<?php
/**
 * NIS2 Agile - Ingest fonti ISO 27000 nella Knowledge Base (RAG)
 * ----------------------------------------------------------------------------
 * Indicizza nella collection Qdrant `nis2_kb` (scope SYSTEM) il CATALOGO CURATO
 * dei controlli ISO (dalla tabella iso27001_annex_controls) + le descrizioni
 * delle fonti ISO registrate in application/config/nis2_sources.php, cosi che
 * AIService::askWithRag() e il modulo SGSI abbiano grounding sui controlli
 * 27001/27017/27018.
 *
 * IMPORTANTE (copyright): NON viene indicizzato il testo letterale delle norme
 * ISO (coperto da copyright), ma solo titoli/sintesi proprietari e riferimenti.
 *
 * PREREQUISITI: migration 038 applicata (tabella iso27001_annex_controls
 * popolata) + accesso a Qdrant + Voyage. ESEGUIRE SU HETZNER:
 *   docker exec -i nis2-app php /var/www/nis2-agile/scripts/ingest-iso-sources.php
 *
 * Opzioni:
 *   --dry-run    costruisce i chunk e stampa le statistiche senza upsert
 * ============================================================================
 */

if (PHP_SAPI !== 'cli') { fwrite(STDERR, "Solo CLI\n"); exit(1); }

if (!defined('BASE_PATH')) define('BASE_PATH', dirname(__DIR__));
if (!defined('APP_PATH'))  define('APP_PATH', BASE_PATH . '/application');

require_once APP_PATH . '/config/env.php';
require_once APP_PATH . '/config/config.php';
require_once APP_PATH . '/config/database.php';
require_once APP_PATH . '/services/EmbedService.php';
require_once APP_PATH . '/services/VectorService.php';

$opts = getopt('', ['dry-run']);
$dryRun = isset($opts['dry-run']);

function logln(string $m): void { echo '[' . date('Y-m-d H:i:s') . "] $m\n"; }

function uuid(): string
{
    $b = random_bytes(16);
    $b[6] = chr((ord($b[6]) & 0x0f) | 0x40);
    $b[8] = chr((ord($b[8]) & 0x3f) | 0x80);
    return vsprintf('%s%s-%s-%s-%s-%s%s%s', str_split(bin2hex($b), 4));
}

function chunkText(string $text, int $size = 2000, int $overlap = 200): array
{
    $text = mb_convert_encoding($text, 'UTF-8', 'UTF-8');
    $chunks = []; $len = mb_strlen($text, 'UTF-8'); $start = 0;
    while ($start < $len) {
        $take = min($size, $len - $start);
        $piece = mb_substr($text, $start, $take, 'UTF-8');
        if (trim($piece) !== '') $chunks[] = $piece;
        if ($start + $take >= $len) break;
        $start += ($size - $overlap);
    }
    return $chunks;
}

$sources = require APP_PATH . '/config/nis2_sources.php';
$isoKeys = ['iso_27001_2022','iso_27002_2022','iso_27017_2015','iso_27018_2019'];

// Costruisce un documento di testo per ciascuno standard ISO: descrizione fonte
// + elenco dei controlli (codice, titolo IT/EN, tema) presi dal DB curato.
$standardMap = [
    'iso_27001_2022' => 'iso27001',
    'iso_27017_2015' => 'iso27017',
    'iso_27018_2019' => 'iso27018',
];

logln('=== Ingest fonti ISO 27000 nella KB (scope SYSTEM) ===');
if ($dryRun) logln('MODALITA DRY-RUN: nessun upsert.');

$embed = null; $vector = null;
if (!$dryRun) {
    $embed = new EmbedService();
    $vector = new VectorService();
    $vector->ensureCollection($embed->dims);
}

$totalChunks = 0; $done = 0;
foreach ($isoKeys as $key) {
    if (!isset($sources[$key])) { logln("SKIP {$key}: non nel registry"); continue; }
    $src = $sources[$key];

    $body = "FONTE: {$src['citation']}\nAUTORITA: {$src['authority']}\n\n{$src['full']}\n\n";

    // Catalogo controlli (se questo standard ha una mappatura DB)
    if (isset($standardMap[$key])) {
        $std = $standardMap[$key];
        try {
            $rows = Database::fetchAll(
                'SELECT control_code, theme, title_it, title_en, iso27002_ref, condition_tag
                   FROM iso27001_annex_controls WHERE standard = ? ORDER BY sort_order',
                [$std]
            );
        } catch (Throwable $e) {
            logln("  ERRORE lettura iso27001_annex_controls (migration 038 applicata?): " . $e->getMessage());
            $rows = [];
        }
        if ($rows) {
            $body .= "CATALOGO CONTROLLI ({$src['short']}):\n";
            foreach ($rows as $r) {
                $cond = $r['condition_tag'] ? " [condizionale: {$r['condition_tag']}]" : '';
                $ref  = $r['iso27002_ref'] ? " ({$r['iso27002_ref']})" : '';
                $body .= "- {$r['control_code']} - {$r['title_it']} / {$r['title_en']} [tema: {$r['theme']}]{$ref}{$cond}\n";
            }
        }
    }

    $body = preg_replace('/[ \t]+/', ' ', $body);
    $chunks = chunkText($body, 2000, 200);
    logln("Fonte: {$src['short']} -> " . count($chunks) . ' chunk');
    $totalChunks += count($chunks);

    if ($dryRun) { $done++; continue; }

    // Idempotenza: rimuovi i chunk SYSTEM esistenti per questa fonte
    try {
        $vector->deleteByFilter(['must' => [
            ['key' => 'scope', 'match' => ['value' => 'SYSTEM']],
            ['key' => 'source', 'match' => ['value' => $src['citation']]],
        ]]);
    } catch (Exception $e) { logln('  (warning) delete precedente: ' . $e->getMessage()); }

    $docUuid = uuid();
    $points = [];
    foreach ($chunks as $i => $chunk) {
        $vec = null;
        for ($try = 1; $try <= 5; $try++) {
            try { $vec = $embed->embed($chunk); break; }
            catch (Throwable $e) {
                if ($try === 5) { logln("  ERRORE embed chunk {$i}: " . $e->getMessage()); throw $e; }
                sleep($try);
            }
        }
        $points[] = [
            'id' => uuid(),
            'vector' => $vec,
            'payload' => [
                'doc_uuid'           => $docUuid,
                'title'              => $src['short'] . ($i > 0 ? ' (parte ' . ($i + 1) . ')' : ''),
                'chunk'              => $chunk,
                'entity_type'        => 'standard_iso',
                'source'             => $src['citation'],
                'lang'               => 'it',
                'scope'              => 'SYSTEM',
                'consulting_firm_id' => null,
                'organization_id'    => null,
                'shared_with_orgs'   => [],
                'uploaded_by'        => 0,
            ],
        ];
    }
    foreach (array_chunk($points, 64) as $batch) {
        $vector->upsertBatch($batch);
    }

    try {
        $stmt = Database::getInstance()->prepare(
            "INSERT INTO kb_uploaded_documents
             (qdrant_doc_uuid, scope, consulting_firm_id, organization_id, uploaded_by, title, entity_type, source, lang, chunk_count, shared_with_orgs, status)
             VALUES (?, 'SYSTEM', NULL, NULL, 0, ?, 'standard_iso', ?, 'it', ?, '[]', 'ready')"
        );
        $stmt->execute([$docUuid, $src['short'], $src['citation'], count($chunks)]);
    } catch (Exception $e) { logln('  (warning) tracking insert: ' . $e->getMessage()); }

    logln("  OK indicizzato (doc_uuid={$docUuid})");
    $done++;
}

logln("=== Completato: {$done} fonti ISO, {$totalChunks} chunk totali ===");