sepia-search-motore-di-ricerca-video/server/lib/elastic-search-videos.ts

import { difference } from 'lodash'
import { buildIndex, buildSort, elasticSearch, extractQueryResult, indexDocuments } from '../helpers/elastic-search'
import { logger } from '../helpers/logger'
import { buildUrl } from '../helpers/utils'
import { CONFIG } from '../initializers/constants'
import { VideosSearchQuery } from '../types/video-search.model'
import { DBVideo, DBVideoDetails, EnhancedVideo, IndexableVideo, IndexableVideoDetails } from '../types/video.model'
import { buildAvatarMapping, formatAvatarForAPI, formatAvatarForDB } from './elastic-search-avatar'

function initVideosIndex () {
  return buildIndex(CONFIG.ELASTIC_SEARCH.INDEXES.VIDEOS, buildVideosMapping())
}

async function indexVideos (videos: IndexableVideo[], replace = false) {
  return indexDocuments({
    objects: videos,
    formatter: v => formatVideoForDB(v),
    replace,
    index: CONFIG.ELASTIC_SEARCH.INDEXES.VIDEOS
  })
}

function refreshVideosIndex () {
  return elasticSearch.indices.refresh({ index: CONFIG.ELASTIC_SEARCH.INDEXES.VIDEOS })
}

function removeVideosFromHosts (hosts: string[]) {
  if (hosts.length === 0) return

  logger.info({ hosts }, 'Will remove videos from hosts.')

  return elasticSearch.delete_by_query({
    index: CONFIG.ELASTIC_SEARCH.INDEXES.VIDEOS,
    body: {
      query: {
        bool: {
          filter: {
            terms: {
              host: hosts
            }
          }
        }
      }
    }
  })
}

async function removeNotExistingVideos (host: string, existingVideos: Set<number>) {
  const idsFromDB = await getVideoIdsOf(host)

  const idsToRemove = difference(idsFromDB, Array.from(existingVideos))

  logger.info({ idsToRemove }, 'Will remove %d videos from %s.', idsToRemove.length, host)

  return elasticSearch.delete_by_query({
    index: CONFIG.ELASTIC_SEARCH.INDEXES.VIDEOS,
    body: {
      query: {
        bool: {
          filter: [
            {
              terms: {
                id: idsToRemove
              }
            },
            {
              term: {
                host
              }
            }
          ]
        }
      }
    }
  })
}

async function getVideoIdsOf (host: string) {
  const res = await elasticSearch.search({
    index: CONFIG.ELASTIC_SEARCH.INDEXES.VIDEOS,
    body: {
      size: 0,
      aggs: {
        ids: {
          terms: {
            size: 500000,
            field: 'id'
          }
        }
      },
      query: {
        bool: {
          filter: [
            {
              term: {
                host
              }
            }
          ]
        }
      }
    }
  })

  return res.body.aggregations.ids.buckets.map(b => b.key)
}

async function queryVideos (search: VideosSearchQuery) {
  const bool: any = {}
  const filter: any[] = []
  const mustNot: any[] = []

  if (search.search) {
    Object.assign(bool, {
      must: [
        {
          multi_match: {
            query: search.search,
            fields: [ 'name', 'description', 'tags' ],
            fuzziness: 'AUTO'
          }
        }
      ]
    })
  }

  if (search.blockedAccounts) {
    mustNot.push({
      terms: {
        'account.handle': search.blockedAccounts
      }
    })
  }

  if (search.blockedHosts) {
    mustNot.push({
      terms: {
        host: search.blockedHosts
      }
    })
  }

  if (search.startDate) {
    filter.push({
      range: {
        publishedAt: {
          gte: search.startDate
        }
      }
    })
  }

  if (search.endDate) {
    filter.push({
      range: {
        publishedAt: {
          lte: search.endDate
        }
      }
    })
  }

  if (search.originallyPublishedStartDate) {
    filter.push({
      range: {
        originallyPublishedAt: {
          gte: search.startDate
        }
      }
    })
  }

  if (search.originallyPublishedEndDate) {
    filter.push({
      range: {
        originallyPublishedAt: {
          lte: search.endDate
        }
      }
    })
  }

  if (search.nsfw && search.nsfw !== 'both') {
    filter.push({
      term: {
        nsfw: (search.nsfw + '') === 'true'
      }
    })
  }

  if (search.categoryOneOf) {
    filter.push({
      terms: {
        'category.id': search.categoryOneOf
      }
    })
  }

  if (search.licenceOneOf) {
    filter.push({
      terms: {
        'licence.id': search.licenceOneOf
      }
    })
  }

  if (search.languageOneOf) {
    filter.push({
      terms: {
        'language.id': search.languageOneOf
      }
    })
  }

  if (search.tagsOneOf) {
    filter.push({
      terms: {
        tags: search.tagsOneOf
      }
    })
  }

  if (search.tagsAllOf) {
    for (const t of search.tagsAllOf) {
      filter.push({
        term: {
          tags: t
        }
      })
    }
  }

  if (search.durationMin) {
    filter.push({
      range: {
        duration: {
          gte: search.durationMin
        }
      }
    })
  }

  if (search.durationMax) {
    filter.push({
      range: {
        duration: {
          lte: search.durationMax
        }
      }
    })
  }

  Object.assign(bool, { filter })

  if (mustNot.length !== 0) {
    Object.assign(bool, { must_not: mustNot })
  }

  const body = {
    from: search.start,
    size: search.count,
    sort: buildSort(search.sort),
    query: { bool }
  }

  logger.debug({ body }, 'Will query Elastic Search for videos.')

  const res = await elasticSearch.search({
    index: CONFIG.ELASTIC_SEARCH.INDEXES.VIDEOS,
    body
  })

  return extractQueryResult(res)
}

export {
  indexVideos,
  removeNotExistingVideos,
  queryVideos,
  refreshVideosIndex,
  removeVideosFromHosts,
  initVideosIndex,
  formatVideoForAPI
}

// ############################################################################

function formatVideoForDB (v: IndexableVideo | IndexableVideoDetails): DBVideo | DBVideoDetails {
  return {
    id: v.id,
    uuid: v.uuid,

    indexedAt: new Date(),
    createdAt: v.createdAt,
    updatedAt: v.updatedAt,
    publishedAt: v.publishedAt,
    originallyPublishedAt: v.originallyPublishedAt,

    category: {
      id: v.category.id,
      label: v.category.label
    },
    licence: {
      id: v.licence.id,
      label: v.licence.label
    },
    language: {
      id: v.language.id,
      label: v.language.label
    },
    privacy: {
      id: v.privacy.id,
      label: v.privacy.label
    },

    name: v.name,
    description: v.description,
    duration: v.duration,
    thumbnailPath: v.thumbnailPath,
    previewPath: v.previewPath,
    embedPath: v.embedPath,

    views: v.views,
    likes: v.likes,
    dislikes: v.dislikes,

    nsfw: v.nsfw,
    host: v.host,
    url: v.url,

    tags: (v as IndexableVideoDetails).tags ? (v as IndexableVideoDetails).tags : undefined,

    account: {
      id: v.account.id,
      name: v.account.name,
      displayName: v.account.displayName,
      url: v.account.url,
      host: v.account.host,

      handle: `${v.account.name}@${v.account.host}`,

      avatar: formatAvatarForDB(v.account)
    },

    channel: {
      id: v.channel.id,
      name: v.channel.name,
      displayName: v.channel.displayName,
      url: v.channel.url,
      host: v.channel.host,

      handle: `${v.channel.name}@${v.channel.host}`,

      avatar: formatAvatarForDB(v.channel)
    }
  }
}

function formatVideoForAPI (v: DBVideoDetails, fromHost?: string): EnhancedVideo {
  return {
    id: v.id,
    uuid: v.uuid,

    score: v.score,

    createdAt: new Date(v.createdAt),
    updatedAt: new Date(v.updatedAt),
    publishedAt: new Date(v.publishedAt),
    originallyPublishedAt: v.originallyPublishedAt,

    category: {
      id: v.category.id,
      label: v.category.label
    },
    licence: {
      id: v.licence.id,
      label: v.licence.label
    },
    language: {
      id: v.language.id,
      label: v.language.label
    },
    privacy: {
      id: v.privacy.id,
      label: v.privacy.label
    },

    name: v.name,
    description: v.description,
    duration: v.duration,

    tags: v.tags,

    thumbnailPath: v.thumbnailPath,
    thumbnailUrl: buildUrl(v.host, v.thumbnailPath),

    previewPath: v.previewPath,
    previewUrl: buildUrl(v.host, v.previewPath),

    embedPath: v.embedPath,
    embedUrl: buildUrl(v.host, v.embedPath),

    url: v.url,

    isLocal: fromHost && fromHost === v.host,

    views: v.views,
    likes: v.likes,
    dislikes: v.dislikes,

    nsfw: v.nsfw,

    account: {
      id: v.account.id,
      name: v.account.name,
      displayName: v.account.displayName,
      url: v.account.url,
      host: v.account.host,

      avatar: formatAvatarForAPI(v.account)
    },

    channel: {
      id: v.channel.id,
      name: v.channel.name,
      displayName: v.channel.displayName,
      url: v.channel.url,
      host: v.channel.host,

      avatar: formatAvatarForAPI(v.channel)
    }
  }
}

function buildChannelOrAccountMapping () {
  return {
    id: {
      type: 'long'
    },

    name: {
      type: 'text',
      fields: {
        raw: {
          type: 'keyword'
        }
      }
    },
    displayName: {
      type: 'text'
    },
    url: {
      type: 'keyword'
    },
    host: {
      type: 'keyword'
    },
    handle: {
      type: 'keyword'
    },

    avatar: {
      properties: buildAvatarMapping()
    }
  }
}

function buildVideosMapping () {
  return {
    id: {
      type: 'long'
    },

    uuid: {
      type: 'keyword'
    },
    createdAt: {
      type: 'date',
      format: 'date_optional_time'
    },
    updatedAt: {
      type: 'date',
      format: 'date_optional_time'
    },
    publishedAt: {
      type: 'date',
      format: 'date_optional_time'
    },
    originallyPublishedAt: {
      type: 'date',
      format: 'date_optional_time'
    },
    indexedAt: {
      type: 'date',
      format: 'date_optional_time'
    },

    category: {
      properties: {
        id: {
          type: 'keyword'
        },
        label: {
          type: 'text'
        }
      }
    },

    licence: {
      properties: {
        id: {
          type: 'keyword'
        },
        label: {
          type: 'text'
        }
      }
    },

    language: {
      properties: {
        id: {
          type: 'keyword'
        },
        label: {
          type: 'text'
        }
      }
    },

    privacy: {
      properties: {
        id: {
          type: 'keyword'
        },
        label: {
          type: 'text'
        }
      }
    },

    name: {
      type: 'text'
    },

    description: {
      type: 'text'
    },

    tags: {
      type: 'text',

      fields: {
        raw: {
          type: 'keyword'
        }
      }
    },

    duration: {
      type: 'long'
    },

    thumbnailPath: {
      type: 'keyword'
    },
    previewPath: {
      type: 'keyword'
    },
    embedPath: {
      type: 'keyword'
    },

    url: {
      type: 'keyword'
    },

    views: {
      type: 'long'
    },
    likes: {
      type: 'long'
    },
    dislikes: {
      type: 'long'
    },
    nsfw: {
      type: 'boolean'
    },

    host: {
      type: 'keyword'
    },

    account: {
      properties: buildChannelOrAccountMapping()
    },

    channel: {
      properties: buildChannelOrAccountMapping()
    }
  }
}