Brancheneinstufung2/connector-superoffice/build_matrix.py

import sqlite3
import os
import json
import requests
from dotenv import load_dotenv

load_dotenv(override=True)

DB_FILE = "marketing_matrix.db"
GEMINI_API_KEY = os.getenv("GEMINI_API_KEY")
NOTION_API_KEY = os.getenv("NOTION_API_KEY")
NOTION_DB_INDUSTRIES = "2ec88f4285448014ab38ea664b4c2b81"

# --- MAPPINGS ---
# SuperOffice ID -> Notion Vertical Name
VERTICAL_MAP = {
    23: "Logistics - Warehouse"
}

# SuperOffice ID -> Persona Name & Pains (aus unserer Definition)
ROLE_MAP = {
    19: {"name": "Operativer Entscheider", "pains": "Zuverlässigkeit, einfache Bedienbarkeit, Personaleinsatz-Optimierung, minimale Störungen"},
    20: {"name": "Infrastruktur-Verantwortlicher", "pains": "Technische Machbarkeit, IT-Sicherheit, Integration, Brandschutz"},
    21: {"name": "Wirtschaftlicher Entscheider", "pains": "ROI, Amortisationszeit, Kostenstruktur, Einsparpotenziale"},
    22: {"name": "Innovations-Treiber", "pains": "Wettbewerbsfähigkeit, Modernisierung, Employer Branding, Kundenerlebnis"}
}

# --- DATABASE SETUP ---
def init_db():
    conn = sqlite3.connect(DB_FILE)
    c = conn.cursor()
    c.execute('''CREATE TABLE IF NOT EXISTS text_blocks
                 (vertical_id INTEGER, role_id INTEGER, 
                  subject TEXT, intro TEXT, social_proof TEXT,
                  PRIMARY KEY (vertical_id, role_id))''')
    conn.commit()
    conn.close()
    print("✅ Database initialized.")

# --- NOTION FETCHER ---
def get_vertical_pains_gains(vertical_name):
    url = f"https://api.notion.com/v1/databases/{NOTION_DB_INDUSTRIES}/query"
    headers = {
        "Authorization": f"Bearer {NOTION_API_KEY}",
        "Notion-Version": "2022-06-28",
        "Content-Type": "application/json"
    }
    payload = {
        "filter": {
            "property": "Vertical",
            "title": {"contains": vertical_name}
        }
    }
    resp = requests.post(url, headers=headers, json=payload)
    if resp.status_code == 200:
        results = resp.json().get("results", [])
        if results:
            props = results[0]['properties']
            pains = props.get('Pains', {}).get('rich_text', [])
            gains = props.get('Gains', {}).get('rich_text', [])
            return {
                "pains": pains[0]['plain_text'] if pains else "",
                "gains": gains[0]['plain_text'] if gains else ""
            }
    print(f"⚠️ Warning: No data found for {vertical_name}")
    return {"pains": "N/A", "gains": "N/A"}

# --- GEMINI GENERATOR ---
def generate_text(vertical_name, v_data, role_id, role_data):
    url = f"https://generativelanguage.googleapis.com/v1beta/models/gemini-2.0-flash:generateContent?key={GEMINI_API_KEY}"
    
    prompt = f"""
    Du bist ein B2B-Copywriter. Erstelle 3 Textbausteine für eine Cold-Outreach E-Mail.
    
    KONTEXT:
    Branche: {vertical_name}
    Branchen-Pains: {v_data['pains']}
    Lösung-Gains: {v_data['gains']}
    
    Rolle: {role_data['name']}
    Rollen-Pains: {role_data['pains']}
    
    AUFGABE:
    1. Subject: Betreffzeile (max 40 Zeichen!). Knackig, Pain-bezogen.
    2. Intro: Einleitungssatz (max 40 Zeichen!). Brücke Pain -> Lösung.
    3. SocialProof: Referenzsatz (max 40 Zeichen!). "Wir arbeiten mit X..."
    
    FORMAT (JSON):
    {{ "Subject": "...", "Intro": "...", "SocialProof": "..." }}
    """
    
    payload = {
        "contents": [{"parts": [{"text": prompt}]}],
        "generationConfig": {"responseMimeType": "application/json"}
    }
    
    try:
        resp = requests.post(url, json=payload)
        if resp.status_code == 200:
            return json.loads(resp.json()['candidates'][0]['content']['parts'][0]['text'])
    except Exception as e:
        print(f"Gemini Error: {e}")
    return None

# --- MAIN ---
def run_matrix():
    init_db()
    conn = sqlite3.connect(DB_FILE)
    c = conn.cursor()
    
    # Iterate Verticals
    for v_id, v_name in VERTICAL_MAP.items():
        print(f"\nProcessing Vertical: {v_name} (ID: {v_id})")
        v_data = get_vertical_pains_gains(v_name)
        
        # Iterate Roles
        for r_id, r_data in ROLE_MAP.items():
            print(f"  > Generating for Role: {r_data['name']} (ID: {r_id})...", end="", flush=True)
            
            # Check if exists (optional: skip if exists)
            # ...
            
            text_block = generate_text(v_name, v_data, r_id, r_data)
            
            if text_block:
                # Robustness: Handle list return from Gemini
                if isinstance(text_block, list):
                    if len(text_block) > 0 and isinstance(text_block[0], dict):
                        text_block = text_block[0] # Take first item if list of dicts
                    else:
                        print(" ❌ Failed (Unexpected JSON format: List without dicts).")
                        continue

                # Cut to 40 chars hard limit (Safety)
                subj = text_block.get("Subject", "")[:40]
                intro = text_block.get("Intro", "Intro")[:40] # Fallback key name check
                if "Introduction_Textonly" in text_block: intro = text_block["Introduction_Textonly"][:40]
                proof = text_block.get("SocialProof", "")[:40]
                if "Industry_References_Textonly" in text_block: proof = text_block["Industry_References_Textonly"][:40]

                c.execute("INSERT OR REPLACE INTO text_blocks VALUES (?, ?, ?, ?, ?)",
                          (v_id, r_id, subj, intro, proof))
                conn.commit()
                print(" ✅ Done.")
            else:
                print(" ❌ Failed.")

    conn.close()
    print("\nMatrix generation complete.")

if __name__ == "__main__":
    run_matrix()