fix: correct file count and table count display in collection info (#223)

Issue 1 - File count mismatch:
- CollectionCatalogInfo now accepts documentCount prop from actual API response
- CollectionDrawer fetches document count from /api/documents endpoint
- Uses authoritative source instead of potentially stale collection_info

Issue 2 - Table count not showing:
- Normalize doc_type_counts keys in _get_document_type_counts()
- Convert 'structured' to 'table' when no subtype is specified
- Ensures consistency with frontend expectations and derive_boolean_flags()

Added unit tests for both frontend and backend changes.

Author: otavionvidia

frontend/src/components/collections/CollectionCatalogInfo.tsx

@@ -19,14 +19,19 @@ import type { Collection } from "../../types/collections";
 
 interface CollectionCatalogInfoProps {
   collection: Collection;
+  /** Optional override for file count - use actual document count from API for accuracy */
+  documentCount?: number;
 }
 
-export function CollectionCatalogInfo({ collection }: CollectionCatalogInfoProps) {
+export function CollectionCatalogInfo({ collection, documentCount }: CollectionCatalogInfoProps) {
   const info = collection.collection_info;
 
   if (!info) {
     return null;
   }
+  
+  // Use actual document count if provided, otherwise fall back to collection_info
+  const fileCount = documentCount ?? info.number_of_files;
 
   const formatDate = (dateString?: string) => {
     if (!dateString) return 'N/A';
@@ -124,12 +129,12 @@ export function CollectionCatalogInfo({ collection }: CollectionCatalogInfoProps
 
         {/* Content Metrics */}
         <Flex gap="density-md" style={{ flexWrap: 'wrap' }}>
-          {/* File Count */}
-          {info.number_of_files !== undefined && (
+          {/* File Count - use actual document count if available */}
+          {fileCount !== undefined && (
             <Flex align="center" gap="density-xs">
               <FileText size={14} style={{ color: 'var(--text-color-subtle)' }} />
               <Text kind="body/regular/sm">
-                {info.number_of_files} files
+                {fileCount} files
               </Text>
             </Flex>
           )}

frontend/src/components/collections/CollectionDrawer.tsx

@@ -18,6 +18,7 @@ import { useNewCollectionStore } from "../../store/useNewCollectionStore";
 import { useCollectionDrawerStore } from "../../store/useCollectionDrawerStore";
 import { useCollectionActions } from "../../hooks/useCollectionActions";
 import { useCollections } from "../../api/useCollectionsApi";
+import { useCollectionDocuments } from "../../api/useCollectionDocuments";
 import { DrawerActions } from "../drawer/DrawerActions";
 import { ConfirmationModal } from "../modals/ConfirmationModal";
 import { Notification, SidePanel, Stack } from "@kui/react";
@@ -44,6 +45,9 @@ export default function CollectionDrawer() {
   // Subscribe to collections query to sync activeCollection with fresh data
   const { data: collections } = useCollections();
 
+  // Fetch actual document count for accurate display
+  const { data: documentsData } = useCollectionDocuments(activeCollection?.collection_name || "");
+  
   // Sync activeCollection with fresh data from query when collections change
   useEffect(() => {
     if (activeCollection && collections) {
@@ -117,7 +121,12 @@ export default function CollectionDrawer() {
       closeOnClickOutside
     >
       <Stack gap="density-md">
-        {activeCollection && <CollectionCatalogInfo collection={activeCollection} />}
+        {activeCollection && (
+          <CollectionCatalogInfo 
+            collection={activeCollection} 
+            documentCount={documentsData?.total_documents}
+          />
+        )}
         <DocumentsList />
       </Stack>
 

frontend/src/components/collections/__tests__/CollectionCatalogInfo.test.tsx

@@ -0,0 +1,170 @@
+// SPDX-FileCopyrightText: Copyright (c) 2025 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
+// SPDX-License-Identifier: Apache-2.0
+
+import { describe, it, expect } from 'vitest';
+import { render, screen } from '../../../test/utils';
+import { CollectionCatalogInfo } from '../CollectionCatalogInfo';
+import type { Collection } from '../../../types/collections';
+
+describe('CollectionCatalogInfo', () => {
+  const createMockCollection = (overrides: Partial<Collection['collection_info']> = {}): Collection => ({
+    collection_name: 'test-collection',
+    num_entities: 100,
+    metadata_schema: [],
+    collection_info: {
+      status: 'Active',
+      date_created: '2026-01-15T10:00:00Z',
+      last_updated: '2026-01-15T12:00:00Z',
+      number_of_files: 5,
+      description: 'Test collection description',
+      ...overrides,
+    },
+  });
+
+  describe('File Count Display', () => {
+    it('displays number_of_files from collection_info when documentCount is not provided', () => {
+      const collection = createMockCollection({ number_of_files: 7 });
+      render(<CollectionCatalogInfo collection={collection} />);
+      
+      expect(screen.getByText('7 files')).toBeInTheDocument();
+    });
+
+    it('displays documentCount prop when provided (overrides collection_info)', () => {
+      const collection = createMockCollection({ number_of_files: 7 });
+      render(<CollectionCatalogInfo collection={collection} documentCount={4} />);
+      
+      // Should show actual document count (4) not the stale collection_info value (7)
+      expect(screen.getByText('4 files')).toBeInTheDocument();
+      expect(screen.queryByText('7 files')).not.toBeInTheDocument();
+    });
+
+    it('displays 0 files when documentCount is 0', () => {
+      const collection = createMockCollection({ number_of_files: 5 });
+      render(<CollectionCatalogInfo collection={collection} documentCount={0} />);
+      
+      expect(screen.getByText('0 files')).toBeInTheDocument();
+    });
+
+    it('falls back to collection_info when documentCount is undefined', () => {
+      const collection = createMockCollection({ number_of_files: 10 });
+      render(<CollectionCatalogInfo collection={collection} documentCount={undefined} />);
+      
+      expect(screen.getByText('10 files')).toBeInTheDocument();
+    });
+
+    it('does not show file count when both are undefined', () => {
+      const collection = createMockCollection({ number_of_files: undefined });
+      render(<CollectionCatalogInfo collection={collection} documentCount={undefined} />);
+      
+      expect(screen.queryByText(/files/)).not.toBeInTheDocument();
+    });
+  });
+
+  describe('Table Count Display', () => {
+    it('displays table count when doc_type_counts.table is present', () => {
+      const collection = createMockCollection({
+        doc_type_counts: { table: 3, text: 10 },
+      });
+      render(<CollectionCatalogInfo collection={collection} />);
+      
+      expect(screen.getByText('3 Tables')).toBeInTheDocument();
+    });
+
+    it('does not display table count when table count is 0', () => {
+      const collection = createMockCollection({
+        doc_type_counts: { table: 0, text: 10 },
+      });
+      render(<CollectionCatalogInfo collection={collection} />);
+      
+      expect(screen.queryByText(/Tables/)).not.toBeInTheDocument();
+    });
+
+    it('does not display table count when doc_type_counts is undefined', () => {
+      const collection = createMockCollection({
+        doc_type_counts: undefined,
+      });
+      render(<CollectionCatalogInfo collection={collection} />);
+      
+      expect(screen.queryByText(/Tables/)).not.toBeInTheDocument();
+    });
+  });
+
+  describe('Chart Count Display', () => {
+    it('displays chart count when doc_type_counts.chart is present', () => {
+      const collection = createMockCollection({
+        doc_type_counts: { chart: 5, text: 10 },
+      });
+      render(<CollectionCatalogInfo collection={collection} />);
+      
+      expect(screen.getByText('5 Charts')).toBeInTheDocument();
+    });
+
+    it('does not display chart count when chart count is 0', () => {
+      const collection = createMockCollection({
+        doc_type_counts: { chart: 0, text: 10 },
+      });
+      render(<CollectionCatalogInfo collection={collection} />);
+      
+      expect(screen.queryByText(/Charts/)).not.toBeInTheDocument();
+    });
+  });
+
+  describe('Image Count Display', () => {
+    it('displays image count when doc_type_counts.image is present', () => {
+      const collection = createMockCollection({
+        doc_type_counts: { image: 12, text: 10 },
+      });
+      render(<CollectionCatalogInfo collection={collection} />);
+      
+      expect(screen.getByText('12 Images')).toBeInTheDocument();
+    });
+
+    it('does not display image count when image count is 0', () => {
+      const collection = createMockCollection({
+        doc_type_counts: { image: 0, text: 10 },
+      });
+      render(<CollectionCatalogInfo collection={collection} />);
+      
+      expect(screen.queryByText(/Images/)).not.toBeInTheDocument();
+    });
+  });
+
+  describe('Combined Content Metrics', () => {
+    it('displays all content types when present', () => {
+      const collection = createMockCollection({
+        number_of_files: 10,
+        doc_type_counts: { table: 3, chart: 2, image: 5, text: 20 },
+      });
+      render(<CollectionCatalogInfo collection={collection} documentCount={10} />);
+      
+      expect(screen.getByText('10 files')).toBeInTheDocument();
+      expect(screen.getByText('3 Tables')).toBeInTheDocument();
+      expect(screen.getByText('2 Charts')).toBeInTheDocument();
+      expect(screen.getByText('5 Images')).toBeInTheDocument();
+    });
+  });
+
+  describe('Status Badge', () => {
+    it('displays Active status with correct styling', () => {
+      const collection = createMockCollection({ status: 'Active' });
+      render(<CollectionCatalogInfo collection={collection} />);
+      
+      expect(screen.getByText('Active')).toBeInTheDocument();
+    });
+  });
+
+  describe('Edge Cases', () => {
+    it('returns null when collection_info is undefined', () => {
+      const collection: Collection = {
+        collection_name: 'test',
+        num_entities: 0,
+        metadata_schema: [],
+        collection_info: undefined,
+      };
+      const { container } = render(<CollectionCatalogInfo collection={collection} />);
+      
+      expect(container.firstChild).toBeNull();
+    });
+  });
+});
+

frontend/src/components/collections/__tests__/CollectionDrawer.test.tsx

@@ -24,10 +24,14 @@ vi.mock('../../../store/useCollectionDrawerStore', () => ({
   useCollectionDrawerStore: vi.fn(() => ({
     activeCollection: {
       collection_name: 'Test Collection',
-      metadata_schema: []
+      metadata_schema: [],
+      collection_info: { status: 'Active' }
     },
     closeDrawer: vi.fn(),
-    toggleUploader: vi.fn()
+    toggleUploader: vi.fn(),
+    deleteError: null,
+    showUploader: false,
+    updateActiveCollection: vi.fn()
   }))
 }));
 
@@ -38,14 +42,34 @@ vi.mock('../../../hooks/useCollectionActions', () => ({
   }))
 }));
 
+vi.mock('../../../api/useCollectionDocuments', () => ({
+  useCollectionDocuments: vi.fn(() => ({
+    data: { total_documents: 5, documents: [] },
+    isLoading: false,
+    error: null
+  }))
+}));
+
+vi.mock('../../../api/useCollectionsApi', () => ({
+  useCollections: vi.fn(() => ({
+    data: [{
+      collection_name: 'Test Collection',
+      metadata_schema: [],
+      collection_info: { status: 'Active' }
+    }],
+    isLoading: false,
+    error: null
+  }))
+}));
+
 // Mock child components that are still used
 interface MockDrawerActionsProps {
   onDelete: () => void;
   onAddSource: () => void;
   isDeleting: boolean;
 }
 
-vi.mock('../../drawer/DrawerActions', () => ({
+vi.mock('../../../components/drawer/DrawerActions', () => ({
   DrawerActions: ({ onDelete, onAddSource, isDeleting }: MockDrawerActionsProps) => (
     <div data-testid="drawer-actions">
       <button data-testid="delete-button" onClick={onDelete} disabled={isDeleting}>
@@ -56,14 +80,22 @@ vi.mock('../../drawer/DrawerActions', () => ({
   )
 }));
 
-vi.mock('../tasks/DocumentsList', () => ({
+vi.mock('../../tasks/DocumentsList', () => ({
   DocumentsList: () => <div data-testid="documents-list">Loading documents...</div>
 }));
 
-vi.mock('../../drawer/UploaderSection', () => ({
+vi.mock('../../../components/drawer/UploaderSection', () => ({
   UploaderSection: () => <div data-testid="uploader-section">Uploader Section</div>
 }));
 
+vi.mock('../CollectionCatalogInfo', () => ({
+  CollectionCatalogInfo: ({ documentCount }: { documentCount?: number }) => (
+    <div data-testid="collection-catalog-info">
+      {documentCount !== undefined && <span>{documentCount} files</span>}
+    </div>
+  )
+}));
+
 describe('CollectionDrawer', () => {
   beforeEach(() => {
     vi.clearAllMocks();

src/nvidia_rag/ingestor_server/main.py

@@ -2764,8 +2764,16 @@ def _get_document_type_counts(
         self, results: list[list[dict[str, str | dict]]]
     ) -> dict[str, int]:
         """
-        Get document type counts from the results
+        Get document type counts from the results.
+        
+        Note: Document types are normalized to standard keys (table, chart, image, text)
+        to ensure consistency with frontend expectations and derive_boolean_flags().
         """
+        # Mapping from nv-ingest types/subtypes to normalized keys
+        type_normalization = {
+            "structured": "table",  # Structured data defaults to table
+        }
+        
         doc_type_counts = defaultdict(int)
         total_documents = 0
         total_elements = 0
@@ -2781,11 +2789,16 @@ def _get_document_type_counts(
                     .get("content_metadata", {})
                     .get("subtype", "")
                 )
+                # Use subtype if available, otherwise use document_type
                 if document_subtype:
-                    document_type_subtype = document_subtype
+                    doc_type_key = document_subtype
                 else:
-                    document_type_subtype = document_type
-                doc_type_counts[document_type_subtype] += 1
+                    doc_type_key = document_type
+                
+                # Normalize the key to standard names (table, chart, image, text)
+                doc_type_key = type_normalization.get(doc_type_key, doc_type_key)
+                
+                doc_type_counts[doc_type_key] += 1
                 if document_type == "text":
                     content = result_element.get("metadata", {}).get("content", "")
                     if isinstance(content, str):