<?php
// OPUS5 — Orphans Classifier (doctrine 91)
// Lit /api/wevia-pages-registry.php?action=orphans et classifie :
// - LEGITIMATE_ARCHIVE : 404.html, test-*, v2/v3/legacy, hidden, deprecated → OK à archiver
// - ACTIVE_ORPHAN : page active non-branchée → À rebrancher
// - DORMANT_CANDIDATE : page ancienne mais potentiellement utile → Décision user
header('Content-Type: application/json');
header('Access-Control-Allow-Origin: *');
$t0 = microtime(true);
$R = ['ts'=>date('c'), 'source'=>'opus5-orphans-classifier'];

// Fetch orphans list depuis filesystem cache direct (bypass nginx 301)
$cache_path = '/var/www/html/api/wevia-pages-registry.cache.json';
if (file_exists($cache_path)) {
    $raw = file_get_contents($cache_path);
} else {
    $ch = curl_init('http://127.0.0.1/api/wevia-pages-registry.php?action=orphans');
    curl_setopt_array($ch, [CURLOPT_RETURNTRANSFER=>true, CURLOPT_TIMEOUT=>10, CURLOPT_FOLLOWLOCATION=>true]);
    $raw = curl_exec($ch);
    curl_close($ch);
}
$data = @json_decode((string)$raw, true) ?: [];
$orphans = $data['orphans'] ?? [];

// Patterns de classification
$archive_patterns = [
    '/^404\.html$/'                    => 'error_page_default',
    '/-v\d+\.html$/'                   => 'versioned_legacy',
    '/-legacy\.html$/'                 => 'marked_legacy',
    '/^test-/'                         => 'test_page',
    '/-test\.html$/'                   => 'test_page',
    '/-hidden\.html$/'                 => 'hidden_intentional',
    '/^google[a-z0-9]+\.html$/'        => 'google_verification',
    '/-iso3d\.html$/'                  => 'experimental_3d',
    '/-alive\.html$/'                  => 'demo_variant',
    '/-final\.html$/'                  => 'deprecated_final',
    '/-hd2\.html$/'                    => 'hd_variant',
    '/-3d\.html$/'                     => 'demo_3d',
    '/^data-deletion\.html$/'          => 'gdpr_required_page',
    '/-demo-/'                         => 'demo_page',
    '/^dormant-/'                      => 'explicitly_dormant',
    '/-offline\.html$/'                => 'offline_variant',
    '/^all-screens-live\.html$/'       => 'dev_internal'
];

// Patterns pages actives à rebrancher (whitelist métier)
$active_patterns = [
    '/^ethica-(login|chatbot)\.html$/' => 'ethica_module_active',
    '/^office-login\.html$/'           => 'office_sso_active',
    '/^qa-hub\.html$/'                 => 'qa_hub_active',
    '/^infra-monitor\.html$/'          => 'infra_live_active',
    '/^monitoring\.html$/'             => 'monitoring_active',
    '/^sso-monitor\.html$/'            => 'sso_monitoring_active',
    '/^cron-control\.html$/'           => 'cron_admin_active',
    '/^wevia-widget\.html$/'           => 'wevia_widget_consumer_facing',
    '/^plan-du-site\.html$/'           => 'sitemap_required_seo',
    '/^claw-(chat|code)\.html$/'       => 'openclaw_active',
    '/^droid-terminal\.html$/'         => 'terminal_active',
    '/^dmaic-workbench\.html$/'        => 'leansigma_active',
    '/^bpmn-studio-live\.html$/'       => 'process_modeler_active',
    '/^candidate-detail\.html$/'       => 'hr_candidate_active',
    '/^ecosysteme-ia-maroc\.html$/'    => 'ia_ecosystem_active',
    '/^weval-(arena|ops-screens)\.html$/' => 'internal_ops_active',
    '/^acquired-dashboard\.html$/'     => 'acquired_dashboard_active',
    '/^weval-enterprise-management\.html$/' => 'enterprise_mgmt_active',
    '/^enterprise-complete-v73\.html$/' => 'enterprise_v73_retained'
];

// Classification
$out = [
    'LEGITIMATE_ARCHIVE' => [],
    'ACTIVE_ORPHAN' => [],
    'DORMANT_CANDIDATE' => []
];

foreach ($orphans as $page => $meta) {
    $classified = false;
    
    // Check archive patterns first
    foreach ($archive_patterns as $pat => $reason) {
        if (preg_match($pat, $page)) {
            $out['LEGITIMATE_ARCHIVE'][] = [
                'page' => $page,
                'reason' => $reason,
                'class' => $meta['class'] ?? '?',
                'size_kb' => $meta['size_kb'] ?? 0,
                'mtime' => $meta['mtime'] ?? ''
            ];
            $classified = true;
            break;
        }
    }
    if ($classified) continue;
    
    // Check active patterns
    foreach ($active_patterns as $pat => $reason) {
        if (preg_match($pat, $page)) {
            $out['ACTIVE_ORPHAN'][] = [
                'page' => $page,
                'reason' => $reason,
                'class' => $meta['class'] ?? '?',
                'size_kb' => $meta['size_kb'] ?? 0,
                'mtime' => $meta['mtime'] ?? '',
                'action_required' => 'LINK_FROM_WTP_OR_HUB'
            ];
            $classified = true;
            break;
        }
    }
    if ($classified) continue;
    
    // Par défaut : DORMANT_CANDIDATE (décision user)
    $out['DORMANT_CANDIDATE'][] = [
        'page' => $page,
        'class' => $meta['class'] ?? '?',
        'size_kb' => $meta['size_kb'] ?? 0,
        'mtime' => $meta['mtime'] ?? '',
        'action_required' => 'USER_DECISION'
    ];
}

$R['classification'] = $out;
$R['summary'] = [
    'total_orphans' => count($orphans),
    'legitimate_archive' => count($out['LEGITIMATE_ARCHIVE']),
    'active_orphan_to_link' => count($out['ACTIVE_ORPHAN']),
    'dormant_candidate' => count($out['DORMANT_CANDIDATE']),
    'legit_pct' => count($orphans) > 0 ? round(count($out['LEGITIMATE_ARCHIVE']) / count($orphans) * 100, 1) : 0,
    'actionable_count' => count($out['ACTIVE_ORPHAN']) + count($out['DORMANT_CANDIDATE']),
    'archived_pct' => count($orphans) > 0 ? round(count($out['LEGITIMATE_ARCHIVE']) / count($orphans) * 100, 1) : 0
];

$R['recommendation'] = [];
if (count($out['ACTIVE_ORPHAN']) > 0) {
    $R['recommendation'][] = 'LINK ' . count($out['ACTIVE_ORPHAN']) . ' active orphans depuis WTP drawer ou Unified Hub';
}
if (count($out['DORMANT_CANDIDATE']) > 0) {
    $R['recommendation'][] = 'REVIEW ' . count($out['DORMANT_CANDIDATE']) . ' dormant candidates : decide keep+link ou archive';
}
if (count($out['LEGITIMATE_ARCHIVE']) > 0) {
    $R['recommendation'][] = 'ARCHIVE OK ' . count($out['LEGITIMATE_ARCHIVE']) . ' pages : test/legacy/deprecated (decision user si move vers /archive/)';
}

$R['doctrine'] = '91 — orphans classifier : separe archive legitime / active orphan / dormant candidate';
$R['total_ms'] = round((microtime(true) - $t0) * 1000);
echo json_encode($R, JSON_PRETTY_PRINT | JSON_UNESCAPED_UNICODE);