<?php
/**
 * fix_joomla_links.php
 *
 * Replaces Joomla internal links in WordPress post_content with correct WP URLs.
 *
 * Handles:
 *  1. index.php?option=com_content&view=article&id=NNN  → jos_content ID → WP post_name
 *  2. es/.../NNN-slug.html (relative)                   → K2 item ID    → WP post_name
 *  3. http://feadulta.com/es/.../NNN-slug.html           → K2 item ID    → WP post_name
 *  4. https://farmer.taild3aaf6.ts.net/fea/es/.../NNN-slug.html → K2 ID → WP post_name
 *
 * Usage: php fix_joomla_links.php [--dry-run]
 */

$dry_run = in_array('--dry-run', $argv ?? []);

// DB config
$db_host = 'wordpress-mysql';
$db_name = 'wordpress_db';
$db_user = 'wordpress_user';
$db_pass = 'wordpress_pass';
$wp_site_url = 'https://farmer.taild3aaf6.ts.net/fea';

$pdo = new PDO("mysql:host=$db_host;dbname=$db_name;charset=utf8mb4", $db_user, $db_pass, [
    PDO::ATTR_ERRMODE => PDO::ERRMODE_EXCEPTION,
]);

echo "=== Fix Joomla Internal Links ===\n";
echo $dry_run ? "[DRY RUN - no changes will be saved]\n\n" : "[LIVE RUN - changes will be saved]\n\n";

// -------------------------------------------------------------------------
// Step 1: Build lookup maps from wp_postmeta
// -------------------------------------------------------------------------
echo "Building lookup maps from wp_postmeta...\n";

// Map: K2 item ID → WP post_name
$k2_map = [];
$stmt = $pdo->query("
    SELECT pm.meta_value AS k2_id, p.post_name
    FROM wp_postmeta pm
    JOIN wp_posts p ON pm.post_id = p.ID
    WHERE pm.meta_key = '_fgj2wp_old_k2_id'
      AND p.post_status IN ('publish', 'draft')
      AND p.post_type = 'post'
      AND p.post_name != ''
");
foreach ($stmt as $row) {
    $k2_map[(int)$row['k2_id']] = $row['post_name'];
}
echo "  K2 map: " . count($k2_map) . " entries\n";

// Map: jos_content ID → WP post_name
$joomla_map = [];
$stmt = $pdo->query("
    SELECT pm.meta_value AS joomla_id, p.post_name
    FROM wp_postmeta pm
    JOIN wp_posts p ON pm.post_id = p.ID
    WHERE pm.meta_key = '_fgj2wp_old_id'
      AND p.post_status IN ('publish', 'draft')
      AND p.post_type = 'post'
      AND p.post_name != ''
");
foreach ($stmt as $row) {
    $joomla_map[(int)$row['joomla_id']] = $row['post_name'];
}
echo "  jos_content map: " . count($joomla_map) . " entries\n\n";

// -------------------------------------------------------------------------
// Step 2: Fetch posts with Joomla links
// -------------------------------------------------------------------------
$stmt = $pdo->query("
    SELECT ID, post_title, post_content
    FROM wp_posts
    WHERE post_type = 'post'
      AND post_status IN ('publish', 'draft')
      AND (
        post_content LIKE '%index.php?option=%'
        OR post_content LIKE '%\"es/%'
        OR post_content LIKE '%/es/%'
        OR post_content LIKE '%feadulta.com%'
        OR post_content LIKE '%farmer.taild3aaf6%'
      )
");
$posts = $stmt->fetchAll(PDO::FETCH_ASSOC);
echo "Posts to process: " . count($posts) . "\n\n";

// -------------------------------------------------------------------------
// Step 3: Process each post
// -------------------------------------------------------------------------
$stats = [
    'posts_changed'  => 0,
    'posts_skipped'  => 0,
    'links_replaced' => 0,
    'links_not_found'=> 0,
];
$not_found_log = [];

// Regex patterns (note: href values may be HTML-entity encoded: & → &amp;)
$patterns = [

    // Pattern A: index.php?option=com_content&[amp;]view=article&[amp;]id=NNN[;alias]
    'joomla_content' => '/href="index\.php\?option=com_content(?:&(?:amp;)?)[^"]*?(?:&(?:amp;)?)id=(\d+)[^"]*"/i',

    // Pattern B: K2 item links — only on known Joomla-origin domains/paths
    // Matches:
    //   href="es/[path/]NNN-slug.html"                               (relative)
    //   href="http://feadulta.com/[path/]NNN-slug.html"              (old domain)
    //   href="https://farmer.taild3aaf6.ts.net/fea/[path/]NNN-slug.html" (staging domain)
    'k2_item' => '/href="(?:(?:https?:\/\/feadulta\.com|https?:\/\/farmer\.taild3aaf6\.ts\.net\/fea)\/)?es\/[^"]*?\/(\d+)-[^"\/]+\.html[^"]*"/i',
];

$update_stmt = $pdo->prepare("UPDATE wp_posts SET post_content = ? WHERE ID = ?");

foreach ($posts as $post) {
    $original = $post['post_content'];
    $content  = $original;
    $changed  = false;

    // --- Pattern A: jos_content links ---
    $content = preg_replace_callback(
        $patterns['joomla_content'],
        function ($m) use ($joomla_map, $wp_site_url, &$stats, &$not_found_log, $post) {
            $id = (int)$m[1];
            if (isset($joomla_map[$id])) {
                $stats['links_replaced']++;
                $new_url = $wp_site_url . '/' . $joomla_map[$id] . '/';
                return 'href="' . $new_url . '"';
            }
            $stats['links_not_found']++;
            $not_found_log[] = "jos_content ID=$id not found (post {$post['ID']}: {$post['post_title']})";
            return $m[0]; // keep original
        },
        $content
    );

    // --- Pattern B: K2 item links ---
    $content = preg_replace_callback(
        $patterns['k2_item'],
        function ($m) use ($k2_map, $wp_site_url, &$stats, &$not_found_log, $post) {
            $id = (int)$m[1];

            // Skip if ID 0, or if this looks like a year (4 digits in 1900-2100 range) in a date URL
            if ($id === 0) return $m[0];

            // Skip pure numbers that are years in date-based URLs (e.g. /2024/01/post.html)
            // We check: if the full match contains /YYYY/ before the filename, skip
            if ($id >= 1990 && $id <= 2100 && preg_match('/\/\d{4}\//', $m[0])) {
                return $m[0];
            }

            if (isset($k2_map[$id])) {
                $stats['links_replaced']++;
                $new_url = $wp_site_url . '/' . $k2_map[$id] . '/';
                return 'href="' . $new_url . '"';
            }

            $stats['links_not_found']++;
            $not_found_log[] = "K2 ID=$id not found in map (post {$post['ID']}: {$post['post_title']}) | original: " . substr($m[0], 0, 100);
            return $m[0]; // keep original
        },
        $content
    );

    if ($content !== $original) {
        $changed = true;
        $stats['posts_changed']++;
        if (!$dry_run) {
            $update_stmt->execute([$content, $post['ID']]);
        } else {
            echo "  [DRY] Would update post {$post['ID']}: {$post['post_title']}\n";
        }
    } else {
        $stats['posts_skipped']++;
    }
}

// -------------------------------------------------------------------------
// Step 4: Summary
// -------------------------------------------------------------------------
echo "\n=== Results ===\n";
echo "Posts changed:       {$stats['posts_changed']}\n";
echo "Posts unchanged:     {$stats['posts_skipped']}\n";
echo "Links replaced:      {$stats['links_replaced']}\n";
echo "Links not resolved:  {$stats['links_not_found']}\n";

if (!empty($not_found_log)) {
    $log_path = '/tmp/fix_joomla_links_unresolved.log';
    file_put_contents($log_path, implode("\n", $not_found_log) . "\n");
    echo "\nUnresolved links logged to: $log_path\n";
    echo "First 10 unresolved:\n";
    foreach (array_slice($not_found_log, 0, 10) as $line) {
        echo "  $line\n";
    }
}

echo "\nDone.\n";