saufi/ChatbotAI
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

First commit

Browse Source
This commit is contained in:
Saufi
2026-05-18 08:56:23 +08:00
commit fd3d3a4d2b
147 changed files with 22099 additions and 0 deletions
Download Patch File Download Diff File Expand all files Collapse all files

229
app/Services/KnowledgeBase/AuditService.php Normal file
View File

@@ -0,0 +1,229 @@
<?php
namespace App\Services\KnowledgeBase;
use App\Models\AuditLog;
use Illuminate\Support\Facades\Auth;
use Illuminate\Support\Facades\Request;
/**
* AuditService
*
* Simpan audit trail untuk semua tindakan penting dalam sistem.
* Append-only — tiada delete atau update audit log.
*/
class AuditService
{
/**
* Log satu event.
*
* @param string $event Nama event (e.g. 'document.uploaded')
* @param mixed $model Model yang terlibat (optional)
* @param array $oldValues Data sebelum perubahan
* @param array $newValues Data selepas perubahan
* @param ?string $description Huraian untuk manusia
*/
public function log(
string $event,
mixed $model = null,
array $oldValues = [],
array $newValues = [],
?string $description = null
): AuditLog {
return AuditLog::create([
'user_id' => Auth::id(),
'event' => $event,
'auditable_type' => $model ? get_class($model) : null,
'auditable_id' => $model?->getKey(),
'old_values' => empty($oldValues) ? null : $oldValues,
'new_values' => empty($newValues) ? null : $newValues,
'description' => $description,
'ip_address' => Request::ip(),
'user_agent' => Request::userAgent(),
]);
}
// Shortcut methods untuk event biasa
public function documentUploaded($document, $version): void
{
$this->log(
'document.uploaded',
$document,
[],
[
'document_id' => $document->id,
'version_number' => $version->version_number,
'filename' => $version->original_filename,
],
"Dokumen '{$document->title}' versi {$version->version_number} diupload."
);
}
public function documentActivated($document): void
{
$this->log(
'document.activated',
$document,
['is_active' => false],
['is_active' => true],
"Dokumen '{$document->title}' diaktifkan."
);
}
public function documentDeactivated($document): void
{
$this->log(
'document.deactivated',
$document,
['is_active' => true],
['is_active' => false],
"Dokumen '{$document->title}' dinyahaktifkan."
);
}
public function documentReindexed($document, $version): void
{
$this->log(
'document.reindexed',
$version,
[],
['document_id' => $document->id, 'version_id' => $version->id],
"Dokumen '{$document->title}' versi {$version->version_number} diindeks semula."
);
}
public function knowledgeItemCreated($item): void
{
$this->log(
'knowledge_item.created',
$item,
[],
['title' => $item->title, 'type' => $item->item_type],
"Knowledge item '{$item->title}' ({$item->item_type}) dicipta."
);
}
public function knowledgeItemUpdated($item, array $oldValues): void
{
$this->log(
'knowledge_item.updated',
$item,
$oldValues,
$item->getAttributes(),
"Knowledge item '{$item->title}' dikemaskini."
);
}
public function knowledgeItemDeactivated($item): void
{
$this->log(
'knowledge_item.deactivated',
$item,
['is_active' => true],
['is_active' => false],
"Knowledge item '{$item->title}' dinyahaktifkan."
);
}
public function faqConvertedFromFeedback($feedback, $knowledgeItem): void
{
$this->log(
'faq.converted_from_feedback',
$knowledgeItem,
[],
['feedback_id' => $feedback->id, 'knowledge_item_id' => $knowledgeItem->id],
"FAQ baru '{$knowledgeItem->title}' dicipta dari feedback chat."
);
}
public function categoryCreated($category): void
{
$this->log(
'category.created',
$category,
[],
['name' => $category->name, 'slug' => $category->slug],
"Kategori '{$category->name}' dicipta."
);
}
public function systemReindexStarted(string $scope): void
{
$this->log(
'system.reindex_started',
null,
[],
['scope' => $scope],
"Reindeks sistem dimulakan untuk: {$scope}"
);
}
// =========================================================================
// CHUNK REVIEW & EDITING EVENTS
// =========================================================================
public function chunkFinalTextEdited($chunk, ?string $oldText, string $newText): void
{
$this->log(
'chunk.final_text_edited',
$chunk,
['final_text' => mb_substr($oldText ?? '[content asal]', 0, 200)],
['final_text' => mb_substr($newText, 0, 200)],
"final_text chunk #{$chunk->chunk_index} (ID: {$chunk->id}) diedit. Reindex diantrikan."
);
}
public function chunkExcluded($chunk, string $oldStatus): void
{
$this->log(
'chunk.excluded',
$chunk,
['chunk_status' => $oldStatus, 'is_active' => true],
['chunk_status' => 'excluded', 'is_active' => false],
"Chunk #{$chunk->chunk_index} (ID: {$chunk->id}) dikecualikan dari indexing."
);
}
public function chunkIncluded($chunk, string $oldStatus): void
{
$this->log(
'chunk.included',
$chunk,
['chunk_status' => $oldStatus, 'is_active' => false],
['chunk_status' => $chunk->chunk_status, 'is_active' => true],
"Chunk #{$chunk->chunk_index} (ID: {$chunk->id}) dikembalikan ke indexing."
);
}
public function chunkReindexTriggered($chunk): void
{
$this->log(
'chunk.reindex_triggered',
$chunk,
[],
['chunk_status' => 'needs_reindex'],
"Reindex manual dicetuskan untuk chunk #{$chunk->chunk_index} (ID: {$chunk->id})."
);
}
public function chunkSplit($parentChunk, array $children, string $splitGroupId): void
{
$childIds = array_map(fn($c) => $c->id, $children);
$this->log(
'chunk.split',
$parentChunk,
['chunk_status' => 'indexed', 'is_active' => true],
[
'chunk_status' => 'superseded',
'is_active' => false,
'split_group_id' => $splitGroupId,
'child_count' => count($children),
'child_chunk_ids' => $childIds,
],
"Chunk #{$parentChunk->chunk_index} (ID: {$parentChunk->id}) di-split kepada "
. count($children) . " chunk baharu. Split group: {$splitGroupId}"
);
}
}

438
app/Services/KnowledgeBase/IngestionService.php Normal file
View File

@@ -0,0 +1,438 @@
<?php
namespace App\Services\KnowledgeBase;
use App\Models\DocumentChunk;
use App\Models\DocumentVersion;
use App\Models\KnowledgeItem;
use App\Models\ProcessingLog;
use App\Services\Document\ChunkingService;
use App\Services\Document\PdfExtractorService;
use App\Services\Ollama\OllamaService;
use App\Services\Qdrant\QdrantService;
use Illuminate\Support\Facades\DB;
use Illuminate\Support\Facades\Log;
use Illuminate\Support\Str;
use RuntimeException;
/**
* IngestionService
*
* Menyelaras keseluruhan proses ingestion dokumen:
* Extract → Chunk → Embed → Qdrant Sync
*
* Ini adalah "orchestrator" — ia koordinasi semua service lain.
* Setiap langkah dilog dalam processing_logs untuk monitoring.
*/
class IngestionService
{
public function __construct(
private readonly PdfExtractorService $extractor,
private readonly ChunkingService $chunker,
private readonly OllamaService $ollama,
private readonly QdrantService $qdrant,
) {}
private function normalizeExtractedText(string $text): string
{
$text = str_replace(["\r\n", "\r"], "\n", $text);
// Buang control character pelik kecuali newline dan tab
$text = preg_replace('/[^\P{C}\n\t]+/u', '', $text);
// Tukar multiple whitespace kepada satu space, tapi kekalkan line break asas
$text = preg_replace("/[ \t]+/u", ' ', $text);
$text = preg_replace("/\n{3,}/u", "\n\n", $text);
return trim($text);
}
/**
* Proses penuh satu document version.
* Dipanggil oleh ProcessUploadedDocumentJob.
*
* @throws RuntimeException Jika proses gagal pada mana-mana langkah
*/
public function processDocumentVersion(DocumentVersion $version): void
{
$startTime = microtime(true);
Log::info("Mula proses document version {$version->id}", [
'document_id' => $version->document_id,
'version' => $version->version_number,
]);
// ── Langkah 1: Extract ──────────────────────────────────────────────
$version->updateStatus(DocumentVersion::STATUS_EXTRACTING);
ProcessingLog::record(
DocumentVersion::class,
$version->id,
ProcessingLog::STAGE_EXTRACT,
ProcessingLog::STATUS_STARTED
);
$extraction = $this->extractor->extract(
$version->stored_path,
config('knowledgebase.upload.storage_disk', 'local')
);
if (!$extraction['success']) {
$version->updateStatus(DocumentVersion::STATUS_EXTRACTION_FAILED, $extraction['error']);
ProcessingLog::record(
DocumentVersion::class,
$version->id,
ProcessingLog::STAGE_EXTRACT,
ProcessingLog::STATUS_FAILED,
$extraction['error']
);
throw new RuntimeException(
"Pengekstrakan teks gagal: " . $extraction['error']
);
}
// Kemaskini page count jika dapat
if ($extraction['page_count'] > 0) {
$version->update(['page_count' => $extraction['page_count']]);
}
ProcessingLog::record(
DocumentVersion::class,
$version->id,
ProcessingLog::STAGE_EXTRACT,
ProcessingLog::STATUS_COMPLETED,
null,
['page_count' => $extraction['page_count']]
);
// ── Langkah 2: Chunk ─────────────────────────────────────────────────
$version->updateStatus(DocumentVersion::STATUS_CHUNKING);
ProcessingLog::record(
DocumentVersion::class,
$version->id,
ProcessingLog::STAGE_CHUNK,
ProcessingLog::STATUS_STARTED
);
// Normalize teks sebelum dihantar ke chunker
$normalizedText = $this->normalizeExtractedText($extraction['full_text']);
$chunks = $this->chunker->chunk(
$normalizedText,
$extraction['pages']
);
if (empty($chunks)) {
$version->updateStatus(DocumentVersion::STATUS_FAILED, 'Tiada chunk dihasilkan dari teks.');
ProcessingLog::record(
DocumentVersion::class,
$version->id,
ProcessingLog::STAGE_CHUNK,
ProcessingLog::STATUS_FAILED,
'Tiada chunk dihasilkan'
);
throw new RuntimeException('Tiada chunk dihasilkan dari dokumen.');
}
// Deactivate chunk versi sebelumnya (jika ini bukan versi pertama)
$this->deactivatePreviousChunks($version);
// Simpan chunk baru dalam MySQL
$savedChunks = $this->saveChunks($version, $chunks);
ProcessingLog::record(
DocumentVersion::class,
$version->id,
ProcessingLog::STAGE_CHUNK,
ProcessingLog::STATUS_COMPLETED,
null,
['chunk_count' => count($savedChunks)]
);
// ── Langkah 3: Embed & Qdrant ────────────────────────────────────────
$version->updateStatus(DocumentVersion::STATUS_EMBEDDING);
ProcessingLog::record(
DocumentVersion::class,
$version->id,
ProcessingLog::STAGE_EMBED,
ProcessingLog::STATUS_STARTED
);
$this->embedAndSyncChunks($version, $savedChunks);
// ── Selesai ──────────────────────────────────────────────────────────
$version->updateStatus(DocumentVersion::STATUS_INDEXED);
// Aktifkan dokumen jika ini versi pertama yang berjaya
$document = $version->document;
if ($document->status !== 'active') {
$document->update([
'status' => 'active',
'is_active' => true,
]);
}
$duration = round(microtime(true) - $startTime, 2);
ProcessingLog::record(
DocumentVersion::class,
$version->id,
ProcessingLog::STAGE_COMPLETE,
ProcessingLog::STATUS_COMPLETED,
null,
['duration_seconds' => $duration, 'chunk_count' => count($savedChunks)]
);
Log::info("Dokumen version {$version->id} berjaya diproses dalam {$duration}s", [
'chunk_count' => count($savedChunks),
]);
}
/**
* Embed dan sync satu knowledge item ke Qdrant.
* Dipanggil selepas create/update knowledge item.
*/
public function processKnowledgeItem(KnowledgeItem $item): void
{
$text = $item->getEmbeddableText();
if (empty(trim($text))) {
throw new RuntimeException('Knowledge item tidak mempunyai kandungan untuk di-embed.');
}
// Jika ada qdrant_point_id lama, update
// Jika tiada, jana UUID baru
$pointId = $item->qdrant_point_id ?? (string) Str::uuid();
$vector = $this->ollama->embed($text);
$payload = $this->buildKnowledgeItemPayload($item);
$this->qdrant->ensureCollectionExists();
$this->qdrant->upsertPoint($pointId, $vector, $payload);
$item->markAsEmbedded($pointId);
Log::info("KnowledgeItem {$item->id} berjaya di-embed.", [
'type' => $item->item_type,
'category_id' => $item->category_id,
]);
}
/**
* Deactivate semua chunk dalam Qdrant untuk versi lama.
* Chunk dalam MySQL kekal — hanya is_active di Qdrant dikemaskini.
*/
public function deactivateVersionInQdrant(DocumentVersion $version): void
{
$chunks = $version->chunks()
->whereNotNull('qdrant_point_id')
->where('is_embedded', true)
->get();
if ($chunks->isEmpty()) {
return;
}
$pointIds = $chunks->pluck('qdrant_point_id')->toArray();
$this->qdrant->updatePayloadBatch($pointIds, [
'is_active' => false,
'status' => 'inactive',
]);
// Kemaskini MySQL juga
$version->chunks()->update(['is_active' => false]);
}
/**
* Deactivate knowledge item dalam Qdrant.
*/
public function deactivateKnowledgeItemInQdrant(KnowledgeItem $item): void
{
if ($item->qdrant_point_id) {
$this->qdrant->updatePayload($item->qdrant_point_id, [
'is_active' => false,
'status' => 'inactive',
]);
}
}
// =========================================================================
// PRIVATE HELPERS
// =========================================================================
/**
* Deactivate chunk dari versi sebelumnya.
*/
private function deactivatePreviousChunks(DocumentVersion $currentVersion): void
{
$previousVersions = DocumentVersion::where('document_id', $currentVersion->document_id)
->where('id', '!=', $currentVersion->id)
->where('processing_status', DocumentVersion::STATUS_INDEXED)
->get();
foreach ($previousVersions as $prev) {
$this->deactivateVersionInQdrant($prev);
// Tandakan versi lama bukan current lagi
$prev->update(['is_current' => false]);
}
}
/**
* Simpan semua chunk dalam MySQL.
*
* @return DocumentChunk[]
*/
private function saveChunks(DocumentVersion $version, array $chunks): array
{
$document = $version->document;
return DB::transaction(function () use ($version, $document, $chunks) {
$saved = [];
foreach ($chunks as $chunk) {
$saved[] = DocumentChunk::create([
'document_id' => $document->id,
'document_version_id' => $version->id,
'chunk_index' => $chunk['chunk_index'],
'page_number' => $chunk['page_number'] ?? null,
'content' => $chunk['content'],
'token_count' => $chunk['word_count'] ?? null,
'section_heading' => $chunk['section_heading'] ?? null,
'is_active' => true,
'is_embedded' => false,
]);
}
// Set versi ini sebagai current
$version->update(['is_current' => true]);
return $saved;
});
}
/**
* Jana embedding dan sync semua chunk ke Qdrant.
*/
private function embedAndSyncChunks(DocumentVersion $version, array $chunks): void
{
$document = $version->document;
$category = $document->category;
$this->qdrant->ensureCollectionExists();
$batchSize = 10; // Proses 10 chunk sekali untuk elak timeout Ollama
$chunkBatches = array_chunk($chunks, $batchSize);
foreach ($chunkBatches as $batch) {
$points = [];
foreach ($batch as $chunk) {
try {
// Guna getEmbeddableText() — final_text > cleaned_text > content
// Semasa ingestion pertama, final_text dan cleaned_text adalah null
// jadi ia akan fallback ke content (raw extraction)
$vector = $this->ollama->embed($chunk->getEmbeddableText());
$pointId = (string) Str::uuid();
$points[] = [
'id' => $pointId,
'vector' => $vector,
'payload' => $this->buildChunkPayload($chunk, $version, $document, $category),
];
$chunk->markAsEmbedded($pointId);
} catch (RuntimeException $e) {
Log::error("Gagal embed chunk {$chunk->id}", [
'error' => $e->getMessage(),
]);
throw $e;
}
}
if (!empty($points)) {
$this->qdrant->upsertPoints($points);
}
}
ProcessingLog::record(
DocumentVersion::class,
$version->id,
ProcessingLog::STAGE_QDRANT,
ProcessingLog::STATUS_COMPLETED,
null,
['synced_points' => count($chunks)]
);
}
/**
* Bina Qdrant payload untuk chunk PDF.
* Payload ini yang akan digunakan untuk filter dan display sumber.
*/
private function buildChunkPayload(
DocumentChunk $chunk,
DocumentVersion $version,
$document,
$category
): array {
return [
'knowledge_type' => 'pdf_chunk',
'source_type' => 'pdf',
'category_id' => $category->id,
'category_name' => $category->name,
'category_slug' => $category->slug,
'document_id' => $document->id,
'document_version_id' => $version->id,
'document_chunk_id' => $chunk->id,
'knowledge_item_id' => null,
'title' => $document->title,
'page_number' => $chunk->page_number,
'chunk_index' => $chunk->chunk_index,
'section_heading' => $chunk->section_heading,
'text' => mb_substr($chunk->getEmbeddableText(), 0, 1000),
// Excerpt teks yang di-embed (final_text > cleaned_text > content)
'is_active' => true,
'status' => 'active',
'tags' => $document->tags ?? [],
'effective_date' => $document->effective_date?->toDateString(),
'language' => $document->language,
'created_at' => now()->toIso8601String(),
];
}
/**
* Bina Qdrant payload untuk knowledge item (FAQ, polisi, dll.)
*/
private function buildKnowledgeItemPayload(KnowledgeItem $item): array
{
return [
'knowledge_type' => $item->item_type,
'source_type' => 'manual',
'category_id' => $item->category_id,
'category_name' => $item->category->name,
'category_slug' => $item->category->slug,
'document_id' => null,
'document_version_id' => null,
'document_chunk_id' => null,
'knowledge_item_id' => $item->id,
'title' => $item->title,
'page_number' => null,
'chunk_index' => 0,
'section_heading' => null,
'text' => mb_substr($item->getEmbeddableText(), 0, 1000),
'is_active' => $item->is_active,
'status' => $item->is_active ? 'active' : 'inactive',
'tags' => $item->tags ?? [],
'effective_date' => $item->effective_date?->toDateString(),
'language' => $item->language,
'created_at' => now()->toIso8601String(),
];
}
}

249
app/Services/KnowledgeBase/RAGService.php Normal file
View File

@@ -0,0 +1,249 @@
<?php
namespace App\Services\KnowledgeBase;
use App\Services\Ollama\OllamaService;
use App\Services\Qdrant\QdrantService;
use Illuminate\Support\Facades\Log;
use RuntimeException;
/**
* RAGService (Retrieval-Augmented Generation)
*
* Koordinasi proses RAG:
* 1. Jana embedding untuk soalan user
* 2. Cari context paling relevan dari Qdrant
* 3. Bina context string
* 4. Hantar ke Ollama untuk jawapan
* 5. Return jawapan + source references
*/
class RAGService
{
private int $maxContextChunks;
private int $maxContextWords;
public function __construct(
private readonly OllamaService $ollama,
private readonly QdrantService $qdrant,
) {
$this->maxContextChunks = config('knowledgebase.rag.max_context_chunks', 5);
$this->maxContextWords = config('knowledgebase.rag.max_context_words', 2000);
}
/**
* Jawab soalan menggunakan RAG.
*
* @param string $question Soalan pengguna
* @param ?int $categoryId Filter kategori (null = semua)
* @return array{
* answer: string,
* has_answer: bool,
* sources: array[],
* context_chunks: array[],
* model_used: string,
* tokens_used: ?int,
* response_time: float
* }
* @throws RuntimeException Jika Ollama atau Qdrant tidak tersedia
*/
public function ask(string $question, ?int $categoryId = null): array
{
$startTime = microtime(true);
// ── Langkah 1: Jana embedding untuk soalan ─────────────────────────
$queryVector = $this->ollama->embed($question);
// ── Langkah 2: Cari context relevan dari Qdrant ─────────────────────
$filter = $this->qdrant->buildFilter(
categoryId: $categoryId,
isActive: true,
);
$scoreThreshold = config('qdrant.search.score_threshold', 0.3);
$searchResults = $this->qdrant->searchSimilar(
vector: $queryVector,
limit: $this->maxContextChunks,
filter: $filter,
scoreThreshold: $scoreThreshold,
);
//log search result
\Log::info('Qdrant search raw results', [
'question' => $question,
'results' => $searchResults,
]);
\Log::info('Qdrant raw results', [
'scores' => array_map(fn($r) => $r['score'] ?? null, $searchResults),
]);
if (empty($searchResults)) {
$responseTime = round(microtime(true) - $startTime, 3);
return [
'answer' => config('ollama.rag_system_prompt_no_result',
'Maaf, saya tidak menemui maklumat berkaitan dalam pangkalan pengetahuan kami. ' .
'Sila hubungi pejabat kami untuk maklumat lanjut.'),
'has_answer' => false,
'sources' => [],
'context_chunks' => [],
'model_used' => config('ollama.chat_model'),
'tokens_used' => null,
'response_time' => $responseTime,
];
}
// ── Langkah 3: Bina context string ─────────────────────────────────
[$context, $contextChunksData] = $this->buildContext($searchResults);
// ── Langkah 4: Hantar ke Ollama ─────────────────────────────────────
$chatResult = $this->ollama->chat($question, $context);
// ── Langkah 5: Bina source references ──────────────────────────────
$sources = $this->buildSourceReferences($searchResults);
$responseTime = round(microtime(true) - $startTime, 3);
// Tentukan sama ada model ada jawapan atau tidak
$hasAnswer = $this->detectHasAnswer($chatResult['answer']);
return [
'answer' => $chatResult['answer'],
'has_answer' => $hasAnswer,
'sources' => $sources,
'context_chunks' => $contextChunksData,
'model_used' => $chatResult['model'],
'tokens_used' => $chatResult['tokens'],
'response_time' => $responseTime,
];
}
/**
* Bina context string dari search results.
* Had bilangan perkataan supaya tidak melebihi context window model.
*
* @return array{0: string, 1: array[]}
*/
private function buildContext(array $searchResults): array
{
$contextParts = [];
$chunksData = [];
$totalWords = 0;
foreach ($searchResults as $result) {
$payload = $result['payload'] ?? [];
$text = $payload['text'] ?? '';
if (empty($text)) {
continue;
}
$words = str_word_count($text);
if ($totalWords + $words > $this->maxContextWords) {
// Potong jika context dah terlalu panjang
if (empty($contextParts)) {
// Sekurang-kurangnya masukkan satu chunk
$contextParts[] = $text;
$chunksData[] = $this->extractChunkData($result);
}
break;
}
$source = $this->formatSourceLabel($payload);
$contextParts[] = "[Sumber: {$source}]\n{$text}";
$chunksData[] = $this->extractChunkData($result);
$totalWords += $words;
}
return [implode("\n\n---\n\n", $contextParts), $chunksData];
}
/**
* Bina array source references untuk paparan kepada pengguna.
*/
private function buildSourceReferences(array $searchResults): array
{
$sources = [];
$seen = []; // Elak duplikasi sumber yang sama
foreach ($searchResults as $result) {
$payload = $result['payload'] ?? [];
$sourceKey = ($payload['document_id'] ?? '') . '_' .
($payload['knowledge_item_id'] ?? '') . '_' .
($payload['page_number'] ?? '');
if (isset($seen[$sourceKey])) {
continue;
}
$seen[$sourceKey] = true;
$sources[] = [
'type' => $payload['source_type'] ?? 'unknown',
'knowledge_type' => $payload['knowledge_type'] ?? '',
'title' => $payload['title'] ?? 'Tiada tajuk',
'category' => $payload['category_name'] ?? '',
'category_id' => $payload['category_id'] ?? null,
'page_number' => $payload['page_number'] ?? null,
'section_heading' => $payload['section_heading'] ?? null,
'document_id' => $payload['document_id'] ?? null,
'knowledge_item_id' => $payload['knowledge_item_id'] ?? null,
'score' => round($result['score'] ?? 0, 4),
];
}
return $sources;
}
/**
* Extract data chunk untuk disimpan dalam chat_logs.
*/
private function extractChunkData(array $result): array
{
return [
'point_id' => $result['id'] ?? null,
'score' => round($result['score'] ?? 0, 4),
'title' => $result['payload']['title'] ?? '',
'category' => $result['payload']['category_name'] ?? '',
'source_type' => $result['payload']['source_type'] ?? '',
'page_number' => $result['payload']['page_number'] ?? null,
];
}
private function formatSourceLabel(array $payload): string
{
$title = $payload['title'] ?? 'Tanpa tajuk';
$page = isset($payload['page_number']) ? ", ms. {$payload['page_number']}" : '';
$category = $payload['category_name'] ?? '';
return "{$title}{$page} ({$category})";
}
/**
* Detect sama ada model sebenarnya ada jawapan atau tidak.
* Semak jika jawapan adalah "tidak tahu" / fallback.
*/
private function detectHasAnswer(string $answer): bool
{
$noAnswerPatterns = [
'tidak menemui',
'tiada maklumat',
'tidak terdapat dalam',
'sila hubungi',
'tidak dapat menjawab',
'maklumat tidak tersedia',
];
$answerLower = mb_strtolower($answer);
foreach ($noAnswerPatterns as $pattern) {
if (str_contains($answerLower, $pattern)) {
return false;
}
}
return !empty(trim($answer));
}
}