简体繁体 English

在Objectify中查询实体中的选定字段-AppEngine

[英]Querying selected fields in an entity in objectify - appengine

原文 2014-07-21 10:30:08 5 2 android/ google-app-engine/ objectify

I am using appengine and objectify as a backend for my app. 我正在使用appengine和objectify作为我的应用程序的后端。 And when i query in the datastore i get a Entity object which has the data of required row. 当我在数据存储区中查询时，我得到一个实体对象，该对象具有所需行的数据。 But, using objectify how will i query entities and get oly selected fileds from it? 但是，使用objectify我将如何查询实体并从中获取全部选定的文件？ because querying the whole entity will be heavy and it needs more data bandwidth. 因为查询整个实体会很麻烦，并且需要更多的数据带宽。

Eg : In a entity with 4 columns, --> Id,name,description,age. 例如：在具有4列的实体中，-> ID，名称，描述，年龄。 I should query oly Id,name,age. 我应该查询身份证号，姓名，年龄。 I dont want description to be queried. 我不希望查询描述。

2 个解决方案

The GAE datastore does not work like an RDBMS; GAE数据存储无法像RDBMS那样工作； you can't arbitrarily pick and choose which fields to query out of an entity. 您不能随意选择要从实体中查询的字段。 The standard behavior of a datastore query is to follow an index (which maps attribute value to entity key), then fetch-by-key all the entities found. 数据存储区查询的标准行为是遵循索引（将属性值映射到实体键），然后按键获取所有找到的实体。

There is a feature called "projection queries" (which Objectify supports; look for the project() method on the query command object), however it is not a general purpose SELECT statement like you get in SQL. 有一个称为“投影查询”的功能（Objectify支持；在查询命令对象上查找project()方法），但是它不是像SQL中那样的通用SELECT语句。 Projection queries capitalize on the fact that the index itself contains the index values, so if you only want data that's in the index, you don't need to perform a subsequent fetch of the whole Entity. 投影查询充分利用了索引本身包含索引值的事实，因此，如果您只希望索引中的数据，则无需随后对整个Entity进行提取。 However, this comes with some restrictions: 但是，这有一些限制：

You must maintain a multiproperty index with all the data you wish to project. 您必须维护一个包含所有要投影数据的多属性索引。
You must maintain single-property indexes for each of the fields in the multiproperty index. 您必须为多属性索引中的每个字段维护单属性索引。
You can only project on queries that follow that particular index. 您只能对遵循该特定索引的查询进行投影。
Queries bypass Objectify's memcache-based entity cache. 查询绕过Objectify的基于内存缓存的实体缓存。

Be aware of the cost of using projection queries. 注意使用投影查询的成本。 In your example, you will need single-property indexes on Name and Age plus a multiproperty index on {__key__, Name, Age} . 在您的示例中，您将需要在Name和Age上使用单属性索引，并在{__key__, Name, Age}上使用多属性索引。 Instead of 3 write operations per entity written, your new entity will cost 8 write ops. 您的新实体将花费8个写入操作，而不是每个写入的实体进行3个写入操作。 On the other hand, the cost of a projection query is a constant 1 read op. 另一方面，投影查询的成本是恒定的1次读取操作。

On the other other hand, the cost of a batch get from memcache is 0, and the worst-case cost is 1 read op. 而另一方面，一个批次的成本从内存缓存得到的是0，最坏情况下的成本为1读取运算。 Unless your description field is known to be causing you problems, this is a massive premature optimization. 除非您知道您的描述字段会引起问题，否则这是一个过早的优化。