Backup #2074

Dateiinhalt

<?php

declare(strict_types=1);

namespace UseCases\Chat;

// @responsibility: Orchestriert Chat-Nachrichten mit SSE-Progress-Events

use Domain\Constants;
use Domain\Repository\ChatMessageRepositoryInterface;
use Domain\Repository\ChatSessionRepositoryInterface;
use Infrastructure\AI\ClaudeService;
use Infrastructure\AI\ContentQualityValidator;
use Infrastructure\AI\OllamaService;
use Infrastructure\AI\QdrantService;
use Infrastructure\AI\ScoringService;

class StreamingChatMessageUseCase
{
    /** @var callable|null */
    private $progressCallback;

    private float $stepStart = 0.0;

    public function __construct(
        private OllamaService $ollama,
        private QdrantService $qdrant,
        private ClaudeService $claude,
        ScoringService $scoring,
        private ChatSessionRepositoryInterface $sessionRepo,
        private ChatMessageRepositoryInterface $messageRepo,
        private ChatPromptLoader $promptLoader,
        private ContentQualityValidator $qualityValidator,
        private RagContextBuilder $ragBuilder
    ) {
        // $scoring kept for BC but no longer used - ContentSearchService handles reranking
        unset($scoring);
    }

    /**
     * Set progress callback for SSE events
     *
     * @param callable $callback fn(string $step, string $message, ?int $durationMs): void
     */
    public function setProgressCallback(callable $callback): void
    {
        $this->progressCallback = $callback;
    }

    private function emit(string $step, string $message, ?int $durationMs = null): void
    {
        if ($this->progressCallback !== null) {
            ($this->progressCallback)($step, $message, $durationMs);
        }
    }

    private function startStep(): void
    {
        $this->stepStart = microtime(true);
    }

    private function endStep(string $step, string $message): void
    {
        $durationMs = (int) round((microtime(true) - $this->stepStart) * Constants::MS_PER_SECOND);
        $this->emit($step, $message, $durationMs);
    }

    /** Execute chat with streaming progress. @param array<string> $collections */
    public function execute(
        string $sessionUuid, string $message, string $model, array $collections = ['documents'],
        int $contextLimit = 5, int $authorProfileId = 0, int $systemPromptId = 1,
        float $temperature = 0.7, int $maxTokens = 4096, int $structureId = 0, bool $qualityCheck = false
    ): ChatResponse {
        $totalStart = microtime(true);
        // Step 1: Validate session
        $this->emit('session', 'Session validieren...');
        $this->startStep();
        $session = $this->sessionRepo->findByUuid($sessionUuid);
        if ($session === null) {
            $this->emit('error', 'Session nicht gefunden');
            return ChatResponse::error('Session nicht gefunden.');
        }
        $sessionId = $session->getId() ?? 0;
        $this->endStep('session_done', 'Session validiert');
        // Step 2: Validate message
        $message = trim($message);
        if ($message === '') {
            $this->emit('error', 'Keine Nachricht');
            return ChatResponse::error('Bitte gib eine Frage ein.');
        }
        // Step 3: Save user message
        $this->emit('save_user', 'User-Nachricht speichern...');
        $this->startStep();
        $this->messageRepo->save(sessionId: $sessionId, role: 'user', content: $message, model: $model);
        $this->endStep('save_user_done', 'User-Nachricht gespeichert');
        // Step 4: Auto-set title
        $currentTitle = $session->getTitle();
        if ($currentTitle === null || $currentTitle === 'Neuer Chat') {
            $this->sessionRepo->updateTitle($sessionId, mb_substr($message, 0, 50) . (mb_strlen($message) > 50 ? '...' : ''));
        }
        // Step 5: Get prompts
        $this->emit('prompts', 'Prompts laden...');
        $this->startStep();
        $stylePrompt = $this->getStylePromptFromProfile($authorProfileId);
        $systemPrompt = $this->getSystemPromptById($systemPromptId);
        $structurePrompt = $this->getStructurePrompt($structureId);
        if ($structurePrompt !== null) { $systemPrompt = ($systemPrompt ?? '') . "\n\n" . $structurePrompt; }
        $this->endStep('prompts_done', 'Prompts geladen');
        // RAG Pipeline
        $searchResults = [];
        $context = '';
        if ($collections !== []) {
            // Step 6+7: Semantic search via ContentSearchService
            $this->emit('search', 'Semantische Suche in ' . count($collections) . ' Collection(s)...');
            $this->startStep();
            $searchResults = $this->searchWithSemantics($message, $collections, $contextLimit);
            $semanticCount = count(array_filter($searchResults, static fn ($r) => isset($r['intent'])));
            $this->endStep('search_done', count($searchResults) . ' Chunks gefunden (' . $semanticCount . ' mit Semantik)');
            // Step 8: Build context with semantic metadata
            if ($searchResults !== []) {
                $this->emit('context', 'Kontext aufbauen...');
                $this->startStep();
                $context = $this->buildSemanticContext($searchResults);
                $this->endStep('context_done', 'Kontext erstellt (' . strlen($context) . ' Zeichen)');
            }
        }

        // Step 9: LLM Request
        $isOllama = str_starts_with($model, 'ollama:');
        $isClaude = str_starts_with($model, 'claude-');
        $hasContext = $context !== '';
        $this->emit('llm', 'Anfrage an ' . ($isOllama ? substr($model, 7) : $model) . '...');
        $this->startStep();
        $llmStart = microtime(true);
        try {
            if ($isClaude) {
                $userPrompt = $hasContext ? $this->claude->buildRagPrompt($message, $context) : $message;
                $effectiveSystemPrompt = $systemPrompt ?? ($hasContext ? $this->claude->getDefaultSystemPrompt() : 'Du bist ein hilfreicher Assistent. Antworte auf Deutsch, präzise und hilfreich.');
                if ($stylePrompt !== null && $stylePrompt !== '') { $effectiveSystemPrompt .= "\n\n" . $stylePrompt; }
                $llmResponse = $this->claude->ask($userPrompt, $effectiveSystemPrompt, $model, $maxTokens, $temperature);
                $answer = $llmResponse['text'];
                $usage = $llmResponse['usage'];
            } elseif ($isOllama) {
                $ollamaModel = substr($model, 7);
                $instructions = array_filter([$systemPrompt, $stylePrompt]);
                $instructionBlock = $instructions !== [] ? implode("\n\n", $instructions) . "\n\n" : '';
                $userPrompt = $hasContext ? sprintf("%sKontext:\n\n%s\n\n---\n\nFrage: %s", $instructionBlock, $context, $message) : $instructionBlock . $message;
                $answer = $this->ollama->generate($userPrompt, $ollamaModel, $temperature);
                $usage = null;
            } else {
                $this->emit('error', "Unbekanntes Modell: {$model}");
                return ChatResponse::error("Unknown model \"{$model}\".");
            }
        } catch (\RuntimeException $e) {
            $this->emit('error', 'LLM-Fehler: ' . $e->getMessage());
            return ChatResponse::error('LLM request failed: ' . $e->getMessage());
        }
        $llmDuration = (int) round((microtime(true) - $llmStart) * Constants::MS_PER_SECOND);
        $tokenInfo = $usage !== null ? " ({$usage['input_tokens']} in / {$usage['output_tokens']} out)" : '';
        $this->emit('llm_done', "Antwort erhalten{$tokenInfo}", $llmDuration);
        // Step 10: Extract sources
        $this->emit('sources', 'Quellen extrahieren...');
        $this->startStep();
        $sources = $this->extractSources($searchResults);
        $this->endStep('sources_done', count($sources) . ' Quellen extrahiert');
        // Step 11: Save assistant message
        $this->emit('save_assistant', 'Antwort speichern...');
        $this->startStep();
        $sourcesForStorage = array_map(static fn (array $s): string => json_encode($s, JSON_THROW_ON_ERROR), $sources);
        $this->messageRepo->save(
            sessionId: $sessionId, role: 'assistant', content: $answer, model: $model,
            tokensInput: $usage['input_tokens'] ?? null, tokensOutput: $usage['output_tokens'] ?? null,
            sources: $sourcesForStorage, startMicrotime: $llmStart, endMicrotime: microtime(true),
            authorProfileId: $authorProfileId > 0 ? $authorProfileId : null,
            systemPromptId: $systemPromptId > 0 ? $systemPromptId : null,
            collectionsJson: json_encode($collections), contextLimit: $contextLimit
        );
        $this->endStep('save_assistant_done', 'Antwort gespeichert');

        // Step 12: Quality check
        $qualityValidation = null;
        if ($qualityCheck) {
            $this->emit('quality', 'Qualitätsprüfung...');
            $this->startStep();
            $structureName = $structureId > 0 ? $this->getStructureName($structureId) : null;
            $qualityValidation = $this->qualityValidator->validate(
                question: $message,
                answer: $answer,
                sources: $sources,
                structureName: $structureName
            );
            $score = round($qualityValidation['score']);
            $this->endStep('quality_done', "Qualität: {$score}%");
        }

        // Final
        $totalDuration = (int) round((microtime(true) - $totalStart) * Constants::MS_PER_SECOND);
        $this->emit('complete', "Fertig in {$totalDuration}ms", $totalDuration);

        // Build response
        $result = [
            'answer' => $answer,
            'sources' => $sources,
            'usage' => $usage,
            'chunks_used' => count($searchResults),
        ];

        $response = ChatResponse::fromServiceResponse($result, (microtime(true) - $totalStart));
        if ($qualityValidation !== null) {
            $response = $response->withQualityValidation($qualityValidation);
        }

        return $response;
    }

    /** @param array<string> $collections */
    private function searchWithSemantics(string $query, array $collections, int $limit): array
    {
        $allResults = [];

        foreach ($collections as $collection) {
            if ($collection === 'documents') {
                // Use ContentSearchService for semantic-enriched search
                $semanticResults = $this->searchService->search($query, [], $limit);
                foreach ($semanticResults as $result) {
                    $allResults[] = [
                        'chunk_id' => $result['chunk_id'],
                        'content' => $result['content'],
                        'title' => $result['source_path'] ?? $result['heading_path'] ?? 'Unbekannt',
                        'score' => $result['relevance_score'],
                        // Semantic metadata
                        'summary' => $result['summary'] ?? null,
                        'keywords' => $result['keywords'] ?? [],
                        'intent' => $result['intent'] ?? null,
                        'discourse_role' => $result['discourse_role'] ?? null,
                        'sentiment' => $result['sentiment'] ?? null,
                        'frame' => $result['frame'] ?? null,
                        '_collection' => 'documents',
                    ];
                }
            } else {
                // Fallback to QdrantService for other collections
                try {
                    $embedding = $this->ollama->getEmbedding($query);
                    $results = $this->qdrant->search($embedding, $collection, $limit);
                    foreach ($results as $result) {
                        $payload = $result['payload'];
                        $allResults[] = [
                            'chunk_id' => $payload['chunk_id'] ?? 0,
                            'content' => $payload['content'] ?? $payload['content_preview'] ?? '',
                            'title' => $payload['document_title'] ?? $payload['title'] ?? 'Unbekannt',
                            'score' => $result['score'],
                            '_collection' => $collection,
                        ];
                    }
                } catch (\RuntimeException) {
                    continue;
                }
            }
        }

        // Sort by score and limit
        usort($allResults, static fn ($a, $b) => ($b['score'] ?? 0.0) <=> ($a['score'] ?? 0.0));

        return array_slice($allResults, 0, $limit);
    }

    private function buildSemanticContext(array $searchResults, int $maxTokens = 3000): string
    {
        $contextParts = [];
        $totalChars = 0;
        $maxChars = $maxTokens * 4;

        foreach ($searchResults as $index => $result) {
            $content = (string) ($result['content'] ?? '');
            $title = (string) ($result['title'] ?? 'Unbekannt');

            if ($totalChars + strlen($content) > $maxChars) {
                break;
            }

            // Build semantic header if available
            $semanticInfo = [];
            if (isset($result['discourse_role'])) {
                $semanticInfo[] = match ($result['discourse_role']) {
                    'definition' => 'Definition',
                    'thesis' => 'These',
                    'evidence' => 'Beleg',
                    'example' => 'Beispiel',
                    'summary' => 'Zusammenfassung',
                    default => $result['discourse_role'],
                };
            }
            if (isset($result['intent'])) {
                $semanticInfo[] = match ($result['intent']) {
                    'explain' => 'erklärt',
                    'define' => 'definiert',
                    'argue' => 'argumentiert',
                    'compare' => 'vergleicht',
                    'exemplify' => 'veranschaulicht',
                    default => $result['intent'],
                };
            }

            $header = sprintf('[Quelle %d: %s]', $index + 1, $title);
            if ($semanticInfo !== []) {
                $header .= ' (' . implode(', ', $semanticInfo) . ')';
            }

            $contextParts[] = $header . "\n" . $content;
            $totalChars += strlen($content);
        }

        return implode("\n\n---\n\n", $contextParts);
    }

    private function extractSources(array $searchResults): array
    {
        $sources = [];
        $seen = [];

        foreach ($searchResults as $result) {
            $title = (string) ($result['title'] ?? '');
            if ($title === '' || isset($seen[$title])) {
                continue;
            }

            $source = [
                'title' => $title,
                'score' => round((float) ($result['score'] ?? 0), 3),
            ];

            // Add content preview
            $content = $result['content'] ?? null;
            if (is_string($content) && $content !== '') {
                $source['content'] = mb_substr($content, 0, 300);
            }

            // Add collection
            if (isset($result['_collection'])) {
                $source['collection'] = $result['_collection'];
            }

            // Add semantic metadata if available
            if (isset($result['discourse_role'])) {
                $source['discourse_role'] = $result['discourse_role'];
            }
            if (isset($result['intent'])) {
                $source['intent'] = $result['intent'];
            }
            if (isset($result['summary'])) {
                $source['summary'] = $result['summary'];
            }

            $sources[] = $source;
            $seen[$title] = true;
        }

        return $sources;
    }

    private function getStylePromptFromProfile(int $profileId): ?string
    {
        if ($profileId === 0) { return null; }
        $profile = $this->configRepo->findByIdAndType($profileId, 'author_profile');
        if ($profile === null) { return null; }
        $config = json_decode($profile['content'] ?? '{}', true);
        if ($config === null) { return null; }
        $parts = [];
        if (isset($config['stimme']['ton'])) { $parts[] = 'Ton: ' . $config['stimme']['ton']; }
        if (isset($config['stimme']['perspektive'])) { $parts[] = 'Perspektive: ' . $config['stimme']['perspektive']; }
        if (isset($config['stil']['fachsprache']) && $config['stil']['fachsprache']) { $parts[] = 'Verwende Fachsprache'; }
        if (isset($config['stil']['beispiele']) && $config['stil']['beispiele'] === 'häufig') { $parts[] = 'Nutze häufig Beispiele'; }
        if (isset($config['stil']['listen']) && $config['stil']['listen'] === 'bevorzugt') { $parts[] = 'Bevorzuge Listen und Bullet-Points'; }
        if (isset($config['tabus']) && is_array($config['tabus'])) { $parts[] = 'Vermeide: ' . implode(', ', $config['tabus']); }
        if ($parts === []) { return null; }
        return 'Schreibstil (' . ($profile['name'] ?? 'Profil') . '): ' . implode('. ', $parts) . '.';
    }

    private function getSystemPromptById(int $promptId): ?string
    {
        if ($promptId === 0) { return null; }
        $prompt = $this->configRepo->findByIdAndType($promptId, 'system_prompt');
        if ($prompt === null) { return null; }
        $content = json_decode($prompt['content'] ?? '{}', true);
        return $content['prompt'] ?? null;
    }

    private function getStructurePrompt(int $structureId): ?string
    {
        if ($structureId === 0) { return null; }
        $structure = $this->configRepo->findByIdAndType($structureId, 'structure');
        if ($structure === null) { return null; }
        $name = $structure['name'] ?? 'Struktur';
        $config = json_decode($structure['content'] ?? '{}', true);
        if ($config === null) { return null; }
        $parts = ["Formatiere deine Antwort als: {$name}"];
        if (isset($config['sections']) && is_array($config['sections'])) { $parts[] = 'Struktur: ' . implode(' → ', $config['sections']); }
        if (isset($config['max_chars'])) { $parts[] = 'Maximale Länge: ' . $config['max_chars'] . ' Zeichen'; }
        if (isset($config['min_words'])) { $parts[] = 'Mindestens ' . $config['min_words'] . ' Wörter'; }
        if (isset($config['max_words'])) { $parts[] = 'Maximal ' . $config['max_words'] . ' Wörter'; }
        if (isset($config['format']) && $config['format'] === 'qa') { $parts[] = 'FAQ mit ' . ($config['min_questions'] ?? 3) . ' Paaren'; }
        if (isset($config['hashtags']) && $config['hashtags']) { $parts[] = 'Füge passende Hashtags hinzu'; }
        if (isset($config['cta']) && $config['cta']) { $parts[] = 'Schließe mit einem Call-to-Action ab'; }
        return implode('. ', $parts) . '.';
    }

    private function getStructureName(int $structureId): ?string
    {
        $structure = $this->configRepo->findByIdAndType($structureId, 'structure');
        return $structure['name'] ?? null;
    }
}

Vollständig herunterladen

Aktionen

Andere Versionen dieser Datei

← Zurück zur Übersicht