feat(silo): error if 2 sequences link to same nodeId, do not error if sequence has no node or node has no sequence

Author: anna-parker

src/silo/common/phylo_tree.cpp

@@ -272,6 +272,9 @@ roaring::Roaring PhyloTree::getDescendents(const TreeNodeId& node_id) {
             dfs(child);
          }
       };
+   if (child_it->second->isLeaf()) {
+      return result_bitmap;
+   }
    dfs(child_it->second);
    return result_bitmap;
 }

src/silo/common/phylo_tree.h

@@ -21,6 +21,7 @@ class TreeNode {
    int depth;
 
    bool isLeaf() { return children.empty(); }
+   bool rowIndexExists() const { return row_index.has_value(); }
 };
 
 class PhyloTree {

src/silo/storage/column/string_column.cpp

@@ -5,6 +5,7 @@
 #include "silo/common/bidirectional_map.h"
 #include "silo/common/string.h"
 #include "silo/common/tree_node_id.h"
+#include "silo/initialize/initialize_exception.h"
 
 using silo::common::String;
 using silo::common::STRING_SIZE;
@@ -42,8 +43,15 @@ void StringColumnPartition::insert(const std::string& value) {
    const String<STRING_SIZE> tmp(value, metadata->dictionary);
    values.push_back(tmp);
    if (metadata->phylo_tree.has_value()) {
-      metadata->phylo_tree->validateNodeExists(value);
       auto child_it = (metadata->phylo_tree->nodes).find(TreeNodeId{value});
+      if (child_it == metadata->phylo_tree->nodes.end()) {
+         return;
+      }
+      if (child_it->second->rowIndexExists()) {
+         throw silo::initialize::InitializeException(
+            fmt::format("Node '{}' already exists in the phylogenetic tree.", value)
+         );
+      }
       child_it->second->row_index = values.size() - 1;
    }
 }

src/silo/storage/column/string_column.test.cpp

@@ -1,6 +1,7 @@
 #include "silo/storage/column/string_column.h"
 
 #include <gtest/gtest.h>
+#include "silo/common/phylo_tree.h"
 
 using silo::storage::column::StringColumnMetadata;
 using silo::storage::column::StringColumnPartition;
@@ -27,6 +28,25 @@ TEST(StringColumnPartition, rawInsertedValuesRequeried) {
    EXPECT_EQ(under_test.getValues()[5].toString(metadata.dictionary), "value 1");
 }
 
+TEST(StringColumnPartition, rawInsertedValuesWithPhyloTreeRequeried) {
+   auto phylo_tree = silo::common::PhyloTree::fromNewickString(
+      "((CHILD2:0.5, CHILD3:1)CHILD:0.1, NOT_IN_DATASET:1.5)ROOT;"
+   );
+   StringColumnMetadata metadata{"string_column", std::move(phylo_tree)};
+   StringColumnPartition under_test(&metadata);
+
+   under_test.insert("CHILD2");
+   under_test.insert("CHILD3");
+   under_test.insert("NOT_IN_TREE");
+
+   EXPECT_EQ(under_test.getValues()[0].toString(metadata.dictionary), "CHILD2");
+   EXPECT_EQ(under_test.getValues()[1].toString(metadata.dictionary), "CHILD3");
+   EXPECT_EQ(under_test.getValues()[2].toString(metadata.dictionary), "NOT_IN_TREE");
+   EXPECT_EQ(under_test.getDescendents("CHILD2").cardinality(), 0);
+   EXPECT_EQ(under_test.getDescendents("CHILD").cardinality(), 2);
+   EXPECT_EQ(under_test.getDescendents("ROOT").cardinality(), 2);
+}
+
 TEST(StringColumn, rawInsertedValuesRequeried) {
    StringColumnMetadata column("string_column");
    StringColumnPartition under_test{&column};