pounce/backend/app/api/seo.py

"""
SEO Data API - "SEO Juice Detector"

This implements Strategie 3 from analysis_3.md:
"Das Feature: 'SEO Juice Detector'
Wenn eine Domain droppt, prüfst du nicht nur den Namen, 
sondern ob Backlinks existieren.
Monetarisierung: Das ist ein reines Tycoon-Feature ($29/Monat)."

Endpoints:
- GET /seo/{domain} - Get SEO data for a domain (TYCOON ONLY)
- POST /seo/batch - Analyze multiple domains (TYCOON ONLY)
"""
import logging
from typing import List
from fastapi import APIRouter, Depends, HTTPException, Query
from pydantic import BaseModel
from sqlalchemy.ext.asyncio import AsyncSession

from app.database import get_db
from app.api.deps import get_current_user
from app.models.user import User
from app.services.seo_analyzer import seo_analyzer

logger = logging.getLogger(__name__)
router = APIRouter()


# ============== Schemas ==============

class SEOMetrics(BaseModel):
    domain_authority: int | None
    page_authority: int | None
    spam_score: int | None
    total_backlinks: int | None
    referring_domains: int | None


class NotableLinks(BaseModel):
    has_wikipedia: bool
    has_gov: bool
    has_edu: bool
    has_news: bool
    notable_domains: List[str]


class BacklinkInfo(BaseModel):
    domain: str
    authority: int
    page: str = ""


class SEOResponse(BaseModel):
    domain: str
    seo_score: int
    value_category: str
    metrics: SEOMetrics
    notable_links: NotableLinks
    top_backlinks: List[BacklinkInfo]
    estimated_value: float | None
    data_source: str
    last_updated: str | None
    is_estimated: bool


class BatchSEORequest(BaseModel):
    domains: List[str]


class BatchSEOResponse(BaseModel):
    results: List[SEOResponse]
    total_requested: int
    total_processed: int


# ============== Helper ==============

def _check_tycoon_access(user: User) -> None:
    """Verify user has Tycoon tier access."""
    if not user.subscription:
        raise HTTPException(
            status_code=403,
            detail="SEO data is a Tycoon feature. Please upgrade your subscription."
        )
    
    tier = user.subscription.tier.lower() if user.subscription.tier else ""
    if tier != "tycoon":
        raise HTTPException(
            status_code=403,
            detail="SEO data is a Tycoon-only feature. Please upgrade to access backlink analysis."
        )


# ============== Endpoints ==============

@router.get("/{domain}", response_model=SEOResponse)
async def get_seo_data(
    domain: str,
    force_refresh: bool = Query(False, description="Force refresh from API"),
    current_user: User = Depends(get_current_user),
    db: AsyncSession = Depends(get_db),
):
    """
    Get SEO data for a domain.
    
    TYCOON FEATURE ONLY.
    
    Returns:
    - Domain Authority (0-100)
    - Page Authority (0-100)
    - Spam Score (0-100)
    - Total Backlinks
    - Referring Domains
    - Notable links (Wikipedia, .gov, .edu, news sites)
    - Top backlinks with authority scores
    - Estimated SEO value
    
    From analysis_3.md:
    "Domain `alte-bäckerei-münchen.de` ist frei. 
    Hat Links von `sueddeutsche.de` und `wikipedia.org`."
    """
    # Check Tycoon access
    _check_tycoon_access(current_user)
    
    # Clean domain input
    domain = domain.lower().strip()
    if domain.startswith('http://'):
        domain = domain[7:]
    if domain.startswith('https://'):
        domain = domain[8:]
    if domain.startswith('www.'):
        domain = domain[4:]
    domain = domain.rstrip('/')
    
    # Get SEO data
    result = await seo_analyzer.analyze_domain(domain, db, force_refresh)
    
    return SEOResponse(**result)


@router.post("/batch", response_model=BatchSEOResponse)
async def batch_seo_analysis(
    request: BatchSEORequest,
    current_user: User = Depends(get_current_user),
    db: AsyncSession = Depends(get_db),
):
    """
    Analyze multiple domains for SEO data.
    
    TYCOON FEATURE ONLY.
    
    Limited to 10 domains per request to prevent abuse.
    """
    # Check Tycoon access
    _check_tycoon_access(current_user)
    
    # Limit batch size
    domains = request.domains[:10]
    
    results = []
    for domain in domains:
        try:
            # Clean domain
            domain = domain.lower().strip()
            if domain.startswith('http://'):
                domain = domain[7:]
            if domain.startswith('https://'):
                domain = domain[8:]
            if domain.startswith('www.'):
                domain = domain[4:]
            domain = domain.rstrip('/')
            
            result = await seo_analyzer.analyze_domain(domain, db)
            results.append(SEOResponse(**result))
        except Exception as e:
            logger.error(f"Error analyzing {domain}: {e}")
            # Skip failed domains
            continue
    
    return BatchSEOResponse(
        results=results,
        total_requested=len(request.domains),
        total_processed=len(results),
    )


@router.get("/{domain}/quick")
async def get_seo_quick_summary(
    domain: str,
    current_user: User = Depends(get_current_user),
    db: AsyncSession = Depends(get_db),
):
    """
    Get a quick SEO summary for a domain.
    
    This is a lighter version that shows basic metrics without full backlink analysis.
    Available to Trader+ users.
    """
    # Check at least Trader access
    if not current_user.subscription:
        raise HTTPException(
            status_code=403,
            detail="SEO data requires a paid subscription."
        )
    
    tier = current_user.subscription.tier.lower() if current_user.subscription.tier else ""
    if tier == "scout":
        raise HTTPException(
            status_code=403,
            detail="SEO data requires Trader or higher subscription."
        )
    
    # Clean domain
    domain = domain.lower().strip().rstrip('/')
    if domain.startswith('http://'):
        domain = domain[7:]
    if domain.startswith('https://'):
        domain = domain[8:]
    if domain.startswith('www.'):
        domain = domain[4:]
    
    result = await seo_analyzer.analyze_domain(domain, db)
    
    # Return limited data for non-Tycoon
    if tier != "tycoon":
        return {
            'domain': result['domain'],
            'seo_score': result['seo_score'],
            'value_category': result['value_category'],
            'domain_authority': result['metrics']['domain_authority'],
            'has_notable_links': (
                result['notable_links']['has_wikipedia'] or 
                result['notable_links']['has_gov'] or 
                result['notable_links']['has_news']
            ),
            'is_estimated': result['is_estimated'],
            'upgrade_for_details': True,
            'message': "Upgrade to Tycoon for full backlink analysis"
        }
    
    return result