add data viewer

app/main.py

@@ -765,6 +765,19 @@ def build_chunk_tree(chunks: List[Dict[str, Any]]) -> List[Dict[str, Any]]:
     if not chunks:
         return []
     
+    # Debug: Kiểm tra tất cả chunks
+    root_count = 0
+    child_count = 0
+    for chunk in chunks:
+        if chunk.get('cha') is None:
+            root_count += 1
+            logger.debug(f"[TREE] Root chunk: {chunk.get('content', '')[:100]}")
+        else:
+            child_count += 1
+            logger.debug(f"[TREE] Child chunk: {chunk.get('content', '')[:100]} -> parent: {chunk.get('cha')}")
+    
+    logger.info(f"[TREE] Found {root_count} root chunks and {child_count} child chunks from {len(chunks)} total chunks")
+    
     # Tạo dictionary để truy cập nhanh
     chunks_dict = {chunk['id']: chunk for chunk in chunks}
     
@@ -789,21 +802,29 @@ def build_chunk_tree(chunks: List[Dict[str, Any]]) -> List[Dict[str, Any]]:
         }
         
         # Tìm tất cả children của node này
+        children_count = 0
         for other_chunk in chunks:
             if other_chunk.get('cha') == chunk_id:
                 child_node = build_node(other_chunk['id'])
                 node["children"].append(child_node)
+                children_count += 1
         
+        logger.debug(f"[TREE] Node {chunk_id[:8]}... has {children_count} children")
         return node
     
-    # Tìm tất cả root nodes (cha=None)
+    # Tìm tất cả root nodes (cha=None) và sắp xếp theo thứ tự xuất hiện
     root_chunks = []
+    processed_ids = set()  # Để tránh xử lý trùng lặp
+    
     for chunk in chunks:
-        if chunk.get('cha') is None:
+        if chunk.get('cha') is None and chunk['id'] not in processed_ids:
             root_node = build_node(chunk['id'])
             root_chunks.append(root_node)
+            processed_ids.add(chunk['id'])
+            logger.info(f"[TREE] Added root node: {chunk.get('content', '')[:100]}")
     
     logger.info(f"[TREE] Built tree with {len(root_chunks)} root nodes from {len(chunks)} total chunks")
+    logger.info(f"[TREE] Root chunks: {[chunk.get('content', '')[:50] for chunk in root_chunks]}")
     return root_chunks
 
 @app.get("/api/document-chunks/status")