修复：CI 部署环境与 ref 错配、迁移碎片化、图片意图 source_span、章节物化脏版式、会话历史与本地语音不一致
新增：TTS 上传 COS 与分片、章节 reading_segments 物化与快照、markdown 清洗、会话消息 repository、语音 store 重构与相关测试
This commit is contained in:
Kevin
188
api/app/features/memoir/reading_segment_materialize.py
Normal file
188
api/app/features/memoir/reading_segment_materialize.py
Normal file
@@ -0,0 +1,188 @@
|
||||
"""
|
||||
章节阅读片段物化：与 canonical 同一生成时机写入 reading_segments_json（无签名 URL）；
|
||||
API 读时 hydrate 或（dirty / 无快照）回退为运行时物化。
|
||||
"""
|
||||
|
||||
from __future__ import annotations
|
||||
|
||||
from typing import Any
|
||||
|
||||
from app.features.memoir.asset_resolver import (
|
||||
collect_asset_ids_from_markdown,
|
||||
resolve_asset_refs_in_markdown,
|
||||
)
|
||||
from app.features.memoir.markdown_sanitize import sanitize_story_for_chapter_compose
|
||||
from app.features.memoir.models import Chapter
|
||||
|
||||
|
||||
def _primary_story_intent_asset_id(story: Any) -> str | None:
|
||||
for it in getattr(story, "image_intents", None) or []:
|
||||
if getattr(it, "intent_role", None) == "primary":
|
||||
aid = getattr(it, "asset_id", None)
|
||||
return str(aid) if aid else None
|
||||
return None
|
||||
|
||||
|
||||
def _cover_intent_snapshot_from_story(story: Any) -> dict | None:
|
||||
"""primary intent 元数据（无 url），供 JSON 持久化。"""
|
||||
intents = getattr(story, "image_intents", None) or []
|
||||
primary = None
|
||||
for it in intents:
|
||||
if getattr(it, "intent_role", None) == "primary":
|
||||
primary = it
|
||||
break
|
||||
if not primary:
|
||||
return None
|
||||
aid = getattr(primary, "asset_id", None)
|
||||
if not aid:
|
||||
return None
|
||||
status = getattr(primary, "status", None) or "pending"
|
||||
return {
|
||||
"asset_id": str(aid),
|
||||
"status": status,
|
||||
"description": getattr(primary, "caption", None) or "故事配图",
|
||||
"prompt": getattr(primary, "prompt_brief", None),
|
||||
"style": getattr(primary, "style_profile", None),
|
||||
"error": getattr(primary, "error", None),
|
||||
"created_at": primary.created_at.isoformat() if primary.created_at else None,
|
||||
"updated_at": primary.updated_at.isoformat() if primary.updated_at else None,
|
||||
}
|
||||
|
||||
|
||||
def _cover_dict_from_snapshot_row(
|
||||
snap: dict[str, Any], asset_url_map: dict[str, str]
|
||||
) -> dict:
|
||||
aid = snap.get("asset_id")
|
||||
url = asset_url_map.get(str(aid)) if aid else None
|
||||
return {
|
||||
"placeholder": "",
|
||||
"description": snap.get("description") or "故事配图",
|
||||
"index": 0,
|
||||
"status": snap.get("status") or "pending",
|
||||
"prompt": snap.get("prompt"),
|
||||
"url": url,
|
||||
"storage_key": None,
|
||||
"provider": None,
|
||||
"style": snap.get("style"),
|
||||
"size": None,
|
||||
"error": snap.get("error"),
|
||||
"retryable": None,
|
||||
"created_at": snap.get("created_at"),
|
||||
"updated_at": snap.get("updated_at"),
|
||||
}
|
||||
|
||||
|
||||
def build_reading_segments_snapshot(ch: Chapter) -> list[dict[str, Any]]:
|
||||
"""
|
||||
物化阅读片段快照：body 保留 asset://；cover 仅存 intent 元数据（正文已含同 asset 则省略）。
|
||||
与 append_chapter_compose_version 同路径写入。
|
||||
"""
|
||||
links = sorted(
|
||||
list(getattr(ch, "story_links", None) or []),
|
||||
key=lambda x: getattr(x, "order_index", 0),
|
||||
)
|
||||
out: list[dict[str, Any]] = []
|
||||
for link in links:
|
||||
st = getattr(link, "story", None)
|
||||
if st is None:
|
||||
continue
|
||||
title = (getattr(st, "title", None) or "").strip()
|
||||
raw = (getattr(st, "canonical_markdown", None) or "").strip()
|
||||
body = sanitize_story_for_chapter_compose(raw, title)
|
||||
if not body:
|
||||
continue
|
||||
primary_aid = _primary_story_intent_asset_id(st)
|
||||
inline_ids = set(collect_asset_ids_from_markdown(body))
|
||||
cover: dict | None = None
|
||||
if primary_aid and primary_aid not in inline_ids:
|
||||
cover = _cover_intent_snapshot_from_story(st)
|
||||
out.append(
|
||||
{
|
||||
"story_id": st.id,
|
||||
"body_markdown": body,
|
||||
"cover_image": cover,
|
||||
}
|
||||
)
|
||||
return out
|
||||
|
||||
|
||||
def materialize_chapter_reading_segments(
|
||||
ch: Chapter,
|
||||
asset_url_map: dict[str, str] | None = None,
|
||||
) -> list[dict[str, Any]]:
|
||||
"""运行时物化（解析签名 URL），与旧 build_reading_segments 行为一致。"""
|
||||
from app.features.memoir import helpers as h
|
||||
|
||||
asset_url_map = asset_url_map or {}
|
||||
resolve = lambda aid: asset_url_map.get(aid) # noqa: E731
|
||||
links = sorted(
|
||||
list(getattr(ch, "story_links", None) or []),
|
||||
key=lambda x: getattr(x, "order_index", 0),
|
||||
)
|
||||
segments: list[dict[str, Any]] = []
|
||||
for link in links:
|
||||
st = getattr(link, "story", None)
|
||||
if st is None:
|
||||
continue
|
||||
title = (getattr(st, "title", None) or "").strip()
|
||||
raw = (getattr(st, "canonical_markdown", None) or "").strip()
|
||||
body = sanitize_story_for_chapter_compose(raw, title)
|
||||
if not body:
|
||||
continue
|
||||
body_md = resolve_asset_refs_in_markdown(body, resolve)
|
||||
img_raw = h.story_primary_cover_image_dict(st, asset_url_map=asset_url_map)
|
||||
primary_aid = _primary_story_intent_asset_id(st)
|
||||
inline_ids = set(collect_asset_ids_from_markdown(body))
|
||||
if img_raw and primary_aid and primary_aid in inline_ids:
|
||||
img_raw = None
|
||||
img_norm = h.first_normalized_image_for_api(img_raw) if img_raw else None
|
||||
segments.append(
|
||||
{
|
||||
"story_id": st.id,
|
||||
"body_markdown": body_md,
|
||||
"cover_image": img_norm,
|
||||
}
|
||||
)
|
||||
return segments
|
||||
|
||||
|
||||
def hydrate_reading_segments_from_snapshot(
|
||||
ch: Chapter,
|
||||
asset_url_map: dict[str, str] | None = None,
|
||||
) -> list[dict[str, Any]]:
|
||||
"""将持久化快照解析为 API 形态（签名 URL）。"""
|
||||
from app.features.memoir import helpers as h
|
||||
|
||||
asset_url_map = asset_url_map or {}
|
||||
resolve = lambda aid: asset_url_map.get(aid) # noqa: E731
|
||||
rows = getattr(ch, "reading_segments_json", None) or []
|
||||
out: list[dict[str, Any]] = []
|
||||
for row in rows:
|
||||
body = resolve_asset_refs_in_markdown(row["body_markdown"], resolve)
|
||||
ci = row.get("cover_image")
|
||||
if ci:
|
||||
img_raw = _cover_dict_from_snapshot_row(ci, asset_url_map)
|
||||
img_norm = h.first_normalized_image_for_api(img_raw)
|
||||
else:
|
||||
img_norm = None
|
||||
out.append(
|
||||
{
|
||||
"story_id": row["story_id"],
|
||||
"body_markdown": body,
|
||||
"cover_image": img_norm,
|
||||
}
|
||||
)
|
||||
return out
|
||||
|
||||
|
||||
def resolve_reading_segments_for_chapter_detail(
|
||||
ch: Chapter,
|
||||
asset_url_map: dict[str, str] | None = None,
|
||||
) -> list[dict[str, Any]]:
|
||||
"""章节详情：dirty 或无快照列时运行时物化；否则 hydrate。"""
|
||||
asset_url_map = asset_url_map or {}
|
||||
dirty = getattr(ch, "markdown_compose_dirty", True)
|
||||
has_snapshot = getattr(ch, "reading_segments_json", None) is not None
|
||||
if has_snapshot and not dirty:
|
||||
return hydrate_reading_segments_from_snapshot(ch, asset_url_map=asset_url_map)
|
||||
return materialize_chapter_reading_segments(ch, asset_url_map=asset_url_map)
|
||||
Reference in New Issue
Block a user