Adding support for handling YAML Merge Key (#41) - proper keys comparison for merging

include/yaml-cpp/exceptions.h

@@ -87,6 +87,8 @@ const char* const INVALID_ANCHOR = "invalid anchor";
 const char* const INVALID_ALIAS = "invalid alias";
 const char* const INVALID_TAG = "invalid tag";
 const char* const BAD_FILE = "bad file";
+const char* const MERGE_KEY_NEEDS_SINGLE_OR_SEQUENCE_OF_MAPS =
+    "merge key needs either single map or sequence of maps";
 
 template <typename T>
 inline const std::string KEY_NOT_FOUND_WITH_KEY(

src/nodebuilder.cpp

@@ -1,3 +1,4 @@
+#include <algorithm>
 #include <cassert>
 
 #include "nodebuilder.h"
@@ -15,6 +16,7 @@ NodeBuilder::NodeBuilder()
       m_stack{},
       m_anchors{},
       m_keys{},
+      m_mergeDicts{},
       m_mapDepth(0) {
   m_anchors.push_back(nullptr);  // since the anchors start at 1
 }
@@ -69,11 +71,38 @@ void NodeBuilder::OnMapStart(const Mark& mark, const std::string& tag,
   node.set_tag(tag);
   node.set_style(style);
   m_mapDepth++;
+  m_mergeDicts.emplace_back();
+}
+
+void MergeMapCollection(detail::node& map_to, detail::node& map_from,
+                        detail::shared_memory_holder& pMemory) {
+  for (auto j = map_from.begin(); j != map_from.end(); j++) {
+    const auto from_key = j->first;
+    /// NOTE: const_map_to.get(*j->first) cannot be used here, since it
+    /// compares only the shared_ptr's, while we need to compare the key
+    /// itself.
+    ///
+    /// NOTE: get() also iterates over elements
+    bool found = std::any_of(map_to.begin(), map_to.end(), [&](const detail::node_iterator_value<detail::node> & kv)
+    {
+        const auto key_node = kv.first;
+        return key_node->scalar() == from_key->scalar();
+    });
+    if (!found)
+      map_to.insert(*from_key, *j->second, pMemory);
+  }
 }
 
 void NodeBuilder::OnMapEnd() {
   assert(m_mapDepth > 0);
+  detail::node& collection = *m_stack.back();
+  auto& toMerge = *m_mergeDicts.rbegin();
+  /// The elements for merging should be traversed in reverse order to prefer last values.
+  for (auto it = toMerge.rbegin(); it != toMerge.rend(); ++it) {
+    MergeMapCollection(collection, **it, m_pMemory);
+  }
   m_mapDepth--;
+  m_mergeDicts.pop_back();
   Pop();
 }
 
@@ -107,15 +136,40 @@ void NodeBuilder::Pop() {
   m_stack.pop_back();
 
   detail::node& collection = *m_stack.back();
-
   if (collection.type() == NodeType::Sequence) {
     collection.push_back(node, m_pMemory);
   } else if (collection.type() == NodeType::Map) {
     assert(!m_keys.empty());
     PushedKey& key = m_keys.back();
     if (key.second) {
-      collection.insert(*key.first, node, m_pMemory);
-      m_keys.pop_back();
+      detail::node& nk = *key.first;
+      if (nk.type() == NodeType::Scalar &&
+          ((nk.tag() == "tag:yaml.org,2002:merge" && nk.scalar() == "<<") ||
+           (nk.tag() == "?" && nk.scalar() == "<<"))) {
+        if (node.type() == NodeType::Map) {
+          m_mergeDicts.rbegin()->emplace_back(&node);
+          m_keys.pop_back();
+        } else if (node.type() == NodeType::Sequence) {
+          for (auto i = node.begin(); i != node.end(); i++) {
+            auto v = *i;
+            if ((*v).type() == NodeType::Map) {
+              m_mergeDicts.rbegin()->emplace_back(&(*v));
+            } else {
+              throw ParserException(
+                  node.mark(),
+                  ErrorMsg::MERGE_KEY_NEEDS_SINGLE_OR_SEQUENCE_OF_MAPS);
+            }
+          }
+          m_keys.pop_back();
+        } else {
+          throw ParserException(
+              node.mark(),
+              ErrorMsg::MERGE_KEY_NEEDS_SINGLE_OR_SEQUENCE_OF_MAPS);
+        }
+      } else {
+        collection.insert(*key.first, node, m_pMemory);
+        m_keys.pop_back();
+      }
     } else {
       key.second = true;
     }

src/nodebuilder.h

@@ -67,6 +67,7 @@ class NodeBuilder : public EventHandler {
 
   using PushedKey = std::pair<detail::node*, bool>;
   std::vector<PushedKey> m_keys;
+  std::vector<Nodes> m_mergeDicts;
   std::size_t m_mapDepth;
 };
 }  // namespace YAML

test/integration/load_node_test.cpp

@@ -1,6 +1,7 @@
 #include "yaml-cpp/yaml.h"  // IWYU pragma: keep
 
 #include "gtest/gtest.h"
+#include <algorithm>
 
 namespace YAML {
 namespace {
@@ -173,6 +174,78 @@ TEST(LoadNodeTest, CloneAlias) {
   EXPECT_EQ(clone[0], clone);
 }
 
+TEST(LoadNodeTest, MergeKeyA) {
+  Node node = Load(
+      "{x: &foo {a : 1,b : 1,c : 1}, y: &bar {d: 2, e : 2, f : 2, a : 2}, z: "
+      "&stuff { << : *foo, b : 3} }");
+  EXPECT_EQ(NodeType::Map, node["z"].Type());
+  EXPECT_FALSE(node["z"]["<<"]);
+  EXPECT_EQ(1, node["z"]["a"].as<int>());
+  EXPECT_EQ(3, node["z"]["b"].as<int>());
+  EXPECT_EQ(1, node["z"]["c"].as<int>());
+}
+
+TEST(LoadNodeTest, MergeKeyAIterator) {
+  Node node = Load(
+      "{x: &foo {a : 1,b : 1,c : 1}, y: &bar {d: 2, e : 2, f : 2, a : 2}, z: "
+      "&stuff { << : *foo, b : 3} }");
+  EXPECT_EQ(NodeType::Map, node["z"].Type());
+
+  const auto& z = node["z"];
+  size_t z_b_keys = std::count_if(z.begin(), z.end(), [&](const detail::iterator_value & kv)
+  {
+    return kv.first.as<std::string>() == "b";
+  });
+  ASSERT_EQ(z_b_keys, 1);
+}
+
+TEST(LoadNodeTest, MergeKeyTwoOverrides) {
+  Node node = Load(R"(
+trait1: &t1
+  foo: 1
+
+trait2: &t2
+  foo: 2
+
+merged:
+  <<: *t1
+  <<: *t2
+)");
+  EXPECT_EQ(NodeType::Map, node["merged"].Type());
+  EXPECT_FALSE(node["merged"]["<<"]);
+  EXPECT_EQ(2, node["merged"]["foo"].as<int>());
+}
+
+TEST(LoadNodeTest, MergeKeyB) {
+  Node node = Load(
+      "{x: &foo {a : 1,b : 1,c : 1}, y: &bar {d: 2, e : 2, f : 2, a : 2}, z: "
+      "&stuff { << : *foo, b : 3}, w: { << : [*stuff, *bar], c: 4 }, v: { '<<' "
+      ": *foo } , u : {!!merge << : *bar}, t: {!!merge << : *bar, h: 3} }");
+  EXPECT_EQ(NodeType::Map, node["z"].Type());
+  EXPECT_EQ(NodeType::Map, node["w"].Type());
+  EXPECT_FALSE(node["z"]["<<"]);
+  EXPECT_EQ(1, node["z"]["a"].as<int>());
+  EXPECT_EQ(3, node["z"]["b"].as<int>());
+  EXPECT_EQ(1, node["z"]["c"].as<int>());
+
+  EXPECT_EQ(2, node["w"]["a"].as<int>());
+  EXPECT_EQ(3, node["w"]["b"].as<int>());
+  EXPECT_EQ(4, node["w"]["c"].as<int>());
+  EXPECT_EQ(2, node["w"]["d"].as<int>());
+  EXPECT_EQ(2, node["w"]["e"].as<int>());
+  EXPECT_EQ(2, node["w"]["f"].as<int>());
+
+  EXPECT_TRUE(node["v"]["<<"]);
+  EXPECT_EQ(1, node["v"]["<<"]["a"].as<int>());
+
+  EXPECT_FALSE(node["u"]["<<"]);
+  EXPECT_EQ(2, node["u"]["d"].as<int>());
+
+  EXPECT_FALSE(node["t"]["<<"]);
+  EXPECT_EQ(2, node["t"]["d"].as<int>());
+  EXPECT_EQ(3, node["t"]["h"].as<int>());
+}
+
 TEST(LoadNodeTest, ForceInsertIntoMap) {
   Node node;
   node["a"] = "b";