breakpilot-lehrer/geo-service/utils/license_checker.py

"""
License Checker Utility
Validates data source licenses and generates attribution
"""
from typing import Optional
from enum import Enum
import structlog

logger = structlog.get_logger(__name__)


class LicenseType(Enum):
    """Supported data source license types."""
    ODBL = "odbl"  # OpenStreetMap
    COPERNICUS = "copernicus"  # Copernicus DEM
    CC_BY = "cc-by"  # Creative Commons Attribution
    CC_BY_SA = "cc-by-sa"  # Creative Commons Attribution-ShareAlike
    CC0 = "cc0"  # Public Domain
    PROPRIETARY = "proprietary"  # Not allowed


class DataSource(Enum):
    """Known data sources."""
    OPENSTREETMAP = "openstreetmap"
    COPERNICUS_DEM = "copernicus_dem"
    OPENAERIAL = "openaerial"
    WIKIMEDIA = "wikimedia"
    GOOGLE = "google"  # FORBIDDEN
    BING = "bing"  # FORBIDDEN
    APPLE = "apple"  # FORBIDDEN
    HERE = "here"  # FORBIDDEN


# License information for allowed sources
ALLOWED_SOURCES = {
    DataSource.OPENSTREETMAP: {
        "license": LicenseType.ODBL,
        "attribution": "© OpenStreetMap contributors",
        "url": "https://www.openstreetmap.org/copyright",
        "commercial": True,
        "derivative_allowed": True,
    },
    DataSource.COPERNICUS_DEM: {
        "license": LicenseType.COPERNICUS,
        "attribution": "© Copernicus Service Information",
        "url": "https://spacedata.copernicus.eu/",
        "commercial": True,
        "derivative_allowed": True,
    },
    DataSource.OPENAERIAL: {
        "license": LicenseType.CC_BY,
        "attribution": "© OpenAerialMap contributors",
        "url": "https://openaerialmap.org/",
        "commercial": True,
        "derivative_allowed": True,
    },
    DataSource.WIKIMEDIA: {
        "license": LicenseType.CC_BY_SA,
        "attribution": "Wikimedia Commons",
        "url": "https://commons.wikimedia.org/",
        "commercial": True,
        "derivative_allowed": True,
    },
}

# Forbidden sources
FORBIDDEN_SOURCES = {
    DataSource.GOOGLE: "Google Maps ToS prohibit derivatives and offline use",
    DataSource.BING: "Bing Maps has restrictive licensing",
    DataSource.APPLE: "Apple Maps prohibits commercial use",
    DataSource.HERE: "HERE requires paid licensing",
}


class LicenseChecker:
    """
    Utility for validating data source licenses and generating attributions.

    Ensures DSGVO/GDPR compliance and proper licensing for educational use.
    """

    @staticmethod
    def is_source_allowed(source: DataSource) -> bool:
        """Check if a data source is allowed for use."""
        return source in ALLOWED_SOURCES

    @staticmethod
    def get_forbidden_reason(source: DataSource) -> Optional[str]:
        """Get the reason why a source is forbidden."""
        return FORBIDDEN_SOURCES.get(source)

    @staticmethod
    def validate_url(url: str) -> tuple[bool, str]:
        """
        Validate if a URL is from an allowed source.

        Returns:
            Tuple of (is_allowed, message)
        """
        url_lower = url.lower()

        # Check for forbidden sources
        forbidden_patterns = {
            "google": DataSource.GOOGLE,
            "googleapis": DataSource.GOOGLE,
            "gstatic": DataSource.GOOGLE,
            "bing.com": DataSource.BING,
            "virtualearth": DataSource.BING,
            "apple.com/maps": DataSource.APPLE,
            "here.com": DataSource.HERE,
        }

        for pattern, source in forbidden_patterns.items():
            if pattern in url_lower:
                reason = FORBIDDEN_SOURCES.get(source, "Not allowed")
                return False, f"FORBIDDEN: {source.value} - {reason}"

        # Check for allowed sources
        allowed_patterns = {
            "openstreetmap": DataSource.OPENSTREETMAP,
            "tile.osm": DataSource.OPENSTREETMAP,
            "copernicus": DataSource.COPERNICUS_DEM,
            "openaerialmap": DataSource.OPENAERIAL,
            "wikimedia": DataSource.WIKIMEDIA,
        }

        for pattern, source in allowed_patterns.items():
            if pattern in url_lower:
                info = ALLOWED_SOURCES[source]
                return True, f"ALLOWED: {source.value} ({info['license'].value})"

        # Unknown source - warn but allow with custom attribution
        return True, "UNKNOWN: Verify license manually"

    @staticmethod
    def get_attribution_for_sources(
        sources: list[DataSource],
    ) -> list[dict]:
        """
        Get attribution information for a list of data sources.

        Args:
            sources: List of data sources used

        Returns:
            List of attribution dictionaries
        """
        attributions = []

        for source in sources:
            if source in ALLOWED_SOURCES:
                info = ALLOWED_SOURCES[source]
                attributions.append({
                    "name": source.value.replace("_", " ").title(),
                    "license": info["license"].value.upper(),
                    "attribution": info["attribution"],
                    "url": info["url"],
                    "required": True,
                })

        return attributions

    @staticmethod
    def generate_attribution_html(sources: list[DataSource]) -> str:
        """
        Generate HTML attribution footer for web display.

        Args:
            sources: List of data sources used

        Returns:
            HTML string with attribution
        """
        attributions = LicenseChecker.get_attribution_for_sources(sources)

        if not attributions:
            return ""

        parts = []
        for attr in attributions:
            parts.append(
                f'<a href="{attr["url"]}" target="_blank" rel="noopener">'
                f'{attr["attribution"]}</a>'
            )

        return " | ".join(parts)

    @staticmethod
    def generate_attribution_text(sources: list[DataSource]) -> str:
        """
        Generate plain text attribution.

        Args:
            sources: List of data sources used

        Returns:
            Plain text attribution string
        """
        attributions = LicenseChecker.get_attribution_for_sources(sources)

        if not attributions:
            return ""

        return " | ".join(attr["attribution"] for attr in attributions)

    @staticmethod
    def check_commercial_use(sources: list[DataSource]) -> tuple[bool, list[str]]:
        """
        Check if all sources allow commercial use.

        Returns:
            Tuple of (all_allowed, list_of_issues)
        """
        issues = []

        for source in sources:
            if source in FORBIDDEN_SOURCES:
                issues.append(f"{source.value}: {FORBIDDEN_SOURCES[source]}")
            elif source in ALLOWED_SOURCES:
                if not ALLOWED_SOURCES[source]["commercial"]:
                    issues.append(f"{source.value}: Commercial use not allowed")

        return len(issues) == 0, issues