Neo4j 密码查询 : using ORDER BY with COLLECT(S)

Question

我很难从两个不同的来源收集数据并合并这些集合，以便最后一个是一组按“dateCreated”排序的对象。

上下文

用户可以分组提问。 问题可以是一般性的，也可以是与特定视频游戏相关的。 如果群组中提出的问题与视频游戏相关，则该问题也会出现在视频游戏的问题页面中。

目前，我有两个一般性问题，一个是针对一款视频游戏的。 因此，在提取问题时，我应该有 3 个问题。

查询

这是查询:

START group = node(627)
MATCH generalQuestions-[?:GENERAL_QUESTION]->group
WITH group, generalQuestions
MATCH gamesQuestions-[?:GAME_QUESTION]->games<-[:GAMES]-group
WITH (collect(generalQuestions) + collect(gamesQuestions)) as questions
RETURN questions
ORDER BY questions.dateCreated

第一期:使用ORDER BY

Cached(questions of type Collection) expected to be of type Map but it is of type Collection - maybe aggregation removed it?

实现我想要做的事情的正确方法是什么？

第二个问题:错误的结果

如果我删除 ORDER BY 子句，我得到 14 个结果而不是 3 个结果......:

[
Node[641]{dateCreated:1380892636,dateUpdated:1380892636,title:"GENERAL TITLE 1",type:1,content:"GENERAL CONTENT 1"},
Node[641]{dateCreated:1380892636,dateUpdated:1380892636,title:"GENERAL TITLE 1",type:1,content:"GENERAL CONTENT 1"},
Node[641]{dateCreated:1380892636,dateUpdated:1380892636,title:"GENERAL TITLE 1",type:1,content:"GENERAL CONTENT 1"},
Node[641]{dateCreated:1380892636,dateUpdated:1380892636,title:"GENERAL TITLE 1",type:1,content:"GENERAL CONTENT 1"},
Node[641]{dateCreated:1380892636,dateUpdated:1380892636,title:"GENERAL TITLE 1",type:1,content:"GENERAL CONTENT 1"},
Node[641]{dateCreated:1380892636,dateUpdated:1380892636,title:"GENERAL TITLE 1",type:1,content:"GENERAL CONTENT 1"},
Node[642]{dateCreated:1380892642,dateUpdated:1380892642,title:"GENERAL TITLE 2",type:1,content:"GENERAL CONTENT 2"},
Node[642]{dateCreated:1380892642,dateUpdated:1380892642,title:"GENERAL TITLE 2",type:1,content:"GENERAL CONTENT 2"},
Node[642]{dateCreated:1380892642,dateUpdated:1380892642,title:"GENERAL TITLE 2",type:1,content:"GENERAL CONTENT 2"},
Node[642]{dateCreated:1380892642,dateUpdated:1380892642,title:"GENERAL TITLE 2",type:1,content:"GENERAL CONTENT 2"},
Node[642]{dateCreated:1380892642,dateUpdated:1380892642,title:"GENERAL TITLE 2",type:1,content:"GENERAL CONTENT 2"},
Node[642]{dateCreated:1380892642,dateUpdated:1380892642,title:"GENERAL TITLE 2",type:1,content:"GENERAL CONTENT 2"},
Node[632]{dateCreated:1380889484,dateUpdated:1380889484,title:"GTA5 TITLE",type:2,content:"GTA5 CONTENT"},
Node[632]{dateCreated:1380889484,dateUpdated:1380889484,title:"GTA5 TITLE",type:2,content:"GTA5 CONTENT"}
]

我收集结果的方式有问题吗？

编辑

获取游戏问题的扩展查询:

gamesQuestions-[:GAME_QUESTION]->()<-[:QUESTIONS]-games-[:INTERESTS]->()<-[:HAS_‌​INTEREST_FOR]-interests<-[:INTERESTS]-group

谢谢你的帮助，

Answer 1

“Order by”期望节点或关系上的属性。您查询中的“问题”是节点的集合，而不是节点/关系，您不能使用“排序依据”对集合进行排序，您只能对节点或关系的属性进行排序。

为了使用“Order by”，您需要将问题作为一列而不是一个集合返回。根据原始查询中指示的关系，以下查询应将一般和特定游戏问题作为一列行返回，并在属性“dateCreated”上对它们进行排序，

START group = node(627) 
Match question-[?:GENERAL_QUESTION|GAME_QUESTION]->()<-[:GAMES*0..1]-(group)
Return distinct question
Order by question.dateCreated

对于游戏问题通过关系序列“gamesQuestions-[?:GAME_QUESTION]->games<-[:GAMES]-group 与组相关的扩展案例，我有 gamesQuestions-[:GAME_QUESTION]->()<-[:QUESTIONS]-games-[:INTERESTS]->()<-[:HAS_‌ INTEREST_FOR]-interests<-[:INTERESTS]-group"，您可以简单地扩展上一个查询中的模式如下，

START group = node(627) 
Match question-[:GENERAL_QUESTION|GAME_QUESTION]->()-[*0..4]-(group)
Return distinct question
Order by question.dateCreated

想法是匹配可以通过一步或 4 步以上到达组节点的问题。

另一种选择是在 where 子句中指定两种模式，

START group = node(627) 
MATCH question-[*]-group
Where question-[:GENERAL_QUESTION]->group or (question-[:GAME_QUESTION]->()<-[:QUESTIONS]-()-[:INTERESTS]->()<-[:HAS_INTERESTS_FOR]-()<-[:INTERESTS]-group)
Return distinct q
Order by q.dateCreated