Remove selectUploadParser: documents now inherit parser_id from KB directly (#17083)

internal/ingestion/pipeline/builtin_registry.go

@@ -8,7 +8,6 @@ import (
 	"io/fs"
 	"os"
 	"path/filepath"
-	"regexp"
 	"sort"
 	"strings"
 	"sync"
@@ -188,39 +187,6 @@ func (r *Registry) IsValid(ref string) bool {
 	return ok
 }
 
-// fileTypeParserOverrides maps a file type value to the canonical parser_id that
-// must be used for all files of that type. The type values are FileType constants
-// defined in the utility package.
-var fileTypeParserOverrides = map[string]string{
-	"visual": "picture",
-	"aural":  "audio",
-}
-
-var (
-	presentationExtPattern = regexp.MustCompile(`(?i)\.(ppt|pptx|pages)$`)
-	emailExtPattern        = regexp.MustCompile(`(?i)\.(msg|eml)$`)
-)
-
-// DefaultParserID returns the canonical parser_id for a document given its file
-// type, filename and the dataset-level parser_id. File-type-based overrides
-// (e.g. "visual" → "picture", "aural" → "audio") take precedence, followed by
-// filename extension heuristics (presentation / email), then falling back to
-// the dataset's parser_id.
-func (r *Registry) DefaultParserID(fileType, filename, fallback string) string {
-	if override, ok := fileTypeParserOverrides[strings.ToLower(fileType)]; ok {
-		return override
-	}
-	base := filepath.Base(strings.TrimSpace(filename))
-	switch {
-	case presentationExtPattern.MatchString(base):
-		return "presentation"
-	case emailExtPattern.MatchString(base):
-		return "email"
-	default:
-		return fallback
-	}
-}
-
 func parseTemplate(filename string, raw []byte) (*BuiltinPipeline, error) {
 	decoder := json.NewDecoder(bytes.NewReader(raw))
 	decoder.UseNumber()

internal/ingestion/pipeline/builtin_registry_test.go

@@ -241,37 +241,3 @@ func TestLoadBuiltinDSL_UnknownFails(t *testing.T) {
 		t.Fatal("LoadBuiltinDSL(nonexistent) = nil, want error")
 	}
 }
-
-func TestDefaultRegistry_DefaultParserID(t *testing.T) {
-	r, err := DefaultRegistry()
-	if err != nil {
-		t.Fatalf("DefaultRegistry: %v", err)
-	}
-
-	tests := []struct {
-		name     string
-		fileType string
-		filename string
-		fallback string
-		want     string
-	}{
-		{name: "visual", fileType: "visual", filename: "img.png", fallback: "naive", want: "picture"},
-		{name: "aural", fileType: "aural", filename: "audio.mp3", fallback: "naive", want: "audio"},
-		{name: "presentation by ext", fileType: "doc", filename: "deck.pptx", fallback: "naive", want: "presentation"},
-		{name: "email by ext", fileType: "doc", filename: "mail.eml", fallback: "naive", want: "email"},
-		{name: "pages as presentation", fileType: "doc", filename: "slides.pages", fallback: "naive", want: "presentation"},
-		{name: "msg as email", fileType: "doc", filename: "message.msg", fallback: "naive", want: "email"},
-		{name: "fallback default", fileType: "doc", filename: "notes.txt", fallback: "manual", want: "manual"},
-		{name: "unknown filetype fallback", fileType: "unknown", filename: "file.bin", fallback: "general", want: "general"},
-	}
-
-	for _, tt := range tests {
-		t.Run(tt.name, func(t *testing.T) {
-			got := r.DefaultParserID(tt.fileType, tt.filename, tt.fallback)
-			if got != tt.want {
-				t.Fatalf("DefaultParserID(%q, %q, %q) = %q, want %q",
-					tt.fileType, tt.filename, tt.fallback, got, tt.want)
-			}
-		})
-	}
-}

internal/service/document/document.go

@@ -63,7 +63,6 @@ func NewDocumentService() *DocumentService {
 	}
 }
 
-// CreateDocumentRequest create document request
 // UpdateDocumentRequest update document request
 type UpdateDocumentRequest struct {
 	Name        *string  `json:"name"`

internal/service/document/document_crud.go

@@ -100,7 +100,6 @@ func (s *DocumentService) DownloadDocument(datasetID, docID string) (*DownloadDo
 	}, nil
 }
 
-// CreateDocument create document
 // GetDocumentByID get document by ID
 func (s *DocumentService) GetDocumentByID(id string) (*DocumentResponse, error) {
 	document, err := s.documentDAO.GetByID(id)

internal/service/document/document_test.go

@@ -40,7 +40,6 @@ import (
 	"ragflow/internal/service"
 	"ragflow/internal/service/file"
 	"ragflow/internal/storage"
-	"ragflow/internal/utility"
 )
 
 // recordingTaskPublisher implements service.TaskPublisher and records published messages.
@@ -711,29 +710,6 @@ func TestDeleteDocumentFull_SharedFilePreserved(t *testing.T) {
 	}
 }
 
-func TestSelectUploadParser_MirrorsPython(t *testing.T) {
-	tests := []struct {
-		name         string
-		docType      utility.FileType
-		filename     string
-		defaultValue string
-		want         string
-	}{
-		{name: "visual", docType: utility.FileTypeVISUAL, filename: "img.png", defaultValue: "naive", want: "picture"},
-		{name: "aural", docType: utility.FileTypeAURAL, filename: "audio.mp3", defaultValue: "naive", want: "audio"},
-		{name: "presentation by ext", docType: utility.FileTypeDOC, filename: "deck.pptx", defaultValue: "naive", want: "presentation"},
-		{name: "email by ext", docType: utility.FileTypeDOC, filename: "mail.eml", defaultValue: "naive", want: "email"},
-		{name: "fallback default", docType: utility.FileTypeDOC, filename: "notes.txt", defaultValue: "manual", want: "manual"},
-	}
-	for _, tt := range tests {
-		t.Run(tt.name, func(t *testing.T) {
-			if got := selectUploadParser(tt.docType, tt.filename, tt.defaultValue); got != tt.want {
-				t.Fatalf("selectUploadParser(%q)=%q, want %q", tt.filename, got, tt.want)
-			}
-		})
-	}
-}
-
 func TestContentHashHex_MatchesPythonXXH128(t *testing.T) {
 	tests := []struct {
 		data []byte

internal/service/document/document_upload.go

@@ -6,15 +6,12 @@ import (
 	"mime/multipart"
 	"net/http"
 	"path/filepath"
-	"ragflow/internal/service"
 	"strings"
 
 	"ragflow/internal/common"
 	"ragflow/internal/entity"
 	"ragflow/internal/storage"
 	"ragflow/internal/utility"
-
-	"go.uber.org/zap"
 )
 
 // UploadLocalDocuments stores each uploaded file in object storage and inserts a
@@ -313,7 +310,7 @@ func (s *DocumentService) newDatasetDocument(kb *entity.Knowledgebase, tenantID,
 	if i := strings.LastIndex(filename, "."); i >= 0 {
 		suffix = filename[i+1:]
 	}
-	parserID := selectUploadParser(utility.FileType(filetype), filename, kb.ParserID)
+	parserID := kb.ParserID
 	if kb.PipelineID != nil {
 		parserID = "" // canvas pipeline mode — parser_id not applicable
 	}
@@ -338,19 +335,6 @@ func (s *DocumentService) newDatasetDocument(kb *entity.Knowledgebase, tenantID,
 		hash := contentHashHex(blob)
 		doc.ContentHash = &hash
 	}
-
-	// When the document's builtin parser_id differs from the KB's (e.g. visual→picture,
-	// aural→audio), re-resolve component_params defaults from the document's own DSL
-	// template so cpnIDs and param keys match the pipeline that will actually execute.
-	if kb.PipelineID == nil && parserID != kb.ParserID {
-		if cp, err := service.ResolveComponentParamsDefaults(parserID, nil); err != nil {
-			common.Warn("newDatasetDocument: resolve component_params defaults",
-				zap.String("parserID", parserID), zap.Error(err))
-		} else if cp != nil {
-			doc.ParserConfig = cp
-		}
-	}
-
 	return doc
 }
 

internal/service/document/document_upload_helpers.go

@@ -3,22 +3,9 @@ package document
 import (
 	"encoding/hex"
 
-	pipelinepkg "ragflow/internal/ingestion/pipeline"
-	"ragflow/internal/utility"
-
 	"github.com/zeebo/xxh3"
 )
 
-// selectUploadParser resolves the parser_id for a document by delegating to
-// the builtin pipeline registry, which owns the file-type-to-parser mapping.
-func selectUploadParser(docType utility.FileType, filename, defaultParser string) string {
-	registry, err := pipelinepkg.DefaultRegistry()
-	if err != nil || registry == nil {
-		return defaultParser
-	}
-	return registry.DefaultParserID(string(docType), filename, defaultParser)
-}
-
 // contentHashHex mirrors Python xxhash.xxh128(blob).hexdigest().
 func contentHashHex(blob []byte) string {
 	sum := xxh3.Hash128(blob).Bytes()

internal/service/document/file2document.go

@@ -198,7 +198,7 @@ func (s *File2DocumentService) convertFiles(fileIDs, kbIDs []string, userID stri
 				continue
 			}
 
-			parserID := selectUploadParser(utility.FileType(file.Type), file.Name, kb.ParserID)
+			parserID := kb.ParserID
 			suffix := strings.TrimPrefix(filepath.Ext(file.Name), ".")
 			doc := &entity.Document{
 				ID:           utility.GenerateUUID(),
@@ -219,19 +219,6 @@ func (s *File2DocumentService) convertFiles(fileIDs, kbIDs []string, userID stri
 				doc.ParserID = "" // canvas pipeline mode — parser_id not applicable
 			}
 
-			// When the document's builtin parser_id differs from the KB's
-			// (e.g. visual→picture, aural→audio), re-resolve component_params
-			// defaults from the document's own DSL template so cpnIDs and
-			// param keys match the pipeline that will actually execute.
-			if kb.PipelineID == nil && parserID != kb.ParserID {
-				if cp, err := service.ResolveComponentParamsDefaults(parserID, nil); err != nil {
-					common.Warn("convertFiles: resolve component_params defaults",
-						zap.String("parserID", parserID), zap.Error(err))
-				} else if cp != nil {
-					doc.ParserConfig = cp
-				}
-			}
-
 			// InsertDocument creates the row and increments KB doc_num in one
 			// transaction, so a failed insert never leaves a stale counter.
 			if err := s.documentSvc.InsertDocument(doc); err != nil {