company-ai-platform/app/modules/personalization/services/preferences.py

import json
import re
import unicodedata
from typing import Any

from fastapi import HTTPException, status
from pydantic import ValidationError
from sqlalchemy import delete, select
from sqlalchemy.exc import IntegrityError
from sqlalchemy.orm import Session

from app.modules.personalization.constants import (
    PREFERENCE_MAX_VALUE_LENGTH,
    PREFERENCE_SIGNAL_TERMS,
    SENSITIVE_PREFERENCE_TERMS,
    UNAVAILABLE_AI_PROVIDERS,
    PreferenceCategory,
    PreferenceSource,
)
from app.modules.personalization.models import UserPreference
from app.modules.personalization.schemas import PreferenceExtractionPayload


class PreferenceService:
    """Manage explicit and safely extracted preferences inside one owner boundary."""

    def __init__(self, db: Session):
        self.db = db

    def list_preferences(
        self,
        owner_id: int,
        category: str | PreferenceCategory | None = None,
    ) -> list[dict[str, Any]]:
        _validate_owner(owner_id)
        stmt = (
            select(UserPreference)
            .where(UserPreference.owner_id == owner_id)
            .order_by(UserPreference.category.asc(), UserPreference.id.asc())
        )
        if category is not None:
            stmt = stmt.where(UserPreference.category == _category_value(category))
        return [_serialize(item) for item in self.db.execute(stmt).scalars()]

    def upsert(
        self,
        owner_id: int,
        category: str | PreferenceCategory,
        value: str,
        source: str | PreferenceSource = PreferenceSource.EXPLICIT,
        *,
        commit: bool = True,
    ) -> dict[str, Any]:
        """Create one owner-scoped preference or reuse its normalized equivalent."""

        _validate_owner(owner_id)
        category_value, clean_value, normalized = validate_preference(category, value)
        source_value = _source_value(source)
        existing = self.db.execute(
            select(UserPreference).where(
                UserPreference.owner_id == owner_id,
                UserPreference.category == category_value,
                UserPreference.normalized_value == normalized,
            )
        ).scalar_one_or_none()
        if existing is not None:
            existing.value = clean_value
            if source_value == PreferenceSource.EXPLICIT:
                existing.source = source_value
            if commit:
                self.db.commit()
                self.db.refresh(existing)
            return _serialize(existing)

        record = UserPreference(
            owner_id=owner_id,
            category=category_value,
            value=clean_value,
            normalized_value=normalized,
            source=source_value,
        )
        try:
            with self.db.begin_nested():
                self.db.add(record)
                self.db.flush()
        except IntegrityError:
            record = self.db.execute(
                select(UserPreference).where(
                    UserPreference.owner_id == owner_id,
                    UserPreference.category == category_value,
                    UserPreference.normalized_value == normalized,
                )
            ).scalar_one()
        if commit:
            self.db.commit()
            self.db.refresh(record)
        return _serialize(record)

    def update(
        self,
        owner_id: int,
        code: str,
        *,
        category: str | PreferenceCategory | None = None,
        value: str | None = None,
    ) -> dict[str, Any]:
        record = self._owned_record(owner_id, code)
        next_category = category if category is not None else record.category
        next_value = value if value is not None else record.value
        category_value, clean_value, normalized = validate_preference(
            next_category,
            next_value,
        )
        try:
            with self.db.begin_nested():
                record.category = category_value
                record.value = clean_value
                record.normalized_value = normalized
                self.db.flush()
        except IntegrityError as exc:
            raise HTTPException(
                status_code=status.HTTP_409_CONFLICT,
                detail="Preference already exists",
            ) from exc
        self.db.commit()
        self.db.refresh(record)
        return _serialize(record)

    def delete(self, owner_id: int, code: str) -> None:
        record = self._owned_record(owner_id, code)
        self.db.delete(record)
        self.db.commit()

    def delete_matching(
        self,
        owner_id: int,
        *,
        category: str | PreferenceCategory,
        value: str,
    ) -> bool:
        category_value, _, normalized = validate_preference(category, value)
        record = self.db.execute(
            select(UserPreference).where(
                UserPreference.owner_id == owner_id,
                UserPreference.category == category_value,
                UserPreference.normalized_value == normalized,
            )
        ).scalar_one_or_none()
        if record is None:
            return False
        self.db.delete(record)
        self.db.commit()
        return True

    def save_auto_extraction(
        self,
        owner_id: int,
        *,
        provider_name: str,
        user_text: str,
        structured_payload: str | dict[str, Any] | list[Any],
    ) -> list[dict[str, Any]]:
        """Persist only allowlisted, non-sensitive output from a real AI provider."""

        if provider_name.strip().lower() in UNAVAILABLE_AI_PROVIDERS:
            return []
        if not contains_preference_signal(user_text):
            return []
        candidates = _parse_extraction_payload(structured_payload)
        saved: list[dict[str, Any]] = []
        for candidate in candidates:
            try:
                saved.append(
                    self.upsert(
                        owner_id=owner_id,
                        category=candidate.category,
                        value=candidate.value,
                        source=PreferenceSource.AUTO,
                        commit=False,
                    )
                )
            except HTTPException:
                # Automatic extraction is intentionally silent. Invalid or sensitive
                # candidates are discarded without creating a rejected profile row.
                continue
        if saved:
            self.db.commit()
        return saved

    def delete_owner_preferences(self, owner_id: int) -> int:
        """Stage deletion of every preference for an owner."""

        result = self.db.execute(
            delete(UserPreference).where(UserPreference.owner_id == owner_id)
        )
        return max(0, int(result.rowcount or 0))

    def _owned_record(self, owner_id: int, code: str) -> UserPreference:
        _validate_owner(owner_id)
        record = self.db.execute(
            select(UserPreference).where(
                UserPreference.owner_id == owner_id,
                UserPreference.code == code,
            )
        ).scalar_one_or_none()
        if record is None:
            # Deliberately indistinguishable from a nonexistent code.
            raise HTTPException(
                status_code=status.HTTP_404_NOT_FOUND,
                detail="Preference not found",
            )
        return record


def contains_preference_signal(value: str) -> bool:
    text = unicodedata.normalize("NFKC", value).casefold()
    return any(term in text for term in PREFERENCE_SIGNAL_TERMS)


def validate_preference(
    category: str | PreferenceCategory,
    value: str,
) -> tuple[str, str, str]:
    category_value = _category_value(category)
    clean_value = _clean_value(value)
    if is_sensitive_preference(clean_value):
        raise HTTPException(
            status_code=status.HTTP_422_UNPROCESSABLE_ENTITY,
            detail="Sensitive preference content is not allowed",
        )
    return category_value, clean_value, normalize_preference_value(clean_value)


def normalize_preference_value(value: str) -> str:
    normalized = unicodedata.normalize("NFKC", value)
    normalized = re.sub(r"\s+", " ", normalized).strip()
    return normalized.casefold()


def is_sensitive_preference(value: str) -> bool:
    normalized = unicodedata.normalize("NFKC", value).casefold()
    return any(term.casefold() in normalized for term in SENSITIVE_PREFERENCE_TERMS)


def _parse_extraction_payload(
    payload: str | dict[str, Any] | list[Any],
) -> list[Any]:
    parsed: Any = payload
    if isinstance(payload, str):
        try:
            parsed = json.loads(payload)
        except json.JSONDecodeError:
            return []
    if isinstance(parsed, list):
        parsed = {"preferences": parsed}
    elif isinstance(parsed, dict) and "preferences" not in parsed:
        parsed = {"preferences": [parsed]}
    try:
        return PreferenceExtractionPayload.model_validate(parsed).preferences
    except ValidationError:
        return []


def _category_value(category: str | PreferenceCategory) -> str:
    try:
        return str(PreferenceCategory(category))
    except ValueError as exc:
        raise HTTPException(
            status_code=status.HTTP_422_UNPROCESSABLE_ENTITY,
            detail="Unsupported preference category",
        ) from exc


def _source_value(source: str | PreferenceSource) -> str:
    try:
        return str(PreferenceSource(source))
    except ValueError as exc:
        raise HTTPException(
            status_code=status.HTTP_422_UNPROCESSABLE_ENTITY,
            detail="Unsupported preference source",
        ) from exc


def _clean_value(value: str) -> str:
    clean_value = unicodedata.normalize("NFKC", str(value)).strip()
    if not clean_value:
        raise HTTPException(
            status_code=status.HTTP_422_UNPROCESSABLE_ENTITY,
            detail="Preference value is required",
        )
    if len(clean_value) > PREFERENCE_MAX_VALUE_LENGTH:
        raise HTTPException(
            status_code=status.HTTP_422_UNPROCESSABLE_ENTITY,
            detail="Preference value is too long",
        )
    return clean_value


def _validate_owner(owner_id: int) -> None:
    if owner_id <= 0:
        raise HTTPException(
            status_code=status.HTTP_422_UNPROCESSABLE_ENTITY,
            detail="Valid preference owner is required",
        )


def _serialize(record: UserPreference) -> dict[str, Any]:
    return {
        "code": record.code,
        "category": record.category,
        "value": record.value,
        "source": record.source,
        "created_at": record.created_at.isoformat(),
        "updated_at": record.updated_at.isoformat(),
    }