[ 
https://issues.apache.org/jira/browse/CALCITE-5677?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17717294#comment-17717294
 ] 

Tanner Clary commented on CALCITE-5677:
---------------------------------------

[~dmsysolyatin] I personally am in favor of removing the unparseCall override 
in {{SqlSubstringFunction}} (which would change the function signature to be 
comma-separated rather than use keywords). I think if there are dialects which 
only support {{SUBSTRING(value FROM position FOR length)}} as [~julianhyde] 
mentioned, then perhaps we could modify those dialect files rather than have it 
be the default as it currently is. I can look more into which dialects support 
which form and will add a commit/updated comment once that's done. 

[~julianhyde] Sorry about that, let me know if the new commit message works. 

> Use comma-separated arguments in BigQuery SUBSTR
> ------------------------------------------------
>
>                 Key: CALCITE-5677
>                 URL: https://issues.apache.org/jira/browse/CALCITE-5677
>             Project: Calcite
>          Issue Type: Bug
>            Reporter: Tanner Clary
>            Assignee: Tanner Clary
>            Priority: Major
>              Labels: pull-request-available
>
> BigQuery's {{SUBSTR(value, position[, length])}} function currently gets 
> translated to the standard {{SUBSTRING(value FROM position[ FOR length])}}. 
> BigQuery expects the arguments to be comma-separated rather than the use of 
> the {{FROM}} and {{FOR}} keywords. 
> EXAMPLE: {{SUBSTR('hello', 2)}} would get translated to {{SUBSTRING('hello' 
> FROM 2)}} which BigQuery does not recognize.
> Hive addresses this issue by adding onto {{HiveSqlDialect#unparseCall}} 
> method. I believe BigQuery should do something similar to avoid this 
> incorrect translation. The docs for the {{SUBSTR}} (and alias {{SUBSTRING}}) 
> function may be found 
> [here|https://cloud.google.com/bigquery/docs/reference/standard-sql/string_functions#substr].
> I will open a PR shortly that implements my suggestion and adds a couple of 
> tests as well.



--
This message was sent by Atlassian Jira
(v8.20.10#820010)

Reply via email to