在 WHERE 子句中使用子查询的函数值 + 使用 DBIx::Class 获取结果

问题描述 投票：0回答：1
我尝试在 DBIx::Class 中构建一个复杂的查询

    预取相关行
    • 计算子查询中的聚合函数
    • 根据函数值过滤结果
    • 使用行获取函数值
在我的应用程序中，这是行数最多的表之一。因此我想在 1 个查询中进行计算和过滤。

我尝试使用 DBIx::Class 的 Git 存储库中的 DBICTest 制作一个简单的示例。请参阅

https://github.com/Perl5/DBIx-Class/tree/maint/0.0828xx/t/lib/DBICTest

BEGIN { do "./t/lib/ANFANG.pm" or die( $@ || $! ) } use strict; use warnings; use lib 'lib'; use DBICTest; $ENV{DBIC_TRACE} = 1; $ENV{DBIC_TRACE_PROFILE} = 'console'; my $schema = DBICTest->init_schema(); my $rs = $schema->resultset('CD'); my $subquery_rs = $rs->search( undef, { '+select' => [ { '' => $rs->search_related('tracks') # in reality I use correlate() ->count_rs->as_query, -as => 'correlated_subquery_col' } ], '+as' => ['correlated_subquery_key'], alias => 'subquery', # from approach number 4 and 6 } )->as_subselect_rs; $subquery_rs->search( { correlated_subquery_col => { '<' => 3 }, }, { alias => 'subquery', # from approach number 5 and 6 prefetch => 'artist', '+columns' => { correlated_subquery_key => 'correlated_subquery_col' }, } )->all; 
像这样运行代码：


$ git clone https://github.com/Perl5/DBIx-Class.git
$ cd DBIx-Class/
$ git checkout master
$ perl test.pl # file shown above

（在我的实际项目中，我使用 DBIx::Class::Helper::Correlate 来汇总相关值，但我无法在示例中使用它。）


对于示例模式，所需的 SQL 可以归结为：


-- pseudo code
SELECT * FROM ( -- all columns from main and related table as well as function value
    SELECT cd.*, -- main table
           artist.*, -- prefetched related table
           (SELECT COUNT(*) FROM track WHERE track.cd = cd.cdid) AS track_count -- subquery with COUNT/SUM
    FROM cds, -- main table
    JOIN artist ON artist.artistid = cd.artist -- related table
)
WHERE track_count < ?

我尝试通过以下步骤更接近解决方案：




  1. 起初我遇到了 PostgreSQL 不允许我的问题
参考附加列中函数的结果，但是
    这个问题帮助了我。
所以我需要使用另一个子查询。我发现并使用了ResultSet->as_subselect_rs()
    第一次创造这个。
    

    

  2. 我需要从外部 
    search()
     调用中的子查询中主动选择函数列。通常我使用 
    '+columns' => { result_hashref_key => 'sql_column_name' }
     语法来选择
额外的列，但我发现下面的语法是必要的，以便定义
列名称 
    correlated_subquery_col
    。名字必须是
也在 WHERE 子句中引用。我不知道这方面的文档
带有空哈希键的奇怪语法
    select => [ { '' => ..., -as => ... } ]
    但受到
    这个问题的启发尝试这个。

    

  3. 如果没有 
    alias
     属性，这实际上在 SQLite 的示例中有效，但在具有我真实模式的 PostgreSQL 中，我收到了错误消息：
    

    

    4. 

DBIx::Class::Storage::DBI::_dbh_execute(): DBI Exception: DBD::Pg::st execute failed:
 ERROR:  column reference "id" is ambiguous
LINE 1: SELECT "items"."id", "items"."purchase_list_id", "items"."va...
               ^ [for Statement "SELECT "items"."id", "items"."purchase_list_id", "items"."value", "items"."offset", "items"."unit_id", "items"."article_id", "items"."purchased", "items"."comment", "items"."servings_sum", "items"."preorder_servings" FROM (SELECT "items"."id", "items"."purchase_list_id", "items"."value", "items"."offset", "items"."unit_id", "items"."article_id", "items"."purchased", "items"."comment", ( (SELECT SUM( "dish"."servings" ) FROM "dish_ingredients" "ingredients_alias"  JOIN "dishes" "dish" ON "dish"."id" = "ingredients_alias"."dish_id" WHERE ( "ingredients_alias"."item_id" = "items"."id" )) ) AS "servings_sum", "article"."id", "article"."project_id", "article"."shop_section_id", "article"."shelf_life_days", "article"."preorder_servings", "article"."preorder_workdays", "article"."name", "article"."comment" FROM "purchase_lists" "me"  JOIN "items" "items" ON "items"."purchase_list_id" = "me"."id"  JOIN "articles" "article" ON "article"."id" = "items"."article_id" WHERE ( ( "article"."preorder_servings" IS NOT NULL AND "article"."preorder_workdays" IS NOT NULL AND "me"."project_id" = ? ) )) "items" WHERE ( "servings_sum" >= "preorder_servings" )" with ParamValues: 1='150']



    在子查询中使用 
  1. alias
     属性 
    search()
     我尝试为子查询提供不同的表别名。这会在 SQLite 中生成以下 SQL：
    

    2. 

SELECT subquery.cdid, subquery.artist, subquery.title, subquery.year, subquery.genreid, subquery.single_track, subquery.correlated_subquery_col, artist.artistid, artist.name, artist.rank, artist.charfield FROM (SELECT subquery.cdid, subquery.artist, subquery.title, subquery.year, subquery.genreid, subquery.single_track, ( (SELECT COUNT( * ) FROM cd me  JOIN track tracks ON tracks.cd = me.cdid) ) AS correlated_subquery_col FROM cd subquery) subquery  JOIN artist artist ON artist.artistid = subquery.artist WHERE ( correlated_subquery_col < ? )

不幸的是，不仅外部查询的子查询别名为 
subquery
，而且在子查询内部，DBIC 尝试从 
subquery.
 进行 SELECT 并使用 
subquery.
 列进行 JOIN。在 PostgreSQL 中，这会产生以下错误：


DBIx::Class::Storage::DBI::_dbh_execute(): DBI Exception: DBD::Pg::st execute failed:
 ERROR:  missing FROM-clause entry for table "subquery"
LINE 1: ...d"  JOIN "articles" "article" ON "article"."id" = "subquery"...
                                                             ^ [for Statement "SELECT "subquery"."id", "subquery"."purchase_list_id", "subquery"."value", "subquery"."offset", "subquery"."unit_id", "subquery"."article_id", "subquery"."purchased", "subquery"."comment", "subquery"."servings_sum", "subquery"."preorder_servings" FROM (SELECT "subquery"."id", "subquery"."purchase_list_id", "subquery"."value", "subquery"."offset", "subquery"."unit_id", "subquery"."article_id", "subquery"."purchased", "subquery"."comment", ( (SELECT SUM( "dish"."servings" ) FROM "dish_ingredients" "ingredients_alias"  JOIN "dishes" "dish" ON "dish"."id" = "ingredients_alias"."dish_id" WHERE ( "ingredients_alias"."item_id" = "items"."id" )) ) AS "servings_sum", "article"."id", "article"."project_id", "article"."shop_section_id", "article"."shelf_life_days", "article"."preorder_servings", "article"."preorder_workdays", "article"."name", "article"."comment" FROM "purchase_lists" "me"  JOIN "items" "items" ON "items"."purchase_list_id" = "me"."id"  JOIN "articles" "article" ON "article"."id" = "subquery"."article_id" WHERE ( ( "article"."preorder_servings" IS NOT NULL AND "article"."preorder_workdays" IS NOT NULL AND "me"."project_id" = ? ) )) "subquery" WHERE ( "servings_sum" >= "preorder_servings" )" with ParamValues: 1='150']



    或者，我尝试在外部查询的 
  1. alias
     中使用 
    search()
     属性。那么别名仅在外部查询中使用，但子查询根本没有别名，并且找不到列。
    

    2. 

DBIx::Class::Storage::DBI::_dbh_execute(): DBI Exception: DBD::Pg::st execute failed:
 ERROR:  missing FROM-clause entry for table "subquery"
LINE 1: SELECT "subquery"."id", "subquery"."purchase_list_id", "subq...
               ^ [for Statement "SELECT "subquery"."id", "subquery"."purchase_list_id", "subquery"."value", "subquery"."offset", "subquery"."unit_id", "subquery"."article_id", "subquery"."purchased", "subquery"."comment", "subquery"."servings_sum", "subquery"."preorder_servings" FROM (SELECT "items"."id", "items"."purchase_list_id", "items"."value", "items"."offset", "items"."unit_id", "items"."article_id", "items"."purchased", "items"."comment", ( (SELECT SUM( "dish"."servings" ) FROM "dish_ingredients" "ingredients_alias"  JOIN "dishes" "dish" ON "dish"."id" = "ingredients_alias"."dish_id" WHERE ( "ingredients_alias"."item_id" = "items"."id" )) ) AS "servings_sum", "article"."id", "article"."project_id", "article"."shop_section_id", "article"."shelf_life_days", "article"."preorder_servings", "article"."preorder_workdays", "article"."name", "article"."comment" FROM "purchase_lists" "me"  JOIN "items" "items" ON "items"."purchase_list_id" = "me"."id"  JOIN "articles" "article" ON "article"."id" = "items"."article_id" WHERE ( ( "article"."preorder_servings" IS NOT NULL AND "article"."preorder_workdays" IS NOT NULL AND "me"."project_id" = ? ) )) "items" WHERE ( "servings_sum" >= "preorder_servings" )" with ParamValues: 1='150']



    最后我尝试在两个 
  1. alias => 'subquery'
     调用中设置 
    search()
    。尽管现在所有 SELECT 列名称中都使用了别名，但这会给出相同的错误消息。但奇怪的是，DBIC 仍然使用默认的表别名 
    me
     并在 PostgreSQL 中写入 
    FROM "purchase_lists" "me"
    ：
    

    2. 

DBIx::Class::Storage::DBI::_dbh_execute(): DBI Exception: DBD::Pg::st execute failed:
 ERROR:  missing FROM-clause entry for table "subquery"
LINE 1: ...d"  JOIN "articles" "article" ON "article"."id" = "subquery"...
                                                             ^ [for Statement "SELECT "subquery"."id", "subquery"."purchase_list_id", "subquery"."value", "subquery"."offset", "subquery"."unit_id", "subquery"."article_id", "subquery"."purchased", "subquery"."comment", "subquery"."servings_sum", "subquery"."preorder_servings" FROM (SELECT "subquery"."id", "subquery"."purchase_list_id", "subquery"."value", "subquery"."offset", "subquery"."unit_id", "subquery"."article_id", "subquery"."purchased", "subquery"."comment", ( (SELECT SUM( "dish"."servings" ) FROM "dish_ingredients" "ingredients_alias"  JOIN "dishes" "dish" ON "dish"."id" = "ingredients_alias"."dish_id" WHERE ( "ingredients_alias"."item_id" = "items"."id" )) ) AS "servings_sum", "article"."id", "article"."project_id", "article"."shop_section_id", "article"."shelf_life_days", "article"."preorder_servings", "article"."preorder_workdays", "article"."name", "article"."comment" FROM "purchase_lists" "me"  JOIN "items" "items" ON "items"."purchase_list_id" = "me"."id"  JOIN "articles" "article" ON "article"."id" = "subquery"."article_id" WHERE ( ( "article"."preorder_servings" IS NOT NULL AND "article"."preorder_workdays" IS NOT NULL AND "me"."project_id" = ? ) )) "subquery" WHERE ( "servings_sum" >= "preorder_servings" )" with ParamValues: 1='150']

...但是 
FROM cd subquery
 在 SQLite 示例中：


SELECT subquery.cdid, subquery.artist, subquery.title, subquery.year, subquery.genreid, subquery.single_track, subquery.correlated_subquery_col, artist.artistid, artist.name, artist.rank, artist.charfield FROM (
    SELECT subquery.cdid, subquery.artist, subquery.title, subquery.year, subquery.genreid, subquery.single_track, ( (
        SELECT COUNT( * ) FROM cd me  JOIN track tracks ON tracks.cd = me.cdid
    ) ) AS correlated_subquery_col
    FROM cd subquery -- uses alias here in SQLite but not in PostgreSQL
) subquery
JOIN artist artist ON artist.artistid = subquery.artist
WHERE ( correlated_subquery_col < ? )

问题：



    如何用DBIC实现我想要的查询？
  • 

    • 我发现使用相同 DBIC 语法的 SQLite 和 PostgreSQL 之间存在差异。这可能是一个错误吗？我可能会尝试进一步减少示例，以便将潜在的错误报告发布到 DBIC 邮件列表。
  • 

    • 

如果您好奇，我真正的疑问是什么：我的应用程序是用于收集食谱和制定饮食计划的。我需要获取所有购买列表项，预取他们的文章，并计算每个列表项的总和，即需要购买该项目的份量。示例行：采购清单项目 100 公斤苹果与文章“苹果”相关，并且两盘菜总共有 25 份，每盘 10 份和 15 份。对于预订，仅显示超过给定份数的购买清单项目。

    



    
    sql
    
    postgresql
    
    perl
    
    dbix-class
    
    
1个回答

    

      
      
      

        

          

            

              
1
投票

              
            

          

        

        
感谢 Mastodon 上的 
barubary 帮助我找到了解决方案。

问题出在 
as
 和 
alias
 键上。这更加简单并且有效：


my $rs = $schema->resultset('CD');

my $subquery_rs = $rs->search(
    undef,
    {       
        '+select' => [
            {     
                '' => $rs->search_related('tracks')    # in reality I use correlate()
                  ->count_rs->as_query,
                -as => 'correlated_subquery_col'
            },
        ],
    }
)->as_subselect_rs;

$subquery_rs->search(
    {
        correlated_subquery_col => { '<' => 3 },
    },
    {
        prefetch   => 'artist',
        '+columns' => { correlated_subquery_key => 'correlated_subquery_col' },
    }
)->all;

它会生成以下 SQL：


SELECT me.cdid, me.artist, me.title, me.year, me.genreid, me.single_track, me.correlated_subquery_col, artist.artistid, artist.name, artist.rank, artist.charfield 
  FROM ( 
    SELECT me.cdid, me.artist, me.title, me.year, me.genreid, me.single_track, ( 
        SELECT COUNT( * ) 
          FROM cd me 
          JOIN track tracks 
            ON tracks.cd = me.cdid
       ) AS correlated_subquery_col 
      FROM cd me
   ) me 
  JOIN artist artist 
    ON artist.artistid = me.artist 
WHERE correlated_subquery_col < '3'

    



      

      
    

  

  

    
最新问题


  





  

    © www.soinside.com 2019 - 2024. All rights reserved.