<?php
/**
 * fix_image_paths.php
 *
 * Reescribe rutas relativas Joomla `images/...` en wp_posts.post_content
 * a rutas absolutas del site `/fea/wp-content/uploads/...`, pero solo cuando
 * el fichero correspondiente existe en /var/www/html/wp-content/uploads/.
 *
 * Cubre src= y href= con comillas dobles o simples.
 * URL-decodifica antes de comprobar el filesystem (mp3 con espacios/tildes).
 *
 * Issue: rafa/feadulta#34
 *
 * Usage:
 *   docker exec wordpress-web php /tmp/fix_image_paths.php --dry-run
 *   docker exec wordpress-web php /tmp/fix_image_paths.php           # live
 */

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

$db_host = 'wordpress-mysql';
$db_name = 'wordpress_db';
$db_user = 'wordpress_user';
$db_pass = 'wordpress_pass';

$uploads_fs   = '/var/www/html/wp-content/uploads';
$uploads_url  = '/fea/wp-content/uploads';

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

echo "=== Fix image paths (Joomla `images/...` → WP uploads) ===\n";
echo $dry_run ? "[DRY RUN]\n\n" : "[LIVE RUN]\n\n";

$stmt = $pdo->query("
    SELECT ID, post_title, post_content
    FROM wp_posts
    WHERE post_status IN ('publish','draft')
      AND post_type IN ('post','page')
      AND (
        post_content LIKE '%src=\"images/%'
        OR post_content LIKE \"%src='images/%\"
        OR post_content LIKE '%href=\"images/%'
        OR post_content LIKE \"%href='images/%\"
      )
");
$posts = $stmt->fetchAll(PDO::FETCH_ASSOC);
echo "Posts candidatos: " . count($posts) . "\n\n";

$stats = [
    'posts_changed'     => 0,
    'posts_unchanged'   => 0,
    'refs_rewritten'    => 0,
    'refs_missing_file' => 0,
];
$missing = [];          // path => count
$missing_per_post = []; // ID => [path,...]

// (src|href)=  ( " | ' ) images/...  ( " | ' )
$pattern = '/\b(src|href)=("|\')images\/([^"\']+)\2/i';

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

foreach ($posts as $post) {
    $original = $post['post_content'];
    $pid      = (int)$post['ID'];

    $content = preg_replace_callback(
        $pattern,
        function ($m) use ($uploads_fs, $uploads_url, &$stats, &$missing, &$missing_per_post, $pid) {
            $attr     = $m[1];
            $quote    = $m[2];
            $rel_enc  = $m[3];               // tal como aparece en HTML (puede ir URL-encoded)
            $rel_dec  = urldecode($rel_enc); // para mirar el filesystem
            $fs_path  = $uploads_fs . '/' . $rel_dec;

            if (is_file($fs_path)) {
                $stats['refs_rewritten']++;
                return $attr . '=' . $quote . $uploads_url . '/' . $rel_enc . $quote;
            }

            $stats['refs_missing_file']++;
            $missing[$rel_dec] = ($missing[$rel_dec] ?? 0) + 1;
            $missing_per_post[$pid][] = $rel_dec;
            return $m[0]; // dejar sin tocar
        },
        $original
    );

    if ($content !== $original) {
        $stats['posts_changed']++;
        if (!$dry_run) {
            $update->execute([$content, $pid]);
        }
    } else {
        $stats['posts_unchanged']++;
    }
}

echo "=== Resumen ===\n";
echo "Posts modificados:        {$stats['posts_changed']}\n";
echo "Posts sin cambios:        {$stats['posts_unchanged']}\n";
echo "Referencias reescritas:   {$stats['refs_rewritten']}\n";
echo "Referencias sin fichero:  {$stats['refs_missing_file']}\n";
echo "Rutas faltantes únicas:   " . count($missing) . "\n";

if (!empty($missing)) {
    arsort($missing);
    $log_path = '/tmp/fix_image_paths_missing.log';
    $lines = [];
    foreach ($missing as $path => $n) {
        $lines[] = sprintf("%4d  %s", $n, $path);
    }
    file_put_contents($log_path, implode("\n", $lines) . "\n");
    echo "\nLog rutas faltantes (orden por #ocurrencias): $log_path\n";
    echo "Top 15:\n";
    $i = 0;
    foreach ($missing as $path => $n) {
        echo sprintf("  %4d  %s\n", $n, $path);
        if (++$i >= 15) break;
    }

    // breakdown por carpeta raíz (segmento tras `images/`)
    $by_root = [];
    foreach ($missing as $path => $n) {
        $root = explode('/', $path)[0] ?? '?';
        $by_root[$root] = ($by_root[$root] ?? 0) + $n;
    }
    arsort($by_root);
    echo "\nFaltantes por carpeta raíz:\n";
    foreach ($by_root as $root => $n) {
        echo sprintf("  %4d  images/%s/\n", $n, $root);
    }
}

echo "\nHecho.\n";