如何筛选出拥有年龄大于5岁父节点的狗图节点？

先创建狗节点及父子关系：

INSERT INTO dog (id, age) VALUES ("mr_whaf", 8), ("mrs_whaf", 3), ("whaf_jr", 0), ("whaf_whaf", 1);
RELATE [dog:mr_whaf, dog:mrs_whaf]->parent_of->dog:whaf_jr;
RELATE dog:mrs_whaf->parent_of->dog:whaf_whaf;

需求：仅筛选出年龄小于2岁且拥有至少一个年龄大于5岁父节点的狗节点，预期结果只返回whaf_jr。

遇到的问题 #

执行以下查询时，会返回不符合要求的whaf_whaf（其parents为空数组）：

SELECT id,age, <-parent_of<-(dog WHERE age > 5).* AS parents FROM dog WHERE age < 2;

返回结果：

[
    {
        age: 0,
        id: dog:whaf_jr,
        parents: [
            {
                age: 8,
                id: dog:mr_whaf
            }
        ]
    },
    {
        age: 1,
        id: dog:whaf_whaf,
        parents: []
    }
]

尝试用count(parents)或parents.len()过滤时，要么返回空结果，要么报错：

SELECT id,age, <-parent_of<-(dog WHERE age > 5).* AS parents FROM dog WHERE age < 2 AND count(parents) > 0;
-- 返回空数组

SELECT id,age, <-parent_of<-(dog WHERE age > 5).* AS parents FROM dog WHERE age < 2 AND parents.len() > 0;
-- 报错：'There was a problem running the len() function. no such method found for the none type'

解决思路 #

问题核心是：当没有匹配的父节点时，parents字段可能为null而非空数组，导致len()报错；count(parents)的统计逻辑也不符合预期。正确做法是在WHERE子句中直接检查是否存在符合条件的父节点关联，而非先投影字段再过滤。

方案1：使用EXISTS关键字验证关联存在性 #

SELECT id, age, <-parent_of<-(dog WHERE age > 5).* AS parents
FROM dog
WHERE age < 2 AND EXISTS (<-parent_of<-(dog WHERE age > 5));

方案2：用MATCH语句清晰表达关联逻辑 #

MATCH $child isa dog, has age $child_age;
MATCH $parent isa dog, has age $parent_age;
MATCH ($parent)-[parent_of]->($child);
WHERE $child_age < 2 AND $parent_age > 5;
SELECT $child.id AS id, $child.age AS age, $parent.* AS parents;

两种查询都会精准返回whaf_jr节点，符合预期需求。

内容的提问来源于stack exchange，提问作者Thomas Hügel