refactor: XYNE-103 Attachments are moved to file.sd schema

frontend/src/components/AttachmentPreview.tsx

@@ -10,7 +10,7 @@ import {
   DialogHeader,
   DialogTitle,
 } from "@/components/ui/dialog"
-import { getFileType } from "@/utils/fileUtils"
+import { getFileType } from "shared/fileUtils"
 import { getFileIcon } from "@/components/ChatBox"
 
 interface AttachmentPreviewProps {

frontend/src/components/ChatBox.tsx

@@ -74,8 +74,8 @@ import {
   validateAndDeduplicateFiles,
   createImagePreview,
   cleanupPreviewUrls,
-  getFileType,
 } from "@/utils/fileUtils"
+import { getFileType } from "shared/fileUtils"
 import { authFetch } from "@/utils/authFetch"
 
 interface SelectedFile {
@@ -977,15 +977,37 @@ export const ChatBox = React.forwardRef<ChatBoxRef, ChatBoxProps>(
       return ext || "file"
     }
 
-    const removeFile = useCallback((id: string) => {
+    const removeFile = useCallback(async (id: string) => {
+      const fileToRemove = selectedFiles.find((f) => f.id === id)
+
+      // If the file has metadata with fileId (meaning it's already uploaded), delete it from the server
+      if (fileToRemove?.metadata?.fileId) {
+        try {
+          const response = await api.files.delete.$post({
+            json: {
+              attachment: fileToRemove.metadata
+            }
+          })
+
+          if (!response.ok) {
+            const errorText = await response.text()
+            console.error(`Failed to delete attachment: ${errorText}`)
+            // Still remove from UI even if server deletion fails
+          }
+        } catch (error) {
+          console.error('Error deleting attachment:', error)
+          // Still remove from UI even if server deletion fails
+        }
+      }
+
+      // Remove from UI
       setSelectedFiles((prev) => {
-        const fileToRemove = prev.find((f) => f.id === id)
         if (fileToRemove?.preview) {
           URL.revokeObjectURL(fileToRemove.preview)
         }
         return prev.filter((f) => f.id !== id)
       })
-    }, [])
+    }, [selectedFiles])
 
     const { handleFileSelect, handleFileChange } = createFileSelectionHandlers(
       fileInputRef,

frontend/src/lib/common.tsx

@@ -43,10 +43,9 @@ import {
   DataSourceEntity,
   WebSearchEntity,
   FileType,
-  MIME_TYPE_MAPPINGS,
-  EXTENSION_MAPPINGS,
 } from "shared/types"
 import { LoadingSpinner } from "@/routes/_authenticated/admin/integrations/google"
+import { getFileType } from "shared/fileUtils"
 
 // Define placeholder entities if they don't exist in shared/types
 const PdfEntity = { Default: "pdf_default" } as const
@@ -165,30 +164,9 @@ export const getIcon = (
   }
 }
 
-// Helper function to determine FileType from a file
-export const getFileType = (file: File): FileType => {
-  // First check by MIME type
-  for (const [fileType, mimeTypes] of Object.entries(MIME_TYPE_MAPPINGS)) {
-    if ((mimeTypes as readonly string[]).includes(file.type)) {
-      return fileType as FileType
-    }
-  }
-
-  // Fallback to extension checking
-  const fileName = file.name.toLowerCase()
-  for (const [fileType, extensions] of Object.entries(EXTENSION_MAPPINGS)) {
-    if ((extensions as readonly string[]).some(ext => fileName.endsWith(ext))) {
-      return fileType as FileType
-    }
-  }
-
-  // Default fallback
-  return FileType.FILE
-}
-
 // Icon mapping from FileType to SVG component
 export const getFileIcon = (file: File) => {
-  const fileType = getFileType(file)
+  const fileType = getFileType({type: file.type, name: file.name})
 
   switch (fileType) {
     case FileType.TEXT:

frontend/src/utils/fileUtils.ts

@@ -1,35 +1,11 @@
 import { isValidFile, isImageFile } from "shared/fileUtils"
 import { SelectedFile } from "@/components/ClFileUpload"
 import { authFetch } from "./authFetch"
-import { FileType, MIME_TYPE_MAPPINGS, EXTENSION_MAPPINGS, UploadStatus } from "shared/types"
+import { UploadStatus } from "shared/types"
 
 // Generate unique ID for files
 export const generateFileId = () => Math.random().toString(36).substring(2, 9)
 
-export const getFileType = ({ type, name }: { type: string, name: string }): FileType => {
-  const fileName = name.toLowerCase()
-  const mimeType = type.toLowerCase()
-  const baseMime = mimeType.split(";")[0]
-
-  // Check each file type category using the mappings
-  for (const [fileType, mimeTypes] of Object.entries(MIME_TYPE_MAPPINGS)) {
-    // Check MIME type first (more reliable)
-    if (mimeTypes.some(mime => baseMime === mime)) {
-      return fileType as FileType
-    }
-  }
-
-  // Fallback to extension-based detection
-  for (const [fileType, extensions] of Object.entries(EXTENSION_MAPPINGS)) {
-    if (extensions.some(ext => fileName.endsWith(ext))) {
-      return fileType as FileType
-    }
-  }
-
-  // Default fallback
-  return FileType.FILE
-}
-
 // Create preview URL for image files
 export const createImagePreview = (file: File): string | undefined => {
   if (isImageFile(file.type)) {

package.json

@@ -2,4 +2,4 @@
   "dependencies": {
     "zustand": "^5.0.8"
   }
-}
+}

server/api/chat/agents.ts

@@ -83,6 +83,7 @@ import {
   VespaChatUserSchema,
   type VespaSearchResult,
   type VespaSearchResults,
+  AttachmentEntity,
 } from "@xyne/vespa-ts/types"
 import { APIError } from "openai"
 import { insertChatTrace } from "@/db/chatTrace"
@@ -121,7 +122,6 @@ import { getModelValueFromLabel } from "@/ai/modelConfig"
 import {
   buildContext,
   buildUserQuery,
-  expandSheetIds,
   getThreadContext,
   isContextSelected,
   UnderstandMessageAndAnswer,
@@ -156,6 +156,7 @@ import { getDateForAI } from "@/utils/index"
 import { validateVespaIdInAgentIntegrations } from "@/search/utils"
 import { getAuth, safeGet } from "../agent"
 import { applyFollowUpContext } from "@/utils/parseAttachment"
+import { expandSheetIds } from "@/search/utils"
 const {
   JwtPayloadKey,
   defaultBestModel,
@@ -392,7 +393,7 @@ const checkAndYieldCitationsForAgent = async function* (
           }
 
           // we dont want citations for attachments in the chat
-          if (item.source.entity === KnowledgeBaseEntity.Attachment) {
+          if (Object.values(AttachmentEntity).includes(item.source.entity as AttachmentEntity)) {
             continue
           }
 

server/api/chat/chat.ts

@@ -128,6 +128,8 @@ import {
   type VespaSearchResultsSchema,
   KnowledgeBaseEntity,
   KbItemsSchema,
+  AttachmentEntity,
+  fileSchema,
 } from "@xyne/vespa-ts/types"
 import { APIError } from "openai"
 import {
@@ -213,31 +215,12 @@ import { getDateForAI } from "@/utils/index"
 import type { User } from "@microsoft/microsoft-graph-types"
 import { getAuth, safeGet } from "../agent"
 import { getChunkCountPerDoc } from "./chunk-selection"
+import { handleAttachmentDelete } from "../files"
+import { expandSheetIds } from "@/search/utils"
 
 const METADATA_NO_DOCUMENTS_FOUND = "METADATA_NO_DOCUMENTS_FOUND_INTERNAL"
 const METADATA_FALLBACK_TO_RAG = "METADATA_FALLBACK_TO_RAG_INTERNAL"
 
-export function expandSheetIds(fileId: string): string[] {
-  // Check if the fileId matches the pattern docId_sheet_number
-  const sheetMatch = fileId.match(/^(.+)_sheet_(\d+)$/)
-
-  if (!sheetMatch) {
-    // Not a sheet ID, return as is
-    return [fileId]
-  }
-
-  const [, docId, sheetNumberStr] = sheetMatch
-  const sheetNumber = parseInt(sheetNumberStr, 10)
-  // Generate IDs from docId_sheet_0 to docId_sheet_number
-  const expandedIds: string[] = []
-  const upper = Number.isFinite(sheetNumber) ? sheetNumber : 1
-  for (let i = 0; i < upper; i++) {
-    expandedIds.push(`${docId}_sheet_${i}`)
-  }
-
-  return expandedIds
-}
-
 export async function resolveNamesToEmails(
   intent: Intent,
   email: string,
@@ -506,7 +489,7 @@ const checkAndYieldCitations = async function* (
           const f = (item as any)?.fields
           if (
             f?.sddocname === dataSourceFileSchema ||
-            f?.entity === KnowledgeBaseEntity.Attachment
+            Object.values(AttachmentEntity).includes(f?.entity)
           ) {
             // Skip datasource and attachment files from citations
             continue
@@ -784,123 +767,21 @@ export const ChatDeleteApi = async (c: Context) => {
     email = sub || ""
     // @ts-ignore
     const { chatId } = c.req.valid("json")
+    const attachmentsToDelete: AttachmentMetadata[] = []
     await db.transaction(async (tx) => {
       // Get the chat's internal ID first
       const chat = await getChatByExternalIdWithAuth(tx, chatId, email)
       if (!chat) {
         throw new HTTPException(404, { message: "Chat not found" })
       }
 
-      // Get all messages for the chat to find attachments
+      // Get all messages for the chat to delete attachments
       const messagesToDelete = await getChatMessagesWithAuth(tx, chatId, email)
 
-      // Collect all attachment file IDs that need to be deleted
-      const imageAttachmentFileIds: string[] = []
-      const nonImageAttachmentFileIds: string[] = []
-
       for (const message of messagesToDelete) {
         if (message.attachments && Array.isArray(message.attachments)) {
-          const attachments =
-            message.attachments as unknown as AttachmentMetadata[]
-          for (const attachment of attachments) {
-            if (attachment && typeof attachment === "object") {
-              if (attachment.fileId) {
-                // Check if this is an image attachment using both isImage field and fileType
-                const isImageAttachment =
-                  attachment.isImage ||
-                  (attachment.fileType && isImageFile(attachment.fileType))
-
-                if (isImageAttachment) {
-                  imageAttachmentFileIds.push(attachment.fileId)
-                } else {
-                  nonImageAttachmentFileIds.push(attachment.fileId)
-                }
-              }
-            }
-          }
-        }
-      }
-
-      // Delete image attachments and their thumbnails from disk
-      if (imageAttachmentFileIds.length > 0) {
-        loggerWithChild({ email: email }).info(
-          `Deleting ${imageAttachmentFileIds.length} image attachment files and their thumbnails for chat ${chatId}`,
-        )
-
-        for (const fileId of imageAttachmentFileIds) {
-          try {
-            // Validate fileId to prevent path traversal
-            if (
-              fileId.includes("..") ||
-              fileId.includes("/") ||
-              fileId.includes("\\")
-            ) {
-              loggerWithChild({ email: email }).error(
-                `Invalid fileId detected: ${fileId}. Skipping deletion for security.`,
-              )
-              continue
-            }
-            const imageBaseDir = path.resolve(
-              process.env.IMAGE_DIR || "downloads/xyne_images_db",
-            )
-
-            const imageDir = path.join(imageBaseDir, fileId)
-            try {
-              await fs.access(imageDir)
-              await fs.rm(imageDir, { recursive: true, force: true })
-              loggerWithChild({ email: email }).info(
-                `Deleted image attachment directory: ${imageDir}`,
-              )
-            } catch (attachmentError) {
-              loggerWithChild({ email: email }).warn(
-                `Image attachment file ${fileId} not found in either directory during chat deletion`,
-              )
-            }
-          } catch (error) {
-            loggerWithChild({ email: email }).error(
-              error,
-              `Failed to delete image attachment file ${fileId} during chat deletion: ${getErrorMessage(error)}`,
-            )
-          }
-        }
-      }
-
-      // Delete non-image attachments from Vespa kb_items schema
-      if (nonImageAttachmentFileIds.length > 0) {
-        loggerWithChild({ email: email }).info(
-          `Deleting ${nonImageAttachmentFileIds.length} non-image attachments from Vespa kb_items schema for chat ${chatId}`,
-        )
-
-        for (const fileId of nonImageAttachmentFileIds) {
-          try {
-            // Delete from Vespa kb_items schema using the proper Vespa function
-            const vespaIds = expandSheetIds(fileId)
-            for (const id of vespaIds) {
-              try {
-                await DeleteDocument(id, KbItemsSchema)
-                loggerWithChild({ email }).info(
-                  `Successfully deleted non-image attachment ${id} from Vespa kb_items schema`,
-                )
-              } catch (error) {
-                loggerWithChild({ email }).error(
-                  `Failed to delete non-image attachment ${id} from Vespa kb_items schema`,
-                  { error: getErrorMessage(error) }
-                )
-              }
-            }
-          } catch (error) {
-            const errorMessage = getErrorMessage(error)
-            if (errorMessage.includes("404 Not Found")) {
-              loggerWithChild({ email: email }).warn(
-                `Non-image attachment ${fileId} not found in Vespa kb_items schema (may have been already deleted)`,
-              )
-            } else {
-              loggerWithChild({ email: email }).error(
-                error,
-                `Failed to delete non-image attachment ${fileId} from Vespa kb_items schema: ${errorMessage}`,
-              )
-            }
-          }
+          const attachments = message.attachments as AttachmentMetadata[]
+          attachmentsToDelete.push(...attachments)
         }
       }
 
@@ -913,6 +794,9 @@ export const ChatDeleteApi = async (c: Context) => {
       await deleteMessagesByChatId(tx, chatId)
       await deleteChatByExternalIdWithAuth(tx, chatId, email)
     })
+    if (attachmentsToDelete.length) {
+      await handleAttachmentDelete(attachmentsToDelete, email)
+    }
     return c.json({ success: true })
   } catch (error) {
     const errMsg = getErrorMessage(error)
@@ -2086,6 +1970,7 @@ async function* generateAnswerFromGivenContext(
         results = await searchVespaInFiles(builtUserQuery, email, nonCollectionFileIds, {
             limit: fileIds?.length,
             alpha: userAlpha,
+            rankProfile: SearchModes.attachmentRank,
           })
         if (results.root.children) {
           combinedSearchResponse.push(...results.root.children)