feat: enhance community handling - add GraphFingerprint method and skip finalization if unchanged; update min community size default

Author: aksOps

internal/config/config.go

@@ -81,7 +81,7 @@ func Load(cfgFile string) (*Config, error) {
 	v.SetDefault("indexing.extract_graph", true)
 	v.SetDefault("indexing.extract_claims", true)
 	v.SetDefault("indexing.max_gleanings", 1)
-	v.SetDefault("community.min_community_size", 3)
+	v.SetDefault("community.min_community_size", 2)
 	v.SetDefault("community.max_levels", 3)
 	v.SetDefault("server.host", "127.0.0.1")
 	v.SetDefault("server.port", 8080)

internal/pipeline/pipeline.go

@@ -181,6 +181,12 @@ func (p *Pipeline) IndexURL(ctx context.Context, rootURL string, opts IndexOptio
 
 // indexFile processes a single file through Phases 1-2.
 func (p *Pipeline) indexFile(ctx context.Context, path string, opts IndexOptions) error {
+	absPath, err := filepath.Abs(path)
+	if err != nil {
+		return fmt.Errorf("resolve absolute path: %w", err)
+	}
+	path = absPath
+
 	data, err := os.ReadFile(path)
 	if err != nil {
 		return err
@@ -550,6 +556,17 @@ func (p *Pipeline) Finalize(ctx context.Context, verbose bool) error {
 	if err != nil {
 		return fmt.Errorf("load relationships: %w", err)
 	}
+
+	// Check if finalization can be skipped: communities already exist and
+	// entity/relationship counts haven't changed since last run.
+	existingEntities, existingRels, existingComms, fpErr := p.store.GraphFingerprint(ctx)
+	if fpErr == nil && existingComms > 0 &&
+		existingEntities == len(entities) && existingRels == len(rels) {
+		slog.Info("⏭️ skipping finalization — graph unchanged since last run",
+			"entities", len(entities), "relationships", len(rels), "communities", existingComms)
+		return nil
+	}
+
 	slog.Info("🧩 Phase 3: running Louvain community detection",
 		"entities", len(entities), "relationships", len(rels))
 
@@ -640,6 +657,12 @@ func (p *Pipeline) Finalize(ctx context.Context, verbose bool) error {
 		}
 	}
 
+	if len(workItems) == 0 {
+		slog.Warn("⚠️ all communities filtered out by min_community_size",
+			"min_community_size", p.cfg.Community.MinCommunitySize,
+			"hint", "lower community.min_community_size in config or index more documents")
+		return nil
+	}
 	slog.Info("🧩 Phase 4: summarising communities", "communities", len(workItems))
 
 	type commResult struct {

internal/store/store.go

@@ -778,6 +778,17 @@ func (s *Store) ClearCommunities(ctx context.Context) error {
 	return err
 }
 
+// GraphFingerprint returns counts of entities, relationships, and communities
+// to detect whether the graph has changed since last finalization.
+func (s *Store) GraphFingerprint(ctx context.Context) (entities, relationships, communities int, err error) {
+	err = s.db.QueryRowContext(ctx, `
+		SELECT
+			(SELECT COUNT(*) FROM entities),
+			(SELECT COUNT(*) FROM relationships),
+			(SELECT COUNT(*) FROM communities)`).Scan(&entities, &relationships, &communities)
+	return
+}
+
 func scanCommunity(row *sql.Row) (*Community, error) {
 	var c Community
 	var blob []byte