Opencart - Crawls over database looking for images/pdfs used and outputs them as a simple text list. CP/Rclone/PHP can use this list to truncate out unused files.

dev_opencart_img_check.php

<?php
# Idea from: https://github.com/zenseo/opencart-needless-image
# PHP is not good enough to run this through a large store with many images
# Bash cp cant hold enough arguments from file output unless its batched
# Rclone is the best bet:
# rsync -asv --dry-run --recursive --remove-source-files --checksum --files-from=some-file.txt . destination/

function findImages() {

  $tables_to_check = array(
    'banner_image' => 'image',
    'category' => 'image',
    'download' => 'filename',
    'manufacturer' => 'image',
    'option_value' => 'image',
    'order_download' => 'filename',
    'product' => 'image',
    'product_image' => 'image',
    'setting' => 'value',
    'voucher_theme' => 'image',
  );

  $tables_desc_to_check = array(
    'bf_attribute_value' => 'value',
    'category_description' => 'description',
    //'category_description' => 'description_two',
    'information_description' => 'description',
    'product_attribute' => 'text',
    'product_description' => 'description',
    'product_tab' => 'text'
  );

  $escape_spaces = false;
  $img_matches = array();
  $img_regex = "/data\/.*?\.(jp(e)?g|png|gif|bmp)/i";

  foreach ($tables_desc_to_check as $name => $column) {

    $_parts_sql = 'SELECT `' . $column . '` image FROM `' . DB_PREFIX . $name . '` WHERE `' . $column . "` LIKE '%data/%'";
    $_parts = $this->db->query($_parts_sql);

    foreach ($_parts->rows as $_part_row) {
      preg_match_all($img_regex, $_part_row['image'], $img_matches);

      //$img_matches = array_unique($img_matches[0]);

      if (isset($img_matches[0])) {
        foreach ($img_matches[0] as $img_path) {

          $img_path = str_replace('%20', '', $img_path);

          if ($escape_spaces) {
            $img_path = str_replace(' ', '\\ ', $img_path);
          }

          $parts[] = $img_path;
        }

        unset($img_matches);
      }
    }
  }


  foreach ($tables_to_check as $name => $column) {

    $_parts_sql = 'SELECT `' . $column . '` image FROM `' . DB_PREFIX . $name . '` WHERE `' . $column . "` LIKE 'data/%'";
    $_parts = $this->db->query($_parts_sql);

    foreach ($_parts->rows as $_part_row) {

      $img_path = str_replace('%20', '', $_part_row['image']);

      if ($escape_spaces) {
        $img_path = str_replace(' ', '\\ ', $_part_row['image']);
      }

      $parts[] = $img_path;

    }
  }

  $parts = array_values(array_unique($parts));

  foreach ($parts as $echo) {
    echo $echo . PHP_EOL;
  }

  exit();
}

public function findPdfs() {
	
	$tables_desc_to_check = array(
		'category_description' => 'description',
		//'category_description' => 'description_two',
		'information_description' => 'description',
		'product_attribute' => 'text',
		'product_description' => 'description',
		'product_tab' => 'text'
	);
	
	$escape_spaces = false;
	$pdf_matches = array();
	$pdf_regex = "/PDFFILES\/.*?\.(jp(e)?g|png|gif|bmp|pdf)/i";
	
	foreach ($tables_desc_to_check as $name => $column) {

		$_parts_sql = 'SELECT `' . $column . '` pdf FROM `' . DB_PREFIX . $name . '` WHERE `' . $column . "` LIKE '%.pdf%'";
		$_parts = $this->db->query($_parts_sql);
		
		foreach ($_parts->rows as $_part_row) {
			preg_match_all($pdf_regex, $_part_row['pdf'], $pdf_matches);
			
			//$pdf_matches = array_unique($pdf_matches[0]);
			
			if (isset($pdf_matches[0])) {
				foreach ($pdf_matches[0] as $pdf_path) {
					
					$pdf_path = str_replace('%20', '', $pdf_path);
					
					if ($escape_spaces) {
						$pdf_path = str_replace(' ', '\\ ', $pdf_path);
					}
					
					$parts[] = $pdf_path;
				}
				
				unset($pdf_matches);
			}
		}
	}
	
	$parts = array_values(array_unique($parts));
	
	foreach ($parts as $echo) {
		echo $echo . PHP_EOL;
	}
	
	exit();
}