feat: don't create temporary objects for serialization

Fix: #646

core/CMakeLists.txt

@@ -10,6 +10,7 @@ add_library(core
   src/gauge.cc
   src/histogram.cc
   src/info.cc
+  src/iovector.cc
   src/registry.cc
   src/serializer.cc
   src/summary.cc

core/include/prometheus/collectable.h

@@ -1,12 +1,7 @@
 #pragma once
 
-#include <vector>
-
 #include "prometheus/detail/core_export.h"
-
-namespace prometheus {
-struct MetricFamily;
-}
+#include "prometheus/serializer.h"
 
 namespace prometheus {
 
@@ -19,7 +14,7 @@ class PROMETHEUS_CPP_CORE_EXPORT Collectable {
   virtual ~Collectable() = default;
 
   /// \brief Returns a list of metrics and their samples.
-  virtual std::vector<MetricFamily> Collect() const = 0;
+  virtual void Collect(const Serializer& out) const = 0;
 };
 
 }  // namespace prometheus

core/include/prometheus/family.h

@@ -140,7 +140,7 @@ class PROMETHEUS_CPP_CORE_EXPORT Family : public Collectable {
   /// Collect is called by the Registry when collecting metrics.
   ///
   /// \return Zero or more samples for each dimensional data.
-  std::vector<MetricFamily> Collect() const override;
+  void Collect(const Serializer& out) const override;
 
  private:
   std::unordered_map<Labels, std::unique_ptr<T>, detail::LabelHasher> metrics_;

core/include/prometheus/iovector.h

@@ -0,0 +1,27 @@
+#pragma once
+
+#include <cstddef>
+#include <cstdint>
+#include <string>
+#include <vector>
+
+#include "prometheus/detail/core_export.h"
+
+namespace prometheus {
+
+struct PROMETHEUS_CPP_CORE_EXPORT IOVector {
+  using ByteVector = std::vector<std::uint8_t>;
+
+  bool empty() const;
+
+  std::size_t size() const;
+
+  std::size_t copy(std::size_t offset, void* buffer,
+                   std::size_t bufferSize) const;
+
+  void add(const std::string& str, std::size_t chunkSize);
+
+  std::vector<ByteVector> data;
+};
+
+}  // namespace prometheus

core/include/prometheus/metric_family.h

@@ -13,6 +13,5 @@ struct PROMETHEUS_CPP_CORE_EXPORT MetricFamily {
   std::string name;
   std::string help;
   MetricType type = MetricType::Untyped;
-  std::vector<ClientMetric> metric;
 };
 }  // namespace prometheus

core/include/prometheus/registry.h

@@ -79,7 +79,7 @@ class PROMETHEUS_CPP_CORE_EXPORT Registry : public Collectable {
   /// function.
   ///
   /// \return Zero or more metrics and their samples.
-  std::vector<MetricFamily> Collect() const override;
+  void Collect(const Serializer& out) const override;
 
   /// \brief Removes a metrics family from the registry.
   ///

core/include/prometheus/serializer.h

@@ -12,9 +12,10 @@ namespace prometheus {
 class PROMETHEUS_CPP_CORE_EXPORT Serializer {
  public:
   virtual ~Serializer() = default;
-  virtual std::string Serialize(const std::vector<MetricFamily>&) const;
-  virtual void Serialize(std::ostream& out,
-                         const std::vector<MetricFamily>& metrics) const = 0;
+
+  virtual void SerializeHelp(const MetricFamily& family) const = 0;
+  virtual void SerializeMetrics(const MetricFamily& family,
+                                const ClientMetric& metric) const = 0;
 };
 
 }  // namespace prometheus

core/include/prometheus/text_serializer.h

@@ -1,19 +1,27 @@
 #pragma once
 
-#include <iosfwd>
-#include <vector>
+#include <cstddef>
 
 #include "prometheus/detail/core_export.h"
+#include "prometheus/iovector.h"
 #include "prometheus/metric_family.h"
 #include "prometheus/serializer.h"
 
 namespace prometheus {
 
 class PROMETHEUS_CPP_CORE_EXPORT TextSerializer : public Serializer {
  public:
-  using Serializer::Serialize;
-  void Serialize(std::ostream& out,
-                 const std::vector<MetricFamily>& metrics) const override;
+  TextSerializer(IOVector& ioVector);
+
+  void SerializeHelp(const MetricFamily& family) const override;
+  void SerializeMetrics(const MetricFamily& family,
+                        const ClientMetric& metric) const override;
+
+ private:
+  void Add(const std::ostringstream& stream) const;
+
+  IOVector& ioVector_;
+  static constexpr std::size_t chunkSize_ = 1 * 1024 * 1024;
 };
 
 }  // namespace prometheus

core/src/family.cc

@@ -86,22 +86,24 @@ const Labels& Family<T>::GetConstantLabels() const {
 }
 
 template <typename T>
-std::vector<MetricFamily> Family<T>::Collect() const {
+void Family<T>::Collect(const Serializer& out) const {
   std::lock_guard<std::mutex> lock{mutex_};
 
   if (metrics_.empty()) {
-    return {};
+    return;
   }
 
   auto family = MetricFamily{};
   family.name = name_;
   family.help = help_;
   family.type = T::metric_type;
-  family.metric.reserve(metrics_.size());
+
+  out.SerializeHelp(family);
+
   for (const auto& m : metrics_) {
-    family.metric.push_back(std::move(CollectMetric(m.first, m.second.get())));
+    auto&& metric = CollectMetric(m.first, m.second.get());
+    out.SerializeMetrics(family, metric);
   }
-  return {family};
 }
 
 template <typename T>

core/src/iovector.cc

@@ -0,0 +1,56 @@
+#include "prometheus/iovector.h"
+
+#include <algorithm>
+#include <numeric>
+
+namespace prometheus {
+
+bool IOVector::empty() const { return data.empty() || !size(); }
+
+std::size_t IOVector::size() const {
+  return std::accumulate(begin(data), end(data), std::size_t{0},
+                         [](std::size_t size, const ByteVector& chunk) {
+                           return size + chunk.size();
+                         });
+}
+
+std::size_t IOVector::copy(std::size_t offset, void* buffer,
+                           std::size_t bufferSize) const {
+  std::size_t copied = 0;
+  for (const auto& chunk : data) {
+    if (offset >= chunk.size()) {
+      offset -= chunk.size();
+      continue;
+    }
+
+    auto chunkSize = std::min(chunk.size() - offset, bufferSize - copied);
+    std::copy_n(chunk.data() + offset, chunkSize,
+                reinterpret_cast<std::uint8_t*>(buffer) + copied);
+    copied += chunkSize;
+    offset = 0;
+
+    if (copied == bufferSize) {
+      break;
+    }
+  }
+  return copied;
+}
+
+void IOVector::add(const std::string& str, std::size_t chunkSize) {
+  std::size_t size = str.size();
+  std::size_t offset = 0;
+
+  while (size > 0U) {
+    if (data.empty() || data.back().size() >= chunkSize) {
+      data.emplace_back();
+      data.back().reserve(chunkSize);
+    }
+    auto&& chunk = data.back();
+    const auto toAdd = std::min(size, chunkSize - chunk.size());
+    chunk.insert(chunk.end(), str.data() + offset, str.data() + offset + toAdd);
+
+    size -= toAdd;
+    offset += toAdd;
+  }
+}
+}  // namespace prometheus

core/src/registry.cc

@@ -16,11 +16,9 @@ namespace prometheus {
 
 namespace {
 template <typename T>
-void CollectAll(std::vector<MetricFamily>& results, const T& families) {
+void CollectAll(const Serializer& out, const T& families) {
   for (auto&& collectable : families) {
-    auto metrics = collectable->Collect();
-    results.insert(results.end(), std::make_move_iterator(metrics.begin()),
-                   std::make_move_iterator(metrics.end()));
+    collectable->Collect(out);
   }
 }
 
@@ -43,17 +41,14 @@ Registry::Registry(InsertBehavior insert_behavior)
 
 Registry::~Registry() = default;
 
-std::vector<MetricFamily> Registry::Collect() const {
+void Registry::Collect(const Serializer& out) const {
   std::lock_guard<std::mutex> lock{mutex_};
-  auto results = std::vector<MetricFamily>{};
 
-  CollectAll(results, counters_);
-  CollectAll(results, gauges_);
-  CollectAll(results, histograms_);
-  CollectAll(results, infos_);
-  CollectAll(results, summaries_);
-
-  return results;
+  CollectAll(out, counters_);
+  CollectAll(out, gauges_);
+  CollectAll(out, histograms_);
+  CollectAll(out, infos_);
+  CollectAll(out, summaries_);
 }
 
 template <>

core/src/serializer.cc

@@ -4,10 +4,4 @@
 
 namespace prometheus {
 
-std::string Serializer::Serialize(
-    const std::vector<MetricFamily>& metrics) const {
-  std::ostringstream ss;
-  Serialize(ss, metrics);
-  return ss.str();
-}
 }  // namespace prometheus

core/src/text_serializer.cc

@@ -4,6 +4,7 @@
 #include <limits>
 #include <locale>
 #include <ostream>
+#include <sstream>
 #include <string>
 
 #include "prometheus/client_metric.h"
@@ -153,67 +154,77 @@ void SerializeHistogram(std::ostream& out, const MetricFamily& family,
     WriteTail(out, metric);
   }
 }
+}  // namespace
+
+TextSerializer::TextSerializer(IOVector& ioVector) : ioVector_(ioVector) {}
+
+void TextSerializer::SerializeHelp(const MetricFamily& family) const {
+  std::ostringstream out;
 
-void SerializeFamily(std::ostream& out, const MetricFamily& family) {
   if (!family.help.empty()) {
     out << "# HELP " << family.name << " " << family.help << "\n";
   }
   switch (family.type) {
     case MetricType::Counter:
       out << "# TYPE " << family.name << " counter\n";
-      for (auto& metric : family.metric) {
-        SerializeCounter(out, family, metric);
-      }
       break;
     case MetricType::Gauge:
       out << "# TYPE " << family.name << " gauge\n";
-      for (auto& metric : family.metric) {
-        SerializeGauge(out, family, metric);
-      }
       break;
     // info is not handled by prometheus, we use gauge as workaround
     // (https://github.com/OpenObservability/OpenMetrics/blob/98ae26c87b1c3bcf937909a880b32c8be643cc9b/specification/OpenMetrics.md#info-1)
     case MetricType::Info:
       out << "# TYPE " << family.name << " gauge\n";
-      for (auto& metric : family.metric) {
-        SerializeInfo(out, family, metric);
-      }
       break;
     case MetricType::Summary:
       out << "# TYPE " << family.name << " summary\n";
-      for (auto& metric : family.metric) {
-        SerializeSummary(out, family, metric);
-      }
       break;
     case MetricType::Untyped:
       out << "# TYPE " << family.name << " untyped\n";
-      for (auto& metric : family.metric) {
-        SerializeUntyped(out, family, metric);
-      }
       break;
     case MetricType::Histogram:
       out << "# TYPE " << family.name << " histogram\n";
-      for (auto& metric : family.metric) {
-        SerializeHistogram(out, family, metric);
-      }
       break;
   }
+
+  Add(out);
 }
-}  // namespace
 
-void TextSerializer::Serialize(std::ostream& out,
-                               const std::vector<MetricFamily>& metrics) const {
-  auto saved_locale = out.getloc();
-  auto saved_precision = out.precision();
+void TextSerializer::SerializeMetrics(const MetricFamily& family,
+                                      const ClientMetric& metric) const {
+  std::ostringstream out;
 
   out.imbue(std::locale::classic());
   out.precision(std::numeric_limits<double>::max_digits10 - 1);
 
-  for (auto& family : metrics) {
-    SerializeFamily(out, family);
+  switch (family.type) {
+    case MetricType::Counter:
+      SerializeCounter(out, family, metric);
+      break;
+    case MetricType::Gauge:
+      SerializeGauge(out, family, metric);
+      break;
+    // info is not handled by prometheus, we use gauge as workaround
+    // (https://github.com/OpenObservability/OpenMetrics/blob/98ae26c87b1c3bcf937909a880b32c8be643cc9b/specification/OpenMetrics.md#info-1)
+    case MetricType::Info:
+      SerializeInfo(out, family, metric);
+      break;
+    case MetricType::Summary:
+      SerializeSummary(out, family, metric);
+      break;
+    case MetricType::Untyped:
+      SerializeUntyped(out, family, metric);
+      break;
+    case MetricType::Histogram:
+      SerializeHistogram(out, family, metric);
+      break;
   }
 
-  out.imbue(saved_locale);
-  out.precision(saved_precision);
+  Add(out);
 }
+
+void TextSerializer::Add(const std::ostringstream& stream) const {
+  ioVector_.add(stream.str(), chunkSize_);
+}
+
 }  // namespace prometheus

core/tests/CMakeLists.txt

@@ -7,8 +7,8 @@ add_executable(prometheus_core_test
   family_test.cc
   gauge_test.cc
   histogram_test.cc
+  iovector_test.cc
   registry_test.cc
-  serializer_test.cc
   summary_test.cc
   text_serializer_test.cc
   utils_test.cc