C# Mongodb cartesian product of multiple object array documents

Question

Trying to understand with either C#/Linq or even raw Mongodb query itself how to join multiple arrays as a cartesian product.

Say for example I had a collection that I filtered down to the following two documents:

[
{"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"}
]

How can I join each item in each array to produce the following type of result?

[{"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"}]

I am looking for a solution that could work with any number of documents. So for example if in this example there was a 3rd document that also had 3 object arrays that would produce a result set of 27 items in the array - or 27 rows as it were.

Hoping for a solution of how to use C# (Linq?) Mongodb Driver to query and return data like this but would be open to even a mongodb specific query as I can hopefully reverse the logic from there. thank you

Answer 1

You can try below aggregation pipeline.

Note mergeObjects aggregation operator is available in the 3.5.6 + development release which will be rolled into upcoming 3.6 release.

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"
   }
  ]
 ]
}

Aggregation using conditional expression.

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
         }
        }
       }]
      }
     }
    }
   }
  }
 }
});

Aggregation( using $setUnion ).

This solution was only added to suppress the conditional expression to give more readable pipeline.

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
        }
       }
      }
     }
    }
   }
  }
 }
});

Update

Both the above solutions will not work with repeating values across arrays as noted by Asya Kamsky's comments below because of the incorrect $cond in the first solution and $setUnion in second solution.

The correct fix is to

start with initialValue of [ { } ]

Or

change input to exclude the first element like input: {$slice:["$data", 1, {$subtract:[{$size:"$data"},1]}]},

Complete aggregation pipeline

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"] 
        }
       }
      }
     }
    }
   }
  }
 }
});

Reference: Cartesian product of multiple arrays in JavaScript