BetterCallPraskovia/backend/src/application/use_cases/document_use_cases.py

"""
Use cases для работы с документами
"""
from uuid import UUID
from typing import BinaryIO, Optional
from src.domain.entities.document import Document
from src.domain.repositories.document_repository import IDocumentRepository
from src.domain.repositories.collection_repository import ICollectionRepository
from src.domain.repositories.collection_access_repository import ICollectionAccessRepository
from src.application.services.document_parser_service import DocumentParserService
from src.application.services.rag_service import RAGService
from src.shared.exceptions import NotFoundError, ForbiddenError


class DocumentUseCases:
    """Use cases для документов"""
    
    def __init__(
        self,
        document_repository: IDocumentRepository,
        collection_repository: ICollectionRepository,
        access_repository: ICollectionAccessRepository,
        parser_service: DocumentParserService,
        rag_service: Optional[RAGService] = None
    ):
        self.document_repository = document_repository
        self.collection_repository = collection_repository
        self.access_repository = access_repository
        self.parser_service = parser_service
        self.rag_service = rag_service
    
    async def _check_collection_access(self, user_id: UUID, collection) -> bool:
        """Проверить доступ пользователя к коллекции"""
        if collection.owner_id == user_id:
            return True
        
        if collection.is_public:
            return True
        
        access = await self.access_repository.get_by_user_and_collection(user_id, collection.collection_id)
        return access is not None
    
    async def create_document(
        self,
        collection_id: UUID,
        title: str,
        content: str,
        metadata: dict | None = None
    ) -> Document:
        """Создать документ"""
        collection = await self.collection_repository.get_by_id(collection_id)
        if not collection:
            raise NotFoundError(f"Коллекция {collection_id} не найдена")
        
        document = Document(
            collection_id=collection_id,
            title=title,
            content=content,
            metadata=metadata or {}
        )
        return await self.document_repository.create(document)
    
    async def upload_and_parse_document(
        self,
        collection_id: UUID,
        file: BinaryIO,
        filename: str,
        user_id: UUID
    ) -> Document:
        """Загрузить и распарсить документ, затем автоматически проиндексировать"""
        collection = await self.collection_repository.get_by_id(collection_id)
        if not collection:
            raise NotFoundError(f"Коллекция {collection_id} не найдена")
        
        has_access = await self._check_collection_access(user_id, collection)
        if not has_access:
            raise ForbiddenError("У вас нет доступа к этой коллекции")
        
        title, content = await self.parser_service.parse_pdf(file, filename)
        
        document = Document(
            collection_id=collection_id,
            title=title,
            content=content,
            metadata={"filename": filename}
        )
        document = await self.document_repository.create(document)

        if self.rag_service:
            try:
                await self.rag_service.index_document(document)
            except Exception as e:
                print(f"Ошибка при автоматической индексации документа {document.document_id}: {e}")
        
        return document
    
    async def get_document(self, document_id: UUID) -> Document:
        """Получить документ по ID"""
        document = await self.document_repository.get_by_id(document_id)
        if not document:
            raise NotFoundError(f"Документ {document_id} не найден")
        return document
    
    async def update_document(
        self,
        document_id: UUID,
        user_id: UUID,
        title: str | None = None,
        content: str | None = None,
        metadata: dict | None = None
    ) -> Document:
        """Обновить документ"""
        document = await self.get_document(document_id)
        
        collection = await self.collection_repository.get_by_id(document.collection_id)
        if not collection:
            raise NotFoundError(f"Коллекция {document.collection_id} не найдена")
        has_access = await self._check_collection_access(user_id, collection)
        if not has_access:
            raise ForbiddenError("У вас нет доступа к этой коллекции")
        
        if title is not None:
            document.title = title
        if content is not None:
            document.content = content
        if metadata is not None:
            document.metadata = metadata
        
        return await self.document_repository.update(document)
    
    async def delete_document(self, document_id: UUID, user_id: UUID) -> bool:
        """Удалить документ"""
        document = await self.get_document(document_id)
        
        collection = await self.collection_repository.get_by_id(document.collection_id)
        if not collection or collection.owner_id != user_id:
            raise ForbiddenError("Только владелец коллекции может удалять документы")
        
        return await self.document_repository.delete(document_id)
    
    async def list_collection_documents(self, collection_id: UUID, skip: int = 0, limit: int = 100) -> list[Document]:
        """Получить документы коллекции"""
        collection = await self.collection_repository.get_by_id(collection_id)
        if not collection:
            raise NotFoundError(f"Коллекция {collection_id} не найдена")
        
        return await self.document_repository.list_by_collection(collection_id, skip=skip, limit=limit)