Hi, dawangli ~
Usually people build the lineage of tables through a self-built platform,
there was a DB to persist the relationship between the tables, for each
job, you may need to analyze each SQL which are source tables and which are
sink.
E.G. The INSERT target table is a sink and table after the scan or join is
a source.
If you got the rel tree, you can get the info by a shuttle, if an AST
instead, you can have a SqlVisitor.
Danny Chan 于2020年10月22日周四 下午5:24写道:
> Hi, dawangli ~
>
> Usually people build the lineage of tables through a self-built platform,
> there was a DB to persist the relationship between the tables, for each
> job, you may need to analyze each SQL which are source tables and which are
> sink.
>
> E.G. The INSERT target table is a sink and table after the scan or join is
> a source.
>
> If you got the rel tree, you can get the info by a shuttle, if an AST
> instead, you can have a SqlVisitor.
>
> dawangli 于2020年10月20日周二 下午9:46写道:
>
>> I want to build a lineage system for a real-time data warehouse,how can I
>> extract table and column lineage from flink sql?
>>
>>
>>
>>
>>
>>
>>
>