<?php
/**
 * WEVAL SITEMAP API
 * Énumère exhaustivement les 250+ pages HTML
 * Détecte orphelins (pas de href vers cette page depuis autre page)
 * Catégorise via patterns de nom
 * Source unique pour /weval-sitemap.html
 */
header('Content-Type: application/json; charset=utf-8');
header('Cache-Control: max-age=300');

$ROOT = '/var/www/html';
$pages = [];
$all_html = glob($ROOT . '/*.html');

// Build set of all referenced pages (to detect orphans)
$referenced = [];
foreach ($all_html as $f) {
    $name = basename($f);
    $content = @file_get_contents($f);
    if (!$content) continue;
    if (preg_match_all('#href\s*=\s*["\']\/?([a-zA-Z0-9_\-]+\.html)["\']#', $content, $m)) {
        foreach ($m[1] as $ref) {
            $referenced[$ref] = ($referenced[$ref] ?? 0) + 1;
        }
    }
}

// Categorize via patterns
function categorize($name) {
    $n = strtolower($name);
    // PRIORITY 1: Entry points / Catalogs (les portail/sitemap/index)
    if (preg_match('/^(weval-portal|pages-index|weval-sitemap|all-screens|weval-master-inventory|cartographie|weval-wiring|plan-du-site|integrations-marketplace|ia-registre|index\.html|apps\.html|solution-finder)/', $n)) return 'Entry·Portal';
    if (preg_match('/^(wevia|wedroid|wevcode|claw|droid-terminal)/', $n)) return 'WEVIA·Brain';
    if (preg_match('/(erp|launchpad)/', $n)) return 'ERP·Unified';
    if (preg_match('/(architecture|archi(?!ve)|sitemap)/', $n)) return 'Architecture';
    if (preg_match('/(agent|persona|avatar)/', $n)) return 'Agents';
    if (preg_match('/(dashboard|hub|center|cockpit)/', $n)) return 'Dashboards/Hubs';
    // PRIORITY 2: Lean Six Sigma + Quality (avant Admin)
    if (preg_match('/(value-stream|value-chain|vsm|dmaic|bpmn|kaizen|gemba|kanban|lean|kpi|maturity|mgmt|management|methodologie|mthodologie|playbook|ultimate-quality)/', $n)) return 'Lean·6σ·Quality';
    if (preg_match('/(test|nonreg|sigma|l99|playwright|living|proof|qa|debug|monitor|monitoring|sessions-monitor|sso-monitor|cyber-monitor|realtime-monitor|world-map-live|nl-autowire-status|wevads-performance)/', $n)) return 'Quality/Monitoring';
    if (preg_match('/(admin|director|manager|control|cron|maintenance|onboarding-em|tasks-live)/', $n)) return 'Admin/Ops';
    if (preg_match('/(crm|ethica|hcp|business|growth|pipeline|deal|consultant|candidate|kaouther|partners-emails|mission-billing|medreach)/', $n)) return 'Business/CRM';
    // PRIORITY 3: Email/Office Operations
    if (preg_match('/(office|gmail|gws-setup|azure|decision-gmail|smtp|warmup|sender|email)/', $n)) return 'Email·Office';
    if (preg_match('/(arena)/', $n)) return 'Operations·Arena';
    if (preg_match('/(deepseek|openclaw|mistral|ollama|gemini|nvidia|provider|sovereign|claude|ai-benchmark|oss-discovery|wiki(?!-)|faq|doctrine)/', $n)) return 'IA·Providers·Docs';
    if (preg_match('/(blade|gpu|infra(?!ction)|server|hetzner|cloudflare|huawei|google-hub|github-hub|caps|keys-hub|api-key)/', $n)) return 'Infrastructure';
    if (preg_match('/(visual)/', $n)) return 'Visual·Mgmt';
    if (preg_match('/(api|token|catalog|registry|capability)/', $n)) return 'APIs/Capabilities';
    if (preg_match('/(login|booking|cgu|consent|data-deletion|signup|register|pricing|privacy|terms|404|gosignup|googlecba|maintenance|pitch|pain-points|ecosysteme)/', $n)) return 'Public/Marketing';
    if (preg_match('/(deer|paperclip|qdrant|n8n|mattermost|listmonk|loki|plausible|arsenal)/', $n)) return 'Infra·Tools';
    if (preg_match('/(case|study|model|enterprise|wtp|technology|platform|golive|value-streaming|sales)/', $n)) return 'Marketing/Sales';
    return 'Autre';
}

// Build pages list
foreach ($all_html as $f) {
    $name = basename($f);
    $size = filesize($f);
    $mtime = filemtime($f);
    $is_orphan = !isset($referenced[$name]);
    $ref_count = $referenced[$name] ?? 0;
    $cat = categorize($name);
    $pages[] = [
        'name' => $name,
        'url' => '/' . $name,
        'category' => $cat,
        'size' => $size,
        'size_kb' => round($size / 1024, 1),
        'mtime' => $mtime,
        'mtime_iso' => date('c', $mtime),
        'mtime_human' => date('Y-m-d H:i', $mtime),
        'is_orphan' => $is_orphan,
        'ref_count' => $ref_count,
    ];
}

// Sort: by category, then by mtime desc
usort($pages, function($a, $b) {
    $c = strcmp($a['category'], $b['category']);
    if ($c !== 0) return $c;
    return $b['mtime'] - $a['mtime'];
});

// Group by category
$by_cat = [];
foreach ($pages as $p) {
    $by_cat[$p['category']][] = $p;
}

// Stats
$stats = [
    'total_pages' => count($pages),
    'total_categories' => count($by_cat),
    'orphan_count' => count(array_filter($pages, fn($p) => $p['is_orphan'])),
    'by_category_count' => array_map('count', $by_cat),
];

echo json_encode([
    'ok' => true,
    'ts' => date('c'),
    'source' => 'live filesystem scan',
    'stats' => $stats,
    'pages' => $pages,
    'by_category' => $by_cat,
], JSON_PRETTY_PRINT | JSON_UNESCAPED_UNICODE);