SIMON Master Console Rebuild

File viewer

guarded to /htdocs

/tools/site_map.php

<?php

/**
 * /tools/site_map.php — SIMON Site Map + Audit Intelligence (FULL + LOG BLOCK AUDIT)
 *
 * Adds:
 * - Plan-based classification: public | private | blocked | unknown
 * - Issue flags:
 *     - not_in_plan
 *     - stale_year
 *     - duplicate_index
 *     - exposes_log
 *     - log_file
 *     - missing_log_htaccess
 * - Folder rollups (dirAgg) and top-level rollups (topAgg)
 * - Findings section + JSON export: ?json=1
 *
 * Security:
 * - Protect /tools/ with Basic Auth or IP allowlist.
 */

header('Content-Type: text/html; charset=utf-8');

function h(string $s): string
{
    return htmlspecialchars($s, ENT_QUOTES, 'UTF-8');
}

$root = rtrim((string)($_SERVER['DOCUMENT_ROOT'] ?? ''), '/');
if ($root === '' || !is_dir($root)) {
    http_response_code(500);
    echo 'DOCUMENT_ROOT not found.';
    exit;
}

$host = (string)($_SERVER['HTTP_HOST'] ?? '');
$scheme = (!empty($_SERVER['HTTPS']) && $_SERVER['HTTPS'] !== 'off') ? 'https' : 'http';
$baseUrl = ($host !== '') ? ($scheme . '://' . $host) : '';

/* -----------------------------
   SIMON NDJSON logger
------------------------------ */
function simon_log_dir(string $root): string
{
    $d = $root . '/_logs';
    if (is_dir($d) || @mkdir($d, 0755, true)) {
        return $d;
    }
    return sys_get_temp_dir();
}

function simon_log_event(string $root, array $evt): void
{
    $dir = simon_log_dir($root);
    $file = rtrim($dir, '/') . '/simon_site_map.ndjson';
    $evt['ts'] = $evt['ts'] ?? gmdate('c');
    $evt['ip'] = $evt['ip'] ?? (string)($_SERVER['REMOTE_ADDR'] ?? '');
    $evt['ua'] = $evt['ua'] ?? (string)($_SERVER['HTTP_USER_AGENT'] ?? '');
    $evt['rid'] = $evt['rid'] ?? bin2hex(random_bytes(8));
    $line = json_encode($evt, JSON_UNESCAPED_SLASHES | JSON_UNESCAPED_UNICODE) . "\n";
    @file_put_contents($file, $line, FILE_APPEND | LOCK_EX);
}

/* -----------------------------
   Plan loader
------------------------------ */
function normalizePath(string $p): string
{
    $p = str_replace('\\', '/', $p);
    $p = preg_replace('#/+#', '/', $p);
    return $p ?: '/';
}

function load_plan(string $root): array
{
    $path = $root . '/tools/site_plan.json';
    $plan = [
        'expected_year' => (int)gmdate('Y'),
        'public_roots' => ['/'],
        'public_pages' => ['/index.php', '/sitemap.xml', '/robots.txt'],
        'private_roots' => ['/tools/', '/_inc/', '/_secrets/', '/_logs/', '/orb/_inc/', '/sitelogs/'],
        'blocked_roots' => ['/_secrets/', '/_logs/', '/sitelogs/'],
        'ignore_patterns' => [],
        'duplicate_index_patterns' => [],
    ];

if (is_file($path)) {
        $txt = @file_get_contents($path);
        $j = is_string($txt) ? json_decode($txt, true) : null;
        if (is_array($j)) {
            $plan = array_replace_recursive($plan, $j);
        }
    }

foreach (['public_roots', 'public_pages', 'private_roots', 'blocked_roots'] as $k) {
        $plan[$k] = array_values(array_unique(array_map('normalizePath', (array)($plan[$k] ?? []))));
    }

$plan['expected_year'] = (int)($plan['expected_year'] ?? (int)gmdate('Y'));
    return $plan;
}

$PLAN = load_plan($root);

function relPath(string $abs, string $root): string
{
    $abs = normalizePath($abs);
    $root = normalizePath($root);
    if (strpos($abs, $root) === 0) {
        $r = substr($abs, strlen($root));
        return normalizePath('/' . ltrim($r, '/'));
    }
    return $abs;
}

function fmtBytes(int $b): string
{
    $u = ['B', 'KB', 'MB', 'GB', 'TB'];
    $i = 0;
    $v = (float)$b;
    while ($v >= 1024 && $i < count($u) - 1) {
        $v /= 1024;
        $i++;
    }
    return sprintf(($i === 0 ? '%d %s' : '%.2f %s'), $v, $u[$i]);
}

/* -----------------------------
   Classification helpers
------------------------------ */
function starts_with_dir(string $path, string $dir): bool
{
    $path = normalizePath($path);
    $dir = normalizePath($dir);
    if (!str_ends_with($dir, '/')) {
        $dir .= '/';
    }
    return $path === rtrim($dir, '/') || str_starts_with($path, $dir);
}

function in_roots(string $path, array $roots): bool
{
    foreach ($roots as $r) {
        if (starts_with_dir($path, $r)) {
            return true;
        }
    }
    return false;
}

function classify_path(string $path, array $PLAN): string
{
    if (in_roots($path, (array)$PLAN['blocked_roots'])) {
        return 'blocked';
    }
    if (in_roots($path, (array)$PLAN['private_roots'])) {
        return 'private';
    }

foreach ((array)$PLAN['public_pages'] as $p) {
        if (normalizePath($p) === normalizePath($path)) {
            return 'public';
        }
    }

if (in_roots($path, (array)$PLAN['public_roots'])) {
        return 'public';
    }
    return 'unknown';
}

/* -----------------------------
   Log detection helpers
   Goal: BLOCK logs but not other files
------------------------------ */
function is_log_path(string $path): bool
{
    $p = normalizePath($path);

if (preg_match('#^/(?:_logs|sitelogs|web360/log)(?:/|$)#i', $p)) {
        return true;
    }
    if (preg_match('#^/(?:assets|assetts)/_logs(?:/|$)#i', $p)) {
        return true;
    }
    if (preg_match('#^/simon_ai/_logs(?:/|$)#i', $p)) {
        return true;
    }

if (preg_match('#\.(?:ndjson|log)$#i', $p)) {
        return true;
    }

return false;
}

function dir_has_htaccess(string $root, string $dirRel): bool
{
    $dirRel = normalizePath($dirRel);
    if (!str_ends_with($dirRel, '/')) {
        $dirRel .= '/';
    }
    $abs = rtrim($root, '/') . $dirRel;
    return is_dir($abs) && is_file($abs . '/.htaccess');
}

/* -----------------------------
   Config
------------------------------ */
$MAX_FILES = 20000;
$MAX_DEPTH = 25;
$INCLUDE_HIDDEN = false;
$FOLLOW_SYMLINKS = false;

$excludeDirs = [
    '/.git', '/.svn', '/.idea',
    '/vendor', '/node_modules',
    '/tmp', '/cache',
];
$excludeFiles = [
    '/.htaccess', '/.htpasswd',
    '/wp-config.php', '/config.php',
];

$sort = (string)($_GET['sort'] ?? 'path'); // path|size|mtime|type
$dirFilter = (string)($_GET['dir'] ?? '');
$q = trim((string)($_GET['q'] ?? ''));
$only = (string)($_GET['only'] ?? ''); // php|html|js|css|img|media|data|other|all
$jsonOut = isset($_GET['json']);
$wantAudit = isset($_GET['audit']) || $jsonOut;

/* -----------------------------
   Scan target
------------------------------ */
$dirFilter = normalizePath('/' . ltrim($dirFilter, '/'));
$scanBaseAbs = $root . ($dirFilter === '/' ? '' : $dirFilter);
if (!is_dir($scanBaseAbs)) {
    $scanBaseAbs = $root;
    $dirFilter = '/';
}

/* -----------------------------
   Ignore via plan patterns
------------------------------ */
function isExcluded(string $rel, array $excludeDirs, array $excludeFiles, array $PLAN): bool
{
    foreach ($excludeFiles as $f) {
        if ($rel === normalizePath($f)) {
            return true;
        }
    }
    foreach ($excludeDirs as $d) {
        $d = normalizePath($d);
        if ($rel === $d || str_starts_with($rel, $d . '/')) {
            return true;
        }
    }
    if (matches_any_pattern($rel, (array)($PLAN['ignore_patterns'] ?? []))) {
        return true;
    }
    return false;
}

/* -----------------------------
   Locked detection + error capture
------------------------------ */
$lockedDirs = [];
$lockedFiles = [];
$scanErrors = [];

set_error_handler(function ($severity, $message, $file, $line) use (&$scanErrors) {
    $scanErrors[] = [
        'op' => 'php_warning',
        'path' => (string)$file,
        'reason' => (string)$message,
        'line' => (int)$line,
        'severity' => (int)$severity,
    ];
    return true;
});

/* -----------------------------
   Inventory scan
------------------------------ */
$files = [];
$count = 0;
$stack = [['abs' => $scanBaseAbs, 'depth' => 0]];

while ($stack) {
    $node = array_pop($stack);
    $absDir = (string)$node['abs'];
    $depth = (int)$node['depth'];
    if ($depth > $MAX_DEPTH) {
        continue;
    }

$relDir = relPath($absDir, $root);
    if (isExcluded($relDir, $excludeDirs, $excludeFiles, $PLAN)) {
        continue;
    }

if (!is_readable($absDir) || !is_executable($absDir)) {
        $reason = [];
        if (!is_readable($absDir)) {
            $reason[] = 'not_readable';
        }
        if (!is_executable($absDir)) {
            $reason[] = 'not_executable';
        }
        $lockedDirs[] = ['path' => normalizePath($relDir) . '/', 'reason' => implode('+', $reason) ?: 'permission'];

$files[] = [
            'path' => normalizePath($relDir) . '/',
            'type' => 'dir',
            'size' => 0,
            'mtime' => @filemtime($absDir) ?: 0,
            'ext' => '',
            'group' => 'dir',
            'url' => '',
            'locked' => true,
            'lock_reason' => implode('+', $reason) ?: 'permission',
        ];
        continue;
    }

$items = @scandir($absDir);
    if (!is_array($items)) {
        $scanErrors[] = ['op' => 'scandir', 'path' => normalizePath($relDir) . '/', 'reason' => 'scandir_failed'];
        $lockedDirs[] = ['path' => normalizePath($relDir) . '/', 'reason' => 'scandir_failed'];

$files[] = [
            'path' => normalizePath($relDir) . '/',
            'type' => 'dir',
            'size' => 0,
            'mtime' => @filemtime($absDir) ?: 0,
            'ext' => '',
            'group' => 'dir',
            'url' => '',
            'locked' => true,
            'lock_reason' => 'scandir_failed',
        ];
        continue;
    }

$files[] = [
        'path' => normalizePath($relDir) . '/',
        'type' => 'dir',
        'size' => 0,
        'mtime' => @filemtime($absDir) ?: 0,
        'ext' => '',
        'group' => 'dir',
        'url' => '',
    ];

foreach ($items as $name) {
        if ($name === '.' || $name === '..') {
            continue;
        }
        if (!$INCLUDE_HIDDEN && str_starts_with($name, '.')) {
            continue;
        }

$abs = normalizePath($absDir . '/' . $name);
        $rel = relPath($abs, $root);

if (isExcluded($rel, $excludeDirs, $excludeFiles, $PLAN)) {
            continue;
        }

if (is_link($abs) && !$FOLLOW_SYMLINKS) {
            $files[] = [
                'path' => $rel,
                'type' => 'symlink',
                'size' => 0,
                'mtime' => @filemtime($abs) ?: 0,
                'ext' => '',
                'group' => 'other',
                'url' => '',
            ];
            continue;
        }

if (is_dir($abs)) {
            $stack[] = ['abs' => $abs, 'depth' => $depth + 1];
            continue;
        }

if (is_file($abs)) {
            $count++;
            if ($count > $MAX_FILES) {
                break 2;
            }

$ext = pathinfo($abs, PATHINFO_EXTENSION) ?: '';
            $group = extGroup($ext);

if ($only !== '' && $only !== 'all') {
                if ($only === 'img') {
                    if ($group !== 'img') {
                        continue;
                    }
                } else {
                    if ($group !== $only) {
                        continue;
                    }
                }
            }
            if ($q !== '' && stripos($rel, $q) === false) {
                continue;
            }

if (!is_readable($abs)) {
                $lockedFiles[] = ['path' => $rel, 'reason' => 'not_readable'];
                $files[] = [
                    'path' => $rel,
                    'type' => 'file',
                    'size' => 0,
                    'mtime' => @filemtime($abs) ?: 0,
                    'ext' => strtolower((string)$ext),
                    'group' => $group,
                    'url' => '',
                    'locked' => true,
                    'lock_reason' => 'not_readable',
                ];
                continue;
            }

$size = @filesize($abs) ?: 0;
            $mtime = @filemtime($abs) ?: 0;
            $url = $baseUrl ? ($baseUrl . $rel) : $rel;

$files[] = [
                'path' => $rel,
                'type' => 'file',
                'size' => $size,
                'mtime' => $mtime,
                'ext' => strtolower((string)$ext),
                'group' => $group,
                'url' => $url,
            ];
        }
    }
}

restore_error_handler();

/* -----------------------------
   Sort
------------------------------ */
usort($files, function ($a, $b) use ($sort) {
    $aSize = (int)($a['size'] ?? 0);
    $bSize = (int)($b['size'] ?? 0);
    $aM = (int)($a['mtime'] ?? 0);
    $bM = (int)($b['mtime'] ?? 0);
    $aP = (string)($a['path'] ?? '');
    $bP = (string)($b['path'] ?? '');
    $aG = (string)($a['group'] ?? '');
    $bG = (string)($b['group'] ?? '');

if ($sort === 'size') {
        return ($bSize <=> $aSize) ?: strcmp($aP, $bP);
    }
    if ($sort === 'mtime') {
        return ($bM <=> $aM) ?: strcmp($aP, $bP);
    }
    if ($sort === 'type') {
        return strcmp($aG, $bG) ?: strcmp($aP, $bP);
    }
    return strcmp($aP, $bP);
});

/* -----------------------------
   Audit intelligence
------------------------------ */
$expectedYear = (int)($PLAN['expected_year'] ?? (int)gmdate('Y'));

$byGroup = [];
$totalSize = 0;
$latest = [];
$largest = [];
$dirAgg = [];
$topAgg = [];

$issuesCount = [
    'not_in_plan' => 0,
    'stale_year' => 0,
    'duplicate_index' => 0,
    'exposes_log' => 0,
    'log_file' => 0,
    'missing_log_htaccess' => 0,
];
$issuesTop = [
    'not_in_plan' => [],
    'stale_year' => [],
    'duplicate_index' => [],
    'exposes_log' => [],
    'log_file' => [],
    'missing_log_htaccess' => [],
];

function top_level(string $path): string
{
    $p = normalizePath($path);
    if ($p === '/' || $p === '') {
        return '/';
    }
    $p = ltrim($p, '/');
    $first = explode('/', $p, 2)[0] ?? '';
    return $first === '' ? '/' : '/' . $first . '/';
}

foreach ($files as &$f) {
    $g = $f['group'] ?? 'other';
    $byGroup[$g] = ($byGroup[$g] ?? 0) + 1;
    $totalSize += (int)($f['size'] ?? 0);

$path = (string)($f['path'] ?? '');
    $type = (string)($f['type'] ?? '');
    $isFile = ($type === 'file' && empty($f['locked']));

$class = classify_path($path, $PLAN);
    $f['class'] = $class;

$dir = rtrim(dirname($path), '/');
    if ($dir === '.' || $dir === '') {
        $dir = '/';
    }
    $dirAgg[$dir]['count'] = ($dirAgg[$dir]['count'] ?? 0) + 1;
    $dirAgg[$dir]['size'] = ($dirAgg[$dir]['size'] ?? 0) + (int)($f['size'] ?? 0);

$top = top_level($path);
    $topAgg[$top]['count'] = ($topAgg[$top]['count'] ?? 0) + 1;
    $topAgg[$top]['size'] = ($topAgg[$top]['size'] ?? 0) + (int)($f['size'] ?? 0);
    $topAgg[$top]['class_' . $class] = ($topAgg[$top]['class_' . $class] ?? 0) + 1;

$f['issues'] = [];
    $f['plan_ok'] = true;

if ($isFile && is_log_path($path)) {
        $f['class'] = 'blocked';
        $f['issues'][] = 'log_file';
        $f['plan_ok'] = false;

$status = log_block_status($root, $path);
        $f['log_block_status'] = $status;

if ($status !== 'blocked_htaccess') {
            $f['issues'][] = 'missing_log_htaccess';
        }

$topAgg[$top]['logs'] = ($topAgg[$top]['logs'] ?? 0) + 1;
    }

if ($f['class'] === 'unknown' && $isFile) {
        $f['issues'][] = 'not_in_plan';
        $f['plan_ok'] = false;
    }

if (
        $isFile &&
        (
            preg_match('#/(_logs|assetts/_logs|assets/_logs)/#i', $path) ||
            preg_match('#\.ndjson$#i', $path) ||
            preg_match('#\.log$#i', $path)
        )
    ) {
        $f['issues'][] = 'exposes_log';
        $f['plan_ok'] = false;
    }

if ($isFile && matches_any_pattern($path, (array)($PLAN['duplicate_index_patterns'] ?? []))) {
        $f['issues'][] = 'duplicate_index';
        $f['plan_ok'] = false;
    }

if (
        $wantAudit &&
        $isFile &&
        in_array((string)($f['group'] ?? ''), ['php', 'html'], true) &&
        (int)($f['size'] ?? 0) > 0 &&
        (int)($f['size'] ?? 0) <= 250000
    ) {
        $abs = $root . $path;
        $txt = @file_get_contents($abs);
        if (is_string($txt) && $txt !== '') {
            if (preg_match_all('/\b(20\d{2})\b/', $txt, $m)) {
                $years = array_map('intval', $m[1]);
                $maxY = $years ? max($years) : 0;
                if ($maxY > 0 && $maxY < $expectedYear) {
                    $f['issues'][] = 'stale_year';
                    $f['plan_ok'] = false;
                    $f['year_max'] = $maxY;
                }
            }
        }
    }

if ($isFile) {
        $latest[] = $f;
        $largest[] = $f;
    }

foreach ($f['issues'] as $iss) {
        $issuesCount[$iss] = ($issuesCount[$iss] ?? 0) + 1;
        if (count($issuesTop[$iss]) < 50 && $isFile) {
            $issuesTop[$iss][] = $path;
        }
    }
}
unset($f);

usort($latest, fn($a, $b) => ((int)$b['mtime'] <=> (int)$a['mtime']) ?: strcmp($a['path'], $b['path']));
usort($largest, fn($a, $b) => ((int)$b['size'] <=> (int)$a['size']) ?: strcmp($a['path'], $b['path']));

$hotDirs = [];
foreach ($dirAgg as $d => $info) {
    $hotDirs[] = [
        'dir' => $d,
        'count' => (int)$info['count'],
        'size' => (int)$info['size'],
    ];
}
usort($hotDirs, fn($a, $b) => ($b['size'] <=> $a['size']) ?: ($b['count'] <=> $a['count']) ?: strcmp($a['dir'], $b['dir']));

$topDirs = [];
foreach ($topAgg as $d => $info) {
    $topDirs[] = [
        'top' => $d,
        'count' => (int)($info['count'] ?? 0),
        'size' => (int)($info['size'] ?? 0),
        'public' => (int)($info['class_public'] ?? 0),
        'private' => (int)($info['class_private'] ?? 0),
        'blocked' => (int)($info['class_blocked'] ?? 0),
        'unknown' => (int)($info['class_unknown'] ?? 0),
        'logs' => (int)($info['logs'] ?? 0),
    ];
}
usort($topDirs, fn($a, $b) => ($b['size'] <=> $a['size']) ?: ($b['count'] <=> $a['count']) ?: strcmp($a['top'], $b['top']));

/* -----------------------------
   Findings
------------------------------ */
$missingLogHtDirs = [];
foreach ($files as $f2) {
    if (($f2['type'] ?? '') !== 'file') {
        continue;
    }
    if (!empty($f2['locked'])) {
        continue;
    }
    $p2 = (string)($f2['path'] ?? '');
    if (!is_log_path($p2)) {
        continue;
    }

$status = (string)($f2['log_block_status'] ?? '');
    if ($status === 'blocked_htaccess') {
        continue;
    }

$d2 = normalizePath(dirname($p2));
    if ($d2 === '.' || $d2 === '') {
        $d2 = '/';
    }
    $missingLogHtDirs[$d2] = true;
}
$missingLogHtDirs = array_values(array_keys($missingLogHtDirs));
sort($missingLogHtDirs, SORT_STRING);

/* -----------------------------
   Log Intelligence events
------------------------------ */
simon_log_event($root, [
    'event' => 'site_map_view',
    'dir' => $dirFilter,
    'q' => $q,
    'only' => $only ?: 'all',
    'count' => count($files),
]);

simon_log_event($root, [
    'event' => 'site_map_audit_findings',
    'expected_year' => $expectedYear,
    'counts' => [
        'items' => count($files),
        'locked_dirs' => count($lockedDirs),
        'locked_files' => count($lockedFiles),
        'scan_errors' => count($scanErrors),
        'issues' => $issuesCount,
        'missing_log_htaccess_dirs' => count($missingLogHtDirs),
    ],
    'top' => [
        'not_in_plan' => array_slice($issuesTop['not_in_plan'], 0, 25),
        'stale_year' => array_slice($issuesTop['stale_year'], 0, 25),
        'duplicate_index' => array_slice($issuesTop['duplicate_index'], 0, 25),
        'exposes_log' => array_slice($issuesTop['exposes_log'], 0, 25),
        'log_file' => array_slice($issuesTop['log_file'], 0, 25),
        'missing_log_htaccess' => array_slice($issuesTop['missing_log_htaccess'], 0, 25),
    ],
    'logs' => [
        'folders_missing_htaccess' => array_slice($missingLogHtDirs, 0, 50),
    ],
]);

/* -----------------------------
   JSON export
------------------------------ */
if ($jsonOut) {
    header('Content-Type: application/json; charset=utf-8');
    echo json_encode([
        'ok' => true,
        'root' => $root,
        'base_url' => $baseUrl,
        'dir' => $dirFilter,
        'count' => count($files),
        'plan' => [
            'expected_year' => $expectedYear,
            'public_roots' => $PLAN['public_roots'],
            'public_pages' => $PLAN['public_pages'],
            'private_roots' => $PLAN['private_roots'],
            'blocked_roots' => $PLAN['blocked_roots'],
        ],
        'summary' => [
            'total_size' => $totalSize,
            'by_group' => $byGroup,
            'locked_dirs' => count($lockedDirs),
            'locked_files' => count($lockedFiles),
            'scan_errors' => count($scanErrors),
            'issues' => $issuesCount,
            'missing_log_htaccess_dirs' => count($missingLogHtDirs),
        ],
        'findings' => [
            'missing_log_htaccess_dirs' => $missingLogHtDirs,
        ],
        'insights' => [
            'largest_files' => array_slice($largest, 0, 25),
            'latest_changes' => array_slice($latest, 0, 25),
            'hot_directories' => array_slice($hotDirs, 0, 20),
            'top_level_directories' => array_slice($topDirs, 0, 200),
        ],
        'locked' => [
            'dirs' => $lockedDirs,
            'files' => $lockedFiles,
            'errors' => $scanErrors,
        ],
        'files' => $files,
    ], JSON_UNESCAPED_SLASHES | JSON_UNESCAPED_UNICODE | JSON_PRETTY_PRINT);
    exit;
}
?>
<!doctype html>
<html lang="en">
<head>
<meta charset="utf-8"/>
<meta name="viewport" content="width=device-width,initial-scale=1,viewport-fit=cover"/>
<title>SIMON Intelligence — Site Map (Full)</title>
<meta name="theme-color" content="#0b1022"/>
<style>
  :root{--bg:#070b16;--edge:rgba(255,255,255,.12);--ink:#eaf2ff;--muted:rgba(234,242,255,.70);--r:18px}
  *{box-sizing:border-box}
  body{margin:0;background:radial-gradient(1200px 700px at 50% 0%, rgba(107,77,255,.16), transparent 60%),radial-gradient(900px 600px at 10% 40%, rgba(106,224,255,.10), transparent 60%),var(--bg);color:var(--ink);font-family:ui-sans-serif,system-ui,-apple-system,Segoe UI,Roboto,Helvetica,Arial}
  .wrap{max-width:1280px;margin:22px auto;padding:0 14px}
  .card{border:1px solid var(--edge);background:linear-gradient(180deg, rgba(255,255,255,.07), rgba(255,255,255,.03));border-radius:var(--r);padding:14px;box-shadow:0 20px 70px rgba(0,0,0,.55)}
  .row{display:flex;gap:10px;flex-wrap:wrap;margin-top:10px}
  .pill{padding:8px 10px;border-radius:999px;border:1px solid var(--edge);background:rgba(0,0,0,.22);font-size:12px;color:var(--muted)}
  h2{margin:0 0 10px;font-size:13px;letter-spacing:.10em;text-transform:uppercase;color:rgba(234,242,255,.82)}
  .meta{color:var(--muted);font-size:13px;line-height:1.5}
  input,select{background:rgba(0,0,0,.25);border:1px solid var(--edge);color:var(--ink);border-radius:12px;padding:10px 12px;font-size:13px}
  button,.btn{background:rgba(107,77,255,.20);border:1px solid rgba(107,77,255,.45);color:var(--ink);border-radius:12px;padding:10px 12px;font-size:13px;cursor:pointer;text-decoration:none;display:inline-block}
  table{width:100%;border-collapse:collapse;margin-top:12px;overflow:hidden;border-radius:14px}
  th,td{padding:10px;border-bottom:1px solid rgba(255,255,255,.10);font-size:13px;vertical-align:top}
  th{text-align:left;color:rgba(234,242,255,.82);letter-spacing:.08em;text-transform:uppercase;font-size:11px}
  .path{font-family:ui-monospace,SFMono-Regular,Menlo,Monaco,Consolas,"Liberation Mono",monospace;word-break:break-all}
  .tag{padding:3px 8px;border-radius:999px;border:1px solid rgba(255,255,255,.14);background:rgba(0,0,0,.18);color:rgba(234,242,255,.78);font-size:11px}
  .tag.lock{border-color:rgba(255,120,120,.35);background:rgba(255,120,120,.10)}
  .tag.public{border-color:rgba(120,255,180,.30)}
  .tag.private{border-color:rgba(255,220,120,.30)}
  .tag.blocked{border-color:rgba(255,120,120,.45)}
  .tag.unknown{border-color:rgba(170,170,255,.30)}
  .issues{display:flex;gap:6px;flex-wrap:wrap;margin-top:6px}
  .issue{padding:2px 8px;border-radius:999px;border:1px solid rgba(255,80,160,.30);background:rgba(255,80,160,.10);font-size:11px;color:rgba(234,242,255,.85)}
  .issue.log_file{border-color:rgba(255,80,80,.45)}
  .issue.missing_log_htaccess{border-color:rgba(255,180,80,.45)}
  .controls{display:flex;gap:10px;flex-wrap:wrap;margin-top:12px}
  code{font-family:ui-monospace,SFMono-Regular,Menlo,Monaco,Consolas,"Liberation Mono",monospace}
</style>
</head>
<body>
<div class="wrap">
  <div class="card">
    <h2>SIMON Site Map (Full) + Audit + Log Block Audit</h2>
    <div class="meta">
      Root: <span class="path"><?= h($root) ?></span><br/>
      Base URL: <span class="path"><?= h($baseUrl ?: '(unknown)') ?></span><br/>
      Scanning: <span class="path"><?= h($dirFilter) ?></span><br/>
      Expected Year: <span class="path"><?= h((string)$expectedYear) ?></span><br/>
      NDJSON: <span class="path">/_logs/simon_site_map.ndjson</span>
    </div>

<div class="row">
      <div class="pill">Items: <b><?= h((string)count($files)) ?></b></div>
      <div class="pill">Total size: <b><?= h(fmtBytes($totalSize)) ?></b></div>
      <div class="pill">Locked dirs: <b><?= h((string)count($lockedDirs)) ?></b></div>
      <div class="pill">Locked files: <b><?= h((string)count($lockedFiles)) ?></b></div>
      <div class="pill">Scan errors: <b><?= h((string)count($scanErrors)) ?></b></div>
      <div class="pill">Issues: <b><?= h((string)array_sum($issuesCount)) ?></b></div>
    </div>

<div class="row">
      <div class="pill">not_in_plan: <b><?= h((string)$issuesCount['not_in_plan']) ?></b></div>
      <div class="pill">stale_year: <b><?= h((string)$issuesCount['stale_year']) ?></b></div>
      <div class="pill">duplicate_index: <b><?= h((string)$issuesCount['duplicate_index']) ?></b></div>
      <div class="pill">log_file: <b><?= h((string)$issuesCount['log_file']) ?></b></div>
      <div class="pill">missing_log_htaccess: <b><?= h((string)$issuesCount['missing_log_htaccess']) ?></b></div>
      <a class="btn" href="/tools/site_map_summary.php">Open Folder Summary</a>
    </div>

<form class="controls" method="get" action="">
      <input name="dir" placeholder="dir filter (e.g. simon_ai)" value="<?= h($dirFilter === '/' ? '' : ltrim($dirFilter, '/')) ?>" />
      <input name="q" placeholder="search path contains..." value="<?= h($q) ?>" />
      <select name="only">
        <?php
          $opts = ['all'=>'all','php'=>'php','html'=>'html','js'=>'js','css'=>'css','img'=>'img','media'=>'media','data'=>'data','other'=>'other'];
          foreach ($opts as $k=>$v) {
            $sel = ($only === $k) ? 'selected' : '';
            echo '<option value="' . h($k) . '" ' . $sel . '>' . h($v) . '</option>';
          }
        ?>
      </select>
      <select name="sort">
        <?php
          $sopts = ['path'=>'path','size'=>'size','mtime'=>'mtime','type'=>'type'];
          foreach ($sopts as $k=>$v) {
            $sel = ($sort === $k) ? 'selected' : '';
            echo '<option value="' . h($k) . '" ' . $sel . '>' . h($v) . '</option>';
          }
        ?>
      </select>
      <button type="submit">Refresh</button>
      <a class="btn" href="?<?= h(http_build_query(array_filter([
        'dir' => $dirFilter === '/' ? '' : ltrim($dirFilter, '/'),
        'q' => $q ?: null,
        'only' => $only ?: null,
        'sort' => $sort ?: null,
        'audit' => 1,
        'json' => 1,
      ], fn($v) => $v !== null && $v !== ''))) ?>">Export Audit JSON</a>
    </form>

<h2 style="margin-top:14px;">Findings</h2>
    <div class="meta">
      Log folders missing <code>.htaccess</code> (should be blocked):
      <?php if (!$missingLogHtDirs): ?>
        <b>none</b>
      <?php else: ?>
        <ul>
          <?php foreach (array_slice($missingLogHtDirs, 0, 30) as $d): ?>
            <li class="path"><?= h($d) ?>/</li>
          <?php endforeach; ?>
        </ul>
        <div class="meta">Fix: add a <code>.htaccess</code> inside each folder with “Require all denied”.</div>
      <?php endif; ?>
    </div>

<h2 style="margin-top:14px;">Top-level directory rollup</h2>
    <table>
      <thead>
        <tr>
          <th>Top Folder</th><th>Items</th><th>Size</th><th>Public</th><th>Private</th><th>Blocked</th><th>Unknown</th><th>Logs</th><th>Drill</th>
        </tr>
      </thead>
      <tbody>
        <?php foreach (array_slice($topDirs, 0, 60) as $t): ?>
          <tr>
            <td class="path"><?= h($t['top']) ?></td>
            <td><?= h((string)$t['count']) ?></td>
            <td><?= h(fmtBytes((int)$t['size'])) ?></td>
            <td><?= h((string)$t['public']) ?></td>
            <td><?= h((string)$t['private']) ?></td>
            <td><?= h((string)$t['blocked']) ?></td>
            <td><?= h((string)$t['unknown']) ?></td>
            <td><?= h((string)$t['logs']) ?></td>
            <td><a class="btn" href="/tools/site_map.php?dir=<?= h(urlencode(trim($t['top'], '/'))) ?>">open</a></td>
          </tr>
        <?php endforeach; ?>
      </tbody>
    </table>

<h2 style="margin-top:14px;">Full inventory</h2>
    <table>
      <thead>
        <tr>
          <th>Path</th><th>Class</th><th>Status</th><th>Size</th><th>Modified</th><th>URL</th>
        </tr>
      </thead>
      <tbody>
        <?php foreach ($files as $f): ?>
          <tr>
            <td class="path"><?= h((string)$f['path']) ?></td>
            <td>
              <?php
                $cls = (string)($f['class'] ?? 'unknown');
                echo '<span class="tag ' . h($cls) . '">' . h(strtoupper($cls)) . '</span>';
              ?>
              <?php if (!empty($f['issues'])): ?>
                <div class="issues">
                  <?php foreach ((array)$f['issues'] as $iss): ?>
                    <span class="issue <?= h((string)$iss) ?>"><?= h((string)$iss) ?></span>
                  <?php endforeach; ?>
                </div>
              <?php endif; ?>
              <?php if (!empty($f['log_block_status'])): ?>
                <div class="meta">log_block_status: <span class="path"><?= h((string)$f['log_block_status']) ?></span></div>
              <?php endif; ?>
              <?php if (!empty($f['year_max'])): ?>
                <div class="meta">max_year_found: <span class="path"><?= h((string)$f['year_max']) ?></span></div>
              <?php endif; ?>
            </td>
            <td>
              <?php if (!empty($f['locked'])): ?>
                <span class="tag lock">LOCKED</span>
                <small><?= h((string)($f['lock_reason'] ?? '')) ?></small>
              <?php else: ?>
                <span class="tag"><?= h(strtoupper((string)($f['group'] ?? 'OTHER'))) ?></span>
                <small><?= h((string)($f['type'] ?? '')) ?></small>
              <?php endif; ?>
            </td>
            <td><?= (($f['type'] ?? '') === 'file' && empty($f['locked'])) ? h(fmtBytes((int)($f['size'] ?? 0))) : '—' ?></td>
            <td><?= !empty($f['mtime']) ? h(date('Y-m-d H:i:s', (int)$f['mtime'])) : '—' ?></td>
            <td>
              <?php if (!empty($f['url'])): ?>
                <a href="<?= h((string)$f['url']) ?>" target="_blank" rel="noopener">open</a>
              <?php else: ?>—<?php endif; ?>
            </td>
          </tr>
        <?php endforeach; ?>
      </tbody>
    </table>

<div class="meta" style="margin-top:12px;">
      Classification uses <span class="path">/tools/site_plan.json</span>. Unknown files are “not in master plan” candidates.<br/>
      Logs are forced to <b>BLOCKED</b> by detection (folders like <code>/_logs</code>, <code>/assetts/_logs</code>, <code>/sitelogs</code>, <code>/web360/log</code>, and extensions <code>.log</code>/<code>.ndjson</code>).
    </div>
  </div>
</div>
</body>
</html>

Quick jump

open a path

True command center for SIMON + WEB360

File viewer

Quick jump