MongoDB group by array inner-elements

Question

I've got a list of articles, and each of them has an array property which lists various individuals mentioned in them:

_id: {
    $oid: "52b632a9e4f2ba13c82ccd23"
},
providerName: "The Guardian",
url: "http://feeds.theguardian.com/c/34708/f/663860/s/3516cebc/sc/38/l/0L0Stheguardian0N0Cmusic0C20A130Cdec0C220Cwaterboys0Efishermans0Eblues0Etour0Ehammersmith/story01.htm",
subject: "The Waterboys – review",
class_artist: [
    "paul mccartney"
]

I've been trying (unsuccessfully) to get a list of all the individual artists (class_artist), based on the number of articles they've been tagged in within the past 7 days.

I've gotten as far as:

var date = new Date();
date.setDate(date.getDate() - 7);

db.articles.group({
    key: { class_artist: 1 },
    cond: { class_date: { $gt: date } },
    reduce: function ( curr, result ) { result.cnt++; },
    initial: { cnt : 0 }
}).sort({cnt: -1});

But unfortunately, it doesn't count them based on the individual array values, but by array compositions (that is, lists of artists).

I tried using the $unwind function, but have not been able to make it work.

Neil Lunn · Accepted Answer · 2014-02-02 10:11:49Z

167

What framework are you using? This is not MongoDB shell and looks like some weird wrapper around MapReduce. In that case $unwind would not be available, and you need it for user in the aggregation framework. Here's what you want in the mongo shell:

db.articles.aggregate([
  {$match: { class_date: { $gte: date } } },
  {$project: { _id: 0, class_artist: 1 } },
  {$unwind: "$class_artist" },
  {$group: { _id: "$class_artist", tags: { $sum: 1 } }},
  {$project: { _id: 0,class_artist: "$_id", tags: 1 } },
  {$sort: { tags: -1 } }
])

So efficiently:

Filter by date because you already set a var for the last 7 days
Project only the field(s) we need { We need only one! }
Unwind the array so we now have a record for every array element in every document
Group on the Artist from the expanded documents
Project into a document format you can use as group messed around with _id
Sort the results in reverse order to see the top tagged first

And the great thing about aggregation is you can gradually build up those stages to see what is going on.

Shake and bake into your own driver implmentation or ODM framework as required.

answered Feb 2, 2014 at 10:11

Neil Lunn

151k36 gold badges356 silver badges327 bronze badges

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

1 Comment

Stennie Over a year ago

FYI, the "weird wrapper" format is the group() command which is implemented in JavaScript and predates the Aggregation Framework. See also: MongoDB aggregation comparison: group(), $group and MapReduce.

Collectives™ on Stack Overflow

MongoDB group by array inner-elements

1 Answer 1

1 Comment

Your Answer

Linked

Hot Network Questions

Collectives™ on Stack Overflow

1 Answer 1

1 Comment

Your Answer

Sign up or log in

Post as a guest

Linked

Related