614 lines
31 KiB
TypeScript
614 lines
31 KiB
TypeScript
import { CheerioCrawler, RequestQueue } from 'crawlee';
|
|
import * as path from 'node:path';
|
|
import * as fs from 'node:fs/promises';
|
|
import { existsSync } from 'node:fs';
|
|
import { URL } from 'node:url';
|
|
import { execSync } from 'node:child_process';
|
|
import axios from 'axios';
|
|
import { FileCacheAdapter } from '../src/utils/cache/file-adapter.js';
|
|
|
|
import { initialState } from '../src/logic/pricing/constants.js';
|
|
|
|
async function main() {
|
|
const OPENROUTER_KEY = process.env.OPENROUTER_KEY;
|
|
if (!OPENROUTER_KEY) {
|
|
console.error('❌ Error: OPENROUTER_KEY not found in environment.');
|
|
process.exit(1);
|
|
}
|
|
|
|
let briefing = '';
|
|
let targetUrl: string | null = null;
|
|
let comments: string | null = null;
|
|
let cacheKey: string | null = null;
|
|
|
|
let jsonStatePath: string | null = null;
|
|
|
|
const args = process.argv.slice(2);
|
|
for (let i = 0; i < args.length; i++) {
|
|
const arg = args[i];
|
|
if (arg === '--url') {
|
|
targetUrl = args[++i];
|
|
} else if (arg === '--comments' || arg === '--notes') {
|
|
comments = args[++i];
|
|
} else if (arg === '--cache-key') {
|
|
cacheKey = args[++i];
|
|
} else if (arg === '--json') {
|
|
jsonStatePath = args[++i];
|
|
} else if (!arg.startsWith('--')) {
|
|
briefing = arg;
|
|
}
|
|
}
|
|
|
|
if (briefing && briefing.startsWith('@')) {
|
|
const rawPath = briefing.substring(1);
|
|
const filePath = rawPath.startsWith('/') ? rawPath : path.resolve(process.cwd(), rawPath);
|
|
briefing = await fs.readFile(filePath, 'utf8');
|
|
}
|
|
|
|
// Discovery ONLY if not provided
|
|
if (!targetUrl && briefing) {
|
|
const urlMatch = briefing.match(/https?:\/\/[^\s]+/);
|
|
if (urlMatch) {
|
|
targetUrl = urlMatch[0];
|
|
console.log(`🔗 Discovered URL in briefing: ${targetUrl}`);
|
|
}
|
|
}
|
|
|
|
if (!briefing && !targetUrl && !comments && !jsonStatePath) {
|
|
console.error('❌ Usage: npm run ai-estimate -- "Briefing text" [--url https://example.com] [--comments "Manual notes"]');
|
|
console.error(' Or: npm run ai-estimate -- @briefing.txt [--url https://example.com]');
|
|
console.error(' Or: npm run ai-estimate -- --json path/to/state.json');
|
|
process.exit(1);
|
|
}
|
|
|
|
const clearCache = process.argv.includes('--clear-cache');
|
|
if (clearCache) {
|
|
console.log('🧹 Clearing cache...');
|
|
const cacheFiles = await fs.readdir(path.join(process.cwd(), '.cache'));
|
|
for (const file of cacheFiles) {
|
|
if (file.startsWith('ai_est_')) {
|
|
await fs.unlink(path.join(process.cwd(), '.cache', file));
|
|
}
|
|
}
|
|
}
|
|
|
|
const cache = new FileCacheAdapter({ prefix: 'ai_est_' });
|
|
const finalCacheKey = cacheKey || `${briefing}_${targetUrl}_${comments}`;
|
|
|
|
// 1. Crawl if URL provided
|
|
let crawlContext = '';
|
|
if (targetUrl) {
|
|
console.log(`🔍 Crawling ${targetUrl} for context...`);
|
|
const cachedCrawl = await cache.get<string>(`crawl_${targetUrl}`);
|
|
if (cachedCrawl && !clearCache) {
|
|
console.log('📦 Using cached crawl results.');
|
|
crawlContext = cachedCrawl;
|
|
} else {
|
|
crawlContext = await performCrawl(targetUrl);
|
|
await cache.set(`crawl_${targetUrl}`, crawlContext, 86400); // 24h cache
|
|
}
|
|
}
|
|
|
|
// 2. Distill Crawl Context (Context Filtering)
|
|
let distilledCrawl = '';
|
|
if (crawlContext) {
|
|
const cachedDistilled = await cache.get<string>(`distilled_${targetUrl}`);
|
|
if (cachedDistilled && !clearCache) {
|
|
distilledCrawl = cachedDistilled;
|
|
} else {
|
|
distilledCrawl = await distillCrawlContext(crawlContext, OPENROUTER_KEY);
|
|
await cache.set(`distilled_${targetUrl}`, distilledCrawl, 86400);
|
|
}
|
|
}
|
|
|
|
// 3. AI Prompting
|
|
console.log('🤖 Consultating Gemini 3 Flash...');
|
|
const cachedAi = !clearCache ? await cache.get<any>(finalCacheKey) : null;
|
|
let formState: any;
|
|
let usage: { prompt: number, completion: number, cost: number } = { prompt: 0, completion: 0, cost: 0 };
|
|
|
|
// Load Context Documents
|
|
const principles = await fs.readFile(path.resolve(process.cwd(), 'docs/PRINCIPLES.md'), 'utf8');
|
|
const techStandards = await fs.readFile(path.resolve(process.cwd(), 'docs/TECH.md'), 'utf8');
|
|
const tone = await fs.readFile(path.resolve(process.cwd(), 'docs/TONE.md'), 'utf8');
|
|
|
|
if (jsonStatePath) {
|
|
console.log(`📂 Loading state from JSON: ${jsonStatePath}`);
|
|
const rawJson = await fs.readFile(path.resolve(process.cwd(), jsonStatePath), 'utf8');
|
|
formState = JSON.parse(rawJson);
|
|
} else if (cachedAi) {
|
|
console.log('📦 Using cached AI response.');
|
|
formState = cachedAi;
|
|
} else {
|
|
const result = await getAiEstimation(briefing, distilledCrawl, comments, OPENROUTER_KEY, principles, techStandards, tone);
|
|
formState = result.state;
|
|
usage = result.usage;
|
|
await cache.set(finalCacheKey, formState);
|
|
}
|
|
|
|
// 3. Save Data & Generate PDF
|
|
const timestamp = new Date().toISOString().replace(/[:.]/g, '-');
|
|
const jsonOutDir = path.resolve(process.cwd(), 'out/estimations/json');
|
|
if (!existsSync(jsonOutDir)) await fs.mkdir(jsonOutDir, { recursive: true });
|
|
|
|
const finalJsonPath = path.join(jsonOutDir, `${formState.companyName || 'unknown'}_${timestamp}.json`);
|
|
await fs.writeFile(finalJsonPath, JSON.stringify(formState, null, 2));
|
|
|
|
const tempJsonPath = path.resolve(process.cwd(), '.cache', `temp_state_${Date.now()}.json`);
|
|
await fs.writeFile(tempJsonPath, JSON.stringify(formState, null, 2));
|
|
|
|
console.log(`📦 Saved detailed state to: ${finalJsonPath}`);
|
|
console.log('📄 Generating PDF estimation...');
|
|
try {
|
|
execSync(`npx tsx ./scripts/generate-quote.ts --input ${tempJsonPath}`, { stdio: 'inherit' });
|
|
} finally {
|
|
// await fs.unlink(tempJsonPath);
|
|
}
|
|
|
|
console.log('\n✨ AI Estimation Complete!');
|
|
if (usage.prompt > 0) {
|
|
console.log('--------------------------------------------------');
|
|
console.log('📊 ACCUMULATED API USAGE (SUM OF 6 PASSES)');
|
|
console.log(` Model: google/gemini-3-flash-preview`);
|
|
console.log(` Total Prompt: ${usage.prompt.toLocaleString()}`);
|
|
console.log(` Total Completion: ${usage.completion.toLocaleString()}`);
|
|
console.log(` Total Tokens: ${(usage.prompt + usage.completion).toLocaleString()}`);
|
|
console.log(` Total Cost (USD): $${usage.cost.toFixed(6)}`);
|
|
console.log('--------------------------------------------------\n');
|
|
}
|
|
}
|
|
|
|
async function distillCrawlContext(rawCrawl: string, apiKey: string): Promise<string> {
|
|
if (!rawCrawl || rawCrawl.trim().length === 0) return "Keine Crawl-Daten vorhanden.";
|
|
|
|
console.log(' ↳ Distilling Crawl Context (Noise Filtering)...');
|
|
const systemPrompt = `
|
|
You are a context distiller. Your goal is to strip away HTML noise, legal footers, and generic fluff from a website crawl.
|
|
Extract the "Company DNA" in 5-8 bullet points (GERMAN).
|
|
|
|
### FOCUS ON:
|
|
1. Core Business / Services.
|
|
2. Unique Selling Points (USPs).
|
|
3. Target Audience (if clear).
|
|
4. Tech Stack or industry-specific equipment mentioned.
|
|
5. Brand tone (e.g. "industrial", "friendly", "technical").
|
|
|
|
### OUTPUT:
|
|
Return ONLY the bullet points. No intro/outro.
|
|
`;
|
|
const resp = await axios.post('https://openrouter.ai/api/v1/chat/completions', {
|
|
model: 'google/gemini-3-flash-preview',
|
|
messages: [{ role: 'system', content: systemPrompt }, { role: 'user', content: `RAW_CRAWL:\n${rawCrawl.substring(0, 30000)}` }],
|
|
}, { headers: { 'Authorization': `Bearer ${apiKey}`, 'Content-Type': 'application/json' } });
|
|
|
|
return resp.data.choices[0].message.content;
|
|
}
|
|
|
|
async function performCrawl(url: string): Promise<string> {
|
|
const pages: { url: string, content: string, type: string }[] = [];
|
|
const origin = new URL(url).origin;
|
|
|
|
const crawler = new CheerioCrawler({
|
|
maxRequestsPerCrawl: 20,
|
|
async requestHandler({ $, request, enqueueLinks }) {
|
|
const title = $('title').text();
|
|
const urlObj = new URL(request.url);
|
|
const urlPath = urlObj.pathname.toLowerCase();
|
|
|
|
let type = 'other';
|
|
if (urlPath === '/' || urlPath === '') type = 'home';
|
|
else if (urlPath.includes('service') || urlPath.includes('leistung')) type = 'service';
|
|
else if (urlPath.includes('blog') || urlPath.includes('news') || urlPath.includes('aktuelles') || urlPath.includes('magazin')) type = 'blog';
|
|
else if (urlPath.includes('contact') || urlPath.includes('kontakt')) type = 'contact';
|
|
else if (urlPath.includes('job') || urlPath.includes('karriere') || urlPath.includes('career') || urlPath.includes('human-resources')) type = 'career';
|
|
else if (urlPath.includes('portfolio') || urlPath.includes('referenz') || urlPath.includes('projekt') || urlPath.includes('case-study')) type = 'portfolio';
|
|
else if (urlPath.includes('legal') || urlPath.includes('impressum') || urlPath.includes('datenschutz') || urlPath.includes('privacy')) type = 'legal';
|
|
|
|
const h1s = $('h1').map((_, el) => $(el).text()).get();
|
|
const navLinks = $('nav a').map((_, el) => $(el).text().trim()).get().filter(t => t.length > 0);
|
|
const bodyText = $('body').text().replace(/\s+/g, ' ').substring(0, 50000);
|
|
const html = $.html();
|
|
const hexColors = html.match(/#(?:[0-9a-fA-F]{3}){1,2}\b/g) || [];
|
|
const uniqueColors = Array.from(new Set(hexColors)).slice(0, 5);
|
|
|
|
pages.push({
|
|
url: request.url,
|
|
type,
|
|
content: `Title: ${title}\nType: ${type}\nHeadings: ${h1s.join(', ')}\nNav: ${navLinks.join(', ')}\nColors: ${uniqueColors.join(', ')}\nText: ${bodyText}`
|
|
});
|
|
|
|
await enqueueLinks({
|
|
limit: 15,
|
|
transformRequestFunction: (req) => {
|
|
const reqUrl = new URL(req.url);
|
|
if (reqUrl.origin !== origin) return false;
|
|
// Skip assets
|
|
if (reqUrl.pathname.match(/\.(pdf|zip|jpg|png|svg|webp)$/i)) return false;
|
|
return req;
|
|
}
|
|
});
|
|
},
|
|
});
|
|
|
|
await crawler.run([url]);
|
|
|
|
const typeCounts = pages.reduce((acc, p) => {
|
|
acc[p.type] = (acc[p.type] || 0) + 1;
|
|
return acc;
|
|
}, {} as Record<string, number>);
|
|
|
|
let summary = `\nCrawl Summary: Identified ${pages.length} pages total on ${origin}.\n`;
|
|
summary += Object.entries(typeCounts).map(([type, count]) => `- ${type}: ${count}`).join('\n') + '\n\n';
|
|
|
|
return summary + pages.map(p => `--- PAGE: ${p.url} ---\n${p.content}`).join('\n\n');
|
|
}
|
|
|
|
async function getAiEstimation(briefing: string, distilledCrawl: string, comments: string | null, apiKey: string, principles: string, techStandards: string, tone: string) {
|
|
let usage = { prompt: 0, completion: 0, cost: 0 };
|
|
const addUsage = (data: any) => {
|
|
if (data?.usage) {
|
|
usage.prompt += data.usage.prompt_tokens || 0;
|
|
usage.completion += data.usage.completion_tokens || 0;
|
|
// OpenRouter provides 'cost' field in USD (as per documentation)
|
|
// If missing, we use a fallback calculation for transparency
|
|
if (data.usage.cost !== undefined) {
|
|
usage.cost += data.usage.cost;
|
|
} else {
|
|
// Fallback: Gemini 3 Flash Flash pricing (~$0.1 / 1M prompt, ~$0.4 / 1M completion)
|
|
usage.cost += (data.usage.prompt_tokens || 0) * (0.1 / 1000000) + (data.usage.completion_tokens || 0) * (0.4 / 1000000);
|
|
}
|
|
}
|
|
};
|
|
|
|
// 1. PASS 1: Fact Extraction (Briefing Sensor)
|
|
console.log(' ↳ Pass 1: Fact Extraction (Briefing Sensor)...');
|
|
const pass1SystemPrompt = `
|
|
You are a precision sensor. Analyze the BRIEFING and extract ONLY the raw facts.
|
|
Tone: Literal, non-interpretive.
|
|
Output language: GERMAN (Strict).
|
|
|
|
### MISSION:
|
|
Focus 100% on the BRIEFING text provided by the user. Use the DISTILLED_CRAWL only as background context for terms or company details. If there is a conflict, the BRIEFING is the absolute source of truth.
|
|
|
|
### OBJECTIVES:
|
|
- Extract **companyName**: The full legal and brand name (e.g., "E-TIB GmbH"). Use signatures and crawl data.
|
|
- Extract **personName**: The name of the primary human contact (e.g., "Danny Joseph"). **CRITICAL**: Check email signatures and "Mit freundlichen Grüßen" blocks. DO NOT use "Sie", "Firma" or generic terms if a name exists.
|
|
- Extract **existingWebsite**: The primary URL mentioned in the briefing or signature (e.g., "www.e-tib.com").
|
|
- Extract **websiteTopic**: A short descriptor of the CORE BUSINESS (e.g., "Kabeltiefbau"). MAX 3 WORDS.
|
|
- **isRelaunch**: Set to TRUE if the briefing mentions an existing website, a URL, or if the company is an established entity (e.g. "Gruppe", "GmbH seit 20XX"). Assume a presence exists that needs a modern "Zentrale Webpräsenz".
|
|
- **CRITICAL LOGIC**: If a URL is mentioned, isRelaunch MUST be TRUE.
|
|
- For all textual values (deadline, websiteTopic, targetAudience etc.): USE GERMAN.
|
|
- **multilang**: ONLY if the briefing mentions multiple target languages (e.g., DE/EN).
|
|
- **maps**: If "Google Maps" or location maps are mentioned or implicit (Contact page).
|
|
- **CRITICAL**: Do NOT include "social" in apiSystems unless the user explicitly wants to SYNC/POST content to social media. "Existing social media links" are NOT apiSystems.
|
|
- **CRITICAL**: "Video Player", "Cookie Banner", "Animations" are NOT features. They are visual/base. Do NOT map them to features.
|
|
|
|
### CATEGORY MAPPING (IDs ONLY):
|
|
- **selectedPages**: [Home, About, Services, Contact, Landing, Legal]
|
|
- **features**: [blog_news, products, jobs, refs, events]
|
|
- **functions**: [search, filter, pdf, forms, members, calendar, multilang, chat]
|
|
- **apiSystems**: [crm_erp, payment, marketing, ecommerce, maps, social, analytics]
|
|
- **assets**: [existing_website, logo, styleguide, content_concept, media, icons, illustrations, fonts]
|
|
|
|
### OUTPUT FORMAT (Strict JSON):
|
|
{
|
|
"companyName": string,
|
|
"companyAddress": string,
|
|
"personName": string,
|
|
"existingWebsite": string,
|
|
"websiteTopic": string,
|
|
"isRelaunch": boolean,
|
|
"selectedPages": string[],
|
|
"features": string[],
|
|
"functions": string[],
|
|
"apiSystems": string[],
|
|
"assets": string[],
|
|
"deadline": string (GERMAN),
|
|
"targetAudience": "B2B" | "B2C" | "Internal" | string (GERMAN),
|
|
"expectedAdjustments": "low" | "medium" | "high" | string (GERMAN),
|
|
"employeeCount": "ca. 10+" | "ca. 50+" | "ca. 100+" | "ca. 250+" | "ca. 500+" | "ca. 1000+"
|
|
}
|
|
`;
|
|
const pass1UserPrompt = `BRIEFING (TRUTH SOURCE):\n${briefing}\n\nCOMMENTS:\n${comments}\n\nDISTILLED_CRAWL (CONTEXT ONLY):\n${distilledCrawl}`;
|
|
const p1Resp = await axios.post('https://openrouter.ai/api/v1/chat/completions', {
|
|
model: 'google/gemini-3-flash-preview',
|
|
messages: [{ role: 'system', content: pass1SystemPrompt }, { role: 'user', content: pass1UserPrompt }],
|
|
response_format: { type: 'json_object' }
|
|
}, { headers: { 'Authorization': `Bearer ${apiKey}`, 'Content-Type': 'application/json' } });
|
|
addUsage(p1Resp.data);
|
|
const facts = JSON.parse(p1Resp.data.choices[0].message.content);
|
|
|
|
// 2. PASS 2: Feature Deep-Dive
|
|
console.log(' ↳ Pass 2: Feature Deep-Dive...');
|
|
const pass2SystemPrompt = `
|
|
You are a detail-oriented Solution Architect.
|
|
For EVERY item selected in Pass 1 (pages, features, functions, apiSystems), write a specific justification and technical scope.
|
|
|
|
### RULES:
|
|
1. **CONCRETE & SPECIFIC**: Do NOT say "Implementation of X". Say "Displaying X with Y filters".
|
|
2. **NO EFFECTS**: Do not mention "fade-ins", "animations" or "visual styling". Focus on FUNCTION.
|
|
3. **ABSOLUTE RULE**: EVERYTHING MUST BE GERMAN.
|
|
4. **TRANSPARENCY**: Explain exactly what the USER gets.
|
|
5. **API NOTE**: For 'media' or 'video', explicitly state "Upload & Integration" (NO STREAMING).
|
|
|
|
### INPUT (from Pass 1):
|
|
${JSON.stringify(facts, null, 2)}
|
|
|
|
### OUTPUT FORMAT (Strict JSON):
|
|
{
|
|
"pageDetails": { "Home": string, ... },
|
|
"featureDetails": { "blog_news": string, ... },
|
|
"functionDetails": { "search": string, ... },
|
|
"apiDetails": { "crm_erp": string, ... }
|
|
}
|
|
`;
|
|
const p2Resp = await axios.post('https://openrouter.ai/api/v1/chat/completions', {
|
|
model: 'google/gemini-3-flash-preview',
|
|
messages: [{ role: 'system', content: pass2SystemPrompt }, { role: 'user', content: briefing }],
|
|
response_format: { type: 'json_object' }
|
|
}, { headers: { 'Authorization': `Bearer ${apiKey}`, 'Content-Type': 'application/json' } });
|
|
addUsage(p2Resp.data);
|
|
const details = JSON.parse(p2Resp.data.choices[0].message.content);
|
|
|
|
// 3. PASS 3: Strategic Content (Bespoke Strategy)
|
|
console.log(' ↳ Pass 3: Strategic Content (Bespoke Strategy)...');
|
|
const pass3SystemPrompt = `
|
|
You are a high-end Digital Architect. Your goal is to make the CUSTOMER feel 100% understood.
|
|
Analyze the BRIEFING and the EXISTING WEBSITE context.
|
|
|
|
### TONE & COMMUNICATION PRINCIPLES (STRICT):
|
|
${tone}
|
|
|
|
### OBJECTIVE:
|
|
1. **briefingSummary**: A deep, respectful summary of the status quo and the target state.
|
|
- **LENGTH**: EXACTLY TWO PARAGRAPHS. Minimum 8 sentences total.
|
|
- **MIRROR TEST**: Acknowledge the EXISTING website specifically. Why does the new project make sense NOW?
|
|
- **ABSOLUTE RULE**: DO NOT claim "keine digitale Repräsentation", "erstmals abgebildet", "Erstplatzierung" or "kommunikative Lücke" regarding existence if Pass 1 identified this as a RELAUNCH (isRelaunch=true). EXPLICITLY acknowledge the existing context and the NEED FOR EVOLUTION/MODERNIZATION.
|
|
- **RESPECT**: Explicitly incorporate the customer's expressed wishes.
|
|
- **ABSOLUTE RULE**: DO NOT INVENT DETAILS. Do not mention specific people (e.g., "Frieder Helmich"), software versions, or internal details NOT present in the briefing.
|
|
- **TONE**: Natural Ich-Form. Clear, direct, zero marketing fluff.
|
|
2. **designVision**: A high-density, professional vision of the future execution.
|
|
- **LENGTH**: EXACTLY TWO PARAGRAPHS. Minimum 6 sentences total.
|
|
- **INVESTMENT VALUE**: Plant a clear picture of a stable, high-quality system.
|
|
- **TECHNICAL PRECISION**: Focus on execution (Typografie, Visual Logic, Performance).
|
|
- **NO FLUFF**: Do NOT focus on "Full-Screen Hero Video" as the main thing. Focus on the FIRM's essence and how we translate it into a professional tool.
|
|
- **ABSOLUTE RULE**: NO HALLUCINATIONS. Stay general yet precise. No "Verzicht auf Stockmaterial" unless explicitly stated.
|
|
|
|
### OUTPUT FORMAT (Strict JSON):
|
|
{
|
|
"briefingSummary": string,
|
|
"designVision": string
|
|
}
|
|
`;
|
|
const p3Resp = await axios.post('https://openrouter.ai/api/v1/chat/completions', {
|
|
model: 'google/gemini-3-flash-preview',
|
|
messages: [
|
|
{ role: 'system', content: pass3SystemPrompt },
|
|
{ role: 'user', content: `BRIEFING (TRUTH SOURCE):\n${briefing}\n\nEXISTING WEBSITE (CONTEXT):\n${distilledCrawl}\n\nEXTRACTED FACTS:\n${JSON.stringify(facts, null, 2)}` }
|
|
],
|
|
response_format: { type: 'json_object' }
|
|
}, { headers: { 'Authorization': `Bearer ${apiKey}`, 'Content-Type': 'application/json' } });
|
|
addUsage(p3Resp.data);
|
|
const strategy = JSON.parse(p3Resp.data.choices[0].message.content);
|
|
|
|
// 4. PASS 4: Information Architecture (Sitemap)
|
|
console.log(' ↳ Pass 4: Information Architecture...');
|
|
const pass4SystemPrompt = `
|
|
You are a Senior UX Architect. Design a hierarchical sitemap following the 'Industrial Logic' principle.
|
|
EVERYTHING MUST BE IN GERMAN.
|
|
|
|
### SITEMAP RULES:
|
|
1. **HIERARCHY**: Build a logical tree. Group by category (e.g., "Kern-Präsenz", "Lösungen", "Vertrauen", "Rechtliches").
|
|
2. **INTENT**: Each page MUST have a title and a brief functional conversion intent (desc).
|
|
3. **COMPREHENSIVENESS**: Ensure all 'selectedPages' and 'features' from Pass 1 are represented.
|
|
4. **LANGUAGE**: STRICT GERMAN TITLES. Do NOT use "Home", "About", "Services". Use "Startseite", "Über uns", "Leistungen".
|
|
|
|
### DATA CONTEXT:
|
|
${JSON.stringify({ facts, strategy }, null, 2)}
|
|
|
|
### OUTPUT FORMAT (Strict JSON):
|
|
{
|
|
"websiteTopic": string,
|
|
"sitemap": [ { "category": string, "pages": [ { "title": string, "desc": string } ] } ]
|
|
}
|
|
`;
|
|
const p4Resp = await axios.post('https://openrouter.ai/api/v1/chat/completions', {
|
|
model: 'google/gemini-3-flash-preview',
|
|
messages: [{ role: 'system', content: pass4SystemPrompt }, { role: 'user', content: `BRIEFING (TRUTH SOURCE):\n${briefing}\n\nDISTILLED_CRAWL (CONTEXT):\n${distilledCrawl}` }],
|
|
response_format: { type: 'json_object' }
|
|
}, { headers: { 'Authorization': `Bearer ${apiKey}`, 'Content-Type': 'application/json' } });
|
|
addUsage(p4Resp.data);
|
|
const ia = JSON.parse(p4Resp.data.choices[0].message.content);
|
|
|
|
// 5. PASS 5: Position Synthesis & Pricing Transparency
|
|
console.log(' ↳ Pass 5: Position Synthesis...');
|
|
const pass5SystemPrompt = `
|
|
You are a Senior Solution Architect. Your goal is ABSOLUTE TRANSPARENCY for the customer.
|
|
Each position in the quote must be perfectly justified and detailed.
|
|
|
|
### POSITION TITLES:
|
|
"Basis Website Setup", "Individuelle Seiten", "System-Module", "Logik-Funktionen", "Visuelle Inszenierung", "Komplexe Interaktion", "Schnittstellen (API)", "Inhaltsverwaltung (CMS)", "Sorglos-Betrieb (Hosting)".
|
|
|
|
### MAPPING RULES (STRICTLY BASED ON PRICING.MD):
|
|
- **Basis Website Setup**: Infrastructure, Hosting, SEO-Basics, Cookie-Consent, Staging/Production.
|
|
- **Individuelle Seiten**: Layout and structure for specific pages (e.g. Home, Services).
|
|
- **System-Module (Features)**: Closed systems with data structures (Blog, News, Products, Jobs, References).
|
|
- **Logik-Funktionen**: Pure logic (Search, Filter, Forms, PDF-Export, Multi-lang).
|
|
- **Visuelle Inszenierung**: Hero-Story, visual flows, Scroll-effects.
|
|
- **Komplexe Interaktion**: UI-experiences like Configurators or multi-step processes.
|
|
- **Schnittstellen (API)**: REAL Data Syncs (CRM, ERP, Stripe).
|
|
- **Inhaltsverwaltung (CMS)**: Setup and mapping for Headless CMS.
|
|
- **Sorglos-Betrieb (Hosting)**: Hosting, Updates, Backups.
|
|
|
|
### RULES FOR positionDescriptions (STRICT):
|
|
1. **NO "ICH-FORM"**: Do NOT use "Ich" or "Mein". Lead with the action or component.
|
|
2. **CONCISE & ITEM-BASED**: Use short, technical sentences. Focus on WHAT is delivered.
|
|
3. **ZERO GENERALIZATION**: Do NOT say "Verschiedene Funktionen".
|
|
4. **ITEMIZED SYNTHESIS**: Mention EVERY component selected in Pass 1.
|
|
5. **HARD SPECIFICS**: Preserve technical details from the briefing (e.g., "110 kV", "HDD-Bohrtechnik").
|
|
6. **STYLE**: Direct, engineering-grade, 100% GERMAN.
|
|
7. **SPECIFIC - PAGES**: For "Individuelle Seiten", list the pages as a comma-separated list.
|
|
8. **SPECIFIC - HOSTING**: Always append: "Inkl. 20GB Speicher."
|
|
9. **SPECIFIC - LOGIC**: Describe the ACTUAL logic. NEVER use generic terms.
|
|
10. **STRICT KEYS**: Keys MUST be EXACTLY the ones defined in POSITION TITLES.
|
|
|
|
### EXAMPLES (FEW-SHOT):
|
|
- **BAD**: "Ich entwickle die Seiten: Startseite, Leistungen, Kontakt."
|
|
- **GOOD**: "Erstellung der Seiten: Startseite (Video-Hero), Über uns (Timeline), Leistungen, Kontakt."
|
|
- **BAD**: "Ich binde Google Maps an."
|
|
- **GOOD**: "Native API-Integration von Google Maps mit individueller Standort-Visualisierung."
|
|
- **BAD**: "Ich programmiere Scroll-Effekte."
|
|
- **GOOD**: "Visuelle Inszenierung der Meilensteine durch Scroll-aktivierte Timeline-Elemente."
|
|
|
|
### POSITION RULES (STRICT):
|
|
1. **Basis Website Setup**: This position MUST ALWAYS contain exactly these 7 points: "Projekt-Setup & Infrastruktur, Hosting-Bereitstellung, Grundstruktur & Design-Vorlage, technisches SEO-Basics, Analytics (mit automatischem Mail-Report), Testing-, Staging- & Production-Umgebung, Livegang."
|
|
2. **Sorglos-Betrieb (Hosting)**: Describe the service (Hosting, SSL, Security-Updates, 24/7 Monitoring, Portfolio-Update-Service). NEVER mention "Inklusive Basis-Infrastruktur".
|
|
3. **LOGIC CONSISTENCY**: If Pass 1 extracted 1 function, you MUST describe exactly 1 function scope. If you describe two things (e.g., "Formular AND Search") but the count is 1, it is a FAIL.
|
|
4. **SIMPLICITY**: Write in "Simple German". High density of information, but easy for a CEO. No jargon.
|
|
5. **NO IMPLEMENTATION DETAILS**: Focus on WHAT is done, not HOW (no libraries, no technical "under-the-hood" talk).
|
|
|
|
### DATA CONTEXT:
|
|
${JSON.stringify({ facts, details, strategy, ia }, null, 2)}
|
|
|
|
### OUTPUT FORMAT (Strict JSON):
|
|
{
|
|
"positionDescriptions": { "Basis Website Setup": string, ... }
|
|
}
|
|
`;
|
|
const p5Resp = await axios.post('https://openrouter.ai/api/v1/chat/completions', {
|
|
model: 'google/gemini-3-flash-preview',
|
|
messages: [{ role: 'system', content: pass5SystemPrompt }, { role: 'user', content: briefing }],
|
|
response_format: { type: 'json_object' }
|
|
}, { headers: { 'Authorization': `Bearer ${apiKey}`, 'Content-Type': 'application/json' } });
|
|
addUsage(p5Resp.data);
|
|
const positionsData = JSON.parse(p5Resp.data.choices[0].message.content);
|
|
|
|
// 6. PASS 6: The Industrial Critic
|
|
console.log(' ↳ Pass 6: The Industrial Critic (Quality Gate)...');
|
|
const pass6SystemPrompt = `
|
|
You are the "Industrial Critic". Your goal is to catch quality regressions and ensure the document is bespoke, technical, and professional.
|
|
Analyze the CURRENT_STATE against the BRIEFING_TRUTH.
|
|
|
|
### CRITICAL ERROR CHECKLIST (FAIL IF FOUND):
|
|
1. **Hallucination Leakage**: FAIL if names of people (e.g., "Frieder Helmich"), specific software versions, or invented details are used unless they appear EXACTLY in the BRIEFING.
|
|
- **CRITICAL**: Forbid "Sie", "Ansprechpartner" or "Unternehmen" for personName if a name IS in the briefing. If none is in briefing, use empty string.
|
|
2. **Logic Conflict**: FAIL if isRelaunch is true but briefingSummary claims no website exists or uses phrases like "Da aktuell keine digitale Repräsentation vorliegt", "erstmals abgebildet", "Erstplatzierung" or "Lücke schließen" (regarding existence).
|
|
- FAIL if the description in positionDescriptions mentions more items than extracted in facts.
|
|
3. **Implementation Fluff**: FAIL if "React", "Next.js", "TypeScript", "Tailwind" or other tech-stack details are mentioned. Focus on Concept & Result.
|
|
4. **Length Check**: Briefing and Vision MUST be significantly long (EXACTLY 2 paragraphs each, minimum 8 sentences for briefing, 6 for vision).
|
|
|
|
### MISSION:
|
|
Return updated fields ONLY. Specifically focus on hardening 'positionDescriptions', 'sitemap', 'briefingSummary', and 'designVision'.
|
|
|
|
### DATA CONTEXT:
|
|
${JSON.stringify({ facts, strategy, ia, positionsData }, null, 2)}
|
|
`;
|
|
const p6Resp = await axios.post('https://openrouter.ai/api/v1/chat/completions', {
|
|
model: 'google/gemini-3-flash-preview',
|
|
messages: [{ role: 'system', content: pass6SystemPrompt }, { role: 'user', content: `BRIEFING_TRUTH:\n${briefing}` }],
|
|
response_format: { type: 'json_object' }
|
|
}, { headers: { 'Authorization': `Bearer ${apiKey}`, 'Content-Type': 'application/json' } });
|
|
addUsage(p6Resp.data);
|
|
const reflection = JSON.parse(p6Resp.data.choices[0].message.content);
|
|
|
|
// 6. Reflection Merge Utility
|
|
const mergeReflection = (state: any, reflection: any) => {
|
|
let result = { ...state };
|
|
const unwrap = (obj: any): any => {
|
|
if (!obj || typeof obj !== 'object' || Array.isArray(obj)) return obj;
|
|
if (obj["0"]) return unwrap(obj["0"]);
|
|
if (obj.state) return unwrap(obj.state);
|
|
if (obj.facts) return unwrap(obj.facts);
|
|
if (obj.strategy) return unwrap(obj.strategy);
|
|
if (obj.ia) return unwrap(obj.ia);
|
|
if (obj.positionsData) return unwrap(obj.positionsData);
|
|
return obj;
|
|
};
|
|
|
|
const cleanedReflection = unwrap(reflection);
|
|
Object.entries(cleanedReflection).forEach(([key, value]) => {
|
|
if (value && value !== "" && value !== "null") {
|
|
result[key] = value;
|
|
}
|
|
});
|
|
return result;
|
|
};
|
|
|
|
let finalState = mergeReflection({
|
|
...initialState,
|
|
...facts,
|
|
...strategy,
|
|
...ia,
|
|
...positionsData
|
|
}, reflection);
|
|
|
|
finalState.statusQuo = facts.isRelaunch ? 'Relaunch' : 'Neuentwicklung';
|
|
|
|
// Normalization Layer: Map hallucinated German keys back to internal keys
|
|
const normalizationMap: Record<string, string> = {
|
|
"Briefing-Zusammenfassung": "briefingSummary",
|
|
"Design-Vision": "designVision",
|
|
"Zusammenfassung": "briefingSummary",
|
|
"Vision": "designVision",
|
|
"BRIEFING_SUMMARY": "briefingSummary",
|
|
"DESIGN_VISION": "designVision"
|
|
};
|
|
|
|
Object.entries(normalizationMap).forEach(([gerKey, intKey]) => {
|
|
if (finalState[gerKey] && !finalState[intKey]) {
|
|
if (typeof finalState[gerKey] === 'object' && !Array.isArray(finalState[gerKey])) {
|
|
finalState[intKey] = Object.values(finalState[gerKey]).join('\n\n');
|
|
} else {
|
|
finalState[intKey] = finalState[gerKey];
|
|
}
|
|
}
|
|
});
|
|
|
|
// Sitemap Normalization (German keys to internal)
|
|
if (Array.isArray(finalState.sitemap)) {
|
|
finalState.sitemap = finalState.sitemap.map((cat: any) => ({
|
|
category: cat.category || cat.kategorie || cat.Kategorie || cat.title || "Allgemein",
|
|
pages: (cat.pages || cat.seiten || cat.Seiten || []).map((page: any) => ({
|
|
title: page.title || page.titel || page.Titel || "Seite",
|
|
desc: page.desc || page.beschreibung || page.Beschreibung || page.description || ""
|
|
}))
|
|
}));
|
|
}
|
|
|
|
// Position Descriptions Normalization
|
|
if (finalState.positionDescriptions) {
|
|
const normalized: Record<string, string> = {};
|
|
Object.entries(finalState.positionDescriptions).forEach(([key, value]) => {
|
|
const normalizedKey = key === 'titel' || key === 'Title' ? 'title' : key;
|
|
const normalizedValue = typeof value === 'object' ? (value as any).beschreibung || (value as any).description || JSON.stringify(value) : value;
|
|
normalized[normalizedKey] = normalizedValue as string;
|
|
});
|
|
finalState.positionDescriptions = normalized;
|
|
}
|
|
|
|
// Normalize final state
|
|
if (Array.isArray(finalState.positionDescriptions)) {
|
|
const normalized: Record<string, string> = {};
|
|
finalState.positionDescriptions.forEach((item: any) => {
|
|
const key = item.feature || item.id || item.title || item.pos;
|
|
if (key) normalized[key] = item.description || item.desc;
|
|
});
|
|
finalState.positionDescriptions = normalized;
|
|
}
|
|
|
|
if (finalState.sitemap && !Array.isArray(finalState.sitemap)) {
|
|
if (finalState.sitemap.categories) finalState.sitemap = finalState.sitemap.categories;
|
|
else if (finalState.sitemap.sitemap) finalState.sitemap = finalState.sitemap.sitemap;
|
|
else {
|
|
const entries = Object.entries(finalState.sitemap);
|
|
if (entries.every(([_, v]) => Array.isArray(v))) {
|
|
finalState.sitemap = entries.map(([category, pages]) => ({ category, pages }));
|
|
}
|
|
}
|
|
}
|
|
|
|
return { state: finalState, usage };
|
|
}
|
|
|
|
main().catch(console.error);
|