geograsim/App/sims/fluggesellschaft/scripts/generate-city-questions.py

#!/usr/bin/env python3
"""
Fluggesellschaft — Quiz-Fragen-Generator.

Pro Stadt werden ZWEI Fragen erzeugt:
  question:        "In welchem Land liegt X?"  (analog Busfahrt)
  questionContinent: "Auf welchem Kontinent liegt X?"  (neu, weltweit relevant)

Die Engine wird pro Auftrag eine der beiden Fragen zufällig (oder
deterministisch pro Auftrags-ID) wählen.

Deterministischer RNG via mulberry32-Hash auf city.id — stabile Re-Runs.
"""

import json
import sys
from pathlib import Path

ROOT = Path(__file__).resolve().parents[1] / "assets" / "data"
CITIES_PATH = ROOT / "cities.json"

CONTINENTS = ["Europa", "Asien", "Afrika", "Nordamerika", "Südamerika", "Ozeanien", "Antarktis"]

def mulberry32(seed):
    state = [seed & 0xFFFFFFFF]
    def nx():
        state[0] = (state[0] + 0x6D2B79F5) & 0xFFFFFFFF
        t = state[0]
        r = ((t ^ (t >> 15)) * (1 | t)) & 0xFFFFFFFF
        r = ((r + ((r ^ (r >> 7)) * (61 | r))) & 0xFFFFFFFF) ^ r
        return ((r ^ (r >> 14)) & 0xFFFFFFFF) / 4294967296.0
    return nx

def slug_hash(s):
    h = 2166136261
    for ch in s:
        h = (h ^ ord(ch)) & 0xFFFFFFFF
        h = (h * 16777619) & 0xFFFFFFFF
    return h

# Heuristik: Kontinent aus Land ableiten (mehr Genauigkeit als auf lat/lon basieren)
COUNTRY_CONTINENT = {
    # Europa (alle Bus-Länder)
    "Österreich": "Europa", "Deutschland": "Europa", "Schweiz": "Europa", "Italien": "Europa",
    "Frankreich": "Europa", "Spanien": "Europa", "Portugal": "Europa", "Belgien": "Europa",
    "Niederlande": "Europa", "Luxemburg": "Europa", "Polen": "Europa", "Tschechien": "Europa",
    "Slowakei": "Europa", "Ungarn": "Europa", "Slowenien": "Europa", "Kroatien": "Europa",
    "Bosnien und Herzegowina": "Europa", "Serbien": "Europa", "Montenegro": "Europa",
    "Nordmazedonien": "Europa", "Albanien": "Europa", "Bulgarien": "Europa", "Rumänien": "Europa",
    "Griechenland": "Europa", "Türkei": "Europa",
    "Vereinigtes Königreich": "Europa", "Großbritannien": "Europa", "Irland": "Europa",
    "Dänemark": "Europa", "Schweden": "Europa", "Norwegen": "Europa", "Finnland": "Europa",
    "Island": "Europa", "Estland": "Europa", "Lettland": "Europa", "Litauen": "Europa",
    "Ukraine": "Europa", "Belarus": "Europa", "Moldawien": "Europa", "Russland": "Europa",
    "Liechtenstein": "Europa", "Monaco": "Europa", "Malta": "Europa", "Zypern": "Europa",
    "Andorra": "Europa", "San Marino": "Europa", "Vatikan": "Europa",
    # Nordamerika
    "USA": "Nordamerika", "Vereinigte Staaten": "Nordamerika", "Kanada": "Nordamerika",
    "Mexiko": "Nordamerika", "Kuba": "Nordamerika",
    # Südamerika
    "Brasilien": "Südamerika", "Argentinien": "Südamerika", "Ecuador": "Südamerika",
    "Kolumbien": "Südamerika", "Peru": "Südamerika", "Bolivien": "Südamerika",
    "Chile": "Südamerika", "Venezuela": "Südamerika", "Uruguay": "Südamerika",
    "Paraguay": "Südamerika",
    # Afrika
    "Marokko": "Afrika", "Tunesien": "Afrika", "Ägypten": "Afrika", "Senegal": "Afrika",
    "Nigeria": "Afrika", "Kenia": "Afrika", "Äthiopien": "Afrika", "Südafrika": "Afrika",
    "Algerien": "Afrika", "Libyen": "Afrika", "Ghana": "Afrika",
    # Asien
    "Japan": "Asien", "China": "Asien", "Indien": "Asien", "Thailand": "Asien",
    "Singapur": "Asien", "Südkorea": "Asien", "Indonesien": "Asien",
    "Vietnam": "Asien", "Philippinen": "Asien", "Malaysia": "Asien",
    "Iran": "Asien", "Saudi-Arabien": "Asien", "Vereinigte Arabische Emirate": "Asien",
    # Ozeanien
    "Australien": "Ozeanien", "Neuseeland": "Ozeanien",
}

def pick_distractors(seed_str, correct, pool):
    pool = [p for p in pool if p != correct]
    rng = mulberry32(slug_hash(seed_str))
    shuffled = pool[:]
    for i in range(len(shuffled)-1, 0, -1):
        j = int(rng() * (i+1))
        shuffled[i], shuffled[j] = shuffled[j], shuffled[i]
    return shuffled[:2]

def build_country_question(city, all_countries):
    distractors = pick_distractors(city["id"], city["country"], all_countries)
    rng = mulberry32(slug_hash(city["id"] + "_pos"))
    pos = int(rng() * 3)
    choices = distractors[:]
    choices.insert(pos, city["country"])
    return {
        "prompt": {"de": f"In welchem Land liegt {city['title']}?", "easy": f"In welchem Land liegt {city['title']}?"},
        "choices": choices, "correctIndex": pos,
    }

def build_continent_question(city):
    correct = city.get("continent") or COUNTRY_CONTINENT.get(city["country"], "Europa")
    pool = [c for c in CONTINENTS if c != correct]
    rng = mulberry32(slug_hash(city["id"] + "_cont"))
    shuffled = pool[:]
    for i in range(len(shuffled)-1, 0, -1):
        j = int(rng() * (i+1))
        shuffled[i], shuffled[j] = shuffled[j], shuffled[i]
    distractors = shuffled[:2]
    pos = int(mulberry32(slug_hash(city["id"] + "_cpos"))() * 3)
    choices = distractors[:]
    choices.insert(pos, correct)
    return {
        "prompt": {"de": f"Auf welchem Kontinent liegt {city['title']}?", "easy": f"Auf welchem Kontinent liegt {city['title']}?"},
        "choices": choices, "correctIndex": pos,
    }

def main():
    with CITIES_PATH.open(encoding="utf-8") as f:
        cities = json.load(f)
    all_countries = sorted({c["country"] for c in cities})
    print(f"Cities: {len(cities)}, Countries: {len(all_countries)}")

    # Continent-Property setzen falls fehlt
    for c in cities:
        if not c.get("continent"):
            c["continent"] = COUNTRY_CONTINENT.get(c["country"], "Europa")

    added = 0
    for c in cities:
        if "question" not in c:
            c["question"] = build_country_question(c, all_countries)
            added += 1
        if "questionContinent" not in c:
            c["questionContinent"] = build_continent_question(c)

    with CITIES_PATH.open("w", encoding="utf-8") as f:
        json.dump(cities, f, ensure_ascii=False, indent=2)
        f.write("\n")
    print(f"OK -- {added} Land-Fragen + Kontinent-Fragen ergaenzt.")

if __name__ == "__main__":
    sys.exit(main())