Intern label sets

src/metric.cpp

@@ -16,6 +16,7 @@
 #include <thread>
 #include <type_traits>
 #include <unordered_map>
+#include <unordered_set>
 #include <vector>
 
 #include <immintrin.h>
@@ -81,6 +82,7 @@ static std::string_view arena_copy_string(std::string_view str,
 struct LabelsKey {
   ArenaVector<std::pair<std::string_view, std::string_view>> labels;
 
+  // Arena-owning constructor (copies strings into arena)
   LabelsKey(std::span<const std::pair<std::string_view, std::string_view>> l,
             ArenaAllocator &arena)
       : labels(&arena) {
@@ -302,6 +304,16 @@ struct Metric {
     return *histogramFamilies;
   }
 
+  // Global label interning set to avoid duplicate LabelsKey allocations
+  static auto &get_interned_labels() {
+    using InternSet = std::unordered_set<LabelsKey, std::hash<LabelsKey>,
+                                         std::equal_to<LabelsKey>,
+                                         ArenaStlAllocator<LabelsKey>>;
+    static InternSet *internedLabels =
+        new InternSet(ArenaStlAllocator<LabelsKey>(&get_global_arena()));
+    return *internedLabels;
+  }
+
   // Thread cleanup for per-family thread-local storage
   struct ThreadInit {
     ArenaAllocator arena;
@@ -379,6 +391,27 @@ struct Metric {
 
   // Thread cleanup now handled by ThreadInit RAII
 
+  // Intern labels to avoid duplicate arena allocations
+  static const LabelsKey &intern_labels(
+      std::span<const std::pair<std::string_view, std::string_view>> labels) {
+    auto &interned_set = get_interned_labels();
+
+    // Create temporary lookup key using stack-allocated arena
+    ArenaAllocator lookup_arena(1024); // Small arena for lookups
+    LabelsKey lookup_key{labels, lookup_arena};
+
+    // Use standard hash set lookup
+    auto it = interned_set.find(lookup_key);
+    if (it != interned_set.end()) {
+      return *it;
+    }
+
+    // Not found - create and intern new key
+    LabelsKey new_key{labels, get_global_arena()};
+    auto result = interned_set.emplace(std::move(new_key));
+    return *result.first;
+  }
+
   static Counter create_counter_instance(
       Family<Counter> *family,
       std::span<const std::pair<std::string_view, std::string_view>> labels) {
@@ -386,7 +419,7 @@ struct Metric {
     (void)thread_init;
 
     std::unique_lock<std::mutex> _{mutex};
-    LabelsKey key{labels, get_global_arena()};
+    const LabelsKey &key = intern_labels(labels);
 
     // Validate that labels aren't already registered as callback
     validate_or_abort(family->p->callbacks.find(key) ==
@@ -419,7 +452,7 @@ struct Metric {
       Family<Gauge> *family,
       std::span<const std::pair<std::string_view, std::string_view>> labels) {
     std::unique_lock<std::mutex> _{mutex};
-    LabelsKey key{labels, get_global_arena()};
+    const LabelsKey &key = intern_labels(labels);
 
     // Validate that labels aren't already registered as callback
     validate_or_abort(family->p->callbacks.find(key) ==
@@ -444,7 +477,7 @@ struct Metric {
     (void)thread_init;
 
     std::unique_lock<std::mutex> _{mutex};
-    LabelsKey key{labels, get_global_arena()};
+    const LabelsKey &key = intern_labels(labels);
 
     // Ensure thread state exists
     auto thread_id = std::this_thread::get_id();
@@ -1163,7 +1196,7 @@ void Family<Counter>::register_callback(
     std::span<const std::pair<std::string_view, std::string_view>> labels,
     MetricCallback<Counter> callback) {
   std::unique_lock<std::mutex> _{Metric::mutex};
-  LabelsKey key{labels, Metric::get_global_arena()};
+  const LabelsKey &key = Metric::intern_labels(labels);
 
   // Validate that labels aren't already in use by create() calls
   for (const auto &[thread_id, per_thread] : p->per_thread_state) {
@@ -1186,7 +1219,7 @@ void Family<Gauge>::register_callback(
     std::span<const std::pair<std::string_view, std::string_view>> labels,
     MetricCallback<Gauge> callback) {
   std::unique_lock<std::mutex> _{Metric::mutex};
-  LabelsKey key{labels, Metric::get_global_arena()};
+  const LabelsKey &key = Metric::intern_labels(labels);
 
   // Validate that labels aren't already in use by create() calls
   validate_or_abort(p->instances.find(key) == p->instances.end(),