blog/bp/blog/ghost_db.py

from __future__ import annotations

from typing import Any, Dict, List, Optional, Sequence, Tuple
from sqlalchemy import select, func, asc, desc, and_, or_
from sqlalchemy.ext.asyncio import AsyncSession
from sqlalchemy.orm import selectinload, joinedload

from models.ghost_content import Post, Author, Tag, PostTag
from models.page_config import PageConfig
from models.tag_group import TagGroup, TagGroupTag


class DBAPIError(Exception):
    """Raised when our local DB returns something unexpected."""


def _author_to_public(a: Optional[Author]) -> Optional[Dict[str, Any]]:
    if a is None:
        return None
    if a.deleted_at is not None:
        # treat deleted authors as missing
        return None
    return {
        "id": a.ghost_id,
        "slug": a.slug,
        "name": a.name,
        "profile_image": a.profile_image,
        "cover_image": a.cover_image,
        # expose more (bio, etc.) if needed
    }


def _tag_to_public(t: Tag) -> Dict[str, Any]:
    return {
        "id": t.ghost_id,
        "slug": t.slug,
        "name": t.name,
        "description": t.description,
        "feature_image": t.feature_image,  # fixed key
        "visibility": t.visibility,
        "deleted_at": t.deleted_at,
    }


def _post_to_public(p: Post) -> Dict[str, Any]:
    """
    Shape a Post to the public JSON used by the app, mirroring GhostClient._normalise_post.
    """
    # Primary author: explicit or first available
    primary_author = p.primary_author or (p.authors[0] if p.authors else None)

    # Primary tag: prefer explicit relationship, otherwise first public/non-deleted tag
    primary_tag = getattr(p, "primary_tag", None)
    if primary_tag is None:
        public_tags = [
            t for t in (p.tags or [])
            if t.deleted_at is None and (t.visibility or "public") == "public"
        ]
        primary_tag = public_tags[0] if public_tags else None

    return {
        "id": p.id,
        "ghost_id": p.ghost_id,
        "slug": p.slug,
        "title": p.title,
        "html": p.html,
        "is_page": p.is_page,
        "excerpt": p.custom_excerpt or p.excerpt,
        "custom_excerpt": p.custom_excerpt,
        "published_at": p.published_at,
        "updated_at": p.updated_at,
        "visibility": p.visibility,
        "status": p.status,
        "deleted_at": p.deleted_at,
        "feature_image": p.feature_image,
        "user_id": p.user_id,
        "publish_requested": p.publish_requested,
        "primary_author": _author_to_public(primary_author),
        "primary_tag": _tag_to_public(primary_tag) if primary_tag else None,
        "tags": [
            _tag_to_public(t)
            for t in (p.tags or [])
            if t.deleted_at is None and (t.visibility or "public") == "public"
        ],
        "authors": [
            _author_to_public(a)
            for a in (p.authors or [])
            if a and a.deleted_at is None
        ],
    }


class DBClient:
    """
    Drop-in replacement for GhostClient, but served from our mirrored tables.
    Call methods with an AsyncSession.
    """

    def __init__(self, session: AsyncSession):
        self.sess = session

    async def list_posts(
        self,
        limit: int = 10,
        page: int = 1,
        selected_tags: Optional[Sequence[str]] = None,
        selected_authors: Optional[Sequence[str]] = None,
        search: Optional[str] = None,
        drafts: bool = False,
        drafts_user_id: Optional[int] = None,
        exclude_covered_tag_ids: Optional[Sequence[int]] = None,
    ) -> Tuple[List[Dict[str, Any]], Dict[str, Any]]:
        """
        List published posts, optionally filtered by tags/authors and a search term.
        When drafts=True, lists draft posts instead (filtered by drafts_user_id if given).
        Returns (posts, pagination).
        """

        # ---- base visibility filters
        if drafts:
            base_filters = [
                Post.deleted_at.is_(None),
                Post.status == "draft",
                Post.is_page.is_(False),
            ]
            if drafts_user_id is not None:
                base_filters.append(Post.user_id == drafts_user_id)
        else:
            base_filters = [
                Post.deleted_at.is_(None),
                Post.status == "published",
                Post.is_page.is_(False),
            ]

        q = select(Post).where(*base_filters)

        # ---- TAG FILTER (matches any tag on the post)
        if selected_tags:
            tag_slugs = list(selected_tags)
            q = q.where(
                Post.tags.any(
                    and_(
                        Tag.slug.in_(tag_slugs),
                        Tag.deleted_at.is_(None),
                    )
                )
            )

        # ---- EXCLUDE-COVERED FILTER ("etc" mode: posts NOT covered by any group)
        if exclude_covered_tag_ids:
            covered_sq = (
                select(PostTag.post_id)
                .join(Tag, Tag.id == PostTag.tag_id)
                .where(
                    Tag.id.in_(list(exclude_covered_tag_ids)),
                    Tag.deleted_at.is_(None),
                )
            )
            q = q.where(Post.id.notin_(covered_sq))

        # ---- AUTHOR FILTER (matches primary or any author)
        if selected_authors:
            author_slugs = list(selected_authors)
            q = q.where(
                or_(
                    Post.primary_author.has(
                        and_(
                            Author.slug.in_(author_slugs),
                            Author.deleted_at.is_(None),
                        )
                    ),
                    Post.authors.any(
                        and_(
                            Author.slug.in_(author_slugs),
                            Author.deleted_at.is_(None),
                        )
                    ),
                )
            )

        # ---- SEARCH FILTER (title OR excerpt OR plaintext contains)
        if search:
            term = f"%{search.strip().lower()}%"
            q = q.where(
                or_(
                    func.lower(func.coalesce(Post.title,    "")).like(term),
                    func.lower(func.coalesce(Post.excerpt,  "")).like(term),
                    func.lower(func.coalesce(Post.plaintext,"")).like(term),
                )
            )

        # ---- ordering
        if drafts:
            q = q.order_by(desc(Post.updated_at))
        else:
            q = q.order_by(desc(Post.published_at))

        # ---- pagination math
        if page < 1:
            page = 1
        offset_val = (page - 1) * limit

        # ---- total count with SAME filters (including tag/author/search)
        q_no_limit = q.with_only_columns(Post.id).order_by(None)
        count_q = select(func.count()).select_from(q_no_limit.subquery())
        total = int((await self.sess.execute(count_q)).scalar() or 0)

        # ---- eager load relationships to avoid N+1 / greenlet issues
        q = (
            q.options(
                joinedload(Post.primary_author),
                joinedload(Post.primary_tag),
                selectinload(Post.authors),
                selectinload(Post.tags),
            )
            .limit(limit)
            .offset(offset_val)
        )

        rows: List[Post] = list((await self.sess.execute(q)).scalars())
        posts = [_post_to_public(p) for p in rows]

        # ---- search_count: reflect same filters + search (i.e., equals total once filters applied)
        search_count = total

        pages_total = (total + limit - 1) // limit if limit else 1
        pagination = {
            "page": page,
            "limit": limit,
            "pages": pages_total,
            "total": total,
            "search_count": search_count,
            "next": page + 1 if page < pages_total else None,
            "prev": page - 1 if page > 1 else None,
        }

        return posts, pagination

    async def list_pages(
        self,
        limit: int = 10,
        page: int = 1,
        search: Optional[str] = None,
    ) -> Tuple[List[Dict[str, Any]], Dict[str, Any]]:
        """
        List published pages (is_page=True) with their PageConfig eagerly loaded.
        Returns (pages, pagination).
        """
        base_filters = [
            Post.deleted_at.is_(None),
            Post.status == "published",
            Post.is_page.is_(True),
        ]

        q = select(Post).where(*base_filters)

        if search:
            term = f"%{search.strip().lower()}%"
            q = q.where(
                or_(
                    func.lower(func.coalesce(Post.title, "")).like(term),
                    func.lower(func.coalesce(Post.excerpt, "")).like(term),
                    func.lower(func.coalesce(Post.plaintext, "")).like(term),
                )
            )

        q = q.order_by(desc(Post.published_at))

        if page < 1:
            page = 1
        offset_val = (page - 1) * limit

        q_no_limit = q.with_only_columns(Post.id).order_by(None)
        count_q = select(func.count()).select_from(q_no_limit.subquery())
        total = int((await self.sess.execute(count_q)).scalar() or 0)

        q = (
            q.options(
                joinedload(Post.primary_author),
                joinedload(Post.primary_tag),
                selectinload(Post.authors),
                selectinload(Post.tags),
                joinedload(Post.page_config),
            )
            .limit(limit)
            .offset(offset_val)
        )

        rows: List[Post] = list((await self.sess.execute(q)).scalars())

        def _page_to_public(p: Post) -> Dict[str, Any]:
            d = _post_to_public(p)
            pc = p.page_config
            d["features"] = pc.features if pc else {}
            return d

        pages_list = [_page_to_public(p) for p in rows]

        pages_total = (total + limit - 1) // limit if limit else 1
        pagination = {
            "page": page,
            "limit": limit,
            "pages": pages_total,
            "total": total,
            "next": page + 1 if page < pages_total else None,
            "prev": page - 1 if page > 1 else None,
        }

        return pages_list, pagination

    async def posts_by_slug(
        self,
        slug: str,
        include: Sequence[str] = ("tags", "authors"),
        fields: Sequence[str] = (
            "id",
            "slug",
            "title",
            "html",
            "excerpt",
            "custom_excerpt",
            "published_at",
            "feature_image",
        ),
        include_drafts: bool = False,
    ) -> List[Dict[str, Any]]:
        """
        Return posts (usually 1) matching this slug.

        Only returns published, non-deleted posts by default.
        When include_drafts=True, also returns draft posts (for admin access).

        Eager-load related objects via selectinload/joinedload so we don't N+1 when
        serializing in _post_to_public().
        """

        # Build .options(...) dynamically based on `include`
        load_options = []

        # Tags
        if "tags" in include:
            load_options.append(selectinload(Post.tags))
            if hasattr(Post, "primary_tag"):
                # joinedload is fine too; selectin keeps a single extra roundtrip
                load_options.append(selectinload(Post.primary_tag))

        # Authors
        if "authors" in include:
            if hasattr(Post, "primary_author"):
                load_options.append(selectinload(Post.primary_author))
            if hasattr(Post, "authors"):
                load_options.append(selectinload(Post.authors))

        filters = [Post.deleted_at.is_(None), Post.slug == slug]
        if not include_drafts:
            filters.append(Post.status == "published")

        q = (
            select(Post)
            .where(*filters)
            .order_by(desc(Post.published_at))
            .options(*load_options)
        )

        result = await self.sess.execute(q)
        rows: List[Post] = list(result.scalars())

        return [(_post_to_public(p), p) for p in rows]

    async def list_tags(
        self,
        limit: int = 5000,
        page: int = 1,
        is_page=False,
    ) -> List[Dict[str, Any]]:
        """
        Return public, not-soft-deleted tags.
        Include published_post_count = number of published (not deleted) posts using that tag.
        """

        if page < 1:
            page = 1
        offset_val = (page - 1) * limit

        # Subquery: count published posts per tag
        tag_post_counts_sq = (
            select(
                PostTag.tag_id.label("tag_id"),
                func.count().label("published_post_count"),
            )
            .select_from(PostTag)
            .join(Post, Post.id == PostTag.post_id)
            .where(
                Post.deleted_at.is_(None),
                Post.published_at.is_not(None),
                Post.is_page.is_(is_page),
            )
            .group_by(PostTag.tag_id)
            .subquery()
        )

        q = (
            select(
                Tag,
                func.coalesce(tag_post_counts_sq.c.published_post_count, 0).label(
                    "published_post_count"
                ),
            )
            .outerjoin(
                tag_post_counts_sq,
                tag_post_counts_sq.c.tag_id == Tag.id,
            )
            .where(
                Tag.deleted_at.is_(None),
                (Tag.visibility == "public") | (Tag.visibility.is_(None)),
                func.coalesce(tag_post_counts_sq.c.published_post_count, 0) > 0,
            )
            .order_by(desc(func.coalesce(tag_post_counts_sq.c.published_post_count, 0)), asc(Tag.name))
            .limit(limit)
            .offset(offset_val)
        )

        result = await self.sess.execute(q)

        # result will return rows like (Tag, published_post_count)
        rows = list(result.all())

        tags = [
            {
                "id": tag.ghost_id,
                "slug": tag.slug,
                "name": tag.name,
                "description": tag.description,
                "feature_image": tag.feature_image,
                "visibility": tag.visibility,
                "published_post_count": count,
            }
            for (tag, count) in rows
        ]

        return tags

    async def list_authors(
        self,
        limit: int = 5000,
        page: int = 1,
        is_page=False,
    ) -> List[Dict[str, Any]]:
        """
        Return non-deleted authors.
        Include published_post_count = number of published (not deleted) posts by that author
        (counted via Post.primary_author_id).
        """

        if page < 1:
            page = 1
        offset_val = (page - 1) * limit

        # Subquery: count published posts per primary author
        author_post_counts_sq = (
            select(
                Post.primary_author_id.label("author_id"),
                func.count().label("published_post_count"),
            )
            .where(
                Post.deleted_at.is_(None),
                Post.published_at.is_not(None),
                Post.is_page.is_(is_page),
            )
            .group_by(Post.primary_author_id)
            .subquery()
        )

        q = (
            select(
                Author,
                func.coalesce(author_post_counts_sq.c.published_post_count, 0).label(
                    "published_post_count"
                ),
            )
            .outerjoin(
                author_post_counts_sq,
                author_post_counts_sq.c.author_id == Author.id,
            )
            .where(
                Author.deleted_at.is_(None),
            )
            .order_by(asc(Author.name))
            .limit(limit)
            .offset(offset_val)
        )

        result = await self.sess.execute(q)
        rows = list(result.all())

        authors = [
            {
                "id": a.ghost_id,
                "slug": a.slug,
                "name": a.name,
                "bio": a.bio,
                "profile_image": a.profile_image,
                "cover_image": a.cover_image,
                "website": a.website,
                "location": a.location,
                "facebook": a.facebook,
                "twitter": a.twitter,
                "published_post_count": count,
            }
            for (a, count) in rows
        ]

        return authors

    async def count_drafts(self, user_id: Optional[int] = None) -> int:
        """Count draft (non-page, non-deleted) posts, optionally for a single user."""
        q = select(func.count()).select_from(Post).where(
            Post.deleted_at.is_(None),
            Post.status == "draft",
            Post.is_page.is_(False),
        )
        if user_id is not None:
            q = q.where(Post.user_id == user_id)
        return int((await self.sess.execute(q)).scalar() or 0)

    async def list_tag_groups_with_counts(self) -> List[Dict[str, Any]]:
        """
        Return all tag groups with aggregated published post counts.
        Each group dict includes a `tag_slugs` list and `tag_ids` list.
        Count = distinct published posts having ANY member tag.
        Ordered by sort_order, name.
        """
        # Subquery: distinct published post IDs per tag group
        post_count_sq = (
            select(
                TagGroupTag.tag_group_id.label("group_id"),
                func.count(func.distinct(PostTag.post_id)).label("post_count"),
            )
            .select_from(TagGroupTag)
            .join(PostTag, PostTag.tag_id == TagGroupTag.tag_id)
            .join(Post, Post.id == PostTag.post_id)
            .where(
                Post.deleted_at.is_(None),
                Post.published_at.is_not(None),
                Post.is_page.is_(False),
            )
            .group_by(TagGroupTag.tag_group_id)
            .subquery()
        )

        q = (
            select(
                TagGroup,
                func.coalesce(post_count_sq.c.post_count, 0).label("post_count"),
            )
            .outerjoin(post_count_sq, post_count_sq.c.group_id == TagGroup.id)
            .order_by(asc(TagGroup.sort_order), asc(TagGroup.name))
        )

        rows = list((await self.sess.execute(q)).all())

        groups = []
        for tg, count in rows:
            # Fetch member tag slugs + ids for this group
            tag_rows = list(
                (await self.sess.execute(
                    select(Tag.slug, Tag.id)
                    .join(TagGroupTag, TagGroupTag.tag_id == Tag.id)
                    .where(
                        TagGroupTag.tag_group_id == tg.id,
                        Tag.deleted_at.is_(None),
                        (Tag.visibility == "public") | (Tag.visibility.is_(None)),
                    )
                )).all()
            )
            groups.append({
                "id": tg.id,
                "name": tg.name,
                "slug": tg.slug,
                "feature_image": tg.feature_image,
                "colour": tg.colour,
                "sort_order": tg.sort_order,
                "post_count": count,
                "tag_slugs": [r[0] for r in tag_rows],
                "tag_ids": [r[1] for r in tag_rows],
            })

        return groups

    async def count_etc_posts(self, assigned_tag_ids: List[int]) -> int:
        """
        Count published posts not covered by any tag group.
        Includes posts with no tags and posts whose tags are all unassigned.
        """
        base = [
            Post.deleted_at.is_(None),
            Post.published_at.is_not(None),
            Post.is_page.is_(False),
        ]
        if assigned_tag_ids:
            covered_sq = (
                select(PostTag.post_id)
                .join(Tag, Tag.id == PostTag.tag_id)
                .where(
                    Tag.id.in_(assigned_tag_ids),
                    Tag.deleted_at.is_(None),
                )
            )
            base.append(Post.id.notin_(covered_sq))

        q = select(func.count()).select_from(Post).where(*base)
        return int((await self.sess.execute(q)).scalar() or 0)

    async def list_drafts(self) -> List[Dict[str, Any]]:
        """Return all draft (non-page, non-deleted) posts, newest-updated first."""
        q = (
            select(Post)
            .where(
                Post.deleted_at.is_(None),
                Post.status == "draft",
                Post.is_page.is_(False),
            )
            .order_by(desc(Post.updated_at))
            .options(
                joinedload(Post.primary_author),
                joinedload(Post.primary_tag),
                selectinload(Post.authors),
                selectinload(Post.tags),
            )
        )
        rows: List[Post] = list((await self.sess.execute(q)).scalars())
        return [_post_to_public(p) for p in rows]