dify - 💡(How to fix) Fix Dataset detail response exposes null summary_index_setting enable [1 pull requests]

AI disclosure: This issue was drafted and analyzed with Codex using GPT-5.5. I have reviewed the analysis, and I am responsible for the content.

Context

Datasets created before summary index settings were populated can have Dataset.summary_index_setting = NULL in the database.

That persistence state means summary index generation has not been configured and should behave as disabled.

Problem

The dataset detail response currently expands that NULL value into an empty nested response model and sends a shape like:

{
  "enable": null,
  "model_name": null,
  "model_provider_name": null,
  "summary_prompt": null
}

This leaks the nullable persistence representation into the Console API contract. enable: null has no useful business meaning for the Console client; the UI already treats missing or null enable values as disabled.

This also creates an invalid shape for downstream request payloads that model summary_index_setting.enable as a boolean discriminator. In the document creation / preview flow, the frontend can keep the dataset detail response in local state and include it in process_rule.summary_index_setting when requesting an indexing estimate.

Expected behavior

The dataset detail API should represent Dataset.summary_index_setting = NULL as summary index generation disabled:

{
  "enable": false,
  "model_name": null,
  "model_provider_name": null,
  "summary_prompt": null
}

The nullable storage detail should stay inside the backend persistence layer instead of becoming part of the Console response contract.

Actual behavior

The dataset detail response exposes enable: null for legacy NULL summary index settings.

Impact

Console clients receive an ambiguous value for a boolean setting and may propagate that invalid shape into follow-up requests such as /datasets/indexing-estimate.