Projecting specific fields present inside an array, based on the value of some other field

Question

Overview :

The documents, that I'm working upon, have two nested arrays in them - contentMetaData & text_content. Within contentMetaData, we have the text_content and content_flag. Based on the value of the content_flag, I need to hide specific field within the text_content.

Requirement :

If the content_flag is true, text_content should have a single child - the text_note.
If the content_flag is false, text_content should have a single child - the text_description.
The structure and other details need to be preserved.
Documents SHOULD NOT be updated; the values need to be only hidden during projection.

Version Used : Mongo 2.6

Sample Document :

{
  "_id": ObjectId("56f8dd19e4b0365115927b0f"),
  "contentId": "cbc91805-2faa-4eff-8f84-02547173c152",
  "contentMetaData": [
    {
      "_id": "1574b58f-b7fa-4cd5-b34f-98beeb657c97",
      "name": "text_content",
      "attributes": [],
      "children": [
        {
          "_id": "97340ecf-fdbd-41e5-a6b2-01cc542f16ee",
          "name": "text_note",
          "value": "abc",
          "type": "java.lang.String",
          "attributes": [],
          "children": [],
          "noOfChildren": 0,
          "positionIndex": 1
        },
        {
          "_id": "19c5a3fb-54a2-4368-a89d-ea1d2554402d",
          "name": "text_description",
          "value": "def",
          "type": "java.lang.String",
          "attributes": [],
          "children": [],
          "noOfChildren": 0,
          "positionIndex": 2
        }
      ],
      "noOfChildren": 2,
      "positionIndex": 1
    },
    {
      "_id": "4e8ef7c9-cffd-4b36-9109-89b263dff3c8",
      "name": "content_flag",
      "value": "true",
      "type": "java.lang.String",
      "attributes": [],
      "children": [],
      "noOfChildren": 0,
      "positionIndex": 2
    }
  ]
}

Sample Output :

{
  "_id":  ObjectId("56f8dd19e4b0365115927b0f"),
  "contentId": "cbc91805-2faa-4eff-8f84-02547173c152",
  "contentMetaData": [
    {
      "_id": "1574b58f-b7fa-4cd5-b34f-98beeb657c97",
      "name": "text_content",
      "attributes": [],
      "children": [
        {
          "_id": "97340ecf-fdbd-41e5-a6b2-01cc542f16ee",
          "name": "text_note",
          "value": "abc",
          "type": "java.lang.String",
          "attributes": [],
          "children": [],
          "noOfChildren": 0,
          "positionIndex": 1
        }
      ],
      "noOfChildren": 2,
      "positionIndex": 1
    },
    {
      "_id": "4e8ef7c9-cffd-4b36-9109-89b263dff3c8",
      "name": "content_flag",
      "value": "true",
      "type": "java.lang.String",
      "attributes": [],
      "children": [],
      "noOfChildren": 0,
      "positionIndex": 2
    }
  ]
}

I attempted using $map but it didn't work. I tried using $unwind, but was unable to $push the data back, in the desired format.

Sample Mongo Code :

db.content.aggregate([
{
    $project: {
        _id: 1,
        contentId: 1,
        contentMetaData: 1
        tempMetaData: "$contentMetaData"
    }
},
{
    $unwind: "$contentMetaData"
},
{
    $match: {
        "contentMetaData.name": "content_flag"
    }
},
{
    $project: {
        _id: 1,
        contentId: 1,
        contentMetaData: "$tempMetaData",
        content_flag_value: "$contentMetaData.value"
    }
},
{
    $project: {
        _id: 1,
        contentId: 1,
        contentMetaData: 1,
        tempMetaData: "$contentMetaData",
        content_flag_value: 1
    }
},
{
        $unwind: "$contentMetaData"
},
{
        $match: {
                "contentMetaData.name": "text_content"
        }
},
{
        $project: {
                _id: 1,
                contentId: 1,
                contentMetaData: 1,
                tempMetaData: "$contentMetaData",
                content_flag_value: 1,
                text_content : "$contentMetaData.children",
                temp_text_content: "$text_content"
        }
},
{
        $unwind: "$text_content"
},
{
        $group:{
            _id:"$_id",
            contentId:{$first:"$contentId"},
            text_content:
            {$max:
                {$cond: 
                    [
                        {$eq: ["$content_flag_value", "true"]},
                        {$cond:
                            [{$or:[
                                {$eq: ["$text_content.name","wk_link_url"]},
                                {$eq: ["$text_content.name","wk_link_description"]}
                            ]},
                            "$text_content",
                            null]
                        },
                        null
                    ]
                }
            },
            contentMetaData:{$first:"$contentMetaData"}
        }
},
{
        $group:{
            _id:"$_id",
            contentId:{$first:"$contentId"},
            contentMetaData:{$push:{"text_content":"$text_content"}}
        }
},
{
        $project: {
                _id: 0,
                contentId: 1,
                contentMetaData: 1
        }
}]).pretty()

I'm new to Mongo. Can somebody help me out with this?

I see $map in your post. Does it mean you can use 3.x mongo version ? — s7vr
– s7vr, Commented Jan 27, 2017 at 14:54
@Veeram - As Chridam mentioned, it's available in 2.6 as well. — Subhashree Pradhan
– Subhashree Pradhan, Commented Jan 28, 2017 at 6:05

s7vr · Accepted Answer · 2017-01-31 05:16:14Z

1

You can try the below aggregation.

$map in combination with $setDifference to extract text_content and content_flag array.

$unwind to content_flag document.

$map to keep the current values in text_content and $map in combination with $setDifference to filter the children on the criteria.

$setUnion to join back the text_content and content_flag array into contentMetaData

db.collection.aggregate({
    $project: {
        _id: 1,
        contentId: 1,
        text_content: {
            "$setDifference": [{
                    "$map": {
                        "input": "$contentMetaData",
                        "as": "text",
                        "in": {
                            "$cond": [{
                                    $eq: ['$$text.name', "text_content"]
                                },
                                "$$text",
                                false
                            ]
                        }
                    }
                },
                [false]
            ]
        },
        content_flag: {
            "$setDifference": [{
                    "$map": {
                        "input": "$contentMetaData",
                        "as": "content",
                        "in": {
                            "$cond": [{
                                    $eq: ['$$content.name', "content_flag"]
                                },
                                "$$content",
                                false
                            ]
                        }
                    }
                },
                [false]
            ]
        }
    }
}, {
    $unwind: "$content_flag"
}, {
    $project: {
        "_id": 1,
        contentId: 1,
        "contentMetaData": {
            $setUnion: [{
                    $map: {
                        input: "$text_content",
                        as: "text",
                        in: {
                            "_id": "$$text._id",
                            "name": "$$text.name",
                            "attributes": "$$text.attributes",
                            "noOfChildren": "$$text.noOfChildren",
                            "positionIndex": "$$text.positionIndex",
                            "children": {
                                "$setDifference": [{
                                        "$map": {
                                            "input": "$$text.children",
                                            "as": "child",
                                            "in": {
                                                "$cond": [{
                                                        "$cond": [{
                                                            $eq: ["$content_flag.value", "true"]
                                                        }, {
                                                            $eq: ["$$child.name", "text_note"]
                                                        }, {
                                                            $eq: ["$$child.name", "text_description"]
                                                        }]
                                                    },
                                                    "$$child",
                                                    false
                                                ]
                                            }
                                        }
                                    },
                                    [false]
                                ]
                            }
                        }
                    }
                },
                ["$content_flag"]
            ]
        }
    }
})

Update:

$map in combination with $setDifference to extract content_flag array.

$unwind to content_flag document.

$redact to go through a document level at a time and look for name field recursively and perform $$DESCEND and $$PRUNE on the criteria.

$project to format the final response.

db.collection.aggregate({
    $project: {
        _id: 1,
        contentId: 1,
        contentMetaData: 1,
        content_flag: {
            "$setDifference": [{
                    "$map": {
                        "input": "$contentMetaData",
                        "as": "content",
                        "in": {
                            "$cond": [{
                                    $eq: ['$$content.name', "content_flag"]
                                },
                                "$$content",
                                false
                            ]
                        }
                    }
                },
                [false]
            ]
        }
    }
}, {
    $unwind: "$content_flag"
}, {
    $redact: {
        $cond: [{
                $or: [{
                    $eq: ["$name", "text_content"]
                }, {
                    $not: "$name"
                }, {
                    $eq: ["$name", "content_flag"]
                }, {
                    $and: [{
                        $eq: ["$name", "text_note"]
                    }, {
                        $eq: ["$$ROOT.content_flag.value", "true"]
                    }]
                }, {
                    $and: [{
                        $eq: ["$name", "text_description"]
                    }, {
                        $eq: ["$$ROOT.content_flag.value", "false"]
                    }]
                }]
            },
            "$$DESCEND",
            "$$PRUNE"
        ]
    }
}, {
    $project: {
        _id: 1,
        contentId: 1,
        contentMetaData: 1
    }
});

edited Jan 31, 2017 at 5:16

answered Jan 30, 2017 at 2:53

s7vr

76.2k12 gold badges118 silver badges135 bronze badges

Sign up to request clarification or add additional context in comments.

5 Comments

Subhashree Pradhan Over a year ago

Thank you for your response; I appreciate your efforts. I tried your code, but I got an exception: invalid operator '$slice' in the shell. The use of $slice within the $project is not supported in Mongo 2.6. Kindly refer : docs.mongodb.com/v2.6/reference/operator/projection/slice . Moreover, I would suggest not using $slice, because we shouldn't assume the array length; it may vary in future. If you could just stick to the field names, that would be much helpful.

Subhashree Pradhan Over a year ago

As per the updated answer, I'm able to get the desired text_content. But the rest of the fields present within contentMetadata is not getting updated. I found that the $setUnion is only meant for merging two arrays. Refer : docs.mongodb.com/v2.6/reference/operator/aggregation/setUnion/… . Since content_flag is just an object, this approach isn't working.

s7vr Over a year ago

Looks like wrapping obj as [] only works from 3.2 version in project stage. Seems like it [false] is working in setDifference. Can you replace ["$content_flag"] with {"$setDifference": [["$content_flag"], [false]]} and see if it works ?

Subhashree Pradhan Over a year ago

No, that didn't work as well. Instead of printing the entire object, it is printing "$content_flag".

s7vr Over a year ago

Added an alternative.

Collectives™ on Stack Overflow

Projecting specific fields present inside an array, based on the value of some other field

1 Answer 1

5 Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

1 Answer 1

5 Comments

Your Answer

Sign up or log in

Post as a guest

Related