Aniworld/src/server/services/setup_service.py

"""Setup service for first-time database initialization.

This service runs during initial application setup to:
1. Scan anime folders in the data directory
2. Extract title and year from folder names
3. Create AnimeSeries records in the database
4. Resolve provider keys via search (if single match found)

The run_once logic is handled by the caller (perform_initial_setup)
via _check_initial_scan_status, not by this service itself.
"""
import os
import re
from dataclasses import dataclass
from datetime import datetime, timezone
from pathlib import Path
from typing import Optional

import structlog

from src.config.settings import settings
from src.server.database.connection import get_db_session
from src.server.database.service import AnimeSeriesService, UnresolvedFolderService
from src.server.utils.dependencies import get_series_app

logger = structlog.get_logger(__name__)


@dataclass
class SeriesProperties:
    """Filesystem-derived properties for an AnimeSeries."""
    has_nfo: bool = False
    nfo_path: Optional[str] = None
    nfo_created_at: Optional[datetime] = None
    nfo_updated_at: Optional[datetime] = None
    logo_loaded: bool = False
    images_loaded: bool = False


class SetupService:
    """Service for setup operations during application initialization."""

    @staticmethod
    def _extract_year_from_folder_name(folder_name: str) -> Optional[int]:
        """Extract year from folder name if present.
        
        Looks for year in format "(YYYY)" at the end of folder name.
        
        Args:
            folder_name: The folder name to parse
            
        Returns:
            Year as integer if found, None otherwise
        """
        if not folder_name:
            return None
        
        match = re.search(r'\((\d{4})\)', folder_name)
        if match:
            year = int(match.group(1))
            if 1900 <= year <= 2100:
                return year
        return None

    @staticmethod
    def _extract_title_from_folder_name(folder_name: str) -> str:
        """Extract title from folder name by removing year suffix.
        
        Args:
            folder_name: The folder name to parse
            
        Returns:
            Title with year suffix and surrounding whitespace removed
        """
        return re.sub(r'\s*\(\d{4}\)\s*$', '', folder_name).strip()

    @staticmethod
    def _normalize_title(title: str) -> str:
        """Normalize title for fuzzy matching.
        
        Strips common suffixes and lowercases for comparison.
        
        Args:
            title: The title to normalize
            
        Returns:
            Normalized title string
        """
        # Remove common anime suffixes (case-insensitive)
        suffixes = [
            r'\s*\(TV\)\s*$',
            r'\s*\(Anime\)\s*$',
            r'\s*\(OAD\)\s*$',
            r'\s*\(OVA\)\s*$',
            r'\s*\(Special\)\s*$',
            r'\s*\(Movie\)\s*$',
            r'\s*\(Spin-Off\)\s*$',
        ]
        normalized = title.lower().strip()
        for suffix_pattern in suffixes:
            normalized = re.sub(suffix_pattern, '', normalized, flags=re.IGNORECASE).strip()
        return normalized

    @staticmethod
    def _titles_match(title1: str, title2: str, threshold: float = 0.85) -> bool:
        """Check if two titles match using fuzzy comparison.
        
        Args:
            title1: First title
            title2: Second title
            threshold: Similarity threshold (0.0 to 1.0)
            
        Returns:
            True if titles match within threshold
        """
        norm1 = SetupService._normalize_title(title1)
        norm2 = SetupService._normalize_title(title2)
        
        # Direct match after normalization
        if norm1 == norm2:
            return True
        
        # Containment check (e.g., "Attack on Titan" in "Attack on Titan (TV)")
        if norm1 in norm2 or norm2 in norm1:
            return True
        
        # Similarity ratio check using SequenceMatcher
        from difflib import SequenceMatcher
        ratio = SequenceMatcher(None, norm1, norm2).ratio()
        return ratio >= threshold

    @staticmethod
    async def _resolve_key_via_search(title: str) -> str:
        """Resolve provider key by searching for the title.
        
        Args:
            title: The title to search for
            
        Returns:
            Provider key if exactly one match with same name found,
            empty string otherwise
        """
        if not title:
            return ""
        
        try:
            series_app = get_series_app()
            results = await series_app.search(title)
            
            if len(results) == 1:
                result_name = results[0].get('title', '')
                result_link = results[0].get('link', '')
                
                if SetupService._titles_match(result_name, title):
                    if result_link and '/anime/stream/' in result_link:
                        return result_link.split('/anime/stream/')[-1].split('/')[0]
                    elif result_link:
                        # Link is already the key (e.g., "shinobi-no-ittoki")
                        return result_link
                    else:
                        logger.debug(
                            "Series key resolved but link format unexpected",
                            folder_title=title,
                            result_title=result_name,
                            link=result_link
                        )
                else:
                    logger.debug(
                        "Series search result title mismatch",
                        folder_title=title,
                        result_title=result_name,
                        link=result_link
                    )
            elif len(results) > 1:
                logger.debug(
                    "Multiple search results for title, skipping fuzzy match",
                    title=title,
                    result_count=len(results)
                )
        except Exception as e:
            logger.warning(
                "Provider search failed for folder",
                title=title,
                error=str(e)
            )
        
        return ""

    @staticmethod
    def _check_nfo_file(folder_path: Path) -> tuple[bool, Optional[str], Optional[datetime], Optional[datetime]]:
        """Check if tvshow.nfo exists and return its metadata.
        
        Args:
            folder_path: Path to the series folder
            
        Returns:
            Tuple of (has_nfo, nfo_path, nfo_created_at, nfo_updated_at)
        """
        nfo_path = folder_path / "tvshow.nfo"
        if nfo_path.is_file():
            stat = nfo_path.stat()
            created = datetime.fromtimestamp(stat.st_ctime, tz=timezone.utc)
            updated = datetime.fromtimestamp(stat.st_mtime, tz=timezone.utc)
            return True, str(nfo_path), created, updated
        return False, None, None, None

    @staticmethod
    def _check_logo_file(folder_path: Path) -> bool:
        """Check if logo.png exists.
        
        Args:
            folder_path: Path to the series folder
            
        Returns:
            True if logo.png exists, False otherwise
        """
        return (folder_path / "logo.png").is_file()

    @staticmethod
    def _check_image_files(folder_path: Path) -> bool:
        """Check if any image files (poster, fanart) exist.
        
        Args:
            folder_path: Path to the series folder
            
        Returns:
            True if any poster.jpg/jpeg/png or fanart.jpg/jpeg/png exists
        """
        image_extensions = {'.jpg', '.jpeg', '.png'}
        for child in folder_path.iterdir():
            if child.is_file():
                name_lower = child.name.lower()
                if name_lower.startswith(('poster', 'fanart')) and child.suffix.lower() in image_extensions:
                    return True
        return False

    @classmethod
    def _get_series_properties(cls, folder_path: Path) -> SeriesProperties:
        """Get all filesystem-derived properties for a series folder.
        
        Args:
            folder_path: Path to the series folder
            
        Returns:
            SeriesProperties with all detected values
        """
        has_nfo, nfo_path, nfo_created_at, nfo_updated_at = cls._check_nfo_file(folder_path)
        logo_loaded = cls._check_logo_file(folder_path)
        images_loaded = cls._check_image_files(folder_path)
        
        return SeriesProperties(
            has_nfo=has_nfo,
            nfo_path=nfo_path,
            nfo_created_at=nfo_created_at,
            nfo_updated_at=nfo_updated_at,
            logo_loaded=logo_loaded,
            images_loaded=images_loaded,
        )

    @classmethod
    async def run(cls) -> int:
        """Run the setup service.
        
        Scans anime folders, creates AnimeSeries records, and resolves
        provider keys via search. Should only be called after checking
        that initial scan hasn't been completed yet (via _check_initial_scan_status).
        
        Returns:
            Number of new series created
        """
        if not settings.anime_directory:
            logger.info("Anime directory not configured, skipping setup")
            return 0
        
        anime_dir = Path(settings.anime_directory)
        if not anime_dir.is_dir():
            logger.info(
                "Anime directory does not exist, skipping setup: %s",
                anime_dir
            )
            return 0
        
        logger.info("Running setup service...")
        
        created_count = 0
        skipped_existing = 0
        unresolved_count = 0
        
        try:
            series_app = get_series_app()
            
            async with get_db_session() as db:
                for folder in anime_dir.iterdir():
                    if not folder.is_dir():
                        continue
                    
                    folder_name = folder.name
                    
                    # Check if series already exists in DB
                    existing = await AnimeSeriesService.get_by_folder(
                        db, folder_name
                    )
                    if existing:
                        skipped_existing += 1
                        continue
                    
                    # Check if already tracked as unresolved
                    existing_unresolved = await UnresolvedFolderService.get_by_folder_name(
                        db, folder_name
                    )
                    if existing_unresolved and existing_unresolved.is_resolved:
                        # Was previously unresolved but now resolved - create the series
                        resolved_key = existing_unresolved.provider_key
                        year = cls._extract_year_from_folder_name(folder_name)
                        title = cls._extract_title_from_folder_name(folder_name)
                        props = cls._get_series_properties(folder)
                        
                        series = await AnimeSeriesService.create(
                            db=db,
                            key=resolved_key,
                            name=title,
                            site="https://aniworld.to",
                            folder=folder_name,
                            year=year,
                            loading_status="completed",
                            episodes_loaded=True,
                            logo_loaded=props.logo_loaded,
                            images_loaded=props.images_loaded,
                            has_nfo=props.has_nfo,
                            nfo_path=props.nfo_path,
                            nfo_created_at=props.nfo_created_at,
                            nfo_updated_at=props.nfo_updated_at,
                        )
                        created_count += 1
                        
                        # Delete the unresolved tracking now that series is created
                        await UnresolvedFolderService.delete(db, folder_name)
                        continue
                    elif existing_unresolved:
                        # Already tracked as unresolved, skip
                        unresolved_count += 1
                        continue
                    
                    # Extract title and year from folder name
                    year = cls._extract_year_from_folder_name(folder_name)
                    title = cls._extract_title_from_folder_name(folder_name)
                    
                    if not title:
                        logger.warning(
                            "Could not extract title from folder: %s",
                            folder_name
                        )
                        continue
                    
                    # Resolve key via provider search
                    resolved_key = await cls._resolve_key_via_search(title)
                    
                    if not resolved_key:
                        # Track unresolved folder for later manual resolution
                        import json
                        try:
                            series_results = await series_app.search(title)
                            search_result_json = json.dumps(series_results) if series_results else None
                        except Exception:
                            search_result_json = None
                        
                        await UnresolvedFolderService.create(
                            db=db,
                            folder_name=folder_name,
                            title=title,
                            year=year,
                            search_attempts=1,
                            last_search_result=search_result_json,
                        )
                        logger.warning(
                            "Could not resolve series key for folder, tracking as unresolved: %s",
                            folder_name
                        )
                        continue
                    
                    # Also check if a series with this key already exists (different folder, same anime)
                    existing_by_key = await AnimeSeriesService.get_by_key(db, resolved_key)
                    if existing_by_key:
                        logger.debug(
                            "Series with key already exists, skipping",
                            folder=folder_name,
                            key=resolved_key,
                            existing_folder=existing_by_key.folder
                        )
                        skipped_existing += 1
                        continue
                    
                    # Check filesystem properties
                    props = cls._get_series_properties(folder)
                    
                    # Create AnimeSeries record
                    series = await AnimeSeriesService.create(
                        db=db,
                        key=resolved_key,
                        name=title,
                        site="https://aniworld.to",
                        folder=folder_name,
                        year=year,
                        loading_status="completed",
                        episodes_loaded=True,
                        logo_loaded=props.logo_loaded,
                        images_loaded=props.images_loaded,
                        has_nfo=props.has_nfo,
                        nfo_path=props.nfo_path,
                        nfo_created_at=props.nfo_created_at,
                        nfo_updated_at=props.nfo_updated_at,
                    )
                    created_count += 1
                    
                    logger.debug(
                        "Created series from folder",
                        folder=folder_name,
                        title=title,
                        year=year,
                        key=resolved_key or "(unresolved)",
                        has_nfo=props.has_nfo,
                        logo_loaded=props.logo_loaded,
                        images_loaded=props.images_loaded,
                    )
            
            logger.info(
                "Setup complete",
                created=created_count,
                skipped_existing=skipped_existing,
                unresolved=unresolved_count
            )
            
        except Exception as e:
            logger.error(
                "Setup failed",
                error=str(e),
                exc_info=True
            )
            return created_count
        
        return created_count