admin/MuMuAINovel
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

feature: 新增支持拆书自选保留章节数，或者整本拆书功能

Browse Source
This commit is contained in:
xiamuceer
2026-04-02 11:03:58 +08:00
parent 099d12a1c9
commit 5968a3d29e
6 changed files with 158 additions and 23 deletions
Download Patch File Download Diff File Expand all files Collapse all files
backend/app/api/book_import.py
+20
View File
@@ -15,6 +15,7 @@ from app.schemas.book_import import (
BookImportPreviewResponse,
BookImportRetryRequest,
BookImportTaskCreateResponse,
BookImportTaskCreateRequest,
BookImportTaskStatusResponse,
)
from app.services.book_import_service import book_import_service
@@ -33,6 +34,8 @@ async def create_book_import_task(
project_id: str | None = Form(default=None, description="兼容参数：当前版本固定新建项目，不支持传入"),
create_new_project: bool = Form(default=True, description="兼容参数：当前版本仅支持 true"),
import_mode: str = Form(default="append", description="导入模式：append/overwrite"),
extract_mode: str = Form(default="tail", description="解析范围：tail=截取末章，full=整本"),
tail_chapter_count: int = Form(default=10, description="当 extract_mode=tail 时，截取末尾章节数，需为5的倍数；超过50按整本拆处理"),
):
user_id = getattr(request.state, "user_id", None)
if not user_id:
@@ -44,11 +47,26 @@ async def create_book_import_task(
if import_mode not in {"append", "overwrite"}:
raise HTTPException(status_code=400, detail="import_mode 仅支持 append 或 overwrite")
if extract_mode not in {"tail", "full"}:
raise HTTPException(status_code=400, detail="extract_mode 仅支持 tail 或 full")
if tail_chapter_count < 5:
raise HTTPException(status_code=400, detail="tail_chapter_count 不能小于 5")
if tail_chapter_count % 5 != 0:
raise HTTPException(status_code=400, detail="tail_chapter_count 必须是 5 的倍数")
if tail_chapter_count > 50:
extract_mode = "full"
if project_id:
raise HTTPException(status_code=400, detail="当前仅支持新建项目导入，不支持指定 project_id")
if not create_new_project:
raise HTTPException(status_code=400, detail="当前仅支持新建项目导入")
create_payload = BookImportTaskCreateRequest(
extract_mode=extract_mode,
tail_chapter_count=tail_chapter_count,
)
content = await file.read()
if len(content) > MAX_TXT_SIZE:
raise HTTPException(status_code=413, detail="文件大小超过 50MB 限制")
@@ -60,6 +78,8 @@ async def create_book_import_task(
project_id=None,
create_new_project=True,
import_mode=import_mode,
extract_mode=create_payload.extract_mode,
tail_chapter_count=create_payload.tail_chapter_count,
)
return task
backend/app/schemas/book_import.py
+8 -1
View File
@@ -9,6 +9,7 @@ TaskStatus = Literal["pending", "running", "completed", "failed", "cancelled"]
ImportMode = Literal["append", "overwrite"]
ExtractLevel = Literal["basic", "standard", "deep"]
WarningLevel = Literal["info", "warning", "error"]
BookImportExtractMode = Literal["tail", "full"]
class BookImportWarning(BaseModel):
@@ -45,6 +46,12 @@ class BookImportOutline(BaseModel):
structure: Optional[dict[str, Any]] = Field(None, description="结构化大纲（与系统大纲生成结构一致）")
class BookImportTaskCreateRequest(BaseModel):
"""创建拆书任务请求"""
extract_mode: BookImportExtractMode = Field(default="tail", description="提取范围：tail=截取末章，full=整本")
tail_chapter_count: int = Field(default=10, ge=5, le=9999, description="当 extract_mode=tail 时，截取末尾章节数；需为5的倍数，超过50将按整本处理")
class BookImportTaskCreateResponse(BaseModel):
"""创建任务响应"""
task_id: str
@@ -89,4 +96,4 @@ class BookImportApplyResponse(BaseModel):
class BookImportRetryRequest(BaseModel):
"""重试失败步骤请求"""
steps: list[str] = Field(..., min_length=1, description="需要重试的步骤名列表，如 world_building / career_system / characters")
steps: list[str] = Field(..., min_length=1, description="需要重试的步骤名列表，如 world_building / career_system / characters")
backend/app/services/book_import_service.py
+75 -18
View File
@@ -34,6 +34,7 @@ from app.schemas.book_import import (
BookImportApplyRequest,
BookImportApplyResponse,
BookImportChapter,
BookImportExtractMode,
BookImportOutline,
BookImportPreviewResponse,
BookImportTaskCreateResponse,
@@ -65,6 +66,8 @@ class _BookImportTask:
project_id: Optional[str]
create_new_project: bool
import_mode: str
extract_mode: BookImportExtractMode = "tail"
tail_chapter_count: int = 10
status: str = "pending"
progress: int = 0
message: Optional[str] = "任务已创建"
@@ -95,7 +98,16 @@ class BookImportService:
project_id: Optional[str],
create_new_project: bool,
import_mode: str,
extract_mode: BookImportExtractMode = "tail",
tail_chapter_count: int = 10,
) -> BookImportTaskCreateResponse:
normalized_tail_count = max(5, int(tail_chapter_count))
normalized_extract_mode = extract_mode
if normalized_tail_count % 5 != 0:
normalized_tail_count = ((normalized_tail_count + 4) // 5) * 5
if normalized_tail_count > 50:
normalized_extract_mode = "full"
task_id = str(uuid.uuid4())
task = _BookImportTask(
task_id=task_id,
@@ -104,6 +116,8 @@ class BookImportService:
project_id=project_id,
create_new_project=create_new_project,
import_mode=import_mode,
extract_mode=normalized_extract_mode,
tail_chapter_count=normalized_tail_count,
)
async with self._tasks_lock:
self._tasks[task_id] = task
@@ -150,15 +164,17 @@ class BookImportService:
}
warnings = list(task.preview.warnings) if task.preview else []
chapters_to_import, outlines_to_import, was_trimmed = self._trim_last_ten_for_apply(
chapters_to_import, outlines_to_import, was_trimmed = self._select_chapters_for_import(
chapters=payload.chapters,
outlines=payload.outlines,
extract_mode=task.extract_mode,
tail_chapter_count=task.tail_chapter_count,
)
if was_trimmed:
warnings.append(
BookImportWarning(
code="apply_trimmed_to_last_ten",
message=f"导入阶段已强制仅保留最后 {len(chapters_to_import)} 章",
code="apply_trimmed_for_extract_mode",
message=f"导入阶段已按解析配置仅保留 {len(chapters_to_import)} 章",
level="info",
)
)
@@ -248,15 +264,17 @@ class BookImportService:
}
warnings = list(task.preview.warnings) if task.preview else []
chapters_to_import, outlines_to_import, was_trimmed = self._trim_last_ten_for_apply(
chapters_to_import, outlines_to_import, was_trimmed = self._select_chapters_for_import(
chapters=payload.chapters,
outlines=payload.outlines,
extract_mode=task.extract_mode,
tail_chapter_count=task.tail_chapter_count,
)
if was_trimmed:
warnings.append(
BookImportWarning(
code="apply_trimmed_to_last_ten",
message=f"导入阶段已强制仅保留最后 {len(chapters_to_import)} 章",
code="apply_trimmed_for_extract_mode",
message=f"导入阶段已按解析配置仅保留 {len(chapters_to_import)} 章",
level="info",
)
)
@@ -580,7 +598,7 @@ class BookImportService:
return
try:
# 进度分配：编码识别 5%，文本清洗 10%，章节切分 15%，截取末10章 18%，AI反向生成 20%-95%，完成 100%
# 进度分配：编码识别 5%，文本清洗 10%，章节切分 15%，按配置筛选章节 18%，AI反向生成 20%-95%，完成 100%
self._set_task_state(task, status="running", progress=5, message="正在识别编码并读取文本...")
self._check_cancelled(task)
@@ -600,7 +618,7 @@ class BookImportService:
)
self._check_cancelled(task)
self._set_task_state(task, status="running", progress=18, message="仅保留末10章并重建预览结构...")
self._set_task_state(task, status="running", progress=18, message="正在按解析配置筛选章节并构建预览...")
preview = await self._build_preview(
task=task,
filename=task.filename,
@@ -798,18 +816,26 @@ class BookImportService:
return count, total_words
def _trim_last_ten_for_apply(
def _select_chapters_for_import(
self,
*,
chapters: list[BookImportChapter],
outlines: list[BookImportOutline],
extract_mode: BookImportExtractMode,
tail_chapter_count: int,
) -> tuple[list[BookImportChapter], list[BookImportOutline], bool]:
if not chapters:
return [], [], False
sorted_chapters = sorted(chapters, key=lambda x: x.chapter_number)
selected = sorted_chapters[-10:]
was_trimmed = len(sorted_chapters) > len(selected) or len(outlines) > 10
normalized_tail_count = max(5, int(tail_chapter_count))
if normalized_tail_count > 50 or extract_mode == "full":
selected = sorted_chapters
else:
normalized_tail_count = min(normalized_tail_count, len(sorted_chapters))
selected = sorted_chapters[-normalized_tail_count:]
was_trimmed = len(sorted_chapters) > len(selected)
normalized_chapters: list[BookImportChapter] = []
for idx, item in enumerate(selected, start=1):
@@ -826,7 +852,10 @@ class BookImportService:
normalized_outlines: list[BookImportOutline] = []
sorted_outlines = sorted(outlines, key=lambda x: x.order_index) if outlines else []
if sorted_outlines:
selected_outlines = sorted_outlines[-len(normalized_chapters):]
if extract_mode == "full":
selected_outlines = sorted_outlines[:len(normalized_chapters)]
else:
selected_outlines = sorted_outlines[-len(normalized_chapters):]
for idx, item in enumerate(selected_outlines, start=1):
normalized_outlines.append(
BookImportOutline(
@@ -853,6 +882,30 @@ class BookImportService:
return normalized_chapters, normalized_outlines, was_trimmed
def _select_raw_chapters_for_preview(
self,
*,
chapters_data: list[dict],
extract_mode: BookImportExtractMode,
tail_chapter_count: int,
) -> tuple[list[dict], bool]:
if not chapters_data:
return [], False
normalized_tail_count = max(5, int(tail_chapter_count))
if normalized_tail_count > 50 or extract_mode == "full":
return chapters_data, False
normalized_tail_count = min(normalized_tail_count, len(chapters_data))
selected = chapters_data[-normalized_tail_count:]
return selected, len(selected) < len(chapters_data)
def _get_extract_mode_label(self, extract_mode: BookImportExtractMode, selected_total: int) -> str:
if extract_mode == "full" or selected_total > 50:
return "整本"
return f"末{selected_total}章"
def _derive_world_settings(
self,
*,
@@ -974,9 +1027,13 @@ class BookImportService:
chapters: list[BookImportChapter] = []
warnings: list[BookImportWarning] = []
# 仅保留最后10章用于最终导入，重建章节序号为 1..N
selected_chapters_raw = chapters_data[-10:] if len(chapters_data) > 10 else chapters_data
selected_chapters_raw, was_trimmed = self._select_raw_chapters_for_preview(
chapters_data=chapters_data,
extract_mode=task.extract_mode,
tail_chapter_count=task.tail_chapter_count,
)
selected_total = len(selected_chapters_raw)
selection_label = self._get_extract_mode_label(task.extract_mode, selected_total)
title_counter: Counter[str] = Counter()
for idx, chapter in enumerate(selected_chapters_raw, start=1):
@@ -1020,7 +1077,7 @@ class BookImportService:
task,
status="running",
progress=chapter_progress,
message=f"已处理末章 {idx}/{selected_total} 个章节结构...",
message=f"已处理{selection_label} {idx}/{selected_total} 个章节结构...",
)
for title, count in title_counter.items():
@@ -1033,11 +1090,11 @@ class BookImportService:
)
)
if len(chapters_data) > selected_total:
if was_trimmed:
warnings.append(
BookImportWarning(
code="trimmed_to_last_ten_chapters",
message=f"已按规则仅保留最后 {selected_total} 章用于导入（原始识别 {len(chapters_data)} 章）",
code="trimmed_for_extract_mode",
message=f"已按解析配置仅保留{selection_label} {selected_total} 章用于导入（原始识别 {len(chapters_data)} 章）",
level="info",
)
)