Solr 查询不在嵌套子文档上

Question

如何查询父文档，而子文档没有具有特定字段值？

例如： 假设我们有以下数据结构：

    {
        "type_s": "book",
        "id_l": 4294967298,
        "title_s": "The Little Mermaid"
        {
            "type_s": "review",
            "id_l": "4294967451",       
            "reviewer_s": "Freeman, Gordon",        
            "comment_s": "Great book!"      
        },
        {
            "type_s": "review",
            "id_l": "4294967452",       
            "reviewer_s": "Denton, J.C.",       
            "comment_s": "My daughter loved it!"        
        }
    },
{
        "type_s": "book",
        "id_l": 4294967298,
        "title_s": "Lion King"
        {
            "type_s": "review",
            "id_l": "4294967457",       
            "reviewer_s": "Woods, Susanne",     
            "comment_s": "One of the best!"     
        },
        {
            "type_s": "review",
            "id_l": "4294967458",       
            "reviewer_s": "Denver, Michel",     
            "comment_s": "Liked the ending!"        
        }
    },
    {
        "type_s": "book",
        "id_l": 4294967298,
        "title_s": "7 dwarves"
        {
            "type_s": "review",
            "id_l": "4294967453",       
            "reviewer_s": "Freeman, Gordon",        
            "comment_s": "Great book!"      
        },
        {
            "type_s": "review",
            "id_l": "4294967454",       
            "reviewer_s": "Delacroix, Marie",       
            "comment_s": "Too many dwarves!"        
        }
    }

如果我想让所有的书都得到“弗里曼”的评论，我会这样做：

&fq={!parent which='type_s:book'}type_s:review AND reviewer_s:Freeman

这会给我两本书。

但是，如果我想让所有的书都没有“弗里曼”的评论，我该怎么办？

我试过这样

&fq={!parent which='type_s:book'}type_s:review AND reviewer_s:(NOT Freeman)

这给了我 0 个结果

还有这个

&fq={!parent which='type_s:book'}type_s:review AND NOT reviewer_s:Freeman)

这给了我所有的父文件。

下面的内容更有希望，它给了我一些结果（在我的真实用例中）

&fq={!parent which='type_s:book'}type_s:review AND -reviewer_s:["" TO *]

请注意，我也尝试使用单引号中的搜索词进行查询。

Answer 1

如果 solr 具有 elasticsearch 中的 include_in_parent 之类的功能，那将是可以实现的。但是，如果您以不同的方式对数据进行索引，您将能够实现您想要的。您需要将审阅者索引为父文档中的多值字段（实际上，当您使用 include_in_parent 时，elasticsearch 在幕后做了这件事情）：

[{                                                                                                                                                                                                        
    "type_s": "book",                                                                                                                                                                                     
    "id": 4294967298,                                                                                                                                                                                     
    "title_s": "The Little Mermaid",                                                                                                                                                                      
    "reviewers_ms": ["Freeman, Gordon", "Denton, J.C."],                                                                                                                                                  
    ...                                                                                                                                                                                                   
}, {                                                                                                                                                                                                      
    "type_s": "book",                                                                                                                                                                                     
    "id": 4294967299,                                                                                                                                                                                     
    "title_s": "Lion King",                                                                                                                                                                               
    "reviewers_ms": ["Woods, Susanne", "Denver, Michel"],                                                                                                                                                 
    ...                                                                                                                                                                                                   
}, {                                                                                                                                                                                                      
    "type_s": "book",                                                                                                                                                                                     
    "id": 4294967300,                                                                                                                                                                                     
    "title_s": "7 dwarves",                                                                                                                                                                               
    "reviewers_ms": ["Freeman, Gordon", "Delacroix, Marie"],                                                                                                                                              
    ...                                                                                                                                                                                                   
}]

然后，如果您按以下条件过滤，您将得到想要的结果：

type_s:book AND -reviewers_ms:"Freeman, Gordon"

更新

我找到了一种无需特殊索引即可解决问题的方法。这个过滤器对我有用：

type_s:book AND -{!parent which='type_s:book' v='reviewer_s:"Freeman, Gordon"'}

此语法还应该解决您在评论中提到的问题

但是，如果我有另一个字段，在 嵌套文档，我想以组合方式过滤？

type_s:book AND -{!parent which='type_s:book' v='reviewer_s:"Freeman, Gordon" AND type_s:"review"'}