feat(knowledge): 支持按知识库配置文档数量上限

dashboard/src/api/modules/knowledgeBases.ts

@@ -37,6 +37,7 @@ export interface KnowledgeBase {
   embedding_model: string;
   embedding_dim: number;
   doc_count: number;
+  max_documents: number;
   created_at: number;
   updated_at: number;
 }
@@ -151,6 +152,7 @@ export const knowledgeBasesApi = {
     default_open?: boolean;
     shared?: boolean;
     icon_name?: string;
+    max_documents?: number;
   }) =>
     request<KnowledgeBase>("/knowledge-bases", {
       method: "POST",
@@ -165,6 +167,7 @@ export const knowledgeBasesApi = {
       default_open?: boolean;
       shared?: boolean;
       icon_name?: string;
+      max_documents?: number;
     },
   ) =>
     request<KnowledgeBase>(`/knowledge-bases/${id}`, {

dashboard/src/locales/en.json

@@ -322,6 +322,8 @@
     "documents": "Documents",
     "documentLimit": "{{count}} / {{max}} documents",
     "documentLimitReached": "This knowledge base already contains the maximum of {{count}} documents.",
+    "maxDocuments": "Document limit",
+    "maxDocumentsHint": "Maximum number of files in this knowledge base. 0 = unlimited, default 100.",
     "documentTooLarge": "Each document must be at most {{sizeMb}} MB.",
     "baseLimitReached": "You can create at most {{count}} knowledge bases.",
     "uploadHint": "Supports md / txt / pdf / docx / pptx. Max {{sizeMb}} MB per file.",

dashboard/src/locales/zh.json

@@ -322,6 +322,8 @@
     "documents": "文档",
     "documentLimit": "{{count}} / {{max}} 个文档",
     "documentLimitReached": "此知识库已达到 {{count}} 个文档的上限。",
+    "maxDocuments": "文档数量上限",
+    "maxDocumentsHint": "本知识库可容纳的最大文档数，0 表示不限制，默认为 100。",
     "documentTooLarge": "单个文档不能超过 {{sizeMb}} MB。",
     "baseLimitReached": "每个用户最多可创建 {{count}} 个知识库。",
     "uploadHint": "支持 md / txt / pdf / docx / pptx，单文件不超过 {{sizeMb}} MB。",

dashboard/src/pages/KnowledgeBases/index.tsx

@@ -12,6 +12,7 @@ import {
   Empty,
   Form,
   Input,
+  InputNumber,
   List,
   Modal,
   Popconfirm,
@@ -94,6 +95,7 @@ type BaseFormValues = {
   name: string;
   description?: string;
   icon_name?: string;
+  max_documents?: number;
 };
 
 type DocsViewMode = "card" | "table";
@@ -327,7 +329,8 @@ export default function KnowledgeBasesPage() {
     : 0;
   const atBaseLimit = ownedBaseCount >= limits.max_bases_per_owner;
   const fileCount = documents.filter((document) => !document.is_dir).length;
-  const isAtDocumentLimit = fileCount >= limits.max_docs_per_kb;
+  const isAtDocumentLimit =
+    fileCount >= (selected?.max_documents ?? limits.max_docs_per_kb);
   const folderEntries = documents
     .filter((document) =>
       isDirectKnowledgeChild(document.path || document.filename, currentFolder),
@@ -598,6 +601,7 @@ export default function KnowledgeBasesPage() {
       name: "",
       description: "",
       icon_name: "book-open",
+      max_documents: 100,
     });
     setDefaultOpenChecked(false);
     setSharedChecked(false);
@@ -611,6 +615,7 @@ export default function KnowledgeBasesPage() {
       name: selected.name,
       description: selected.description,
       icon_name: selected.icon_name || undefined,
+      max_documents: selected.max_documents,
     });
     setDefaultOpenChecked(selected.default_open);
     setSharedChecked(selected.shared);
@@ -813,7 +818,10 @@ export default function KnowledgeBasesPage() {
 
   const uploadDocuments = async (files: FileList | null) => {
     if (!selected || !files || !usable || isAtDocumentLimit) return;
-    const remaining = Math.max(0, limits.max_docs_per_kb - fileCount);
+    const remaining = Math.max(
+      0,
+      (selected?.max_documents ?? limits.max_docs_per_kb) - fileCount,
+    );
     const chosen = Array.from(files).slice(0, remaining);
     const oversized = chosen.filter(
       (file) => file.size > limits.max_document_bytes,
@@ -1476,7 +1484,7 @@ export default function KnowledgeBasesPage() {
                       <span className={skillStyles.gridCount}>
                         {t("knowledgeBases.documentLimit", {
                           count: fileCount,
-                          max: limits.max_docs_per_kb,
+                          max: selected?.max_documents ?? limits.max_docs_per_kb,
                         })}
                       </span>
                       <div className={skillStyles.gridToolbarRight}>
@@ -1607,7 +1615,7 @@ export default function KnowledgeBasesPage() {
                         type="info"
                         showIcon
                         message={t("knowledgeBases.documentLimitReached", {
-                          count: limits.max_docs_per_kb,
+                          count: selected?.max_documents ?? limits.max_docs_per_kb,
                         })}
                       />
                     ) : null}
@@ -1937,6 +1945,19 @@ export default function KnowledgeBasesPage() {
               showCount
             />
           </Form.Item>
+          <Form.Item
+            name="max_documents"
+            label={t("knowledgeBases.maxDocuments")}
+            extra={t("knowledgeBases.maxDocumentsHint")}
+          >
+            <InputNumber
+              min={0}
+              max={10000}
+              step={1}
+              precision={0}
+              style={{ width: "100%" }}
+            />
+          </Form.Item>
           <Form.Item name="icon_name" label={t("knowledgeBases.icon")}>
             <KnowledgeIconPicker />
           </Form.Item>

src/octop/api/routers/knowledge_bases.py

@@ -69,6 +69,12 @@ class CreateBaseBody(BaseModel):
     default_open: bool = False
     shared: bool = False
     icon_name: str = Field(default="", max_length=64)
+    max_documents: int | None = Field(
+        default=None,
+        ge=0,
+        le=10_000,
+        description="Per-base document limit. 0 = unlimited, default 100.",
+    )
 
 
 class CreateFolderBody(BaseModel):
@@ -98,6 +104,12 @@ class UpdateBaseBody(BaseModel):
     default_open: bool | None = None
     shared: bool | None = None
     icon_name: str | None = Field(default=None, max_length=64)
+    max_documents: int | None = Field(
+        default=None,
+        ge=0,
+        le=10_000,
+        description="Per-base document limit. 0 = unlimited.",
+    )
 
 
 class RenameDocumentBody(BaseModel):
@@ -418,6 +430,7 @@ async def create_base(
             default_open=body.default_open,
             shared=body.shared,
             icon_name=body.icon_name.strip(),
+            max_documents=body.max_documents if body.max_documents is not None else MAX_DOCS_PER_KB,
         )
         return _base_payload(server, base)
     except Exception as exc:
@@ -480,6 +493,7 @@ async def update_base(
                 default_open=body.default_open,
                 shared=body.shared,
                 icon_name=body.icon_name.strip() if body.icon_name is not None else None,
+                max_documents=body.max_documents,
                 is_admin=_is_admin(user),
             ),
         )

src/octop/infra/db/migrate.py

@@ -637,6 +637,8 @@ def _ensure_knowledge_bases_schema(db: DatabasePool) -> None:
     """Create or rebuild knowledge tables to the integer-PK identity schema."""
     _rebuild_knowledge_identity_schema(db)
     _drop_knowledge_base_members(db)
+    # Schema v10: per-knowledge-base configurable document limit.
+    _ensure_column(db, "knowledge_bases", "max_documents", "INTEGER NOT NULL DEFAULT 100")
 
 
 def _ensure_sso_oidc_schema(db: DatabasePool) -> None:

src/octop/infra/db/migrations/010_kb_max_documents.pg.sql

@@ -0,0 +1,7 @@
+-- Schema v10: per-knowledge-base configurable document limit.
+-- PostgreSQL: ADD COLUMN IF NOT EXISTS is safe (skips if already present).
+
+ALTER TABLE knowledge_bases
+  ADD COLUMN IF NOT EXISTS max_documents INTEGER NOT NULL DEFAULT 100;
+
+UPDATE _schema_version SET version = 10;

src/octop/infra/db/migrations/010_kb_max_documents.sql

@@ -0,0 +1,6 @@
+-- Schema v10: per-knowledge-base configurable document limit.
+-- The column is added by migrate.py::_ensure_knowledge_bases_schema so that
+-- boot-time repair covers pre-v10 databases. This file only bumps _schema_version.
+-- 100 is the previous system-wide default; 0 means unlimited.
+
+UPDATE _schema_version SET version = 10;

src/octop/infra/db/repos/knowledge.py

@@ -31,11 +31,16 @@ class KnowledgeBaseRow:
     embedding_model: str
     embedding_dim: int
     doc_count: int
+    max_documents: int
     created_at: int
     updated_at: int
 
     @classmethod
     def from_row(cls, r: DbRow) -> KnowledgeBaseRow:
+        # Schema v10 adds max_documents. Fall back to 100 for pre-v10 DBs.
+        # sqlite3.Row has no __contains__; use keys() (like users.py).
+        keys = frozenset(r.keys()) if hasattr(r, "keys") else frozenset()
+        max_doc = int(r["max_documents"]) if "max_documents" in keys else 100
         return cls(
             id=str(r["knowledge_base_id"]),
             pk=int(r["id"]),
@@ -48,6 +53,7 @@ def from_row(cls, r: DbRow) -> KnowledgeBaseRow:
             embedding_model=r["embedding_model"],
             embedding_dim=r["embedding_dim"],
             doc_count=r["doc_count"],
+            max_documents=max_doc,
             created_at=r["created_at"],
             updated_at=r["updated_at"],
         )
@@ -112,29 +118,54 @@ def create_base(
         icon_name: str = "",
         embedding_model: str = "",
         embedding_dim: int = 0,
+        max_documents: int | None = None,
     ) -> KnowledgeBaseRow:
         kb_id = self._allocate_base_id()
         ts = now_ts()
         with self._db.transaction() as conn:
-            conn.execute(
-                "INSERT INTO knowledge_bases("
-                "knowledge_base_id, owner_user_id, name, description, default_open, shared, "
-                "icon_name, embedding_model, embedding_dim, doc_count, created_at, updated_at"
-                ") VALUES (?, ?, ?, ?, ?, ?, ?, ?, ?, 0, ?, ?)",
-                (
-                    kb_id,
-                    owner_user_id,
-                    name,
-                    description,
-                    bool_int(default_open),
-                    bool_int(shared),
-                    icon_name,
-                    embedding_model,
-                    embedding_dim,
-                    ts,
-                    ts,
-                ),
-            )
+            if max_documents is None:
+                # Rely on the column DEFAULT (100) for max_documents.
+                conn.execute(
+                    "INSERT INTO knowledge_bases("
+                    "knowledge_base_id, owner_user_id, name, description, default_open, shared, "
+                    "icon_name, embedding_model, embedding_dim, doc_count, created_at, updated_at"
+                    ") VALUES (?, ?, ?, ?, ?, ?, ?, ?, ?, 0, ?, ?)",
+                    (
+                        kb_id,
+                        owner_user_id,
+                        name,
+                        description,
+                        bool_int(default_open),
+                        bool_int(shared),
+                        icon_name,
+                        embedding_model,
+                        embedding_dim,
+                        ts,
+                        ts,
+                    ),
+                )
+            else:
+                conn.execute(
+                    "INSERT INTO knowledge_bases("
+                    "knowledge_base_id, owner_user_id, name, description, default_open, shared, "
+                    "icon_name, embedding_model, embedding_dim, doc_count, max_documents, "
+                    "created_at, updated_at"
+                    ") VALUES (?, ?, ?, ?, ?, ?, ?, ?, ?, 0, ?, ?, ?)",
+                    (
+                        kb_id,
+                        owner_user_id,
+                        name,
+                        description,
+                        bool_int(default_open),
+                        bool_int(shared),
+                        icon_name,
+                        embedding_model,
+                        embedding_dim,
+                        max_documents,
+                        ts,
+                        ts,
+                    ),
+                )
         row = self.get_base(kb_id)
         if row is None:
             raise RuntimeError(f"knowledge base insert failed: {kb_id}")
@@ -183,6 +214,7 @@ def update_base(
         embedding_model: str | None = None,
         embedding_dim: int | None = None,
         doc_count: int | None = None,
+        max_documents: int | None = None,
     ) -> None:
         fields, params = partial_updates(
             [
@@ -194,6 +226,7 @@ def update_base(
                 ("embedding_model", embedding_model),
                 ("embedding_dim", embedding_dim),
                 ("doc_count", doc_count),
+                ("max_documents", max_documents),
             ]
         )
         if not fields:
@@ -264,8 +297,12 @@ def create_document(
             self.ensure_folder(kb_id, folder)
         doc_id = new_ulid()
         ts = now_ts()
+        # Treat both None (caller did not specify) and 0 (per-base "unlimited"
+        # sentinel) as unbounded. Otherwise 0 would be enforced literally as
+        # "at most 0 documents" and reject every create.
+        enforce_limit = max_documents is not None and max_documents > 0
         with self._db.transaction() as conn:
-            if max_documents is not None:
+            if enforce_limit:
                 cursor = conn.execute(
                     "UPDATE knowledge_bases SET doc_count = doc_count + 1, updated_at = ? "
                     "WHERE knowledge_base_id = ? AND doc_count < ?",
@@ -280,7 +317,7 @@ def create_document(
                 ") VALUES (?, ?, ?, ?, 0, ?, ?, ?, ?, '', 0, ?, ?)",
                 (doc_id, kb_id, rel, name, content_type, byte_size, content_hash, status, ts, ts),
             )
-            if max_documents is None:
+            if not enforce_limit:
                 conn.execute(
                     "UPDATE knowledge_bases SET doc_count = doc_count + 1, updated_at = ? "
                     "WHERE knowledge_base_id = ?",