feat: support manual file vectorization

backend/docs/swagger.yaml

@@ -3139,6 +3139,40 @@ definitions:
     - data
     - errorMsg
     type: object
+  FileEmbeddingSkipResponse:
+    properties:
+      fileID:
+        type: string
+      reason:
+        type: string
+    required:
+    - fileID
+    - reason
+    type: object
+  FileEmbeddingSubmissionResponse:
+    properties:
+      skipped:
+        items:
+          $ref: '#/definitions/FileEmbeddingSkipResponse'
+        type: array
+      submittedFileIDs:
+        items:
+          type: string
+        type: array
+    required:
+    - skipped
+    - submittedFileIDs
+    type: object
+  FileEmbeddingSubmissionResponseDoc:
+    properties:
+      data:
+        $ref: '#/definitions/FileEmbeddingSubmissionResponse'
+      errorMsg:
+        type: string
+    required:
+    - data
+    - errorMsg
+    type: object
   FileListResponse:
     properties:
       quota:
@@ -3166,6 +3200,8 @@ definitions:
     type: object
   FileObjectResponse:
     properties:
+      canVectorize:
+        type: boolean
       chunkCount:
         type: integer
       createdAt:
@@ -3214,7 +3250,10 @@ definitions:
         type: string
       updatedAt:
         type: string
+      vectorizationReason:
+        type: string
     required:
+    - canVectorize
     - chunkCount
     - createdAt
     - detectedMIME
@@ -3237,9 +3276,12 @@ definitions:
     - sizeBytes
     - status
     - updatedAt
+    - vectorizationReason
     type: object
   FileProcessingStatusResponse:
     properties:
+      canVectorize:
+        type: boolean
       chunkCount:
         type: integer
       completedAt:
@@ -3284,7 +3326,10 @@ definitions:
         x-omitempty: false
       updatedAt:
         type: string
+      vectorizationReason:
+        type: string
     required:
+    - canVectorize
     - chunkCount
     - completedAt
     - detectedMIME
@@ -3305,6 +3350,7 @@ definitions:
     - ragReason
     - startedAt
     - updatedAt
+    - vectorizationReason
     type: object
   FileUpdateResponseDoc:
     properties:
@@ -3758,6 +3804,40 @@ definitions:
     required:
     - file
     type: object
+  KnowledgeBaseFileEmbeddingSkipResponse:
+    properties:
+      fileID:
+        type: string
+      reason:
+        type: string
+    required:
+    - fileID
+    - reason
+    type: object
+  KnowledgeBaseFileEmbeddingSubmissionResponse:
+    properties:
+      skipped:
+        items:
+          $ref: '#/definitions/KnowledgeBaseFileEmbeddingSkipResponse'
+        type: array
+      submittedFileIDs:
+        items:
+          type: string
+        type: array
+    required:
+    - skipped
+    - submittedFileIDs
+    type: object
+  KnowledgeBaseFileEmbeddingSubmissionResponseDoc:
+    properties:
+      data:
+        $ref: '#/definitions/KnowledgeBaseFileEmbeddingSubmissionResponse'
+      errorMsg:
+        type: string
+    required:
+    - data
+    - errorMsg
+    type: object
   KnowledgeBaseFileMutationDataResponse:
     properties:
       updated:
@@ -3809,12 +3889,18 @@ definitions:
     type: object
   KnowledgeBaseFileProcessingStatusResponse:
     properties:
+      canVectorize:
+        type: boolean
       chunkCount:
         type: integer
       detectedMIME:
         type: string
+      embedError:
+        type: string
       embedStatus:
         type: string
+      extractStatus:
+        type: string
       fileCategory:
         type: string
       fileID:
@@ -3829,28 +3915,40 @@ definitions:
         type: boolean
       updatedAt:
         type: string
+      vectorizationReason:
+        type: string
     required:
+    - canVectorize
     - chunkCount
     - detectedMIME
+    - embedError
     - embedStatus
+    - extractStatus
     - fileCategory
     - fileID
     - processing
     - processingReady
     - processingStatus
     - ragOptOut
     - updatedAt
+    - vectorizationReason
     type: object
   KnowledgeBaseFileResponse:
     properties:
+      canVectorize:
+        type: boolean
       chunkCount:
         type: integer
       createdAt:
         type: string
       detectedMIME:
         type: string
+      embedError:
+        type: string
       embedStatus:
         type: string
+      extractStatus:
+        type: string
       fileCategory:
         type: string
       fileID:
@@ -3871,11 +3969,16 @@ definitions:
         type: integer
       updatedAt:
         type: string
+      vectorizationReason:
+        type: string
     required:
+    - canVectorize
     - chunkCount
     - createdAt
     - detectedMIME
+    - embedError
     - embedStatus
+    - extractStatus
     - fileCategory
     - fileID
     - fileName
@@ -3886,6 +3989,7 @@ definitions:
     - ragOptOut
     - sizeBytes
     - updatedAt
+    - vectorizationReason
     type: object
   KnowledgeBaseFileResponseDoc:
     properties:
@@ -7459,6 +7563,28 @@ definitions:
     - usedBytes
     - userID
     type: object
+  SubmitFileEmbeddingsRequest:
+    properties:
+      fileIDs:
+        items:
+          type: string
+        maxItems: 100
+        minItems: 1
+        type: array
+    required:
+    - fileIDs
+    type: object
+  SubmitPlatformFileEmbeddingsRequest:
+    properties:
+      fileIDs:
+        items:
+          type: string
+        maxItems: 100
+        minItems: 1
+        type: array
+    required:
+    - fileIDs
+    type: object
   SubscribeRequest:
     properties:
       cycles:
@@ -11387,6 +11513,42 @@ paths:
       summary: 获取平台资料内容
       tags:
       - admin-knowledge-bases
+  /admin/knowledge-bases/files/embeddings:
+    post:
+      consumes:
+      - application/json
+      description: 为管理员选中的平台资料提交向量化任务，最多100个；重复提交会幂等跳过
+      parameters:
+      - description: 平台资料ID，最多100个
+        in: body
+        name: request
+        required: true
+        schema:
+          $ref: '#/definitions/SubmitPlatformFileEmbeddingsRequest'
+      produces:
+      - application/json
+      responses:
+        "200":
+          description: OK
+          schema:
+            $ref: '#/definitions/KnowledgeBaseFileEmbeddingSubmissionResponseDoc'
+        "400":
+          description: Bad Request
+          schema:
+            $ref: '#/definitions/KnowledgebaseErrorDoc'
+        "500":
+          description: Internal Server Error
+          schema:
+            $ref: '#/definitions/KnowledgebaseErrorDoc'
+        "503":
+          description: Service Unavailable
+          schema:
+            $ref: '#/definitions/KnowledgebaseErrorDoc'
+      security:
+      - BearerAuth: []
+      summary: 批量提交平台资料向量化
+      tags:
+      - admin-knowledge-bases
   /admin/llm/icon-assets:
     get:
       description: 分页查询仍在图标库中的已上传图标，按上传时间倒序返回
@@ -17296,6 +17458,42 @@ paths:
       summary: 获取文件内容
       tags:
       - chat
+  /files/embeddings:
+    post:
+      consumes:
+      - application/json
+      description: 为当前用户已完成文本提取的文件提交向量化任务，最多100个；重复提交会幂等跳过
+      parameters:
+      - description: 文件ID，最多100个
+        in: body
+        name: request
+        required: true
+        schema:
+          $ref: '#/definitions/SubmitFileEmbeddingsRequest'
+      produces:
+      - application/json
+      responses:
+        "200":
+          description: OK
+          schema:
+            $ref: '#/definitions/FileEmbeddingSubmissionResponseDoc'
+        "400":
+          description: Bad Request
+          schema:
+            $ref: '#/definitions/ConversationErrorDoc'
+        "500":
+          description: Internal Server Error
+          schema:
+            $ref: '#/definitions/ConversationErrorDoc'
+        "503":
+          description: Service Unavailable
+          schema:
+            $ref: '#/definitions/ConversationErrorDoc'
+      security:
+      - BearerAuth: []
+      summary: 批量提交指定文件向量化
+      tags:
+      - chat
   /files/processing/statuses:
     post:
       consumes:

backend/internal/app/app.go

@@ -412,6 +412,7 @@ func NewApp() (*App, error) {
 	knowledgeBaseService.SetFileCleaner(conversationService)
 	knowledgeBaseService.SetFileContentOpener(conversationService)
 	knowledgeBaseService.SetFileUploader(conversationService)
+	knowledgeBaseService.SetFileEmbeddingSubmitter(processingService)
 	knowledgeBaseService.SetLogger(log)
 	conversationService.SetKnowledgeBaseResolver(knowledgeBaseService)
 	knowledgeBaseHandler := knowledgebasehttp.NewHandler(knowledgeBaseService, runtimeCfg)

backend/internal/application/conversation/service_file.go

@@ -5,6 +5,7 @@ import (
 	"errors"
 	"strings"
 
+	appembedding "github.com/DEEIX-AI/DEEIX-Chat/backend/internal/application/embedding"
 	appupload "github.com/DEEIX-AI/DEEIX-Chat/backend/internal/application/upload"
 	model "github.com/DEEIX-AI/DEEIX-Chat/backend/internal/domain/conversation"
 	"github.com/DEEIX-AI/DEEIX-Chat/backend/internal/repository"
@@ -52,6 +53,25 @@ func (s *Service) ListFiles(
 	return s.uploadSvc.ListFiles(ctx, userID, page, pageSize, searchQuery, filterKind, sortBy)
 }
 
+// SubmitFileEmbeddings 提交当前用户指定文件的向量化任务。
+func (s *Service) SubmitFileEmbeddings(ctx context.Context, userID uint, fileIDs []string) (appembedding.TargetedSubmissionResult, error) {
+	if s.processingSvc == nil {
+		return appembedding.TargetedSubmissionResult{}, appembedding.ErrEmbeddingServiceNotConfigured
+	}
+	return s.processingSvc.SubmitFileEmbeddings(ctx, userID, fileIDs)
+}
+
+// ResolveFileVectorizationCapabilities 返回文件显式向量化能力的后端事实状态。
+func (s *Service) ResolveFileVectorizationCapabilities(
+	ctx context.Context,
+	files []model.FileObject,
+) map[string]appembedding.FileVectorizationCapability {
+	if s.processingSvc == nil {
+		return map[string]appembedding.FileVectorizationCapability{}
+	}
+	return s.processingSvc.ResolveFileVectorizationCapabilities(ctx, files)
+}
+
 // UploadFile 上传文件并扣减用户配额。
 func (s *Service) UploadFile(ctx context.Context, input appupload.UploadFileInput) (*appupload.UploadFileResult, error) {
 	return s.uploadSvc.UploadFile(ctx, input)