"""
Intake Extractor — derives UCCA intake flags from DETECTED SERVICES,
not from website text content.

The actual data processing happens through APIs, scripts, and cookies —
NOT through visible text on the page. A news website reporting about
healthcare does NOT process health data.

Flags are derived deterministically from:
1. Which third-party services are embedded (Google Analytics → tracking)
2. Which payment providers are used (Stripe → payment_data)
3. Which CDN/fonts are loaded (Google Fonts → cross_border_transfer)
"""

import logging

logger = logging.getLogger(__name__)

# Service category → intake flags mapping
# This is the ONLY source of truth for what a service implies
SERVICE_TO_FLAGS: dict[str, dict[str, bool]] = {
    # Tracking & Analytics → personal_data + tracking
    "tracking": {
        "personal_data": True,
        "tracking": True,
    },
    # Marketing → marketing + tracking + third_party_sharing
    "marketing": {
        "personal_data": True,
        "tracking": True,
        "marketing": True,
        "third_party_sharing": True,
    },
    # Heatmap/Session Recording → tracking + profiling
    "heatmap": {
        "personal_data": True,
        "tracking": True,
        "profiling": True,
    },
    # Payment → payment_data
    "payment": {
        "personal_data": True,
        "payment_data": True,
    },
    # Chatbot → personal_data (user sends messages)
    "chatbot": {
        "personal_data": True,
        "customer_data": True,
    },
    # CRM → customer_data + profiling
    "crm": {
        "personal_data": True,
        "customer_data": True,
        "profiling": True,
    },
    # CDN from non-EU → cross_border_transfer (IP sent to US)
    "cdn": {
        "personal_data": True,
    },
}

# Specific services with special flags
SPECIFIC_SERVICE_FLAGS: dict[str, dict[str, bool]] = {
    "klarna": {"automated_decisions": True, "payment_data": True},
    "paypal": {"cross_border_transfer": True, "payment_data": True},
    "stripe": {"cross_border_transfer": True, "payment_data": True},
    "google_analytics": {"cross_border_transfer": True, "tracking": True},
    "facebook_pixel": {"cross_border_transfer": True, "marketing": True, "profiling": True},
    "hotjar": {"profiling": True, "tracking": True},
    "ms_clarity": {"cross_border_transfer": True, "profiling": True},
    "tiktok_pixel": {"cross_border_transfer": True, "marketing": True},
    "intercom": {"cross_border_transfer": True, "ai_usage": True},
}


def extract_intake_flags_from_services(detected_services: list[dict]) -> dict:
    """Derive intake flags from detected third-party services.

    This is deterministic and 100% accurate — if Google Analytics is
    embedded, tracking IS happening. No guessing needed.
    """
    flags = {
        "personal_data": False,
        "customer_data": False,
        "payment_data": False,
        "location_data": False,
        "biometric_data": False,
        "minor_data": False,
        "health_data": False,
        "marketing": False,
        "profiling": False,
        "automated_decisions": False,
        "third_party_sharing": False,
        "cross_border_transfer": False,
        "tracking": False,
        "ai_usage": False,
    }

    for svc in detected_services:
        category = svc.get("category", "other")
        service_id = svc.get("id", "")
        eu_adequate = svc.get("eu_adequate", True)

        # Apply category-level flags
        cat_flags = SERVICE_TO_FLAGS.get(category, {})
        for key, value in cat_flags.items():
            if value:
                flags[key] = True

        # Apply service-specific flags
        svc_flags = SPECIFIC_SERVICE_FLAGS.get(service_id, {})
        for key, value in svc_flags.items():
            if value:
                flags[key] = True

        # Non-EU service → cross_border_transfer
        if not eu_adequate:
            flags["cross_border_transfer"] = True
            flags["third_party_sharing"] = True

    # Any website with detected services processes personal data (IP at minimum)
    if detected_services:
        flags["personal_data"] = True

    active = {k: v for k, v in flags.items() if v}
    logger.info("Intake flags from %d services: %s", len(detected_services), active)
    return flags


# Keep backward compatibility
async def extract_intake_flags(text: str) -> dict:
    """DEPRECATED — use extract_intake_flags_from_services() instead.

    This function used LLM to guess flags from text content.
    Text content does NOT represent actual data processing.
    """
    logger.warning(
        "extract_intake_flags(text) called — DEPRECATED. "
        "Use extract_intake_flags_from_services(detected_services) instead."
    )
    # Return minimal flags — website exists = personal_data (IP)
    return {"personal_data": True, "tracking": False}


def flags_to_ucca_intake(flags: dict) -> dict:
    """Convert extracted flags to UCCA intake format."""
    return {
        "data_types": {
            "personal_data": flags.get("personal_data", False),
            "customer_data": flags.get("customer_data", False),
            "location_data": flags.get("location_data", False),
            "biometric_data": flags.get("biometric_data", False),
            "minor_data": flags.get("minor_data", False),
            "images": False,
            "audio": False,
            "financial_data": flags.get("payment_data", False),
            "employee_data": False,
            "article_9_data": flags.get("health_data", False) or flags.get("biometric_data", False),
        },
        "purpose": {
            "marketing": flags.get("marketing", False),
            "analytics": flags.get("tracking", False),
            "profiling": flags.get("profiling", False),
            "automation": flags.get("ai_usage", False),
            "customer_support": False,
            "evaluation_scoring": flags.get("automated_decisions", False),
            "decision_making": flags.get("automated_decisions", False),
        },
        "automation": "fully_automated" if flags.get("automated_decisions") else
                      "partially_automated" if flags.get("ai_usage") else "manual",
        "outputs": {
            "recommendations_to_users": flags.get("profiling", False),
            "data_export": flags.get("cross_border_transfer", False),
            "legal_effects": flags.get("automated_decisions", False),
        },
        "hosting": {
            "region": "non_eu" if flags.get("cross_border_transfer") else "eu",
        },
    }