MongoDB聚合/展开/组/项目查询组合

我有以下格式的集合中的记录：

"_id" : "2013-05-23", "authors_who_sold_books" : [ { "id" : "Charles Dickens", "num_sold" : 1, "customers" : [ { "time_bought" : 1368627290, "customer_id" : 9715923 } ] }, { "id" : "JRR Tolkien", "num_sold" : 2, "customers" : [ { "date_bought" : 1368540890, "customer_id" : 9872345 }, { "date_bought" : 1368537290, "customer_id" : 9163893 } ] } ] } 

每个日期都有一个记录，其中许多日期将包含同一作者。 我正在查询返回以下内容：

 { "_id" : "Charles Dickens", "num_sold" : 235, "customers" : [ { "date_bought" : 1368627290, "customer_id" : 9715923 }, { "date_bought" : 1368622358, "customer_id" : 9876234 }, etc... ] } 

我已经尝试了聚合，组合，放松和项目的各种组合，但仍然无法完全实现并且非常感谢任何建议。

对于额外的点，我实际上是使用Ruby gem做的，所以特定于此的代码会很棒。 但是，我可以转换普通的MongoDB查询语言。

Answer 1

我获取了您的示例数据，稍微修改了第二个文档，然后将它们添加到测试集合中。 我使用的文件如下：

 { "_id" : "2013-05-23", "authors_who_sold_books" : [ { "id" : "Charles Dickens", "num_sold" : 1, "customers" : [ { "time_bought" : 1368627290, "customer_id" : 9715923 } ] }, { "id" : "JRR Tolkien", "num_sold" : 2, "customers" : [ { "date_bought" : 1368540890, "customer_id" : 9872345 }, { "date_bought" : 1368537290, "customer_id" : 9163893 } ] } ] } { "_id" : "2013-05-21", "authors_who_sold_books" : [ { "id" : "Charles Dickens", "num_sold" : 3, "customers" : [ { "time_bought" : 1368627290, "customer_id" : 9715923 }, { "time_bought" : 1368627290, "customer_id" : 9715923 }, { "time_bought" : 1368627290, "customer_id" : 9715923 } ] }, { "id" : "JRR Tolkien", "num_sold" : 1, "customers" : [ { "date_bought" : 1368540890, "customer_id" : 9872345 } ] } ] } 

现在，为了获得预期的结果，我使用了聚合框架并运行了这个查询：

 db.collection.aggregate([ { // First we unwind all the authors that sold books $unwind: '$authors_who_sold_books', }, { // Next, we unwind each of the customers that purchased a book $unwind: '$authors_who_sold_books.customers' }, { // Now we group them by "Author Name" (hoping they are unique!) $group: { _id: '$authors_who_sold_books.id', // Increment the number sold by each author num_sold: { $sum: 1 }, // Add the customer data to the array customers: { $push: '$authors_who_sold_books.customers' } } } ]); 

我试着记录上面的代码，这样就更有意义了。 基本上，它将数据展开两次，以便为作者的每次销售创建文档。 首先由authors_who_sold_books展开，然后展开authors_who_sold_books.customers 。

下一步就是将它们分组并将所有客户推送到customers数组中，并为我们拥有的每个unwinded文档将num_sold递增1。

结果看起来像：

 { "result" : [ { "_id" : "JRR Tolkien", "num_sold" : 3, "customers" : [ { "date_bought" : 1368540890, "customer_id" : 9872345 }, { "date_bought" : 1368537290, "customer_id" : 9163893 }, { "date_bought" : 1368540890, "customer_id" : 9872345 } ] }, { "_id" : "Charles Dickens", "num_sold" : 4, "customers" : [ { "time_bought" : 1368627290, "customer_id" : 9715923 }, { "time_bought" : 1368627290, "customer_id" : 9715923 }, { "time_bought" : 1368627290, "customer_id" : 9715923 }, { "time_bought" : 1368627290, "customer_id" : 9715923 } ] } ], "ok" : 1 } 

希望这可以帮助您找出真正的解决方案:)