fix: drain HTTP body, paginate Qdrant scroll, deduplicate extractUniqueDirs

internal/indexer/indexer.go

@@ -509,7 +509,7 @@ func Run(configPath string, force bool, logger *slog.Logger) error {
 	var wg sync.WaitGroup
 	var progress atomic.Int64
 
-	totalDirs := len(extractUniqueDirs(walkResult.Files))
+	totalDirs := len(currentDirs)
 	totalItems := len(walkResult.Files) + totalDirs
 
 	// Create directory tracker — directories are analyzed automatically

internal/store/qdrant.go

@@ -39,6 +39,7 @@ func (q *QdrantStore) EnsureCollection() error {
 	if err != nil {
 		return fmt.Errorf("check collection: %w", err)
 	}
+	io.Copy(io.Discard, resp.Body)
 	resp.Body.Close()
 
 	if resp.StatusCode == http.StatusOK {
@@ -159,28 +160,22 @@ func (q *QdrantStore) GetAllFilePoints() ([]*Point, error) {
 	q.logger.Debug("GetAllFilePoints")
 	start := time.Now()
 
-	body := map[string]any{
-		"filter": map[string]any{
-			"must": []map[string]any{
-				{
-					"key": "type",
-					"match": map[string]any{
-						"value": "file",
-					},
+	filter := map[string]any{
+		"must": []map[string]any{
+			{
+				"key": "type",
+				"match": map[string]any{
+					"value": "file",
 				},
 			},
 		},
-		"limit":        1000,
-		"with_payload": true,
-		"with_vector":  false,
 	}
 
-	var result qdrantScrollResponse
-	if err := q.postJSON("/collections/"+q.collection+"/points/scroll", body, &result); err != nil {
+	points, err := q.scrollAll(filter)
+	if err != nil {
 		return nil, err
 	}
 
-	points := parsePoints(result.Result.Points)
 	q.logger.Debug("GetAllFilePoints completed",
 		"count", len(points),
 		"duration", time.Since(start),
@@ -193,28 +188,22 @@ func (q *QdrantStore) GetAllDirPoints() ([]*Point, error) {
 	q.logger.Debug("GetAllDirPoints")
 	start := time.Now()
 
-	body := map[string]any{
-		"filter": map[string]any{
-			"must": []map[string]any{
-				{
-					"key": "type",
-					"match": map[string]any{
-						"value": "directory",
-					},
+	filter := map[string]any{
+		"must": []map[string]any{
+			{
+				"key": "type",
+				"match": map[string]any{
+					"value": "directory",
 				},
 			},
 		},
-		"limit":        1000,
-		"with_payload": true,
-		"with_vector":  false,
 	}
 
-	var result qdrantScrollResponse
-	if err := q.postJSON("/collections/"+q.collection+"/points/scroll", body, &result); err != nil {
+	points, err := q.scrollAll(filter)
+	if err != nil {
 		return nil, err
 	}
 
-	points := parsePoints(result.Result.Points)
 	q.logger.Debug("GetAllDirPoints completed",
 		"count", len(points),
 		"duration", time.Since(start),
@@ -307,6 +296,40 @@ func (q *QdrantStore) Search(vector []float32, limit int) ([]*SearchResult, erro
 	return results, nil
 }
 
+// scrollAll paginates through all points matching the given filter.
+func (q *QdrantStore) scrollAll(filter map[string]any) ([]*Point, error) {
+	const scrollLimit = 1000
+
+	var allPoints []*Point
+	var offset *string
+
+	for {
+		body := map[string]any{
+			"filter":       filter,
+			"limit":        scrollLimit,
+			"with_payload": true,
+			"with_vector":  false,
+		}
+		if offset != nil {
+			body["offset"] = *offset
+		}
+
+		var result qdrantScrollResponse
+		if err := q.postJSON("/collections/"+q.collection+"/points/scroll", body, &result); err != nil {
+			return nil, err
+		}
+
+		allPoints = append(allPoints, parsePoints(result.Result.Points)...)
+
+		if result.Result.NextPageOffset == nil {
+			break
+		}
+		offset = result.Result.NextPageOffset
+	}
+
+	return allPoints, nil
+}
+
 // --- HTTP helpers ---
 
 func (q *QdrantStore) put(path string, body any) error {
@@ -428,7 +451,8 @@ func (q *QdrantStore) postExpectOK(path string, body any) error {
 
 type qdrantScrollResponse struct {
 	Result struct {
-		Points []qdrantPoint `json:"points"`
+		Points        []qdrantPoint `json:"points"`
+		NextPageOffset *string      `json:"next_page_offset"`
 	} `json:"result"`
 }
 

internal/store/qdrant_test.go

@@ -558,3 +558,61 @@ func TestInterfaceCompliance(t *testing.T) {
 	// Compile-time check that QdrantStore implements Store
 	var _ Store = (*QdrantStore)(nil)
 }
+
+func TestGetAllFilePoints_Pagination(t *testing.T) {
+	callCount := 0
+	srv := httptest.NewServer(http.HandlerFunc(func(w http.ResponseWriter, r *http.Request) {
+		body, _ := io.ReadAll(r.Body)
+		var req map[string]any
+		json.Unmarshal(body, &req)
+
+		callCount++
+		var resp string
+		if callCount == 1 {
+			// First page: has next_page_offset
+			if _, hasOffset := req["offset"]; hasOffset {
+				t.Error("first request should not have offset")
+			}
+			resp = `{
+				"result": {
+					"points": [
+						{"id": "uuid-1", "payload": {"file_path": "a.go", "type": "file", "summary": "file a"}},
+						{"id": "uuid-2", "payload": {"file_path": "b.go", "type": "file", "summary": "file b"}}
+					],
+					"next_page_offset": "uuid-2"
+				}
+			}`
+		} else {
+			// Second page: no next_page_offset
+			if req["offset"] != "uuid-2" {
+				t.Errorf("expected offset uuid-2, got %v", req["offset"])
+			}
+			resp = `{
+				"result": {
+					"points": [
+						{"id": "uuid-3", "payload": {"file_path": "c.go", "type": "file", "summary": "file c"}}
+					]
+				}
+			}`
+		}
+		w.WriteHeader(http.StatusOK)
+		w.Write([]byte(resp))
+	}))
+	defer srv.Close()
+
+	s := NewQdrantStore(srv.URL, "vedcode_", "test", 3072, noopLogger)
+	points, err := s.GetAllFilePoints()
+	if err != nil {
+		t.Fatalf("GetAllFilePoints failed: %v", err)
+	}
+
+	if callCount != 2 {
+		t.Errorf("expected 2 scroll requests, got %d", callCount)
+	}
+	if len(points) != 3 {
+		t.Fatalf("expected 3 points across pages, got %d", len(points))
+	}
+	if points[2].FilePath != "c.go" {
+		t.Errorf("expected third point file_path c.go, got %s", points[2].FilePath)
+	}
+}