Hi Chang,

The hive-hcatalog-core-1.1.0-cdh5.13.1.jar jar file is not something
maintained by Apache. For vendor specific problems you should reach
out to the respective support team from where you obtained the
product.

Apart from that the version that you are using (5.13.1) is quite old.
Please re-try your use-case with the latest Apache Hive 4.0.0-beta-1
release [1] and report back if you still observe unexpected behavior.

Best,
Stamatis

[1] https://hive.apache.org/general/downloads/

On Mon, Jan 29, 2024 at 5:52 AM chang.wd <chang...@foxmail.com> wrote:
>
> Dear Hive Support Team,
>
> I hope you are doing well. I am writing to inquire about a specific behavior 
> I encountered in Hive, related to the 
> org.apache.hadoop.hive.serde2.objectinspector.StandardStructObjectInspector 
> class.
>
> Sql to reply this behavior:
> ```
> -- add JsonSerDe jar
> ADD JAR hive-hcatalog-core-1.1.0-cdh5.13.1.jar;
> -- create json table, the `struct<MD5:string>` will become to lower case: 
> `struct<md5:string>`.
> CREATE TABLE `test.hive_json_struct_schema`(
>   `cond_keys` struct<MD5:string>
> )
> ROW FORMAT SERDE
>   'org.apache.hive.hcatalog.data.JsonSerDe'
> STORED AS INPUTFORMAT
>   'org.apache.hadoop.mapred.TextInputFormat'
> OUTPUTFORMAT
>   'org.apache.hadoop.hive.ql.io.HiveIgnoreKeyTextOutputFormat'
> ```
>
> When using the StandardStructObjectInspector class, it appears that field 
> names are being automatically converted to lowercase in the following code 
> snippet:
>
> ```
> this.fieldName = fieldName.toLowerCase();
> ```
>
> This behavior subsequently causes issues when querying JSON formatted tables, 
> particularly when nested Struct field names within the JSON data contain a 
> mix of uppercase and lowercase characters. Since field names are being 
> changed to lowercase by the StandardStructObjectInspector class, the actual 
> field names no longer match the expected field names, which leads to errors 
> when reading the data.(Not with SQL)
>
> I would appreciate if you could kindly provide an explanation for this design 
> choice and whether there are any available workarounds or alternative 
> solutions for this scenario. I understand that the class may have been 
> implemented to avoid case sensitivity issues, but in cases like mine where 
> field name case matters, it would be helpful to have a better understanding 
> of how to handle this situation.
>
> Thank you in advance for your assistance and guidance. I look forward to 
> hearing from you.
>
> Best regards,
>
> Chang

Reply via email to