codebee 0.1.95 → 0.1.96

package/app/core/bookmeta.py

@@ -1,1036 +1,1036 @@
-# -*- coding: utf-8 -*-
-"""作品信息一键生成：按发布平台（番茄/七猫）产出建书表单要填的开书资料。
-
-用户在详情页「成果」分区点按钮触发（不做自动生成），后台线程调编排者模型
-提炼 大纲 + 故事圣经 + 已有章节 的精华，按平台表单字段归一化后存任务
-book_meta 字段，同时在工作目录落一份 Markdown 归档。降级链与大纲一致：
-编排者 API → 作者 CLI → 模板兜底（模板只做搬运归纳，必填字段仍保证可建书）。
-"""
-from __future__ import annotations
-
-import os
-import re
-import time
-from pathlib import Path
-
-PLATFORMS = {
-    "fanqie": {"label": "番茄", "file": "作品信息-番茄.md"},
-    "qimao": {"label": "七猫", "file": "作品信息-七猫.md"},
-}
-# 单平台生成上限：编排者提炼一段简介级文本是轻任务，900s 足够含重试
-BOOKMETA_TIMEOUT = 900
-_FIRST_CH_HEAD = 1500   # 注入提示词的第一章开头字符数
-
-
-def template_only():
-    """只走模板兜底（跳过编排者/作者 CLI）：env TUTTI_BOOKMETA_TEMPLATE_ONLY=1。
-
-    离线环境、不想为开书资料再花一次模型调用、以及测试要确定性输出时使用；
-    产出仍是结构完整的字段，书名/简介直接取大纲与目标，平台必填字段由生成阶段
-    的官方目录兜底补齐。"""
-    return str(os.environ.get("TUTTI_BOOKMETA_TEMPLATE_ONLY") or "").strip() in ("1", "true", "yes")
-
-
-# 各平台建书表单的字段展示顺序（前端卡片与 Markdown 归档共用）。
-# 番茄官方表单（2026-09 实抓）：签约模式=连载模式/完本模式；阅读标签=主分类(必选1)
-# +主题/角色/情节(各≤2)；内容标签=情节(≤4)/情感(≤2)/人设(≤4)/世界观(≤1)。
-# 七猫官方表单：一级→二级级联按频道；作品标签四组每组必选 1-3 个。
-FIELD_LABELS = {
-    "fanqie": [
-        ("book_name", "作品名"), ("signing_mode", "签约模式"),
-        ("target_reader", "目标读者"), ("category", "主分类"),
-        ("tags_theme", "主题标签"), ("tags_role", "角色标签"),
-        ("tags_plot", "情节标签"),
-        ("content_plot", "内容·情节"), ("content_emotion", "内容·情感"),
-        ("content_character", "内容·人设"), ("content_world", "内容·世界观"),
-        ("protagonist_1", "主角名1"),
-        ("protagonist_2", "主角名2"), ("summary", "作品简介"),
-    ],
-    "qimao": [
-        ("book_name", "作品名称"), ("target_reader", "目标读者"),
-        ("category_main", "一级分类"), ("category_sub", "二级分类"),
-        ("tags_style", "风格标签"), ("tags_role", "角色标签"),
-        ("tags_plot", "情节标签"), ("tags_bg", "背景标签"),
-        ("protagonist_1", "主角名1"),
-        ("protagonist_2", "主角名2"), ("status", "作品状态"),
-        ("summary", "作品简介"),
-    ],
-}
-
-PROMPT_HEAD = """你是网文编辑，熟悉 __PLATFORM__ 建书表单的填法。
-
-__SKILLS__
-请依据下方的小说素材（目标/故事圣经/大纲/已有章节开头），为这本书生成一套
-「作品信息」，直接可用于在 __PLATFORM__ 创建作品。只输出一个 ```json 代码块，
-不要输出其他内容。JSON 结构：
-__SCHEMA__
-硬性要求：
-- 字段值必须贴合素材内容，不编造素材里没有的设定；简介不超过 500 字，突出卖点与钩子；
-- 每个标签数组字段都必须至少给 1 个、并尽量贴着上限给满，只从对应组的选项表里选；
-  任何一组留空数组、漏掉字段或自造表外词都算错误；
-- __OPTIONS_RULE__
-- 题材健康，无违规内容，符合平台签约调性。
-"""
-
-# 签约模式真实选项（番茄建书表单 2026-09 实抓）：连载模式 / 完本模式
-FANQIE_MODES = ("连载模式", "完本模式")
-
-FANQIE_SCHEMA = ('{"book_name": "作品名(15字内)", "signing_mode": "连载模式 或 完本模式", '
-                 '"target_reader": "男频 或 女频", "category": "主分类(必选且只能选一个，从下方分类表选)", '
-                 '"tags_theme": ["主题标签", …最多2个], "tags_role": ["角色标签", …最多2个], '
-                 '"tags_plot": ["情节标签", …最多2个], '
-                 '"content_plot": ["内容标签·情节", …最多4个], '
-                 '"content_emotion": ["内容标签·情感", …最多2个], '
-                 '"content_character": ["内容标签·人设", …最多4个], '
-                 '"content_world": ["内容标签·世界观", …最多1个], '
-                 '"protagonist_1": "主角名1(5字内)", '
-                 '"protagonist_2": "主角名2(5字内，没有就空串)", "summary": "作品简介(500字内)"}')
-
-QIMAO_SCHEMA = ('{"book_name": "作品名称(18字内)", "target_reader": "男生 或 女生", '
-                '"category_main": "一级分类(从该频道一级分类表选一个)", '
-                '"category_sub": "二级分类(必须是该一级下的二级)", '
-                '"tags_style": ["风格标签", 必选1-3个], '
-                '"tags_role": ["角色标签", 必选1-3个], '
-                '"tags_plot": ["情节标签", 必选1-3个], '
-                '"tags_bg": ["背景标签", 必选1-3个], '
-                '"protagonist_1": "主角名1(5字内)", '
-                '"protagonist_2": "主角名2(5字内，没有就空串)", "status": "连载中 或 已完结", '
-                '"summary": "作品简介(200字内)"}')
-
-
-def _fq_tables(reader):
-    """番茄按目标读者取对应选项表（男频/女频两套完全独立的分类与标签）。"""
-    from . import bookmeta_catalog as cat
-    if reader == "女频":
-        return (cat.FANQIE_FEMALE_CATEGORIES, cat.FANQIE_FEMALE_TAGS_THEME,
-                cat.FANQIE_FEMALE_TAGS_ROLE, cat.FANQIE_FEMALE_TAGS_PLOT)
-    return (cat.FANQIE_MALE_CATEGORIES, cat.FANQIE_MALE_TAGS_THEME,
-            cat.FANQIE_MALE_TAGS_ROLE, cat.FANQIE_MALE_TAGS_PLOT)
-
-
-def _fq_content_tables():
-    """番茄内容标签四组（男女性共用）：情节/情感/人设/世界观。"""
-    from . import bookmeta_catalog as cat
-    return (cat.FANQIE_CONTENT_PLOT, cat.FANQIE_CONTENT_EMOTION,
-            cat.FANQIE_CONTENT_CHARACTER, cat.FANQIE_CONTENT_WORLD)
-
-
-def _fq_options_block(goal):
-    """番茄提示词的真实选项注入：按 goal 推断读者给出对应分类与标签表。
-
-    2026-09-24 起首行加频道判定规则：素材明显属于另一频时 target_reader
-    如实填、分类留空（规整层按该频表校验并兜底补齐），防「注入男频表+
-    女频素材」时硬凑男频分类。"""
-    reader = "女频" if "女" in (goal or "") else "男频"
-    cats, theme, role, plot = _fq_tables(reader)
-    c_plot, c_emo, c_char, c_world = _fq_content_tables()
-    return ("先按素材判定目标读者（战争/军功/历史争霸/都市异能/系统→男频；"
-            "总裁豪门/宅斗宫斗/年代/婚恋情感线为主→女频），target_reader 如实填；"
-            "主分类从下方按目标读者 %s 预注入的表里选，素材明显属于另一频时"
-            "分类留空待系统按该频表补齐。主分类要贴世界观而非字面词——架空"
-            "王朝/军事/军功题材选历史古代或历史脑洞，西方奇幻仅限西式魔幻"
-            "世界观。\n"
-            "目标读者（%s）可用的主分类表（必选且只能选一个）：%s\n"
-            "阅读标签三组各最多选 2 个——主题组：%s\n角色组：%s\n情节组（部分，选最贴合的）：%s\n"
-            "内容标签四组——情节组（最多4个，部分）：%s\n情感组（最多2个）：%s\n"
-            "人设组（最多4个，部分）：%s\n世界观组（最多1个）：%s"
-            % (reader, reader, "、".join(cats), "、".join(theme), "、".join(role),
-               "、".join(plot[:50]), "、".join(c_plot[:60]), "、".join(c_emo),
-               "、".join(c_char[:60]), "、".join(c_world)))
-
-
-def _qm_options_block(goal):
-    """七猫提示词的真实选项注入：双频道级联 + 频道判定规则 + 跨频道禁令。
-
-    2026-09-24 前只按 goal 猜频道注入单频道表，但挡不住模型自带另一频道
-    的分类知识跨频道自造组合（男频军事文被配成 女生/古代言情/权谋天下，
-    级联还自洽）——必须双频道全量注入并明示判定规则。"""
-    from . import bookmeta_catalog as cat
-    cascades = ["%s频道（一级→二级）：%s" % (ch, "；".join(
-        "%s（二级：%s）" % (m, "、".join(v)) for m, v in mains.items()))
-        for ch, mains in cat.QIMAO_CATS.items()]
-    pool = ["%s组（必选1-3个）：%s" % (g, "、".join(t))
-            for g, t in cat.QIMAO_TAG_GROUPS.items()]
-    return ("先按素材判定目标读者频道：战争/军功/练兵/历史争霸/都市异能/"
-            "系统流/玄幻修仙→男生；总裁豪门/宅斗宫斗/年代/婚恋情感线为主→"
-            "女生。target_reader 必须填判定的频道，一级/二级分类只能从该频道"
-            "的级联表里选，严禁跨频道——如「权谋天下」只存在于女生·古代言情"
-            "下，男频军事历史文不得选用；「架空历史」属男生·历史。\n"
-            "%s\n作品标签四组（男女生同池）：%s"
-            % ("。".join(cascades), "；".join(pool)))
-
-
-# ---- 频道题材信号（2026-09-24 实案：男频军事文《戍边骑奴》被生成端配成
-# 女生/古代言情/权谋天下——「权谋天下」字面贴合朝堂线但只存在于女生频道，
-# 模型跨频道自造组合且级联自洽，目录校验拦不住，只能靠题材信号判频道。
-# 必须保守：信号单侧命中才判，双侧命中/零命中一律不判（宁可不纠不可错纠）。
-# 证据语料只用 goal+简介：分类字段是嫌疑输出、标签池男女同池，都不算证据。
-_CH_SIGNAL_MALE = ("军功", "戍边", "骑奴", "特种兵", "练兵", "杀伐", "阵斩",
-                   "百夫长", "大军", "蛮族", "边军", "边关", "边疆", "战场",
-                   "争霸", "兵王", "战神", "赘婿", "武侠", "修仙", "修真",
-                   "玄幻", "机甲", "星际", "电竞", "网游", "领主")
-_CH_SIGNAL_FEMALE = ("言情", "总裁", "豪门", "宅斗", "宫斗", "嫡女", "庶女",
-                     "甜宠", "虐恋", "宠妻", "宠夫", "王妃", "医妃", "皇后",
-                     "公主", "丫鬟", "千金", "女配", "追妻", "闪婚", "隐婚",
-                     "婆媳", "闺蜜", "女尊", "女强", "军婚", "宫闱")
-
-
-def _channel_anchor(text):
-    """题材信号判频道：男/女信号词单侧命中返回 男生/女生，否则 None。"""
-    t = str(text or "")
-    if not t:
-        return None
-    male = any(w in t for w in _CH_SIGNAL_MALE)
-    female = any(w in t for w in _CH_SIGNAL_FEMALE)
-    if male and not female:
-        return "男生"
-    if female and not male:
-        return "女生"
-    return None
-
-
-_READER_OF = {"fanqie": {"男生": "男频", "女生": "女频"},
-              "qimao": {"男频": "男生", "女频": "女生"}}
-
-
-def _reader_anchor(task, platform, meta):
-    """频道锚：题材信号（goal+简介）优先，其次同书另一平台已定的读者。
-
-    返回 (该平台的读者词, 依据说明) 或 (None, "")。"""
-    corpus = "。".join(x for x in ((task or {}).get("goal") or "",
-                                   (meta or {}).get("summary") or "") if x)
-    sig = _channel_anchor(corpus)
-    if sig:
-        # 信号词是男生/女生：七猫原样用，番茄翻成 男频/女频
-        own = sig if platform == "qimao" else {"男生": "男频", "女生": "女频"}[sig]
-        return own, "题材信号"
-    peer_plat = "qimao" if platform == "fanqie" else "fanqie"
-    peer = (((task or {}).get("book_meta") or {}).get(peer_plat) or {})
-    if peer.get("status") == "done":
-        mapped = _READER_OF[platform].get(
-            (peer.get("data") or {}).get("target_reader"))
-        if mapped:
-            return mapped, "同书另一平台(%s)已定读者" % PLATFORMS[peer_plat]["label"]
-    return None, ""
-
-
-def _apply_fix_note(meta, source):
-    """把 _fill_required_fields 记录的纠偏说明并进 source。
-
-    meta 里的 _fixes 是临时键，随出随清，不落 book_meta 数据（否则
-    repair_existing 的幂等比较会每次不等、启动反复重写）。"""
-    fixes = meta.pop("_fixes", None)
-    if fixes:
-        source = (str(source or "") + "；" + "；".join(str(x) for x in fixes)
-                  ).strip("；")
-    return source
-
-
-def _collect_material(task):
-    """开书资料素材：目标 + 附件上下文 + 故事圣经 + 最近一次大纲 + 第一章开头。"""
-    from . import store  # 惰性导入，同 planner：避免测试环境导入顺序问题
-    parts = ["## 小说目标\n" + (task.get("goal") or "（无）")]
-    ctx = (task.get("context") or "").strip()
-    if ctx:
-        parts.append("## 背景与上下文\n" + ctx[:2000])
-    _, bible, _ = store.read_story_bible(task["id"])
-    if bible:
-        parts.append("## 故事圣经\n" + bible[:6000])
-    outline = None
-    for r in store.task_runs(task["id"]):
-        o = r.get("outline")
-        if o and o.get("chapters") and not o.get("degraded"):
-            outline = o
-            break
-    if outline:
-        lines = ["书名：%s" % (outline.get("book_title") or "（未定）")]
-        for k, c in enumerate(outline["chapters"]):
-            lines.append("第 %d 章《%s》：%s" % (k + 1, c.get("title", ""),
-                                               str(c.get("beats") or "")[:120]))
-        parts.append("## 已有大纲\n" + "\n".join(lines))
-    head = ""
-    try:
-        from . import runner
-        # 章节文件命名并存：连载流水线落 chapter-01.md（两位），有的书是
-        # chapter-001.md（三位）——只认一种会瞎掉「第一章开头」素材
-        # （2026-09-28 马甲书案：明明 8 章成稿在手却报「没有正文」）。
-        for name in ("chapter-001.md", "chapter-01.md", "chapter-1.md"):
-            p = Path(task.get("workdir") or "") / name
-            if p.is_file():
-                head = runner.read_text_any_enc(p)[:_FIRST_CH_HEAD]
-                break
-    except OSError:
-        pass
-    if head:
-        parts.append("## 第一章开头\n" + head)
-    return "\n\n".join(parts), outline
-
-
-def _tags(v, limit=6):
-    """（旧字段兜底保留）通用标签列表清洗：去重/去空/截断。"""
-    if not isinstance(v, list):
-        return []
-    out = []
-    for x in v:
-        s = str(x).strip()[:12]
-        if s and s not in out:
-            out.append(s)
-        if len(out) >= limit:
-            break
-    return out
-
-
-def _name(v, limit):
-    return str(v or "").strip().replace("\n", " ")[:limit]
-
-
-def _summary(v):
-    """剥 markdown 装饰后截断。2026-09-23 实案：简介混成调研报告标题
-    「# 网文选题分析报告…」——模型把研究素材当小说素材时输出的是标题行，
-    至少不能把 #/* 符号和报告腔标题原样带进建书表单。"""
-    t = re.sub(r"^\s*#{1,6}\s*", "", str(v or "").strip())
-    t = re.sub(r"(?m)^\s*[-*•]\s*", "", t)
-    t = t.replace("**", "").replace("__", "")
-    t = re.sub(r"\n{2,}", "\n", t).strip()
-    return t[:500]
-
-
-def _in_table(v, table, limit=10):
-    """值必须在平台真实选项表内；不在返回空串（宁缺勿错——填了假选项用户还得手动改）。"""
-    s = _name(v, limit)
-    return s if s in (table or []) else ""
-
-
-def _filter_table(v, table, limit=6):
-    """多选标签：只保留表内项（保持模型给的顺序），去重截断。"""
-    if not isinstance(v, list):
-        return []
-    out = []
-    for x in v:
-        s = str(x).strip()[:12]
-        if s and s in (table or []) and s not in out:
-            out.append(s)
-        if len(out) >= limit:
-            break
-    return out
-
-
-def _norm(data, platform, goal=""):
-    """规范化平台字段输出；不合规返回 None。
-
-    分类/标签按平台真实选项表校验（bookmeta_catalog）：分类单选不在表内清空、
-    标签多选只留表内项；七猫二级按级联校验（一级无级联记录时宽松放行）。"""
-    if not isinstance(data, dict):
-        return None
-    book_name = _name(data.get("book_name"), 15 if platform == "fanqie" else 15)
-    summary = _summary(data.get("summary"))
-    if not book_name and not summary:
-        return None
-    if platform == "fanqie":
-        reader = str(data.get("target_reader") or "").strip()
-        if reader not in ("男频", "女频"):
-            reader = "女频" if "女" in (goal or "") else "男频"
-        mode = str(data.get("signing_mode") or "").strip()
-        if mode not in FANQIE_MODES:
-            mode = "连载模式"
-        cats, theme, role, plot = _fq_tables(reader)
-        c_plot, c_emo, c_char, c_world = _fq_content_tables()
-        # 兼容旧字段：read_tags/content_tags 并入情节/主题组（老数据不丢）
-        legacy_read = data.get("read_tags") if isinstance(data.get("read_tags"), list) else []
-        return {"book_name": book_name, "signing_mode": mode, "target_reader": reader,
-                "category": _in_table(data.get("category"), cats),
-                "tags_theme": _filter_table(data.get("tags_theme"), theme, 2),
-                "tags_role": _filter_table(data.get("tags_role"), role, 2),
-                "tags_plot": _filter_table(data.get("tags_plot"), plot, 2),
-                "content_plot": _filter_table(data.get("content_plot"), c_plot, 4),
-                "content_emotion": _filter_table(data.get("content_emotion"), c_emo, 2),
-                "content_character": _filter_table(data.get("content_character"), c_char, 4),
-                "content_world": _filter_table(data.get("content_world"), c_world, 1),
-                "legacy_read_tags": _filter_table(legacy_read, plot + theme + role),
-                "protagonist_1": _name(data.get("protagonist_1"), 5),
-                "protagonist_2": _name(data.get("protagonist_2"), 5),
-                "summary": summary}
-    from . import bookmeta_catalog as cat
-    reader = str(data.get("target_reader") or "").strip()
-    if reader not in ("男生", "女生"):
-        reader = "女生" if "女" in (goal or "") else "男生"
-    status = str(data.get("status") or "").strip()
-    if status not in ("连载中", "已完结"):
-        status = "连载中"
-    channel = cat.QIMAO_CATS.get(reader) or {}
-    c_main = _in_table(data.get("category_main"), list(channel))
-    c_sub = _name(data.get("category_sub"), 10)
-    if c_sub and not (c_main and c_sub in (channel.get(c_main) or [])):
-        # 一级/二级不自洽（一级空、或二级不属于该频道——如女生专属的
-        # 「现实故事」配了男生频道）：跨频道反查官方目录纠正，频道与
-        # 一级跟着二级走（表单级联的选项集以此为准）；全目录都没有
-        # 才清空——表单选不出的词不硬填
-        loc = cat.qimao_locate_sub(c_sub)
-        if loc:
-            reader, c_main = loc
-        else:
-            c_sub = ""
-    # 作品标签四组（官方每组必选 1-3）：逐组过滤 + 截到 3；老数据 tags 并入
-    legacy_tags = data.get("tags") if isinstance(data.get("tags"), list) else []
-    groups = {}
-    for key, gname in (("tags_style", "风格"), ("tags_role", "角色"),
-                       ("tags_plot", "情节"), ("tags_bg", "背景")):
-        got = _filter_table(data.get(key), cat.QIMAO_TAG_GROUPS[gname], 3)
-        if len(got) < 3 and legacy_tags:
-            for t in _filter_table(legacy_tags, cat.QIMAO_TAG_GROUPS[gname], 3):
-                if t not in got:
-                    got.append(t)
-                    if len(got) >= 3:
-                        break
-        groups[key] = got
-    return {"book_name": book_name, "target_reader": reader,
-            "category_main": c_main, "category_sub": c_sub,
-            "tags_style": groups["tags_style"], "tags_role": groups["tags_role"],
-            "tags_plot": groups["tags_plot"], "tags_bg": groups["tags_bg"],
-            "protagonist_1": _name(data.get("protagonist_1"), 5),
-            "protagonist_2": _name(data.get("protagonist_2"), 5),
-            "status": status, "summary": summary}
-
-
-def _template_meta(task, platform, outline):
-    """兜底模板：只做素材搬运归纳（书名/简介来自大纲与目标），其余留空待补。
-
-    标签/分类尽力而为：扫描素材文本（goal/圣经/大纲书名）里恰好命中平台
-    选项表的词就带上——模板没有模型理解力，不做语义归纳，命中多少算多少，
-    其余留「待补充」让用户手选（宁缺勿错，不填表外词）。"""
-    goal = task.get("goal") or ""
-    book_name = str((outline or {}).get("book_title") or task.get("title") or "").strip()
-    summary = goal.splitlines()[0][:500] if goal.strip() else ""
-    if not summary:
-        summary = book_name or "（待补充）"    # 全空素材也要有返回，_norm 才不会拒绝
-    material = _material_text(task, outline)
-    data = {"book_name": book_name, "summary": summary}
-    if platform == "fanqie":
-        reader = "女频" if "女" in goal else "男频"
-        cats, theme, role, plot = _fq_tables(reader)
-        c_plot, c_emo, c_char, c_world = _fq_content_tables()
-        data.update({"target_reader": reader,
-                     "category": _scan_table(material, cats, 1),
-                     "tags_theme": _scan_table(material, theme, 2),
-                     "tags_role": _scan_table(material, role, 2),
-                     "tags_plot": _scan_table(material, plot, 2),
-                     "content_plot": _scan_table(material, c_plot, 4),
-                     "content_emotion": _scan_table(material, c_emo, 2),
-                     "content_character": _scan_table(material, c_char, 4),
-                     "content_world": _scan_table(material, c_world, 1)})
-    else:
-        from . import bookmeta_catalog as cat
-        reader = "女生" if "女" in goal else "男生"
-        data["target_reader"] = reader
-        for main, subs in (cat.QIMAO_CATS.get(reader) or {}).items():
-            if main in material:
-                data["category_main"] = main
-                for s in subs:               # 二级名撞素材即选（如「宫闱宅斗」）
-                    if s in material:
-                        data["category_sub"] = s
-                        break
-                break
-        for key, gname in (("tags_style", "风格"), ("tags_role", "角色"),
-                           ("tags_plot", "情节"), ("tags_bg", "背景")):
-            data[key] = _scan_table(material, cat.QIMAO_TAG_GROUPS[gname], 3)
-    return _norm(data, platform, goal)
-
-
-def _scan_table(text, table, limit):
-    """素材文本里命中的选项表词（按表序取，保证真实存在）。"""
-    if not text:
-        return []
-    out = []
-    for word in table:
-        if len(word) >= 2 and word in text:
-            out.append(word)
-            if len(out) >= limit:
-                break
-    return out
-
-
-def _material_text(task, outline):
-    """标签词面扫描用的素材文本：goal + 圣经 + 大纲书名（模板兜底同源）。"""
-    material = task.get("goal") or ""
-    try:
-        _, bible, _ = store_read_bible(task)
-        material += "\n" + (bible or "")
-    except Exception:
-        pass
-    if outline:
-        material += "\n" + str(outline.get("book_title") or "")
-    return material
-
-
-def _fill_tag_gaps(task, platform, meta, outline):
-    """llm 归一化后仍为空的标签组：用素材词面扫描补一轮（命中多少算多少）。
-
-    模型偶尔整组漏给（实测 codex-cli 只填了内容标签前两组）；空着用户就得
-    手选，扫描命中的是官方表内词，补上好过留白——已有的组不动，不覆盖模型。"""
-    material = _material_text(task, outline)
-    if not material:
-        return meta
-    goal = task.get("goal") or ""
-    if platform == "fanqie":
-        reader = meta.get("target_reader") or ("女频" if "女" in goal else "男频")
-        cats, theme, role, plot = _fq_tables(reader)
-        c_plot, c_emo, c_char, c_world = _fq_content_tables()
-        pairs = [("tags_theme", theme, 2), ("tags_role", role, 2), ("tags_plot", plot, 2),
-                 ("content_plot", c_plot, 4), ("content_emotion", c_emo, 2),
-                 ("content_character", c_char, 4), ("content_world", c_world, 1)]
-    else:
-        from . import bookmeta_catalog as cat
-        pairs = [(k, cat.QIMAO_TAG_GROUPS[g], 3) for k, g in
-                 (("tags_style", "风格"), ("tags_role", "角色"),
-                  ("tags_plot", "情节"), ("tags_bg", "背景"))]
-    for key, table, limit in pairs:
-        if not meta.get(key):
-            meta[key] = _scan_table(material, table, limit)
-    return meta
-
-
-def _material_names(material):
-    """从圣经/大纲/章节里提取明确标注的人物名。
-
-    只接受带人物语义的短语（如“女主：阿禾”或“男主叫陆沉”），避免把
-    普通句子里的名词误当成主角。返回顺序去重后的最多两个名字。
-    """
-    text = str(material or "")
-    out = []
-    patterns = (
-        r"(?:主角|主人公|男主(?:角)?|女主(?:角)?|男一|女一|姓名|名字)"
-        r"\s*(?:是|叫|为|：|:)\s*([\u4e00-\u9fffA-Za-z][\u4e00-\u9fffA-Za-z·]{1,7})",
-        r"人物\s*[：:]\s*([^\n。；;]{2,40})",
-    )
-    for pat in patterns:
-        for m in re.finditer(pat, text):
-            raw = m.group(1)
-            chunks = re.split(r"[、,，/和及与]|\s+", raw)
-            for chunk in chunks:
-                name = re.sub(r"[^\u4e00-\u9fffA-Za-z·]", "", chunk)
-                if 2 <= len(name) <= 8 and name not in out:
-                    out.append(name[:5])
-                if len(out) >= 2:
-                    return out
-    return out
-
-
-def _pick_option(material, table, keywords=()):
-    """从官方选项表中按素材命中词选择一个，完全命中不到时用首项。"""
-    text = str(material or "")
-    for item in table:
-        if len(item) >= 2 and item in text:
-            return item
-    for key, item in keywords:
-        if key and key in text and item in table:
-            return item
-    return table[0] if table else ""
-
-
-def _has_value(value):
-    """判断建书字段是否真正有值（兼容列表和历史占位文本）。"""
-    placeholders = {"待补充", "（待补充）", "(待补充)"}
-    if isinstance(value, list):
-        return any((text := str(x or "").strip()) and text not in placeholders
-                   for x in value)
-    text = str(value or "").strip()
-    return bool(text) and text not in placeholders
-
-
-def _md_plain(text):
-    """markdown 轻洗（在 _summary 净化之上）：每行剥标题/列表/引用符与行内
-    强调符，合并成一段正文文字（建书表单简介是单行框）。"""
-    out = []
-    for ln in _summary(text).splitlines():
-        ln = re.sub(r"^\s*(#{1,6}\s*|[-*+]\s+|>\s*|\d+[.、)）]\s*)", "", ln)
-        ln = re.sub(r"[*_`#>]+", "", ln).strip()
-        if ln:
-            out.append(ln)
-    return "".join(out)
-
-
-def _summary_ok(value):
-    """简介够不够格进建书表单：番茄表单硬性要求 50-500 字（占位文案即
-    「请输入50-500字」），markdown 标题/占位符过不了平台提交校验，且自动
-    化流程看不到平台的字段报错，只会误报成「未登录或改版」。"""
-    return 50 <= len(_md_plain(value)) <= 500
-
-
-def _prose_summary(cur, material):
-    """简介兜底清理：markdown 洗成正文；洗完仍不足 50 字就从素材正文取。
-    取不出够长的正文时原样返回——由建书前置闸拦停并提示补写，不硬编。"""
-    text = _md_plain(cur)
-    if len(text) >= 50:
-        return text[:500]
-    body = _md_plain(material)
-    if len(body) >= 50:
-        return body[:500]
-    return str(cur or "").strip()
-
-
-def _fill_required_fields(task, platform, meta, outline, material):
-    """生成阶段的必填字段闸门。
-
-    ``_norm`` 仍保持“表外值清空”的纯校验语义，供旧数据修复和单测使用；
-    这里仅在真正交付作品信息前补齐平台建书表单的必填项，避免 UI 展示一整
-    屏“待补充”而无法直接复制建书。所有分类/标签仍来自官方目录。
-    """
-    from . import bookmeta_catalog as cat
-
-    material = str(material or "")
-    names = _material_names(material)
-    female_words = ("女频", "女生", "女主", "言情", "古言", "恋爱")
-    female = any(x in ((task.get("goal") or "") + material) for x in female_words)
-    defaults = ("沈知微", "顾砚川", "林知夏") if female else ("陆沉", "苏晚", "程野")
-    if not _has_value(meta.get("protagonist_1")):
-        meta["protagonist_1"] = names[0] if names else defaults[0]
-    if not _has_value(meta.get("protagonist_2")):
-        candidate = names[1] if len(names) > 1 else defaults[1]
-        if candidate == meta["protagonist_1"]:
-            candidate = next((name for name in defaults if name != meta["protagonist_1"]), defaults[2])
-        meta["protagonist_2"] = candidate
-
-    # 简介质量：平台建书表单对简介有硬字数校验（番茄 50-500 字），markdown
-    # 标题/占位文本过不了提交且流程看不到报错——洗成正文，不够长从素材取
-    meta["summary"] = _prose_summary(meta.get("summary"), material)
-
-    # 频道锚：题材信号/同书另一平台的读者与模型声称的频道冲突时以锚为准
-    # 翻转（2026-09-24 实案：男频军事文被配成 女生/古代言情/权谋天下，
-    # 级联自洽骗过目录校验）。翻转后旧频道分类作废，由下方素材补齐逻辑
-    # 按正确频道重选；纠偏说明记 _fixes 临时键，由调用方并进 source。
-    anchor, why = _reader_anchor(task, platform, meta)
-    if anchor and meta.get("target_reader") != anchor:
-        meta["_fixes"] = ["频道纠偏：%s→%s（依据：%s）"
-                          % (meta.get("target_reader") or "未填", anchor, why)]
-        meta["target_reader"] = anchor
-        if platform == "qimao":
-            ch = cat.QIMAO_CATS.get(anchor) or {}
-            if meta.get("category_main") not in ch:
-                meta["category_main"] = ""
-            ch_main = meta.get("category_main")
-            if meta.get("category_sub") and not (
-                    ch_main and meta["category_sub"] in (ch.get(ch_main) or [])):
-                meta["category_sub"] = ""
-        else:
-            fq_cats, _, _, _ = _fq_tables(anchor)
-            if meta.get("category") not in fq_cats:
-                meta["category"] = ""
-
-    if platform == "fanqie":
-        reader = meta.get("target_reader") or ("女频" if female else "男频")
-        if reader not in ("男频", "女频"):
-            reader = "女频" if female else "男频"
-        meta["target_reader"] = reader
-        cats, theme, role, plot = _fq_tables(reader)
-        c_plot, c_emo, c_char, c_world = _fq_content_tables()
-        if meta.get("category") not in cats:
-            # 关键词兜底按频道给（2026-09-24 实案《戍边骑奴》：旧表不分频道，
-            # 「戍边→古风世情」是女频词，男频素材永远选不中，全部落到男频表
-            # 首项「西方奇幻」——架空王朝军事文被配成西式魔幻）
-            kw = (("历史", "历史古代"), ("古代", "历史古代"),
-                  ("王朝", "历史脑洞"), ("架空", "历史脑洞"),
-                  ("戍边", "历史古代"), ("边关", "历史古代"),
-                  ("抗战", "抗战谍战"), ("谍战", "抗战谍战"),
-                  ("军", "历史古代"), ("种田", "都市种田"),
-                  ("都市", "都市脑洞"), ("悬疑", "悬疑脑洞")) if reader == "男频" else (
-                 ("宅斗", "宫斗宅斗"), ("宫斗", "宫斗宅斗"),
-                 ("总裁", "豪门总裁"), ("豪门", "豪门总裁"),
-                 ("古代", "古风世情"), ("古言", "古言脑洞"),
-                 ("年代", "年代"), ("种田", "种田"),
-                 ("婚姻", "职场婚恋"), ("恋爱", "职场婚恋"),
-                 ("悬疑", "女频悬疑"), ("快穿", "快穿"))
-            meta["category"] = _pick_option(material, cats, kw)
-        groups = (("tags_theme", theme, 2), ("tags_role", role, 2),
-                  ("tags_plot", plot, 2), ("content_plot", c_plot, 4),
-                  ("content_emotion", c_emo, 2), ("content_character", c_char, 4),
-                  ("content_world", c_world, 1))
-    else:
-        reader = meta.get("target_reader") or ("女生" if female else "男生")
-        if reader not in ("男生", "女生"):
-            reader = "女生" if female else "男生"
-        meta["target_reader"] = reader
-        channel = cat.QIMAO_CATS.get(reader) or cat.QIMAO_CATS["男生"]
-        main = meta.get("category_main")
-        if main not in channel:
-            # 关键词兜底按频道给（2026-09-24 频道锚落地后读方可信，旧的
-            # 「戍边→古代言情」在男生频道下永远选不中白占位）
-            kw = (("戍边", "历史"), ("边关", "历史"), ("军", "军事"),
-                  ("都市", "都市"), ("玄幻", "玄幻奇幻"), ("悬疑", "奇闻异事")
-                  ) if reader == "男生" else (
-                  ("宅斗", "古代言情"), ("宫斗", "古代言情"),
-                  ("古代", "古代言情"), ("总裁", "现代言情"),
-                  ("年代", "现代言情"), ("都市", "现代言情"))
-            main = _pick_option(material, list(channel), kw)
-            meta["category_main"] = main
-        subs = channel.get(main) or []
-        sub = meta.get("category_sub")
-        if sub not in subs:
-            located = cat.qimao_locate_sub(sub)
-            if located:
-                reader, main = located
-                meta["target_reader"] = reader
-                meta["category_main"] = main
-                subs = cat.QIMAO_CATS[reader][main]
-            else:
-                sub = ""
-        if sub not in subs:
-            meta["category_sub"] = _pick_option(material, subs)
-        groups = (("tags_style", cat.QIMAO_TAG_GROUPS["风格"], 3),
-                  ("tags_role", cat.QIMAO_TAG_GROUPS["角色"], 3),
-                  ("tags_plot", cat.QIMAO_TAG_GROUPS["情节"], 3),
-                  ("tags_bg", cat.QIMAO_TAG_GROUPS["背景"], 3))
-    for key, table, limit in groups:
-        values = meta.get(key)
-        if not isinstance(values, list):
-            values = []
-        values = [str(v).strip() for v in values if str(v or "").strip() in table]
-        if not values:
-            values = [_pick_option(material, table)]
-        meta[key] = values[:limit]
-    return meta
-
-
-def store_read_bible(task):
-    """圣经读取的小包装（模板兜底用；异常向上抛由调用方吞掉）。"""
-    from . import store
-    fp, text, err = store.read_story_bible(task.get("id") or "")
-    if err:
-        raise ValueError(err)
-    return fp, text, None
-
-
-def _resolve_author(task):
-    """作者 CLI 兜底用：按「实现」角色选一个 real 智能体（同连载起草的路由口径）。"""
-    try:
-        from . import catalog, manager, registry, router
-        agents = registry.effective_agents(catalog.load(), manager.detect_all())
-        agent, _ = router.pick([a for a in agents if a.get("mode") == "real"],
-                               "implement", task.get("type") or "serial_novel")
-        return agent
-    except Exception:
-        return None
-
-
-def make_book_meta(task, platform, author_agent=None, log_path=None):
-    """生成一个平台的作品信息。返回 (meta dict, source 字符串)；模板兜底永远有返回。"""
-    from . import knowledge, modelhub, planner, runner, skills  # 惰性导入，同 planner
-    material, outline = _collect_material(task)
-    # 素材充分性门：没有大纲也没有章稿/圣经的任务（如「帮我定方向」的选题
-    # 研究任务）没有可推介的小说要素——生成只会产出报告腔垃圾
-    # （2026-09-23 实案：简介=「# 网文选题分析报告…」）。宁明确拒绝。
-    if not outline and "## 第一章开头" not in material and "## 故事圣经" not in material:
-        raise ValueError(
-            "该任务还没有小说大纲、正文或故事圣经，无法生成作品信息——"
-            "先完成大纲/首批章节后再来一键生成")
-    schema = {"fanqie": FANQIE_SCHEMA, "qimao": QIMAO_SCHEMA}.get(platform) or ""
-    plat_label = PLATFORMS[platform]["label"]
-    sk_block, _ = skills.block_for(task)
-    # 实案教训（2026-09-23）：素材偏研究向时模型把简介写成「分析报告标题」
-    summary_rule = ("- summary 必须是面向读者的小说推介文案：点出题材、主角、核心冲突与悬念，"
-                    "200/500 字内成段；绝不允许写成报告标题、要点清单或任务说明。")
-    kb_block = knowledge.block_for(task)
-    if kb_block:
-        sk_block = (sk_block + "\n\n" + kb_block) if sk_block else kb_block
-    goal = task.get("goal") or ""
-    # 真实选项注入：分类/标签表按平台与读者频段给全（模型只能从表里选）
-    options = (_fq_options_block(goal) if platform == "fanqie" else _qm_options_block(goal))
-    prompt = (PROMPT_HEAD.replace("__PLATFORM__", plat_label)
-              .replace("__SKILLS__", sk_block)
-              .replace("__SCHEMA__", schema)
-              .replace("__OPTIONS_RULE__", options + "\n" + summary_rule)
-              + "\n\n## 小说素材\n" + material)
-
-    orch = None
-    if not template_only():
-        try:
-            orch = modelhub.resolve_orchestrator()
-        except Exception:
-            orch = None
-    orch_err = ""
-    if orch:
-        prov, model = orch
-        res = modelhub.chat(prov["id"], model, prompt, max_tokens=4000, timeout=300)
-        planner._log_usage("bookmeta", "bookmeta", task, res, model=model,
-                           provider=prov.get("name", prov.get("id", "")),
-                           provider_id=prov.get("id", ""))
-        if res["ok"]:
-            meta = _norm(runner.extract_json(res.get("text") or ""), platform, goal)
-            if meta:
-                meta = _fill_tag_gaps(task, platform, meta, outline)
-                meta = _fill_required_fields(task, platform, meta, outline, material)
-                return meta, _apply_fix_note(meta, "编排者(%s · %s)"
-                                             % (prov.get("name", prov["id"]), model))
-        orch_err = str(res.get("error") or "返回内容无法解析为作品信息")[:200]
-
-    if author_agent and author_agent.get("mode") == "real" and not template_only():
-        res = runner.run_agent(modelhub.bind_agent(author_agent), prompt,
-                               workdir=task.get("workdir"), readonly=True,
-                               timeout=BOOKMETA_TIMEOUT, log_path=log_path)
-        planner._log_usage("bookmeta", "bookmeta", task, res, agent=author_agent)
-        if res["ok"]:
-            meta = _norm(runner.extract_json(res.get("text") or ""), platform, goal)
-            if meta:
-                meta = _fill_tag_gaps(task, platform, meta, outline)
-                meta = _fill_required_fields(task, platform, meta, outline, material)
-                return meta, _apply_fix_note(meta, "llm(%s)" % author_agent["id"])
-        orch_err = orch_err or str(res.get("error") or "")[:200]
-
-    meta = _template_meta(task, platform, outline)
-    meta = _fill_required_fields(task, platform, meta, outline, material)
-    meta["source"] = ("模板兜底（编排者不可用：%s）" % orch_err) if orch_err else "模板兜底"
-    return meta, _apply_fix_note(meta, meta["source"])
-
-
-# ---- 换名（2026-10-02 用户实案：生成的书名与本站已有作品重名，用户要的是
-# 「换一个名字」而不是整套重摇——重摇会覆盖已核对过的分类/标签/简介）
-
-RENAME_TIMEOUT = 120
-# 排除名单注入上限：全站书名可能上百个，超出的按序丢弃（提示词预算保护；
-# 排序后字母序前 80 个 + 调用方传入的近期撞名足够防撞，撞了还有重试兜底）
-RENAME_EXCLUDE_LIMIT = 80
-
-RENAME_PROMPT = """你是网文编辑。这本书《__OLD__》的书名与已有作品重名了，请另起一个新书名。
-
-书的基本情况：
-__MATERIAL__
-
-硬性要求：
-- 只输出一个 ```json 代码块：{"book_name": "新书名"}，不要输出其他内容；
-- 新书名不超过 15 个字，贴合题材、有网文点击欲，与旧书名同一题材但用词明显不同；
-- 严禁与下面任何一本书重名，也不得只加「2」「新」「之」等后缀敷衍：
-__EXCLUDE__"""
-
-
-def used_book_names():
-    """本站全部已用书名（换名排除名单）：已建书登记（books.json 的 title）
-    + 所有任务 book_meta 各平台 done 态的 book_name。读取异常只降级不阻断
-    （名单不全最坏结果是换出来的名字仍撞，用户再点一次）。"""
-    from .publish import ledger
-    names = set()
-    try:
-        # books.json 结构：{task_id: {platform: {book_id, title, url, ...}}}
-        for by_plat in (ledger.load_books() or {}).values():
-            if not isinstance(by_plat, dict):
-                continue
-            for rec in by_plat.values():
-                t = str((rec or {}).get("title") or "").strip()
-                if t:
-                    names.add(t)
-    except Exception:
-        pass
-    try:
-        from . import store
-        for task in store.list_tasks(limit=10 ** 9):
-            for entry in (task.get("book_meta") or {}).values():
-                if isinstance(entry, dict) and entry.get("status") == "done":
-                    n = str((entry.get("data") or {}).get("book_name") or "").strip()
-                    if n:
-                        names.add(n)
-    except Exception:
-        pass
-    return names
-
-
-def rename_book(task, platform, old_name, exclude_names=()):
-    """给已生成的作品信息换一个不重名的书名。返回 (new_name, "") 或 ("", 错误)。
-
-    只产出新名，不写盘——调用方负责更新 book_meta 与归档同步；这样模型失败
-    时现有数据原样保留（区别于整套生成的线程+状态机，轻调用同步返回即可）。
-    新名若仍撞排除名单，把撞名项并入名单重试一次。"""
-    from . import modelhub, planner, runner
-    old_name = str(old_name or "").strip()
-    exclude = {str(x).strip() for x in (exclude_names or ()) if str(x).strip()}
-    exclude.add(old_name)
-    goal_lines = [x.strip() for x in (task.get("goal") or "").splitlines() if x.strip()]
-    meta_lines = []
-    if goal_lines:
-        meta_lines.append("目标：" + goal_lines[0][:200])
-    try:
-        _, bible, _ = store_read_bible(task)
-        if bible:
-            meta_lines.append("故事圣经摘录：" + bible[:400])
-    except Exception:
-        pass
-    material = "\n".join(meta_lines) if meta_lines else "（素材缺失，按旧书名题材另起）"
-    orch = None
-    try:
-        orch = modelhub.resolve_orchestrator()
-    except Exception:
-        orch = None
-    if not orch:
-        return "", "编排者模型不可用，无法自动换名（可稍后重试）"
-    prov, model = orch
-    last = ""
-    for _ in range(2):                           # 首轮撞名单 → 带撞名项重试一次
-        ex_block = "\n".join("- " + n for n in sorted(exclude)[:RENAME_EXCLUDE_LIMIT])
-        prompt = (RENAME_PROMPT.replace("__OLD__", old_name)
-                  .replace("__MATERIAL__", material)
-                  .replace("__EXCLUDE__", ex_block or "（暂无）"))
-        # max_tokens 4000 对齐主生成（2026-10-03 实案：glm-5.3 思考计入
-        # max_tokens，给 400 三次全被思考烧穿 output 打满，JSON 没成形）
-        res = modelhub.chat(prov["id"], model, prompt, max_tokens=4000,
-                            timeout=RENAME_TIMEOUT)
-        planner._log_usage("bookmeta", "bookmeta-rename", task, res, model=model,
-                           provider=prov.get("name", prov.get("id", "")),
-                           provider_id=prov.get("id", ""))
-        if not res["ok"]:
-            last = str(res.get("error") or "模型调用失败")[:200]
-            continue
-        text = str(res.get("text") or "")
-        data = runner.extract_json(text)
-        new_name = _name(data.get("book_name") if isinstance(data, dict) else "", 15)
-        if not new_name:
-            # 兜底：思考烧预算时 JSON 可能没成形，正文里有《书名》/「书名」
-            # 形态就抠出来用——抠出的名字仍要过下面的排除名单检查
-            m = re.search(r"[《「]([^《》「」]{1,15})[》」]", text)
-            if m:
-                new_name = _name(m.group(1), 15)
-        if not new_name:
-            last = "返回内容无法解析出新书名"
-            continue
-        if new_name in exclude:
-            last = "新书名《%s》仍与本站已有作品重名" % new_name
-            exclude.add(new_name)
-            continue
-        return new_name, ""
-    return "", last or "换名失败，请稍后重试"
-
-
-def render_markdown(task, platform, meta):
-    """落工作目录的归档 Markdown（与前端弹框同一份字段顺序）。"""
-    plat = PLATFORMS[platform]["label"]
-    lines = ["# 作品信息（%s）" % plat, "",
-             "> 任务：%s　生成于 %s　来源：%s" % (
-                 task.get("title") or task.get("id") or "",
-                 time.strftime("%Y-%m-%d %H:%M"), meta.get("source") or "")]
-    lines.append("")
-    for key, label in FIELD_LABELS[platform]:
-        v = meta.get(key)
-        if isinstance(v, list):
-            v = "、".join(v) if v else "（待补充）"
-        elif key == "summary":
-            v = (v or "").strip() or "（待补充）"
-        else:
-            v = str(v or "").strip() or "（待补充）"
-        lines.append("- **%s**：%s" % (label, v))
-    lines.append("")
-    lines.append("> 各字段可直接复制进 %s 建书表单；分类与标签均选自平台真实选项表。" % plat)
-    return "\n".join(lines)
-
-
-def generate_async(task_id, platform, author_agent=None):
-    """后台线程入口：跑生成并把终态写回任务 book_meta；成功时落工作目录归档。
-
-    状态机：running（路由置入）→ done / failed。任何异常都落 failed，不留悬挂。"""
-    from . import store
-    task = store.get_task(task_id)
-    if not task:
-        return
-    try:
-        meta, source = make_book_meta(task, platform, author_agent=author_agent)
-        entry = {"status": "done", "data": meta, "source": source,
-                 "at": time.strftime("%Y-%m-%d %H:%M:%S")}
-        fp = Path(task.get("workdir") or "") / PLATFORMS[platform]["file"]
-        if str(fp.parent) and fp.parent.is_dir():
-            fp.write_text(render_markdown(task, platform, meta), encoding="utf-8")
-            entry["file"] = PLATFORMS[platform]["file"]
-    except Exception as e:                       # 兜底线程不能静默死掉
-        entry = {"status": "failed", "error": str(e)[:300],
-                 "at": time.strftime("%Y-%m-%d %H:%M:%S")}
-    cur = store.get_task(task_id)
-    if not cur:
-        return
-    prev = (cur.get("book_meta") or {}).get(platform) or {}
-    if prev.get("status") != "running":          # 期间被删/被重置：丢弃结果
-        return
-    store.set_book_meta(task_id, platform, entry)
-
-
-def recover_orphans():
-    """启动收尸：后台生成线程随进程重启死掉后，落盘的 running 无人写终态——
-    前端永远「生成中」（按钮禁用）+ 接口幂等拒绝，卡死不可自愈。启动时把
-    遗留 running 统一改判 failed，交还「重试」按钮。返回改判条数。"""
-    from . import store
-    n = 0
-    for task in store.list_tasks(limit=10 ** 9):
-        for plat, entry in (task.get("book_meta") or {}).items():
-            if isinstance(entry, dict) and entry.get("status") == "running":
-                store.set_book_meta(task["id"], plat, {
-                    "status": "failed",
-                    "error": "生成随服务重启中断，请点重试",
-                    "at": time.strftime("%Y-%m-%d %H:%M:%S")})
-                n += 1
-    return n
-
-
-def repair_existing():
-    """补齐历史已生成记录中的必填字段。
-
-    旧版本允许分类/主角为空并把结果标成 done；只要字段缺失就按当前官方
-    目录和已有素材补一次，已有非空字段保持不变。返回修复条数，供启动日志
-    和升级诊断使用。
-    """
-    from . import store
-    repaired = 0
-    for task in store.list_tasks(limit=10 ** 9):
-        book_meta = task.get("book_meta") or {}
-        for platform in PLATFORMS:
-            entry = book_meta.get(platform) or {}
-            if entry.get("status") != "done" or not isinstance(entry.get("data"), dict):
-                continue
-            data = dict(entry["data"])
-            if platform == "fanqie":
-                reader = data.get("target_reader")
-                cats, _, _, _ = _fq_tables(reader if reader in ("男频", "女频") else "男频")
-                required_ok = (
-                    data.get("category") in cats
-                    and all(_has_value(data.get(key)) for key in (
-                        "tags_theme", "tags_role", "tags_plot", "content_plot",
-                        "content_emotion", "content_character", "content_world"))
-                    and _has_value(data.get("protagonist_1"))
-                    and _has_value(data.get("protagonist_2"))
-                    and _summary_ok(data.get("summary")))
-            else:
-                from . import bookmeta_catalog as cat
-                reader = data.get("target_reader")
-                channel = cat.QIMAO_CATS.get(reader) or {}
-                main = data.get("category_main")
-                required_ok = (
-                    main in channel
-                    and data.get("category_sub") in (channel.get(main) or [])
-                    and all(_has_value(data.get(key)) for key in (
-                        "tags_style", "tags_role", "tags_plot", "tags_bg"))
-                    and _has_value(data.get("protagonist_1"))
-                    and _has_value(data.get("protagonist_2"))
-                    and _summary_ok(data.get("summary")))
-            # 频道失配也视为待修复（2026-09-24 实案：级联自洽的跨频道组合
-            # 字段全齐，required_ok 挡不住，只能靠题材信号识别）
-            anchor, _ = _reader_anchor(task, platform,
-                                       {"summary": data.get("summary") or ""})
-            if anchor and anchor != data.get("target_reader"):
-                required_ok = False
-            if required_ok:
-                continue
-            try:
-                material, outline = _collect_material(task)
-                fixed = _fill_required_fields(task, platform, data, outline, material)
-                fixes = fixed.pop("_fixes", None)   # 先清临时键再比幂等
-                if fixed == entry["data"]:
-                    continue
-                payload = dict(entry)
-                payload["data"] = fixed
-                note = "已补齐必填字段"
-                if fixes:
-                    note += "；" + "；".join(str(x) for x in fixes)
-                payload["source"] = (str(entry.get("source") or "") + "；" + note).lstrip("；")
-                store.set_book_meta(task["id"], platform, payload)
-                try:
-                    fp = Path(task.get("workdir") or "") / PLATFORMS[platform]["file"]
-                    if fp.parent.is_dir():
-                        fp.write_text(render_markdown(task, platform, fixed), encoding="utf-8")
-                except OSError:
-                    pass
-                repaired += 1
-            except Exception:
-                continue
-    return repaired
+# -*- coding: utf-8 -*-
+"""作品信息一键生成：按发布平台（番茄/七猫）产出建书表单要填的开书资料。
+
+用户在详情页「成果」分区点按钮触发（不做自动生成），后台线程调编排者模型
+提炼 大纲 + 故事圣经 + 已有章节 的精华，按平台表单字段归一化后存任务
+book_meta 字段，同时在工作目录落一份 Markdown 归档。降级链与大纲一致：
+编排者 API → 作者 CLI → 模板兜底（模板只做搬运归纳，必填字段仍保证可建书）。
+"""
+from __future__ import annotations
+
+import os
+import re
+import time
+from pathlib import Path
+
+PLATFORMS = {
+    "fanqie": {"label": "番茄", "file": "作品信息-番茄.md"},
+    "qimao": {"label": "七猫", "file": "作品信息-七猫.md"},
+}
+# 单平台生成上限：编排者提炼一段简介级文本是轻任务，900s 足够含重试
+BOOKMETA_TIMEOUT = 900
+_FIRST_CH_HEAD = 1500   # 注入提示词的第一章开头字符数
+
+
+def template_only():
+    """只走模板兜底（跳过编排者/作者 CLI）：env TUTTI_BOOKMETA_TEMPLATE_ONLY=1。
+
+    离线环境、不想为开书资料再花一次模型调用、以及测试要确定性输出时使用；
+    产出仍是结构完整的字段，书名/简介直接取大纲与目标，平台必填字段由生成阶段
+    的官方目录兜底补齐。"""
+    return str(os.environ.get("TUTTI_BOOKMETA_TEMPLATE_ONLY") or "").strip() in ("1", "true", "yes")
+
+
+# 各平台建书表单的字段展示顺序（前端卡片与 Markdown 归档共用）。
+# 番茄官方表单（2026-09 实抓）：签约模式=连载模式/完本模式；阅读标签=主分类(必选1)
+# +主题/角色/情节(各≤2)；内容标签=情节(≤4)/情感(≤2)/人设(≤4)/世界观(≤1)。
+# 七猫官方表单：一级→二级级联按频道；作品标签四组每组必选 1-3 个。
+FIELD_LABELS = {
+    "fanqie": [
+        ("book_name", "作品名"), ("signing_mode", "签约模式"),
+        ("target_reader", "目标读者"), ("category", "主分类"),
+        ("tags_theme", "主题标签"), ("tags_role", "角色标签"),
+        ("tags_plot", "情节标签"),
+        ("content_plot", "内容·情节"), ("content_emotion", "内容·情感"),
+        ("content_character", "内容·人设"), ("content_world", "内容·世界观"),
+        ("protagonist_1", "主角名1"),
+        ("protagonist_2", "主角名2"), ("summary", "作品简介"),
+    ],
+    "qimao": [
+        ("book_name", "作品名称"), ("target_reader", "目标读者"),
+        ("category_main", "一级分类"), ("category_sub", "二级分类"),
+        ("tags_style", "风格标签"), ("tags_role", "角色标签"),
+        ("tags_plot", "情节标签"), ("tags_bg", "背景标签"),
+        ("protagonist_1", "主角名1"),
+        ("protagonist_2", "主角名2"), ("status", "作品状态"),
+        ("summary", "作品简介"),
+    ],
+}
+
+PROMPT_HEAD = """你是网文编辑，熟悉 __PLATFORM__ 建书表单的填法。
+
+__SKILLS__
+请依据下方的小说素材（目标/故事圣经/大纲/已有章节开头），为这本书生成一套
+「作品信息」，直接可用于在 __PLATFORM__ 创建作品。只输出一个 ```json 代码块，
+不要输出其他内容。JSON 结构：
+__SCHEMA__
+硬性要求：
+- 字段值必须贴合素材内容，不编造素材里没有的设定；简介不超过 500 字，突出卖点与钩子；
+- 每个标签数组字段都必须至少给 1 个、并尽量贴着上限给满，只从对应组的选项表里选；
+  任何一组留空数组、漏掉字段或自造表外词都算错误；
+- __OPTIONS_RULE__
+- 题材健康，无违规内容，符合平台签约调性。
+"""
+
+# 签约模式真实选项（番茄建书表单 2026-09 实抓）：连载模式 / 完本模式
+FANQIE_MODES = ("连载模式", "完本模式")
+
+FANQIE_SCHEMA = ('{"book_name": "作品名(15字内)", "signing_mode": "连载模式 或 完本模式", '
+                 '"target_reader": "男频 或 女频", "category": "主分类(必选且只能选一个，从下方分类表选)", '
+                 '"tags_theme": ["主题标签", …最多2个], "tags_role": ["角色标签", …最多2个], '
+                 '"tags_plot": ["情节标签", …最多2个], '
+                 '"content_plot": ["内容标签·情节", …最多4个], '
+                 '"content_emotion": ["内容标签·情感", …最多2个], '
+                 '"content_character": ["内容标签·人设", …最多4个], '
+                 '"content_world": ["内容标签·世界观", …最多1个], '
+                 '"protagonist_1": "主角名1(5字内)", '
+                 '"protagonist_2": "主角名2(5字内，没有就空串)", "summary": "作品简介(500字内)"}')
+
+QIMAO_SCHEMA = ('{"book_name": "作品名称(18字内)", "target_reader": "男生 或 女生", '
+                '"category_main": "一级分类(从该频道一级分类表选一个)", '
+                '"category_sub": "二级分类(必须是该一级下的二级)", '
+                '"tags_style": ["风格标签", 必选1-3个], '
+                '"tags_role": ["角色标签", 必选1-3个], '
+                '"tags_plot": ["情节标签", 必选1-3个], '
+                '"tags_bg": ["背景标签", 必选1-3个], '
+                '"protagonist_1": "主角名1(5字内)", '
+                '"protagonist_2": "主角名2(5字内，没有就空串)", "status": "连载中 或 已完结", '
+                '"summary": "作品简介(200字内)"}')
+
+
+def _fq_tables(reader):
+    """番茄按目标读者取对应选项表（男频/女频两套完全独立的分类与标签）。"""
+    from . import bookmeta_catalog as cat
+    if reader == "女频":
+        return (cat.FANQIE_FEMALE_CATEGORIES, cat.FANQIE_FEMALE_TAGS_THEME,
+                cat.FANQIE_FEMALE_TAGS_ROLE, cat.FANQIE_FEMALE_TAGS_PLOT)
+    return (cat.FANQIE_MALE_CATEGORIES, cat.FANQIE_MALE_TAGS_THEME,
+            cat.FANQIE_MALE_TAGS_ROLE, cat.FANQIE_MALE_TAGS_PLOT)
+
+
+def _fq_content_tables():
+    """番茄内容标签四组（男女性共用）：情节/情感/人设/世界观。"""
+    from . import bookmeta_catalog as cat
+    return (cat.FANQIE_CONTENT_PLOT, cat.FANQIE_CONTENT_EMOTION,
+            cat.FANQIE_CONTENT_CHARACTER, cat.FANQIE_CONTENT_WORLD)
+
+
+def _fq_options_block(goal):
+    """番茄提示词的真实选项注入：按 goal 推断读者给出对应分类与标签表。
+
+    2026-09-24 起首行加频道判定规则：素材明显属于另一频时 target_reader
+    如实填、分类留空（规整层按该频表校验并兜底补齐），防「注入男频表+
+    女频素材」时硬凑男频分类。"""
+    reader = "女频" if "女" in (goal or "") else "男频"
+    cats, theme, role, plot = _fq_tables(reader)
+    c_plot, c_emo, c_char, c_world = _fq_content_tables()
+    return ("先按素材判定目标读者（战争/军功/历史争霸/都市异能/系统→男频；"
+            "总裁豪门/宅斗宫斗/年代/婚恋情感线为主→女频），target_reader 如实填；"
+            "主分类从下方按目标读者 %s 预注入的表里选，素材明显属于另一频时"
+            "分类留空待系统按该频表补齐。主分类要贴世界观而非字面词——架空"
+            "王朝/军事/军功题材选历史古代或历史脑洞，西方奇幻仅限西式魔幻"
+            "世界观。\n"
+            "目标读者（%s）可用的主分类表（必选且只能选一个）：%s\n"
+            "阅读标签三组各最多选 2 个——主题组：%s\n角色组：%s\n情节组（部分，选最贴合的）：%s\n"
+            "内容标签四组——情节组（最多4个，部分）：%s\n情感组（最多2个）：%s\n"
+            "人设组（最多4个，部分）：%s\n世界观组（最多1个）：%s"
+            % (reader, reader, "、".join(cats), "、".join(theme), "、".join(role),
+               "、".join(plot[:50]), "、".join(c_plot[:60]), "、".join(c_emo),
+               "、".join(c_char[:60]), "、".join(c_world)))
+
+
+def _qm_options_block(goal):
+    """七猫提示词的真实选项注入：双频道级联 + 频道判定规则 + 跨频道禁令。
+
+    2026-09-24 前只按 goal 猜频道注入单频道表，但挡不住模型自带另一频道
+    的分类知识跨频道自造组合（男频军事文被配成 女生/古代言情/权谋天下，
+    级联还自洽）——必须双频道全量注入并明示判定规则。"""
+    from . import bookmeta_catalog as cat
+    cascades = ["%s频道（一级→二级）：%s" % (ch, "；".join(
+        "%s（二级：%s）" % (m, "、".join(v)) for m, v in mains.items()))
+        for ch, mains in cat.QIMAO_CATS.items()]
+    pool = ["%s组（必选1-3个）：%s" % (g, "、".join(t))
+            for g, t in cat.QIMAO_TAG_GROUPS.items()]
+    return ("先按素材判定目标读者频道：战争/军功/练兵/历史争霸/都市异能/"
+            "系统流/玄幻修仙→男生；总裁豪门/宅斗宫斗/年代/婚恋情感线为主→"
+            "女生。target_reader 必须填判定的频道，一级/二级分类只能从该频道"
+            "的级联表里选，严禁跨频道——如「权谋天下」只存在于女生·古代言情"
+            "下，男频军事历史文不得选用；「架空历史」属男生·历史。\n"
+            "%s\n作品标签四组（男女生同池）：%s"
+            % ("。".join(cascades), "；".join(pool)))
+
+
+# ---- 频道题材信号（2026-09-24 实案：男频军事文《戍边骑奴》被生成端配成
+# 女生/古代言情/权谋天下——「权谋天下」字面贴合朝堂线但只存在于女生频道，
+# 模型跨频道自造组合且级联自洽，目录校验拦不住，只能靠题材信号判频道。
+# 必须保守：信号单侧命中才判，双侧命中/零命中一律不判（宁可不纠不可错纠）。
+# 证据语料只用 goal+简介：分类字段是嫌疑输出、标签池男女同池，都不算证据。
+_CH_SIGNAL_MALE = ("军功", "戍边", "骑奴", "特种兵", "练兵", "杀伐", "阵斩",
+                   "百夫长", "大军", "蛮族", "边军", "边关", "边疆", "战场",
+                   "争霸", "兵王", "战神", "赘婿", "武侠", "修仙", "修真",
+                   "玄幻", "机甲", "星际", "电竞", "网游", "领主")
+_CH_SIGNAL_FEMALE = ("言情", "总裁", "豪门", "宅斗", "宫斗", "嫡女", "庶女",
+                     "甜宠", "虐恋", "宠妻", "宠夫", "王妃", "医妃", "皇后",
+                     "公主", "丫鬟", "千金", "女配", "追妻", "闪婚", "隐婚",
+                     "婆媳", "闺蜜", "女尊", "女强", "军婚", "宫闱")
+
+
+def _channel_anchor(text):
+    """题材信号判频道：男/女信号词单侧命中返回 男生/女生，否则 None。"""
+    t = str(text or "")
+    if not t:
+        return None
+    male = any(w in t for w in _CH_SIGNAL_MALE)
+    female = any(w in t for w in _CH_SIGNAL_FEMALE)
+    if male and not female:
+        return "男生"
+    if female and not male:
+        return "女生"
+    return None
+
+
+_READER_OF = {"fanqie": {"男生": "男频", "女生": "女频"},
+              "qimao": {"男频": "男生", "女频": "女生"}}
+
+
+def _reader_anchor(task, platform, meta):
+    """频道锚：题材信号（goal+简介）优先，其次同书另一平台已定的读者。
+
+    返回 (该平台的读者词, 依据说明) 或 (None, "")。"""
+    corpus = "。".join(x for x in ((task or {}).get("goal") or "",
+                                   (meta or {}).get("summary") or "") if x)
+    sig = _channel_anchor(corpus)
+    if sig:
+        # 信号词是男生/女生：七猫原样用，番茄翻成 男频/女频
+        own = sig if platform == "qimao" else {"男生": "男频", "女生": "女频"}[sig]
+        return own, "题材信号"
+    peer_plat = "qimao" if platform == "fanqie" else "fanqie"
+    peer = (((task or {}).get("book_meta") or {}).get(peer_plat) or {})
+    if peer.get("status") == "done":
+        mapped = _READER_OF[platform].get(
+            (peer.get("data") or {}).get("target_reader"))
+        if mapped:
+            return mapped, "同书另一平台(%s)已定读者" % PLATFORMS[peer_plat]["label"]
+    return None, ""
+
+
+def _apply_fix_note(meta, source):
+    """把 _fill_required_fields 记录的纠偏说明并进 source。
+
+    meta 里的 _fixes 是临时键，随出随清，不落 book_meta 数据（否则
+    repair_existing 的幂等比较会每次不等、启动反复重写）。"""
+    fixes = meta.pop("_fixes", None)
+    if fixes:
+        source = (str(source or "") + "；" + "；".join(str(x) for x in fixes)
+                  ).strip("；")
+    return source
+
+
+def _collect_material(task):
+    """开书资料素材：目标 + 附件上下文 + 故事圣经 + 最近一次大纲 + 第一章开头。"""
+    from . import store  # 惰性导入，同 planner：避免测试环境导入顺序问题
+    parts = ["## 小说目标\n" + (task.get("goal") or "（无）")]
+    ctx = (task.get("context") or "").strip()
+    if ctx:
+        parts.append("## 背景与上下文\n" + ctx[:2000])
+    _, bible, _ = store.read_story_bible(task["id"])
+    if bible:
+        parts.append("## 故事圣经\n" + bible[:6000])
+    outline = None
+    for r in store.task_runs(task["id"]):
+        o = r.get("outline")
+        if o and o.get("chapters") and not o.get("degraded"):
+            outline = o
+            break
+    if outline:
+        lines = ["书名：%s" % (outline.get("book_title") or "（未定）")]
+        for k, c in enumerate(outline["chapters"]):
+            lines.append("第 %d 章《%s》：%s" % (k + 1, c.get("title", ""),
+                                               str(c.get("beats") or "")[:120]))
+        parts.append("## 已有大纲\n" + "\n".join(lines))
+    head = ""
+    try:
+        from . import runner
+        # 章节文件命名并存：连载流水线落 chapter-01.md（两位），有的书是
+        # chapter-001.md（三位）——只认一种会瞎掉「第一章开头」素材
+        # （2026-09-28 马甲书案：明明 8 章成稿在手却报「没有正文」）。
+        for name in ("chapter-001.md", "chapter-01.md", "chapter-1.md"):
+            p = Path(task.get("workdir") or "") / name
+            if p.is_file():
+                head = runner.read_text_any_enc(p)[:_FIRST_CH_HEAD]
+                break
+    except OSError:
+        pass
+    if head:
+        parts.append("## 第一章开头\n" + head)
+    return "\n\n".join(parts), outline
+
+
+def _tags(v, limit=6):
+    """（旧字段兜底保留）通用标签列表清洗：去重/去空/截断。"""
+    if not isinstance(v, list):
+        return []
+    out = []
+    for x in v:
+        s = str(x).strip()[:12]
+        if s and s not in out:
+            out.append(s)
+        if len(out) >= limit:
+            break
+    return out
+
+
+def _name(v, limit):
+    return str(v or "").strip().replace("\n", " ")[:limit]
+
+
+def _summary(v):
+    """剥 markdown 装饰后截断。2026-09-23 实案：简介混成调研报告标题
+    「# 网文选题分析报告…」——模型把研究素材当小说素材时输出的是标题行，
+    至少不能把 #/* 符号和报告腔标题原样带进建书表单。"""
+    t = re.sub(r"^\s*#{1,6}\s*", "", str(v or "").strip())
+    t = re.sub(r"(?m)^\s*[-*•]\s*", "", t)
+    t = t.replace("**", "").replace("__", "")
+    t = re.sub(r"\n{2,}", "\n", t).strip()
+    return t[:500]
+
+
+def _in_table(v, table, limit=10):
+    """值必须在平台真实选项表内；不在返回空串（宁缺勿错——填了假选项用户还得手动改）。"""
+    s = _name(v, limit)
+    return s if s in (table or []) else ""
+
+
+def _filter_table(v, table, limit=6):
+    """多选标签：只保留表内项（保持模型给的顺序），去重截断。"""
+    if not isinstance(v, list):
+        return []
+    out = []
+    for x in v:
+        s = str(x).strip()[:12]
+        if s and s in (table or []) and s not in out:
+            out.append(s)
+        if len(out) >= limit:
+            break
+    return out
+
+
+def _norm(data, platform, goal=""):
+    """规范化平台字段输出；不合规返回 None。
+
+    分类/标签按平台真实选项表校验（bookmeta_catalog）：分类单选不在表内清空、
+    标签多选只留表内项；七猫二级按级联校验（一级无级联记录时宽松放行）。"""
+    if not isinstance(data, dict):
+        return None
+    book_name = _name(data.get("book_name"), 15 if platform == "fanqie" else 15)
+    summary = _summary(data.get("summary"))
+    if not book_name and not summary:
+        return None
+    if platform == "fanqie":
+        reader = str(data.get("target_reader") or "").strip()
+        if reader not in ("男频", "女频"):
+            reader = "女频" if "女" in (goal or "") else "男频"
+        mode = str(data.get("signing_mode") or "").strip()
+        if mode not in FANQIE_MODES:
+            mode = "连载模式"
+        cats, theme, role, plot = _fq_tables(reader)
+        c_plot, c_emo, c_char, c_world = _fq_content_tables()
+        # 兼容旧字段：read_tags/content_tags 并入情节/主题组（老数据不丢）
+        legacy_read = data.get("read_tags") if isinstance(data.get("read_tags"), list) else []
+        return {"book_name": book_name, "signing_mode": mode, "target_reader": reader,
+                "category": _in_table(data.get("category"), cats),
+                "tags_theme": _filter_table(data.get("tags_theme"), theme, 2),
+                "tags_role": _filter_table(data.get("tags_role"), role, 2),
+                "tags_plot": _filter_table(data.get("tags_plot"), plot, 2),
+                "content_plot": _filter_table(data.get("content_plot"), c_plot, 4),
+                "content_emotion": _filter_table(data.get("content_emotion"), c_emo, 2),
+                "content_character": _filter_table(data.get("content_character"), c_char, 4),
+                "content_world": _filter_table(data.get("content_world"), c_world, 1),
+                "legacy_read_tags": _filter_table(legacy_read, plot + theme + role),
+                "protagonist_1": _name(data.get("protagonist_1"), 5),
+                "protagonist_2": _name(data.get("protagonist_2"), 5),
+                "summary": summary}
+    from . import bookmeta_catalog as cat
+    reader = str(data.get("target_reader") or "").strip()
+    if reader not in ("男生", "女生"):
+        reader = "女生" if "女" in (goal or "") else "男生"
+    status = str(data.get("status") or "").strip()
+    if status not in ("连载中", "已完结"):
+        status = "连载中"
+    channel = cat.QIMAO_CATS.get(reader) or {}
+    c_main = _in_table(data.get("category_main"), list(channel))
+    c_sub = _name(data.get("category_sub"), 10)
+    if c_sub and not (c_main and c_sub in (channel.get(c_main) or [])):
+        # 一级/二级不自洽（一级空、或二级不属于该频道——如女生专属的
+        # 「现实故事」配了男生频道）：跨频道反查官方目录纠正，频道与
+        # 一级跟着二级走（表单级联的选项集以此为准）；全目录都没有
+        # 才清空——表单选不出的词不硬填
+        loc = cat.qimao_locate_sub(c_sub)
+        if loc:
+            reader, c_main = loc
+        else:
+            c_sub = ""
+    # 作品标签四组（官方每组必选 1-3）：逐组过滤 + 截到 3；老数据 tags 并入
+    legacy_tags = data.get("tags") if isinstance(data.get("tags"), list) else []
+    groups = {}
+    for key, gname in (("tags_style", "风格"), ("tags_role", "角色"),
+                       ("tags_plot", "情节"), ("tags_bg", "背景")):
+        got = _filter_table(data.get(key), cat.QIMAO_TAG_GROUPS[gname], 3)
+        if len(got) < 3 and legacy_tags:
+            for t in _filter_table(legacy_tags, cat.QIMAO_TAG_GROUPS[gname], 3):
+                if t not in got:
+                    got.append(t)
+                    if len(got) >= 3:
+                        break
+        groups[key] = got
+    return {"book_name": book_name, "target_reader": reader,
+            "category_main": c_main, "category_sub": c_sub,
+            "tags_style": groups["tags_style"], "tags_role": groups["tags_role"],
+            "tags_plot": groups["tags_plot"], "tags_bg": groups["tags_bg"],
+            "protagonist_1": _name(data.get("protagonist_1"), 5),
+            "protagonist_2": _name(data.get("protagonist_2"), 5),
+            "status": status, "summary": summary}
+
+
+def _template_meta(task, platform, outline):
+    """兜底模板：只做素材搬运归纳（书名/简介来自大纲与目标），其余留空待补。
+
+    标签/分类尽力而为：扫描素材文本（goal/圣经/大纲书名）里恰好命中平台
+    选项表的词就带上——模板没有模型理解力，不做语义归纳，命中多少算多少，
+    其余留「待补充」让用户手选（宁缺勿错，不填表外词）。"""
+    goal = task.get("goal") or ""
+    book_name = str((outline or {}).get("book_title") or task.get("title") or "").strip()
+    summary = goal.splitlines()[0][:500] if goal.strip() else ""
+    if not summary:
+        summary = book_name or "（待补充）"    # 全空素材也要有返回，_norm 才不会拒绝
+    material = _material_text(task, outline)
+    data = {"book_name": book_name, "summary": summary}
+    if platform == "fanqie":
+        reader = "女频" if "女" in goal else "男频"
+        cats, theme, role, plot = _fq_tables(reader)
+        c_plot, c_emo, c_char, c_world = _fq_content_tables()
+        data.update({"target_reader": reader,
+                     "category": _scan_table(material, cats, 1),
+                     "tags_theme": _scan_table(material, theme, 2),
+                     "tags_role": _scan_table(material, role, 2),
+                     "tags_plot": _scan_table(material, plot, 2),
+                     "content_plot": _scan_table(material, c_plot, 4),
+                     "content_emotion": _scan_table(material, c_emo, 2),
+                     "content_character": _scan_table(material, c_char, 4),
+                     "content_world": _scan_table(material, c_world, 1)})
+    else:
+        from . import bookmeta_catalog as cat
+        reader = "女生" if "女" in goal else "男生"
+        data["target_reader"] = reader
+        for main, subs in (cat.QIMAO_CATS.get(reader) or {}).items():
+            if main in material:
+                data["category_main"] = main
+                for s in subs:               # 二级名撞素材即选（如「宫闱宅斗」）
+                    if s in material:
+                        data["category_sub"] = s
+                        break
+                break
+        for key, gname in (("tags_style", "风格"), ("tags_role", "角色"),
+                           ("tags_plot", "情节"), ("tags_bg", "背景")):
+            data[key] = _scan_table(material, cat.QIMAO_TAG_GROUPS[gname], 3)
+    return _norm(data, platform, goal)
+
+
+def _scan_table(text, table, limit):
+    """素材文本里命中的选项表词（按表序取，保证真实存在）。"""
+    if not text:
+        return []
+    out = []
+    for word in table:
+        if len(word) >= 2 and word in text:
+            out.append(word)
+            if len(out) >= limit:
+                break
+    return out
+
+
+def _material_text(task, outline):
+    """标签词面扫描用的素材文本：goal + 圣经 + 大纲书名（模板兜底同源）。"""
+    material = task.get("goal") or ""
+    try:
+        _, bible, _ = store_read_bible(task)
+        material += "\n" + (bible or "")
+    except Exception:
+        pass
+    if outline:
+        material += "\n" + str(outline.get("book_title") or "")
+    return material
+
+
+def _fill_tag_gaps(task, platform, meta, outline):
+    """llm 归一化后仍为空的标签组：用素材词面扫描补一轮（命中多少算多少）。
+
+    模型偶尔整组漏给（实测 codex-cli 只填了内容标签前两组）；空着用户就得
+    手选，扫描命中的是官方表内词，补上好过留白——已有的组不动，不覆盖模型。"""
+    material = _material_text(task, outline)
+    if not material:
+        return meta
+    goal = task.get("goal") or ""
+    if platform == "fanqie":
+        reader = meta.get("target_reader") or ("女频" if "女" in goal else "男频")
+        cats, theme, role, plot = _fq_tables(reader)
+        c_plot, c_emo, c_char, c_world = _fq_content_tables()
+        pairs = [("tags_theme", theme, 2), ("tags_role", role, 2), ("tags_plot", plot, 2),
+                 ("content_plot", c_plot, 4), ("content_emotion", c_emo, 2),
+                 ("content_character", c_char, 4), ("content_world", c_world, 1)]
+    else:
+        from . import bookmeta_catalog as cat
+        pairs = [(k, cat.QIMAO_TAG_GROUPS[g], 3) for k, g in
+                 (("tags_style", "风格"), ("tags_role", "角色"),
+                  ("tags_plot", "情节"), ("tags_bg", "背景"))]
+    for key, table, limit in pairs:
+        if not meta.get(key):
+            meta[key] = _scan_table(material, table, limit)
+    return meta
+
+
+def _material_names(material):
+    """从圣经/大纲/章节里提取明确标注的人物名。
+
+    只接受带人物语义的短语（如“女主：阿禾”或“男主叫陆沉”），避免把
+    普通句子里的名词误当成主角。返回顺序去重后的最多两个名字。
+    """
+    text = str(material or "")
+    out = []
+    patterns = (
+        r"(?:主角|主人公|男主(?:角)?|女主(?:角)?|男一|女一|姓名|名字)"
+        r"\s*(?:是|叫|为|：|:)\s*([\u4e00-\u9fffA-Za-z][\u4e00-\u9fffA-Za-z·]{1,7})",
+        r"人物\s*[：:]\s*([^\n。；;]{2,40})",
+    )
+    for pat in patterns:
+        for m in re.finditer(pat, text):
+            raw = m.group(1)
+            chunks = re.split(r"[、,，/和及与]|\s+", raw)
+            for chunk in chunks:
+                name = re.sub(r"[^\u4e00-\u9fffA-Za-z·]", "", chunk)
+                if 2 <= len(name) <= 8 and name not in out:
+                    out.append(name[:5])
+                if len(out) >= 2:
+                    return out
+    return out
+
+
+def _pick_option(material, table, keywords=()):
+    """从官方选项表中按素材命中词选择一个，完全命中不到时用首项。"""
+    text = str(material or "")
+    for item in table:
+        if len(item) >= 2 and item in text:
+            return item
+    for key, item in keywords:
+        if key and key in text and item in table:
+            return item
+    return table[0] if table else ""
+
+
+def _has_value(value):
+    """判断建书字段是否真正有值（兼容列表和历史占位文本）。"""
+    placeholders = {"待补充", "（待补充）", "(待补充)"}
+    if isinstance(value, list):
+        return any((text := str(x or "").strip()) and text not in placeholders
+                   for x in value)
+    text = str(value or "").strip()
+    return bool(text) and text not in placeholders
+
+
+def _md_plain(text):
+    """markdown 轻洗（在 _summary 净化之上）：每行剥标题/列表/引用符与行内
+    强调符，合并成一段正文文字（建书表单简介是单行框）。"""
+    out = []
+    for ln in _summary(text).splitlines():
+        ln = re.sub(r"^\s*(#{1,6}\s*|[-*+]\s+|>\s*|\d+[.、)）]\s*)", "", ln)
+        ln = re.sub(r"[*_`#>]+", "", ln).strip()
+        if ln:
+            out.append(ln)
+    return "".join(out)
+
+
+def _summary_ok(value):
+    """简介够不够格进建书表单：番茄表单硬性要求 50-500 字（占位文案即
+    「请输入50-500字」），markdown 标题/占位符过不了平台提交校验，且自动
+    化流程看不到平台的字段报错，只会误报成「未登录或改版」。"""
+    return 50 <= len(_md_plain(value)) <= 500
+
+
+def _prose_summary(cur, material):
+    """简介兜底清理：markdown 洗成正文；洗完仍不足 50 字就从素材正文取。
+    取不出够长的正文时原样返回——由建书前置闸拦停并提示补写，不硬编。"""
+    text = _md_plain(cur)
+    if len(text) >= 50:
+        return text[:500]
+    body = _md_plain(material)
+    if len(body) >= 50:
+        return body[:500]
+    return str(cur or "").strip()
+
+
+def _fill_required_fields(task, platform, meta, outline, material):
+    """生成阶段的必填字段闸门。
+
+    ``_norm`` 仍保持“表外值清空”的纯校验语义，供旧数据修复和单测使用；
+    这里仅在真正交付作品信息前补齐平台建书表单的必填项，避免 UI 展示一整
+    屏“待补充”而无法直接复制建书。所有分类/标签仍来自官方目录。
+    """
+    from . import bookmeta_catalog as cat
+
+    material = str(material or "")
+    names = _material_names(material)
+    female_words = ("女频", "女生", "女主", "言情", "古言", "恋爱")
+    female = any(x in ((task.get("goal") or "") + material) for x in female_words)
+    defaults = ("沈知微", "顾砚川", "林知夏") if female else ("陆沉", "苏晚", "程野")
+    if not _has_value(meta.get("protagonist_1")):
+        meta["protagonist_1"] = names[0] if names else defaults[0]
+    if not _has_value(meta.get("protagonist_2")):
+        candidate = names[1] if len(names) > 1 else defaults[1]
+        if candidate == meta["protagonist_1"]:
+            candidate = next((name for name in defaults if name != meta["protagonist_1"]), defaults[2])
+        meta["protagonist_2"] = candidate
+
+    # 简介质量：平台建书表单对简介有硬字数校验（番茄 50-500 字），markdown
+    # 标题/占位文本过不了提交且流程看不到报错——洗成正文，不够长从素材取
+    meta["summary"] = _prose_summary(meta.get("summary"), material)
+
+    # 频道锚：题材信号/同书另一平台的读者与模型声称的频道冲突时以锚为准
+    # 翻转（2026-09-24 实案：男频军事文被配成 女生/古代言情/权谋天下，
+    # 级联自洽骗过目录校验）。翻转后旧频道分类作废，由下方素材补齐逻辑
+    # 按正确频道重选；纠偏说明记 _fixes 临时键，由调用方并进 source。
+    anchor, why = _reader_anchor(task, platform, meta)
+    if anchor and meta.get("target_reader") != anchor:
+        meta["_fixes"] = ["频道纠偏：%s→%s（依据：%s）"
+                          % (meta.get("target_reader") or "未填", anchor, why)]
+        meta["target_reader"] = anchor
+        if platform == "qimao":
+            ch = cat.QIMAO_CATS.get(anchor) or {}
+            if meta.get("category_main") not in ch:
+                meta["category_main"] = ""
+            ch_main = meta.get("category_main")
+            if meta.get("category_sub") and not (
+                    ch_main and meta["category_sub"] in (ch.get(ch_main) or [])):
+                meta["category_sub"] = ""
+        else:
+            fq_cats, _, _, _ = _fq_tables(anchor)
+            if meta.get("category") not in fq_cats:
+                meta["category"] = ""
+
+    if platform == "fanqie":
+        reader = meta.get("target_reader") or ("女频" if female else "男频")
+        if reader not in ("男频", "女频"):
+            reader = "女频" if female else "男频"
+        meta["target_reader"] = reader
+        cats, theme, role, plot = _fq_tables(reader)
+        c_plot, c_emo, c_char, c_world = _fq_content_tables()
+        if meta.get("category") not in cats:
+            # 关键词兜底按频道给（2026-09-24 实案《戍边骑奴》：旧表不分频道，
+            # 「戍边→古风世情」是女频词，男频素材永远选不中，全部落到男频表
+            # 首项「西方奇幻」——架空王朝军事文被配成西式魔幻）
+            kw = (("历史", "历史古代"), ("古代", "历史古代"),
+                  ("王朝", "历史脑洞"), ("架空", "历史脑洞"),
+                  ("戍边", "历史古代"), ("边关", "历史古代"),
+                  ("抗战", "抗战谍战"), ("谍战", "抗战谍战"),
+                  ("军", "历史古代"), ("种田", "都市种田"),
+                  ("都市", "都市脑洞"), ("悬疑", "悬疑脑洞")) if reader == "男频" else (
+                 ("宅斗", "宫斗宅斗"), ("宫斗", "宫斗宅斗"),
+                 ("总裁", "豪门总裁"), ("豪门", "豪门总裁"),
+                 ("古代", "古风世情"), ("古言", "古言脑洞"),
+                 ("年代", "年代"), ("种田", "种田"),
+                 ("婚姻", "职场婚恋"), ("恋爱", "职场婚恋"),
+                 ("悬疑", "女频悬疑"), ("快穿", "快穿"))
+            meta["category"] = _pick_option(material, cats, kw)
+        groups = (("tags_theme", theme, 2), ("tags_role", role, 2),
+                  ("tags_plot", plot, 2), ("content_plot", c_plot, 4),
+                  ("content_emotion", c_emo, 2), ("content_character", c_char, 4),
+                  ("content_world", c_world, 1))
+    else:
+        reader = meta.get("target_reader") or ("女生" if female else "男生")
+        if reader not in ("男生", "女生"):
+            reader = "女生" if female else "男生"
+        meta["target_reader"] = reader
+        channel = cat.QIMAO_CATS.get(reader) or cat.QIMAO_CATS["男生"]
+        main = meta.get("category_main")
+        if main not in channel:
+            # 关键词兜底按频道给（2026-09-24 频道锚落地后读方可信，旧的
+            # 「戍边→古代言情」在男生频道下永远选不中白占位）
+            kw = (("戍边", "历史"), ("边关", "历史"), ("军", "军事"),
+                  ("都市", "都市"), ("玄幻", "玄幻奇幻"), ("悬疑", "奇闻异事")
+                  ) if reader == "男生" else (
+                  ("宅斗", "古代言情"), ("宫斗", "古代言情"),
+                  ("古代", "古代言情"), ("总裁", "现代言情"),
+                  ("年代", "现代言情"), ("都市", "现代言情"))
+            main = _pick_option(material, list(channel), kw)
+            meta["category_main"] = main
+        subs = channel.get(main) or []
+        sub = meta.get("category_sub")
+        if sub not in subs:
+            located = cat.qimao_locate_sub(sub)
+            if located:
+                reader, main = located
+                meta["target_reader"] = reader
+                meta["category_main"] = main
+                subs = cat.QIMAO_CATS[reader][main]
+            else:
+                sub = ""
+        if sub not in subs:
+            meta["category_sub"] = _pick_option(material, subs)
+        groups = (("tags_style", cat.QIMAO_TAG_GROUPS["风格"], 3),
+                  ("tags_role", cat.QIMAO_TAG_GROUPS["角色"], 3),
+                  ("tags_plot", cat.QIMAO_TAG_GROUPS["情节"], 3),
+                  ("tags_bg", cat.QIMAO_TAG_GROUPS["背景"], 3))
+    for key, table, limit in groups:
+        values = meta.get(key)
+        if not isinstance(values, list):
+            values = []
+        values = [str(v).strip() for v in values if str(v or "").strip() in table]
+        if not values:
+            values = [_pick_option(material, table)]
+        meta[key] = values[:limit]
+    return meta
+
+
+def store_read_bible(task):
+    """圣经读取的小包装（模板兜底用；异常向上抛由调用方吞掉）。"""
+    from . import store
+    fp, text, err = store.read_story_bible(task.get("id") or "")
+    if err:
+        raise ValueError(err)
+    return fp, text, None
+
+
+def _resolve_author(task):
+    """作者 CLI 兜底用：按「实现」角色选一个 real 智能体（同连载起草的路由口径）。"""
+    try:
+        from . import catalog, manager, registry, router
+        agents = registry.effective_agents(catalog.load(), manager.detect_all())
+        agent, _ = router.pick([a for a in agents if a.get("mode") == "real"],
+                               "implement", task.get("type") or "serial_novel")
+        return agent
+    except Exception:
+        return None
+
+
+def make_book_meta(task, platform, author_agent=None, log_path=None):
+    """生成一个平台的作品信息。返回 (meta dict, source 字符串)；模板兜底永远有返回。"""
+    from . import knowledge, modelhub, planner, runner, skills  # 惰性导入，同 planner
+    material, outline = _collect_material(task)
+    # 素材充分性门：没有大纲也没有章稿/圣经的任务（如「帮我定方向」的选题
+    # 研究任务）没有可推介的小说要素——生成只会产出报告腔垃圾
+    # （2026-09-23 实案：简介=「# 网文选题分析报告…」）。宁明确拒绝。
+    if not outline and "## 第一章开头" not in material and "## 故事圣经" not in material:
+        raise ValueError(
+            "该任务还没有小说大纲、正文或故事圣经，无法生成作品信息——"
+            "先完成大纲/首批章节后再来一键生成")
+    schema = {"fanqie": FANQIE_SCHEMA, "qimao": QIMAO_SCHEMA}.get(platform) or ""
+    plat_label = PLATFORMS[platform]["label"]
+    sk_block, _ = skills.block_for(task)
+    # 实案教训（2026-09-23）：素材偏研究向时模型把简介写成「分析报告标题」
+    summary_rule = ("- summary 必须是面向读者的小说推介文案：点出题材、主角、核心冲突与悬念，"
+                    "200/500 字内成段；绝不允许写成报告标题、要点清单或任务说明。")
+    kb_block = knowledge.block_for(task)
+    if kb_block:
+        sk_block = (sk_block + "\n\n" + kb_block) if sk_block else kb_block
+    goal = task.get("goal") or ""
+    # 真实选项注入：分类/标签表按平台与读者频段给全（模型只能从表里选）
+    options = (_fq_options_block(goal) if platform == "fanqie" else _qm_options_block(goal))
+    prompt = (PROMPT_HEAD.replace("__PLATFORM__", plat_label)
+              .replace("__SKILLS__", sk_block)
+              .replace("__SCHEMA__", schema)
+              .replace("__OPTIONS_RULE__", options + "\n" + summary_rule)
+              + "\n\n## 小说素材\n" + material)
+
+    orch = None
+    if not template_only():
+        try:
+            orch = modelhub.resolve_orchestrator()
+        except Exception:
+            orch = None
+    orch_err = ""
+    if orch:
+        prov, model = orch
+        res = modelhub.chat(prov["id"], model, prompt, max_tokens=4000, timeout=300)
+        planner._log_usage("bookmeta", "bookmeta", task, res, model=model,
+                           provider=prov.get("name", prov.get("id", "")),
+                           provider_id=prov.get("id", ""))
+        if res["ok"]:
+            meta = _norm(runner.extract_json(res.get("text") or ""), platform, goal)
+            if meta:
+                meta = _fill_tag_gaps(task, platform, meta, outline)
+                meta = _fill_required_fields(task, platform, meta, outline, material)
+                return meta, _apply_fix_note(meta, "编排者(%s · %s)"
+                                             % (prov.get("name", prov["id"]), model))
+        orch_err = str(res.get("error") or "返回内容无法解析为作品信息")[:200]
+
+    if author_agent and author_agent.get("mode") == "real" and not template_only():
+        res = runner.run_agent(modelhub.bind_agent(author_agent), prompt,
+                               workdir=task.get("workdir"), readonly=True,
+                               timeout=BOOKMETA_TIMEOUT, log_path=log_path)
+        planner._log_usage("bookmeta", "bookmeta", task, res, agent=author_agent)
+        if res["ok"]:
+            meta = _norm(runner.extract_json(res.get("text") or ""), platform, goal)
+            if meta:
+                meta = _fill_tag_gaps(task, platform, meta, outline)
+                meta = _fill_required_fields(task, platform, meta, outline, material)
+                return meta, _apply_fix_note(meta, "llm(%s)" % author_agent["id"])
+        orch_err = orch_err or str(res.get("error") or "")[:200]
+
+    meta = _template_meta(task, platform, outline)
+    meta = _fill_required_fields(task, platform, meta, outline, material)
+    meta["source"] = ("模板兜底（编排者不可用：%s）" % orch_err) if orch_err else "模板兜底"
+    return meta, _apply_fix_note(meta, meta["source"])
+
+
+# ---- 换名（2026-10-02 用户实案：生成的书名与本站已有作品重名，用户要的是
+# 「换一个名字」而不是整套重摇——重摇会覆盖已核对过的分类/标签/简介）
+
+RENAME_TIMEOUT = 120
+# 排除名单注入上限：全站书名可能上百个，超出的按序丢弃（提示词预算保护；
+# 排序后字母序前 80 个 + 调用方传入的近期撞名足够防撞，撞了还有重试兜底）
+RENAME_EXCLUDE_LIMIT = 80
+
+RENAME_PROMPT = """你是网文编辑。这本书《__OLD__》的书名与已有作品重名了，请另起一个新书名。
+
+书的基本情况：
+__MATERIAL__
+
+硬性要求：
+- 只输出一个 ```json 代码块：{"book_name": "新书名"}，不要输出其他内容；
+- 新书名不超过 15 个字，贴合题材、有网文点击欲，与旧书名同一题材但用词明显不同；
+- 严禁与下面任何一本书重名，也不得只加「2」「新」「之」等后缀敷衍：
+__EXCLUDE__"""
+
+
+def used_book_names():
+    """本站全部已用书名（换名排除名单）：已建书登记（books.json 的 title）
+    + 所有任务 book_meta 各平台 done 态的 book_name。读取异常只降级不阻断
+    （名单不全最坏结果是换出来的名字仍撞，用户再点一次）。"""
+    from .publish import ledger
+    names = set()
+    try:
+        # books.json 结构：{task_id: {platform: {book_id, title, url, ...}}}
+        for by_plat in (ledger.load_books() or {}).values():
+            if not isinstance(by_plat, dict):
+                continue
+            for rec in by_plat.values():
+                t = str((rec or {}).get("title") or "").strip()
+                if t:
+                    names.add(t)
+    except Exception:
+        pass
+    try:
+        from . import store
+        for task in store.list_tasks(limit=10 ** 9):
+            for entry in (task.get("book_meta") or {}).values():
+                if isinstance(entry, dict) and entry.get("status") == "done":
+                    n = str((entry.get("data") or {}).get("book_name") or "").strip()
+                    if n:
+                        names.add(n)
+    except Exception:
+        pass
+    return names
+
+
+def rename_book(task, platform, old_name, exclude_names=()):
+    """给已生成的作品信息换一个不重名的书名。返回 (new_name, "") 或 ("", 错误)。
+
+    只产出新名，不写盘——调用方负责更新 book_meta 与归档同步；这样模型失败
+    时现有数据原样保留（区别于整套生成的线程+状态机，轻调用同步返回即可）。
+    新名若仍撞排除名单，把撞名项并入名单重试一次。"""
+    from . import modelhub, planner, runner
+    old_name = str(old_name or "").strip()
+    exclude = {str(x).strip() for x in (exclude_names or ()) if str(x).strip()}
+    exclude.add(old_name)
+    goal_lines = [x.strip() for x in (task.get("goal") or "").splitlines() if x.strip()]
+    meta_lines = []
+    if goal_lines:
+        meta_lines.append("目标：" + goal_lines[0][:200])
+    try:
+        _, bible, _ = store_read_bible(task)
+        if bible:
+            meta_lines.append("故事圣经摘录：" + bible[:400])
+    except Exception:
+        pass
+    material = "\n".join(meta_lines) if meta_lines else "（素材缺失，按旧书名题材另起）"
+    orch = None
+    try:
+        orch = modelhub.resolve_orchestrator()
+    except Exception:
+        orch = None
+    if not orch:
+        return "", "编排者模型不可用，无法自动换名（可稍后重试）"
+    prov, model = orch
+    last = ""
+    for _ in range(2):                           # 首轮撞名单 → 带撞名项重试一次
+        ex_block = "\n".join("- " + n for n in sorted(exclude)[:RENAME_EXCLUDE_LIMIT])
+        prompt = (RENAME_PROMPT.replace("__OLD__", old_name)
+                  .replace("__MATERIAL__", material)
+                  .replace("__EXCLUDE__", ex_block or "（暂无）"))
+        # max_tokens 4000 对齐主生成（2026-10-03 实案：glm-5.3 思考计入
+        # max_tokens，给 400 三次全被思考烧穿 output 打满，JSON 没成形）
+        res = modelhub.chat(prov["id"], model, prompt, max_tokens=4000,
+                            timeout=RENAME_TIMEOUT)
+        planner._log_usage("bookmeta", "bookmeta-rename", task, res, model=model,
+                           provider=prov.get("name", prov.get("id", "")),
+                           provider_id=prov.get("id", ""))
+        if not res["ok"]:
+            last = str(res.get("error") or "模型调用失败")[:200]
+            continue
+        text = str(res.get("text") or "")
+        data = runner.extract_json(text)
+        new_name = _name(data.get("book_name") if isinstance(data, dict) else "", 15)
+        if not new_name:
+            # 兜底：思考烧预算时 JSON 可能没成形，正文里有《书名》/「书名」
+            # 形态就抠出来用——抠出的名字仍要过下面的排除名单检查
+            m = re.search(r"[《「]([^《》「」]{1,15})[》」]", text)
+            if m:
+                new_name = _name(m.group(1), 15)
+        if not new_name:
+            last = "返回内容无法解析出新书名"
+            continue
+        if new_name in exclude:
+            last = "新书名《%s》仍与本站已有作品重名" % new_name
+            exclude.add(new_name)
+            continue
+        return new_name, ""
+    return "", last or "换名失败，请稍后重试"
+
+
+def render_markdown(task, platform, meta):
+    """落工作目录的归档 Markdown（与前端弹框同一份字段顺序）。"""
+    plat = PLATFORMS[platform]["label"]
+    lines = ["# 作品信息（%s）" % plat, "",
+             "> 任务：%s　生成于 %s　来源：%s" % (
+                 task.get("title") or task.get("id") or "",
+                 time.strftime("%Y-%m-%d %H:%M"), meta.get("source") or "")]
+    lines.append("")
+    for key, label in FIELD_LABELS[platform]:
+        v = meta.get(key)
+        if isinstance(v, list):
+            v = "、".join(v) if v else "（待补充）"
+        elif key == "summary":
+            v = (v or "").strip() or "（待补充）"
+        else:
+            v = str(v or "").strip() or "（待补充）"
+        lines.append("- **%s**：%s" % (label, v))
+    lines.append("")
+    lines.append("> 各字段可直接复制进 %s 建书表单；分类与标签均选自平台真实选项表。" % plat)
+    return "\n".join(lines)
+
+
+def generate_async(task_id, platform, author_agent=None):
+    """后台线程入口：跑生成并把终态写回任务 book_meta；成功时落工作目录归档。
+
+    状态机：running（路由置入）→ done / failed。任何异常都落 failed，不留悬挂。"""
+    from . import store
+    task = store.get_task(task_id)
+    if not task:
+        return
+    try:
+        meta, source = make_book_meta(task, platform, author_agent=author_agent)
+        entry = {"status": "done", "data": meta, "source": source,
+                 "at": time.strftime("%Y-%m-%d %H:%M:%S")}
+        fp = Path(task.get("workdir") or "") / PLATFORMS[platform]["file"]
+        if str(fp.parent) and fp.parent.is_dir():
+            fp.write_text(render_markdown(task, platform, meta), encoding="utf-8")
+            entry["file"] = PLATFORMS[platform]["file"]
+    except Exception as e:                       # 兜底线程不能静默死掉
+        entry = {"status": "failed", "error": str(e)[:300],
+                 "at": time.strftime("%Y-%m-%d %H:%M:%S")}
+    cur = store.get_task(task_id)
+    if not cur:
+        return
+    prev = (cur.get("book_meta") or {}).get(platform) or {}
+    if prev.get("status") != "running":          # 期间被删/被重置：丢弃结果
+        return
+    store.set_book_meta(task_id, platform, entry)
+
+
+def recover_orphans():
+    """启动收尸：后台生成线程随进程重启死掉后，落盘的 running 无人写终态——
+    前端永远「生成中」（按钮禁用）+ 接口幂等拒绝，卡死不可自愈。启动时把
+    遗留 running 统一改判 failed，交还「重试」按钮。返回改判条数。"""
+    from . import store
+    n = 0
+    for task in store.list_tasks(limit=10 ** 9):
+        for plat, entry in (task.get("book_meta") or {}).items():
+            if isinstance(entry, dict) and entry.get("status") == "running":
+                store.set_book_meta(task["id"], plat, {
+                    "status": "failed",
+                    "error": "生成随服务重启中断，请点重试",
+                    "at": time.strftime("%Y-%m-%d %H:%M:%S")})
+                n += 1
+    return n
+
+
+def repair_existing():
+    """补齐历史已生成记录中的必填字段。
+
+    旧版本允许分类/主角为空并把结果标成 done；只要字段缺失就按当前官方
+    目录和已有素材补一次，已有非空字段保持不变。返回修复条数，供启动日志
+    和升级诊断使用。
+    """
+    from . import store
+    repaired = 0
+    for task in store.list_tasks(limit=10 ** 9):
+        book_meta = task.get("book_meta") or {}
+        for platform in PLATFORMS:
+            entry = book_meta.get(platform) or {}
+            if entry.get("status") != "done" or not isinstance(entry.get("data"), dict):
+                continue
+            data = dict(entry["data"])
+            if platform == "fanqie":
+                reader = data.get("target_reader")
+                cats, _, _, _ = _fq_tables(reader if reader in ("男频", "女频") else "男频")
+                required_ok = (
+                    data.get("category") in cats
+                    and all(_has_value(data.get(key)) for key in (
+                        "tags_theme", "tags_role", "tags_plot", "content_plot",
+                        "content_emotion", "content_character", "content_world"))
+                    and _has_value(data.get("protagonist_1"))
+                    and _has_value(data.get("protagonist_2"))
+                    and _summary_ok(data.get("summary")))
+            else:
+                from . import bookmeta_catalog as cat
+                reader = data.get("target_reader")
+                channel = cat.QIMAO_CATS.get(reader) or {}
+                main = data.get("category_main")
+                required_ok = (
+                    main in channel
+                    and data.get("category_sub") in (channel.get(main) or [])
+                    and all(_has_value(data.get(key)) for key in (
+                        "tags_style", "tags_role", "tags_plot", "tags_bg"))
+                    and _has_value(data.get("protagonist_1"))
+                    and _has_value(data.get("protagonist_2"))
+                    and _summary_ok(data.get("summary")))
+            # 频道失配也视为待修复（2026-09-24 实案：级联自洽的跨频道组合
+            # 字段全齐，required_ok 挡不住，只能靠题材信号识别）
+            anchor, _ = _reader_anchor(task, platform,
+                                       {"summary": data.get("summary") or ""})
+            if anchor and anchor != data.get("target_reader"):
+                required_ok = False
+            if required_ok:
+                continue
+            try:
+                material, outline = _collect_material(task)
+                fixed = _fill_required_fields(task, platform, data, outline, material)
+                fixes = fixed.pop("_fixes", None)   # 先清临时键再比幂等
+                if fixed == entry["data"]:
+                    continue
+                payload = dict(entry)
+                payload["data"] = fixed
+                note = "已补齐必填字段"
+                if fixes:
+                    note += "；" + "；".join(str(x) for x in fixes)
+                payload["source"] = (str(entry.get("source") or "") + "；" + note).lstrip("；")
+                store.set_book_meta(task["id"], platform, payload)
+                try:
+                    fp = Path(task.get("workdir") or "") / PLATFORMS[platform]["file"]
+                    if fp.parent.is_dir():
+                        fp.write_text(render_markdown(task, platform, fixed), encoding="utf-8")
+                except OSError:
+                    pass
+                repaired += 1
+            except Exception:
+                continue
+    return repaired