MongoDb 聚合数据操作 - 对象到数组答案

【问题标题】：MongoDb Aggregation Data manipulation - Objects to ArraysMongoDb 聚合数据操作 - 对象到数组
【发布时间】：2021-11-03 21:29:18
【问题描述】：

我有以下示例数据集

[{
  "_id": {
    "$oid": "60f83d3cd66842301905aa77"
  },
  "id": 527438,
  "name": "CryptoPunk #4050",
  "asset_contract": {
    "name": "CryptoPunks",
    "address": "0xb47e3cd837ddf8e4c57f05d70ab865de6e193bbb"
  },
  "traits": [
    {
      "trait_type": "type",
      "value": "Male",
      "display_type": null,
      "max_value": null,
      "trait_count": 6039,
      "order": null
    },
    {
      "trait_type": "accessory",
      "value": "Mohawk",
      "display_type": null,
      "max_value": null,
      "trait_count": 441,
      "order": null
    },
    {
      "trait_type": "accessory",
      "value": "Earring",
      "display_type": null,
      "max_value": null,
      "trait_count": 2459,
      "order": null
    },
    {
      "trait_type": "accessory",
      "value": "Frown",
      "display_type": null,
      "max_value": null,
      "trait_count": 261,
      "order": null
    }
  ],
  "token_id": "4050",
  "permalink": "https://opensea.io/assets/0xb47e3cd837ddf8e4c57f05d70ab865de6e193bbb/4050",
  "background_color": null,
  "image_url": "https://lh3.googleusercontent.com/sO18rDQYhC5yIcj12RVsv31pbbsZo_2muQQbTJMQHn47EKGhnirs8mxzohm58HAZ7taBoe4pU6x1qntlExk_TtJ-",
  "image_preview_url": "https://lh3.googleusercontent.com/sO18rDQYhC5yIcj12RVsv31pbbsZo_2muQQbTJMQHn47EKGhnirs8mxzohm58HAZ7taBoe4pU6x1qntlExk_TtJ-=s250",
  "animation_url": null,
  "vault_contract": "0x269616d549d7e8eaa82dfb17028d0b212d11232a"
},{
  "_id": { "$oid": "60f83d3cbc3f0161da2141f7" },
  "id": 17736625,
  "name": "OJ Simpson",
  "asset_contract": {
    "name": "Hashmasks",
    "address": "0xc2c747e0f7004f9e8817db2ca4997657a7746928"
  },
  "traits": [
    {
      "trait_type": "Character",
      "value": "Male",
      "display_type": null,
      "max_value": null,
      "trait_count": 8659,
      "order": null
    },
    {
      "trait_type": "Mask",
      "value": "Doodle",
      "display_type": null,
      "max_value": null,
      "trait_count": 2187,
      "order": null
    },
    {
      "trait_type": "Eye Color",
      "value": "Dark",
      "display_type": null,
      "max_value": null,
      "trait_count": 7419,
      "order": null
    },
    {
      "trait_type": "Item",
      "value": "No Item",
      "display_type": null,
      "max_value": null,
      "trait_count": 14533,
      "order": null
    },
    {
      "trait_type": "Skin Color",
      "value": "Dark",
      "display_type": null,
      "max_value": null,
      "trait_count": 3784,
      "order": null
    },
    {
      "trait_type": "Token ID",
      "value": 3535,
      "display_type": "number",
      "max_value": null,
      "trait_count": 0,
      "order": null
    },
    {
      "trait_type": "Background",
      "value": "Doodle",
      "display_type": null,
      "max_value": null,
      "trait_count": 5538,
      "order": null
    }
  ],
  "token_id": "3535",
  "permalink": "https://opensea.io/assets/0xc2c747e0f7004f9e8817db2ca4997657a7746928/3535",
  "background_color": null,
  "image_url": "https://lh3.googleusercontent.com/NZQu7CNjgJ_1uhbUVwEb-14rZPJmPCaqaXy0qnUpgm5Qll0BvmmF7tPMjBhFH6ZZp_qzOPxHi0NFmRkOjHoBQ0BODcWI8NlyBXLu",
  "image_preview_url": "https://lh3.googleusercontent.com/NZQu7CNjgJ_1uhbUVwEb-14rZPJmPCaqaXy0qnUpgm5Qll0BvmmF7tPMjBhFH6ZZp_qzOPxHi0NFmRkOjHoBQ0BODcWI8NlyBXLu=s250",
  "animation_url": null,
  "vault_contract": "0xc7a8b45e184138114e6085c82936a8db93dd156a"
}]

我想更新到的

[{
    "_id": {
      "$oid": "60f83d3cd66842301905aa77"
    },
    "id": 527438,
    "name": "CryptoPunk #4050",
    "asset_contract": {
      "name": "CryptoPunks",
      "address": "0xb47e3cd837ddf8e4c57f05d70ab865de6e193bbb"
    },
    "traits":
      {
        "type": "Male",
        "accessory": ["Mohawk", "Earing", "Frown"], 
      },
    "token_id": "4050",
    "permalink": "https://opensea.io/assets/0xb47e3cd837ddf8e4c57f05d70ab865de6e193bbb/4050",
    "background_color": null,
    "image_url": "https://lh3.googleusercontent.com/sO18rDQYhC5yIcj12RVsv31pbbsZo_2muQQbTJMQHn47EKGhnirs8mxzohm58HAZ7taBoe4pU6x1qntlExk_TtJ-",
    "image_preview_url": "https://lh3.googleusercontent.com/sO18rDQYhC5yIcj12RVsv31pbbsZo_2muQQbTJMQHn47EKGhnirs8mxzohm58HAZ7taBoe4pU6x1qntlExk_TtJ-=s250",
    "animation_url": null,
    "vault_contract": "0x269616d549d7e8eaa82dfb17028d0b212d11232a"
  },{
  "_id": { "$oid": "60f83d3cbc3f0161da2141f7" },
  "id": 17736625,
  "name": "OJ Simpson",
  "asset_contract": {
    "name": "Hashmasks",
    "address": "0xc2c747e0f7004f9e8817db2ca4997657a7746928"
  },
  "traits": {
      "character": "Male",
      "mask": "Doodle",
      "eye_color": "Dark",
      "item": "No Item",
      "skin_color": "Dark",
      "token_id": 3535,
      "background": "Doodle",
    },
  "token_id": "3535",
  "permalink": "https://opensea.io/assets/0xc2c747e0f7004f9e8817db2ca4997657a7746928/3535",
  "background_color": null,
  "image_url": "https://lh3.googleusercontent.com/NZQu7CNjgJ_1uhbUVwEb-14rZPJmPCaqaXy0qnUpgm5Qll0BvmmF7tPMjBhFH6ZZp_qzOPxHi0NFmRkOjHoBQ0BODcWI8NlyBXLu",
  "image_preview_url": "https://lh3.googleusercontent.com/NZQu7CNjgJ_1uhbUVwEb-14rZPJmPCaqaXy0qnUpgm5Qll0BvmmF7tPMjBhFH6ZZp_qzOPxHi0NFmRkOjHoBQ0BODcWI8NlyBXLu=s250",
  "animation_url": null,
  "vault_contract": "0xc7a8b45e184138114e6085c82936a8db93dd156a"
}]

其背后的逻辑是

查看 Traits 数组对象
获取 trait_type 值并使用小写名称（空格下划线）创建新键
将新键的值设置为"value"的值

所以，

      "trait_type": "type",
      "value": "Male",

//becomes

            "type": "Male"

如果有多个相同特征类型的实例，请创建一个值数组。

所以，

{
      "trait_type": "accessory",
      "value": "Mohawk",
      "display_type": null,
      "max_value": null,
      "trait_count": 441,
      "order": null
    },
    {
      "trait_type": "accessory",
      "value": "Earring",
      "display_type": null,
      "max_value": null,
      "trait_count": 2459,
      "order": null
    },

// becomes

            "accessory": ["Mohawk", "Earring"]

【问题讨论】：

标签： json mongodb elasticsearch aggregation-framework algolia

【解决方案1】：

查询

即使管道不允许我们使用此处使用的组查找等阶段，它也是一个聚合更新。（您可以使用$out 和替换集合之后或$merge 来替换文档（类似于更新））
第一张地图
- 对于每个特征（特征的文档成员），将其放入数组
  [["trait_type": "type"] ["value": "Male"] ["display_type": null] ...]
- 减少该数组以仅从它们构造 1 个文档
  {"type" "type","value" :"Male"}（小写字母和“_”也是如此）

现在的特征就像

"traits": [
  {
    "type": "type",
    "value": "Male"
  },
  {
    "type": "accessory",
    "value": "Mohawk"
  },
  {
    "type": "accessory",
    "value": "Earring"
  },
  {
    "type": "accessory",
    "value": "Frown"
  }
]

使用虚拟集合 [{}] 进行查找（我们这样做是为了在该数组中创建一个组），这就像一个技巧，允许我们在 1 个文档中使用阶段运算符
- 查找管道按类型展开和分组
```
"traits": [
  {
    "values": [
      "Mohawk",
      "Earring",
      "Frown"
    ],
    "type": "accessory"
  },
  {
    "values": [
      "Male"
    ],
    "type": "type"
  }
]
```
- 然后它是一个替换根来获取类型的值，使其成为字段名称和值作为值(if size=1 removes the array)

查找后我们有

"traits": [
  {
    "accessory": [
      "Mohawk",
      "Earring",
      "Frown"
    ]
  },
  {
    "type": "Male"
  }
]

所以我们要做的就是减少这些特征并合并对象（无论如何，键都是唯一的，因为我们按它们分组）
我们得到了预期的输出（至少我认为还可以）

Test code here

db.collection.aggregate([
  {
    "$set": {
      "traits": {
        "$map": {
          "input": "$traits",
          "as": "t",
          "in": {
            "$reduce": {
              "input": {
                "$map": {
                  "input": {
                    "$objectToArray": "$$t"
                  },
                  "as": "m",
                  "in": [
                    "$$m.k",
                    "$$m.v"
                  ]
                }
              },
              "initialValue": {},
              "in": {
                "$let": {
                  "vars": {
                    "type_value": "$$value",
                    "ta": "$$this"
                  },
                  "in": {
                    "$let": {
                      "vars": {
                        "key": {
                          "$arrayElemAt": [
                            "$$ta",
                            0
                          ]
                        },
                        "value": {
                          "$arrayElemAt": [
                            "$$ta",
                            1
                          ]
                        }
                      },
                      "in": {
                        "$switch": {
                          "branches": [
                            {
                              "case": {
                                "$eq": [
                                  "$$key",
                                  "value"
                                ]
                              },
                              "then": {
                                "$mergeObjects": [
                                  "$$type_value",
                                  {
                                    "value": "$$value"
                                  }
                                ]
                              }
                            },
                            {
                              "case": {
                                "$eq": [
                                  "$$key",
                                  "trait_type"
                                ]
                              },
                              "then": {
                                "$mergeObjects": [
                                  "$$type_value",
                                  {
                                    "type": {
                                      "$replaceAll": {
                                        "input": {
                                          "$toLower": "$$value"
                                        },
                                        "find": " ",
                                        "replacement": "_"
                                      }
                                    }
                                  }
                                ]
                              }
                            }
                          ],
                          "default": "$$type_value"
                        }
                      }
                    }
                  }
                }
              }
            }
          }
        }
      }
    }
  },
  {
    "$lookup": {
      "from": "dummy",
      "let": {
        "traits": "$traits"
      },
      "pipeline": [
        {
          "$set": {
            "traits": "$$traits"
          }
        },
        {
          "$unwind": {
            "path": "$traits"
          }
        },
        {
          "$replaceRoot": {
            "newRoot": "$traits"
          }
        },
        {
          "$group": {
            "_id": "$type",
            "values": {
              "$push": "$value"
            }
          }
        },
        {
          "$set": {
            "type": "$_id"
          }
        },
        {
          "$project": {
            "_id": 0
          }
        },
        {
          "$replaceRoot": {
            "newRoot": {
              "$cond": [
                {
                  "$eq": [
                    {
                      "$size": "$values"
                    },
                    1
                  ]
                },
                {
                  "$arrayToObject": {
                    "$let": {
                      "vars": {
                        "pair": [
                          [
                            "$type",
                            {
                              "$arrayElemAt": [
                                "$values",
                                0
                              ]
                            }
                          ]
                        ]
                      },
                      "in": "$$pair"
                    }
                  }
                },
                {
                  "$arrayToObject": {
                    "$let": {
                      "vars": {
                        "pair": [
                          [
                            "$type",
                            "$values"
                          ]
                        ]
                      },
                      "in": "$$pair"
                    }
                  }
                }
              ]
            }
          }
        }
      ],
      "as": "traits"
    }
  },
  {
    "$set": {
      "traits": {
        "$mergeObjects": "$traits"
      }
    }
  }
])

【讨论】：

非常感谢@Takis_。我正在运行它的集合有 57k 条记录，我无法将 dummy 附加到记录集的末尾。当我尝试使用 Mongo shell（以及 Compass Aggregation）运行它时，我也收到了 Invalid $set :: caused by :: Unrecognized expression '$setField'
实际用途是在一个集合上执行聚合并将输出移动到另一个集合，因此db.rawCollection.aggregation() 并输出所有结果以更新辅助集合db.cleanCollection。这将使用脚本每 5 分钟运行一次。
为此，您还需要 1 个名为 dummy 的数据库集合，它将有 1 个空文档 collection=[{}]（或任何名称，只要您在查询时也更改它）。 $setField 错误是因为您没有运行 MongoDB 5，没关系，还有另一种方法，我将更新查询。在您的情况下，您需要 $out 保存到另一个集合。
我更新了查询删除了$setField 将其替换为$mergeObjects 它很好，虚拟是您数据库中的一个新集合，只有一个文档空文档dummy=[{}] 您只需要权限创建该集合。你可以给它任何你想要的名字，但你也必须在$lookup 中替换那个名字。对于$out，看看如何在你的驱动程序中使用它，它将结果输出到一个新的集合中（不更新）新的集合将成为聚合的结果。
感谢 Takis，我又奖励了 50 分并分配给您，非常感谢您一直以来的支持。