Function handleFileForOCR

apps/sim/lib/knowledge/documents/document-processor.ts:312–382

(
  fileUrl: string,
  filename: string,
  mimeType: string,
  userId?: string,
  workspaceId?: string | null
)

}

async function handleFileForOCR(
  fileUrl: string,
  filename: string,
  mimeType: string,
  userId?: string,
  workspaceId?: string | null
) {
  const isExternalHttps = /^https:\/\//i.test(fileUrl) && !isInternalFileUrl(fileUrl)

  if (isExternalHttps) {
    if (mimeType === 'application/pdf') {
      logger.info(`handleFileForOCR: Downloading external PDF to check page count`)
      try {
        const buffer = await downloadFileWithTimeout(fileUrl, userId)
        logger.info(`handleFileForOCR: Downloaded external PDF: ${buffer.length} bytes`)
        return { httpsUrl: fileUrl, buffer }
      } catch (error) {
        logger.warn(
          `handleFileForOCR: Failed to download external PDF for page count check, proceeding without batching`,
          {
            error: toError(error).message,
          }
        )
        return { httpsUrl: fileUrl, buffer: undefined }
      }
    }
    logger.info(`handleFileForOCR: Using external URL directly`)
    return { httpsUrl: fileUrl, buffer: undefined }
  }

  logger.info(`Uploading "${filename}" to cloud storage for OCR`)

  const buffer = await downloadFileWithTimeout(fileUrl, userId)

  logger.info(`Downloaded ${filename}: ${buffer.length} bytes`)

  try {
    const metadata: Record<string, string> = {
      originalName: filename,
      uploadedAt: new Date().toISOString(),
      purpose: 'knowledge-base',
      ...(userId && { userId }),
      ...(workspaceId && { workspaceId }),
    }

    const timestamp = Date.now()
    const uniqueId = randomBytes(8).toString('hex')
    const safeFileName = filename.replace(/[^a-zA-Z0-9.-]/g, '_')
    const customKey = `kb/${timestamp}-${uniqueId}-${safeFileName}`

    const cloudResult = await StorageService.uploadFile({
      file: buffer,
      fileName: filename,
      contentType: mimeType,
      context: 'knowledge-base',
      customKey,
      metadata,
    })