add filter support to model catalog (kubeflow#1436)

* add filter support to model catalog

Signed-off-by: Adysen Rothman <[email protected]>

* add filter support to yaml

Signed-off-by: Adysen Rothman <[email protected]>

* update tests

Signed-off-by: Adysen Rothman <[email protected]>

* more clear model verbiage & logging

Signed-off-by: Adysen Rothman <[email protected]>

* rm sample rhec catalog --> example in readme

Signed-off-by: Adysen Rothman <[email protected]>

---------

Signed-off-by: Adysen Rothman <[email protected]>

Author: adysenrothman

catalog/internal/catalog/catalog_test.go

@@ -21,7 +21,7 @@ func TestLoadCatalogSources(t *testing.T) {
 		{
 			name:    "test-catalog-sources",
 			args:    args{catalogsPath: "testdata/test-catalog-sources.yaml"},
-			want:    []string{"catalog1", "catalog3"},
+			want:    []string{"catalog1", "catalog3", "catalog4"},
 			wantErr: false,
 		},
 	}
@@ -69,6 +69,11 @@ func TestLoadCatalogSourcesEnabledDisabled(t *testing.T) {
 					Name:    "Catalog 3",
 					Enabled: &trueValue,
 				},
+				"catalog4": {
+					Id:      "catalog4",
+					Name:    "Catalog 4",
+					Enabled: &trueValue,
+				},
 			},
 			wantErr: false,
 		},

catalog/internal/catalog/rhec_catalog.go

@@ -12,7 +12,6 @@ import (
 	"time"
 
 	"github.com/Khan/genqlient/graphql"
-	"github.com/golang/glog"
 	"github.com/kubeflow/model-registry/catalog/internal/catalog/genqlient"
 	"github.com/kubeflow/model-registry/catalog/pkg/openapi"
 	model "github.com/kubeflow/model-registry/catalog/pkg/openapi"
@@ -26,9 +25,8 @@ type rhecModel struct {
 
 // rhecCatalogConfig defines the structure of the RHEC catalog configuration.
 type rhecCatalogConfig struct {
-	Models []struct {
-		Repository string `yaml:"repository"`
-	} `yaml:"models"`
+	Models         []string `yaml:"models"`
+	ExcludedModels []string `yaml:"excludedModels"`
 }
 
 type rhecCatalogImpl struct {
@@ -229,23 +227,12 @@ func newRhecModel(repoData *genqlient.GetRepositoryResponse, imageData genqlient
 	}
 }
 
-func (r *rhecCatalogImpl) load(modelsList []any) error {
+func (r *rhecCatalogImpl) load(modelsList []string, excludedModelsList []string) error {
 	graphqlClient := graphql.NewClient("https://catalog.redhat.com/api/containers/graphql/", http.DefaultClient)
 	ctx := context.Background()
 
 	models := make(map[string]*rhecModel)
-	for _, modelEntry := range modelsList {
-		modelMap, ok := modelEntry.(map[string]any)
-		if !ok {
-			glog.Warningf("skipping invalid entry in 'models' list")
-			continue
-		}
-		repo, ok := modelMap["repository"].(string)
-		if !ok {
-			glog.Warningf("skipping model with missing or invalid 'repository'")
-			continue
-		}
-
+	for _, repo := range modelsList {
 		repoData, err := fetchRepository(ctx, graphqlClient, repo)
 		if err != nil {
 			return err
@@ -261,6 +248,11 @@ func (r *rhecCatalogImpl) load(modelsList []any) error {
 				for _, imageTag := range imageRepository.Tags {
 					tagName := imageTag.Name
 					fullModelName := repo + ":" + tagName
+
+					if isModelExcluded(fullModelName, excludedModelsList) {
+						continue
+					}
+
 					model := newRhecModel(repoData, image, tagName, repo)
 					models[fullModelName] = model
 				}
@@ -275,6 +267,19 @@ func (r *rhecCatalogImpl) load(modelsList []any) error {
 	return nil
 }
 
+func isModelExcluded(modelName string, patterns []string) bool {
+	for _, pattern := range patterns {
+		if strings.HasSuffix(pattern, "*") {
+			if strings.HasPrefix(modelName, strings.TrimSuffix(pattern, "*")) {
+				return true
+			}
+		} else if modelName == pattern {
+			return true
+		}
+	}
+	return false
+}
+
 func newRhecCatalog(source *CatalogSourceConfig) (CatalogSourceProvider, error) {
 	modelsData, ok := source.Properties["models"]
 	if !ok {
@@ -286,11 +291,35 @@ func newRhecCatalog(source *CatalogSourceConfig) (CatalogSourceProvider, error)
 		return nil, fmt.Errorf("'models' property should be a list")
 	}
 
+	models := make([]string, len(modelsList))
+	for i, v := range modelsList {
+		models[i], ok = v.(string)
+		if !ok {
+			return nil, fmt.Errorf("invalid entry in 'models' list, expected a string")
+		}
+	}
+
+	// Excluded models is an optional source property.
+	var excludedModels []string
+	if excludedModelsData, ok := source.Properties["excludedModels"]; ok {
+		excludedModelsList, ok := excludedModelsData.([]any)
+		if !ok {
+			return nil, fmt.Errorf("'excludedModels' property should be a list")
+		}
+		excludedModels = make([]string, len(excludedModelsList))
+		for i, v := range excludedModelsList {
+			excludedModels[i], ok = v.(string)
+			if !ok {
+				return nil, fmt.Errorf("invalid entry in 'excludedModels' list, expected a string")
+			}
+		}
+	}
+
 	r := &rhecCatalogImpl{
 		models: make(map[string]*rhecModel),
 	}
 
-	err := r.load(modelsList)
+	err := r.load(models, excludedModels)
 	if err != nil {
 		return nil, fmt.Errorf("error loading rhec catalog: %w", err)
 	}

catalog/internal/catalog/rhec_catalog_test.go

@@ -342,3 +342,52 @@ func TestRhecCatalogListModels(t *testing.T) {
 		})
 	}
 }
+
+func TestIsModelExcluded(t *testing.T) {
+	tests := []struct {
+		name      string
+		modelName string
+		patterns  []string
+		want      bool
+	}{
+		{
+			name:      "exact match",
+			modelName: "model1:v1",
+			patterns:  []string{"model1:v1"},
+			want:      true,
+		},
+		{
+			name:      "wildcard match",
+			modelName: "model1:v2",
+			patterns:  []string{"model1:*"},
+			want:      true,
+		},
+		{
+			name:      "no match",
+			modelName: "model2:v1",
+			patterns:  []string{"model1:*"},
+			want:      false,
+		},
+		{
+			name:      "multiple patterns with match",
+			modelName: "model3:v1",
+			patterns:  []string{"model2:*", "model3:v1"},
+			want:      true,
+		},
+		{
+			name:      "empty patterns",
+			modelName: "model1:v1",
+			patterns:  []string{},
+			want:      false,
+		},
+	}
+
+	for _, tt := range tests {
+		t.Run(tt.name, func(t *testing.T) {
+			got := isModelExcluded(tt.modelName, tt.patterns)
+			if got != tt.want {
+				t.Errorf("isModelExcluded() = %v, want %v", got, tt.want)
+			}
+		})
+	}
+}

catalog/internal/catalog/testdata/test-catalog-sources.yaml

@@ -21,4 +21,13 @@ catalogs:
     enabled: true
     properties:
       models:
-      - repository: rhelai1/modelcar-granite-7b-starter
+      - rhelai1/modelcar-granite-7b-starter
+  - name: "Catalog 4"
+    id: catalog4
+    type: rhec
+    enabled: true
+    properties:
+      models:
+      - rhelai1/modelcar-granite-7b-starter
+      excludedModels:
+      - rhelai1/modelcar-granite-7b-starter:latest

catalog/internal/catalog/yaml_catalog.go

@@ -150,7 +150,7 @@ func (y *yamlCatalogImpl) GetArtifacts(ctx context.Context, name string) (*model
 	return &list, nil
 }
 
-func (y *yamlCatalogImpl) load(path string) error {
+func (y *yamlCatalogImpl) load(path string, excludedModelsList []string) error {
 	bytes, err := os.ReadFile(path)
 	if err != nil {
 		return fmt.Errorf("failed to read %s file: %v", yamlCatalogPath, err)
@@ -161,9 +161,13 @@ func (y *yamlCatalogImpl) load(path string) error {
 		return fmt.Errorf("failed to parse %s file: %v", yamlCatalogPath, err)
 	}
 
-	models := make(map[string]*yamlModel, len(contents.Models))
+	models := make(map[string]*yamlModel)
 	for i := range contents.Models {
-		models[contents.Models[i].Name] = &contents.Models[i]
+		modelName := contents.Models[i].Name
+		if isModelExcluded(modelName, excludedModelsList) {
+			continue
+		}
+		models[modelName] = &contents.Models[i]
 	}
 
 	y.modelsLock.Lock()
@@ -186,8 +190,26 @@ func newYamlCatalog(source *CatalogSourceConfig) (CatalogSourceProvider, error)
 		return nil, fmt.Errorf("abs: %w", err)
 	}
 
-	p := &yamlCatalogImpl{}
-	err = p.load(yamlModelFile)
+	// Excluded models is an optional source property.
+	var excludedModels []string
+	if excludedModelsData, ok := source.Properties["excludedModels"]; ok {
+		excludedModelsList, ok := excludedModelsData.([]any)
+		if !ok {
+			return nil, fmt.Errorf("'excludedModels' property should be a list")
+		}
+		excludedModels = make([]string, len(excludedModelsList))
+		for i, v := range excludedModelsList {
+			excludedModels[i], ok = v.(string)
+			if !ok {
+				return nil, fmt.Errorf("invalid entry in 'excludedModels' list, expected a string")
+			}
+		}
+	}
+
+	p := &yamlCatalogImpl{
+		models: make(map[string]*yamlModel),
+	}
+	err = p.load(yamlModelFile, excludedModels)
 	if err != nil {
 		return nil, err
 	}
@@ -202,7 +224,7 @@ func newYamlCatalog(source *CatalogSourceConfig) (CatalogSourceProvider, error)
 		for range changes {
 			glog.Infof("Reloading YAML catalog %s", yamlModelFile)
 
-			err = p.load(yamlModelFile)
+			err = p.load(yamlModelFile, excludedModels)
 			if err != nil {
 				glog.Errorf("unable to load YAML catalog: %v", err)
 			}

catalog/internal/catalog/yaml_catalog_test.go

@@ -171,16 +171,37 @@ func TestYAMLCatalogListModels(t *testing.T) {
 	assert.Equal(int32(0), emptyModels.Size)
 	assert.Equal(int32(0), emptyModels.PageSize)
 	assert.Len(emptyModels.Items, 0)
+
+	// Test case 13: Test with excluded models
+	excludedProvider := testYAMLProviderWithExclusions(t, "testdata/test-list-models-catalog.yaml", []any{
+		"model-alpha",
+	})
+	excludedModels, err := excludedProvider.ListModels(ctx, ListModelsParams{})
+	if assert.NoError(err) {
+		assert.NotNil(excludedModels)
+		assert.Equal(int32(5), excludedModels.Size)
+		for _, m := range excludedModels.Items {
+			assert.NotEqual("model-alpha", m.Name)
+		}
+	}
 }
 
 func testYAMLProvider(t *testing.T, path string) CatalogSourceProvider {
+	return testYAMLProviderWithExclusions(t, path, nil)
+}
+
+func testYAMLProviderWithExclusions(t *testing.T, path string, excludedModels []any) CatalogSourceProvider {
+	properties := map[string]any{
+		yamlCatalogPath: path,
+	}
+	if excludedModels != nil {
+		properties["excludedModels"] = excludedModels
+	}
 	provider, err := newYamlCatalog(&CatalogSourceConfig{
-		Properties: map[string]any{
-			yamlCatalogPath: path,
-		},
+		Properties: properties,
 	})
 	if err != nil {
-		t.Fatalf("newYamlCatalog(%s) failed: %v", path, err)
+		t.Fatalf("newYamlCatalog(%s) with exclusions failed: %v", path, err)
 	}
 	return provider
 }

manifests/kustomize/options/catalog/README.md

@@ -32,6 +32,7 @@ The `yaml` type sources model metadata from a local YAML file.
 ##### Properties
 
 - **`yamlCatalogPath`** (*string*, required): The path to the YAML file containing the model definitions. This path is relative to the directory where the `sources.yaml` file is located.
+- **`excludedModels`** (*string list*, optional): A list of models to exclude from the catalog. These can be an exact name with a tag (e.g., `model-a:1.0`) or a pattern ending with `*` to exclude all tags for a repository (e.g., `model-b:*`).
 
 ##### Example
 
@@ -43,6 +44,9 @@ catalogs:
     enabled: true
     properties:
       yamlCatalogPath: sample-catalog.yaml
+      excludedModels:
+      - model-a:1.0
+      - model-b:*
 ```
 
 #### `rhec`
@@ -51,8 +55,8 @@ The `rhec` type sources model metadata from the Red Hat Ecosystem Catalog.
 
 ##### Properties
 
-- **`models`** (*list*, required): A list of models to include from the Red Hat Ecosystem Catalog. Each entry in the list must contain a `repository` field.
-  - **`repository`** (*string*, required): The name of the model repository in the Red Hat Ecosystem Catalog (e.g., `rhelai1/modelcar-granite-7b-starter`).
+- **`models`** (*string list*, required): A list of models to include from the Red Hat Ecosystem Catalog. Each entry contains the full name of the model repository in the Red Hat Ecosystem Catalog (e.g., `rhelai1/modelcar-granite-7b-starter`).
+- **`excludedModels`** (*string list*, optional): A list of models to exclude from the catalog. These can be an exact name with a tag (e.g., `rhelai1/modelcar-granite-7b-starter:b9514c3`) or a pattern ending with `*` to exclude all tags for a repository (e.g., `rhelai1/modelcar-granite-7b-starter:*`).
 
 ##### Example
 
@@ -64,5 +68,8 @@ catalogs:
     enabled: true
     properties:
       models:
-      - repository: rhelai1/modelcar-granite-7b-starter
-```
+      - rhelai1/modelcar-granite-7b-starter
+      excludedModels:
+      - rhelai1/modelcar-granite-7b-starter:v0
+      - rhelai1/modelcar-granite-*
+```

manifests/kustomize/options/catalog/sources.yaml

@@ -5,11 +5,3 @@ catalogs:
   enabled: true
   properties:
     yamlCatalogPath: sample-catalog.yaml
-- name: Red Hat Ecosystem Catalog
-  id: sample_rhec_catalog
-  type: rhec
-  enabled: true
-  properties:
-    models:
-    - repository: rhelai1/modelcar-granite-7b-starter
-