addressed review comments

Signed-off-by: Maroon Ayoub <[email protected]>

Author: vMaroon

pkg/epp/scheduling/local_config.go

@@ -20,7 +20,6 @@ import (
 	"context"
 
 	"sigs.k8s.io/controller-runtime/pkg/log"
-	"sigs.k8s.io/gateway-api-inference-extension/pkg/epp/scheduling/plugins/filter"
 	"sigs.k8s.io/gateway-api-inference-extension/pkg/epp/scheduling/plugins/picker"
 	"sigs.k8s.io/gateway-api-inference-extension/pkg/epp/scheduling/plugins/scorer"
 	envutil "sigs.k8s.io/gateway-api-inference-extension/pkg/epp/util/env"
@@ -31,7 +30,7 @@ const (
 	kvCacheScorerEnablementEnvVar   = "ENABLE_KVCACHE_AWARE_SCORER"
 	loadAwareScorerEnablementEnvVar = "ENABLE_LOAD_AWARE_SCORER"
 	prefixScorerEnablementEnvVar    = "ENABLE_PREFIX_AWARE_SCORER"
-	pdFilterEnablementEnvVar = "ENABLE_PD_FILTER"
+	pdFilterEnablementEnvVar        = "ENABLE_PD_FILTER"
 
 	kvCacheScorerWeightEnvVar   = "KVCACHE_AWARE_SCORER_WEIGHT"
 	loadAwareScorerWeightEnvVar = "LOAD_AWARE_SCORER_WEIGHT"
@@ -86,19 +85,6 @@ func setKVCacheAwareScorer() {
 	loggerDebug.Info("Initialized KVCacheAwareScorer", "weight", kvCacheScorerWeight)
 }
 
-func setPDFilter() {
-	ctx := context.Background()
-	loggerDebug := log.FromContext(ctx).WithName("scheduler_config").V(logutil.DEBUG)
-
-	if envutil.GetEnvString(pdFilterEnablementEnvVar, "false", loggerDebug) != "true" {
-		loggerDebug.Info("Skipping PDFilter creation as it is not enabled")
-		return
-	}
-
-	defaultConfig.filters = append(defaultConfig.filters, filter.PDFilter)
-	loggerDebug.Info("Initialized PDFilter")
-}
-
 func setPrefixScorer() {
 	ctx := context.Background()
 	loggerDebug := log.FromContext(ctx).WithName("scheduler_config").V(logutil.DEBUG)

pkg/epp/scheduling/plugins/scorer/prefix_aware_scorer.go

@@ -98,3 +98,33 @@ func (s *PrefixAwareScorer) PostResponse(ctx *types.SchedulingContext, pod types
 func (s *PrefixAwareScorer) GetPrefixStore() *PrefixStore {
 	return s.prefixStore
 }
+
+// podToKey is a function type that converts a Pod to a string key.
+// It returns the key and a boolean indicating success.
+type podToKeyFunc func(pod types.Pod) (string, bool)
+
+func indexedScoresToNormalizedScoredPods(pods []types.Pod, podToKey podToKeyFunc,
+	scores map[string]int) map[types.Pod]float64 {
+	scoredPods := make(map[types.Pod]float64)
+	minScore, maxScore := getMinMax(scores)
+
+	for _, pod := range pods {
+		key, ok := podToKey(pod)
+		if !ok {
+			continue
+		}
+
+		if score, ok := scores[key]; ok {
+			if minScore == maxScore {
+				scoredPods[pod] = 1.0
+				continue
+			}
+
+			scoredPods[pod] = float64(score-minScore) / float64(maxScore-minScore)
+		} else {
+			scoredPods[pod] = 0.0
+		}
+	}
+
+	return scoredPods
+}

pkg/epp/scheduling/plugins/scorer/prefix_store.go

@@ -89,7 +89,7 @@ func NewPrefixStore(config *PrefixStoreConfig) *PrefixStore {
 
 // AddEntry adds a new entry to the prefix store.
 func (s *PrefixStore) AddEntry(modelName string, prompt string, pod *types.NamespacedName) error {
-	if prompt == "" || pod == nil {
+	if prompt == "" || pod == nil || len(prompt) < s.blockSize /* skip if prompt is too short */ {
 		return nil
 	}
 
@@ -111,7 +111,7 @@ func (s *PrefixStore) AddEntry(modelName string, prompt string, pod *types.Names
 	for start := 0; start < len(prompt); start += s.blockSize {
 		end := start + s.blockSize
 		if end > len(prompt) {
-			end = len(prompt)
+			break // skip partial blocks
 		}
 
 		// Compute the hash for the current block
@@ -142,6 +142,10 @@ func (s *PrefixStore) AddEntry(modelName string, prompt string, pod *types.Names
 // FindMatchingPods finds all pods that match the given prompt and model name.
 // It returns a map of pods and the number of blocks they match.
 func (s *PrefixStore) FindMatchingPods(prompt, modelName string) map[string]int {
+	if prompt == "" || modelName == "" || len(prompt) < s.blockSize /* skip if prompt is too short */ {
+		return nil
+	}
+
 	s.RLock()
 	cache, ok := s.store[modelName] // cache is thread-safe
 	s.RUnlock()