fix a bug in saving the tree index after removal.

yahoojapan · Dec 11, 2024 · bf646a3 · bf646a3
1 parent 453265d
commit bf646a3
Show file tree

Hide file tree

Showing 6 changed files with 49 additions and 11 deletions.
diff --git a/VERSION b/VERSION
@@ -1 +1 @@
-2.3.4
+2.3.5
diff --git a/lib/NGT/Command.cpp b/lib/NGT/Command.cpp
@@ -1188,8 +1188,8 @@ void NGT::Command::repair(Args &args) {
       }
     } else {
       if (removedIDs.find(id) == removedIDs.end() && id < objSize) {
-        std::cerr << "Not found an object in the tree. However, it might be a duplicated object. " << id
-                  << std::endl;
+        std::cerr << "Not found an object in the tree. However, it might be a duplicated same object. " << id
+                  << "/" << removedIDs.size() << std::endl;
         uninsertedTreeObjectCount++;
         if (repair) {
           try {

diff --git a/lib/NGT/Graph.cpp b/lib/NGT/Graph.cpp
@@ -1001,8 +1001,23 @@ void NeighborhoodGraph::removeEdgesReliably(ObjectID id) {
 #else
         GraphNode::iterator ei = std::lower_bound(n.begin(), n.end(), obj);
         if ((ei == n.end()) || ((*ei).id != obj.id)) {
-          n.insert(ei, obj);
-          insertionA = true;
+          auto idx = distance(n.begin(), ei);
+          bool found = false;
+          for (int i = idx - 1; idx >= 0 && found == false; i--) {
+            if (n[idx - 1].distance != n[i].distance) break;
+            if (n[i].id == obj.id) found = true;
+          }
+          for (int i = idx + 1; idx < (static_cast<int>(n.size()) - 1) && found == false; i++) {
+            if (n[idx + 1].distance != n[i].distance) break;
+            if (n[i].id == obj.id) found = true;
+          }
+          if (found) {
+            std::cerr << "Warning! The distances calculated earlier are different now, "
+                      << "therefore the index might be inconsistent." << std::endl;
+          } else {
+            n.insert(ei, obj);
+            insertionA = true;
+          }
         }
 #endif
       }
@@ -1024,8 +1039,24 @@ void NeighborhoodGraph::removeEdgesReliably(ObjectID id) {
 #else
         GraphNode::iterator ei = std::lower_bound(n.begin(), n.end(), obj);
         if ((ei == n.end()) || ((*ei).id != obj.id)) {
-          n.insert(ei, obj);
-          insertionB = true;
+          auto idx = distance(n.begin(), ei);
+          bool found = false;
+          for (int i = idx - 1; idx >= 0 && found == false; i--) {
+            if (n[idx - 1].distance != n[i].distance) break;
+            if (n[i].id == obj.id) found = true;
+          }
+          for (int i = idx + 1; idx < (static_cast<int>(n.size()) - 1) && found == false; i++) {
+            if (n[idx + 1].distance != n[i].distance) break;
+            if (n[i].id == obj.id) found = true;
+          }
+          if (found) {
+            std::cerr << "Warning! The distances calculated earlier are different now, "
+                      << "therefore the index might be inconsistent." << std::endl;
+          } else {
+            n.insert(ei, obj);
+            insertionB = true;
+          }
+
         }
 #endif
       }

diff --git a/lib/NGT/Node.h b/lib/NGT/Node.h
@@ -456,10 +456,13 @@ class LeafNode : public Node {
     }
 #endif // NGT_NODE_USE_VECTOR
     if (pivot == 0) {
-      // Before insertion, parent ID == 0 and object size == 0, that indicates an empty index
-      if (parent.getID() != 0 || objectSize != 0) {
-        NGTThrowException("Node::write: pivot is null!");
+      if (parent.getID() != 0) {
+        NGTThrowException("Node::write: The pivot is null and the parent is invalid.");
+      }
+      if (objectSize != 0) {
+        NGTThrowException("Node::write: The pivot is null with no object assigned.");
       }
+      // Before insertion, parent ID == 0 and object size == 0, that indicates an empty index
     } else {
 #ifdef NGT_SHARED_MEMORY_ALLOCATOR
       std::cerr << "not implemented" << std::endl;

diff --git a/lib/NGT/Tree.cpp b/lib/NGT/Tree.cpp
@@ -369,6 +369,7 @@ void DVPTree::removeEmptyNodes(InternalNode &inode) {
     LeafNode *ln = new LeafNode;
 #endif
     ln->parent = target->parent;
+    ln->pivot = PersistentObject::allocate(*objectSpace);
     insertNode(ln);
 
     InternalNode &in = *(InternalNode *)getNode(ln->parent);

diff --git a/lib/NGT/Tree.h b/lib/NGT/Tree.h
@@ -446,7 +446,10 @@ class DVPTree {
         auto objs = ln.getObjectIDs();
 #endif
         for (size_t idx = 0; idx < ln.objectSize; ++idx) {
-          ids.insert(objs[idx].id);
+          auto ret = ids.insert(objs[idx].id);
+          if (ret.second == false) {
+            std::cerr << "Warning! Duplicated ID in the tree. ID=" << objs[idx].id << std::endl;
+          }
         }
       }
     }