File block v5

Author: Sg312

apps/docs/components/ui/icon-mapping.ts

@@ -264,6 +264,7 @@ export const blockTypeToIconMap: Record<string, IconComponent> = {
   file: DocumentIcon,
   file_v2: DocumentIcon,
   file_v4: DocumentIcon,
+  file_v5: DocumentIcon,
   findymail: FindymailIcon,
   firecrawl: FirecrawlIcon,
   fireflies: FirefliesIcon,

apps/docs/content/docs/de/tools/file.mdx

@@ -6,7 +6,7 @@ description: Mehrere Dateien lesen und parsen
 import { BlockInfoCard } from "@/components/ui/block-info-card"
 
 <BlockInfoCard 
-  type="file_v4"
+  type="file_v5"
   color="#40916C"
 />
 

apps/docs/content/docs/en/execution/files.mdx

@@ -25,37 +25,39 @@ You can access any of these properties when referencing files from previous bloc
 
 ## The File Block
 
-The **File block** is the universal entry point for files in your workflows. It accepts files from any source and outputs standardized file objects that work with all integrations.
+The **File block** is the universal entry point for files in your workflows. It has several operations:
 
-**Inputs:**
-- **Uploaded files** - Drag and drop or select files directly
-- **External URLs** - Any publicly accessible file URL
-- **Files from other blocks** - Pass files from Gmail attachments, Slack downloads, etc.
+| Operation | What it does | Key output |
+| --- | --- | --- |
+| **Read** | Returns workspace file objects you select or reference. Does **not** load file text. | `files` (`UserFile[]`) |
+| **Get Content** | Extracts the text of one or more files (PDF, DOCX, CSV, etc. are parsed automatically). | `contents` (`string[]`, one entry per file) |
+| **Fetch** | Downloads and parses a file from an external URL, with optional request headers. | `files`, `combinedContent` |
+| **Write** | Creates a new workspace file. | `id`, `name`, `size`, `url` |
+| **Append** | Appends content to an existing workspace file. | `id`, `name`, `size`, `url` |
 
-**Outputs:**
-- A list of `UserFile` objects with consistent structure (`name`, `url`, `base64`, `type`, `size`)
-- `combinedContent` - Extracted text content from all files (for documents)
+`UserFile` objects share a consistent structure (`name`, `url`, `base64`, `type`, `size`), so they work with every integration.
 
-**Example usage:**
+### Reading file text: chain Read → Get Content
 
-```
-// Get all files from the File block
-<file.files>
+**Read intentionally returns only file objects, not their text.** To read the text of files, pass the file objects into a **Get Content** operation:
 
-// Get the first file
-<file.files[0]>
+```
+// Block 1 — File (Read): returns the selected workspace files
+<file1.files>
 
-// Get combined text content from parsed documents
-<file.combinedContent>
+// Block 2 — File (Get Content): set its file input to <file1.files>
+//   → returns an array of extracted text, one entry per file
+<file2.contents>        // all file contents
+<file2.contents[0]>     // text of the first file
 ```
 
-The File block automatically:
-- Detects file types from URLs and extensions
-- Extracts text from PDFs, CSVs, and documents
-- Generates base64 encoding for binary files
-- Creates presigned URLs for secure access
+The same pattern works for files produced anywhere — feed `<gmail.attachments>`, `<agent.files>`, or `<start.files>` straight into **Get Content**. You never need to extract `base64` or `url` yourself to get text.
+
+<Callout type="info">
+  Extracted `contents` can be large. The File block stores it through the execution large-value system automatically, so big documents don't bloat your run payloads.
+</Callout>
 
-Use the File block when you need to normalize files from different sources before passing them to other blocks like Vision, STT, or email integrations.
+Use the File block when you need to normalize files from different sources, read their text, or hand standardized file objects to other blocks like Vision, STT, or email integrations.
 
 ## Passing Files Between Blocks
 
@@ -138,7 +140,7 @@ Use `url` for direct downloads or `base64` for inline processing.
 ## Blocks That Work with Files
 
 **File inputs:**
-- **File** - Parse documents, images, and text files
+- **File** - Read file objects, extract text with Get Content, fetch URLs, write, and append
 - **Vision** - Analyze images with AI models
 - **Mistral Parser** - Extract text from PDFs
 

apps/docs/content/docs/es/tools/file.mdx

@@ -6,7 +6,7 @@ description: Leer y analizar múltiples archivos
 import { BlockInfoCard } from "@/components/ui/block-info-card"
 
 <BlockInfoCard 
-  type="file_v4"
+  type="file_v5"
   color="#40916C"
 />
 

apps/docs/content/docs/fr/tools/file.mdx

@@ -6,7 +6,7 @@ description: Lire et analyser plusieurs fichiers
 import { BlockInfoCard } from "@/components/ui/block-info-card"
 
 <BlockInfoCard 
-  type="file_v4"
+  type="file_v5"
   color="#40916C"
 />
 

apps/docs/content/docs/ja/tools/file.mdx

@@ -6,7 +6,7 @@ description: 複数のファイルを読み込んで解析する
 import { BlockInfoCard } from "@/components/ui/block-info-card"
 
 <BlockInfoCard 
-  type="file_v4"
+  type="file_v5"
   color="#40916C"
 />
 

apps/docs/content/docs/zh/tools/file.mdx

@@ -6,7 +6,7 @@ description: 读取并解析多个文件
 import { BlockInfoCard } from "@/components/ui/block-info-card"
 
 <BlockInfoCard 
-  type="file_v4"
+  type="file_v5"
   color="#40916C"
 />
 

apps/sim/app/api/tools/file/manage/route.ts

@@ -1,3 +1,4 @@
+import { Buffer, isUtf8 } from 'buffer'
 import { createLogger } from '@sim/logger'
 import { getErrorMessage } from '@sim/utils/errors'
 import { generateShortId } from '@sim/utils/id'
@@ -7,8 +8,10 @@ import { parseRequest } from '@/lib/api/server'
 import { checkInternalAuth } from '@/lib/auth/hybrid'
 import { splitWorkspaceFilePath } from '@/lib/copilot/tools/server/files/workspace-file'
 import { acquireLock, releaseLock } from '@/lib/core/config/redis'
+import { generateRequestId } from '@/lib/core/utils/request'
 import { ensureAbsoluteUrl } from '@/lib/core/utils/urls'
 import { withRouteHandler } from '@/lib/core/utils/with-route-handler'
+import { isSupportedFileType, parseBuffer } from '@/lib/file-parsers'
 import { ensureWorkspaceFileFolderPath } from '@/lib/uploads/contexts/workspace/workspace-file-folder-manager'
 import {
   fetchWorkspaceFileBuffer,
@@ -18,11 +21,14 @@ import {
   uploadWorkspaceFile,
 } from '@/lib/uploads/contexts/workspace/workspace-file-manager'
 import { getFileExtension, getMimeTypeFromExtension } from '@/lib/uploads/utils/file-utils'
+import { downloadFileFromStorage } from '@/lib/uploads/utils/file-utils.server'
 import { performMoveWorkspaceFileItems } from '@/lib/workspace-files/orchestration'
 import {
   assertActiveWorkspaceAccess,
   isWorkspaceAccessDeniedError,
 } from '@/lib/workspaces/permissions/utils'
+import { assertToolFileAccess } from '@/app/api/files/authorization'
+import type { UserFile } from '@/executor/types'
 
 export const dynamic = 'force-dynamic'
 
@@ -122,6 +128,46 @@ const extractFileIdsFromInput = (fileInput: unknown): string[] => {
     .filter((id) => id.length > 0)
 }
 
+/** Per-file download cap for the content operation. Aligned with the durable large-value ceiling. */
+const MAX_GET_CONTENT_FILE_BYTES = 64 * 1024 * 1024
+/** Combined extracted-text cap so the content array stays within the large-value-ref ceiling. */
+const MAX_GET_CONTENT_TOTAL_BYTES = 64 * 1024 * 1024
+
+const isLikelyTextBuffer = (buffer: Buffer): boolean => isUtf8(buffer) && !buffer.includes(0)
+
+/**
+ * Download a stored file and extract its text content. Parseable types (PDF, DOCX,
+ * CSV, etc.) go through the shared file-parsers; other UTF-8 files are returned as
+ * raw text; binary files yield a short placeholder rather than corrupt bytes.
+ */
+const extractUserFileTextContent = async (
+  userFile: UserFile,
+  requestId: string
+): Promise<string> => {
+  const buffer = await downloadFileFromStorage(userFile, requestId, logger, {
+    maxBytes: MAX_GET_CONTENT_FILE_BYTES,
+  })
+
+  const extension = getFileExtension(userFile.name)
+  if (extension && isSupportedFileType(extension)) {
+    try {
+      const result = await parseBuffer(buffer, extension)
+      return result.content ?? ''
+    } catch (error) {
+      logger.warn('Falling back to raw text after parser failure', {
+        name: userFile.name,
+        error: getErrorMessage(error, 'Unknown error'),
+      })
+    }
+  }
+
+  if (isLikelyTextBuffer(buffer)) {
+    return buffer.toString('utf-8')
+  }
+
+  return `[Binary file: ${userFile.name} (${userFile.type || 'application/octet-stream'}, ${buffer.length} bytes). Cannot extract text content.]`
+}
+
 export const POST = withRouteHandler(async (request: NextRequest) => {
   const auth = await checkInternalAuth(request, { requireWorkflowId: false })
   if (!auth.success) {
@@ -231,6 +277,69 @@ export const POST = withRouteHandler(async (request: NextRequest) => {
         })
       }
 
+      case 'content': {
+        const { fileId, fileInput } = body
+        const requestId = generateRequestId()
+
+        const selectedFileIds = Array.isArray(fileId)
+          ? fileId.map((id) => id.trim()).filter(Boolean)
+          : fileId
+            ? normalizeFileIdList(fileId)
+            : extractFileIdsFromInput(fileInput)
+        const selectedInputFiles = fileId ? [] : extractUserFilesFromInput(fileInput)
+
+        if (selectedFileIds.length === 0 && selectedInputFiles.length === 0) {
+          return NextResponse.json({ success: false, error: 'File is required' }, { status: 400 })
+        }
+
+        const workspaceFiles = await Promise.all(
+          selectedFileIds.map((id) => getWorkspaceFile(workspaceId, id))
+        )
+        const missingFileId = selectedFileIds.find((_, index) => !workspaceFiles[index])
+        if (missingFileId) {
+          return NextResponse.json(
+            { success: false, error: `File not found: "${missingFileId}"` },
+            { status: 404 }
+          )
+        }
+
+        const userFiles: UserFile[] = workspaceFiles
+          .map((file) => workspaceFileToUserFile(file))
+          .filter((file): file is NonNullable<ReturnType<typeof workspaceFileToUserFile>> =>
+            Boolean(file)
+          )
+          .concat(selectedInputFiles)
+
+        const contents: string[] = []
+        let totalBytes = 0
+        for (const userFile of userFiles) {
+          const denied = await assertToolFileAccess(userFile.key, userId, requestId, logger)
+          if (denied) return denied
+
+          const content = await extractUserFileTextContent(userFile, requestId)
+          totalBytes += Buffer.byteLength(content, 'utf8')
+          if (totalBytes > MAX_GET_CONTENT_TOTAL_BYTES) {
+            return NextResponse.json(
+              {
+                success: false,
+                error: `Combined file content is too large to return safely. Maximum is ${
+                  MAX_GET_CONTENT_TOTAL_BYTES / (1024 * 1024)
+                } MB.`,
+              },
+              { status: 413 }
+            )
+          }
+          contents.push(content)
+        }
+
+        logger.info('File content extracted', { count: contents.length })
+
+        return NextResponse.json({
+          success: true,
+          data: { contents },
+        })
+      }
+
       case 'write': {
         const { fileName, content, contentType } = body
         const { folderSegments, leafName } = splitWorkspaceFilePath(fileName)

apps/sim/blocks/blocks.test.ts

@@ -109,11 +109,11 @@ describe.concurrent('Blocks Module', () => {
       expect(block?.tools.config?.tool({ operation: 'file_get' })).toBe('file_get')
     })
 
-    it('should expose v4 with read and fetch routed to the expected tools', () => {
+    it('should keep v4 read and fetch routed to the expected tools', () => {
       const block = getBlock('file_v4')
 
       expect(block).toBeDefined()
-      expect(block?.hideFromToolbar).toBe(false)
+      expect(block?.hideFromToolbar).toBe(true)
       expect(block?.subBlocks[0].options?.map((option) => option.id)).toEqual([
         'file_read',
         'file_fetch',
@@ -160,6 +160,70 @@ describe.concurrent('Blocks Module', () => {
         workspaceId: 'workspace-1',
       })
     })
+
+    it('should expose v5 with read (files only) and a get content operation', () => {
+      const block = getBlock('file_v5')
+
+      expect(block).toBeDefined()
+      expect(block?.hideFromToolbar).toBe(false)
+      expect(block?.subBlocks[0].options?.map((option) => option.id)).toEqual([
+        'file_read',
+        'file_get_content',
+        'file_fetch',
+        'file_write',
+        'file_append',
+      ])
+      expect(block?.subBlocks.find((subBlock) => subBlock.id === 'readFile')?.multiple).toBe(true)
+      expect(block?.tools.config?.tool({ operation: 'file_read' })).toBe('file_read')
+      expect(block?.tools.config?.tool({ operation: 'file_get_content' })).toBe('file_get_content')
+      expect(block?.tools.config?.tool({ operation: 'file_fetch' })).toBe('file_fetch')
+
+      // Read exposes files only; the redundant single-file output is gone
+      expect(block?.outputs.files).toBeDefined()
+      expect(block?.outputs.file).toBeUndefined()
+      // Get content exposes a contents array
+      expect(block?.outputs.contents).toBeDefined()
+
+      // Get content resolves canonical IDs
+      expect(
+        block?.tools.config?.params?.({
+          operation: 'file_get_content',
+          getContentInput: '["file-1","file-2"]',
+          _context: { workspaceId: 'workspace-1' },
+        })
+      ).toEqual({
+        fileId: ['file-1', 'file-2'],
+        workspaceId: 'workspace-1',
+      })
+
+      // Get content resolves selected file objects
+      expect(
+        block?.tools.config?.params?.({
+          operation: 'file_get_content',
+          getContentInput: [
+            {
+              key: 'workspace/workspace-1/example.md',
+              name: 'example.md',
+              path: '/api/files/serve/workspace%2Fworkspace-1%2Fexample.md?context=workspace',
+              size: 123,
+              type: 'text/markdown',
+            },
+          ],
+          _context: { workspaceId: 'workspace-1' },
+        })
+      ).toEqual({
+        fileInput: [
+          {
+            key: 'workspace/workspace-1/example.md',
+            name: 'example.md',
+            path: '/api/files/serve/workspace%2Fworkspace-1%2Fexample.md?context=workspace',
+            size: 123,
+            type: 'text/markdown',
+          },
+        ],
+        workspaceId: 'workspace-1',
+      })
+    })
   })
 
   describe('Agent block', () => {