mongodb - 如何根据唯一属性在 MongoDB 集合中进行基于查询的分页？

Question

我有一个名为 ranking 的集合，其中的文档如下所示:

{ "_id" : ObjectId("55dde5e4827ff4e65b684f94"), "round" : 0, "teamId" : "55a9c261a02911b85fdff231", "teamName" : "FORZA JUVE", "userId" : "55a9c209291bc40561cc97d2", "signupDate" : ISODate("2015-07-18T03:03:37Z"), "userName" : "Noé González Ramírez", "signupPoints" : 0, "lfPoints" : 559, "roundPoints" : [ 110, 99, 91, 65, 64, 61, 69 ], "roundRankings" : [ ], "ranking" : 1 }
{ "_id" : ObjectId("55dde591827ff4e65b6846d6"), "round" : 0, "teamId" : "55a58179f82921e922244402", "teamName" : "Dream ★ Team", "userId" : "55a5809fd18e541b2176412f", "signupDate" : ISODate("2015-07-14T21:35:27Z"), "userName" : "Fabio Dionicio López Lazcano", "signupPoints" : 0, "lfPoints" : 555, "roundPoints" : [ 105, 91, 108, 60, 53, 77, 61 ], "roundRankings" : [ ], "ranking" : 2 }
{ "_id" : ObjectId("55dde593827ff4e65b6847d3"), "round" : 0, "teamId" : "55a58889f82921e922245511", "teamName" : "Camel Toes", "userId" : "55a58868f82921e9222454b1", "signupDate" : ISODate("2015-07-14T22:08:40Z"), "userName" : "Luis Fridman Tabak", "signupPoints" : 0, "lfPoints" : 555, "roundPoints" : [ 74, 96, 99, 76, 69, 70, 71 ], "roundRankings" : [ ], "ranking" : 3 }
{ "_id" : ObjectId("55dde60a827ff4e65b6853d5"), "round" : 0, "teamId" : "55ac7ca61532c9284e2a89a7", "teamName" : "S.T.A.R.S.", "userId" : "55ac7be31532c9284e2a895c", "signupDate" : ISODate("2015-07-20T04:41:07Z"), "userName" : "Daniel Ivan Estudillo Lopez", "signupPoints" : 0, "lfPoints" : 555, "roundPoints" : [ 61, 94, 105, 73, 90, 50, 82 ], "roundRankings" : [ ], "ranking" : 4 }
{ "_id" : ObjectId("55dde5d2827ff4e65b684db3"), "round" : 0, "teamId" : "55a857b8fad3a3124ca72edf", "teamName" : "EFRASTEAM", "userId" : "55a8575dfad3a3124ca72ebd", "signupDate" : ISODate("2015-07-17T01:16:13Z"), "userName" : "EFRAÍN BARRIENTOS RODRÍGUEZ", "signupPoints" : 0, "lfPoints" : 549, "roundPoints" : [ 102, 89, 92, 58, 81, 60, 67 ], "roundRankings" : [ ], "ranking" : 5 }
{ "_id" : ObjectId("55dde5c1827ff4e65b684baa"), "round" : 0, "teamId" : "55a800a4ac9c9c3302be1dee", "teamName" : "Lombardi FC", "userId" : "55a722eac0dbc6e577489d34", "signupDate" : ISODate("2015-07-16T03:20:10Z"), "userName" : "J Alejandro Padilla", "signupPoints" : 0, "lfPoints" : 546, "roundPoints" : [ 103, 98, 65, 47, 77, 59, 97 ], "roundRankings" : [ ], "ranking" : 6 }
{ "_id" : ObjectId("55dde5a1827ff4e65b68485f"), "round" : 0, "teamId" : "55a5c2e88a5f0fd66b355ef6", "teamName" : "FC Barcelona", "userId" : "55a5c0458a5f0fd66b355d6f", "signupDate" : ISODate("2015-07-15T02:07:01Z"), "userName" : "Daniel Alvarez Nunez", "signupPoints" : 0, "lfPoints" : 545, "roundPoints" : [ 94, 101, 82, 63, 71, 68, 66 ], "roundRankings" : [ ], "ranking" : 7 }
{ "_id" : ObjectId("55dde55d827ff4e65b684197"), "round" : 0, "teamId" : "55a54fbfef67112c685cf78d", "teamName" : "RUCHINTER FC *", "userId" : "55a54f713711fed46621fb01", "signupDate" : ISODate("2015-07-14T18:05:37Z"), "userName" : "RAUL VERA LUGO", "signupPoints" : 0, "lfPoints" : 544, "roundPoints" : [ 84, 109, 87, 41, 84, 57, 82 ], "roundRankings" : [ ], "ranking" : 8 }
{ "_id" : ObjectId("55dde583827ff4e65b684661"), "round" : 0, "teamId" : "55a57327f82921e9222418ae", "teamName" : "SPARTA FC *", "userId" : "55a571b2f82921e9222413a2", "signupDate" : ISODate("2015-07-14T20:31:46Z"), "userName" : "Alejandro Padilla", "signupPoints" : 0, "lfPoints" : 544, "roundPoints" : [ 86, 98, 93, 29, 80, 65, 93 ], "roundRankings" : [ ], "ranking" : 9 }
{ "_id" : ObjectId("55dde592827ff4e65b684739"), "round" : 0, "teamId" : "55a59941b07aeb00512822d8", "teamName" : "Minions", "userId" : "55a59400d18e541b21766a0e", "signupDate" : ISODate("2015-07-14T22:58:08Z"), "userName" : "Carlos Zarate", "signupPoints" : 0, "lfPoints" : 541, "roundPoints" : [ 100, 91, 69, 66, 76, 68, 71 ], "roundRankings" : [ ], "ranking" : 10 }

如您所见，他们有一个 ranking 属性，它是一个整数，该属性在创建时不会(也不能)分配。

所以我需要获取轮数设置为 0 的整个文档，按降序排列。换句话说，我必须对整个文档进行分页(它们大约有 300K)。

我知道有一种方法可以查询 _id 大于上一页最后结果的文档。

这就是我尝试使用以下索引的原因:

{ranking: 1, _id: 1, round: 1}
{ranking: 1,  _id: 1}

所以首先我查询第一页的前n个元素，例如:

db.ranking.find({round: 0}).sort({ranking: 1}).limit(3)

哪些输出:

{ "_id" : ObjectId("55dde5e4827ff4e65b684f94"), "round" : 0, "teamId" : "55a9c261a02911b85fdff231", "teamName" : "FORZA JUVE", "userId" : "55a9c209291bc40561cc97d2", "signupDate" : ISODate("2015-07-18T03:03:37Z"), "userName" : "Noé González Ramírez", "signupPoints" : 0, "lfPoints" : 559, "roundPoints" : [ 110, 99, 91, 65, 64, 61, 69 ], "roundRankings" : [ ], "ranking" : 1 }
{ "_id" : ObjectId("55dde591827ff4e65b6846d6"), "round" : 0, "teamId" : "55a58179f82921e922244402", "teamName" : "Dream ★ Team", "userId" : "55a5809fd18e541b2176412f", "signupDate" : ISODate("2015-07-14T21:35:27Z"), "userName" : "Fabio Dionicio López Lazcano", "signupPoints" : 0, "lfPoints" : 555, "roundPoints" : [ 105, 91, 108, 60, 53, 77, 61 ], "roundRankings" : [ ], "ranking" : 2 }
{ "_id" : ObjectId("55dde593827ff4e65b6847d3"), "round" : 0, "teamId" : "55a58889f82921e922245511", "teamName" : "Camel Toes", "userId" : "55a58868f82921e9222454b1", "signupDate" : ISODate("2015-07-14T22:08:40Z"), "userName" : "Luis Fridman Tabak", "signupPoints" : 0, "lfPoints" : 555, "roundPoints" : [ 74, 96, 99, 76, 69, 70, 71 ], "roundRankings" : [ ], "ranking" : 3 }

然后我尝试获取以下页面:

db.lineuppointsrecord.find({round: 0, _id: {$gt: ObjectId("55dde593827ff4e65b6847d3")}}).sort({ranking: 1}).limit(3)

那会错误地输出:

{ "_id" : ObjectId("55dde5e4827ff4e65b684f94"), "round" : 0, "teamId" : "55a9c261a02911b85fdff231", "teamName" : "FORZA JUVE", "userId" : "55a9c209291bc40561cc97d2", "signupDate" : ISODate("2015-07-18T03:03:37Z"), "userName" : "Noé González Ramírez", "signupPoints" : 0, "lfPoints" : 559, "roundPoints" : [ 110, 99, 91, 65, 64, 61, 69 ], "roundRankings" : [ ], "ranking" : 1 }
{ "_id" : ObjectId("55dde60a827ff4e65b6853d5"), "round" : 0, "teamId" : "55ac7ca61532c9284e2a89a7", "teamName" : "S.T.A.R.S.", "userId" : "55ac7be31532c9284e2a895c", "signupDate" : ISODate("2015-07-20T04:41:07Z"), "userName" : "Daniel Ivan Estudillo Lopez", "signupPoints" : 0, "lfPoints" : 555, "roundPoints" : [ 61, 94, 105, 73, 90, 50, 82 ], "roundRankings" : [ ], "ranking" : 4 }
{ "_id" : ObjectId("55dde5d2827ff4e65b684db3"), "round" : 0, "teamId" : "55a857b8fad3a3124ca72edf", "teamName" : "EFRASTEAM", "userId" : "55a8575dfad3a3124ca72ebd", "signupDate" : ISODate("2015-07-17T01:16:13Z"), "userName" : "EFRAÍN BARRIENTOS RODRÍGUEZ", "signupPoints" : 0, "lfPoints" : 549, "roundPoints" : [ 102, 89, 92, 58, 81, 60, 67 ], "roundRankings" : [ ], "ranking" : 5 }

第一个结果是第 1 轮的文档出现在初始查询的第一页!

我知道我可以使用 skip 和 limit 但是，我应该如何按性能顺序执行查询？

Answer 1

基于唯一值的页面

如果每个排名在集合中都是唯一的，就像在您的示例文档中一样，并且每个必需的文档都有一个排名，您可以每次使用先前结果的最后排名进行查询，例如:

db.ranking.find({round: 0}).sort({ranking: 1}).limit(3)
db.ranking.find({round: 0, ranking : {$gt : 3}}).sort({ranking: 1}).limit(3)
db.ranking.find({round: 0, ranking : {$gt : 6}}).sort({ranking: 1}).limit(3)

如果排名在集合中不是唯一的，即如果多个团队的排名相同，则此方法将不起作用，因为它依赖于下一个排名始终大于前一个排名。如果您尝试使用它，您最终可能会跳过文档。

这个解决方案可以使用索引{ round: 1, ranking: 1 }

基于范围的分页

如果排名是唯一的，总是递增 1 并且没有差距，您甚至可以按范围查询，因为您知道您期望的排名:

db.ranking.find({round: 0, ranking : {$gte : 1, $lte: 3 }})
db.ranking.find({round: 0, ranking : {$gte : 4, $lte: 6 }})
db.ranking.find({round: 0, ranking : {$gte : 7, $lte: 9 }})

如果排名在集合中不是唯一的，您仍然可以这样做，但需要处理这样一个事实，即它可能会导致页面大小不一甚至是空白页面，尤其是对于小页面大小。

这个解决方案也可以使用索引{ round: 1, ranking: 1 }

对非唯一值进行分页

为了完整起见，这里是一个查询，它将处理非唯一排名的分页，并被索引覆盖。它需要检查排名更大，或者排名相同且 _id 更大。如果可能，我会推荐其他解决方案，因为它们更简单，包括所需的索引和查询。

db.ranking.createIndex({round: 1, ranking: 1, _id: 1})

db.ranking.find({$or : [
       {round: 0, ranking : { $gt : 3}},
       {round: 0, ranking : 3, _id : { $gt : new ObjectId("55a9c261a02911b85fdff231")}}
    ]})
.sort({round:1, ranking: 1, _id: 1})
.limit(3)