Function parseDocument

apps/sim/lib/knowledge/documents/document-processor.ts:277–310

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

}

async function parseDocument(
  fileUrl: string,
  filename: string,
  mimeType: string,
  userId?: string,
  workspaceId?: string | null
): Promise<{
  content: string
  processingMethod: 'file-parser' | 'mistral-ocr'
  cloudUrl?: string
  metadata?: FileParseMetadata
}> {
  const isPDF = mimeType === 'application/pdf'
  const hasAzureMistralOCR =
    env.OCR_AZURE_API_KEY && env.OCR_AZURE_ENDPOINT && env.OCR_AZURE_MODEL_NAME

  const mistralApiKey = await getMistralApiKey(workspaceId)
  const hasMistralOCR = !!mistralApiKey

  if (isPDF && (hasAzureMistralOCR || hasMistralOCR)) {
    if (hasAzureMistralOCR) {
      logger.info(`Using Azure Mistral OCR: ${filename}`)
      return parseWithAzureMistralOCR(fileUrl, filename, mimeType, userId)
    }

    if (hasMistralOCR) {
      logger.info(`Using Mistral OCR: ${filename}`)
      return parseWithMistralOCR(fileUrl, filename, mimeType, userId, workspaceId, mistralApiKey)
    }
  }

  logger.info(`Using file parser: ${filename}`)
  return parseWithFileParser(fileUrl, filename, mimeType, userId)
}

async function handleFileForOCR(
  fileUrl: string,