scanner.php

<?php
/**
 * DNS Sync Validator — Background Scanner
 *
 * Run via api.php which launches: nohup php scanner.php >> data/scanner.log 2>&1 &
 * Do NOT run directly through the web server.
 *
 * For each unique (domain, host, type) record group in domain_list.csv, this
 * script queries both DNS servers and records whether they return identical
 * results. Results are written incrementally to data/results.json so the
 * dashboard can display live progress.
 *
 * To stop a running scan gracefully, create the file data/stop_signal or use
 * the Stop button in the dashboard which calls api.php?action=stop.
 *
 * Copyright (c) 2026 Brad Boegler <bradthx@gmail.com>
 * GNU General Public License v3.0
 */

require_once __DIR__ . '/config.php';
require_once __DIR__ . '/lib/CsvParser.php';
require_once __DIR__ . '/lib/DnsChecker.php';
require_once __DIR__ . '/lib/ResultStore.php';

// Never time out regardless of how long the scan runs
set_time_limit(0);
ignore_user_abort(true);

// Ensure data directory exists
if (!is_dir(DATA_DIR)) {
    mkdir(DATA_DIR, 0755, true);
}

// Write PID file so api.php can verify the process is alive
file_put_contents(PID_FILE, getmypid());

// Remove any stale stop signal from a previous run
if (file_exists(STOP_SIGNAL_FILE)) {
    unlink(STOP_SIGNAL_FILE);
}

// ─── Initialization ───────────────────────────────────────────────────────────

$scanId     = date('Ymd_His');
$issuesFile = DATA_DIR . '/issues_' . $scanId . '.txt';

echo "[{$scanId}] Scanner starting (PID " . getmypid() . ")\n";

// Count total lines once for the progress bar denominator
$parser       = new CsvParser(CSV_FILE);
$totalRecords = $parser->countLines();

echo "[{$scanId}] Total CSV records: {$totalRecords}\n";

// Initialize result state and write immediately so the dashboard shows activity
$results = ResultStore::blank();
$results['scan_id']       = $scanId;
$results['status']        = 'running';
$results['total_records'] = $totalRecords;
$results['started_at']    = date('Y-m-d H:i:s');
$results['issues_file']   = basename($issuesFile);
ResultStore::write($results);

// Open the per-run issues file for appending domain names with mismatches
$issuesFh = fopen($issuesFile, 'w');
if ($issuesFh === false) {
    echo "[{$scanId}] ERROR: Cannot open issues file: {$issuesFile}\n";
    exit(1);
}

// ─── Per-domain tracking ─────────────────────────────────────────────────────

$groupsProcessed    = 0;
$processedLines     = 0;
$currentDomain      = null;
$domainAllMatched   = true;   // tracks if every group for the current domain matched

// ─── Main Scan Loop ───────────────────────────────────────────────────────────

foreach ($parser->streamGroups() as $group) {

    // Check for stop signal before starting each group
    if (file_exists(STOP_SIGNAL_FILE)) {
        echo "[{$scanId}] Stop signal received. Halting after {$groupsProcessed} groups.\n";
        $results['status'] = 'stopped';
        break;
    }

    $fqdn = DnsChecker::buildFqdn($group['domain'], $group['host']);
    $type = $group['type'];

    // Query both authoritative servers
    $primaryValues   = DnsChecker::query(PRIMARY_DNS_SERVER,   $fqdn, $type);
    $secondaryValues = DnsChecker::query(SECONDARY_DNS_SERVER, $fqdn, $type);

    // A match means both servers return the exact same set of values
    $match = DnsChecker::compare($primaryValues, $secondaryValues);

    // Update counters
    $groupsProcessed++;
    $processedLines += $group['lines'];

    $results['total_groups']    = $groupsProcessed;
    $results['processed_lines'] = $processedLines;

    if (!isset($results['by_type'][$type])) {
        $results['by_type'][$type] = ['matched' => 0, 'unmatched' => 0];
    }

    if ($match) {
        $results['matched']++;
        $results['by_type'][$type]['matched']++;
    } else {
        $results['unmatched']++;
        $results['by_type'][$type]['unmatched']++;
    }

    // Track per-domain state for the "last 25 domains" lists
    if ($group['domain'] !== $currentDomain) {
        if ($currentDomain !== null) {
            flushDomain($currentDomain, $domainAllMatched, $results, $issuesFh);
        }
        $currentDomain    = $group['domain'];
        $domainAllMatched = true;
    }
    if (!$match) {
        $domainAllMatched = false;
    }

    // Throttle to avoid hammering iterative DNS servers
    if (THROTTLE_USEC > 0) {
        usleep(THROTTLE_USEC);
    }

    // Flush results to disk periodically
    if ($groupsProcessed % FLUSH_EVERY === 0) {
        ResultStore::write($results);
        $pct = $totalRecords > 0 ? round($processedLines / $totalRecords * 100, 1) : 0;
        echo "[{$scanId}] {$groupsProcessed} groups | {$pct}% | matched: {$results['matched']} | unmatched: {$results['unmatched']}\n";
    }
}

// ─── Finalization ─────────────────────────────────────────────────────────────

// Flush the last domain
if ($currentDomain !== null) {
    flushDomain($currentDomain, $domainAllMatched, $results, $issuesFh);
}

if ($results['status'] !== 'stopped') {
    $results['status'] = 'complete';
}

ResultStore::write($results);
fclose($issuesFh);

// Remove PID file now that we are done
if (file_exists(PID_FILE)) {
    unlink(PID_FILE);
}

echo "[{$scanId}] Scan {$results['status']}. Groups: {$groupsProcessed} | Matched: {$results['matched']} | Unmatched: {$results['unmatched']}\n";

// ─── Helper Functions ─────────────────────────────────────────────────────────

/**
 * Record a completed domain into the appropriate recent-domains list.
 * Domains with any mismatch are also written to the issues file.
 *
 * @param string   $domain      The domain name
 * @param bool     $allMatched  True if every record group for this domain matched
 * @param array    $results     Results array passed by reference
 * @param resource $issuesFh    Open file handle to the issues text file
 */
function flushDomain(string $domain, bool $allMatched, array &$results, $issuesFh): void
{
    if ($allMatched) {
        $results['domains_matched']++;
        array_unshift($results['recent_matched_domains'], $domain);
        $results['recent_matched_domains'] = array_slice(
            $results['recent_matched_domains'], 0, 25
        );
    } else {
        $results['domains_unmatched']++;
        array_unshift($results['recent_unmatched_domains'], $domain);
        $results['recent_unmatched_domains'] = array_slice(
            $results['recent_unmatched_domains'], 0, 25
        );
        fwrite($issuesFh, $domain . "\n");
    }
}