[nitpick] Missing braces around single-statement else clause. For consistency and maintainability, add braces around the else block.

The catch block with '// do nothing' comment silently ignores parsing errors. Consider logging the exception at debug level to aid troubleshooting while maintaining the fallback behavior.

suggest mark RedisSourceOptions.SINGLE_FIELD_NAME is required when data_type = key

I was looking at the code and noticed that the behavior of the KEY and STRING types seems completely identical.
Is there a specific reason why the KEY type is separated out?
Was KEY perhaps introduced to semantically distinguish it from other types that can hold multiple values?

Also, just to confirm: When key_read_enabled=true and data_type=key, should single_field_name be mandatory?

my bad, when data_type = key. SINGLE_FIELD_NAME should not be required.

suggest this field use parameter, user can change this name. the default value is key

Done

please add some log. or we can't find this case.

Done

Now redis source only support JSON format. This logical is look good to me. But we need think about how to deal with other format.

Thanks for the feedback!
Currently, Redis source only supports JSON format as you mentioned.
To prepare for supporting other formats in the future, I have refactored the parsing logic by defining an interface and implementing format-specific parsers.
This design allows easier extension to support additional formats later on.

Please let me know if you have any further suggestions!

Suggest change to:

.conditional(
RedisSourceOptions.READ_KEY_ENABLED,
true,
RedisSourceOptions.SINGLE_FIELD_NAME,
RedisSourceOptions.KEY_FIELD_NAME
)

@dybyte

@liunaijie
Thank you for the suggestion!
However, I’d like to clarify that single_field_name is only applicable when the value in Redis is a single value .
In other cases, setting single_field_name has no effect.
If we want to make a required option, I believe enforcing key_field_name when read_key_enabled=true would be more appropriate.
What kind of constraint do you think would be appropriate to add?

.conditional(
RedisSourceOptions.READ_KEY_ENABLED,
true,
RedisSourceOptions.SINGLE_FIELD_NAME,
RedisSourceOptions.KEY_FIELD_NAME
)

this means when read_key_enabled = true, then the single_field_name and key_field_name is required.

I just wanted to clarify that I think it would be sufficient to enforce only key_field_name as required when read_key_enabled is true.
I believe making single_field_name mandatory might be unnecessary.
Please let me know if you see any reason otherwise.

the optionRule method can used in:

1. runtime validation
   only requires single_field_name when read_key_enabled = true, as key_field_name is optional due to its default value.
   In this case, we don't need set key_field_name
2. configuration creation
   e.g., in WebUI needs stricter rules:
   when enabling read_key_enabled:
   show key_field_name input with pre-filled default (key)
   show single_field_name as required empty field (must be explicitly set)
   block config submission until single_field_name is provided
   In this case, we need set key_field_name and single_field_name are required when read_key_enabled=true

Given the second scenario, it's advisable to explicitly set both parameters.

I see, that makes sense. I understand that stricter validation is necessary, especially at the WebUI level, to ensure configuration consistency. Thank you for the clarification!