html/api/pages-orphans-list.php

<?php
/**
 * Pages orphans list API V96.19
 * User WTP shows "54 pages orphelines détectées. Voir la liste"
 * But no API existed to return structured list with wire suggestions
 * 
 * Endpoint: /api/pages-orphans-list.php
 * Returns: 54 orphans with category, last_modified, size, wire_suggestion
 */
header('Content-Type: application/json; charset=utf-8');
header('Access-Control-Allow-Origin: *');

$DOCROOT = '/var/www/html';

// Read pages-index.php output if exists, else scan directly
$orphans_file = "$DOCROOT/api/orphans-json.json";
$data = @json_decode(@file_get_contents($orphans_file), true);

if (!is_array($data) || empty($data)) {
    // Fallback scan: find .html files not referenced from index.html or WTP
    $all_html = glob("$DOCROOT/*.html");
    $referenced = [];
    $anchor_files = [
        'index.html', 'weval-technology-platform.html', 'wtp.html',
        'admin-v2.html', 'admin.html', 'architecture.html', 'dg-command-center.html',
        'enterprise-model.html', 'wevia-master.html', 'tools-hub.html', 'apps.html',
        'agents-archi.html', 'nonreg.html', 'portal.html', 'portal-executive.html',
        'lean6sigma-dashboard.html', 'cartographie-screens.html', 'pages-index.html',
        'unified-hub.html', 'wevia-autonomy-dashboard.html', 'enterprise.html',
        'ethica-hub.html', 'cloudflare-hub.html', 'huawei-cloud.html', 'orphans-hub.html', 'go-100pct.html',
        'wevia-backoffice.html', 'wevia-business-visual-studio.html',
    ];
    foreach ($anchor_files as $af) {
        $af_path = "$DOCROOT/$af";
        if (file_exists($af_path)) {
            $content = file_get_contents($af_path);
            if (preg_match_all('/href="\/?([a-z0-9_\-\xc0-\xff%]+\.html)"/iu', $content, $matches)) {
                foreach ($matches[1] as $m) $referenced[$m] = true;
            }
        }
    }
    
    $orphans = [];
    foreach ($all_html as $path) {
        $name = basename($path);
        // V96.21 filter out backup/gold files (legitimate archives, not real orphans)
            if (preg_match("/\\.(gold|backup|bak|orig|pre-|\\.pre-|\\.GOLD-)/i", $name)) continue;
            if (preg_match("/^\\./", $name)) continue; // hidden
            if (preg_match("/(\\d{8}|\\d{14})\\.html$/", $name)) continue; // date-stamped backups
            if (!isset($referenced[$name])) {
            $orphans[] = [
                'file' => $name,
                'size_bytes' => filesize($path),
                'last_modified' => date('c', filemtime($path)),
                'age_days' => round((time() - filemtime($path)) / 86400),
            ];
        }
    }
    
    // Enrich with category + suggestion
    $categorize = function($name) {
        $n = strtolower($name);
        if (strpos($n, 'ethica') !== false) return ['marketing', 'Add link in WTP Marketing & Email module'];
        if (strpos($n, 'wevia') !== false) return ['ai', 'Add link in WTP Intelligence IA module'];
        if (strpos($n, 'wevads') !== false) return ['marketing', 'Add link in WTP Marketing & Email module'];
        if (strpos($n, 'sap') !== false || strpos($n, 'vistex') !== false) return ['erp', 'Add link in WTP ERP Integrations module'];
        if (strpos($n, 'huawei') !== false || strpos($n, 'cloud') !== false) return ['cloud', 'Add link in WTP Operations & Infra module'];
        if (strpos($n, 'agent') !== false) return ['ai', 'Add link in WTP Intelligence IA / RH & Talent'];
        if (strpos($n, 'test') !== false || strpos($n, 'demo') !== false) return ['dev', 'Archive to /tests/ or /demos/ folder'];
        if (strpos($n, 'admin') !== false || strpos($n, 'dashboard') !== false) return ['operations', 'Add to WTP Operations & Infra'];
        if (strpos($n, 'security') !== false || strpos($n, 'gdpr') !== false) return ['security', 'Add to WTP Security & Compliance'];
        if (strpos($n, 'consent') !== false || strpos($n, 'privacy') !== false) return ['legal', 'Link from footer legal/privacy'];
        if (strpos($n, 'ia') !== false || strpos($n, '-ai') !== false) return ['ai', 'Add link in WTP Intelligence IA'];
        if (strpos($n, 'landing') !== false || strpos($n, 'nearshore') !== false) return ['marketing-landing', 'Verify SEO referenced, else add to sitemap'];
        return ['other', 'Manual review - may be archived'];
    };
    
    foreach ($orphans as &$o) {
        list($cat, $sug) = $categorize($o['file']);
        $o['category'] = $cat;
        $o['wire_suggestion'] = $sug;
    }
    unset($o); // WAVE 207 fix PHP by-ref leak
    // WAVE 207 dedup by file name (defensive)
    $seen = [];
    $unique = [];
    foreach ($orphans as $o2) {
        if (!isset($seen[$o2['file']])) {
            $seen[$o2['file']] = 1;
            $unique[] = $o2;
        }
    }
    $orphans = $unique;
    
    // Count by category
    $by_cat = [];
    foreach ($orphans as $o) {
        $by_cat[$o['category']] = ($by_cat[$o['category']] ?? 0) + 1;
    }
    arsort($by_cat);
    
    $data = [
        'total_html_pages' => count($all_html),
        'referenced' => count($referenced),
        'orphans_count' => count($orphans),
        'orphans' => $orphans,
        'by_category' => $by_cat,
        'scan_anchors' => $anchor_files,
    ];
}

$data['v'] = 'V96.19-pages-orphans-opus';
$data['ts'] = date('c');
$data['note'] = 'Scan files in /var/www/html/*.html not referenced from index/WTP. Wire suggestions provided per file.';

echo json_encode($data, JSON_PRETTY_PRINT | JSON_UNESCAPED_UNICODE);