aac484463d
回忆录 Story 流水线（同步） - 同步路径仅写入 Story 与章节关联，改为 mark_chapter_dirty_sync，不再内联 compose - 物化由 Celery recompose_chapter 异步完成；compose 不变量与异常时保留 dirty 的语义在 repo 中补充说明 - Evidence：大批次时降低 top_k；路由候选 story 携带 char_count/version_count；append 超长/版本过多时强制新开 story - 叙事 prompt：relevant_chunks 去重，减少重复证据噪声 - 叙事回退与忠实度 gate：返回 fallback 类型并记录结构化日志（含耗时、JSON 有效性等） Post-commit 与任务编排 - 新增 post_commit.enqueue_story_post_commit_effects：统一派发 generate_story_image（Redis 去重）、延迟 recompose_chapter、可选 memory compaction - memoir_tasks / story_service / story_image_tasks 改为调用 post-commit 入口；主图回填后按关联章节重算并调度物化与 compacs（锁委托、Redis 单例、ASR to_thread） - 更新 test_narrative_pipeline 以适配 _apply_narrative_fallbacks 返回值
345 lines
11 KiB
Python
345 lines
11 KiB
Python
"""Memoir repository — Book, Chapter, MemoirState data access."""
|
||
|
||
import uuid
|
||
|
||
from sqlalchemy import delete, func, select, update
|
||
from sqlalchemy.ext.asyncio import AsyncSession
|
||
from sqlalchemy.orm import Session, joinedload
|
||
|
||
from app.core.db import utc_now
|
||
from app.features.asset.models import Asset
|
||
from app.features.memoir.asset_resolver import collect_asset_ids_for_chapter
|
||
from app.features.memoir.chapter_markdown_compose import (
|
||
materialize_chapter_markdown_from_loaded_chapter,
|
||
)
|
||
from app.features.memoir.models import (
|
||
Book,
|
||
Chapter,
|
||
ChapterCoverIntent,
|
||
ChapterStoryLink,
|
||
ChapterVersion,
|
||
MemoirState,
|
||
)
|
||
from app.features.memoir.reading_segment_materialize import (
|
||
build_reading_segments_snapshot,
|
||
)
|
||
from app.features.story.models import Story
|
||
from app.features.story.time_hints import life_sort_key_parts
|
||
|
||
|
||
async def get_current_book(user_id: str, db: AsyncSession) -> Book | None:
|
||
stmt = (
|
||
select(Book)
|
||
.where(Book.user_id == user_id)
|
||
.order_by(Book.updated_at.desc())
|
||
.limit(1)
|
||
)
|
||
result = await db.execute(stmt)
|
||
return result.scalar_one_or_none()
|
||
|
||
|
||
async def get_chapters_for_memoir_list(
|
||
user_id: str,
|
||
db: AsyncSession,
|
||
*,
|
||
active_only: bool = True,
|
||
is_new_only: bool | None = None,
|
||
) -> list[Chapter]:
|
||
"""列表/详情：stories-first，预加载 story_links 与 images。"""
|
||
stmt = (
|
||
select(Chapter)
|
||
.where(Chapter.user_id == user_id)
|
||
.options(
|
||
joinedload(Chapter.images),
|
||
joinedload(Chapter.story_links)
|
||
.joinedload(ChapterStoryLink.story)
|
||
.joinedload(Story.image_intents),
|
||
)
|
||
.order_by(Chapter.order_index)
|
||
)
|
||
if active_only:
|
||
stmt = stmt.where(Chapter.is_active == True) # noqa: E712
|
||
if is_new_only is True:
|
||
stmt = stmt.where(Chapter.is_new == True) # noqa: E712
|
||
result = await db.execute(stmt)
|
||
return list(result.unique().scalars().all())
|
||
|
||
|
||
async def get_chapter_by_id(chapter_id: str, db: AsyncSession) -> Chapter | None:
|
||
stmt = (
|
||
select(Chapter)
|
||
.where(Chapter.id == chapter_id)
|
||
.options(
|
||
joinedload(Chapter.images),
|
||
joinedload(Chapter.story_links)
|
||
.joinedload(ChapterStoryLink.story)
|
||
.joinedload(Story.image_intents),
|
||
)
|
||
)
|
||
result = await db.execute(stmt)
|
||
return result.unique().scalars().one_or_none()
|
||
|
||
|
||
async def get_memoir_state(user_id: str, db: AsyncSession) -> MemoirState | None:
|
||
stmt = select(MemoirState).where(MemoirState.user_id == user_id)
|
||
result = await db.execute(stmt)
|
||
return result.scalar_one_or_none()
|
||
|
||
|
||
async def get_chapter_ids_linked_to_story(db: AsyncSession, story_id: str) -> list[str]:
|
||
stmt = select(ChapterStoryLink.chapter_id).where(
|
||
ChapterStoryLink.story_id == story_id
|
||
)
|
||
result = await db.execute(stmt)
|
||
return list(dict.fromkeys(result.scalars().all()))
|
||
|
||
|
||
async def mark_chapters_dirty_for_story(db: AsyncSession, story_id: str) -> None:
|
||
ids = await get_chapter_ids_linked_to_story(db, story_id)
|
||
if not ids:
|
||
return
|
||
await db.execute(
|
||
update(Chapter).where(Chapter.id.in_(ids)).values(markdown_compose_dirty=True)
|
||
)
|
||
|
||
|
||
def mark_chapters_dirty_for_story_sync(session: Session, story_id: str) -> None:
|
||
stmt = select(ChapterStoryLink.chapter_id).where(
|
||
ChapterStoryLink.story_id == story_id
|
||
)
|
||
ids = list(dict.fromkeys(session.scalars(stmt).all()))
|
||
if not ids:
|
||
return
|
||
session.execute(
|
||
update(Chapter).where(Chapter.id.in_(ids)).values(markdown_compose_dirty=True)
|
||
)
|
||
|
||
|
||
def get_chapter_ids_linked_to_story_sync(session: Session, story_id: str) -> list[str]:
|
||
stmt = select(ChapterStoryLink.chapter_id).where(
|
||
ChapterStoryLink.story_id == story_id
|
||
)
|
||
return list(dict.fromkeys(session.scalars(stmt).all()))
|
||
|
||
|
||
def mark_chapter_dirty_sync(session: Session, chapter_id: str) -> None:
|
||
"""幂等：将章节标为需物化 markdown（多次调用安全）。"""
|
||
ch = session.get(Chapter, chapter_id)
|
||
if ch:
|
||
ch.markdown_compose_dirty = True
|
||
session.flush()
|
||
|
||
|
||
async def get_chapter_with_story_links_for_compose(
|
||
chapter_id: str, db: AsyncSession
|
||
) -> Chapter | None:
|
||
stmt = (
|
||
select(Chapter)
|
||
.where(Chapter.id == chapter_id)
|
||
.options(
|
||
joinedload(Chapter.story_links)
|
||
.joinedload(ChapterStoryLink.story)
|
||
.joinedload(Story.image_intents),
|
||
)
|
||
)
|
||
result = await db.execute(stmt)
|
||
return result.unique().scalar_one_or_none()
|
||
|
||
|
||
async def append_chapter_compose_version_async(
|
||
db: AsyncSession,
|
||
chapter: Chapter,
|
||
markdown: str,
|
||
) -> None:
|
||
count_stmt = select(func.count(ChapterVersion.id)).where(
|
||
ChapterVersion.chapter_id == chapter.id
|
||
)
|
||
version_no = (await db.execute(count_stmt)).scalar() or 0
|
||
version_no += 1
|
||
vid = str(uuid.uuid4())
|
||
version = ChapterVersion(
|
||
id=vid,
|
||
chapter_id=chapter.id,
|
||
version_no=version_no,
|
||
markdown_snapshot=markdown,
|
||
actor_type="system",
|
||
source_type="compose_from_stories",
|
||
)
|
||
db.add(version)
|
||
await db.flush()
|
||
chapter.canonical_markdown = markdown
|
||
chapter.current_version_id = vid
|
||
chapter.markdown_compose_dirty = False
|
||
chapter.markdown_composed_at = utc_now()
|
||
chapter.reading_segments_json = build_reading_segments_snapshot(chapter)
|
||
|
||
|
||
def append_chapter_compose_version_sync(
|
||
session: Session,
|
||
chapter: Chapter,
|
||
markdown: str,
|
||
) -> None:
|
||
count_stmt = select(func.count(ChapterVersion.id)).where(
|
||
ChapterVersion.chapter_id == chapter.id
|
||
)
|
||
version_no = (session.execute(count_stmt).scalar() or 0) + 1
|
||
vid = str(uuid.uuid4())
|
||
version = ChapterVersion(
|
||
id=vid,
|
||
chapter_id=chapter.id,
|
||
version_no=version_no,
|
||
markdown_snapshot=markdown,
|
||
actor_type="system",
|
||
source_type="compose_from_stories",
|
||
)
|
||
session.add(version)
|
||
session.flush()
|
||
chapter.canonical_markdown = markdown
|
||
chapter.current_version_id = vid
|
||
chapter.markdown_compose_dirty = False
|
||
chapter.markdown_composed_at = utc_now()
|
||
chapter.reading_segments_json = build_reading_segments_snapshot(chapter)
|
||
|
||
|
||
def reorder_chapter_story_links_by_life_order_sync(
|
||
session: Session, chapter_id: str
|
||
) -> None:
|
||
"""
|
||
按人生发生顺序（time_start / 标题年 / 创建时间）重排 chapter_story_links.order_index。
|
||
无 story 或单条时 noop。
|
||
"""
|
||
stmt = (
|
||
select(Chapter)
|
||
.where(Chapter.id == chapter_id)
|
||
.options(
|
||
joinedload(Chapter.story_links).joinedload(ChapterStoryLink.story),
|
||
)
|
||
)
|
||
chapter = session.execute(stmt).unique().scalar_one_or_none()
|
||
if not chapter:
|
||
return
|
||
links = list(chapter.story_links or [])
|
||
if len(links) < 2:
|
||
return
|
||
|
||
def _key(link: ChapterStoryLink) -> tuple[int, int, str]:
|
||
st = link.story
|
||
if st is None:
|
||
return (9999, 0, link.story_id or "")
|
||
return life_sort_key_parts(
|
||
time_start=getattr(st, "time_start", None),
|
||
title=getattr(st, "title", None),
|
||
created_at=getattr(st, "created_at", None),
|
||
story_id=st.id,
|
||
)
|
||
|
||
ordered = sorted(links, key=_key)
|
||
for i, link in enumerate(ordered):
|
||
if link.order_index != i:
|
||
link.order_index = i
|
||
session.flush()
|
||
|
||
|
||
def compose_chapter_from_story_links_sync(session: Session, chapter_id: str) -> bool:
|
||
"""
|
||
按 story_links 重组 canonical_markdown 并写入版本链。
|
||
若无 story_links 则清除 dirty 并返回 False。
|
||
|
||
不变量：成功物化或空链接分支均会将 markdown_compose_dirty=False；异常退出保留 dirty。
|
||
"""
|
||
stmt = (
|
||
select(Chapter)
|
||
.where(Chapter.id == chapter_id)
|
||
.options(
|
||
joinedload(Chapter.story_links)
|
||
.joinedload(ChapterStoryLink.story)
|
||
.joinedload(Story.image_intents),
|
||
)
|
||
)
|
||
chapter = session.execute(stmt).unique().scalar_one_or_none()
|
||
if not chapter:
|
||
return False
|
||
links = list(chapter.story_links or [])
|
||
if not links:
|
||
chapter.markdown_compose_dirty = False
|
||
chapter.reading_segments_json = []
|
||
session.flush()
|
||
return False
|
||
md = materialize_chapter_markdown_from_loaded_chapter(chapter)
|
||
append_chapter_compose_version_sync(session, chapter, md)
|
||
return True
|
||
|
||
|
||
async def replace_chapter_story_links_async(
|
||
db: AsyncSession,
|
||
*,
|
||
chapter_id: str,
|
||
user_id: str,
|
||
story_ids: list[str],
|
||
) -> None:
|
||
chapter = await db.get(Chapter, chapter_id)
|
||
if not chapter or chapter.user_id != user_id:
|
||
raise ValueError("Chapter not found or access denied")
|
||
if len(story_ids) != len(set(story_ids)):
|
||
raise ValueError("Duplicate story_id in story_ids")
|
||
if story_ids:
|
||
stmt = select(Story.id).where(
|
||
Story.id.in_(story_ids),
|
||
Story.user_id == user_id,
|
||
)
|
||
result = await db.execute(stmt)
|
||
found = set(result.scalars().all())
|
||
missing = set(story_ids) - found
|
||
if missing:
|
||
raise ValueError(f"Stories not found or not owned: {sorted(missing)}")
|
||
await db.execute(
|
||
delete(ChapterStoryLink).where(ChapterStoryLink.chapter_id == chapter_id)
|
||
)
|
||
await db.flush()
|
||
for i, sid in enumerate(story_ids):
|
||
db.add(
|
||
ChapterStoryLink(
|
||
id=str(uuid.uuid4()),
|
||
chapter_id=chapter_id,
|
||
story_id=sid,
|
||
order_index=i,
|
||
)
|
||
)
|
||
await db.flush()
|
||
|
||
|
||
async def collect_cos_storage_keys_for_chapter(
|
||
db: AsyncSession, chapter: Chapter
|
||
) -> list[str]:
|
||
"""
|
||
章节内插图 MemoirImage、正文 asset:// 引用的 Asset、封面 cover_asset、封面意图绑定的 Asset 的 storage_key。
|
||
用于软删除章节后回收 COS 空间。
|
||
"""
|
||
keys: set[str] = set()
|
||
for img in getattr(chapter, "images", None) or []:
|
||
sk = getattr(img, "storage_key", None)
|
||
if sk:
|
||
keys.add(sk)
|
||
|
||
asset_ids = set(collect_asset_ids_for_chapter(chapter))
|
||
intent_rows = await db.execute(
|
||
select(ChapterCoverIntent.asset_id).where(
|
||
ChapterCoverIntent.chapter_id == chapter.id,
|
||
ChapterCoverIntent.asset_id.isnot(None),
|
||
)
|
||
)
|
||
for aid in intent_rows.scalars().all():
|
||
if aid:
|
||
asset_ids.add(str(aid))
|
||
|
||
if asset_ids:
|
||
row_keys = await db.execute(
|
||
select(Asset.storage_key).where(
|
||
Asset.id.in_(asset_ids),
|
||
Asset.storage_key.isnot(None),
|
||
)
|
||
)
|
||
keys.update(k for k in row_keys.scalars().all() if k)
|
||
|
||
return sorted(keys)
|