import pool from '../db'; import { scanContent } from './compliance/scanner'; import { callLLM } from './grok'; import { editorialGuardrails, classifyNarrative, getBlogNarrativePrompt, NarrativeContext, NarrativeClassification } from './narrative-engine'; export interface BlogPostContent { title: string; metaDescription: string; content: string; excerpt: string; tags: string[]; } const DISALLOWED_TITLE_PATTERNS = [ /\bfortress\b/i, /\bskeleton crew\b/i, /\bskeleton\b/i, /\bghost roster\b/i, /\bghost lineup\b/i, /\bghost starter\b/i, /\bghost spread\b/i, /\bghost\b/i, /\bopening day ghosts\b/i, /\bphantom\b/i, /\bbuzzsaw\b/i, /\bwalking wounded\b/i, /\bwinds of war\b/i, /\bbroken rosters?\b/i, /\bbroken lineups?\b/i, /\bbroken bats?\b/i, /\bshell game\b/i, /\bdesert mirage\b/i, /\btrapdoor\b/i, /\bgraveyard\b/i, ]; const DUPLICATE_MATCHUP_WINDOW_MS = 3 * 60 * 60 * 1000; function cleanSlugPart(value: string): string { return value.toLowerCase().replace(/[^a-z0-9]+/g, '-').replace(/(^-|-$)/g, ''); } function normalizeTeamForDuplicateMatch(value: string): string { return value .normalize('NFD') .replace(/[\u0300-\u036f]/g, '') .toLowerCase() .replace(/&/g, ' and ') .replace(/[^a-z0-9 ]+/g, ' ') .replace(/\s+/g, ' ') .trim(); } function teamNamesLikelySame(a: string, b: string): boolean { const normA = normalizeTeamForDuplicateMatch(a); const normB = normalizeTeamForDuplicateMatch(b); if (!normA || !normB) return false; if (normA === normB) return true; const tokensA = normA.split(' '); const tokensB = normB.split(' '); const shorter = normA.length <= normB.length ? normA : normB; const longer = normA.length <= normB.length ? normB : normA; if ((tokensA.length === 1 || tokensB.length === 1) && longer.startsWith(`${shorter} `)) { return true; } return false; } function buildFallbackBlogTitle(context: { homeTeam: string; awayTeam: string; league: string; }): string { const templates = [ `Why ${context.awayTeam} at ${context.homeTeam} is tighter than it looks`, `${context.awayTeam} at ${context.homeTeam}: where this matchup tilts`, `${context.homeTeam} vs ${context.awayTeam}: the pressure points that matter`, `${context.awayTeam}-${context.homeTeam}: the details that could decide it`, ]; const seed = `${context.league}:${context.awayTeam}:${context.homeTeam}`; const index = Array.from(seed).reduce((sum, char) => sum + char.charCodeAt(0), 0) % templates.length; const title = templates[index]; return title.length <= 80 ? title : title.slice(0, 77).trimEnd() + '...'; } export function sanitizeGeneratedBlogTitle(title: string, context: { homeTeam: string; awayTeam: string; league: string; }): string { const cleaned = title.trim().replace(/\s+/g, ' '); if (!cleaned) return buildFallbackBlogTitle(context); if (DISALLOWED_TITLE_PATTERNS.some((pattern) => pattern.test(cleaned))) { return buildFallbackBlogTitle(context); } return cleaned; } export function blogPostsLikelyDuplicate(candidate: { sport: string; gameDate: string; homeTeam: string; awayTeam: string; startsAt: string; }, existing: { sport: string; game_date: string; home_team: string; away_team: string; starts_at: string | null; }): boolean { if (candidate.sport !== existing.sport || candidate.gameDate !== existing.game_date) { return false; } const directMatch = teamNamesLikelySame(candidate.homeTeam, existing.home_team) && teamNamesLikelySame(candidate.awayTeam, existing.away_team); const reversedMatch = teamNamesLikelySame(candidate.homeTeam, existing.away_team) && teamNamesLikelySame(candidate.awayTeam, existing.home_team); if (!directMatch && !reversedMatch) return false; if (!existing.starts_at) return true; const candidateTs = new Date(candidate.startsAt).getTime(); const existingTs = new Date(existing.starts_at).getTime(); if (Number.isNaN(candidateTs) || Number.isNaN(existingTs)) return true; return Math.abs(candidateTs - existingTs) <= DUPLICATE_MATCHUP_WINDOW_MS; } export function generateSlug(league: string, homeTeam: string, awayTeam: string, gameDate: string): string { return `${cleanSlugPart(league)}-${cleanSlugPart(awayTeam)}-vs-${cleanSlugPart(homeTeam)}-${gameDate}`; } export async function generateBlogPost(context: { league: string; homeTeam: string; awayTeam: string; startsAt: string; sanitizedSummary: string; keyFactors: string[]; spreadAnalysis: string; totalAnalysis: string; historicalTrend: string; injuryNotes: string; oddsData?: any; narrativeClassification?: NarrativeClassification; }): Promise { const oddsInfo = context.oddsData ? `Current markets: - Moneyline: Home ${context.oddsData.moneyline?.home ?? 'N/A'} | Away ${context.oddsData.moneyline?.away ?? 'N/A'} - Spread: Home ${context.oddsData.spread?.home?.line ?? 'N/A'} | Away ${context.oddsData.spread?.away?.line ?? 'N/A'} - Total: ${context.oddsData.total?.over?.line ?? 'N/A'}` : 'Current markets not available.'; const gameDate = new Date(context.startsAt).toLocaleDateString('en-US', { weekday: 'long', year: 'numeric', month: 'long', day: 'numeric' }); // Team logo URLs (ESPN CDN) const teamLogoUrl = (teamName: string, league: string) => { const slug = teamName.toLowerCase().replace(/[^a-z0-9]/g, '-').replace(/-+/g, '-'); return `https://a.espncdn.com/combiner/i?img=/i/teamlogos/${league.toLowerCase()}/500/${slug}.png&h=100&w=100`; }; const prompt = `You are generating a Rain Wire post for a sports matchup within the Rainmaker ecosystem. TERMINOLOGY: Blogs and Rain Wire are the same thing. MATCHUP: ${context.awayTeam} @ ${context.homeTeam} GAME DATE: ${gameDate} LEAGUE: ${context.league.toUpperCase()} BACKGROUND RESEARCH (weave into narrative naturally — do NOT quote directly, do NOT include hard numbers): - Summary: ${context.sanitizedSummary} - Key Factors: ${context.keyFactors.join('; ')} - Spread Context: ${context.spreadAnalysis} - Total Context: ${context.totalAnalysis} - Historical Trend: ${context.historicalTrend} - Injury Notes: ${context.injuryNotes || 'None reported'} PRIMARY OBJECTIVE: The sole objective of this Rain Wire post is to persuade the reader to click the Forecast link by delivering rich context, storytelling, and intrigue — WITHOUT revealing proprietary analysis, picks, or hard numbers. If the reader does not feel compelled to click the Forecast link at the end, the post has failed. VISUAL HEADER (REQUIRED — place at the very top):
${context.awayTeam}

${context.awayTeam}

VS
${context.homeTeam}

${context.homeTeam}

(Use real ESPN team logo URLs if possible. Format: https://a.espncdn.com/i/teamlogos/${context.league.toLowerCase()}/500/ABBREV.png) CONTENT REQUIREMENTS: - Minimum 250 words, target around 300-320 words. Start winding down at 320. Going slightly over 350 is fine but do NOT write novels. - Highly descriptive, visual, and narrative-driven - Written in a confident, authoritative tone - Must present an ORIGINAL angle on the matchup - Informative but INTENTIONALLY INCOMPLETE — hold back deeper insights for the Forecast RESEARCH TO WEAVE IN NATURALLY: - Matchup storylines and situational dynamics - Recent team form (last 5–10 games, trends, momentum) - Injury considerations and roster availability - Weather and environmental factors when relevant - Context that serious analysts would consider before forming an opinion ⚠️ DO NOT INCLUDE: - Exact statistics or projections - Explicit picks or recommendations - Hard numbers of any kind - Internal model name references or probabilities (never mention PIFF, DIGIMON, Grok — only Rain Man) LANGUAGE & TERMINOLOGY STANDARD (CRITICAL — extra weight): Ethos: "Smarter language for better existence." All copy must avoid explicit gambling language. Use market-based, analytical phrasing instead. NEVER USE: betting lines, betting markets, bettors, wagers, gambling. AVOID: sportsbooks (use "market" or "market venues" instead), odds (use "market price" or "market pricing"), betting (use "investing", "market interest", "positioning") USE INSTEAD: - "markets" or "current market" instead of "betting markets" - "market consensus" instead of "betting consensus" - "market positioning" or "market movement" instead of "line movement" - "early market interest" or "speculative interest" instead of "early action" - "sharp interest" or "informed interest" instead of "sharp money" - "pricing behavior" instead of "odds movement" - "consensus signals" instead of "betting signals" - "bookmakers" (sparingly, only when necessary) TONE: Treat markets as signals, not wagers. Frame activity as analysis of collective behavior, not gambling. Maintain a calm, authoritative, institutional voice. Assume the reader understands the underlying mechanics without naming them explicitly. GOAL: Elevate the brand into sports intelligence and forecasting. Keep content compliant, professional, and broadly distributable across media, search, and AI platforms. NARRATIVE STYLE: - You are a skeptical market analyst, not a storyteller. Question the number, not the narrative. - Every headline MUST reference a specific market detail: the spread, total, a key stat, or a situational factor that changes the price. Headlines without a concrete angle get rejected. - Name the edge: is the market too high, too low, ignoring an injury, overreacting to a streak? - Avoid dramatic metaphors, war imagery, and horror-movie language. No "fortress", "skeleton crew", "ghost roster", "buzzsaw", "phantom", "graveyard", "trapdoor", "walking wounded", "winds of war", "broken rosters", "desert mirage", or any similar cliché. - Write titles a sharp reader would click — someone who wants to know WHY a number is wrong, not who is going to win. - Title formula: [Situation or factor] + [market price or stat] + [skeptical take on the mispricing] - Examples of GOOD titles: "Memphis Missing 3 Starters but PHX -11.5 Still Looks Overpriced", "BOS/HOU Opens -125 With Whitlock on Short Rest — Market Isn't Pricing Fatigue", "Avalanche 9-2 Rout Masks a Makar Injury That Shifts the Series Price" - Examples of BAD titles: "Ghost Roster Walks Into a Buzzsaw", "The Fortress Has a Trapdoor", "Skeleton Crew vs. Star Power" - Open with the sharpest analytical angle, not atmospheric drama - Body should show the reasoning: matchup factors, rest advantages, injury context, pace mismatches, travel, weather — the things that move numbers CONTEST COVERAGE & FUNNEL STRUCTURE: - This is exactly ONE Rain Wire post for ONE contest - Each Rain Wire corresponds to a single matching Forecast for the same contest - The Rain Wire is the awareness + intrigue layer; the Forecast is the proof + reasoning layer - NEVER use the word "prediction(s)" — always call it a "Forecast" SUGGESTIVE CTA STANDARD: - End with language that implies we have a strong directional read on the matchup WITHOUT stating a pick or using hard numbers - Position the Forecast as where readers find the math, logic, and model-driven reasoning behind the edge - Be confidence-coded and suggestive — allude to there being a strong read on the likely outcome - Withhold the "why" — direct users to the Forecast where analysis, logic, and model reasoning are shown FORECAST CTA (REQUIRED — end every post with this):

🌧️ Want the Full Forecast?

There are subtle edges and hidden value in this matchup that only deeper analysis reveals. The surface doesn't tell the full story.

View Full Forecast →
SEO & DISTRIBUTION: - SEO-optimized with clean heading hierarchy (h2 → h3) - Semantic HTML:

,

,

,