几个小时以来，我一直在努力思考如何做到这一点，我有一个名为“Jobs”的集合——在内部，它们有一个子集合“Site”，即 Jobs.site。此站点子集合具有属性“UNID”。

我正在尝试基于文本搜索从数据库中检索文档，效果很好。

但我试图仅检索基于该 Job.Site.UNID 的 UNIQUE 文档，并且可能添加了一个计数作为额外属性。结果应如下所示:

工作:{ 站点:{ field1:'EXAMPLE'，UNID:'SITEID'，计数:5 }}

这意味着作业集合中有 5 个作业具有该 site.UNID。

这是我目前所拥有的:

[
// GETTING DOCS BASED ON TEXT SEARCH RESULTS
    {
      $match: {
        // clientId: req.user.client_id,
        $text: { $search: body.searchTerms }
      }
    },
// SORTING THEM BASED ON TEXTSCORE
    { $sort: { score: { $meta: 'textScore' } } },
// THE PROBLEMATIC GROUPING PART
    { $group: { site: { UPRN: '$UPRN', myCount: { $sum: 1 } } } },
// I ONYL WANT TO GET 20 DOCS AT A TIME
    { $limit: 20 },
// THE DATA THAT I WANT IN MY DOCUMENTS, MAYBE COUNT WOULD COME HERE?
    {
      $project: {
        site: true,
        score: { $meta: 'textScore' }
      }
    },
// GETTING RID OF POOR MATCHES BASED ON A SCORE CALCULATED IN ANOTHER 
// FUNCTION BASED ON THE NUMBER OF WORDS IN THE TEXT SEARCH
    {
      $match: {
        score: { $gt: matchScore }
      }
    }
  ]

这里让我印象深刻的是 The field 'site' must be an accumulator object

所以我想不出正确处理该子集合属性的语法。

编辑:感谢@Anthony，V2 完美地工作并且已经对其进行了彻底的测试，除了它似乎没有计算工作总数，它总是 1 或我在 $sum 中设置的任何值:但是有 200 多个结果，仍在工作在上面。

 { $match: { $text: { $search: body.searchTerms } } },
    { $sort: { $score: { $meta: 'textScore' } } },
    // { $match: { score: { $gt: 0.1 } } },
    {
      $group: {
        _id: '$UNID',
        counter: { $sum: 1 },
        score: { $first: { $meta: 'textScore' } },
        title: { $first: '$title' },
        postcode: { $first: '$postcode' },
        addressLine1: { $first: '$addressLine1' },
        city: { $first: '$city' },
        projectName: { $first: '$projectName' },
        jobsCount: { $sum: '$counter' }
      }
    },
    { $limit: 20 },
    {
      $project: {
        UNID: '$_id',
        title: '$title',
        postcode: '$postcode',
        addressLine1: '$addressLine1',
        projectName: '$projectName',
        city: '$city',
        score: 1,
        jobsCount: true
      }
    }

示例数据:

{
  "_id": "randomString0",
  "title": "Quality",
  "site": {
    "_id": "rKFRbvH8CEbJYdzDs",
    "title": "Title 1",
    "addressLine1": "address1",
    "UNID": "001",
    "city": "cityName",
    "createdAt": null
  }
},
{
  "_id": "randomString1",
  "title": "Some2123",
  "site": {
    "_id": "rKFRbvH8CEbJYdzDs",
    "title": "Title 1",
    "addressLine1": "address1",
    "UNID": "001",
    "city": "cityName",
    "createdAt": null
  }
},
{
  "_id": "randomString2",
  "title": "Random title",
  "site": {
    "_id": "rKFRbvH8CEbJYdzDs",
    "title": "Title 1",
    "addressLine1": "address1",
    "UNID": "001",
    "city": "cityName",
    "createdAt": null
  }
},
{
  "_id": "randomString3",
  "title": "Another unique job",
  "site": {
    "_id": "rKFRbvH8CEbJYdzDs",
    "title": "Title 1",
    "addressLine1": "address1",
    "UNID": "001",
    "city": "cityName",
    "createdAt": null
  }
},
{
  "_id": "randomString4",
  "title": "Other thing",
  "site": {
    "_id": "rKFRbvH8CEbJYdzDs",
    "title": "Title 1",
    "addressLine1": "address1",
    "UNID": "001",
    "city": "cityName",
    "createdAt": null
  }
},
{
  "_id": "randomString5",
  "title": "Something else",
  "site": {
    "_id": "rKFRbvH8CEbJYdzDs",
    "title": "Title 1",
    "addressLine1": "address1",
    "UNID": "001",
    "city": "cityName",
    "createdAt": null
  }
}

如您所见，站点数据在所有这 5 个文档下始终是唯一的，但是计数器应该计算有多少文档具有相同的唯一性

最佳答案

在$group暂存 _id(您要分组的依据)表达式是必需的表达式。而且只有少数accumulators你可以使用 $group聚合阶段。

所以你的聚合一定是这样的

[
  { "$match": { "$text": { "$search": body.searchTerms }}},
  { "$sort": { "score": { "$meta": "textScore" } } },
  { "$match": { "score": { "$gt": matchScore }}},
  { "$group": {
    "_id": "$UPRN",
    "myCount": { "$sum": 1 },
    "score": { "$first": "$score" }
  }},
  { "$limit": 20 },
  { "$project": {
    "site": "$_id",
    "score": 1,
    "myCount": 1
  }}
]

关于node.js - Mongodb 基于具有计数的唯一子集合属性聚合搜索结果？，我们在Stack Overflow上找到一个类似的问题： https://stackoverflow.com/questions/55636225/