breakpilot-lehrer/klausur-service/backend/country_metadata.py

"""
Country Metadata for EU/EWR Compliance Coverage.

Provides entry points (DPA URLs, legal portals, languages) for all EU/EWR countries.
DACH countries have full RAG coverage; others have entry-point metadata for system prompt injection.

Usage:
    from country_metadata import COUNTRY_METADATA, get_country_context, DACH_COUNTRIES
"""

from typing import Dict, Optional

COUNTRY_METADATA: Dict[str, dict] = {
    "EU": {
        "country_name": "Europaeische Union",
        "language_codes": ["de", "en", "fr"],
        "legal_portal": {"name": "EUR-Lex", "url": "https://eur-lex.europa.eu/"},
        "dpa": {"name": "European Data Protection Board (EDPB)", "url": "https://edpb.europa.eu/"},
        "rag_coverage": "full",
        "priority_modules": ["DSI", "VVT", "TOMs", "AGB", "Impressum", "Retention", "DSFA", "AVV", "Cookies"],
    },
    "DE": {
        "country_name": "Deutschland",
        "language_codes": ["de"],
        "legal_portal": {"name": "Gesetze im Internet", "url": "https://www.gesetze-im-internet.de/"},
        "dpa": {"name": "Datenschutzkonferenz (DSK)", "url": "https://www.datenschutzkonferenz-online.de/"},
        "rag_coverage": "full",
        "priority_modules": ["DSI", "VVT", "TOMs", "AGB", "Impressum", "Retention", "DSFA", "AVV", "Cookies"],
    },
    "AT": {
        "country_name": "Oesterreich",
        "language_codes": ["de"],
        "legal_portal": {"name": "Rechtsinformationssystem (RIS)", "url": "https://www.ris.bka.gv.at/"},
        "dpa": {"name": "Datenschutzbehoerde (DSB)", "url": "https://www.dsb.gv.at/"},
        "rag_coverage": "full",
        "priority_modules": ["DSI", "VVT", "TOMs", "AGB", "Impressum", "Retention", "DSFA", "AVV", "Cookies"],
    },
    "CH": {
        "country_name": "Schweiz",
        "language_codes": ["de", "fr", "it"],
        "legal_portal": {"name": "Fedlex", "url": "https://www.fedlex.admin.ch/"},
        "dpa": {"name": "EDOEB", "url": "https://www.edoeb.admin.ch/"},
        "rag_coverage": "full",
        "priority_modules": ["DSI", "VVT", "TOMs", "AGB", "Impressum", "Retention", "DSFA", "AVV", "Cookies"],
    },
    "FR": {
        "country_name": "Frankreich",
        "language_codes": ["fr"],
        "legal_portal": {"name": "Legifrance", "url": "https://www.legifrance.gouv.fr/"},
        "dpa": {"name": "CNIL", "url": "https://www.cnil.fr/"},
        "rag_coverage": "entry_point",
        "priority_modules": ["DSI", "VVT", "TOMs", "AGB", "Impressum", "Retention"],
    },
    "IT": {
        "country_name": "Italien",
        "language_codes": ["it"],
        "legal_portal": {"name": "Normattiva", "url": "https://www.normattiva.it/"},
        "dpa": {"name": "Garante per la protezione dei dati personali", "url": "https://www.garanteprivacy.it/"},
        "rag_coverage": "entry_point",
        "priority_modules": ["DSI", "VVT", "TOMs", "AGB", "Impressum", "Retention"],
    },
    "ES": {
        "country_name": "Spanien",
        "language_codes": ["es"],
        "legal_portal": {"name": "BOE", "url": "https://www.boe.es/"},
        "dpa": {"name": "AEPD", "url": "https://www.aepd.es/"},
        "rag_coverage": "entry_point",
        "priority_modules": ["DSI", "VVT", "TOMs", "AGB", "Impressum", "Retention"],
    },
    "NL": {
        "country_name": "Niederlande",
        "language_codes": ["nl"],
        "legal_portal": {"name": "Overheid.nl", "url": "https://wetten.overheid.nl/"},
        "dpa": {"name": "Autoriteit Persoonsgegevens", "url": "https://www.autoriteitpersoonsgegevens.nl/"},
        "rag_coverage": "entry_point",
        "priority_modules": ["DSI", "VVT", "TOMs", "AGB", "Impressum", "Retention"],
    },
    "BE": {
        "country_name": "Belgien",
        "language_codes": ["fr", "nl", "de"],
        "legal_portal": {"name": "eJustice Belgium", "url": "https://www.ejustice.just.fgov.be/"},
        "dpa": {"name": "Autorite de protection des donnees (APD)", "url": "https://www.autoriteprotectiondonnees.be/"},
        "rag_coverage": "entry_point",
        "priority_modules": ["DSI", "VVT", "TOMs", "AGB", "Impressum", "Retention"],
    },
    "LU": {
        "country_name": "Luxemburg",
        "language_codes": ["fr", "de"],
        "legal_portal": {"name": "Legilux", "url": "https://legilux.public.lu/"},
        "dpa": {"name": "CNPD", "url": "https://cnpd.public.lu/"},
        "rag_coverage": "entry_point",
        "priority_modules": ["DSI", "VVT", "TOMs", "AGB", "Impressum", "Retention"],
    },
    "IE": {
        "country_name": "Irland",
        "language_codes": ["en"],
        "legal_portal": {"name": "Irish Statute Book", "url": "https://www.irishstatutebook.ie/"},
        "dpa": {"name": "Data Protection Commission (DPC)", "url": "https://www.dataprotection.ie/"},
        "rag_coverage": "entry_point",
        "priority_modules": ["DSI", "VVT", "TOMs", "AGB", "Impressum", "Retention"],
    },
    "DK": {
        "country_name": "Daenemark",
        "language_codes": ["da"],
        "legal_portal": {"name": "Retsinformation", "url": "https://www.retsinformation.dk/"},
        "dpa": {"name": "Datatilsynet", "url": "https://www.datatilsynet.dk/"},
        "rag_coverage": "entry_point",
        "priority_modules": ["DSI", "VVT", "TOMs"],
    },
    "SE": {
        "country_name": "Schweden",
        "language_codes": ["sv"],
        "legal_portal": {"name": "Riksdagen", "url": "https://www.riksdagen.se/"},
        "dpa": {"name": "IMY (Integritetsskyddsmyndigheten)", "url": "https://www.imy.se/"},
        "rag_coverage": "entry_point",
        "priority_modules": ["DSI", "VVT", "TOMs"],
    },
    "FI": {
        "country_name": "Finnland",
        "language_codes": ["fi", "sv"],
        "legal_portal": {"name": "Finlex", "url": "https://www.finlex.fi/"},
        "dpa": {"name": "Tietosuojavaltuutetun toimisto", "url": "https://tietosuoja.fi/"},
        "rag_coverage": "entry_point",
        "priority_modules": ["DSI", "VVT", "TOMs"],
    },
    "NO": {
        "country_name": "Norwegen",
        "language_codes": ["no"],
        "legal_portal": {"name": "Lovdata", "url": "https://lovdata.no/"},
        "dpa": {"name": "Datatilsynet", "url": "https://www.datatilsynet.no/"},
        "rag_coverage": "entry_point",
        "priority_modules": ["DSI", "VVT", "TOMs"],
    },
    "IS": {
        "country_name": "Island",
        "language_codes": ["is"],
        "legal_portal": {"name": "Althingi", "url": "https://www.althingi.is/"},
        "dpa": {"name": "Personuvernd", "url": "https://www.personuvernd.is/"},
        "rag_coverage": "entry_point",
        "priority_modules": ["DSI", "VVT"],
    },
    "LI": {
        "country_name": "Liechtenstein",
        "language_codes": ["de"],
        "legal_portal": {"name": "Gesetze.li", "url": "https://www.gesetze.li/"},
        "dpa": {"name": "Datenschutzstelle", "url": "https://www.datenschutzstelle.li/"},
        "rag_coverage": "entry_point",
        "priority_modules": ["DSI", "VVT", "TOMs"],
    },
    "UK": {
        "country_name": "Vereinigtes Koenigreich",
        "language_codes": ["en"],
        "legal_portal": {"name": "Legislation.gov.uk", "url": "https://www.legislation.gov.uk/"},
        "dpa": {"name": "Information Commissioner's Office (ICO)", "url": "https://ico.org.uk/"},
        "rag_coverage": "entry_point",
        "priority_modules": ["DSI", "VVT", "TOMs", "AGB", "Impressum", "Retention"],
    },
}

DACH_COUNTRIES = {"DE", "AT", "CH"}


def get_country_context(country_code: str) -> Optional[str]:
    """
    Generate context string for system prompt injection.

    For DACH countries: Full RAG search hint.
    For other countries: Entry point with DPA and legal portal URLs.
    """
    meta = COUNTRY_METADATA.get(country_code.upper())
    if not meta:
        return None

    lines = [f"Land: {meta['country_name']} ({country_code.upper()})"]
    lines.append(f"Sprachen: {', '.join(meta['language_codes'])}")
    lines.append(f"Rechtsportal: {meta['legal_portal']['name']} — {meta['legal_portal']['url']}")
    lines.append(f"Datenschutzbehoerde: {meta['dpa']['name']} — {meta['dpa']['url']}")

    if meta["rag_coverage"] == "full":
        lines.append(f"RAG-Abdeckung: VOLLSTAENDIG — Suche in bp_legal_corpus fuer nationale Gesetze verfuegbar.")
        lines.append(f"Compliance-Module: {', '.join(meta['priority_modules'])}")
    else:
        lines.append(f"RAG-Abdeckung: Einstiegspunkt — Keine nationalen Gesetze im RAG. Verweise auf das Rechtsportal und die Datenschutzbehoerde.")
        lines.append(f"Hinweis: Fuer detaillierte rechtliche Informationen zu {meta['country_name']} bitte das Rechtsportal oder die DPA konsultieren.")

    return "\n".join(lines)


def get_all_countries_summary() -> str:
    """Generate a summary of all covered countries for overview display."""
    lines = ["Laenderabdeckung EU/EWR:"]
    lines.append("")
    lines.append("VOLLSTAENDIG (DACH):")
    for code in sorted(DACH_COUNTRIES):
        meta = COUNTRY_METADATA[code]
        lines.append(f"  {code}: {meta['country_name']} — DPA: {meta['dpa']['name']}")

    lines.append("")
    lines.append("EINSTIEGSPUNKTE:")
    for code, meta in sorted(COUNTRY_METADATA.items()):
        if code not in DACH_COUNTRIES and code != "EU":
            lines.append(f"  {code}: {meta['country_name']} — DPA: {meta['dpa']['name']}")

    return "\n".join(lines)