Using $push with $group with pymongo

问题

Objective

Fix my make_pipeline() function to, using an aggregation query, count the number of tweets for each user, add them to an array and return the 5 users with the most tweets.

Exercise

Using an aggregation query, count the number of tweets for each user. In the same $group stage, use $push to accumulate all the tweet texts for each user.

Limit your output to the 5 users with the most tweets.

Your result documents should include only the fields:

"_id" (screen name of user),
"count" (number of tweets found for the user),
"tweet_texts" (a list of the tweet texts found for the user).

Background

To achieve the previous objective I am testing the following code:

def make_pipeline():
    # complete the aggregation pipeline
    pipeline = [
        {"$group": {"_id": "$user.screen_name", "tweet_texts": {"$push": "$text"}, "count": {"$sum": 1}}},
        {"$project": {"_id": "$user.screen_name", "count": 1, "tweet_texts": 1}},
        {"$sort" : {"count" : -1}},
        {"$limit": 5}
    ]
    return pipeline

Logic

First I group all the tweets by username. Then, in the same stage, I push all the texted tweets to tweet_texts and I count each occurrence that was grouped. I believe this will give me the number of users with most tweets.

Then I make a projection to select only the three fields I want:

_id
count
tweet_texts

I finish by sorting and limiting the amount of results.

Problem

I am passing the test, but not the submission. What am I doing wrong? I now the error must be in the first (group) stage, but I can't find for the love of God what I am doing wrong.

Data Sample

{
    "_id" : ObjectId("5304e2e3cc9e684aa98bef97"),
    "text" : "First week of school is over :P",
    "in_reply_to_status_id" : null,
    "retweet_count" : null,
    "contributors" : null,
    "created_at" : "Thu Sep 02 18:11:25 +0000 2010",
    "geo" : null,
    "source" : "web",
    "coordinates" : null,
    "in_reply_to_screen_name" : null,
    "truncated" : false,
    "entities" : {
        "user_mentions" : [ ],
        "urls" : [ ],
        "hashtags" : [ ]
    },
    "retweeted" : false,
    "place" : null,
    "user" : {
        "friends_count" : 145,
        "profile_sidebar_fill_color" : "E5507E",
        "location" : "Ireland :)",
        "verified" : false,
        "follow_request_sent" : null,
        "favourites_count" : 1,
        "profile_sidebar_border_color" : "CC3366",
        "profile_image_url" : "http://a1.twimg.com/profile_images/1107778717/phpkHoxzmAM_normal.jpg",
        "geo_enabled" : false,
        "created_at" : "Sun May 03 19:51:04 +0000 2009",
        "description" : "",
        "time_zone" : null,
        "url" : null,
        "screen_name" : "Catherinemull",
        "notifications" : null,
        "profile_background_color" : "FF6699",
        "listed_count" : 77,
        "lang" : "en",
        "profile_background_image_url" : "http://a3.twimg.com/profile_background_images/138228501/149174881-8cd806890274b828ed56598091c84e71_4c6fd4d8-full.jpg",
        "statuses_count" : 2475,
        "following" : null,
        "profile_text_color" : "362720",
        "protected" : false,
        "show_all_inline_media" : false,
        "profile_background_tile" : true,
        "name" : "Catherine Mullane",
        "contributors_enabled" : false,
        "profile_link_color" : "B40B43",
        "followers_count" : 169,
        "id" : 37486277,
        "profile_use_background_image" : true,
        "utc_offset" : null
    },
    "favorited" : false,
    "in_reply_to_user_id" : null,
    "id" : NumberLong("22819398300")
}

Please help!

回答1:

The $project step is redundant as the $group pipeline already produces just those three fields so there's no need for a preceding $project stage.

The correct pipeline should be

pipeline = [ 
    {
        "$group": {
            "_id": "$user.screen_name", 
            "tweet_texts": { "$push": "$text" }, 
            "count": { "$sum": 1 }
        }
    }, 
    { "$sort" : { "count" : -1 } }, 
    { "$limit": 5 } 
]

Your $project pipeline didn't work because the previous $group pipeline doesn't produce any field "$user.screen_name" which you attempt to use as the _id field in the $project pipeline.

However, if you wanted to include the $project step then the working pipeline should follow:

pipeline = [ 
    {
        "$group": {
            "_id": "$user.screen_name", 
            "tweet_texts": { "$push": "$text" }, 
            "count": { "$sum": 1 }
        }
    }, 
    { "$project": { "count": 1, "tweet_texts": 1 } },
    { "$sort" : { "count" : -1 } }, 
    { "$limit": 5 } 
]

回答2:

Reading comments

Reading the comments I found out that

pipeline = [
        {"$group": {"_id": "$user.screen_name", "tweet_texts": {"$push": "$text"}, "count": {"$sum": 1}}},
        {"$project": {"_id": "$user.screen_name", "count": 1, "tweet_texts": 1}},
        {"$sort" : {"count" : -1}},
        {"$limit": 5}
    ]

Should in fact be changed to:

pipeline = [ 
        {"$group": {"_id": "$user.screen_name", "tweet_texts": {"$push": "$text"}, "count": {"$sum": 1}}}, 
        {"$sort" : {"count" : -1}}, 
        {"$limit": 5}
    ]

Why?

The full answer and explanation can be seen in the answer:

What is the difference between these two MongoDB queries?

The conclusion of the story is that I am using the $project stage wrongly. Not only was is no needed in the first place, to make it idempotent it should be

{"$project": {"_id": "$_id", "count": 1, "tweet_texts": 1}},

I also highly recommend his answer:

https://stackoverflow.com/a/40970901/1337392

Special Thanks

The following users deserve kudos++:

https://stackoverflow.com/users/122005/chridam

For directing me in to the right path!

来源：https://stackoverflow.com/questions/40868553/using-push-with-group-with-pymongo

标签

python

mongodb

mongodb-query

pymongo