CASSANDRA_RESULT_CSV = 'result_from_cassandra_entries.csv';
+    private const CASSANDRA_HTML_REPORT = 'cassandra.html';
+    private const PHYSICAL_RESULT_CSV = 'result_from_physical_files.csv';
+    private const PHYSICAL_HTML_REPORT = 'physical.html';
+    private const CSV_COLUMN_FILE_ATTACHMENT = 'File/Attachment';
+    private const CSV_COLUMN_FILE_PATH = 'File path';
+    private const CSV_COLUMN_FILE_NAME = 'File name';
+    private const CSV_COLUMN_THUMB1 = 'Thumb 1';
+    private const CSV_COLUMN_THUMB2 = 'Thumb 2';
+    private const CSV_COLUMN_SIZE = 'Size';
+    private const CSV_COLUMN_CREATION_TIME = 'Creation Time';
+    private const CSV_COLUMN_CLIENT_ID = 'ClientId';
+    private const CSV_COLUMN_BUCKET = 'Bucket';
+    private const CSV_COLUMN_ID = 'Id';
+    private const CSV_EXTENSION = '.csv';
+    private const NUM_FIELDS = 3;
+    private const DELIMITER = '-';
 
     public function __construct($directory = null)
     {
@@ -29,26 +51,26 @@ class DataConsistencyChecker
      *
      * @return void
      */
+    
     public function init(): void
     {
-        $config = parse_ini_file("config.ini", true);
+        $config = parse_ini_file('config.ini', true);
+        $cassandraConfig = $config['CASSANDRA'];
         $this->_cluster = Cassandra::cluster()
-            ->withContactPoints($config['CASSANDRA']['host'])
-            ->withPort(9042)
+            ->withContactPoints($cassandraConfig['host'])
+            ->withPort($cassandraConfig['port'])
             ->withCredentials(
-                $config['CASSANDRA']['user'],
-                $config['CASSANDRA']['password']
+                $cassandraConfig['user'],
+                $cassandraConfig['password']
             )
             ->build();
         if ($this->_cluster) {
             try {
-                $this->cassandra = $this->_cluster->connect($config['CASSANDRA']['keyspace']);
+                $this->cassandra = $this->_cluster->connect($cassandraConfig['keyspace']);
             } catch (Exception $e) {
-                echo "err\n";
+                echo "An error occurred: " . $e->getMessage() . "\n";
             }
         }
-
-
     }
 
     /**
@@ -57,61 +79,72 @@ class DataConsistencyChecker
      * @param array $arguments The command line arguments.
      * @return void
      */
-    public function runFromCommandLine($arguments)
-    {
-        $shortOptions = "hd:v:o:r:s:";
-        $longOptions = ["help", "directory:", "version:", "v", "output:", "o", "remove:", "r", "source:", "s"];
+    
+     public function runFromCommandLine(array $arguments): void
+     {
+         $shortOptions = "hd:v:o:r:s:";
+         $longOptions = ["help", "directory:", "version:", "v", "output:", "o", "remove:", "r", "source:", "s"];
+     
+         $options = getopt($shortOptions, $longOptions);
+     
+         if (count($options) === 0 || isset($options['h']) || isset($options['help'])) {
+             $this->displayHelpMessage();
+             exit;
+         }
+     
+         $directory = $options['directory'] ?? $options['d'] ?? null;
+         $schemaVersion = $options['version'] ?? $options['v'] ?? null;
+         $source = $options['source'] ?? $options['s'] ?? null;
+         $remove = $options['remove'] ?? $options['r'] ?? null;
+         $structured_directory = $options['output'] ?? $options['o'] ?? null;
+     
+         if (!file_exists($structured_directory)) {
+             if ($structured_directory === null) {
+                 echo "Output directory is required. Please specify the --output option.\n";
+                 exit;
+             }
+             mkdir($structured_directory, 0777, true);
+         }
+     
+         $this->structured_directory = $structured_directory;
+     
+         if (($directory === null || $schemaVersion === null) && $remove === null && $source === null) {
+             echo "Missing Attachment directory or schema version.\n";
+             exit;
+         }
+     
+         if ($schemaVersion && !in_array($schemaVersion, [1, 2])) {
+             echo "Invalid schema version. Only versions 1 and 2 are supported.\n";
+             exit;
+         }
 
-        $options = getopt($shortOptions, $longOptions);
-
-        if (count($options) == 0 || isset($options['h']) || isset($options['help'])) {
-            $this->displayHelpMessage();
+         if ($structured_directory === null) {
+            echo "Output directory is required. Please specify the --output option.\n";
             exit;
-        }
-
-        $directory = isset($options['directory']) ? $options['directory'] : (isset($options['d']) ? $options['d'] : null);
-        $schemaVersion = isset($options['version']) ? $options['version'] : (isset($options['v']) ? $options['v'] : null);
-
-        $source = isset($options['source']) ? $options['source'] : (isset($options['s']) ? $options['s'] : null);
-        $remove = isset($options['remove']) ? $options['remove'] : (isset($options['r']) ? $options['r'] : null);
-
-        $structured_directory = isset($options['output']) ? $options['output'] : (isset($options['o']) ? $options['o'] : null);
-        if (!file_exists($structured_directory)) {
-            mkdir($structured_directory, 0777, true);
-        }
-        $this->structured_directory = $structured_directory;
-
-        if (($directory === null || $schemaVersion === null) && $remove === null && $source === null) {
-            echo "Missing Attachment directory or schema version.\n";
-            exit;
-        }
-        if ($schemaVersion && !in_array($schemaVersion, [1, 2])) {
-            echo "Invalid schema version. Only versions 1 and 2 are supported.\n";
-            exit;
-        }
-        if ($this->structured_directory == null) {
-            $this->structured_directory = './';
-        }
-
-        static::$schemaVersion = (int) $schemaVersion;
-        $this->directory = $directory;
-        $this->retrived_csv = './result_from_physical_files.csv';
-        if ($remove && $source) {
-            $this->processAttachmentDeletionCSV($remove, $source);
-        } else {
-            $this->checkConsistency('attachment_file_info');
-            if (is_dir($this->structured_directory)) {
-                $this->removeDirectory($this->structured_directory);
-            }
-        }
-        exit;
-    }
+         }
+     
+         static::$schemaVersion = (int) $schemaVersion;
+         $this->directory = $directory;
+         $this->retrived_csv = './result_from_physical_files.csv';
+     
+         if ($remove && $source) {
+             $this->processAttachmentDeletionCSV($remove, $source);
+         } else {
+             $this->checkConsistency('attachment_file_info');
+             if (is_dir($this->structured_directory)) {
+                 $this->removeDirectory($this->structured_directory);
+             }
+         }
+         exit;
+     }
+     
 
     /**
      * Displays the help message with instructions on how to use the script.
      *
      * @return void
      */
+
     private function displayHelpMessage(): void
     {
         $helpMessage = <<<EOT
@@ -140,13 +173,17 @@ class DataConsistencyChecker
      * @param string $directory The directory path to be removed.
      * @return void
      */
-    private function removeDirectory($directory): void
+    
+    private function removeDirectory(string $directory): void
     {
         if (!is_dir($directory)) {
             return;
         }
 
-        $files = array_diff(scandir($directory), ['.', '..']);
+        $files = array_diff(
+            scandir($directory),
+            ['.', '..']
+        );
 
         foreach ($files as $file) {
             $path = $directory . '/' . $file;
@@ -157,7 +194,6 @@ class DataConsistencyChecker
                 unlink($path);
             }
         }
-
         rmdir($directory);
     }
 
@@ -167,7 +203,8 @@ class DataConsistencyChecker
      * @param string $tableName The name of the table in the database to check consistency for.
      * @return void
      */
-    public function checkConsistency($tableName)
+
+    public function checkConsistency(string $tableName): void
     {
         $this->init();
         $dbEntries = $this->getDbEntries($tableName);
@@ -182,7 +219,7 @@ class DataConsistencyChecker
      * @return array An array containing the file entries organized by dynamic values.
      */
 
-    private function getFileEntries($directory): array
+    private function getFileEntries(string $directory): array
     {
         $files = glob($directory . '/*');
         $entries = [];
@@ -190,19 +227,20 @@ class DataConsistencyChecker
         foreach ($files as $file) {
             if (is_file($file)) {
                 $fileName = basename($file);
+
                 if (strpos($fileName, '-thumb1') !== false || strpos($fileName, '-thumb2') !== false) {
                     continue;
                 }
-                $dashParts = explode("-", $fileName, 2);
-                $dotParts = explode(".", $fileName);
+                $dashParts = explode(self::DELIMITER, $fileName, 2);
+                $dotParts = explode('.', $fileName);
                 if (count($dashParts) === 2) {
                     $clientId = $dashParts[0];
                     $id = $dashParts[1];
                     $fileParts = $dashParts;
-                    $delimiter = '-';
+                    $delimiter = self::DELIMITER;
                 } elseif (count($dotParts) >= 2) {
                     $clientId = $dotParts[0];
-                    $id = implode(".", array_slice($dotParts, 1));
+                    $id = implode('.', array_slice($dotParts, 1));
                     $fileParts = $dotParts;
                     $delimiter = '.';
                 } else {
@@ -228,7 +266,6 @@ class DataConsistencyChecker
         foreach ($entries as $clientId => $clientEntries) {
             $this->createPhysicalFileCSV($clientId, $clientEntries);
         }
-
         return $entries;
     }
 
@@ -241,9 +278,9 @@ class DataConsistencyChecker
      * @return void
      */
 
-    private function createPhysicalFileCSV($clientId, $entries)
+    private function createPhysicalFileCSV(string $clientId, array $entries): void
     {
-        $fileName = $this->structured_directory . "physical_" . $clientId . ".csv";
+        $fileName = $this->structured_directory . 'physical_' . $clientId . self::CSV_EXTENSION;
         $csvFile = fopen($fileName, 'w');
         fputcsv($csvFile, ['id', 'size', 'creation_time']);
         foreach ($entries as $entry) {
@@ -265,9 +302,9 @@ class DataConsistencyChecker
      * @return void
      */
 
-    private function createDBFileCSV($clientId, $entries): void
+    private function createDBFileCSV(string $clientId, array $entries): void
     {
-        $fileName = $this->structured_directory . "cassandra_" . (string) $clientId . ".csv";
+        $fileName = $this->structured_directory . 'cassandra_' . (string) $clientId . self::CSV_EXTENSION;
         $csvFile = fopen($fileName, 'w');
 
         $headers = ['id', 'size', 'creation_time', 'filename', 'bucket', 'client_id', 'attachment_id'];
@@ -283,10 +320,8 @@ class DataConsistencyChecker
                 $entry['client_id'],
                 $entry['attachment_id'],
             ];
-
             fputcsv($csvFile, $rowData);
         }
-
         fclose($csvFile);
     }
 
@@ -297,15 +332,13 @@ class DataConsistencyChecker
      * @return array An array containing the retrieved entries.
      */
 
-
-    private function getDbEntries($tableName)
+    private function getDbEntries(string $tableName): array
     {
-
-        if ($this->schema_version() === 1) {
-            $query = "SELECT client_id, id, size, filename, created_on FROM $tableName";
-        } else {
-            $query = "SELECT client_id, id, size, filename, created_on, bucket FROM $tableName";
-        }
+        $query = sprintf(
+            'SELECT client_id, id, size, filename, created_on%s FROM %s',
+            $this->schema_version() === 1 ? '' : ', bucket',
+            $tableName
+        );
         $arguments = [];
         $result = $this->cassandra->execute(
             $query,
@@ -317,7 +350,6 @@ class DataConsistencyChecker
         $entries = [];
         while ($result) {
             foreach ($result as $row) {
-
                 if (preg_match('/^[0-9]+$/', $row['id'])) {
                     $dotParts = explode(".", $row['filename'], 2);
                     $dynamicValue = substr($dotParts[1], 0, 2);
@@ -336,7 +368,7 @@ class DataConsistencyChecker
                     $entries[$dynamicValue][$row['filename']] = $entry;
                 } else {
                     $clientId = substr($row['id'], 0, 2);
-                    $file = $row['client_id'] . '-' . $row['id'];
+                    $file = $row['client_id'] . self::DELIMITER . $row['id'];
                     $timestamp = (int) $row['created_on'];
                     $date = date('Y-m-d H:i:s', $timestamp);
                     $creationTime = str_replace('"', '', $date);
@@ -357,7 +389,6 @@ class DataConsistencyChecker
             }
             $result = $result->nextPage();
         }
-
         // Merge all entries into a single array
         $allEntries = [];
         foreach ($entries as $clientEntries) {
@@ -381,7 +412,7 @@ class DataConsistencyChecker
      * @return array|null An array containing mismatched entries, or null if the entries match.
      */
 
-    private function compareFileEntries($id, $data, $cassandra_file_assoc)
+    private function compareFileEntries(string $id, array $data, array $cassandra_file_assoc)
     {
         if (!isset($cassandra_file_assoc[$id])) {
             return [
@@ -412,7 +443,7 @@ class DataConsistencyChecker
      * @return array An array containing the lines of the file.
      */
 
-    private function getFileLines($file): array
+    private function getFileLines(string $file): array
     {
         $file_contents = file_get_contents($file);
         return explode("\n", $file_contents);
@@ -425,7 +456,7 @@ class DataConsistencyChecker
      * @return array An array of filtered and mapped entries.
      */
 
-    private function filterAndMapEntries($entries): array
+    private function filterAndMapEntries(array $entries): array
     {
         $filtered_entries = array_filter($entries);
         $mapped_entries = array_map(function ($entry) {
@@ -440,18 +471,27 @@ class DataConsistencyChecker
      * @param string $dir The directory path containing the files to be processed.
      * @return void
      */
-    private function process_files_in_directory($dir): void
+
+    private function process_files_in_directory(string $dir): void
     {
         $files = glob($dir . '/*.csv');
         $physical_files = array();
         $cassandra_files = array();
         foreach ($files as $file) {
-            $filename = basename($file, '.csv');
-            $csv_type = substr($filename, 0, strpos($filename, '_'));
-            $file_num = substr($filename, strpos($filename, '_') + 1);
-            if ($csv_type == 'physical') {
+            $filename = basename($file, static::CSV_EXTENSION);
+            $csv_type = substr(
+                $filename,
+                0,
+                strpos($filename, '_')
+            );
+            $file_num = substr(
+                $filename,
+                strpos($filename, '_') + 1
+            );
+
+            if ($csv_type === 'physical') {
                 $physical_files[$file_num] = $file;
-            } elseif ($csv_type == 'cassandra') {
+            } elseif ($csv_type === 'cassandra') {
                 $cassandra_files[$file_num] = $file;
             }
         }
@@ -465,7 +505,7 @@ class DataConsistencyChecker
         $physical_entries = [];
 
         foreach ($physical_files as $file_num => $physical_file) {
-            $cassandra_file = $dir . '/cassandra_' . $file_num . '.csv';
+            $cassandra_file = $dir . '/cassandra_' . $file_num . self::CSV_EXTENSION;
 
             if (file_exists($physical_file) && file_exists($cassandra_file)) {
                 $compared_physical[] = $this->compare_csv_files($physical_file, $cassandra_file);
@@ -476,9 +516,6 @@ class DataConsistencyChecker
             } else {
                 if (!file_exists($physical_file)) {
                     $missing_physical_files[] = $physical_file;
-
-                    echo "cfl: " . var_dump($missing_physical_files) . PHP_EOL;
-
                 }
                 if (!file_exists($cassandra_file)) {
                     $missing_cassandra_files[] = $cassandra_file;
@@ -493,8 +530,8 @@ class DataConsistencyChecker
                     if ($line === reset($physical_file_lines)) {
                         continue;
                     }
-                    $values = explode(",", $line);
-                    if (count($values) == 3) {
+                    $values = explode(',', $line);
+                    if (count($values) === self::NUM_FIELDS) {
                         $values[2] = str_replace('"', '', $values[2]);
                         $file_assoc[$values[0]] = [$values[1], $values[2]];
                     }
@@ -513,7 +550,7 @@ class DataConsistencyChecker
         }
 
         foreach ($cassandra_files as $file_num => $cassandra_file) {
-            $physical_file = $dir . '/physical_' . $file_num . '.csv';
+            $physical_file = $dir . '/physical_' . $file_num . self::CSV_EXTENSION;
 
             if (!file_exists($physical_file)) {
                 $cassandra_file_lines = $this->getFileLines($cassandra_file);
@@ -532,12 +569,14 @@ class DataConsistencyChecker
 
         $result_from_cassandra_entries = array_unique(array_merge($cassandra_entries, $missing_cassandra_entries), SORT_REGULAR);
         $result_from_physical_files = array_unique(array_merge($physical_entries, $missing_physical_files), SORT_REGULAR);
-        $this->generateCsvReportForDbEntries($result_from_cassandra_entries, 'result_from_cassandra_entries.csv');
-        $this->generateHtmlReport($result_from_cassandra_entries, 'cassandra.html');
-        $this->generateCsvReportForPhysicalFiles($result_from_physical_files, 'result_from_physical_files.csv');
-        $this->generateHtmlReport($result_from_physical_files, 'physical.html');
+
+        $this->generateCsvReportForDbEntries($result_from_cassandra_entries, self::CASSANDRA_RESULT_CSV);
+        $this->generateHtmlReport($result_from_cassandra_entries, self::CASSANDRA_HTML_REPORT);
+        $this->generateCsvReportForPhysicalFiles($result_from_physical_files, self::PHYSICAL_RESULT_CSV);
+        $this->generateHtmlReport($result_from_physical_files, self::PHYSICAL_HTML_REPORT);
 
     }
+
     /**
      * Builds an associative array from file lines.
      *
@@ -545,26 +584,23 @@ class DataConsistencyChecker
      * @return array An associative array representing the file association.
      */
 
-    private function buildFileAssociation($file_lines): array
+    private function buildFileAssociation(array $file_lines): array
     {
         $file_assoc = [];
-
         foreach ($file_lines as $line) {
             if ($line === reset($file_lines)) {
                 continue;
             }
 
-            $values = explode(",", $line);
-            if (count($values) == 3) {
-
+            $values = explode(',', $line);
+            if (count($values) === self::NUM_FIELDS) {
                 $file_assoc[$values[0]] = [$values[1], $values[2]];
             }
-            if (count($values) > 3) {
 
+            if (count($values) > 3) {
                 $file_assoc[$values[0]] = [$values[1], $values[2], $values[3], $values[4], $values[5], $values[6]];
             }
         }
-
         return $file_assoc;
     }
 
@@ -576,7 +612,7 @@ class DataConsistencyChecker
      * @return array An array containing missing entries or entries with mismatched data.
      */
 
-    private function compare_csv_files($file1_path, $file2_path): array
+    private function compare_csv_files(string $file1_path, string $file2_path): array
     {
         $file1_data = array_map('str_getcsv', file($file1_path));
         $file2_data = array_map('str_getcsv', file($file2_path));
@@ -597,8 +633,16 @@ class DataConsistencyChecker
         $clientid_index_1 = array_search('client_id', $file1_headers);
         $bucket_index_1 = array_search('bucket', $file1_headers);
         $attachment_id_index_1 = array_search('attachment_id', $file1_headers);
-
-        $file1_assoc = array_reduce($file1_data, function ($result, $row) use ($id_index_1, $size_index_1, $time_index_1, $filename_index_1, $clientid_index_1, $bucket_index_1, $attachment_id_index_1) {
+        $file1_assoc = array_reduce(
+            $file1_data, function ($result, $row) use (
+                $id_index_1, 
+                $size_index_1, 
+                $time_index_1, 
+                $filename_index_1, 
+                $clientid_index_1, 
+                $bucket_index_1, 
+                $attachment_id_index_1
+            ) {
             $result[$row[$id_index_1]] = [
                 'id' => $row[$id_index_1],
                 'file1' => [
@@ -613,7 +657,13 @@ class DataConsistencyChecker
             ];
             return $result;
         }, []);
-        $file2_assoc = array_reduce($file2_data, function ($result, $row) use ($id_index_2, $size_index_2, $time_index_2, $filename_index_2) {
+        $file2_assoc = array_reduce(
+            $file2_data, function ($result, $row) use (
+                $id_index_2, 
+                $size_index_2, 
+                $time_index_2, 
+                $filename_index_2
+            ) {
             $result[$row[$id_index_2]] = [
                 'id' => $row[$id_index_2],
                 'file2' => [
@@ -648,19 +698,28 @@ class DataConsistencyChecker
         return $missing_entries;
     }
 
-
-
-
     /**
      * Generates a CSV report of inconsistent files.
      *
      * @param array $inconsistentFiles An array of inconsistent files.
      * @param string $filename The filename to use for the report.
      */
+
     private function generateCsvReportForPhysicalFiles(array $inconsistentFiles, string $name): void
     {
         $fp = fopen($name, 'w');
-        fputcsv($fp, ['File/Attachment', 'File path', 'File name', 'Thumb 1', 'Thumb 2', 'Size', 'Creation Time', 'ClientId', 'Bucket', 'Id']);
+        fputcsv($fp, [
+            self::CSV_COLUMN_FILE_ATTACHMENT, 
+            self::CSV_COLUMN_FILE_PATH, 
+            self::CSV_COLUMN_FILE_NAME, 
+            self::CSV_COLUMN_THUMB1,  
+            self::CSV_COLUMN_THUMB2, 
+            self::CSV_COLUMN_SIZE, 
+            self::CSV_COLUMN_CREATION_TIME, 
+            self::CSV_COLUMN_CLIENT_ID, 
+            self::CSV_COLUMN_BUCKET, 
+            self::CSV_COLUMN_ID
+        ]);
 
         foreach ($inconsistentFiles as $row) {
             $check_value = $row['id'];
@@ -691,7 +750,6 @@ class DataConsistencyChecker
                 $attachmentId,
             ]);
         }
-
         fclose($fp);
         chmod($name, 0666);
     }
@@ -703,11 +761,19 @@ class DataConsistencyChecker
      * @param string $name The name of the CSV report file to be generated.
      * @return void
      */
+
     private function generateCsvReportForDbEntries(array $inconsistentFiles, string $name): void
     {
-
         $fp = fopen($name, 'w');
-        fputcsv($fp, ['File/Attachment', 'Entry Path', 'Entry Name', 'Thumb 1', 'Thumb 2', 'Size', 'Creation Time']);
+        fputcsv($fp, [
+            self::CSV_COLUMN_FILE_ATTACHMENT, 
+            'Entry Path', 
+            'Entry Name', 
+            self::CSV_COLUMN_THUMB1, 
+            self::CSV_COLUMN_THUMB2, 
+            self::CSV_COLUMN_SIZE, 
+            self::CSV_COLUMN_CREATION_TIME
+        ]);
 
         foreach ($inconsistentFiles as $row) {
             $filePath = $this->directory . '/' . $row['file1'][0] ? $this->directory . '/' . $row['file1'][0] : $this->directory . '/' . $row['file2'][0];
@@ -742,169 +808,51 @@ class DataConsistencyChecker
      * @param array $inconsistentFiles An array of inconsistent files.
      * @param string $filename The filename to use for the report.
      */
-    private function generateHtmlReport(array $inconsistentFiles, string $name): void
-    {
 
-        $file = fopen($name, 'w');
-        if (!$file) {
-            throw new Exception('Failed to open the file for writing.');
-        }
-        $html = '<html><head><style>
-                    table {
-                        font-family: arial, sans-serif;
-                        border-collapse: collapse;
-                        width: 100%;
-                    }
-                    td, th {
-                        border: 1px solid #dddddd;
-                        text-align: left;
-                        padding: 8px;
-                    }
-                    tr:nth-child(even) {
-                        background-color: #dddddd;
-                    }
-                </style></head><body><table><thead><tr><th>File/Attachment</th><th>File path</th><th>File name</th><th>Thumb 1</th><th>Thumb 2</th><th>Size</th><th>Creation Time</th></tr></thead><tbody>';
-
-        foreach ($inconsistentFiles as $row) {
-
-            $filePath = $this->directory . '/' . $row['file1'][0];
-            $size = isset($row['file1'][1]) ? (string) $row['file1'][1] : filesize($filePath);
-            $creationTime = isset($row['file1'][2]) ? (string) $row['file1'][2] : date('Y-m-d H:i:s', filectime($filePath));
-            $thumb1 = isset($row['thumb1']) ? $row['thumb1'] : '';
-            $thumb2 = isset($row['thumb2']) ? $row['thumb2'] : '';
-
-            // Write the properties to the HTML table
-            $html .= '<tr><td>Attachment</td><td>' . htmlspecialchars($filePath) . '</td><td>' . htmlspecialchars($row['file1'][0]) . '</td><td>' . htmlspecialchars($thumb1) . '</td><td>' . htmlspecialchars($thumb2) . '</td><td>' . htmlspecialchars($size) . '</td><td>' . htmlspecialchars($creationTime) . '</td></tr>';
-        }
-
-        $html .= '</tbody></table></body></html>';
-
-        fwrite($file, $html);
-        fclose($file);
-    }
-
-    /**
-     * Compares two CSV files and creates a new CSV file containing the matching entries.
-     *
-     * @param string $firstFile The path to the first CSV file.
-     * @param string $secondFile The path to the second CSV file.
-     * @param string $finalFile The path to the final CSV file to be created.
-     * @return void
-     */
-
-    function compareCSVFilesTransform($firstFile, $secondFile, $finalFile): void
-    {
-        // Read the first CSV file
-        $firstData = array_map('str_getcsv', file($firstFile));
-        $firstHeaders = array_shift($firstData);
-
-        // Read the second CSV file
-        $secondData = array_map('str_getcsv', file($secondFile));
-        $secondHeaders = array_shift($secondData);
-
-        // Find the indexes of the columns to compare in both files
-        $firstIdIndex = array_search('id', $firstHeaders);
-        $firstSizeIndex = array_search('size', $firstHeaders);
-        $firstCreationTimeIndex = array_search('creation_time', $firstHeaders);
-
-        $secondIdIndex = array_search('Entry Path', $secondHeaders);
-        $secondSizeIndex = array_search('size', $secondHeaders);
-        $secondCreationTimeIndex = array_search('creation_time', $secondHeaders);
-
-        // Get the entries from the first file
-        $firstEntries = [];
-        foreach ($firstData as $row) {
-            $firstEntries[$row[$firstIdIndex]] = [
-                'id' => $row[$firstIdIndex],
-                'size' => $row[$firstSizeIndex],
-                'creation_time' => $row[$firstCreationTimeIndex]
-            ];
-        }
-
-        // Create the final result CSV file
-        $finalCsvFile = fopen($finalFile, 'w');
-        fputcsv($finalCsvFile, ['id', 'size', 'creation_time']);
-
-        // Compare the entries from the second file with the entries in the first file
-        foreach ($secondData as $row) {
-            $secondId = $row[$secondIdIndex];
-            $secondSize = $row[$secondSizeIndex];
-            $secondCreationTime = $row[$secondCreationTimeIndex];
-
-            // Check if there is a matching entry in the first file
-            if (
-                isset($firstEntries[$secondId]) &&
-                $firstEntries[$secondId]['size'] === $secondSize &&
-                $firstEntries[$secondId]['creation_time'] === $secondCreationTime
-            ) {
-                fputcsv($finalCsvFile, [
-                    $secondId,
-                    $secondSize,
-                    $secondCreationTime
-                ]);
+     private function generateHtmlReport(array $inconsistentFiles, string $name): void
+     {   
+         $templateFile = 'report_template.html';
+         $templateContent = file_get_contents($templateFile);
+ 
+         if ($templateContent === false) {
+             throw new Exception('Failed to read the HTML template file.');
+         }
+ 
+         $tableRows = '';
+         foreach ($inconsistentFiles as $row) {
+             $filename = $row['file1'][0];
+             $filePath = $this->directory . '/' . $filename;
+             $size = isset($row['file1'][1]) ? (string) $row['file1'][1] : filesize($filePath);
+             $creationTime = isset($row['file1'][2]) ? str_replace('"', '', $row['file1'][2]) : date('Y-m-d H:i:s', filectime($filePath));
+             $thumb1 = $row['file1'][0] . '-thumb1';
+             $thumb2 = $row['file1'][0] . '-thumb2';
+             
+             if (isset($row['file1'][3]) && preg_match('/^[0-9]+\./', $row['file1'][3])) {
+                $filename = $row['file1'][3];
+                $filePath = $this->directory . '/' . $filename;
+                $thumb1 = '';
+                $thumb2 = '';
             }
-        }
-
-        fclose($finalCsvFile);
-    }
-
-    /**
-     * Deletes physical files based on the entries listed in a CSV file.
-     *
-     * @param string $csvFile The path to the CSV file containing the list of files to delete.
-     * @return void
-     */
-
-    private function deletePhysicalFilesFromCsv(string $csvFile): void
-    {
-        $directory = $this->directory . '/';
-        $entries = [];
-        if (($handle = fopen($csvFile, 'r')) !== false) {
-            // Collect the file paths to be deleted
-            while (($data = fgetcsv($handle)) !== false) {
-                // Skip the header row
-                if ($data[0] === 'File/Attachment') {
-                    continue;
-                }
-
-                $filePath = $directory . $data[1];
-                $entries[] = $filePath;
-            }
-            fclose($handle);
-        }
-
-        // Prompt the user for confirmation before deleting the files
-        echo "Before proceeding with the deletion, make sure you have a backup of your data." . PHP_EOL;
-        echo "You can revert back to the backup in case of accidental data loss." . PHP_EOL;
-        echo "Do you want to delete the physical files listed in the CSV report? (yes/no): ";
-        $confirmation = trim(fgets(STDIN));
-
-        if (strtolower($confirmation) === 'yes') {
-            // Create a log file to record the deleted files
-            $logFile = 'deleted_files.log';
-            $logHandle = fopen($logFile, 'a');
-
-            // Delete the physical files
-            foreach ($entries as $filePath) {
-                if (file_exists($filePath)) {
-                    //   unlink($filePath);
-                    echo "File deleted: $filePath" . PHP_EOL;
-
-                    // Write the deleted file path to the log file
-                    fwrite($logHandle, "Deleted file: $filePath" . PHP_EOL);
-                } else {
-                    echo "File not found: $filePath" . PHP_EOL;
-                }
-            }
-
-            fclose($logHandle);
-
-            echo "Deletion completed. The list of deleted files is saved in '$logFile'." . PHP_EOL;
-        } else {
-            echo "Deletion of physical files aborted." . PHP_EOL;
-        }
-    }
-
+            
+             if (is_string($row['id']) && strpos($row['id'], '.') !== false) {
+                 $filename = $row['id'];
+                 $filePath = $this->directory . '/' . $filename;
+                 $thumb1 = '';
+                 $thumb2 = '';
+             }
+            
+             $tableRows .= '<tr><td>Attachment</td><td>' . htmlspecialchars($filePath) . '</td><td>' . htmlspecialchars($filename) . '</td><td>' . htmlspecialchars($thumb1) . '</td><td>' . htmlspecialchars($thumb2) . '</td><td>' . htmlspecialchars($size) . '</td><td>' . htmlspecialchars($creationTime) . '</td></tr>';  
+         }  
+         $html = str_replace('{{table_rows}}', $tableRows, $templateContent); 
+         $file = fopen($name, 'w');
+      
+         if (!$file) {
+             throw new Exception('Failed to open the file for writing.');
+         }
+ 
+         fwrite($file, $html);
+         fclose($file);
+     }
 
     /**
      * Returns cassandra schema version
@@ -912,21 +860,22 @@ class DataConsistencyChecker
      *
      * @return int
      */
-
-    private function schema_version()
+    
+    private function schema_version(): int
     {
         return static::$schemaVersion;
     }
+    
     /**
      * Returns info(id,size) for attachment file
      *
      * @param int $clientId
      * @param string|null $bucketId
      * @param string $id
-     *
-     * @return object
+     * @return object|null 
      */
-    private function get_info(int $clientId, ?string $bucketId, string $id): ?object
+
+    private function get_info(int $clientId, ?string $bucketId, string $id): object
     {
         $attachment = null;
         $args = [
@@ -965,6 +914,7 @@ class DataConsistencyChecker
      * @param string $id
      * @return string
      */
+
     private function _set_bucket(string $id): string
     {
         return substr($id, 0, static::$bucketMagic);
@@ -978,9 +928,10 @@ class DataConsistencyChecker
      *
      * @return void
      */
+
     private function _update_attachment_stats(bool $add, int $size): void
     {
-        $op = $add ? '+' : '-';
+        $op = $add ? '+' : self::DELIMITER;
         $query = $this->cassandra->prepare('UPDATE attachment_stats SET count = count ' . $op . ' 1 where client_id = ?');
         $this->cassandra->execute($query, ['arguments' => ['client_id' => (int) static::$clientId]]);
         $query = $this->cassandra->prepare(
@@ -996,6 +947,7 @@ class DataConsistencyChecker
      *
      * @return object|null
      */
+
     private function _get_attachment_key(string $id): ?object
     {
         $result = null;
@@ -1009,7 +961,6 @@ class DataConsistencyChecker
         if ($data && $data->valid()) {
             $result = (object) $data->current();
         }
-        ;
         return $result;
     }
 
@@ -1018,14 +969,14 @@ class DataConsistencyChecker
      *
      * @param string $attachmentId
      *
-     * @return array
+     * @return object|null
      */
+
     private function _get_attachment_by_id(string $attachmentId): ?object
     {
         $attachment = null;
         try {
             $properties = '*';
-            echo "attid: " . $attachmentId . PHP_EOL;
             $key = $this->_get_attachment_key($attachmentId);
             if ($key) {
                 if ($this->schema_version() === 1) {
@@ -1050,8 +1001,6 @@ class DataConsistencyChecker
             }
         } catch (Cassandra\Exception\InvalidArgumentException $e) {
         }
-        //echo "ATTA: \n";
-        //var_dump($attachment);
 
         return $attachment != null && $attachment->valid() ? $this->_convert_to_object($attachment->current()) : null;
     }
@@ -1061,16 +1010,19 @@ class DataConsistencyChecker
      *
      * @param string $id
      */
+    
     private function _delete_attachment_key(string $id): void
     {
-        $result = null;
-        $query = $this->cassandra->prepare('DELETE FROM attachment_ids WHERE id = ? AND bucket = ? AND client_id = ?');
-        $arguments = [
-            'client_id' => static::$clientId,
-            'bucket' => $this->_set_bucket($id),
-            'id' => $id,
-        ];
-        $this->cassandra->executeAsync($query, ['arguments' => $arguments]);
+        $this->cassandra->executeAsync(
+            $this->cassandra->prepare('DELETE FROM attachment_ids WHERE id = ? AND bucket = ? AND client_id = ?'),
+            [
+                'arguments' => [
+                    'client_id' => static::$clientId,
+                    'bucket' => $this->_set_bucket($id),
+                    'id' => $id
+                ]
+            ]
+        );
     }
 
     /**
@@ -1082,6 +1034,7 @@ class DataConsistencyChecker
      *
      * @return void
      */
+
     private function _update_file_refs(string $data_id, bool $add, string $attachmentId = ''): void
     {
         $queryArguments = [
@@ -1089,7 +1042,7 @@ class DataConsistencyChecker
             'bucket' => $this->_set_bucket($data_id),
             'id' => $data_id,
         ];
-        $query = $this->cassandra->prepare('UPDATE attachment_file_refs SET ref_count = ref_count ' . ($add ? '+' : '-') . ' 1 WHERE bucket = ? AND id = ? AND client_id = ?');
+        $query = $this->cassandra->prepare('UPDATE attachment_file_refs SET ref_count = ref_count ' . ($add ? '+' : self::DELIMITER) . ' 1 WHERE bucket = ? AND id = ? AND client_id = ?');
         $this->cassandra->executeAsync($query, ['arguments' => $queryArguments]);
 
         if (!empty($attachmentId)) {
@@ -1115,6 +1068,7 @@ class DataConsistencyChecker
      *
      * @return bool
      */
+
     public function deleteAttachment(int $clientId, ?string $bucketId, string $id): bool
     {
         static::$clientId = $clientId;
@@ -1135,9 +1089,6 @@ class DataConsistencyChecker
         }
         $query = $this->cassandra->prepare($q);
         $this->cassandra->execute($query, ['arguments' => $refData]);
-
-
-        echo "DEL REFS\n";
         $refData['bucket'] = $this->_set_bucket($id);
         $query = $this->cassandra->prepare(
             'DELETE FROM attachment_file_refs WHERE bucket = ? AND id = ? AND client_id = ?'
@@ -1150,7 +1101,6 @@ class DataConsistencyChecker
         );
         $result = $this->cassandra->execute($query, ['arguments' => $refData]);
 
-        echo "DEL ATTACHMENTS\n";
         if ($this->schema_version() === 1) {
             $delQuery = $this->cassandra->prepare(
                 'DELETE FROM attachments WHERE id = ? AND client_id = ? AND project_id = ? AND entity_type = ?'
@@ -1258,6 +1208,7 @@ class DataConsistencyChecker
      * file,/tmp/testx3f,testx3f,testx3f-thumb1,testx3f-thumb2,4343,20-02-22 13:30,,,
      * cassandra,/test1/testx3,testx3,testx3-thumb1,testx3-thumb2,4343,20-02-22 13:30,1,3,3abc-def
      */
+
     public function processAttachmentDeletionCSV(string $file, string $src): void
     {
 
@@ -1278,7 +1229,7 @@ class DataConsistencyChecker
             if ($line === reset($file_lines)) {
                 continue;
             }
-            $values = explode(",", $line);
+            $values = explode(',', $line);
 
             if ($values[0] === 'File/Attachment') {
                 continue;
@@ -1293,11 +1244,13 @@ class DataConsistencyChecker
                 $data->size = $values[5];
                 $data->created = $values[6];
                 $path = dirname($data->path);
+
                 if (count($values) >= 10) {
                     $data->clientId = (int) $values[7];
                     $data->bucket = $values[8];
                     $data->id = $values[9];
                 }
+
                 if ($data->source === 'Attachment' && $src === 'cassandra') {
                     echo "will delete " . $data->clientId . " : " . $data->bucket . " : " . $data->id . PHP_EOL;
                     $this->deleteAttachment($data->clientId, $data->bucket, $data->id);
@@ -1306,12 +1259,14 @@ class DataConsistencyChecker
                     $filePath = $values[1];
                     $thumb1Path = $path . '/' . $values[3];
                     $thumb2Path = $path . '/' . $values[4];
+
                     if (file_exists($thumb1Path) && file_exists($thumb2Path)) {
                         files::delete($thumb1Path);
                         files::delete($thumb2Path);
                         fwrite($logHandle, "Deleted thumbnail: " . $thumb1Path  . PHP_EOL);
                         fwrite($logHandle, "Deleted thumbnail: " . $thumb2Path  . PHP_EOL);
                     }
+
                     if (file_exists($filePath)) {
                         unlink($filePath);
                         unlink($thumb1Path);
@@ -1325,9 +1280,7 @@ class DataConsistencyChecker
                 }
             }
         }
-
         fclose($logHandle);
-
         echo "Deletion completed. The list of attachments is saved in '$logFile'." . PHP_EOL;
     }
 }
diff --git a/report_template.html b/report_template.html
new file mode 100644
index 0000000..8e3d699
--- /dev/null
+++ b/report_template.html
@@ -0,0 +1,37 @@
+<html>
+<head>
+    <style>
+        table {
+            font-family: arial, sans-serif;
+            border-collapse: collapse;
+            width: 100%;
+        }
+        td, th {
+            border: 1px solid #dddddd;
+            text-align: left;
+            padding: 8px;
+        }
+        tr:nth-child(even) {
+            background-color: #dddddd;
+        }
+    </style>
+</head>
+<body>
+    <table>
+        <thead>
+            <tr>
+                <th>File/Attachment</th>
+                <th>File path</th>
+                <th>File name</th>
+                <th>Thumb 1</th>
+                <th>Thumb 2</th>
+                <th>Size</th>
+                <th>Creation Time</th>
+            </tr>
+        </thead>
+        <tbody>
+            {{table_rows}}
+        </tbody>
+    </table>
+</body>
+</html>

From b0acef47a1e8428107f73bbe67857abced05daa5 Mon Sep 17 00:00:00 2001
From: Spase <spase.papcev@gmail.com>
Date: Tue, 30 May 2023 18:39:44 +0200
Subject: [PATCH 4/4] Fixed a few bug reported from testing

---
 cassandra_consistency_script.php | 108 ++++++++++++++++++-------------
 1 file changed, 64 insertions(+), 44 deletions(-)

diff --git a/cassandra_consistency_script.php b/cassandra_consistency_script.php
index 24fa4e0..51d7e82 100644
--- a/cassandra_consistency_script.php
+++ b/cassandra_consistency_script.php
@@ -58,7 +58,7 @@ class DataConsistencyChecker
         $cassandraConfig = $config['CASSANDRA'];
         $this->_cluster = Cassandra::cluster()
             ->withContactPoints($cassandraConfig['host'])
-            ->withPort($cassandraConfig['port'])
+            ->withPort((int)$cassandraConfig['port'])
             ->withCredentials(
                 $cassandraConfig['user'],
                 $cassandraConfig['password']
@@ -99,11 +99,15 @@ class DataConsistencyChecker
          $structured_directory = $options['output'] ?? $options['o'] ?? null;
      
          if (!file_exists($structured_directory)) {
-             if ($structured_directory === null) {
+
+             if ($structured_directory === null && $source === null) {
                  echo "Output directory is required. Please specify the --output option.\n";
                  exit;
              }
-             mkdir($structured_directory, 0777, true);
+
+	     if ($structured_directory !== null) {
+             	mkdir($structured_directory, 0777, true);
+	     }
          }
      
          $this->structured_directory = $structured_directory;
@@ -118,7 +122,7 @@ class DataConsistencyChecker
              exit;
          }
 
-         if ($structured_directory === null) {
+         if ($structured_directory === null && $source === null) {
             echo "Output directory is required. Please specify the --output option.\n";
             exit;
          }
@@ -131,6 +135,7 @@ class DataConsistencyChecker
              $this->processAttachmentDeletionCSV($remove, $source);
          } else {
              $this->checkConsistency('attachment_file_info');
+
              if (is_dir($this->structured_directory)) {
                  $this->removeDirectory($this->structured_directory);
              }
@@ -153,16 +158,16 @@ class DataConsistencyChecker
 
         Options:
         -h, --help     Display this help screen.
-        --version      Set the schema version.
+        --version      Set the schema version (default: v1).
         --directory    Set the directory path for attachments.
-        --output       Set the folder for temp files
+        --output       Set the folder for temp files - note that this folder will be deleted after script completes
         --remove       We need to choose between file and cassandra what we want to remove
 
         Example:
         php script_name --version schema_version --directory=/path/to/directory --output ./out/
         For Delete:
         php script_name --remove result_from_cassandra_entries.csv --source file - to remove missing physical files
-        php script_name --remove result_from_physical_files.csv --source cassandra - to remove missing cassandra entries
+        php script_name --remove result_from_physical_files.csv --source cassandra --version 2 - to remove missing cassandra entries
         EOT;
         echo $helpMessage;
     }
@@ -186,7 +191,7 @@ class DataConsistencyChecker
         );
 
         foreach ($files as $file) {
-            $path = $directory . '/' . $file;
+            $path = $directory . DIRECTORY_SEPARATOR . $file;
 
             if (is_dir($path)) {
                 $this->removeDirectory($path);
@@ -231,8 +236,10 @@ class DataConsistencyChecker
                 if (strpos($fileName, '-thumb1') !== false || strpos($fileName, '-thumb2') !== false) {
                     continue;
                 }
+
                 $dashParts = explode(self::DELIMITER, $fileName, 2);
                 $dotParts = explode('.', $fileName);
+
                 if (count($dashParts) === 2) {
                     $clientId = $dashParts[0];
                     $id = $dashParts[1];
@@ -278,21 +285,24 @@ class DataConsistencyChecker
      * @return void
      */
 
-    private function createPhysicalFileCSV(string $clientId, array $entries): void
-    {
+      private function createPhysicalFileCSV(string $clientId, array $entries): void
+     {
         $fileName = $this->structured_directory . 'physical_' . $clientId . self::CSV_EXTENSION;
         $csvFile = fopen($fileName, 'w');
-        fputcsv($csvFile, ['id', 'size', 'creation_time']);
+        fputcsv($csvFile, ['id', 'size', 'creation_time', 'filename']);
+        
         foreach ($entries as $entry) {
-            fputcsv($csvFile, [
-                $entry['file_name'],
-                $entry['size'],
-                $entry['creation_time']
-            ]);
+                $idx = strpos($entry['file_name'],'.');
+                 $id = $idx === false ? $entry['file_name'] : substr($entry['file_name'],0,$idx);
+                 fputcsv($csvFile, [
+                    $id,
+                    $entry['size'],
+                    $entry['creation_time'],
+                    $entry['file_name']
+                  ]);
         }
         fclose($csvFile);
-    }
-
+     }
 
     /**
      * Creates a CSV file containing the entries for a specific client ID.
@@ -350,6 +360,11 @@ class DataConsistencyChecker
         $entries = [];
         while ($result) {
             foreach ($result as $row) {
+
+		        if ($row['bucket'] === 'x') {
+                    continue; // Skip the current iteration if bucket is 'x'
+                }
+
                 if (preg_match('/^[0-9]+$/', $row['id'])) {
                     $dotParts = explode(".", $row['filename'], 2);
                     $dynamicValue = substr($dotParts[1], 0, 2);
@@ -365,7 +380,11 @@ class DataConsistencyChecker
                         'client_id' => (string) $row['client_id'],
                         'attachment_id' => (string) $row['id'],
                     ];
-                    $entries[$dynamicValue][$row['filename']] = $entry;
+
+                    if ($row['bucket'] !== 'x') {
+                        $entries[$dynamicValue][$row['filename']] = $entry;
+                    }
+            
                 } else {
                     $clientId = substr($row['id'], 0, 2);
                     $file = $row['client_id'] . self::DELIMITER . $row['id'];
@@ -384,7 +403,9 @@ class DataConsistencyChecker
                     if (!isset($entries[$clientId])) {
                         $entries[$clientId] = [];
                     }
-                    $entries[$clientId][$file] = $entry;
+                    if ($row['bucket'] !== 'x') {
+                        $entries[$clientId][$file] = $entry;
+                    }
                 }
             }
             $result = $result->nextPage();
@@ -569,12 +590,10 @@ class DataConsistencyChecker
 
         $result_from_cassandra_entries = array_unique(array_merge($cassandra_entries, $missing_cassandra_entries), SORT_REGULAR);
         $result_from_physical_files = array_unique(array_merge($physical_entries, $missing_physical_files), SORT_REGULAR);
-
         $this->generateCsvReportForDbEntries($result_from_cassandra_entries, self::CASSANDRA_RESULT_CSV);
         $this->generateHtmlReport($result_from_cassandra_entries, self::CASSANDRA_HTML_REPORT);
         $this->generateCsvReportForPhysicalFiles($result_from_physical_files, self::PHYSICAL_RESULT_CSV);
         $this->generateHtmlReport($result_from_physical_files, self::PHYSICAL_HTML_REPORT);
-
     }
 
     /**
@@ -616,10 +635,8 @@ class DataConsistencyChecker
     {
         $file1_data = array_map('str_getcsv', file($file1_path));
         $file2_data = array_map('str_getcsv', file($file2_path));
-
         $file1_headers = array_shift($file1_data);
         $file2_headers = array_shift($file2_data);
-
         // find indexes of columns in each file
         $id_index_1 = array_search('id', $file1_headers);
         $id_index_2 = array_search('id', $file2_headers);
@@ -629,7 +646,6 @@ class DataConsistencyChecker
         $time_index_2 = array_search('creation_time', $file2_headers);
         $filename_index_1 = array_search('filename', $file1_headers);
         $filename_index_2 = array_search('filename', $file2_headers);
-
         $clientid_index_1 = array_search('client_id', $file1_headers);
         $bucket_index_1 = array_search('bucket', $file1_headers);
         $attachment_id_index_1 = array_search('attachment_id', $file1_headers);
@@ -675,7 +691,6 @@ class DataConsistencyChecker
             ];
             return $result;
         }, []);
-
         $missing_entries = [];
         foreach ($file1_assoc as $id => $data) {
             if (!isset($file2_assoc[$id])) {
@@ -694,7 +709,6 @@ class DataConsistencyChecker
                 }
             }
         }
-
         return $missing_entries;
     }
 
@@ -733,9 +747,9 @@ class DataConsistencyChecker
             $creationTime = isset($row['file1'][2]) ? (string) $row['file1'][2] : date('Y-m-d H:i:s', filectime($filePath));
             $thumb1 = isset($row['thumb1']) ? $row['thumb1'] : '';
             $thumb2 = isset($row['thumb2']) ? $row['thumb2'] : '';
-            $clientId = isset($row['file1'][4]) ? $row['file1'][4] : '';
-            $bucket = isset($row['file1'][5]) ? $row['file1'][5] : '';
-            $attachmentId = isset($row['file1'][5]) ? $row['file1'][6] : '';
+            $bucket = isset($row['file1'][4]) ? $row['file1'][4] : '';
+            $clientId = isset($row['file1'][5]) ? $row['file1'][5] : '';
+            $attachmentId = isset($row['file1'][6]) ? $row['file1'][6] : '';
 
             fputcsv($fp, [
                 'Attachment',
@@ -776,11 +790,11 @@ class DataConsistencyChecker
         ]);
 
         foreach ($inconsistentFiles as $row) {
-            $filePath = $this->directory . '/' . $row['file1'][0] ? $this->directory . '/' . $row['file1'][0] : $this->directory . '/' . $row['file2'][0];
+            $filePath = $this->directory . DIRECTORY_SEPARATOR . $row['file1'][3] ? $this->directory . DIRECTORY_SEPARATOR . $row['file1'][3] : $this->directory . DIRECTORY_SEPARATOR . $row['file2'][3];
             $size = isset($row['file1'][1]) ? (string) $row['file1'][1] : filesize($filePath);
             $creationTime = isset($row['file1'][2]) ? (string) $row['file1'][2] : date('Y-m-d H:i:s', filectime($filePath));
-            $thumb1 = $row['file1'][0] . '-thumb1';
-            $thumb2 = $row['file1'][0] . '-thumb2';
+            $thumb1 = $row['file1'][3] . '-thumb1';
+            $thumb2 = $row['file1'][3] . '-thumb2';
             if (is_string($row['id']) && strpos($row['id'], '.') !== false) {
                 $old_attachment = explode('.', $row['id'])[0];
                 $thumb1 = '';
@@ -796,7 +810,6 @@ class DataConsistencyChecker
                 $creationTime
             ]);
         }
-
         fclose($fp);
         chmod($name, 0666);
     }
@@ -821,7 +834,7 @@ class DataConsistencyChecker
          $tableRows = '';
          foreach ($inconsistentFiles as $row) {
              $filename = $row['file1'][0];
-             $filePath = $this->directory . '/' . $filename;
+             $filePath = $this->directory . DIRECTORY_SEPARATOR . $filename;
              $size = isset($row['file1'][1]) ? (string) $row['file1'][1] : filesize($filePath);
              $creationTime = isset($row['file1'][2]) ? str_replace('"', '', $row['file1'][2]) : date('Y-m-d H:i:s', filectime($filePath));
              $thumb1 = $row['file1'][0] . '-thumb1';
@@ -829,14 +842,14 @@ class DataConsistencyChecker
              
              if (isset($row['file1'][3]) && preg_match('/^[0-9]+\./', $row['file1'][3])) {
                 $filename = $row['file1'][3];
-                $filePath = $this->directory . '/' . $filename;
+                $filePath = $this->directory . DIRECTORY_SEPARATOR . $filename;
                 $thumb1 = '';
                 $thumb2 = '';
             }
             
              if (is_string($row['id']) && strpos($row['id'], '.') !== false) {
                  $filename = $row['id'];
-                 $filePath = $this->directory . '/' . $filename;
+                 $filePath = $this->directory . DIRECTORY_SEPARATOR . $filename;
                  $thumb1 = '';
                  $thumb2 = '';
              }
@@ -875,7 +888,7 @@ class DataConsistencyChecker
      * @return object|null 
      */
 
-    private function get_info(int $clientId, ?string $bucketId, string $id): object
+    private function get_info(int $clientId, ?string $bucketId, string $id): ?object
     {
         $attachment = null;
         $args = [
@@ -887,7 +900,6 @@ class DataConsistencyChecker
             $query = $this->cassandra->prepare('SELECT * FROM attachment_file_info WHERE id = ? AND client_id = ?');
         } else {
             $q = 'SELECT * FROM attachment_file_info WHERE id = ? AND client_id = ? AND bucket = \'' . $bucketId . '\'';
-            echo "Q: [" . $q . "]\n";
             $query = $this->cassandra->prepare($q);
         }
         $res = $this->cassandra->execute(
@@ -958,6 +970,7 @@ class DataConsistencyChecker
             'id' => $id,
         ];
         $data = $this->cassandra->execute($query, ['arguments' => $arguments]);
+
         if ($data && $data->valid()) {
             $result = (object) $data->current();
         }
@@ -1220,12 +1233,15 @@ class DataConsistencyChecker
         if (strtolower($confirmation) !== 'yes') {
             return;
         }
+
+	    $this->init();
         $file_contents = file_get_contents($file);
         $file_lines = explode("\n", $file_contents);
         $logFile = 'deleted_files.log';
         $logHandle = fopen($logFile, 'a');
 
         foreach ($file_lines as $line) {
+
             if ($line === reset($file_lines)) {
                 continue;
             }
@@ -1234,6 +1250,7 @@ class DataConsistencyChecker
             if ($values[0] === 'File/Attachment') {
                 continue;
             }
+            
             if (count($values) >= 7) {
                 $data = (object) array();
                 $data->source = $values[0];
@@ -1257,12 +1274,15 @@ class DataConsistencyChecker
                     fwrite($logHandle, "Deleted attachment: $data->id" . PHP_EOL);
                 } else if ($data->source === 'File' && $src === 'file') {
                     $filePath = $values[1];
-                    $thumb1Path = $path . '/' . $values[3];
-                    $thumb2Path = $path . '/' . $values[4];
+                    $thumb1Path = $path . DIRECTORY_SEPARATOR . $values[3];
+                    $thumb2Path = $path . DIRECTORY_SEPARATOR . $values[4];
+                    $filePath = str_replace('"', '', $filePath);
+                    $thumb1Path = str_replace('"', '', $thumb1Path);
+                    $thumb2Path = str_replace('"', '', $thumb2Path);
 
                     if (file_exists($thumb1Path) && file_exists($thumb2Path)) {
-                        files::delete($thumb1Path);
-                        files::delete($thumb2Path);
+                        unlink($thumb1Path);
+                        unlink($thumb2Path);
                         fwrite($logHandle, "Deleted thumbnail: " . $thumb1Path  . PHP_EOL);
                         fwrite($logHandle, "Deleted thumbnail: " . $thumb2Path  . PHP_EOL);
                     }
@@ -1286,4 +1306,4 @@ class DataConsistencyChecker
 }
 
 $checker = new DataConsistencyChecker();
-$checker->checkConsistency("attachment_file_info", true);
\ No newline at end of file
+$checker->checkConsistency("attachment_file_info", true);