diff --git a/document/data/doc-last-modified.json b/document/data/doc-last-modified.json
index b4dd3674166f..275eab0ecf71 100644
--- a/document/data/doc-last-modified.json
+++ b/document/data/doc-last-modified.json
@@ -355,10 +355,8 @@
   "content/self-host/upgrading/4-16/4161.mdx": "2026-08-24T10:06:28+08:00",
   "content/self-host/upgrading/4-16/4162.en.mdx": "2026-09-03T21:50:14+08:00",
   "content/self-host/upgrading/4-16/4162.mdx": "2026-09-03T21:50:14+08:00",
-  "content/self-host/upgrading/4-17/4170.en.mdx": "2026-09-12T20:05:26+08:00",
-  "content/self-host/upgrading/4-17/4170.mdx": "2026-09-12T20:05:26+08:00",
-  "content/self-host/upgrading/4-17/4171.en.mdx": "2026-09-12T22:05:59+08:00",
-  "content/self-host/upgrading/4-17/4171.mdx": "2026-09-12T22:05:59+08:00",
+  "content/self-host/upgrading/4-17/4170.en.mdx": "2026-09-11T13:12:32+08:00",
+  "content/self-host/upgrading/4-17/4170.mdx": "2026-09-11T13:12:32+08:00",
   "content/self-host/upgrading/outdated/40.en.mdx": "2026-08-16T23:16:43+08:00",
   "content/self-host/upgrading/outdated/40.mdx": "2026-08-16T23:16:43+08:00",
   "content/self-host/upgrading/outdated/41.en.mdx": "2026-08-16T23:16:43+08:00",
@@ -501,6 +499,6 @@
   "content/self-host/upgrading/upgrade-instruction.mdx": "2026-09-08T22:44:33+08:00",
   "content/self-host/upgrading/version-timeline.en.mdx": "2026-08-16T23:16:43+08:00",
   "content/self-host/upgrading/version-timeline.mdx": "2026-08-16T23:16:43+08:00",
-  "content/toc.en.mdx": "2026-09-12T22:05:59+08:00",
-  "content/toc.mdx": "2026-09-12T22:05:59+08:00"
+  "content/toc.en.mdx": "2026-09-03T21:50:14+08:00",
+  "content/toc.mdx": "2026-09-03T21:50:14+08:00"
 }
\ No newline at end of file
diff --git a/packages/global/core/app/type.ts b/packages/global/core/app/type.ts
index 8492bcf402b0..3cbb97d51e18 100644
--- a/packages/global/core/app/type.ts
+++ b/packages/global/core/app/type.ts
@@ -125,22 +125,24 @@ export const EntryPointItemTypeSchema = z.object({
 });
 export type EntryPointItemType = z.infer<typeof EntryPointItemTypeSchema>;
 
-export const EntryPointItemsTypeSchema = z.array(EntryPointItemTypeSchema).superRefine((items, ctx) => {
-  const seenNames = new Map<string, number>();
-  items.forEach((item, index) => {
-    const normalizedName = item.name.trim();
-    const previousIndex = seenNames.get(normalizedName);
-    if (previousIndex !== undefined) {
-      ctx.addIssue({
-        code: 'custom',
-        path: [index, 'name'],
-        message: 'Entry point names must be unique'
-      });
-      return;
-    }
-    seenNames.set(normalizedName, index);
+export const EntryPointItemsTypeSchema = z
+  .array(EntryPointItemTypeSchema)
+  .superRefine((items, ctx) => {
+    const seenNames = new Map<string, number>();
+    items.forEach((item, index) => {
+      const normalizedName = item.name.trim();
+      const previousIndex = seenNames.get(normalizedName);
+      if (previousIndex !== undefined) {
+        ctx.addIssue({
+          code: 'custom',
+          path: [index, 'name'],
+          message: 'Entry point names must be unique'
+        });
+        return;
+      }
+      seenNames.set(normalizedName, index);
+    });
   });
-});
 
 export const AppChatConfigTypeSchema = z.object({
   welcomeText: optionalNullToUndefined(z.string()).meta({
diff --git a/packages/global/core/dataset/type.ts b/packages/global/core/dataset/type.ts
index c3a165d6ca0d..04b27df3d996 100644
--- a/packages/global/core/dataset/type.ts
+++ b/packages/global/core/dataset/type.ts
@@ -119,6 +119,27 @@ export const ChunkSettingsSchema = z.object({
 });
 export type ChunkSettingsType = z.infer<typeof ChunkSettingsSchema>;
 
+/* ===== Iultmzh file parse config ===== */
+// 外部文档解析服务开关,整体校验、整体存取;仅当 collection 的 customPdfParse=true
+// 且系统配置了 customPdfParse.url 时生效。缺失字段由读取层用固定默认值补全。
+export const sangforFileParseConfigSchema = z
+  .object({
+    keep_header_footer: z.boolean().optional().meta({
+      description: '是否保留页眉页脚,默认删除'
+    }),
+    keep_appendix: z.boolean().optional().meta({
+      description: '是否保留附录,默认删除首个附录标题后的内容'
+    }),
+    image_analysis: z.boolean().optional().meta({
+      description: '是否保留图片识别(含印章)结果,默认关闭'
+    }),
+    chart_analysis: z.boolean().optional().meta({
+      description: '是否保留图表转表格结果(同时保留原图片),默认关闭'
+    })
+  })
+  .meta({ description: '外部文档解析配置' });
+export type IultmzhFileParseConfigType = z.infer<typeof sangforFileParseConfigSchema>;
+
 /* ===== Dataset ===== */
 export const DatasetSchema = z
   .object({
@@ -154,6 +175,10 @@ export const DatasetSchema = z
       .meta({ description: '网站配置' }),
     chunkSettings: ChunkSettingsSchema.optional().meta({ description: '分块配置' }),
 
+    sangforFileParseConfig: sangforFileParseConfigSchema
+      .optional()
+      .meta({ description: '外部文档解析配置' }),
+
     apiDatasetServer: ApiDatasetServerSchema.optional().meta({ description: 'API 服务器配置' }),
 
     deleteTime: z.coerce.date().nullish().meta({ description: '删除时间' }),
diff --git a/packages/global/core/workflow/migration/legacy/systemConfig.ts b/packages/global/core/workflow/migration/legacy/systemConfig.ts
index 2ec937737e8a..73bc85af59d1 100644
--- a/packages/global/core/workflow/migration/legacy/systemConfig.ts
+++ b/packages/global/core/workflow/migration/legacy/systemConfig.ts
@@ -147,7 +147,10 @@ export function migrateSystemConfigToChatConfig(input: LegacyWorkflowDataInput)
     ],
     [
       'entryPoints',
-      getSystemConfigInputValue<EntryPointItemType[]>(systemConfigNode, NodeInputKeyEnum.entryPoints)
+      getSystemConfigInputValue<EntryPointItemType[]>(
+        systemConfigNode,
+        NodeInputKeyEnum.entryPoints
+      )
     ]
   ];
   configValues.forEach(([key, value]) => {
diff --git a/packages/global/openapi/core/dataset/api.ts b/packages/global/openapi/core/dataset/api.ts
index caa3c4af1939..153d36b9a4ae 100644
--- a/packages/global/openapi/core/dataset/api.ts
+++ b/packages/global/openapi/core/dataset/api.ts
@@ -8,6 +8,7 @@ import {
   DatasetItemSchema,
   DatasetSchema,
   DatasetListItemSchema,
+  sangforFileParseConfigSchema,
   SearchDataResponseItemSchema
 } from '../../../core/dataset/type';
 import { AppListSortEnum } from '../../../core/app/constants';
@@ -72,6 +73,9 @@ export const CreateDatasetBodySchema = z.object({
   }),
   apiDatasetServer: ApiDatasetServerSchema.optional().meta({
     description: '第三方知识库服务器配置(API/飞书/语雀/钉钉)'
+  }),
+  sangforFileParseConfig: sangforFileParseConfigSchema.optional().meta({
+    description: '外部文档解析开关(页眉页脚/附录/图片识别/图转表),仅对 customPdfParse 解析路径生效'
   })
 });
 
@@ -111,6 +115,10 @@ export const CreateDatasetWithFilesBodySchema = z.object({
       vlmModelId: z.string().nullable().optional().meta({
         description: '视觉语言模型 ID；未传沿用默认，null 或空字符串表示不设置',
         example: ''
+      }),
+      sangforFileParseConfig: sangforFileParseConfigSchema.optional().meta({
+        description:
+          '外部文档解析开关(页眉页脚/附录/图片识别/图转表),仅对 customPdfParse 解析路径生效'
       })
     })
     .meta({ description: '知识库参数' }),
@@ -443,6 +451,10 @@ export const UpdateDatasetBodySchema = z.object({
   }),
   chunkSettings: ChunkSettingsSchema.optional().meta({
     description: '分块配置'
+  }),
+  sangforFileParseConfig: sangforFileParseConfigSchema.optional().meta({
+    description:
+      '外部文档解析开关(页眉页脚/附录/图片识别/图转表),仅对 customPdfParse 解析路径生效;编辑后仅对新解析的文件生效'
   })
 });
 export type UpdateDatasetBody = z.infer<typeof UpdateDatasetBodySchema>;
diff --git a/packages/service/common/file/read/utils.ts b/packages/service/common/file/read/utils.ts
index 50be7d2f6af9..1fb28dd9ae74 100644
--- a/packages/service/common/file/read/utils.ts
+++ b/packages/service/common/file/read/utils.ts
@@ -1,5 +1,6 @@
 import FormData from 'form-data';
 import type { ReadFileResponse } from '../../../worker/readFile/type';
+import type { IultmzhFileParseConfigType } from '@fastgpt/global/core/dataset/type';
 import { axios } from '../../api/axios';
 import { parseMarkdownBase64Images } from '@fastgpt/global/common/string/markdown';
 import { createPdfParseUsage } from '../../../support/wallet/usage/controller';
@@ -7,6 +8,7 @@ import { useDoc2xServer } from '../../../thirdProvider/doc2x';
 import { useTextinServer } from '../../../thirdProvider/textin';
 import { useSomarkServer } from '../../../thirdProvider/somark';
 import { parseFromSangfor, useSangforParse } from '../../../thirdProvider/sangfor';
+import { appendIultmzhFileParseFields } from '../../../thirdProvider/sangfor/parseConfig';
 import { readRawContentFromBuffer, readRawContentFromSource } from '../../../worker/function';
 import { getLogger, LogCategories } from '../../logger';
 import { getImageBuffer } from '../image/utils';
@@ -33,6 +35,7 @@ export const readFileContentByBuffer = async ({
   buffer,
   encoding,
   customPdfParse = false,
+  sangforFileParseConfig,
   usageId,
   getFormatText = true,
   imageKeyOptions,
@@ -46,6 +49,7 @@ export const readFileContentByBuffer = async ({
   encoding: string;
 
   customPdfParse?: boolean;
+  sangforFileParseConfig?: IultmzhFileParseConfigType;
   usageId?: string;
   getFormatText?: boolean;
   imageKeyOptions?: {
@@ -62,6 +66,7 @@ export const readFileContentByBuffer = async ({
     buffer,
     encoding,
     customPdfParse,
+    sangforFileParseConfig,
     usageId,
     getFormatText,
     imageKeyOptions,
@@ -77,6 +82,7 @@ export const readFileContentBySource = async ({
   tmbId,
   source,
   customPdfParse = false,
+  sangforFileParseConfig,
   usageId,
   getFormatText = true,
   imageKeyOptions,
@@ -86,6 +92,7 @@ export const readFileContentBySource = async ({
   tmbId: string;
   source: FileSource;
   customPdfParse?: boolean;
+  sangforFileParseConfig?: IultmzhFileParseConfigType;
   usageId?: string;
   getFormatText?: boolean;
   imageKeyOptions?: {
@@ -101,6 +108,7 @@ export const readFileContentBySource = async ({
     source,
     encoding: source.metadata.encoding ?? '',
     customPdfParse,
+    sangforFileParseConfig,
     usageId,
     getFormatText,
     imageKeyOptions,
@@ -115,6 +123,7 @@ const readFileContent = async ({
   source,
   encoding: initialEncoding,
   customPdfParse,
+  sangforFileParseConfig,
   usageId,
   getFormatText,
   imageKeyOptions,
@@ -127,6 +136,7 @@ const readFileContent = async ({
   source?: FileSource;
   encoding: string;
   customPdfParse: boolean;
+  sangforFileParseConfig?: IultmzhFileParseConfigType;
   usageId?: string;
   getFormatText: boolean;
   imageKeyOptions?: {
@@ -235,6 +245,7 @@ const readFileContent = async ({
     data.append('file', buffer, {
       filename: `file.${materializedExtension}`
     });
+    appendIultmzhFileParseFields(data, sangforFileParseConfig);
     const { data: response } = await axios.post<{
       pages: number;
       markdown: string;
@@ -356,7 +367,8 @@ const readFileContent = async ({
     const { text, pages } = await parseFromSangfor({
       fileBuffer: buffer,
       extension: materializedExtension,
-      imageKeyOptions
+      imageKeyOptions,
+      sangforFileParseConfig
     });
 
     reportPdfParseUsage(pages);
diff --git a/packages/service/common/s3/sources/dataset/index.ts b/packages/service/common/s3/sources/dataset/index.ts
index 049b504f271e..0b2f36c2dae3 100644
--- a/packages/service/common/s3/sources/dataset/index.ts
+++ b/packages/service/common/s3/sources/dataset/index.ts
@@ -140,8 +140,16 @@ export class S3DatasetSource extends S3PrivateBucket {
   }
 
   async getDatasetFileRawText(params: GetDatasetFileContentParams) {
-    const { fileId, teamId, tmbId, customPdfParse, getFormatText, usageId, datasetId } =
-      GetDatasetFileContentParamsSchema.parse(params);
+    const {
+      fileId,
+      teamId,
+      tmbId,
+      customPdfParse,
+      sangforFileParseConfig,
+      getFormatText,
+      usageId,
+      datasetId
+    } = GetDatasetFileContentParamsSchema.parse(params);
 
     if (!isAuthorizedDatasetFileS3Key({ key: fileId, datasetId })) {
       return Promise.reject('Invalid dataset file key');
@@ -149,6 +157,7 @@ export class S3DatasetSource extends S3PrivateBucket {
 
     const rawTextBuffer = await this.rawTextSource.getRawTextBuffer({
       customPdfParse,
+      sangforFileParseConfig,
       sourceId: fileId
     });
     if (rawTextBuffer) {
@@ -166,6 +175,7 @@ export class S3DatasetSource extends S3PrivateBucket {
       tmbId,
       source,
       customPdfParse,
+      sangforFileParseConfig,
       usageId,
       getFormatText,
       imageKeyOptions: {
@@ -177,7 +187,8 @@ export class S3DatasetSource extends S3PrivateBucket {
       sourceId: fileId,
       sourceName: filename,
       text: rawText,
-      customPdfParse
+      customPdfParse,
+      sangforFileParseConfig
     });
 
     return {
diff --git a/packages/service/common/s3/sources/dataset/type.ts b/packages/service/common/s3/sources/dataset/type.ts
index 0b7a89cd9f7d..884554709e71 100644
--- a/packages/service/common/s3/sources/dataset/type.ts
+++ b/packages/service/common/s3/sources/dataset/type.ts
@@ -1,4 +1,5 @@
 import { ObjectIdSchema } from '@fastgpt/global/common/type/mongo';
+import { sangforFileParseConfigSchema } from '@fastgpt/global/core/dataset/type';
 import { ReadStream } from 'fs';
 import z from 'zod';
 
@@ -29,6 +30,7 @@ export const GetDatasetFileContentParamsSchema = z.object({
   tmbId: ObjectIdSchema,
   fileId: z.string().nonempty(), // 这是 ObjectKey
   customPdfParse: z.boolean().optional(),
+  sangforFileParseConfig: sangforFileParseConfigSchema.optional(),
   getFormatText: z.boolean().optional(), // 数据类型都尽可能转化成 markdown 格式
   datasetId: ObjectIdSchema,
   usageId: ObjectIdSchema.optional()
diff --git a/packages/service/common/s3/sources/rawText/index.ts b/packages/service/common/s3/sources/rawText/index.ts
index dfe192fe05f7..6fadd7897adb 100644
--- a/packages/service/common/s3/sources/rawText/index.ts
+++ b/packages/service/common/s3/sources/rawText/index.ts
@@ -25,12 +25,12 @@ export class S3RawTextSource extends S3PrivateBucket {
   }
 
   async addRawTextBuffer(params: AddRawTextBufferParams) {
-    const { sourceId, sourceName, text, customPdfParse } =
+    const { sourceId, sourceName, text, customPdfParse, sangforFileParseConfig } =
       AddRawTextBufferParamsSchema.parse(params);
 
     // 因为 Key 唯一对应一个 Object 所以不需要根据文件内容计算 Hash 直接用 Key 计算 Hash 就行了
     const hash = createHash('md5').update(sourceId).digest('hex');
-    const key = getFileS3Key.rawText({ hash, customPdfParse });
+    const key = getFileS3Key.rawText({ hash, customPdfParse, sangforFileParseConfig });
     const buffer = Buffer.from(text);
 
     await MongoS3TTL.create({
@@ -57,10 +57,10 @@ export class S3RawTextSource extends S3PrivateBucket {
   }
 
   async getRawTextBuffer(params: GetRawTextBufferParams) {
-    const { customPdfParse, sourceId } = params;
+    const { customPdfParse, sangforFileParseConfig, sourceId } = params;
 
     const hash = createHash('md5').update(sourceId).digest('hex');
-    const key = getFileS3Key.rawText({ hash, customPdfParse });
+    const key = getFileS3Key.rawText({ hash, customPdfParse, sangforFileParseConfig });
 
     if (!(await this.isObjectExists(key))) return null;
 
diff --git a/packages/service/common/s3/sources/rawText/type.ts b/packages/service/common/s3/sources/rawText/type.ts
index 4979abcc010b..f6ffb09f2dff 100644
--- a/packages/service/common/s3/sources/rawText/type.ts
+++ b/packages/service/common/s3/sources/rawText/type.ts
@@ -1,10 +1,15 @@
+import { sangforFileParseConfigSchema } from '@fastgpt/global/core/dataset/type';
 import z from 'zod';
 
 export const AddRawTextBufferParamsSchema = z.object({
   customPdfParse: z.boolean().optional(),
+  sangforFileParseConfig: sangforFileParseConfigSchema.optional(),
   sourceId: z.string().nonempty(),
   sourceName: z.string().nonempty(),
   text: z.string()
 });
 export type AddRawTextBufferParams = z.input<typeof AddRawTextBufferParamsSchema>;
-export type GetRawTextBufferParams = Pick<AddRawTextBufferParams, 'customPdfParse' | 'sourceId'>;
+export type GetRawTextBufferParams = Pick<
+  AddRawTextBufferParams,
+  'customPdfParse' | 'sangforFileParseConfig' | 'sourceId'
+>;
diff --git a/packages/service/common/s3/utils.ts b/packages/service/common/s3/utils.ts
index 8aa4f0ec78aa..87ad97ceee9f 100644
--- a/packages/service/common/s3/utils.ts
+++ b/packages/service/common/s3/utils.ts
@@ -1,4 +1,5 @@
 import { isAfter } from 'date-fns';
+import { createHash } from 'node:crypto';
 import type { ClientSession } from 'mongoose';
 import { buffer as consumeStreamToBuffer } from 'node:stream/consumers';
 import type { Readable } from 'node:stream';
@@ -299,9 +300,29 @@ export const getFileS3Key = {
     };
   },
 
-  rawText: ({ hash, customPdfParse }: { hash: string; customPdfParse?: boolean }) => {
+  rawText: ({
+    hash,
+    customPdfParse,
+    sangforFileParseConfig
+  }: {
+    hash: string;
+    customPdfParse?: boolean;
+    sangforFileParseConfig?: Record<string, unknown>;
+  }) => {
+    // 解析配置纳入缓存 key,避免同文件不同开关共享缓存;未传配置保持旧 key 格式
+    const configSuffix = sangforFileParseConfig
+      ? `-${createHash('md5')
+          .update(
+            JSON.stringify(sangforFileParseConfig, (_key, value) =>
+              value instanceof Object && !Array.isArray(value)
+                ? Object.fromEntries(Object.entries(value).sort(([a], [b]) => a.localeCompare(b)))
+                : value
+            )
+          )
+          .digest('hex')}`
+      : '';
     return encodeS3ObjectKey(
-      [S3Sources.rawText, `${hash}${customPdfParse ? '-true' : ''}`].join('/')
+      [S3Sources.rawText, `${hash}${customPdfParse ? '-true' : ''}${configSuffix}`].join('/')
     );
   }
 };
diff --git a/packages/service/core/dataset/apiDataset/custom/api.ts b/packages/service/core/dataset/apiDataset/custom/api.ts
index 7ce94d91977b..c1ce7a701db2 100644
--- a/packages/service/core/dataset/apiDataset/custom/api.ts
+++ b/packages/service/core/dataset/apiDataset/custom/api.ts
@@ -4,6 +4,7 @@ import type {
   ApiDatasetDetailResponse,
   APIFileServerType
 } from '@fastgpt/global/core/dataset/apiDataset/type';
+import type { IultmzhFileParseConfigType } from '@fastgpt/global/core/dataset/type';
 import { type Method } from 'axios';
 import { createProxyAxios } from '../../../../common/api/axios';
 import { readFileRawTextByUrl } from '../../read';
@@ -170,6 +171,7 @@ export const useApiDatasetRequest = ({ apiServer }: { apiServer: APIFileServerTy
     tmbId,
     apiFileId,
     customPdfParse,
+    sangforFileParseConfig,
     datasetId,
     usageId
   }: {
@@ -177,6 +179,7 @@ export const useApiDatasetRequest = ({ apiServer }: { apiServer: APIFileServerTy
     tmbId: string;
     apiFileId: string;
     customPdfParse?: boolean;
+    sangforFileParseConfig?: IultmzhFileParseConfigType;
     datasetId: string;
     usageId?: string;
   }): Promise<ApiFileReadContentResponseType> => {
@@ -204,7 +207,8 @@ export const useApiDatasetRequest = ({ apiServer }: { apiServer: APIFileServerTy
       // Get from buffer
       const rawTextBuffer = await getS3RawTextSource().getRawTextBuffer({
         sourceId: previewUrl,
-        customPdfParse
+        customPdfParse,
+        sangforFileParseConfig
       });
       if (rawTextBuffer) {
         return {
@@ -220,6 +224,7 @@ export const useApiDatasetRequest = ({ apiServer }: { apiServer: APIFileServerTy
         relatedId: apiFileId,
         datasetId,
         customPdfParse,
+        sangforFileParseConfig,
         usageId,
         getFormatText: true
       });
@@ -230,7 +235,8 @@ export const useApiDatasetRequest = ({ apiServer }: { apiServer: APIFileServerTy
         sourceId: previewUrl,
         sourceName,
         text: rawText,
-        customPdfParse
+        customPdfParse,
+        sangforFileParseConfig
       });
 
       return {
diff --git a/packages/service/core/dataset/read.ts b/packages/service/core/dataset/read.ts
index f887eaf51234..ea2e30ed3143 100644
--- a/packages/service/core/dataset/read.ts
+++ b/packages/service/core/dataset/read.ts
@@ -4,6 +4,7 @@ import {
   DatasetCollectionDataProcessModeEnum,
   DatasetSourceReadTypeEnum
 } from '@fastgpt/global/core/dataset/constants';
+import type { IultmzhFileParseConfigType } from '@fastgpt/global/core/dataset/type';
 import { urlsFetch } from '../../common/string/cheerio';
 import { type TextSplitProps } from '../../common/string/textSplitter';
 import { readFileContentBySource } from '../../common/file/read/utils';
@@ -58,6 +59,7 @@ export const readFileRawTextByUrl = async ({
   tmbId,
   url,
   customPdfParse,
+  sangforFileParseConfig,
   getFormatText,
   datasetId,
   usageId,
@@ -67,6 +69,7 @@ export const readFileRawTextByUrl = async ({
   tmbId: string;
   url: string;
   customPdfParse?: boolean;
+  sangforFileParseConfig?: IultmzhFileParseConfigType;
   getFormatText?: boolean;
   relatedId: string; // externalFileId / apiFileId
   datasetId: string;
@@ -94,6 +97,7 @@ export const readFileRawTextByUrl = async ({
   const { rawText } = await retryFn(() =>
     readFileContentBySource({
       customPdfParse,
+      sangforFileParseConfig,
       usageId,
       getFormatText,
       source,
@@ -123,6 +127,7 @@ export const readDatasetSourceRawText = async ({
   externalFileId,
   apiDatasetServer,
   customPdfParse,
+  sangforFileParseConfig,
   getFormatText,
   usageId,
   datasetId
@@ -132,6 +137,8 @@ export const readDatasetSourceRawText = async ({
   type: DatasetSourceReadTypeEnum;
   sourceId: string;
   customPdfParse?: boolean;
+  /** 外部文档解析开关;建议传入 getDatasetIultmzhFileParseConfig(dataset) 补全后的完整配置 */
+  sangforFileParseConfig?: IultmzhFileParseConfigType;
   getFormatText?: boolean;
 
   selector?: string; // link selector
@@ -158,6 +165,7 @@ export const readDatasetSourceRawText = async ({
       fileId: sourceId,
       getFormatText,
       customPdfParse,
+      sangforFileParseConfig,
       usageId,
       datasetId
     });
@@ -190,6 +198,7 @@ export const readDatasetSourceRawText = async ({
       relatedId: externalFileId,
       datasetId,
       customPdfParse,
+      sangforFileParseConfig,
       usageId
     });
     return {
@@ -202,6 +211,7 @@ export const readDatasetSourceRawText = async ({
       teamId,
       tmbId,
       customPdfParse,
+      sangforFileParseConfig,
       datasetId,
       usageId
     });
@@ -222,6 +232,7 @@ export const readApiServerFileContent = async ({
   teamId,
   tmbId,
   customPdfParse,
+  sangforFileParseConfig,
   datasetId,
   usageId
 }: {
@@ -230,6 +241,7 @@ export const readApiServerFileContent = async ({
   teamId: string;
   tmbId: string;
   customPdfParse?: boolean;
+  sangforFileParseConfig?: IultmzhFileParseConfigType;
   datasetId: string;
   usageId?: string;
 }): Promise<{
@@ -241,6 +253,7 @@ export const readApiServerFileContent = async ({
     tmbId,
     apiFileId,
     customPdfParse,
+    sangforFileParseConfig,
     datasetId,
     usageId
   });
diff --git a/packages/service/core/dataset/schema.ts b/packages/service/core/dataset/schema.ts
index 218e03b22355..1d8727669c6b 100644
--- a/packages/service/core/dataset/schema.ts
+++ b/packages/service/core/dataset/schema.ts
@@ -131,6 +131,15 @@ const DatasetSchema = new Schema({
   chunkSettings: {
     type: ChunkSettings
   },
+  // 外部文档解析服务开关,整体存取;缺失字段由读取层用固定默认值补全
+  sangforFileParseConfig: {
+    type: {
+      keep_header_footer: Boolean,
+      keep_appendix: Boolean,
+      image_analysis: Boolean,
+      chart_analysis: Boolean
+    }
+  },
   inheritPermission: {
     type: Boolean,
     default: true
diff --git a/packages/service/test/common/s3/utils.test.ts b/packages/service/test/common/s3/utils.test.ts
index 80c8178c30c7..315bea625cf9 100644
--- a/packages/service/test/common/s3/utils.test.ts
+++ b/packages/service/test/common/s3/utils.test.ts
@@ -873,3 +873,47 @@ describe('getFileS3Key', () => {
     });
   });
 });
+
+describe('getFileS3Key.rawText', () => {
+  const hash = 'abc123def456';
+  const sangforFileParseConfig = {
+    keep_header_footer: true,
+    keep_appendix: false,
+    image_analysis: true,
+    chart_analysis: false
+  };
+
+  it('keeps the legacy key format when no parse config is given', () => {
+    const noFlag = getFileS3Key.rawText({ hash });
+    expect(noFlag).toBe(getFileS3Key.rawText({ hash, customPdfParse: false }));
+    expect(noFlag).toContain(hash);
+    expect(getFileS3Key.rawText({ hash, customPdfParse: true })).toContain(`${hash}-true`);
+  });
+
+  it('is stable for the same config regardless of key order', () => {
+    const reordered = {
+      chart_analysis: false,
+      image_analysis: true,
+      keep_appendix: false,
+      keep_header_footer: true
+    };
+
+    expect(getFileS3Key.rawText({ hash, customPdfParse: true, sangforFileParseConfig })).toBe(
+      getFileS3Key.rawText({ hash, customPdfParse: true, sangforFileParseConfig: reordered })
+    );
+  });
+
+  it('changes the key when config values differ', () => {
+    const flipped = { ...sangforFileParseConfig, keep_header_footer: false };
+
+    expect(getFileS3Key.rawText({ hash, customPdfParse: true, sangforFileParseConfig })).not.toBe(
+      getFileS3Key.rawText({ hash, customPdfParse: true, sangforFileParseConfig: flipped })
+    );
+  });
+
+  it('differs from the legacy key once a config is attached', () => {
+    expect(getFileS3Key.rawText({ hash, customPdfParse: true, sangforFileParseConfig })).not.toBe(
+      getFileS3Key.rawText({ hash, customPdfParse: true })
+    );
+  });
+});
diff --git a/packages/service/test/thirdProvider/sangfor.test.ts b/packages/service/test/thirdProvider/sangfor.test.ts
index af16b07557b3..acfbf3489036 100644
--- a/packages/service/test/thirdProvider/sangfor.test.ts
+++ b/packages/service/test/thirdProvider/sangfor.test.ts
@@ -1,4 +1,5 @@
 import FormData from 'form-data';
+import { getDatasetIultmzhFileParseConfig } from '@fastgpt/service/thirdProvider/sangfor/parseConfig';
 import { afterEach, beforeEach, describe, expect, it, vi } from 'vitest';
 
 const { postMock, parseMarkdownImagesMock, uploadParsedPdfImageMock, mockEnv } = vi.hoisted(() => ({
@@ -208,3 +209,29 @@ describe('Sangfor provider', () => {
     expect(postMock).not.toHaveBeenCalled();
   });
 });
+
+describe('getDatasetIultmzhFileParseConfig', () => {
+  const allDisabled = {
+    keep_header_footer: false,
+    keep_appendix: false,
+    image_analysis: false,
+    chart_analysis: false
+  };
+
+  it('为空数据集/空配置补全四布尔默认值', () => {
+    expect(getDatasetIultmzhFileParseConfig(undefined)).toEqual(allDisabled);
+    expect(getDatasetIultmzhFileParseConfig(null)).toEqual(allDisabled);
+    expect(getDatasetIultmzhFileParseConfig({})).toEqual(allDisabled);
+  });
+
+  it('保留已配置字段,仅补全缺失字段', () => {
+    expect(
+      getDatasetIultmzhFileParseConfig({ sangforFileParseConfig: { keep_header_footer: true } })
+    ).toEqual({
+      keep_header_footer: true,
+      keep_appendix: false,
+      image_analysis: false,
+      chart_analysis: false
+    });
+  });
+});
diff --git a/packages/service/thirdProvider/sangfor/index.ts b/packages/service/thirdProvider/sangfor/index.ts
index 051cf191a0ff..7c09fb19419d 100644
--- a/packages/service/thirdProvider/sangfor/index.ts
+++ b/packages/service/thirdProvider/sangfor/index.ts
@@ -1,11 +1,13 @@
 import FormData from 'form-data';
 import { getErrText } from '@fastgpt/global/common/error/utils';
 import { parseMarkdownBase64Images } from '@fastgpt/global/common/string/markdown';
+import type { IultmzhFileParseConfigType } from '@fastgpt/global/core/dataset/type';
 import z from 'zod';
 import { axios } from '../../common/api/axios';
 import { getImageBuffer } from '../../common/file/image/utils';
 import { uploadParsedPdfImage, type ParsedPdfImageKeyOptions } from '../../common/file/read/image';
 import { serviceEnv } from '../../env';
+import { appendIultmzhFileParseFields } from './parseConfig';
 
 const SangforParseResponseSchema = z.object({
   pages: z.number().int().nonnegative(),
@@ -31,11 +33,13 @@ export const useSangforParse = (extension: string): boolean => {
 export const parseFromSangfor = async ({
   fileBuffer,
   extension,
-  imageKeyOptions
+  imageKeyOptions,
+  sangforFileParseConfig
 }: {
   fileBuffer: Buffer;
   extension: string;
   imageKeyOptions?: ParsedPdfImageKeyOptions;
+  sangforFileParseConfig?: IultmzhFileParseConfigType;
 }) => {
   const { url, key } = global.systemEnv?.customPdfParse ?? {};
   if (!url) {
@@ -45,6 +49,7 @@ export const parseFromSangfor = async ({
   try {
     const form = new FormData();
     form.append('file', fileBuffer, { filename: `file.${extension}` });
+    appendIultmzhFileParseFields(form, sangforFileParseConfig);
     const { data } = await axios.post<unknown>(url, form, {
       timeout: serviceEnv.SANGFOR_PARSE_TIMEOUT_SECONDS * 1000,
       headers: {
diff --git a/packages/service/thirdProvider/sangfor/parseConfig.ts b/packages/service/thirdProvider/sangfor/parseConfig.ts
new file mode 100644
index 000000000000..16859b776e74
--- /dev/null
+++ b/packages/service/thirdProvider/sangfor/parseConfig.ts
@@ -0,0 +1,31 @@
+import type FormData from 'form-data';
+import type {
+  DatasetSchemaType,
+  IultmzhFileParseConfigType
+} from '@fastgpt/global/core/dataset/type';
+
+/**
+ * 从 dataset 取 sangfor 文件解析配置,缺失字段用固定默认值补全为完整四个 boolean,
+ * 避免不同解析引擎对缺失字段的不同兜底行为。
+ */
+export const getDatasetIultmzhFileParseConfig = (
+  dataset?: Pick<DatasetSchemaType, 'sangforFileParseConfig'> | null
+): IultmzhFileParseConfigType => ({
+  keep_header_footer: dataset?.sangforFileParseConfig?.keep_header_footer ?? false,
+  keep_appendix: dataset?.sangforFileParseConfig?.keep_appendix ?? false,
+  image_analysis: dataset?.sangforFileParseConfig?.image_analysis ?? false,
+  chart_analysis: dataset?.sangforFileParseConfig?.chart_analysis ?? false
+});
+
+// 外部解析服务契约:四个开关逐个以独立表单字段下发,值统一转字符串(服务端自行 str→bool)
+export const appendIultmzhFileParseFields = (
+  form: FormData,
+  config?: IultmzhFileParseConfigType
+) => {
+  if (!config) return;
+  Object.entries(config).forEach(([fieldKey, value]) => {
+    if (value !== undefined && value !== null) {
+      form.append(fieldKey, String(value));
+    }
+  });
+};
diff --git a/projects/app/src/pages/api/core/dataset/create.ts b/projects/app/src/pages/api/core/dataset/create.ts
index 15eee3d8f3f2..155231df9864 100644
--- a/projects/app/src/pages/api/core/dataset/create.ts
+++ b/projects/app/src/pages/api/core/dataset/create.ts
@@ -40,7 +40,8 @@ async function handler(req: ApiRequestProps): Promise<CreateDatasetResponse> {
     agentModel,
     vlmModelId,
     vlmModel,
-    apiDatasetServer
+    apiDatasetServer,
+    sangforFileParseConfig
   } = parseApiInput({ req, bodySchema: CreateDatasetBodySchema }).body;
 
   // auth
@@ -95,7 +96,8 @@ async function handler(req: ApiRequestProps): Promise<CreateDatasetResponse> {
           ...(vlmModelStore?.modelId && { vlmModelId: vlmModelStore.modelId }),
           avatar,
           type,
-          apiDatasetServer
+          apiDatasetServer,
+          ...(sangforFileParseConfig && { sangforFileParseConfig })
         }
       ],
       { session, ordered: true }
diff --git a/projects/app/src/pages/api/core/dataset/createWithFiles.ts b/projects/app/src/pages/api/core/dataset/createWithFiles.ts
index f154f9c0b2e1..535e819b596e 100644
--- a/projects/app/src/pages/api/core/dataset/createWithFiles.ts
+++ b/projects/app/src/pages/api/core/dataset/createWithFiles.ts
@@ -42,7 +42,15 @@ async function handler(req: ApiRequestProps): Promise<CreateDatasetWithFilesResp
     req,
     bodySchema: CreateDatasetWithFilesBodySchema
   }).body;
-  const { parentId, name, avatar, vectorModelId, agentModelId, vlmModelId } = datasetParams;
+  const {
+    parentId,
+    name,
+    avatar,
+    vectorModelId,
+    agentModelId,
+    vlmModelId,
+    sangforFileParseConfig
+  } = datasetParams;
   const modelHandle = await getModelHandle();
   const vectorModelData =
     modelHandle.getEmbeddingModelData({ modelId: vectorModelId }, { optional: true }) ??
@@ -89,7 +97,8 @@ async function handler(req: ApiRequestProps): Promise<CreateDatasetWithFilesResp
             ...(vlmModelData?.modelId && { vlmModelId: vlmModelData.modelId }),
             avatar,
             intro: '',
-            type: DatasetTypeEnum.dataset
+            type: DatasetTypeEnum.dataset,
+            ...(sangforFileParseConfig && { sangforFileParseConfig })
           }
         ],
         { session, ordered: true }
diff --git a/projects/app/src/pages/api/core/dataset/file/getPreviewChunks.ts b/projects/app/src/pages/api/core/dataset/file/getPreviewChunks.ts
index f0ef0977f442..be062ead6468 100644
--- a/projects/app/src/pages/api/core/dataset/file/getPreviewChunks.ts
+++ b/projects/app/src/pages/api/core/dataset/file/getPreviewChunks.ts
@@ -1,6 +1,7 @@
 import { getModelHandle } from '@fastgpt/service/core/ai/model';
 import { getDatasetModelReference } from '@fastgpt/service/core/dataset/model';
 import { DatasetSourceReadTypeEnum } from '@fastgpt/global/core/dataset/constants';
+import { getDatasetIultmzhFileParseConfig } from '@fastgpt/service/thirdProvider/sangfor/parseConfig';
 import { rawText2Chunks, readDatasetSourceRawText } from '@fastgpt/service/core/dataset/read';
 import { NextAPI } from '@/service/middleware/entry';
 import type { ApiRequestProps } from '@fastgpt/next/type';
@@ -87,6 +88,8 @@ async function handler(
     selector,
     externalFileId,
     customPdfParse,
+    // 解析开关仅对外部解析路径生效;非 customPdfParse 时传 undefined,rawText 缓存沿用旧 key
+    sangforFileParseConfig: customPdfParse ? getDatasetIultmzhFileParseConfig(dataset) : undefined,
     apiDatasetServer: dataset.apiDatasetServer,
     datasetId
   });
diff --git a/projects/app/src/pages/api/core/dataset/update.ts b/projects/app/src/pages/api/core/dataset/update.ts
index 1933c7f80bc1..0760b20dc4d1 100644
--- a/projects/app/src/pages/api/core/dataset/update.ts
+++ b/projects/app/src/pages/api/core/dataset/update.ts
@@ -70,6 +70,7 @@ async function handler(req: ApiRequestProps<UpdateDatasetBody>) {
       externalReadUrl,
       apiDatasetServer,
       autoSync,
+      sangforFileParseConfig,
       chunkSettings: rawChunkSettings
     }
   } = parseApiInput({
@@ -253,6 +254,8 @@ async function handler(req: ApiRequestProps<UpdateDatasetBody>) {
         ...(externalReadUrl !== undefined && { externalReadUrl }),
         ...(isMove && { inheritPermission: true }),
         ...(typeof autoSync === 'boolean' && { autoSync }),
+        // 传空对象等价于恢复全部开关默认值(读取层补全),旧文件已固化的解析结果不受影响
+        ...(sangforFileParseConfig !== undefined && { sangforFileParseConfig }),
         ...apiDatasetParams,
         ...(!isMove && { updateTime: new Date() })
       },
diff --git a/projects/app/src/service/core/dataset/queues/datasetParse.ts b/projects/app/src/service/core/dataset/queues/datasetParse.ts
index d9598ed526db..6386dd88e65c 100644
--- a/projects/app/src/service/core/dataset/queues/datasetParse.ts
+++ b/projects/app/src/service/core/dataset/queues/datasetParse.ts
@@ -20,6 +20,7 @@ import { addMinutes } from 'date-fns';
 import { checkTeamAiPointsAndLock } from './utils';
 import { getErrText } from '@fastgpt/global/common/error/utils';
 import { delay } from '@fastgpt/global/common/system/utils';
+import { getDatasetIultmzhFileParseConfig } from '@fastgpt/service/thirdProvider/sangfor/parseConfig';
 import { rawText2Chunks, readDatasetSourceRawText } from '@fastgpt/service/core/dataset/read';
 import { getLLMMaxChunkSize } from '@fastgpt/global/core/dataset/training/utils';
 import { checkDatasetIndexLimit } from '@fastgpt/service/support/permission/teamLimit';
@@ -323,6 +324,10 @@ export const datasetParseQueue = async (): Promise<any> => {
           teamId: data.teamId,
           tmbId: data.tmbId,
           customPdfParse: collection.customPdfParse,
+          // 解析开关仅对外部解析路径生效;非 customPdfParse 时传 undefined,rawText 缓存沿用旧 key
+          sangforFileParseConfig: collection.customPdfParse
+            ? getDatasetIultmzhFileParseConfig(dataset)
+            : undefined,
           usageId: data.billId,
           datasetId: data.datasetId,
           ...sourceReadType