<?php
/**
 * Plugin Name: Glossary CSV Importer
 * Description: Imports Glossary CPT items from a CSV file (Term → title, Definition → ACF field glossary_definition).
 * Version: 1.1
 * Author: Your Name
 */
if (! defined('ABSPATH')) exit;
class Glossary_CSV_Importer
{
    public function __construct()
    {
        add_action('admin_menu', array($this, 'register_menu'));
        // Remove filters that strip special characters
        add_action('init', array($this, 'remove_sanitization_filters'));
    }
    
    public function remove_sanitization_filters()
    {
        // Only remove during import - we'll add them back
        // This is handled in process_import()
    }
    
    public function register_menu()
    {
        add_management_page(
            'Glossary CSV Import',
            'Glossary Import',
            'manage_options',
            'glossary-csv-import',
            array($this, 'import_page')
        );
    }
    public function import_page()
    {
        echo '<div class="wrap"><h1>Glossary CSV Import</h1>';
        if (isset($_POST['submit']) && check_admin_referer('glossary_import_form')) {
            $this->process_import();
        }
?>
        <form method="post" enctype="multipart/form-data">
            <?php wp_nonce_field('glossary_import_form'); ?>
            <p><label><strong>Select CSV File:</strong></label><br>
                <input type="file" name="csv_file" required>
            </p>
            <p><input type="submit" name="submit" class="button button-primary" value="Import CSV"></p>
        </form>
<?php
        echo '</div>';
    }
    private function process_import()
    {
        // Temporarily remove sanitization filters
        remove_filter('title_save_pre', 'wp_filter_kses');
        remove_filter('content_save_pre', 'wp_filter_kses');
        
        // force UTF-8 handling
        mb_internal_encoding("UTF-8");
        setlocale(LC_ALL, 'en_US.UTF-8');
        if (!isset($_FILES['csv_file']) || $_FILES['csv_file']['error'] !== UPLOAD_ERR_OK) {
            echo '<div class="error"><p>File upload failed.</p></div>';
            return;
        }
        $tmp = $_FILES['csv_file']['tmp_name'];
        // Read raw bytes
        $csv_raw = file_get_contents($tmp);
        if (!$csv_raw) {
            echo '<div class="error"><p>Could not read CSV.</p></div>';
            return;
        }
        // Remove BOM if present
        $csv_raw = preg_replace('/^\xEF\xBB\xBF/', '', $csv_raw);
        // Detect encoding
        $encoding = mb_detect_encoding($csv_raw, ['UTF-8', 'UTF-16', 'UTF-32', 'ISO-8859-1', 'WINDOWS-1252'], true);
        if (!$encoding) {
            $encoding = 'UTF-8';
        }
        // Convert to UTF-8
        if ($encoding !== 'UTF-8') {
            $csv_raw = mb_convert_encoding($csv_raw, 'UTF-8', $encoding);
        }
        // Write clean UTF-8 csv to real temporary file
        $converted = wp_tempnam('glossary.csv');
        file_put_contents($converted, $csv_raw);
        $handle = fopen($converted, 'r');
        if (!$handle) {
            echo '<div class="error"><p>Failed to open converted CSV.</p></div>';
            return;
        }
        $imported = 0;
        $row = 0;
        while (($data = fgetcsv($handle, 0, ',')) !== false) {
            // Skip empty or invalid rows
            if (!is_array($data) || count($data) < 2) continue;
            $row++;
            // Skip header
            if ($row === 1 && preg_match('/term/i', $data[0])) continue;
            // Keep full Unicode - no sanitization
            $term       = trim($data[0]);
            $definition = trim($data[1]);
            if ($term === '') continue;
            $existing = get_page_by_title($term, OBJECT, 'glossary');
            if ($existing) {
                $post_id = $existing->ID;
            } else {
                // Use wp_insert_post with special flag
                $post_id = wp_insert_post([
                    'post_title'  => sanitize_text_field($term),
                    'post_type'   => 'glossary',
                    'post_status' => 'publish',
                ], false, false); // Third parameter: false = don't sanitize
            }
            if ($post_id) {
                // Update ACF field directly in database to bypass filters
                // update_field() can still sanitize, so use update_post_meta
                update_post_meta($post_id, 'glossary_definition', $definition);
                
                // If you need ACF's internal reference keys:
                $field_object = get_field_object('glossary_definition', $post_id);
                if ($field_object) {
                    update_post_meta($post_id, '_glossary_definition', $field_object['key']);
                }
                
                $imported++;
            }
        }
        fclose($handle);
        
        // Re-add filters
        add_filter('title_save_pre', 'wp_filter_kses');
        add_filter('content_save_pre', 'wp_filter_kses');
        
        echo '<div class="updated"><p>Import complete. ' . $imported . ' items imported.</p></div>';
    }
}
new Glossary_CSV_Importer();