services/knowledge/upload_file.py

from fastapi import UploadFile
# from sqlalchemy.ext.asyncio import AsyncSession
from sqlalchemy.exc import IntegrityError
from externals.storages.azure_blob import upload_pdf as upload_file
from externals.databases.pg_crud import get_file_by_filename, get_file_by_user_id
from externals.databases.pg_models import CVFile, CVUser
from utils.logger import get_logger

logger = get_logger("knowledge.upload")


class KnowledgeFileService:
    def __init__(self, db, user: CVUser):
        self.db = db
        self.user = user

    async def upload(self, file: UploadFile) -> CVFile:
        """
        Upload PDF to Azure Blob Storage and log metadata into cv_file table
        """

        logger.info(
            "📤 Knowledge upload requested",
            extra={"context": {
                "cv_filename": file.filename,
                "file_content_type": file.content_type,
                "user_id": str(self.user.user_id),
                "tenant_id": str(self.user.tenant_id),
            }}
        )

        # 1️⃣ Check DB first (idempotent behavior)
        existing = await get_file_by_filename(self.db, 
                                              filename=file.filename,
                                              user_id=self.user.user_id)
        print(f"DEBUG: KnowledgeFileService/get_file_by_filename/existing: {existing}")
        if existing:
            logger.info(
                "ℹ️ File already exists, skipping upload",
                extra={"context": {
                    "filename": file.filename,
                    "file_id": str(existing.file_id),
                }}
            )
            return existing
        else:
            # 2️⃣ Upload to Azure Blob
            upload_result = await upload_file(file, self.user.tenant_id, self.user.user_id)
            print(f"DEBUG: KnowledgeFileService/upload_result: {upload_result}")

            if not upload_result.get("uploaded"):
                logger.error(
                    "❌ Azure upload failed",
                    extra={"context": {"filename": file.filename}},
                )
                raise RuntimeError(upload_result.get("message", "Upload failed"))


            # 3️⃣ Persist metadata into DB
            cv_file = CVFile(
                user_id=self.user.user_id,
                file_type=file.content_type or "application/pdf",
                filename=file.filename,
                url=upload_result["url"],
                is_extracted=False,
                is_deleted=False,
            )

            try:
                self.db.add(cv_file)
                await self.db.commit()
                await self.db.refresh(cv_file)

                logger.info("✅ File successfully logged", extra={"context": {
                    "file_id": str(cv_file.file_id),
                    "filename": cv_file.filename,
                    "uploaded_by": self.user.email,
                }})

                return cv_file

            except IntegrityError:
                await self.db.rollback()

                logger.warning(
                    "⚠️ Duplicate file detected during insert",
                    extra={"context": {"filename": file.filename}},
                )

                # Fetch the already-existing row
                existing = await get_file_by_filename(self.db, 
                                                    filename=file.filename,
                                                    user_id=self.user.user_id)
                if existing:
                    return existing


    async def get_files_by_user(self, user_id: str):
        """
        Retrieve all file metadata uploaded by a specific user.
        """

        logger.info(
            "📄 Fetching files for user",
            extra={"context": {
                "requested_user_id": user_id,
                "requested_by": str(self.user.user_id),
            }}
        )

        files = await get_file_by_user_id(
            self.db,
            user_id=user_id
        )

        file_list = []
        for file in files:
            file_list.append({
                "file_id": str(file.file_id),
                "filename": file.filename,
                "file_type": file.file_type,
                "url": file.url,
                "is_extracted": file.is_extracted,
                "uploaded_at": file.uploaded_at.isoformat(),
                "date_modified": file.date_modified.isoformat(),
            })

        logger.info(
            "✅ Files retrieved successfully",
            extra={"context": {
                "file_count": len(file_list),
                "requested_user_id": user_id,
            }}
        )

        return file_list

# class KnowledgeUploadService:

#     def __init__(self, db: AsyncSession, user: CVUser):
#         self.db = db
#         self.user = user

#     async def upload(self, file: UploadFile) -> CVFile:
#         """
#         Upload file to Azure Blob and log metadata to cv_file table
#         """

#         logger.info(f"📤 Uploading file: {file.filename}")

#         # 1️⃣ Upload to Azure Blob
#         upload_result = await upload_file(file)

#         if not upload_result.get("uploaded"):
#             raise RuntimeError(upload_result.get("message"))

#         # 2️⃣ Check existing file record
#         existing = await get_file_by_filename(self.db, file.filename)
#         if existing:
#             logger.info(f"ℹ️ File already exists in DB: {file.filename}")
#             return existing

#         # 3️⃣ Persist metadata
#         cv_file = CVFile(
#             file_type=file.content_type or "application/pdf",
#             filename=file.filename,
#             url=upload_result["url"],
#             is_extracted=False,
#         )

#         self.db.add(cv_file)
#         await self.db.commit()
#         await self.db.refresh(cv_file)

#         logger.info(f"✅ File logged into cv_file: {file.filename}")
#         return cv_file