Function videoToDocument

apps/sim/connectors/youtube/youtube.ts:271–310

* Builds the full document for a video, combining title and description as plain-text * content. Returns null for unlisted/private/deleted videos, and (when configured) for * Shorts shorter than 60 seconds. * * Captions/transcripts are intentionally not fetched: `captions.download` requires OAut

(video: VideoItem, excludeShorts: boolean)

 * the video title plus description only.
 */
function videoToDocument(video: VideoItem, excludeShorts: boolean): ExternalDocument | null {
  const videoId = video.id
  if (!videoId) return null

  const privacyStatus = video.status?.privacyStatus
  if (privacyStatus && privacyStatus !== 'public' && privacyStatus !== 'unlisted') {
    return null
  }

  if (excludeShorts) {
    const seconds = parseIso8601Duration(video.contentDetails?.duration)
    if (seconds != null && seconds > 0 && seconds < SHORTS_MAX_DURATION_SECONDS) {
      return null
    }
  }

  const snippet = video.snippet ?? {}
  const title = snippet.title || 'Untitled'
  const description = snippet.description ?? ''
  const publishedAt = snippet.publishedAt ?? ''
  const content = description.trim() ? `${title}\n\n${description}` : title
  const tags = Array.isArray(snippet.tags) ? snippet.tags : []

  return {
    externalId: videoId,
    title,
    content,
    contentDeferred: false,
    mimeType: 'text/plain',
    sourceUrl: `https://www.youtube.com/watch?v=${videoId}`,
    contentHash: buildContentHash(videoId, publishedAt),
    metadata: {
      channelTitle: snippet.channelTitle ?? '',
      publishedAt,
      duration: video.contentDetails?.duration ?? '',
      categoryId: snippet.categoryId ?? '',
      tags,
    },
  }
}

export const youtubeConnector: ConnectorConfig = {
  ...youtubeConnectorMeta,