C# Mongodb 多对象数组文档的笛卡尔积

Question

尝试使用 C#/Linq 甚至原始 Mongodb 查询本身了解如何将多个数组作为笛卡尔积连接。

例如，我有一个集合，我过滤到以下两个文档：

[
{"movie":"starwars","showday":"monday"},
{"movie":"batman","showday":"thursday"},
{"movie":"sleepless","showday":"tuesday"}
]

[
{"actor":"angelina","location":"new york"},
{"actor":"jamie","location":"california"},
{"actor":"mcavoy","location":"arizona"}
]

如何加入每个数组中的每个项目以产生以下类型的结果？

[{"movie":"starwars","showday":"monday","actor":"angelina","location":"new york"},
{"movie":"batman","showday":"thursday","actor":"angelina","location":"new york"},
{"movie":"sleepless","showday":"tuesday","actor":"angelina","location":"new york"},
{"movie":"starwars","showday":"monday","actor":"jamie","location":"california"},
{"movie":"batman","showday":"thursday","actor":"jamie","location":"california"},
{"movie":"sleepless","showday":"tuesday","actor":"jamie","location":"california"},
{"movie":"starwars","showday":"monday","actor":"mcavoy","location":"arizona"},
{"movie":"batman","showday":"thursday","actor":"mcavoy","location":"arizona"},
{"movie":"sleepless","showday":"tuesday","actor":"mcavoy","location":"arizona"}]

我正在寻找一种可以处理任意数量文档的解决方案。因此，例如，如果在本例中，第 3 个文档也有 3 个对象数组，这些对象数组将在数组中生成一个包含 27 个项目的结果集 - 或者说是 27 行。

希望找到如何使用 C#（Linq？）Mongodb 驱动程序来查询和返回这样的数据的解决方案，但即使是特定于 mongodb 的查询也将开放，因为我希望可以从那里反转逻辑。谢谢

Answer 1

您可以尝试以下聚合管道。

注意mergeObjects 聚合运算符在3.5.6 + 开发版本中可用，该版本将在即将推出的3.6 版本中推出。

db.collection.find();
{
 "data" : [
  [
   {
    "movie" : "starwars",
    "showday" : "monday"
   },
   {
    "movie" : "batman",
    "showday" : "thursday"
   },
   {
    "movie" : "sleepless",
    "showday" : "tuesday"
   }
  ],
  [
   {
    "actor" : "angelina",
    "location" : "new york"
   },
   {
    "actor" : "jamie",
    "location" : "california"
   },
   {
    "actor" : "mcavoy",
    "location" : "arizona"
   }
  ]
 ]
}

使用条件表达式进行聚合。

aggregate({
 $project: {
  cp: {
   $reduce: {
    input: "$data",
    initialValue: {
     $arrayElemAt: ["$data", 0] // Set the initial value to the first element of the arrays.
    },
    in: {
     $let: {
      vars: {
       currentr: "$$this", // Current processing element
       currenta: "$$value" // Current accumulated value 
      },
      in: {
       $cond: [{ // Conditional expression to return the accumulated value as initial value for first element
        $eq: ["$$currentr", "$$currenta"]
       },
       "$$currenta",
       { // From second element onwards prepare the cartesian product
        $reduce: {
         input: {
          $map: {
           input: "$$currenta",
           as: a"a",
           in: {
            $map: {
             input: "$$currentr",
             as: r"r",
             in: {
              $mergeObjects: ["$$a", "$$r"] // Merge accumulated value with the current processing element
             }
            }
           }
          }
         },
         initialValue: [],
         in: {
         $concatArrays: ["$$value", "$$this"] // Reduce the merged values which will be used as accumulator for next element
         }
        }
       }]
      }
     }
    }
   }
  }
 }
});

聚合（使用$setUnion）。

添加此解决方案只是为了抑制条件表达式以提供更具可读性的管道。

aggregate({
 $project: {
  cp: {
   $reduce: {
    input: "$data",
    initialValue: {
     $arrayElemAt: ["$data", 0] // Set the initial value to the first element of the arrays.
    },
    in: {
     $let: {
      vars: {
       currentr: "$$this", // Current processing element
       currenta: "$$value" // Current accumulated value 
      },
      in:{ 
       $reduce: {
        input: {
         $map: {
          input: "$$currenta",
          as: "a",
          in: {
           $map: {
            input: "$$currentr",
            as: "r",
            in: {
             $mergeObjects: ["$$a", "$$r"] // Merge accumulated value with the current processing element
            }
           }
          }
         }
        },
        initialValue: [],
        in: {
         $setUnion: ["$$value", "$$this"] // Reduce the merged values which will be used as accumulator for next element
        }
       }
      }
     }
    }
   }
  }
 }
});

更新

由于第一个解决方案中的 $cond 和第二个解决方案中的 $setUnion 不正确，上述两种解决方案都不适用于跨数组重复值，如下面的 Asya Kamsky 的 cmets 所述。

正确的解决方法是

从initialValue 的[ { } ] 开始

或者

更改input 以排除第一个元素，例如input: {$slice:["$data", 1, {$subtract:[{$size:"$data"},1]}]},

完整的聚合管道

aggregate({
 $project: {
  cp: {
   $reduce: {
    input: {$slice:["$data", 1, {$subtract:[{$size:"$data"},1]}]},
    initialValue: {$arrayElemAt:["$data",0]},
    in: {
     $let: {
      vars: {
       currentr: "$$this", 
       currenta: "$$value" 
      },
      in:{ 
       $reduce: {
        input: {
         $map: {
          input: "$$currenta",
          as: "a",
          in: {
           $map: {
            input: "$$currentr",
            as: "r",
            in: {
             $mergeObjects: ["$$a", "$$r"] 
            }
           }
          }
         }
        },
        initialValue: [],
        in: {
         $concatArrays: ["$$value", "$$this"] 
        }
       }
      }
     }
    }
   }
  }
 }
});

参考： Cartesian product of multiple arrays in JavaScript