main

src/modules/content-generation/content-generation.service.ts

@@ -14,11 +14,13 @@ import { SeoService, FullSeoAnalysis as SeoDTO } from '../seo/seo.service';
 import { NeuroMarketingService } from '../neuro-marketing/neuro-marketing.service';
 import { StorageService } from '../visual-generation/services/storage.service';
 import { VisualGenerationService } from '../visual-generation/visual-generation.service';
+import { WebScraperService, ScrapedContent } from '../trends/services/web-scraper.service';
 import { ContentType as PrismaContentType, ContentStatus as PrismaContentStatus, MasterContentType as PrismaMasterContentType } from '@prisma/client';
 
 
 export interface ContentGenerationRequest {
     topic: string;
+    sourceUrl?: string;
     niche?: string;
     platforms: Platform[];
     includeResearch?: boolean;
@@ -76,6 +78,7 @@ export class ContentGenerationService {
         private readonly neuroService: NeuroMarketingService,
         private readonly storageService: StorageService,
         private readonly visualService: VisualGenerationService,
+        private readonly webScraperService: WebScraperService,
     ) { }
 
 
@@ -87,6 +90,7 @@ export class ContentGenerationService {
     async generateContent(request: ContentGenerationRequest): Promise<GeneratedContentBundle> {
         const {
             topic,
+            sourceUrl,
             niche,
             platforms,
             includeResearch = true,
@@ -99,6 +103,26 @@ export class ContentGenerationService {
 
         console.log(`[ContentGenerationService] Starting generation for topic: ${topic}, platforms: ${platforms.join(', ')}`);
 
+        // ========== STEP 1: Scrape source article if URL provided ==========
+        let scrapedSource: ScrapedContent | null = null;
+        if (sourceUrl) {
+            this.logger.log(`Scraping source article: ${sourceUrl}`);
+            try {
+                scrapedSource = await this.webScraperService.scrapeUrl(sourceUrl, {
+                    extractImages: true,
+                    extractLinks: true,
+                    timeout: 15000,
+                }, topic);
+                if (scrapedSource) {
+                    this.logger.log(`Scraped source: ${scrapedSource.wordCount} words, ${scrapedSource.images.length} images, ${scrapedSource.videoLinks.length} videos`);
+                } else {
+                    this.logger.warn(`Failed to scrape source URL: ${sourceUrl}`);
+                }
+            } catch (err) {
+                this.logger.warn(`Source scraping error: ${err.message}`);
+            }
+        }
+
         // Analyze niche if provided
         let nicheAnalysis: NicheAnalysis | undefined;
         if (niche) {
@@ -116,6 +140,23 @@ export class ContentGenerationService {
             });
         }
 
+        // ========== Build enriched context from scraped source ==========
+        let sourceContext = '';
+        if (scrapedSource) {
+            const articleText = scrapedSource.content.substring(0, 3000);
+            const videoInfo = scrapedSource.videoLinks.length > 0
+                ? `\nVİDEO LİNKLERİ: ${scrapedSource.videoLinks.join(', ')}`
+                : '';
+            const importantLinks = scrapedSource.links
+                .filter(l => l.isExternal && !l.href.includes('facebook') && !l.href.includes('twitter'))
+                .slice(0, 5)
+                .map(l => `${l.text}: ${l.href}`)
+                .join('\n');
+            const linkInfo = importantLinks ? `\nÖNEMLİ LİNKLER:\n${importantLinks}` : '';
+
+            sourceContext = `\n\n📰 KAYNAK MAKALE İÇERİĞİ (ZORUNLU REFERANS):\n${articleText}${videoInfo}${linkInfo}\n\n⚠️ ÖNEMLİ: Yukarıdaki kaynak makaledeki TÜM özneleri (kişi, ürün, oyun adları, tarihler, fiyatlar, markalar) habere dahil et. Hiçbir önemli bilgiyi atlama. Video linkleri ve önemli dış linkler varsa bunları da içerikte paylaş.`;
+        }
+
         // Generate content for each platform using AI
         const platformContent: GeneratedContent[] = [];
         for (const platform of platforms) {
@@ -127,11 +168,13 @@ export class ContentGenerationService {
                 const sanitizedSummary = this.sanitizeResearchSummary(
                     research?.summary || `Everything you need to know about ${topic}`
                 );
+                // Append scraped source context to give AI the full article details
+                const enrichedSummary = sanitizedSummary + sourceContext;
                 // Normalize platform to lowercase for consistency
                 const normalizedPlatform = platform.toLowerCase();
                 const aiContent = await this.platformService.generateAIContent(
                     topic,
-                    sanitizedSummary,
+                    enrichedSummary,
                     normalizedPlatform as any, // Cast to any/Platform to resolve type mismatch if Platform is strict union
                     'standard',
                     'tr',
@@ -145,6 +188,9 @@ export class ContentGenerationService {
                     this.logger.warn(`AI Content is empty for ${platform}`);
                 }
 
+                // Use scraped image from source if available
+                const sourceImageUrl = scrapedSource?.images?.[0]?.src || undefined;
+
                 const config = this.platformService.getPlatformConfig(platform);
                 let content: GeneratedContent = {
                     platform,
@@ -163,10 +209,19 @@ export class ContentGenerationService {
                     content.content = voiceApplied.branded;
                 }
 
-                // Add hashtags if requested
+                // Add hashtags using AI (based on actual generated content)
                 if (includeHashtags) {
-                    const hashtagSet = this.hashtagService.generateHashtags(topic, platform);
-                    content.hashtags = hashtagSet.hashtags.map((h) => h.hashtag);
+                    try {
+                        content.hashtags = await this.platformService.generateAIHashtags(
+                            content.content,
+                            topic,
+                            platform as any,
+                            'tr',
+                        );
+                    } catch (hashErr) {
+                        this.logger.warn(`AI hashtag generation failed, skipping: ${hashErr.message}`);
+                        content.hashtags = [];
+                    }
                 }
 
                 // Generate image for visual platforms
@@ -180,11 +235,31 @@ export class ContentGenerationService {
                             platform: platformKey,
                             enhancePrompt: true,
                         });
-                        content.imageUrl = image.url;
-                        this.logger.log(`Image generated for ${platform}: ${image.url}`);
+                        
+                        // Check if image is a real image or just a placeholder
+                        const isPlaceholder = image.url?.includes('placehold.co') || image.url?.includes('placeholder');
+                        if (!isPlaceholder) {
+                            content.imageUrl = image.url;
+                            this.logger.log(`Image generated for ${platform}: ${image.url}`);
+                        } else if (sourceImageUrl) {
+                            // Use scraped source image instead of placeholder
+                            content.imageUrl = sourceImageUrl;
+                            this.logger.log(`Using scraped source image instead of placeholder: ${sourceImageUrl}`);
+                        } else {
+                            content.imageUrl = image.url;
+                            this.logger.log(`Image generated for ${platform}: ${image.url} (placeholder, no source image available)`);
+                        }
                     } catch (imgError) {
                         this.logger.warn(`Image generation failed for ${platform}, continuing without image`, imgError);
+                        // Fallback to scraped source image
+                        if (sourceImageUrl) {
+                            content.imageUrl = sourceImageUrl;
+                            this.logger.log(`Using scraped source image as fallback: ${sourceImageUrl}`);
+                        }
                     }
+                } else if (sourceImageUrl && !content.imageUrl) {
+                    // For non-visual platforms, still attach source image if available
+                    content.imageUrl = sourceImageUrl;
                 }
 
                 platformContent.push(content);
@@ -358,7 +433,7 @@ export class ContentGenerationService {
                             userId: effectiveUserId!,
                             masterContentId: masterContent.id,
                             type: contentType,
-                            title: `${bundle.topic} - ${platformContent.platform}`,
+                            title: this.sanitizeResearchSummary(`${bundle.topic}`) + ` - ${platformContent.platform}`,
                             body: platformContent.content,
                             hashtags: platformContent.hashtags,
                             status: PrismaContentStatus.DRAFT,
@@ -548,6 +623,8 @@ KURALLAR:
 6. Karakter limitini koru
 7. Platformun tonuna uygun yaz
 8. SADECE yayınlanacak metni yaz
+9. Hiçbir haber sitesi, kaynak, ajans veya web sitesi adı kullanma
+10. "...göre", "...haberlere göre", "...kaynağına göre" gibi atıf ifadeleri ASLA kullanma
 
 SADECE yeniden yazılmış metni döndür, açıklama ekleme.`;
 
@@ -589,25 +666,43 @@ SADECE yeniden yazılmış metni döndür, açıklama ekleme.`;
         sanitized = sanitized.replace(/https?:\/\/[^\s]+/gi, '');
         sanitized = sanitized.replace(/www\.[^\s]+/gi, '');
 
-        // Remove common Turkish attribution phrases
+        // Remove common attribution phrases (Turkish and English)
         const attributionPatterns = [
             /\b\w+\.com(\.tr)?\b/gi,
             /\b\w+\.org(\.tr)?\b/gi,
             /\b\w+\.net(\.tr)?\b/gi,
             /\bkaynağına göre\b/gi,
             /\b'e göre\b/gi,
+            /\b'(i|a|e|u|ü|\u0131)n(da|de) (yayınlanan|yer alan|çıkan)\b/gi,
+            /\b(da|de) (çıkan|yayınlanan|yer alan) (haberlere|habere|bilgilere) göre\b/gi,
+            /\bhaberlere göre\b/gi,
+            /\braporuna göre\b/gi,
+            /\bsitesinde yer alan\b/gi,
+            /\baçıklamasına göre\b/gi,
+            /\byazısına göre\b/gi,
+            /\bhaberine göre\b/gi,
+            /\btarafından yapılan\b/gi,
             /\baccording to [^,.]+/gi,
+            /\breported by [^,.]+/gi,
+            /\bas reported in [^,.]+/gi,
             /\bsource:\s*[^,.]+/gi,
             /\breferans:\s*[^,.]+/gi,
             /\bkaynak:\s*[^,.]+/gi,
         ];
 
-        // Common Turkish tech/news source brands to strip
+        // Comprehensive list of Turkish tech/news source brands to strip
         const sourceNames = [
-            'donanımhaber', 'technopat', 'webtekno', 'shiftdelete',
+            'tamindir', 'donanımhaber', 'technopat', 'webtekno', 'shiftdelete',
             'chip online', 'log.com', 'mediatrend', 'bbc', 'cnn',
             'reuters', 'anadolu ajansı', 'hürriyet', 'milliyet',
             'sabah', 'forbes', 'bloomberg', 'techcrunch',
+            'the verge', 'engadget', 'ars technica', 'wired',
+            'mashable', 'gizmodo', 'tom\'s hardware', 'tom\'s guide',
+            'ntv', 'habertürk', 'sozcu', 'sözcü', 'cumhuriyet', 'star',
+            'posta', 'aksam', 'yeni safak', 'yeni şafak', 'takvim',
+            'mynet', 'ensonhaber', 'haber7', 'internethaber',
+            'ad hoc news', 'finanzen.net', 'der aktionär', 'aktionar',
+            'business insider', 'cnbc', 'financial times', 'wall street journal',
         ];
 
         for (const pattern of attributionPatterns) {
@@ -615,12 +710,15 @@ SADECE yeniden yazılmış metni döndür, açıklama ekleme.`;
         }
 
         for (const source of sourceNames) {
-            const regex = new RegExp(`\\b${source}\\b`, 'gi');
+            const regex = new RegExp(`\\b${source.replace(/[.*+?^${}()|[\]\\]/g, '\\$&')}\\b`, 'gi');
             sanitized = sanitized.replace(regex, '');
         }
 
-        // Clean up multiple spaces and trailing commas
-        sanitized = sanitized.replace(/\s{2,}/g, ' ').replace(/,\s*,/g, ',').trim();
+        // Also remove "- site_name" patterns from titles (e.g. "Great News - Tamindir")
+        sanitized = sanitized.replace(/\s*-\s*$/gm, '');
+
+        // Clean up multiple spaces, trailing commas, and orphaned punctuation
+        sanitized = sanitized.replace(/\s{2,}/g, ' ').replace(/,\s*,/g, ',').replace(/\s+([.,;:!?])/g, '$1').trim();
 
         return sanitized;
     }