使用 dbt_utils.union_relations 错误但不知道如何/为啥?

Posted

技术标签:

【中文标题】使用 dbt_utils.union_relations 错误但不知道如何/为啥?【英文标题】:Using dbt_utils.union_relations wrong but don't know how/why?使用 dbt_utils.union_relations 错误但不知道如何/为什么? 【发布时间】:2020-09-03 16:18:23 【问题描述】:

所以我是一个新的 DBT 用户,超级酷的东西,但我遇到了 union_relations 宏的问题。我正在向这个函数提供关系,但是编译/运行的查询没有从关系中找到任何列。

这是我正在运行的代码:


dbt_utils.union_relations(relations=[ref('transform_hist_ca_map_stu_obj_assessment'), ref('transform_hist_sc_map_stu_obj_assessment')])


)
select *
from conformed_obj_assessment
where student_assessment_identifier is not null
and assessment_identifier is not null
and identification_code is not null
and student_unique_id is not null
and performance_level is not null

这是我收到的错误: syntax error at or near "from" LINE 1706: from __dbt__CTE__transform_hist_ca_map_stu_obj_a... ^ compiled SQL at target/run/rally_dw/conformed/conformed_student_objective_assessment.sql

基本上第一列是DBT生成的列,之后应该有来自关系的列,但由于某种原因,这些列没有被拉入。我想知道这是不是因为我想要的关系拉出目前是短暂的,所以没有实现,所以我想知道这是否会导致问题。这是编译后的 SQL,CTE 返回数据,但由于某种原因,它没有被拉入最后一个 CTE。



  create  table "dashboarding"."dev_em_conformed"."conformed_student_objective_assessment__dbt_tmp"
  as (
    
with  __dbt__CTE__historical_ca_map_stu_obj_assessment as (

with hist_ca_map_stu_obj_assess as (
    select * from "dashboarding"."raw_ea"."historical_ca_map_student_obj_assessment"
),
cleaned as (
    select distinct
        source_org,
        assessment_id as assessment_identifier,
        student_assessment_identifier,
        student_unique_id,
        performance_levels  as performance_level,
        scale_score as score,
        assessment_id,
        to_date(test_date, 'YYYY-MM-DD') as test_date,
        identification_code,
        null as parent_objective_assessment_name
    from hist_ca_map_stu_obj_assess
)
select * from cleaned
),  __dbt__CTE__transform_hist_ca_map_stu_obj_assessment as (

with hist_ca_stu_obj_assess as (
    select * from __dbt__CTE__historical_ca_map_stu_obj_assessment
),
final as(
select
    null as source_org,
    student_assessment_identifier,
    assessment_id as assessment_identifier,
    identification_code as identification_code,
    null as school_year,
    student_unique_id,
    null as student_grade_level,
    null as assessment_grade_level,
    NULL as administration_date,
    null as administration_end_date,
    null as objective_assessment_name,
    score,
    performance_level,
    parent_objective_assessment_name,
    null as parent_objective_assessment_id
from hist_ca_stu_obj_assess

)
select * from final
), __dbt__CTE__historical_sc_map_stu_obj_assessment as (

with hist_sc_map_soa as (
    select * from "dashboarding"."raw_ea"."historical_sc_map_student_obj_assessment"
),
cleaned as (
    select distinct
        source_org,
        assessment_id as assessment_identifier,
        student_assessment_identifier,
        student_unique_id,
        performance_levels as performance_level,
        scale_score as score,
        assessment_id,
        to_date(test_date, 'YYYY-MM-DD') as test_date,
        identification_code,
        null as parent_objective_assessment_name
    from hist_sc_map_soa
)
select * from cleaned
),  __dbt__CTE__transform_hist_sc_map_stu_obj_assessment as (

with hist_sc_stu_obj_assess as (
    select * from __dbt__CTE__historical_sc_map_stu_obj_assessment
),
final as(
select
    null as source_org,
    student_assessment_identifier,
    assessment_id as assessment_identifier,
    identification_code as identification_code,
    null as school_year,
    student_unique_id,
    null as student_grade_level,
    null as assessment_grade_level,
    NULL as administration_date,
    null as administration_end_date,
    null as objective_assessment_name,
    score,
    performance_level,
    parent_objective_assessment_name,
    null as parent_objective_assessment_id
from hist_sc_stu_obj_assess

)
select * from final
),  conformed_obj_assessment as(



        (
            select

                cast('__dbt__CTE__transform_hist_ca_map_stu_obj_assessment' as 
    varchar
) as _dbt_source_relation,
                ---NO MORE COLUMNS???

            from __dbt__CTE__transform_hist_ca_map_stu_obj_assessment
        )

        union all
        

        (
            select

                cast('__dbt__CTE__transform_hist_sc_map_stu_obj_assessment' as 
    varchar
) as _dbt_source_relation,
                 ---NO MORE COLUMNS??

            from __dbt__CTE__transform_hist_sc_map_stu_obj_assessment
        )

        


)
select *
from conformed_obj_assessment
where student_assessment_identifier is not null
and assessment_identifier is not null
and identification_code is not null
and student_unique_id is not null
and performance_level is not null

  );

任何想法都将非常感谢谢谢!

【问题讨论】:

您能分享您的packages.yml 文件内容吗?这将有助于告知我的答案! 【参考方案1】:

union_relations 宏依赖于了解存储在信息架构中的关系(表/视图)中有哪些列。由于这个模型是短暂的,信息模式中没有任何记录,这就是为什么会有这样的 SQL:

select

cast('__dbt__CTE__transform_hist_ca_map_stu_obj_assessment' as 
    varchar
) as _dbt_source_relation,

from __dbt__CTE__transform_hist_ca_map_stu_obj_assessment

我注意到您使用的是稍旧版本的 dbt-utils — 虽然我们尚未修复此问题,但我们有 improved the way this issue is handled(在 v0.5.0 中发布)。

较新版本的 dbt-utils 会告诉您以下信息:

Compilation Error in model test_ephemeral (models/test_ephemeral.sql)

  The `union_relations` macro cannot be used with ephemeral models, as it relies on the information schema.

  `__dbt__CTE__my_ephemeral` is an ephemeral model. Consider making is a view or table instead.` is an ephemeral model. Consider making is a view or table instead.

因此,正如(新)错误消息所暗示的那样——解决此问题的唯一方法是让您的上游模型成为视图或表格。

【讨论】:

我想了很多,但很高兴确认我没有错误地考虑这个问题。感谢您的帮助!

以上是关于使用 dbt_utils.union_relations 错误但不知道如何/为啥?的主要内容,如果未能解决你的问题,请参考以下文章

在使用加载数据流步骤的猪中,使用(使用 PigStorage)和不使用它有啥区别?

今目标使用教程 今目标任务使用篇

Qt静态编译时使用OpenSSL有三种方式(不使用,动态使用,静态使用,默认是动态使用)

MySQL db 在按日期排序时使用“使用位置;使用临时;使用文件排序”

使用“使用严格”作为“使用强”的备份

Kettle java脚本组件的使用说明(简单使用升级使用)