【问题标题】:Perform joins in mongodb with three collections?在 mongodb 中使用三个集合执行连接?
【发布时间】:2017-04-13 14:17:44
【问题描述】:

我正在使用 $lookup 来加入 mongoDB 中的不同集合。现在我在这里遇到一个问题,假设我在下面给出了 3 个集合。

user_movies

{
    "_id": ObjectId("5834ecf7432d92675bde9d83"),
    "mobile_no": "7941750156"
    "movies" : ["dallas00", "titanic00", "green_mile00"]
}

电影

{
    "_id" : ObjectId("4834eff7412d9267556d9d52"),
    "movie_name" : "Dallas Buyer's Club",
    "movie_id": "dallas00",
    "active": 0

}

movie_cmets

{
    "_id": ObjectId("1264eff7412d92675567h576"),
    "movie_id" : "dallas00",
    "comment": "what a great movie.",
    "time_posted": "1480516635131"
},
{
    "_id": ObjectId("1264eff7412d92675567h578"),
    "movie_id" : "dallas00",
    "comment": "awesome movie.",
    "time_posted": "1480516635141"
},
{
    "_id": ObjectId("1264eff7412d92675567h567"),
    "movie_id" : "titanic00",
    "comment": "titanic awesome movie.",
    "time_posted": "1480516635132"
},
{
    "_id": ObjectId("1264eff7412d92675567h579"),
    "movie_id" : "green_mile00",
    "comment": "Tom hanks did awesome movie.",
    "time_posted": "1480516635133"
}

User Movies 是我存储用户喜欢的电影的集合,Movies 是我存储有关电影的所有详细信息的集合,movie_cmets 是我存储与电影关联的 cmets 的集合。

现在我想写一个查询,我将向用户显示他们喜欢的电影列表,但所有电影都必须有"active" : 1,以及与该特定电影关联的 cmets。现在我尝试使用$lookup 我的代码如下。

db.user_movies.aggregate(
        {$match : {mobile_no : mobile_no}},
        { "$unwind": "$movies" },
        {$lookup: {from: "movies",localField: "movies",foreignField: "movie_id",as: "bmarks"}},
        {$unwind : "$bmarks"},
        {$match : {"bmarks.active": 1}},
        { $group : { _id : "$_id", movies : {$push : "$bmarks"}, movie_ids: {$push : "$bmarks.movie_id"}}},
        {$lookup: {from: "movie_comments", localField: "",foreignField: "movie_id",as: "comments"}},
        {$unwind : "$comments"},
        {$sort: {time_posted: -1}},
        {$group: {_id: '$_id', comments : {$push : "$comments"}}},

我正在尝试编写一个查询来执行所有这些功能,并且我能够对 user_movies 和 movies 的前两个集合执行连接,但我无法在同一查询中对第三个集合执行查找。我想要的是发送我的 first lookup 的输出数组,即 movie_ids,然后使用 movie_cmets 将其发送到下一个查找。这样我就可以在我的 movie_ids 数组中拥有与电影相关的所有 cmets。

现在谁能告诉我为什么我没有得到任何输出,我可以像我第二次使用它一样使用查找(它使用localfield 作为$group 字段)或者不是怎么能我在一个查询中执行此功能

更新- 现在只剩下一件事了,我想以一种排序的方式得到结果,这样电影就会像它们出现在“user_movies”数组中一样排序(根据它们在数组中的位置) ) 及其对应的 cmets 将按照 'timeposted' 降序排列。

所以对于上述文件,我希望我的输出是这样的

{
     "movie_id": "dallas00"  // dallas00 is first because it appear first in "user_movies" movies array.
     "movie_name": "Dallas Buyer's Club",
     "comments": ["what a great movie.","awesome movie."]
},
{
     "movie_id": "titanic00"  // titanic00 is second because it appear second in "user_movies" movies array.
     "movie_name": "Titanic",
     "comments": ["titanic awesome movie."]
},
{
     "movie_id": "green_mile00"  // green_mile00 is third because it appear third in "user_movies" movies array.
     "movie_name": "Green mile",
     "comments": ["Tom hanks did awesome movie."]
},

Ok 最后一件事,如果可以在我执行第一次查找时获得匹配的电影数组字符串的位置,那么我可以根据我将给予的权重进行排序(对前一个位置的权重更大,依此类推)。所以现在我要做的是{{$sort: { 'movie_weight': -1, 'time_posted': -1}}。现在我将对结果进行排序,这样电影将根据它们在数组"user_movies" 数组中的位置进行排序,而 cmets 将根据与它们的电影相对应的"time_posted" 进行排序。

更新代码

{$match : {mobile_no : mobile_no}},
        {$unwind: { path: "$movies", includeArrayIndex: "movieposition"}},
        {$lookup: {from: "movies",localField: "movies",foreignField: "movie_id",as: "bmarks"}},
        {$unwind : "$bmarks"},
        {$match : {"bmarks.active": 1}},
        {$group : { "_id" : "$bmarks.movie_id", movie_names : {$push : "$bmarks.movie_name"}, movie_ids: {$push : "$bmarks.movie_id"}, movie_position: {$push : "$movieposition"}}},
        {$unwind : "$movie_ids"},
        {$unwind : "$movie_names"},
        {$lookup: {from: "movie_comments", localField: "movie_ids", foreignField: "movie_id", as: "comments"}},
        {$unwind : "$comments"},
        {$sort: { 'movie_position': -1, 'comments.time_posted': -1 }},
        {$group: {_id: { movie_id: '$comments.movie_id', movie_name: '$movie_names' }, cmnts_ids: {$push: '$comments._id'}}},
        {$project: {_id: 0, movie_id: '$_id.movie_id', movie_name: '$_id.movie_name', tot_cmnts: {"$size" : "$cmnts_ids"}, top_cmnts_ids: {$slice: ['$cmnts_ids', 0, 4]}}}},
        {$skip: page*page_size},
        {$limit: page_size}, */

【问题讨论】:

    标签: mongodb join mongoose aggregation-framework lookup


    【解决方案1】:

    好的,看起来你有很多问题。我会试着总结一下。

    问题 1:您提供的聚合代码无法编译。

    db.user_movies.aggregate(
    {$match : {mobile_no : mobile_no}},
    {$unwind: "$movies" },
    {$lookup: {from: "movies",localField: "movies",foreignField: "movie_id",as: "bmarks"}},
    {$unwind : "$bmarks"},
    {$match : {"bmarks.active": 1}},
    {$group : { _id : "$_id", movies : {$push : "$bmarks"}, movie_ids: {$push : "$bmarks.movie_id"}}},
    {$lookup: {from: "movie_comments", localField: "",foreignField: "movie_id",as: "comments"}},
    {$unwind : "$comments"},
    {$sort: {time_posted: -1}},
    {$group: {_id: '$_id', comments : {$push : "$comments"}}},
    

    修复:

    db.user_movies.aggregate([
    {$match : {mobile_no : mobile_no}},
    {$unwind: "$movies" },
    {$lookup: {from: "movies",localField: "movies",foreignField: "movie_id",as: "bmarks"}},
    {$unwind : "$bmarks"},
    {$match : {"bmarks.active": 1}},
    {$group : { _id : "$_id", movies : {$push : "$bmarks"}, movie_ids: {$push : "$bmarks.movie_id"}}},
    {$lookup: {from: "movie_comments", localField: "",foreignField: "movie_id",as: "comments"}},
    {$unwind : "$comments"},
    {$sort: {time_posted: -1}},
    {$group: {_id: '$_id', comments : {$push : "$comments"}}}])
    

    现在假设您使用的数据与您提供的数据相同。

    问题 2:{$match : {"bmarks.active": 1}} 不匹配任何条目。

    修复:{$match : {"bmarks.active": 0}}

    问题 3:未定义查找字段 {$lookup: {from: "movie_comments", localField: "",foreignField: "movie_id",as: "comments"}}

    修复:{$lookup: {from: "movie_comments", localField: "movie_ids",foreignField: "movie_id",as: "comments"}}

    问题 4:movie_ids 没有用于先前查找的展开阶段

    修复:{$unwind : "$movie_ids"}

    问题5:没有时间发布字段{$sort: {time_posted: -1}}

    修复:排序前包含字段

    因此,要将所有内容放在一起,您需要聚合如下所示,以提取每部电影的 cmets。

    db.user_movies.aggregate([
    {$match : {mobile_no : mobile_no}},
    {$unwind: "$movies"},
    {$lookup: {from: "movies",localField: "movies",foreignField: "movie_id",as: "bmarks"}},
    {$unwind : "$bmarks"},
    {$match : {"bmarks.active": 0}},
    {$group : { _id : "$_id", movies : {$push : "$bmarks"}, movie_ids: {$push : "$bmarks.movie_id"}}},
    {$unwind : "$movie_ids"},
    {$lookup: {from: "movie_comments", localField: "movie_ids",foreignField: "movie_id",as: "comments"}},
    {$unwind : "$comments"},
    {$group: {_id: '$_id', comments : {$push : "$comments"}}}])
    

    样本输出

    {
        "_id": ObjectId("5834ecf7432d92675bde9d83"),
        "comments": [{
            "_id": ObjectId("583d96d7e35f6e9c53c9e894"),
            "movie_id": "dallas00",
            "comment": "what a great movie."
        }, {
            "_id": ObjectId("583d96d7e35f6e9c53c9e895"),
            "movie_id": "dallas00",
            "comment": "awesome movie."
        }]
    }
    

    更新:

    db.user_movies.aggregate([
    {$match : {mobile_no : mobile_no}},
    {$unwind: {path: "$movies", includeArrayIndex: "moviePosition"}},
    {$sort :  {moviePosition:1}},
    {$lookup: {from: "movies",localField: "movies",foreignField: "movie_id",as: "bmarks"}},
    {$unwind :"$bmarks"},
    {$group : {_id : "$_id", movies : {$push : {movie_name:"$bmarks.movie_name", movie_id:"$bmarks.movie_id"} }}},
    {$unwind : "$movies"},
    {$lookup: {from: "movie_comments", localField: "movies.movie_id",foreignField: "movie_id",as: "comments"}},
    {$unwind : "$comments"},
    {$group:  {_id: "$movies.movie_id", movie_name: {$first:"$movies.movie_name"}, comments : {$push : {comment:"$comments.comment", time_posted:"$comments.time_posted"}}}},
    {$sort :  {time_posted:-1}},
    {$project:{_id:0, movie_id:"$_id", movie_name:1, comments: "$comments.comment"}}
    ]).pretty();
    

    【讨论】:

    • 实际上它不起作用,因为我的电影位置也通过“time_posted”排序而改变。我认为我们应该这样写“{$sort: { 'moviePosition': -1, 'cmets.time_posted': -1 }}”。你怎么看??
    • 位置真的变了吗?你用了什么例子?
    • 嘿,我找到答案了-1, 'cmets.time_posted': -1 }}" 所以现在我的文档首先按电影排序,然后在 cmets 内部将根据 time_posted 排序,现在只剩下一件简单的事情了
    • 我还想要“user_movies”数组的大小,我考虑过用大小对 bmark 进行分组,例如“{$group : {_id : "$_id", movies : {$push : {movie_name:"$ bmarks.movi​​e_name", movie_id:"$bmarks.movi​​e_id", "movie_arr_size": {"$size" : "$movies"}} }}}" 但是在这里我将从 groupby 获得的每个文档现在都将包含“ movie_arr_size" 元素,我希望它只返回一次你得到它。
    • 我的意思是我不希望movie_arr_size出现在每个文档中传递到下一个管道。
    猜你喜欢
    • 1970-01-01
    • 1970-01-01
    • 1970-01-01
    • 2018-03-06
    • 2019-07-14
    • 2018-08-24
    • 1970-01-01
    • 2021-02-11
    • 2023-03-28
    相关资源
    最近更新 更多