Function processDocumentAsync

apps/sim/lib/knowledge/documents/service.ts:492–848

(
  knowledgeBaseId: string,
  documentId: string,
  docData: {
    filename: string
    fileUrl: string
    fileSize: number
    mimeType: string
  },
  processingOptions: ProcessingOptions = {}
)

}

export async function processDocumentAsync(
  knowledgeBaseId: string,
  documentId: string,
  docData: {
    filename: string
    fileUrl: string
    fileSize: number
    mimeType: string
  },
  processingOptions: ProcessingOptions = {}
): Promise<void> {
  const startTime = Date.now()
  try {
    logger.info(`[${documentId}] Starting document processing: ${docData.filename}`)

    // KB config + workspace billing + doc tags in one JOIN (was 3 SELECTs).
    const contextRows = await db
      .select({
        workspaceId: knowledgeBase.workspaceId,
        chunkingConfig: knowledgeBase.chunkingConfig,
        embeddingModel: knowledgeBase.embeddingModel,
        billedAccountUserId: workspaceTable.billedAccountUserId,
        uploadedBy: document.uploadedBy,
        tag1: document.tag1,
        tag2: document.tag2,
        tag3: document.tag3,
        tag4: document.tag4,
        tag5: document.tag5,
        tag6: document.tag6,
        tag7: document.tag7,
        number1: document.number1,
        number2: document.number2,
        number3: document.number3,
        number4: document.number4,
        number5: document.number5,
        date1: document.date1,
        date2: document.date2,
        boolean1: document.boolean1,
        boolean2: document.boolean2,
        boolean3: document.boolean3,
      })
      .from(document)
      .innerJoin(knowledgeBase, eq(knowledgeBase.id, document.knowledgeBaseId))
      .leftJoin(
        workspaceTable,
        and(eq(workspaceTable.id, knowledgeBase.workspaceId), isNull(workspaceTable.archivedAt))
      )
      .where(
        and(
          eq(document.id, documentId),
          eq(knowledgeBase.id, knowledgeBaseId),
          isNull(document.archivedAt),
          isNull(document.deletedAt),
          isNull(knowledgeBase.deletedAt)
        )
      )
      .limit(1)