Function addDocument

apps/sim/lib/knowledge/connectors/sync-engine.ts:1173–1249

* Upload content to storage as a .txt file, create a document record, * and trigger processing via the existing pipeline.

(
  knowledgeBaseId: string,
  connectorId: string,
  connectorType: string,
  extDoc: ExternalDocument,
  kbOwner: KnowledgeBaseOwner,
  sourceConfig?: Record<string, unknown>
)

 * and trigger processing via the existing pipeline.
 */
async function addDocument(
  knowledgeBaseId: string,
  connectorId: string,
  connectorType: string,
  extDoc: ExternalDocument,
  kbOwner: KnowledgeBaseOwner,
  sourceConfig?: Record<string, unknown>
): Promise<DocumentData> {
  const documentId = generateId()
  const contentBuffer = Buffer.from(extDoc.content, 'utf-8')
  const safeTitle = sanitizeStorageTitle(extDoc.title)
  const customKey = `kb/${Date.now()}-${documentId}-${safeTitle}.txt`

  const fileInfo = await StorageService.uploadFile({
    file: contentBuffer,
    fileName: `${safeTitle}.txt`,
    contentType: 'text/plain',
    context: 'knowledge-base',
    customKey,
    preserveKey: true,
    metadata: kbOwnershipMetadata(kbOwner, `${safeTitle}.txt`),
  })

  const fileUrl = `${getInternalApiBaseUrl()}${fileInfo.path}?context=knowledge-base`

  const tagValues = extDoc.metadata
    ? resolveTagMapping(connectorType, extDoc.metadata, sourceConfig)
    : undefined

  const processingFilename = `${safeTitle}.txt`

  try {
    await db.transaction(async (tx) => {
      const isActive = await isKnowledgeBaseActiveInTx(tx, knowledgeBaseId)
      if (!isActive) {
        throw new Error(`Knowledge base ${knowledgeBaseId} is deleted`)
      }

      await tx.insert(document).values({
        id: documentId,
        knowledgeBaseId,
        filename: extDoc.title,
        fileUrl,
        storageKey: fileInfo.key,
        fileSize: contentBuffer.length,
        mimeType: 'text/plain',
        chunkCount: 0,
        tokenCount: 0,
        characterCount: 0,
        processingStatus: 'pending',
        enabled: true,
        connectorId,
        externalId: extDoc.externalId,
        contentHash: extDoc.contentHash,
        sourceUrl: extDoc.sourceUrl ?? null,
        ...tagValues,
        uploadedAt: new Date(),
      })