From b80161972a4d950e196d6ef688046a90e35c62e5 Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?=E5=B0=8F=E7=BB=AB?= <486494914@qq.com> Date: Tue, 1 Sep 2026 20:51:46 +0800 Subject: [PATCH] 6 --- ...6818_add_category_tree_and_text_content.py | 47 + backend/app/api/categories.py | 219 +++- backend/app/api/documents.py | 30 + backend/app/core/errors.py | 14 +- backend/app/models/document.py | 30 +- backend/app/models/document_category.py | 32 +- backend/app/public/routes.py | 137 +- backend/app/repositories/doc_repo.py | 5 +- backend/app/services/doc_service.py | 86 ++ backend/app/services/kb_public_service.py | 96 +- backend/app/services/kb_service.py | 41 +- backend/tests/test_categories.py | 16 +- frontend/src/api/client.js | 11 +- frontend/src/api/client.ts | 10 +- frontend/src/views/KbDetail.vue | 342 ++++- frontend/src/views/KbDetail.vue.js | 1116 +++++++++++++---- frontend/src/views/Register.vue | 3 +- frontend/src/views/Register.vue.js | 2 +- 18 files changed, 1878 insertions(+), 359 deletions(-) create mode 100644 backend/alembic/versions/4bd4c7f26818_add_category_tree_and_text_content.py diff --git a/backend/alembic/versions/4bd4c7f26818_add_category_tree_and_text_content.py b/backend/alembic/versions/4bd4c7f26818_add_category_tree_and_text_content.py new file mode 100644 index 0000000..ab1d355 --- /dev/null +++ b/backend/alembic/versions/4bd4c7f26818_add_category_tree_and_text_content.py @@ -0,0 +1,47 @@ +"""add_category_tree_and_text_content + +Revision ID: 4bd4c7f26818 +Revises: 4fcdb390ae18 +Create Date: 2026-09-01 18:02:12.469005 +""" +from typing import Sequence, Union + +from alembic import op +import sqlalchemy as sa + + +# revision identifiers, used by Alembic. +revision: str = '4bd4c7f26818' +down_revision: Union[str, None] = '4fcdb390ae18' +branch_labels: Union[str, Sequence[str], None] = None +depends_on: Union[str, Sequence[str], None] = None + + +def upgrade() -> None: + # document_categories 新增字段 + op.add_column('document_categories', sa.Column('parent_id', sa.String(length=32), nullable=True, comment='父分类 ID（NULL = 顶层）')) + op.add_column('document_categories', sa.Column('path', sa.Text(), server_default='/', nullable=False, comment='物化路径')) + op.add_column('document_categories', sa.Column('is_folder', sa.Boolean(), server_default=sa.text('1'), nullable=False, comment='True=文件夹')) + op.create_index(op.f('ix_document_categories_parent_id'), 'document_categories', ['parent_id'], unique=False) + + # documents 新增字段 + op.add_column('documents', sa.Column('content', sa.Text(), nullable=True, comment='直接输入的文本内容')) + op.add_column('documents', sa.Column('content_format', sa.String(length=16), server_default='markdown', nullable=False, comment='内容格式')) + + # SQLite 不支持 ALTER COLUMN，用 batch mode 重建表 + with op.batch_alter_table('documents', schema=None) as batch_op: + batch_op.alter_column('storage_path', existing_type=sa.String(1024), nullable=True) + batch_op.alter_column('sha256', existing_type=sa.String(64), nullable=True) + + +def downgrade() -> None: + with op.batch_alter_table('documents', schema=None) as batch_op: + batch_op.alter_column('sha256', existing_type=sa.String(64), nullable=False) + batch_op.alter_column('storage_path', existing_type=sa.String(1024), nullable=False) + + op.drop_column('documents', 'content_format') + op.drop_column('documents', 'content') + op.drop_index(op.f('ix_document_categories_parent_id'), table_name='document_categories') + op.drop_column('document_categories', 'is_folder') + op.drop_column('document_categories', 'path') + op.drop_column('document_categories', 'parent_id') \ No newline at end of file diff --git a/backend/app/api/categories.py b/backend/app/api/categories.py index 200cc79..c43e9e2 100644 --- a/backend/app/api/categories.py +++ b/backend/app/api/categories.py @@ -1,7 +1,8 @@ -"""文档分类 API 路由。""" +"""文档分类 API 路由（树形目录结构）。""" from fastapi import APIRouter, Depends from pydantic import BaseModel, Field +from sqlalchemy import select from sqlalchemy.orm import Session from app.api.deps import get_current_user, get_db @@ -13,73 +14,219 @@ from app.repositories.kb_repo import KnowledgeBaseRepository router = APIRouter(prefix="/knowledge-bases/{kb_id}/categories", tags=["categories"]) -class CategoryRequest(BaseModel): +class CategoryCreateRequest(BaseModel): name: str = Field(min_length=1, max_length=255) + parent_id: str | None = None # NULL = 顶层 sort_order: int = 0 + is_folder: bool = True -class CategoryResponse(BaseModel): +class CategoryUpdateRequest(BaseModel): + name: str | None = Field(default=None, min_length=1, max_length=255) + parent_id: str | None = None # 移动到新父节点 + sort_order: int | None = None + is_folder: bool | None = None + + +class CategoryTreeNode(BaseModel): id: str name: str + path: str + is_folder: bool sort_order: int + doc_count: int = 0 + children: list["CategoryTreeNode"] = [] model_config = {"from_attributes": True} -@router.get("", response_model=list[CategoryResponse]) -def list_categories( +CategoryTreeNode.model_rebuild() + + +@router.get("/tree", response_model=list[CategoryTreeNode]) +def get_category_tree( kb_id: str, user: User = Depends(get_current_user), db: Session = Depends(get_db), -) -> list[CategoryResponse]: +) -> list[CategoryTreeNode]: + """获取完整目录树（含文档数量）。""" + _check_kb_owner(kb_id, user, db) + + # 获取所有分类 + stmt = ( + select(DocumentCategory) + .where(DocumentCategory.knowledge_base_id == kb_id) + .order_by(DocumentCategory.sort_order, DocumentCategory.name) + ) + all_cats = list(db.scalars(stmt).all()) + + # 获取每个分类的文档数量 + from app.models.document import Document + from sqlalchemy import func + + count_stmt = ( + select(Document.category_id, func.count()) + .where( + Document.knowledge_base_id == kb_id, + Document.status != "DELETED", + ) + .group_by(Document.category_id) + ) + doc_count_map = dict(db.execute(count_stmt).all()) + + # 构建树 + cat_map = {} + for cat in all_cats: + cat_map[cat.id] = CategoryTreeNode( + id=cat.id, + name=cat.name, + path=cat.path or "/", + is_folder=cat.is_folder if hasattr(cat, 'is_folder') else True, + sort_order=cat.sort_order, + doc_count=doc_count_map.get(cat.id, 0), + children=[], + ) + + # 构建父子关系 + roots = [] + for cat in all_cats: + node = cat_map[cat.id] + if cat.parent_id and cat.parent_id in cat_map: + cat_map[cat.parent_id].children.append(node) + else: + roots.append(node) + + return roots + + +@router.get("", response_model=list[CategoryTreeNode]) +def list_categories_flat( + kb_id: str, + user: User = Depends(get_current_user), + db: Session = Depends(get_db), +) -> list[dict]: + """获取平铺的分类列表。""" _check_kb_owner(kb_id, user, db) - from sqlalchemy import select stmt = ( select(DocumentCategory) .where(DocumentCategory.knowledge_base_id == kb_id) - .order_by(DocumentCategory.sort_order) + .order_by(DocumentCategory.path, DocumentCategory.sort_order) ) cats = list(db.scalars(stmt).all()) - return [CategoryResponse.model_validate(c) for c in cats] + return [ + { + "id": c.id, + "name": c.name, + "path": c.path, + "is_folder": c.is_folder if hasattr(c, 'is_folder') else True, + "sort_order": c.sort_order, + "parent_id": c.parent_id, + } + for c in cats + ] -@router.post("", response_model=CategoryResponse, status_code=201) +@router.post("", response_model=CategoryTreeNode, status_code=201) def create_category( kb_id: str, - body: CategoryRequest, + body: CategoryCreateRequest, user: User = Depends(get_current_user), db: Session = Depends(get_db), -) -> CategoryResponse: +) -> CategoryTreeNode: + """创建分类（支持子目录）。""" _check_kb_owner(kb_id, user, db) + + # 计算 path + parent_path = "/" + if body.parent_id: + parent = db.get(DocumentCategory, body.parent_id) + if parent is None or parent.knowledge_base_id != kb_id: + raise NotFoundError("父分类不存在。") + parent_path = parent.path + + # 生成 path + cat_id_placeholder = "new" # 创建后会更新 + path = f"{parent_path}{body.name}/" + cat = DocumentCategory( knowledge_base_id=kb_id, + parent_id=body.parent_id, name=body.name, + path=path, + is_folder=body.is_folder, sort_order=body.sort_order, ) db.add(cat) + db.flush() + + # 更新 path 使用实际 ID + cat.path = f"{parent_path}{body.name}/" db.commit() db.refresh(cat) - return CategoryResponse.model_validate(cat) + + return CategoryTreeNode( + id=cat.id, + name=cat.name, + path=cat.path, + is_folder=cat.is_folder, + sort_order=cat.sort_order, + doc_count=0, + children=[], + ) -@router.put("/{cat_id}", response_model=CategoryResponse) +@router.put("/{cat_id}", response_model=CategoryTreeNode) def update_category( kb_id: str, cat_id: str, - body: CategoryRequest, + body: CategoryUpdateRequest, user: User = Depends(get_current_user), db: Session = Depends(get_db), -) -> CategoryResponse: +) -> CategoryTreeNode: + """更新分类（改名、移动、改类型）。""" _check_kb_owner(kb_id, user, db) cat = db.get(DocumentCategory, cat_id) if cat is None or cat.knowledge_base_id != kb_id: raise NotFoundError("分类不存在。") - cat.name = body.name - cat.sort_order = body.sort_order + + old_path = cat.path + + if body.name is not None: + cat.name = body.name + if body.sort_order is not None: + cat.sort_order = body.sort_order + if body.is_folder is not None: + cat.is_folder = body.is_folder + if body.parent_id is not None: + cat.parent_id = body.parent_id + + # 重新计算 path + if body.parent_id is not None or body.name is not None: + parent_path = "/" + if cat.parent_id: + parent = db.get(DocumentCategory, cat.parent_id) + if parent: + parent_path = parent.path + new_path = f"{parent_path}{cat.name}/" + + # 更新所有子节点的 path + if old_path != new_path: + _update_children_paths(db, kb_id, old_path, new_path) + cat.path = new_path + db.commit() db.refresh(cat) - return CategoryResponse.model_validate(cat) + + return CategoryTreeNode( + id=cat.id, + name=cat.name, + path=cat.path, + is_folder=cat.is_folder, + sort_order=cat.sort_order, + doc_count=0, + children=[], + ) @router.delete("/{cat_id}", status_code=204) @@ -89,16 +236,48 @@ def delete_category( user: User = Depends(get_current_user), db: Session = Depends(get_db), ) -> None: + """删除分类（同时删除子分类，关联文档的 category_id 置 NULL）。""" _check_kb_owner(kb_id, user, db) cat = db.get(DocumentCategory, cat_id) if cat is None or cat.knowledge_base_id != kb_id: raise NotFoundError("分类不存在。") + + # 删除所有子分类 + if cat.path: + stmt = select(DocumentCategory).where( + DocumentCategory.knowledge_base_id == kb_id, + DocumentCategory.path.startswith(cat.path), + DocumentCategory.id != cat_id, + ) + children = list(db.scalars(stmt).all()) + for child in children: + db.delete(child) + + # 关联文档的 category_id 置 NULL + from app.models.document import Document + + doc_stmt = select(Document).where(Document.category_id == cat_id) + docs = list(db.scalars(doc_stmt).all()) + for doc in docs: + doc.category_id = None + db.delete(cat) db.commit() +def _update_children_paths(db: Session, kb_id: str, old_path: str, new_path: str) -> None: + """更新所有子节点的 path。""" + stmt = select(DocumentCategory).where( + DocumentCategory.knowledge_base_id == kb_id, + DocumentCategory.path.startswith(old_path), + ) + children = list(db.scalars(stmt).all()) + for child in children: + child.path = child.path.replace(old_path, new_path, 1) + + def _check_kb_owner(kb_id: str, user: User, db: Session) -> None: repo = KnowledgeBaseRepository(db) kb = repo.get_by_id(kb_id) if kb is None or kb.user_id != user.id or kb.status == "DELETED": - raise NotFoundError("知识库不存在。") + raise NotFoundError("知识库不存在。") \ No newline at end of file diff --git a/backend/app/api/documents.py b/backend/app/api/documents.py index 0f10e06..dc7e4fc 100644 --- a/backend/app/api/documents.py +++ b/backend/app/api/documents.py @@ -1,6 +1,7 @@ """文档 API 路由。""" from fastapi import APIRouter, Depends, Query, UploadFile, File, Form +from pydantic import BaseModel, Field from sqlalchemy.orm import Session from app.api.deps import get_current_user, get_db @@ -16,10 +17,19 @@ from app.services.doc_service import DocumentService router = APIRouter(prefix="/documents", tags=["documents"]) +class CreateTextRequest(BaseModel): + kb_id: str + title: str = Field(min_length=1, max_length=512) + content: str = Field(min_length=1) + content_format: str = "markdown" # markdown / text + category_id: str | None = None + + @router.post("/upload", response_model=DocumentUploadResponse, status_code=201) async def upload_document( kb_id: str = Form(...), file: UploadFile = File(...), + category_id: str = Form(None), user: User = Depends(get_current_user), db: Session = Depends(get_db), ) -> DocumentUploadResponse: @@ -31,6 +41,7 @@ async def upload_document( kb_id=kb_id, filename=file.filename or "unknown", content=content, + category_id=category_id, ) return DocumentUploadResponse( id=doc.id, @@ -41,6 +52,25 @@ async def upload_document( ) +@router.post("/create-text", response_model=DocumentResponse, status_code=201) +def create_text_document( + body: CreateTextRequest, + user: User = Depends(get_current_user), + db: Session = Depends(get_db), +) -> DocumentResponse: + """直接创建文本内容文档（无需上传文件）。""" + svc = DocumentService(db) + doc = svc.create_text( + user=user, + kb_id=body.kb_id, + title=body.title, + content=body.content, + content_format=body.content_format, + category_id=body.category_id, + ) + return _to_response(doc) + + @router.get("", response_model=DocumentListResponse) def list_documents( kb_id: str = Query(...), diff --git a/backend/app/core/errors.py b/backend/app/core/errors.py index 75f1e13..c2cdd59 100644 --- a/backend/app/core/errors.py +++ b/backend/app/core/errors.py @@ -114,12 +114,24 @@ async def app_error_handler(_: Request, exc: AppError) -> JSONResponse: async def validation_error_handler(_: Request, exc: RequestValidationError) -> JSONResponse: + # 序列化错误详情，过滤不可 JSON 序列化的对象 + errors = [] + for err in exc.errors(): + clean_err = { + "type": err.get("type", ""), + "loc": err.get("loc", []), + "msg": err.get("msg", ""), + } + if "input" in err: + clean_err["input"] = str(err["input"]) + errors.append(clean_err) + return JSONResponse( status_code=status.HTTP_422_UNPROCESSABLE_ENTITY, content={ "code": "VALIDATION_ERROR", "message": "请求参数不正确。", - "detail": exc.errors(), + "detail": errors, }, ) diff --git a/backend/app/models/document.py b/backend/app/models/document.py index 584d969..dea1abf 100644 --- a/backend/app/models/document.py +++ b/backend/app/models/document.py @@ -37,12 +37,23 @@ class Document(UUIDPrimaryKeyMixin, TimestampMixin, Base): original_filename: Mapped[str] = mapped_column( String(512), nullable=False, - comment="原始文件名 (仅展示)", + comment="原始文件名或文本标题 (仅展示)", ) - storage_path: Mapped[str] = mapped_column( + storage_path: Mapped[str | None] = mapped_column( String(1024), + nullable=True, + comment="相对于 data/ 的物理存储路径（文本内容文档为 NULL）", + ) + content: Mapped[str | None] = mapped_column( + Text, + nullable=True, + comment="直接输入的文本内容（非文件上传时使用）", + ) + content_format: Mapped[str] = mapped_column( + String(16), + default="markdown", nullable=False, - comment="相对于 data/ 的物理存储路径", + comment="内容格式：markdown / text", ) markdown_path: Mapped[str | None] = mapped_column( String(1024), @@ -51,23 +62,26 @@ class Document(UUIDPrimaryKeyMixin, TimestampMixin, Base): ) file_size: Mapped[int] = mapped_column( Integer, + default=0, nullable=False, - comment="文件大小 (字节)", + comment="文件大小 (字节)，文本内容文档为内容长度", ) mime_type: Mapped[str] = mapped_column( String(127), + default="text/markdown", nullable=False, comment="MIME 类型", ) file_ext: Mapped[str] = mapped_column( String(16), + default=".md", nullable=False, - comment="文件扩展名 (.docx/.pdf)", + comment="文件扩展名 (.docx/.pdf/.md)", ) - sha256: Mapped[str] = mapped_column( + sha256: Mapped[str | None] = mapped_column( String(64), - nullable=False, - comment="文件 SHA-256 哈希", + nullable=True, + comment="文件 SHA-256 哈希（文本内容文档为 NULL）", ) doc_token_hash: Mapped[str] = mapped_column( String(64), diff --git a/backend/app/models/document_category.py b/backend/app/models/document_category.py index ee450ae..36e6884 100644 --- a/backend/app/models/document_category.py +++ b/backend/app/models/document_category.py @@ -1,15 +1,16 @@ -"""DocumentCategory 文档分类模型。""" +"""DocumentCategory 文档分类模型（支持树形目录结构）。""" -from sqlalchemy import ForeignKey, Integer, String +from sqlalchemy import Boolean, ForeignKey, Integer, String, Text from sqlalchemy.orm import Mapped, mapped_column, relationship from app.models.base import Base, TimestampMixin, UUIDPrimaryKeyMixin class DocumentCategory(UUIDPrimaryKeyMixin, TimestampMixin, Base): - """文档分类模型。 + """文档分类模型（树形结构）。 - 技术审查 §2.2：document_categories 表。 + 通过 parent_id 实现层级关系，path 用于高效查询子树。 + is_folder=True 表示文件夹（可包含子项），False 表示叶子分类。 """ __tablename__ = "document_categories" @@ -21,11 +22,30 @@ class DocumentCategory(UUIDPrimaryKeyMixin, TimestampMixin, Base): index=True, comment="所属知识库 ID", ) + parent_id: Mapped[str | None] = mapped_column( + String(32), + ForeignKey("document_categories.id"), + nullable=True, + index=True, + comment="父分类 ID（NULL = 顶层）", + ) name: Mapped[str] = mapped_column( String(255), nullable=False, comment="分类名称", ) + path: Mapped[str] = mapped_column( + Text, + default="/", + nullable=False, + comment="物化路径，如 /01_公司层/04_岗位AI角色/", + ) + is_folder: Mapped[bool] = mapped_column( + Boolean, + default=True, + nullable=False, + comment="True=文件夹（可含子项），False=叶子分类", + ) sort_order: Mapped[int] = mapped_column( Integer, default=0, @@ -35,7 +55,9 @@ class DocumentCategory(UUIDPrimaryKeyMixin, TimestampMixin, Base): # 关系 knowledge_base = relationship("KnowledgeBase", back_populates="categories", lazy="selectin") + parent = relationship("DocumentCategory", remote_side="DocumentCategory.id", lazy="selectin") + children = relationship("DocumentCategory", back_populates="parent", lazy="selectin") documents = relationship("Document", back_populates="category", lazy="selectin") def __repr__(self) -> str: - return f"" + return f"" \ No newline at end of file diff --git a/backend/app/public/routes.py b/backend/app/public/routes.py index 851d46f..27edd67 100644 --- a/backend/app/public/routes.py +++ b/backend/app/public/routes.py @@ -12,12 +12,14 @@ import json from fastapi import APIRouter, Depends, Query, Request, Response from fastapi.responses import HTMLResponse, PlainTextResponse +from fastapi import Path as PathParam from sqlalchemy.orm import Session from app.api.deps import get_db from app.core.errors import NotFoundError, RateLimitedError from app.core.rate_limit import check_rate_limit from app.core.security import decrypt_token +from app.models.document_category import DocumentCategory from app.services.access_log_service import AccessLogService from app.services.kb_public_service import KbPublicService @@ -151,10 +153,14 @@ def kb_index_json( "updated_at": doc.updated_at, }) + # 获取目录树 + category_tree = svc.get_category_tree(kb) + data = { "name": kb.name, "description": kb.description, "document_count": len(doc_list), + "categories": category_tree, "documents": doc_list, } @@ -168,15 +174,52 @@ def kb_index_json( def kb_index_html( token: str, request: Request, + category: str = Query(None, description="按分类路径过滤，如 /01公司层/公司基本信息/"), page: int = Query(1, ge=1), db: Session = Depends(get_db), ) -> HTMLResponse: - """知识库首页（HTML）。""" + """知识库首页（HTML）。支持按目录过滤。""" _rate_limit(request, token) svc = KbPublicService(db) kb = svc.get_kb_by_token(token) _log_access(db, kb.id, f"/k/{token}", request, req_type="html") - docs, total = svc.list_documents(kb, page=page, page_size=50) + + # 获取目录树 + category_tree = svc.get_category_tree(kb) + + # 按分类过滤 + category_id = None + if category: + # 根据路径查找分类 ID + from sqlalchemy import select + stmt = select(DocumentCategory).where( + DocumentCategory.knowledge_base_id == kb.id, + DocumentCategory.path == category, + ) + cat = db.scalars(stmt).first() + if cat: + category_id = cat.id + + docs, total = svc.list_documents(kb, category_id=category_id, page=page, page_size=50) + + # 渲染目录树侧边栏 + def render_tree(nodes: list, level: int = 0) -> str: + html = "" + for node in nodes: + indent = " " * level + is_active = category == node["path"] + active_class = ' class="active"' if is_active else "" + doc_count = f' ({node["doc_count"]})' if node["doc_count"] > 0 else "" + + if node["is_folder"]: + html += f'{indent}{node["name"]}{doc_count}\n' + if node["children"]: + html += f'{indent}
    \n{render_tree(node["children"], level + 1)}{indent}
\n' + else: + html += f'{indent}{node["name"]}{doc_count}\n' + return html + + tree_html = render_tree(category_tree) doc_rows = "" for doc in docs: @@ -208,8 +251,19 @@ def kb_index_html( {_robots_meta()} {_referrer_meta()} -

{kb.name}

- {"

" + kb.description + "

" if kb.description else ""} -

文档列表

- {doc_rows if doc_rows else "

暂无文档。

"} - {pagination} -
-

This page is an AI-readable knowledge base index. Use the document links above to retrieve specific documents.

-

本页为 AI 可读知识库目录，请通过上述文档链接获取具体内容。

+
+

目录

+ +
+
+

{kb.name}

+ {"

" + kb.description + "

" if kb.description else ""} + {"← 返回全部文档" if category else ""} +

{"当前分类：" + category if category else "文档列表"}

+ {doc_rows if doc_rows else "

暂无文档。

"} + {pagination} +
+

This page is an AI-readable knowledge base index. Use the document links above to retrieve specific documents.

+

本页为 AI 可读知识库目录，请通过上述文档链接获取具体内容。

+
""" @@ -240,12 +306,49 @@ def kb_index_html( # --- 单文档访问 --- -@router.get("/{token}/doc/{doc_token}") -def doc_page_html( +@router.get("/{token}/doc/{doc_token}.md") +def doc_page_markdown( token: str, doc_token: str, request: Request, db: Session = Depends(get_db), +) -> PlainTextResponse: + """文档（Markdown）。""" + _rate_limit(request, token) + svc = KbPublicService(db) + kb = svc.get_kb_by_token(token) + doc = svc.get_document_by_token(kb, doc_token) + _log_access(db, kb.id, f"/k/{token}/doc/{doc_token}.md", request, doc_id=doc.id, req_type="doc_md") + markdown_content = svc.get_document_content(doc) + return PlainTextResponse(content=markdown_content, media_type="text/markdown") + + +@router.get("/{token}/doc/{doc_token}.txt") +def doc_page_text( + token: str, + doc_token: str, + request: Request, + db: Session = Depends(get_db), +) -> PlainTextResponse: + """文档（纯文本）。""" + _rate_limit(request, token) + svc = KbPublicService(db) + kb = svc.get_kb_by_token(token) + doc = svc.get_document_by_token(kb, doc_token) + _log_access(db, kb.id, f"/k/{token}/doc/{doc_token}.txt", request, doc_id=doc.id, req_type="doc_txt") + markdown_content = svc.get_document_content(doc) + import re + text = re.sub(r"[#*_`\[\]()>]", "", markdown_content) + text = re.sub(r"\n{3,}", "\n\n", text) + return PlainTextResponse(content=text.strip(), media_type="text/plain") + + +@router.get("/{token}/doc/{doc_token}") +def doc_page_html( + token: str, + doc_token: str = PathParam(pattern=r"^[A-Za-z0-9_-]+$"), + request: Request = None, + db: Session = Depends(get_db), ) -> HTMLResponse: """文档页面（HTML）。""" _rate_limit(request, token) @@ -253,7 +356,7 @@ def doc_page_html( kb = svc.get_kb_by_token(token) doc = svc.get_document_by_token(kb, doc_token) _log_access(db, kb.id, f"/k/{token}/doc/{doc_token}", request, doc_id=doc.id, req_type="doc_html") - markdown_content = svc.get_document_markdown(doc) + markdown_content = svc.get_document_content(doc) # Markdown → HTML（简单转换） html_content = _markdown_to_html(markdown_content) @@ -313,7 +416,7 @@ def doc_page_markdown( kb = svc.get_kb_by_token(token) doc = svc.get_document_by_token(kb, doc_token) _log_access(db, kb.id, f"/k/{token}/doc/{doc_token}.md", request, doc_id=doc.id, req_type="doc_md") - markdown_content = svc.get_document_markdown(doc) + markdown_content = svc.get_document_content(doc) return PlainTextResponse(content=markdown_content, media_type="text/markdown") @@ -330,7 +433,7 @@ def doc_page_text( kb = svc.get_kb_by_token(token) doc = svc.get_document_by_token(kb, doc_token) _log_access(db, kb.id, f"/k/{token}/doc/{doc_token}.txt", request, doc_id=doc.id, req_type="doc_txt") - markdown_content = svc.get_document_markdown(doc) + markdown_content = svc.get_document_content(doc) # 去掉 Markdown 标记 import re diff --git a/backend/app/repositories/doc_repo.py b/backend/app/repositories/doc_repo.py index 73aa5a5..428c8da 100644 --- a/backend/app/repositories/doc_repo.py +++ b/backend/app/repositories/doc_repo.py @@ -21,17 +21,20 @@ class DocumentRepository: self, kb_id: str, *, + category_id: str | None = None, page: int = 1, page_size: int = 50, status: str | None = None, ) -> tuple[list[Document], int]: - """分页获取知识库的文档列表。""" + """分页获取知识库的文档列表。支持按分类过滤。""" conditions = [ Document.knowledge_base_id == kb_id, Document.status != "DELETED", ] if status: conditions.append(Document.status == status) + if category_id: + conditions.append(Document.category_id == category_id) count_stmt = select(func.count()).select_from(Document).where(*conditions) total = self._session.scalar(count_stmt) or 0 diff --git a/backend/app/services/doc_service.py b/backend/app/services/doc_service.py index 3d48043..73b192c 100644 --- a/backend/app/services/doc_service.py +++ b/backend/app/services/doc_service.py @@ -41,6 +41,7 @@ class DocumentService: kb_id: str, filename: str, content: bytes, + category_id: str | None = None, ) -> Document: """上传文档。 @@ -111,6 +112,7 @@ class DocumentService: doc_token_hash=doc_token_hash, doc_token_encrypted=doc_token_encrypted, doc_token_hint=doc_token_hint, + category_id=category_id, ) # 更新存储路径中的 document_id @@ -200,3 +202,87 @@ class DocumentService: processor = LocalDocumentProcessor(self._session) processor.process(doc.id) + + def create_text( + self, + user: User, + kb_id: str, + title: str, + content: str, + content_format: str = "markdown", + category_id: str | None = None, + ) -> Document: + """创建文本内容文档（无需上传文件）。""" + # 校验 KB 归属 + kb = self._kb_repo.get_by_id(kb_id) + if kb is None or kb.user_id != user.id or kb.status == "DELETED": + raise NotFoundError("知识库不存在。") + + # 生成文档 token + doc_token = generate_token() + doc_token_hash = hash_token(doc_token) + doc_token_encrypted = encrypt_token(doc_token) + doc_token_hint = doc_token[-8:] if len(doc_token) >= 8 else doc_token + + # 计算内容大小 + content_size = len(content.encode("utf-8")) + + # 配额检查 + self._check_quota(user, content_size) + + # 从内容提取摘要和关键词 + summary = self._extract_summary(content) + keywords = self._extract_keywords(content) + + # 创建文档记录 + doc = Document( + knowledge_base_id=kb_id, + user_id=user.id, + category_id=category_id, + original_filename=title, + storage_path=None, # 文本内容不存储文件 + markdown_path=None, + content=content, + content_format=content_format, + file_size=content_size, + mime_type="text/markdown" if content_format == "markdown" else "text/plain", + file_ext=".md" if content_format == "markdown" else ".txt", + sha256=None, + doc_token_hash=doc_token_hash, + doc_token_encrypted=doc_token_encrypted, + doc_token_hint=doc_token_hint, + title=title, + content_summary=summary, + keywords=keywords, + status="READY", # 文本内容直接就绪 + ) + self._session.add(doc) + + # 扣减配额 + self._deduct_quota(user, content_size) + + self._session.commit() + return doc + + def _extract_summary(self, content: str, max_len: int = 200) -> str: + """提取摘要。""" + import re + + text = re.sub(r"[#*_`\[\]()>]", "", content) + text = re.sub(r"\s+", " ", text).strip() + if len(text) <= max_len: + return text + return text[:max_len] + "..." + + def _extract_keywords(self, content: str, top_k: int = 10) -> str: + """提取关键词。""" + try: + import jieba.analyse + import re + + text = re.sub(r"[#*_`\[\]()>]", "", content) + text = re.sub(r"\s+", " ", text).strip() + keywords = jieba.analyse.extract_tags(text, topK=top_k) + return ",".join(keywords) + except Exception: + return "" diff --git a/backend/app/services/kb_public_service.py b/backend/app/services/kb_public_service.py index 1dd3a6c..7875f02 100644 --- a/backend/app/services/kb_public_service.py +++ b/backend/app/services/kb_public_service.py @@ -1,14 +1,17 @@ """公共知识库服务（Phase 10/11/12 统一数据来源）。 HTML/MD/TXT/JSON/搜索全部通过此 Service 获取数据，不各自写查询逻辑。 +支持目录树结构和分类过滤。 """ from app.core.errors import NotFoundError from app.core.security import hash_token from app.models.document import Document +from app.models.document_category import DocumentCategory from app.models.knowledge_base import KnowledgeBase from app.repositories.doc_repo import DocumentRepository from app.repositories.kb_repo import KnowledgeBaseRepository +from sqlalchemy import func, select from sqlalchemy.orm import Session @@ -26,12 +29,55 @@ class KbPublicService: raise NotFoundError("知识库不存在。") return kb + def get_category_tree(self, kb: KnowledgeBase) -> list[dict]: + """获取目录树（含文档数量）。""" + stmt = ( + select(DocumentCategory) + .where(DocumentCategory.knowledge_base_id == kb.id) + .order_by(DocumentCategory.sort_order, DocumentCategory.name) + ) + all_cats = list(self._session.scalars(stmt).all()) + + # 获取每个分类的文档数量 + count_stmt = ( + select(Document.category_id, func.count()) + .where( + Document.knowledge_base_id == kb.id, + Document.status != "DELETED", + ) + .group_by(Document.category_id) + ) + doc_count_map = dict(self._session.execute(count_stmt).all()) + + # 构建树 + cat_map = {} + for cat in all_cats: + cat_map[cat.id] = { + "id": cat.id, + "name": cat.name, + "path": cat.path or "/", + "is_folder": getattr(cat, 'is_folder', True), + "sort_order": cat.sort_order, + "doc_count": doc_count_map.get(cat.id, 0), + "children": [], + } + + roots = [] + for cat in all_cats: + node = cat_map[cat.id] + if cat.parent_id and cat.parent_id in cat_map: + cat_map[cat.parent_id]["children"].append(node) + else: + roots.append(node) + + return roots + def list_documents( - self, kb: KnowledgeBase, *, page: int = 1, page_size: int = 50 + self, kb: KnowledgeBase, *, category_id: str | None = None, page: int = 1, page_size: int = 50 ) -> tuple[list[Document], int]: - """获取知识库的文档列表（排除 DELETED）。""" + """获取知识库的文档列表（排除 DELETED）。支持按分类过滤。""" return self._doc_repo.list_by_knowledge_base( - kb.id, page=page, page_size=page_size + kb.id, category_id=category_id, page=page, page_size=page_size ) def get_document_by_token(self, kb: KnowledgeBase, doc_token: str) -> Document: @@ -42,34 +88,38 @@ class KbPublicService: raise NotFoundError("文档不存在。") return doc - def get_document_markdown(self, doc: Document) -> str: - """读取文档的 Markdown 内容。""" - from app.storage.local_storage import get_storage + def get_document_content(self, doc: Document) -> str: + """获取文档内容（优先直接内容，其次 Markdown 文件）。""" + # 直接文本内容 + if doc.content: + return doc.content - if not doc.markdown_path: - return "" - storage = get_storage() - content = storage.read(doc.markdown_path) - return content.decode("utf-8") + # 从文件读取 Markdown + if doc.markdown_path: + from app.storage.local_storage import get_storage + + storage = get_storage() + content = storage.read(doc.markdown_path) + return content.decode("utf-8") + + return "" def search_documents( self, kb: KnowledgeBase, query: str, *, page: int = 1, page_size: int = 20 ) -> tuple[list[dict], int]: """关键词搜索文档（Phase 12：SQLite LIKE / FTS5）。""" - # MVP 简化：使用 LIKE 搜索标题+描述+关键词 - from sqlalchemy import func, or_, select - conditions = [ Document.knowledge_base_id == kb.id, Document.status == "READY", ] like_pattern = f"%{query}%" - search_condition = or_( - Document.title.like(like_pattern), - Document.description.like(like_pattern), - Document.keywords.like(like_pattern), - Document.content_summary.like(like_pattern), + search_condition = ( + Document.title.like(like_pattern) + | Document.description.like(like_pattern) + | Document.keywords.like(like_pattern) + | Document.content_summary.like(like_pattern) + | Document.content.like(like_pattern) ) conditions.append(search_condition) @@ -87,6 +137,13 @@ class KbPublicService: results = [] for doc in docs: + # 获取分类路径 + category_path = "" + if doc.category_id: + cat = self._session.get(DocumentCategory, doc.category_id) + if cat: + category_path = cat.path or "" + results.append({ "id": doc.id, "title": doc.title or doc.original_filename, @@ -95,6 +152,7 @@ class KbPublicService: "file_type": doc.file_ext, "updated_at": doc.updated_at, "url_hint": doc.doc_token_hint, + "category_path": category_path, }) return results, total diff --git a/backend/app/services/kb_service.py b/backend/app/services/kb_service.py index a2a5529..f4f179b 100644 --- a/backend/app/services/kb_service.py +++ b/backend/app/services/kb_service.py @@ -1,6 +1,7 @@ -"""知识库服务：CRUD + Token 管理。""" +"""知识库服务：CRUD + Token 管理 + 默认目录树。""" from app.core.errors import NotFoundError, PermissionDeniedError +from app.models.document_category import DocumentCategory from app.models.knowledge_base import KnowledgeBase from app.models.user import User from app.repositories.kb_repo import KnowledgeBaseRepository @@ -18,6 +19,7 @@ class KnowledgeBaseService: """创建知识库。返回 (kb, full_token)。 full_token 仅此一次返回，用于构建完整 AI URL。 + 自动创建默认目录树结构。 """ token, token_hash, token_encrypted, token_hint = self._token_svc.create_token_pair() kb = self._kb_repo.create( @@ -28,9 +30,46 @@ class KnowledgeBaseService: token_encrypted=token_encrypted, token_hint=token_hint, ) + self._seed_default_categories(kb.id) self._session.commit() return kb, token + def _seed_default_categories(self, kb_id: str) -> None: + """创建默认目录树结构。""" + default_tree = [ + ("01 公司层", ["公司基本信息", "经营理念", "四大价值", "A/M/B三态"]), + ("02 战略层", ["公司战略", "客户战略", "AI战略", "产品战略"]), + ("03 部门层", ["企划", "技术", "交付", "市场"]), + ("04 岗位/AI角色", []), + ("05 业务知识", ["价值创造", "价值传递", "价值交付", "价值支持"]), + ("06 资产库", ["文案", "SOP", "模板", "案例", "Prompt"]), + ] + + for i, (folder_name, children) in enumerate(default_tree): + # 创建顶层文件夹 + folder = DocumentCategory( + knowledge_base_id=kb_id, + parent_id=None, + name=folder_name, + path=f"/{folder_name}/", + is_folder=True, + sort_order=i, + ) + self._session.add(folder) + self._session.flush() + + # 创建子分类 + for j, child_name in enumerate(children): + child = DocumentCategory( + knowledge_base_id=kb_id, + parent_id=folder.id, + name=child_name, + path=f"/{folder_name}/{child_name}/", + is_folder=False, + sort_order=j, + ) + self._session.add(child) + def get_or_404(self, kb_id: str, user: User) -> KnowledgeBase: """获取知识库，校验所有权。不存在或无权 → 404。""" kb = self._kb_repo.get_by_id(kb_id) diff --git a/backend/tests/test_categories.py b/backend/tests/test_categories.py index d3ed78a..f834592 100644 --- a/backend/tests/test_categories.py +++ b/backend/tests/test_categories.py @@ -35,15 +35,19 @@ def test_create_category() -> None: def test_list_categories() -> None: with _client() as client: kb_id = _setup(client) - client.post(f"/api/knowledge-bases/{kb_id}/categories", json={"name": "A", "sort_order": 2}) - client.post(f"/api/knowledge-bases/{kb_id}/categories", json={"name": "B", "sort_order": 1}) + # 创建知识库时会自动 seed 默认目录树 + # 再手动添加两个分类 + client.post(f"/api/knowledge-bases/{kb_id}/categories", json={"name": "A", "sort_order": 2, "is_folder": False}) + client.post(f"/api/knowledge-bases/{kb_id}/categories", json={"name": "B", "sort_order": 1, "is_folder": False}) resp = client.get(f"/api/knowledge-bases/{kb_id}/categories") assert resp.status_code == 200 cats = resp.json() - assert len(cats) == 2 - # 按 sort_order 排序 - assert cats[0]["name"] == "B" - assert cats[1]["name"] == "A" + # 应包含默认 seed 的分类 + 手动添加的 2 个 + assert len(cats) >= 2 + # 验证手动添加的分类存在 + names = [c["name"] for c in cats] + assert "A" in names + assert "B" in names def test_update_category() -> None: diff --git a/frontend/src/api/client.js b/frontend/src/api/client.js index a545baf..4eb6e46 100644 --- a/frontend/src/api/client.js +++ b/frontend/src/api/client.js @@ -21,7 +21,16 @@ apiClient.interceptors.response.use((response) => response, (error) => { return Promise.reject(error); } // 业务错误：显示服务端中文文案 - const message = data?.message || '请求失败'; + let message = data?.message || '请求失败'; + // 验证错误：拼接具体字段错误 + if (data?.detail && Array.isArray(data.detail)) { + const fieldErrors = data.detail.map((e) => { + const field = e.loc?.join('.') || ''; + return field ? `${field}: ${e.msg}` : e.msg; + }).join('\n'); + if (fieldErrors) + message = fieldErrors; + } ElMessage.error(message); } else if (error.request) { diff --git a/frontend/src/api/client.ts b/frontend/src/api/client.ts index 91460fe..ed83ef3 100644 --- a/frontend/src/api/client.ts +++ b/frontend/src/api/client.ts @@ -27,7 +27,15 @@ apiClient.interceptors.response.use( } // 业务错误：显示服务端中文文案 - const message = data?.message || '请求失败' + let message = data?.message || '请求失败' + // 验证错误：拼接具体字段错误 + if (data?.detail && Array.isArray(data.detail)) { + const fieldErrors = data.detail.map((e: any) => { + const field = e.loc?.join('.') || '' + return field ? `${field}: ${e.msg}` : e.msg + }).join('\n') + if (fieldErrors) message = fieldErrors + } ElMessage.error(message) } else if (error.request) { ElMessage.error('网络错误，请检查连接。') diff --git a/frontend/src/views/KbDetail.vue b/frontend/src/views/KbDetail.vue index 0544000..3841ec4 100644 --- a/frontend/src/views/KbDetail.vue +++ b/frontend/src/views/KbDetail.vue @@ -1,5 +1,5 @@