Merge pull request #594 from helixml/fix/knowledge-in-crds

add basic knowledge fields to crd types
helixml · Jan 22, 2025 · ae61de5 · ae61de5
2 parents b26897e + 1e1acd4
commit ae61de5
Show file tree

Hide file tree

Showing 3 changed files with 211 additions and 8 deletions.
diff --git a/operator/api/v1alpha1/aiapp_types.go b/operator/api/v1alpha1/aiapp_types.go
@@ -118,11 +118,48 @@ type TestStep struct {
 	ExpectedOutput string `json:"expected_output"`
 }
 
+// RAGSettings contains settings for RAG processing
+type RAGSettings struct {
+	Threshold          int  `json:"threshold"`
+	ResultsCount       int  `json:"results_count"`
+	ChunkSize          int  `json:"chunk_size"`
+	ChunkOverflow      int  `json:"chunk_overflow"`
+	DisableChunking    bool `json:"disable_chunking"`
+	DisableDownloading bool `json:"disable_downloading"`
+}
+
+// FilestoreSource represents filestore configuration
+type FilestoreSource struct {
+	Path string `json:"path"`
+}
+
+// WebCrawlerConfig contains web crawler settings
+type WebCrawlerConfig struct {
+	Enabled     bool `json:"enabled"`
+	MaxDepth    int  `json:"max_depth"`
+	MaxPages    int  `json:"max_pages"`
+	Readability bool `json:"readability"`
+}
+
+// WebSource represents web source configuration
+type WebSource struct {
+	URLs    []string         `json:"urls"`
+	Crawler WebCrawlerConfig `json:"crawler"`
+}
+
+// KnowledgeSource contains source configuration for knowledge
+type KnowledgeSource struct {
+	Filestore FilestoreSource `json:"filestore"`
+	Web       WebSource       `json:"web"`
+}
+
 // AssistantKnowledge represents knowledge configuration for an assistant
-// Note: This type wasn't in the original types.go, but was referenced.
-// You'll need to define its structure based on your requirements
 type AssistantKnowledge struct {
-	// Add fields as needed
+	Name            string          `json:"name"`
+	RAGSettings     RAGSettings     `json:"rag_settings"`
+	Source          KnowledgeSource `json:"source"`
+	RefreshEnabled  bool            `json:"refresh_enabled"`
+	RefreshSchedule string          `json:"refresh_schedule,omitempty"`
 }
 
 // ToolApiConfig represents API tool configuration

diff --git a/operator/api/v1alpha1/zz_generated.deepcopy.go b/operator/api/v1alpha1/zz_generated.deepcopy.go
diff --git a/operator/config/crd/bases/app.aispec.org_aiapps.yaml b/operator/config/crd/bases/app.aispec.org_aiapps.yaml
@@ -113,10 +113,89 @@ spec:
                     knowledge:
                       description: Knowledge available to the assistant
                       items:
-                        description: |-
-                          AssistantKnowledge represents knowledge configuration for an assistant
-                          Note: This type wasn't in the original types.go, but was referenced.
-                          You'll need to define its structure based on your requirements
+                        description: AssistantKnowledge represents knowledge configuration
+                          for an assistant
+                        properties:
+                          name:
+                            type: string
+                          rag_settings:
+                            description: RAGSettings contains settings for RAG processing
+                            properties:
+                              chunk_overflow:
+                                type: integer
+                              chunk_size:
+                                type: integer
+                              disable_chunking:
+                                type: boolean
+                              disable_downloading:
+                                type: boolean
+                              results_count:
+                                type: integer
+                              threshold:
+                                type: integer
+                            required:
+                            - chunk_overflow
+                            - chunk_size
+                            - disable_chunking
+                            - disable_downloading
+                            - results_count
+                            - threshold
+                            type: object
+                          refresh_enabled:
+                            type: boolean
+                          refresh_schedule:
+                            type: string
+                          source:
+                            description: KnowledgeSource contains source configuration
+                              for knowledge
+                            properties:
+                              filestore:
+                                description: FilestoreSource represents filestore
+                                  configuration
+                                properties:
+                                  path:
+                                    type: string
+                                required:
+                                - path
+                                type: object
+                              web:
+                                description: WebSource represents web source configuration
+                                properties:
+                                  crawler:
+                                    description: WebCrawlerConfig contains web crawler
+                                      settings
+                                    properties:
+                                      enabled:
+                                        type: boolean
+                                      max_depth:
+                                        type: integer
+                                      max_pages:
+                                        type: integer
+                                      readability:
+                                        type: boolean
+                                    required:
+                                    - enabled
+                                    - max_depth
+                                    - max_pages
+                                    - readability
+                                    type: object
+                                  urls:
+                                    items:
+                                      type: string
+                                    type: array
+                                required:
+                                - crawler
+                                - urls
+                                type: object
+                            required:
+                            - filestore
+                            - web
+                            type: object
+                        required:
+                        - name
+                        - rag_settings
+                        - refresh_enabled
+                        - source
                         type: object
                       type: array
                     lora_id: