Missing Source Table for MERGE statement when UNION involved in source subquery

[entilo]

version: sqllineage 1.4.2

when i use LineageRunner to parse sql below, it turned out error like:"return name.strip("`").strip('"').strip("'") AttributeError: 'NoneType' object has no attribute 'strip'":

create or replace view scn.vwn as
SELECT 1 as file_type
FROM scn.tbn

LineageRunner parse error

[entilo]

error sql:
create or replace view scn.vwn as
SELECT 1 as file_type
FROM scn.tbn

correct sql:
create or replace view scn.vwn as
SELECT 1 file_type
FROM scn.tbn

there is something wrong with the "as", i dont konw why

[reata]

This happens only with default non-validating dialect, where sqlparse defines FILE_TYPE as a builtin data type, thus not allowed as alias name.

Can you try switch to ansi dialect? In the long run, we will deprecate non-validating support.

[entilo]

您好，邮件已收到，祝您天天好心情！

[entilo]

This happens only with default non-validating dialect, where sqlparse defines FILE_TYPE as a builtin data type, thus not allowed as alias name.

Can you try switch to ansi dialect? In the long run, we will deprecate non-validating support.

Yes , LineageRunner(sql, dialect="ansi") solved this problem. thank you.

But there are another question : "LineageRunner(sql)" is ok to work with sql like below ,but "LineageRunner(sql, dialect="ansi")" not,

merge into tb_report_risk_day t using (
select * from (
select * from taba a
union all
select * from tabb b
)tb
inner join ext_tab et on tb.aaa = et.aaa)
s on (t.cpdm = s.vc_cpdm) when matched then update set t.upd = s.upd;

its seems another error when using dialect="ansi" to do with "union all" :
if segment.type == "bracketed" and is_union(segment):
AttributeError: 'NoneType' object has no attribute 'type'

[reata]

Can you try upgrading sqllineage to v1.4.5? I believe this is already fixed.

[entilo]

Can you try upgrading sqllineage to v1.4.5? I believe this is already fixed.
sqllineage 1.4.5 用下面sql进行解析有问题：
merge into tb_report_risk_day t using (
select * from (
select * from taba a
union all
select * from tabb b
)tb
inner join ext_tab et on tb.aaa = et.aaa)
s on (t.cpdm = s.vc_cpdm) when matched then update set t.upd = s.upd;

LineageRunner(sql=veryfy_sql)
源表为:[Table: .ext_tab, Table: .taba, Table: .tabb]
目标表为: .tb_report_risk_day

LineageRunner(sql=veryfy_sql, dialect="ansi")
源表为:[]
目标表为: .tb_report_risk_day

LineageRunner方言模式没有报错，但是解析错误 ==

[reata]

I can this is a bug that we need to fix. This is related to UNION ALL in source subquery.

[reata]

I did some research and this test case can be simplified to

SELECT *
FROM (SELECT *
      FROM foo
      UNION ALL
      SELECT *
      FROM bar) s1
         CROSS JOIN baz

where ANSI dialect still has problem identifying all source table correctly, it only recognizes baz.

update: it's still only triggered with merge