unnecessary default dataset_ids in the 'Knowledge Retrieval + Chatbot' and 'Question Classifier + Knowledge + Chatbot' app templates

Self Checks

[X] This is only for bug report, if you would like to ask a question, please head to Discussions.
[X] I have searched for existing issues search for existing issues, including closed ones.
[X] I confirm that I am using English to submit this report (我已阅读并同意 Language Policy).
[X] Please do not modify this template :) and fill in all the required fields.

Dify version

0.6.11

Cloud or Self Hosted

Cloud

Steps to reproduce

Create an app from the template using "Knowledge Retrieval + Chatbot."
On the app detail page, select the "Knowledge Retrieval" node and press the "Add Knowledge" button.
In the knowledge list modal, it shows "1 Knowledge selected" even when no knowledge is selected.
Select a piece of knowledge and press "Add." The Knowledge Retrieval panel then displays two items: one user-selected and one empty.
Attempting to configure the empty knowledge causes the entire page to crash.

✔️ Expected Behavior

Knowledge selected as the user expected

❌ Actual Behavior

empty knowledge selected as default in template "Knowledge Retreival + Chatbot" has one default knowlege "Create app from Question Classifier + Knowledge + Chatbot" has two empty default knowledge

To resolve the issue of unnecessary default dataset_ids in the "Knowledge Retrieval + Chatbot" and "Question Classifier + Knowledge + Chatbot" app templates, you need to ensure that the default values for the dataset_ids are correctly set to an empty array. Here are the steps and code adjustments you can make:

Set Default Values for Knowledge Retrieval Node: Ensure the dataset_ids default value is an empty array:

defaultValue: {
   query_variable_selector: [],
   dataset_ids: [],
   retrieval_mode: RETRIEVE_TYPE.oneWay,
}

Validation Logic for Knowledge Retrieval Node: Ensure the validation logic checks for the presence of required fields:

checkValid(payload: KnowledgeRetrievalNodeType, t: any) {
   let errorMessages = ''
   if (!errorMessages && (!payload.query_variable_selector || payload.query_variable_selector.length === 0))
       errorMessages = t(`${i18nPrefix}.errorMsg.fieldRequired`, { field: t(`${i18nPrefix}.nodes.knowledgeRetrieval.queryVariable`) })

   if (!errorMessages && (!payload.dataset_ids || payload.dataset_ids.length === 0))
       errorMessages = t(`${i18nPrefix}.errorMsg.fieldRequired`, { field: t(`${i18nPrefix}.nodes.knowledgeRetrieval.knowledge`) })

   if (!errorMessages && payload.retrieval_mode === RETRIEVE_TYPE.multiWay && !payload.multiple_retrieval_config?.reranking_model?.provider)
       errorMessages = t(`${i18nPrefix}.errorMsg.fieldRequired`, { field: t(`${i18nPrefix}.errorMsg.fields.rerankModel`) })

   if (!errorMessages && payload.retrieval_mode === RETRIEVE_TYPE.oneWay && !payload.single_retrieval_config?.model?.provider)
       errorMessages = t(`${i18nPrefix}.errorMsg.fieldRequired`, { field: t('common.modelProvider.systemReasoningModel.key') })

   return {
       isValid: !errorMessages,
       errorMessage: errorMessages,
   }
}

Set Default Values for Question Classifier Node: Ensure the dataset_ids default value is an empty array:

defaultValue: {
   query_variable_selector: [],
   model: {
       provider: '',
       name: '',
       mode: 'chat',
       completion_params: {
           temperature: 0.7,
       },
   },
   classes: [
       {
           id: '1',
           name: '',
       },
       {
           id: '2',
           name: '',
       },
   ],
}

Validation Logic for Question Classifier Node: Ensure the validation logic checks for the presence of required fields:

checkValid(payload: QuestionClassifierNodeType, t: any) {
   let errorMessages = ''
   if (!errorMessages && (!payload.query_variable_selector || payload.query_variable_selector.length === 0))
       errorMessages = t(`${i18nPrefix}.errorMsg.fieldRequired`, { field: t(`${i18nPrefix}.nodes.questionClassifiers.inputVars`) })

   if (!errorMessages && !payload.model.provider)
       errorMessages = t(`${i18nPrefix}.errorMsg.fieldRequired`, { field: t(`${i18nPrefix}.nodes.questionClassifiers.model`) })

   if (!errorMessages && (!payload.classes || payload.classes.length === 0))
       errorMessages = t(`${i18nPrefix}.errorMsg.fieldRequired`, { field: t(`${i18nPrefix}.nodes.questionClassifiers.class`) })

   if (!errorMessages && (payload.classes.some(item => !item.name)))
       errorMessages = t(`${i18nPrefix}.errorMsg.fieldRequired`, { field: t(`${i18nPrefix}.nodes.questionClassifiers.topicName`) })

   return {
       isValid: !errorMessages,
       errorMessage: errorMessages,
   }
}

By ensuring these default values and validation checks are correctly implemented, you should be able to resolve the issue of unnecessary default dataset_ids and prevent the app from crashing when configuring empty knowledge items ^[1]^[2].

To continue talking to Dosu, mention @dosu.

langgenius / dify