chore: add admin settings to disable doc-ray and markitdown (#1381)

Author: zjx20

aperag/api/components/schemas/settings.yaml

@@ -7,3 +7,9 @@ Settings:
     mineru_api_token:
       type: string
       description: API token for MinerU
+    use_doc_ray:
+      type: boolean
+      description: Whether to use DocRay
+    use_markitdown:
+      type: boolean
+      description: Whether to use MarkItDown

aperag/docparser/doc_parser.py

@@ -93,6 +93,12 @@ def __init__(self, parser_config: Optional[dict] = None, full_config: list[Parse
                     cfg.settings["api_token"] = token
                 else:
                     cfg.enabled = False
+            elif cfg.name == DocRayParser.name:
+                use_doc_ray = parser_config.get("use_doc_ray", False)
+                cfg.enabled = use_doc_ray
+            elif cfg.name == MarkItDownParser.name:
+                use_markitdown = parser_config.get("use_markitdown", True)
+                cfg.enabled = use_markitdown
 
         for cfg in self.config:
             if not cfg.enabled:

aperag/migration/versions/20250929183757-ef8cf2222205.py

@@ -0,0 +1,26 @@
+"""add use_markitdown as default setting
+
+Revision ID: ef8cf2222205
+Revises: d112e0332219
+Create Date: 2025-09-29 18:37:57.365896
+
+"""
+from typing import Sequence, Union
+
+from alembic import op
+import sqlalchemy as sa
+
+
+# revision identifiers, used by Alembic.
+revision: str = 'ef8cf2222205'
+down_revision: Union[str, None] = 'd112e0332219'
+branch_labels: Union[str, Sequence[str], None] = None
+depends_on: Union[str, Sequence[str], None] = None
+
+
+def upgrade() -> None:
+    op.execute("INSERT INTO setting (key, value, gmt_created, gmt_updated) VALUES ('use_markitdown', 'true', NOW(), NOW())")
+
+
+def downgrade() -> None:
+    op.execute("DELETE FROM setting WHERE key = 'use_markitdown'")

aperag/schema/view_models.py

@@ -14,7 +14,7 @@
 
 # generated by datamodel-codegen:
 #   filename:  openapi.merged.yaml
-#   timestamp: 2025-09-08T06:55:15+00:00
+#   timestamp: 2025-09-29T10:27:18+00:00
 
 from __future__ import annotations
 
@@ -916,6 +916,10 @@ class SearchRequest(BaseModel):
 class Settings(BaseModel):
     use_mineru: Optional[bool] = Field(None, description='Whether to use MinerU')
     mineru_api_token: Optional[str] = Field(None, description='API token for MinerU')
+    use_doc_ray: Optional[bool] = Field(None, description='Whether to use DocRay')
+    use_markitdown: Optional[bool] = Field(
+        None, description='Whether to use MarkItDown'
+    )
 
 
 class GraphLabelsResponse(BaseModel):

aperag/service/setting_service.py

@@ -51,6 +51,18 @@ async def get_use_mineru(self) -> bool:
     async def update_use_mineru(self, use_mineru: bool):
         await self.update_setting("use_mineru", use_mineru)
 
+    async def get_use_doc_ray(self) -> bool:
+        return await self.get_setting("use_doc_ray") or False
+
+    async def update_use_doc_ray(self, use_doc_ray: bool):
+        await self.update_setting("use_doc_ray", use_doc_ray)
+
+    async def get_use_markitdown(self) -> bool:
+        return await self.get_setting("use_markitdown") or True
+
+    async def update_use_markitdown(self, use_markitdown: bool):
+        await self.update_setting("use_markitdown", use_markitdown)
+
     async def get_all_settings(self) -> dict:
         settings = await self.db_ops.query_all_settings()
         return {s.key: json.loads(s.value) for s in settings}

web/src/api/models/default-model-config.ts

@@ -26,6 +26,12 @@ export interface DefaultModelConfig {
      * @memberof DefaultModelConfig
      */
     'scenario': DefaultModelConfigScenarioEnum;
+    /**
+     * 
+     * @type {string}
+     * @memberof DefaultModelConfig
+     */
+    'custom_llm_provider'?: string;
     /**
      * The name of the model provider
      * @type {string}

web/src/api/models/settings.ts

@@ -32,5 +32,17 @@ export interface Settings {
      * @memberof Settings
      */
     'mineru_api_token'?: string;
+    /**
+     * Whether to use DocRay
+     * @type {boolean}
+     * @memberof Settings
+     */
+    'use_doc_ray'?: boolean;
+    /**
+     * Whether to use MarkItDown
+     * @type {boolean}
+     * @memberof Settings
+     */
+    'use_markitdown'?: boolean;
 }
 

web/src/api/openapi.merged.yaml

@@ -5757,6 +5757,12 @@ components:
         mineru_api_token:
           type: string
           description: API token for MinerU
+        use_doc_ray:
+          type: boolean
+          description: Whether to use DocRay
+        use_markitdown:
+          type: boolean
+          description: Whether to use MarkItDown
     graphLabelsResponse:
       type: object
       description: Response containing available graph labels
@@ -6485,6 +6491,8 @@ components:
             - default_for_background_task
           description: The scenario for which this default model is configured
           example: default_for_embedding
+        custom_llm_provider:
+          type: string
         provider_name:
           type: string
           nullable: true

web/src/app/admin/configuration/page.tsx

@@ -8,7 +8,7 @@ import {
 import { getServerApi } from '@/lib/api/server';
 import { Metadata } from 'next';
 import { getTranslations } from 'next-intl/server';
-import { MinerUSettings } from './mineru-settings';
+import { ParserSettings } from './parser-settings';
 import { QuotaSettings } from './quota-settings';
 
 export async function generateMetadata(): Promise<Metadata> {
@@ -40,7 +40,7 @@ export default async function Page() {
         </PageDescription>
 
         <div className="flex flex-col gap-6">
-          <MinerUSettings data={settings} />
+          <ParserSettings data={settings} />
           <QuotaSettings data={resSystemDefaultQuotas.data.quotas} />
         </div>
       </PageContent>

…/admin/configuration/mineru-settings.tsx …/admin/configuration/parser-settings.tsxweb/src/app/admin/configuration/mineru-settings.tsx renamed to web/src/app/admin/configuration/parser-settings.tsx web/src/app/admin/configuration/mineru-settings.tsx renamed to web/src/app/admin/configuration/parser-settings.tsx

@@ -22,9 +22,11 @@ import { toast } from 'sonner';
 const defaultValue = {
   use_mineru: false,
   mineru_api_token: '',
+  use_doc_ray: false,
+  use_markitdown: true,
 };
 
-export const MinerUSettings = ({
+export const ParserSettings = ({
   data: initData = defaultValue,
 }: {
   data: Settings;
@@ -46,9 +48,9 @@ export const MinerUSettings = ({
     toast.success('Saved successfully');
   }, [data]);
 
-  const handleUseMineruChange = useCallback(
-    async (checked: boolean) => {
-      const settings = { ...data, use_mineru: checked };
+  const handleSwitchChange = useCallback(
+    async (key: keyof Settings, checked: boolean) => {
+      const settings = { ...data, [key]: checked };
       setData(settings);
       await apiClient.defaultApi.settingsPut({
         settings,
@@ -98,7 +100,9 @@ export const MinerUSettings = ({
             </div>
             <Switch
               checked={data.use_mineru}
-              onCheckedChange={handleUseMineruChange}
+              onCheckedChange={(checked) =>
+                handleSwitchChange('use_mineru', checked)
+              }
             />
           </div>
         </CardHeader>
@@ -138,6 +142,42 @@ export const MinerUSettings = ({
           </Button>
         </CardFooter>
       </Card>
+      <Card>
+        <CardHeader>
+          <div className="flex flex-row items-center justify-between">
+            <div>
+              <CardTitle>{admin_config('use_doc_ray')}</CardTitle>
+              <CardDescription>
+                {admin_config('use_doc_ray_description')}
+              </CardDescription>
+            </div>
+            <Switch
+              checked={data.use_doc_ray}
+              onCheckedChange={(checked) =>
+                handleSwitchChange('use_doc_ray', checked)
+              }
+            />
+          </div>
+        </CardHeader>
+      </Card>
+      <Card>
+        <CardHeader>
+          <div className="flex flex-row items-center justify-between">
+            <div>
+              <CardTitle>{admin_config('use_markitdown')}</CardTitle>
+              <CardDescription>
+                {admin_config('use_markitdown_description')}
+              </CardDescription>
+            </div>
+            <Switch
+              checked={data.use_markitdown}
+              onCheckedChange={(checked) =>
+                handleSwitchChange('use_markitdown', checked)
+              }
+            />
+          </div>
+        </CardHeader>
+      </Card>
     </>
   );
 };