improvement(vfs): update custom glob impl to use micromatch, fix vfs filename regex (#3680)

* improvement(vfs): update custom glob impl to use micromatch, fix vfs filename regex

* add tests

* file caps

* address comments

* fix open resource

* consolidate files

Author: icecrasher321

apps/sim/app/workspace/[workspaceId]/home/components/mothership-view/components/resource-content/resource-content.tsx

@@ -260,7 +260,7 @@ function EmbeddedFileActions({ workspaceId, fileId }: EmbeddedFileActionsProps)
   }, [file])
 
   const handleOpenInFiles = useCallback(() => {
-    router.push(`/workspace/${workspaceId}/files?fileId=${fileId}`)
+    router.push(`/workspace/${workspaceId}/files?fileId=${encodeURIComponent(fileId)}`)
   }, [router, workspaceId, fileId])
 
   return (

apps/sim/lib/copilot/orchestrator/tool-executor/index.ts

@@ -16,6 +16,7 @@ import { validateMcpDomain } from '@/lib/mcp/domain-check'
 import { mcpService } from '@/lib/mcp/service'
 import { generateMcpServerId } from '@/lib/mcp/utils'
 import { getAllOAuthServices } from '@/lib/oauth/utils'
+import { getWorkspaceFile } from '@/lib/uploads/contexts/workspace/workspace-file-manager'
 import {
   deleteCustomTool,
   getCustomToolById,
@@ -24,7 +25,7 @@ import {
 } from '@/lib/workflows/custom-tools/operations'
 import { deleteSkill, listSkills, upsertSkills } from '@/lib/workflows/skills/operations'
 import { getWorkflowById } from '@/lib/workflows/utils'
-import { isMcpTool } from '@/executor/constants'
+import { isMcpTool, isUuid } from '@/executor/constants'
 import { executeTool } from '@/tools'
 import { getTool, resolveToolId } from '@/tools/utils'
 import {
@@ -1029,15 +1030,42 @@ const SIM_WORKFLOW_TOOL_HANDLERS: Record<
   list: (p, c) => executeVfsList(p, c),
 
   // Resource visibility
-  open_resource: async (p: OpenResourceParams) => {
+  open_resource: async (p: OpenResourceParams, c: ExecutionContext) => {
     const validated = validateOpenResourceParams(p)
     if (!validated.success) {
       return { success: false, error: validated.error }
     }
 
     const params = validated.params
     const resourceType = params.type
-    const resourceId = params.id
+    let resourceId = params.id
+    let title: string = resourceType
+
+    if (resourceType === 'file') {
+      if (!c.workspaceId) {
+        return {
+          success: false,
+          error:
+            'Opening a workspace file requires workspace context. Pass the file UUID from files/<name>/meta.json.',
+        }
+      }
+      if (!isUuid(params.id)) {
+        return {
+          success: false,
+          error:
+            'open_resource for files requires the canonical UUID from files/<name>/meta.json (the "id" field). Do not pass VFS paths, display names, or file_<name> strings.',
+        }
+      }
+      const record = await getWorkspaceFile(c.workspaceId, params.id)
+      if (!record) {
+        return {
+          success: false,
+          error: `No workspace file with id "${params.id}". Confirm the UUID from meta.json.`,
+        }
+      }
+      resourceId = record.id
+      title = record.name
+    }
 
     return {
       success: true,
@@ -1046,7 +1074,7 @@ const SIM_WORKFLOW_TOOL_HANDLERS: Record<
         {
           type: resourceType as 'workflow' | 'table' | 'knowledgebase' | 'file',
           id: resourceId,
-          title: resourceType,
+          title,
         },
       ],
     }

apps/sim/lib/copilot/orchestrator/tool-executor/materialize-file.ts

@@ -2,6 +2,7 @@ import { db } from '@sim/db'
 import { workflow, workspaceFiles } from '@sim/db/schema'
 import { createLogger } from '@sim/logger'
 import { and, eq, isNull } from 'drizzle-orm'
+import { findMothershipUploadRowByChatAndName } from '@/lib/copilot/orchestrator/tool-executor/upload-file-reader'
 import type { ExecutionContext, ToolCallResult } from '@/lib/copilot/orchestrator/types'
 import { getServePathPrefix } from '@/lib/uploads'
 import { downloadWorkspaceFile } from '@/lib/uploads/contexts/workspace/workspace-file-manager'
@@ -12,22 +13,6 @@ import { extractWorkflowMetadata } from '@/app/api/v1/admin/types'
 
 const logger = createLogger('MaterializeFile')
 
-async function findUploadRecord(fileName: string, chatId: string) {
-  const rows = await db
-    .select()
-    .from(workspaceFiles)
-    .where(
-      and(
-        eq(workspaceFiles.originalName, fileName),
-        eq(workspaceFiles.chatId, chatId),
-        eq(workspaceFiles.context, 'mothership'),
-        isNull(workspaceFiles.deletedAt)
-      )
-    )
-    .limit(1)
-  return rows[0] ?? null
-}
-
 function toFileRecord(row: typeof workspaceFiles.$inferSelect) {
   const pathPrefix = getServePathPrefix()
   return {
@@ -41,21 +26,23 @@ function toFileRecord(row: typeof workspaceFiles.$inferSelect) {
     uploadedBy: row.userId,
     deletedAt: row.deletedAt,
     uploadedAt: row.uploadedAt,
+    storageContext: 'mothership' as const,
   }
 }
 
 async function executeSave(fileName: string, chatId: string): Promise<ToolCallResult> {
+  const row = await findMothershipUploadRowByChatAndName(chatId, fileName)
+  if (!row) {
+    return {
+      success: false,
+      error: `Upload not found: "${fileName}". Use glob("uploads/*") to list available uploads.`,
+    }
+  }
+
   const [updated] = await db
     .update(workspaceFiles)
     .set({ context: 'workspace', chatId: null })
-    .where(
-      and(
-        eq(workspaceFiles.originalName, fileName),
-        eq(workspaceFiles.chatId, chatId),
-        eq(workspaceFiles.context, 'mothership'),
-        isNull(workspaceFiles.deletedAt)
-      )
-    )
+    .where(and(eq(workspaceFiles.id, row.id), isNull(workspaceFiles.deletedAt)))
     .returning({ id: workspaceFiles.id, originalName: workspaceFiles.originalName })
 
   if (!updated) {
@@ -84,7 +71,7 @@ async function executeImport(
   workspaceId: string,
   userId: string
 ): Promise<ToolCallResult> {
-  const row = await findUploadRecord(fileName, chatId)
+  const row = await findMothershipUploadRowByChatAndName(chatId, fileName)
   if (!row) {
     return {
       success: false,

apps/sim/lib/copilot/orchestrator/tool-executor/upload-file-reader.ts

@@ -3,6 +3,7 @@ import { workspaceFiles } from '@sim/db/schema'
 import { createLogger } from '@sim/logger'
 import { and, eq, isNull } from 'drizzle-orm'
 import { type FileReadResult, readFileRecord } from '@/lib/copilot/vfs/file-reader'
+import { normalizeVfsSegment } from '@/lib/copilot/vfs/normalize-segment'
 import { getServePathPrefix } from '@/lib/uploads'
 import type { WorkspaceFileRecord } from '@/lib/uploads/contexts/workspace/workspace-file-manager'
 
@@ -21,9 +22,50 @@ function toWorkspaceFileRecord(row: typeof workspaceFiles.$inferSelect): Workspa
     uploadedBy: row.userId,
     deletedAt: row.deletedAt,
     uploadedAt: row.uploadedAt,
+    storageContext: 'mothership',
   }
 }
 
+/**
+ * Resolve a mothership upload row by `originalName`, preferring an exact DB match (limit 1) and
+ * only scanning all chat uploads when that misses (e.g. macOS U+202F vs ASCII space in the name).
+ */
+export async function findMothershipUploadRowByChatAndName(
+  chatId: string,
+  fileName: string
+): Promise<typeof workspaceFiles.$inferSelect | null> {
+  const exactRows = await db
+    .select()
+    .from(workspaceFiles)
+    .where(
+      and(
+        eq(workspaceFiles.chatId, chatId),
+        eq(workspaceFiles.context, 'mothership'),
+        eq(workspaceFiles.originalName, fileName),
+        isNull(workspaceFiles.deletedAt)
+      )
+    )
+    .limit(1)
+
+  if (exactRows[0]) {
+    return exactRows[0]
+  }
+
+  const allRows = await db
+    .select()
+    .from(workspaceFiles)
+    .where(
+      and(
+        eq(workspaceFiles.chatId, chatId),
+        eq(workspaceFiles.context, 'mothership'),
+        isNull(workspaceFiles.deletedAt)
+      )
+    )
+
+  const segmentKey = normalizeVfsSegment(fileName)
+  return allRows.find((r) => normalizeVfsSegment(r.originalName) === segmentKey) ?? null
+}
+
 /**
  * List all chat-scoped uploads for a given chat.
  */
@@ -51,30 +93,18 @@ export async function listChatUploads(chatId: string): Promise<WorkspaceFileReco
 }
 
 /**
- * Read a specific uploaded file by name within a chat session.
+ * Read a specific uploaded file by display name within a chat session.
+ * Resolves names with `normalizeVfsSegment` so macOS screenshot spacing (e.g. U+202F)
+ * matches when the model passes a visually equivalent path.
  */
 export async function readChatUpload(
   filename: string,
   chatId: string
 ): Promise<FileReadResult | null> {
   try {
-    const rows = await db
-      .select()
-      .from(workspaceFiles)
-      .where(
-        and(
-          eq(workspaceFiles.chatId, chatId),
-          eq(workspaceFiles.context, 'mothership'),
-          eq(workspaceFiles.originalName, filename),
-          isNull(workspaceFiles.deletedAt)
-        )
-      )
-      .limit(1)
-
-    if (rows.length === 0) return null
-
-    const record = toWorkspaceFileRecord(rows[0])
-    return readFileRecord(record)
+    const row = await findMothershipUploadRowByChatAndName(chatId, filename)
+    if (!row) return null
+    return readFileRecord(toWorkspaceFileRecord(row))
   } catch (err) {
     logger.warn('Failed to read chat upload', {
       filename,

apps/sim/lib/copilot/vfs/file-reader.ts

@@ -5,8 +5,8 @@ import { isImageFileType } from '@/lib/uploads/utils/file-utils'
 
 const logger = createLogger('FileReader')
 
-const MAX_TEXT_READ_BYTES = 512 * 1024 // 512 KB
-const MAX_IMAGE_READ_BYTES = 5 * 1024 * 1024 // 5 MB
+const MAX_TEXT_READ_BYTES = 5 * 1024 * 1024 // 5 MB
+const MAX_IMAGE_READ_BYTES = 20 * 1024 * 1024 // 20 MB
 
 const TEXT_TYPES = new Set([
   'text/plain',
@@ -53,7 +53,7 @@ export async function readFileRecord(record: WorkspaceFileRecord): Promise<FileR
     if (isImageFileType(record.type)) {
       if (record.size > MAX_IMAGE_READ_BYTES) {
         return {
-          content: `[Image too large: ${record.name} (${(record.size / 1024 / 1024).toFixed(1)}MB, limit 5MB)]`,
+          content: `[Image too large: ${record.name} (${(record.size / 1024 / 1024).toFixed(1)}MB, limit 20MB)]`,
           totalLines: 1,
         }
       }

apps/sim/lib/copilot/vfs/normalize-segment.ts

@@ -0,0 +1,14 @@
+/**
+ * Normalize a string for use as a single VFS path segment (workflow name, file name, etc.).
+ * Applies NFC normalization, trims, strips ASCII control characters, maps `/` to `-`, and
+ * collapses Unicode whitespace (including U+202F as in macOS screenshot names) to a single
+ * ASCII space.
+ */
+export function normalizeVfsSegment(name: string): string {
+  return name
+    .normalize('NFC')
+    .trim()
+    .replace(/[\x00-\x1f\x7f]/g, '')
+    .replace(/\//g, '-')
+    .replace(/\s+/g, ' ')
+}